The Late Roman Army in the Near East from Diocletian to the Arab Conquest: Proceedings of a colloquium held at Potenza, Acerenza and Matera, Italy (May 2005) 9781407331997, 9781407301617

Citation preview

BAR  S1717  2007   LEWIN & PELLEGRINI (Eds)   THE LATE ROMAN ARMY IN THE NEAR EAST

B A R

The Late Roman Army in the Near East from Diocletian to the Arab Conquest Proceedings of a colloquium held at Potenza, Acerenza and Matera, Italy (May 2005) Edited by

Ariel S. Lewin Pietrina Pellegrini with the aid of

Zbigniew T. Fiema and Sylvain Janniard

BAR International Series 1717 2007

The Late Roman Army in the Near East from Diocletian to the Arab Conquest Proceedings of a colloquium held at Potenza, Acerenza and Matera, Italy (May 2005) Edited by

Ariel S. Lewin Pietrina Pellegrini with the aid of

Zbigniew T. Fiema and Sylvain Janniard

BAR International Series 1717 2007

ISBN 9781407301617 paperback ISBN 9781407331997 e-format DOI https://doi.org/10.30861/9781407301617 A catalogue record for this book is available from the British Library

BAR

PUBLISHING

Contents Second to Third Century Prolegomeni ad un congresso: considerazioni sull’esercito romano dell’alto Impero........................................................ 1 Giovanni Brizzi L’armée romaine en mer Rouge et autour de la mer Rouge aux IIème et IIIème siècles apr. J.-C.: à propos de deux inscriptions latines découvertes sur l’archipel Farasan......................................................................... 13 Francois Villeneuve New Light on the Roman Military Base at Dura-Europos: Interim Report on a Pilot Season of Fieldwork in 2005 ................................................................................................... 29 Simon James From Praepositus Pretenturae to Dux Ripae.The Roman ‘Grand Strategy’ on the Middle Euphrates (2nd – 3rd Cent. AD) ........................................................................................................................................................ 49 Tommaso Gnoli The Roman Legionary Camp at Legio, Israel: Results of an Archaeological Survey and Observations on the Roman Military Presence at the Site...................................................................................................................... 57 Yotam Tepper The development of the Roman forces in northeastern Anatolia. New evidence for the history of the exercitus Cappadocicus.......................................................................................................................................... 73 Michael A. Speidel Late Third to Seventh Century The Nabataeaan – Roman Negev in the Third Century CE .............................................................................................. 91 Tali Erickson-Gini The Paved Road from Petra to the ‘Arabah – Commercial Nabataean or Military Roman? ......................................... 101 Chaim Ben-David Be’er Shema – Birsama of the Notitia Dignitatum: A Prolegomenon to the 2006 Excavations..................................... 111 Benjamin J. Dolinka Crossing the Negev in Late Roman Times: the Administrative development of Palaestina Tertia Salutaris and of its Imperial Road Network......................................................................................................... 119 Israel Roll From the Constructive Technique Used in the Military Architecture of the Limes Arabicus to its Conservation and Valorisation ............................................................................................................................. 131 Giuseppe Claudio Infranca Military Architecture of the Limes Arabicus: Strategy, Commercial Routes, Quotidian Life........................................ 143 Maysoun Al-Khouri Economic Conditions, Security Problems and the Deployment of the Army in Later Roman Palestine. Part I: Economy and Population ..................................................................................................................................... 153 Israel Shatzman Fluctuating Provincial Borders in Mid-4th Century Arabia and Palestine ...................................................................... 201 Joonas Sipilä Zosimos II.34.1 and ‘The Constantinian Reform’: Using Johannes Lydos to Expose an Insidious Fabrication ............ 211 Peter Brennan

i

Two Sieges of Amida (AD 359 and 502-503) and the Experience of Combat in the Late Roman Near East................ 219 Noel Lenski Naval Operations During Persian Expedition of Emperor Julian (363 AD)................................................................... 237 Edward Dąbrowa ‘Amr Ibn ‘AdƯ, Mavia, the Phylarchs and the Late Roman Army: Peace and War in the Near East.............................. 243 Ariel S. Lewin L'armée romaine et la défense de la Syrie du Sud. Questions de méthode et nouveau document .................................. 263 Maurice Sartre ࡣͱ͝ࢲͧ͟͢͡, ͇ࢼ͒ͧ͜͟͢͡ and ̈́ͤ͒ͤ͘͡ ͔͘ࢹ͢ ͟͝uࢲ͕ͩ͝ in Roman Arabia (1st – 3rd Century). Central Power, Local Administration, and Nomadic Environment ................................................................................ 275 Thomas Brüggemann Moines, militaires et défense de la frontière orientale au VIe s...................................................................................... 285 Geoffrey Greatrex Archaeological and Ancient Literary Evidence for a Battle near Dara Gap, Turkey, AD 530: Topography, Texts and Trenches.................................................................................................................................... 299 Christopher Lillington-Martin The Byzantine Military in the Petra Papyri - a Summary.............................................................................................. 313 Zbigniew T. Fiema The Operational Methods of the Late Roman Army in the Persian War of 572-591 ..................................................... 321 Frank R. Trombley Una guarnigione bizantina a Roma nell’età di Gregorio Magno: i Theodosiaci............................................................. 357 Pietrina Pellegrini Jews, Rats, and the Battle of Yarmnjk ............................................................................................................................. 367 David Woods The Organization of the Army Cavalry in Late Roman Warfare..................................................................................................................................... 377 Hugh Elton Centuriones ordinarii et ducenarii dans l’armée romaine tardive (IIIe -VIe s. apr. J.-C.) .............................................. 383 Sylvain Janniard Campidoctores Vicarii vel Tribuni: The Senior Regimental Officers of the Late Roman Army and the Rise of the Campidoctor .................................................................................................................................... 395 P. Rance From the Late Roman to the Early Byzantine Army.Two Aspects of Change ............................................................... 411 Oliver Schmitt Egypt L’occupation militaire tardive dans les oasis d’Égypte. L’exemple de Douch............................................................... 421 M. Reddé Africa L’armée romaine d’Afrique de 375 à 439: Mythes et réalités ........................................................................................ 431 Yann Le Bohec

ii

Acknowledgements The congress “The late Roman army in the Near East from Diocletian to the Arab conquest” was held on 10 – 14th May 2005 at the Università della Basilicata at Potenza and Matera and at the Sala per conferenze del comune di Acerenza. The organization of the congress was supported by the following institutions: Ministero dell’Università e della Ricerca; Università della Basilicata; Comunità Montana dell’Alto Bradano; Comune di Acerenza; Provincia di Matera; Ente Parco della Murgia Materana. The editors thank prof. Antonio Lerra for his generous help in the organization of the congress.

Introduction This present volume includes the papers delivered at the Congress, with additional submissions from Dolinka, Rance and Speidel, who were not present at the congress but supplied their papers at a later stage. Fiema chaired one of the sessions and although not delivering a paper at the congress was able to include one for publication. Pierre-Louis Gatier delivered a paper, but was not able to supply it for publication. John Casey, Guido Clemente, Wolf Liebeschuetz and Roger Tomlin did not deliver papers, but chaired sessions. Relating to the theme of the congress, most papers dealt with the Late Roman army in the Near East from Diocletian to the Arab conquest. Nevertheless it should be noted that the paper by Brizzi is intended as an introduction to the passage from an Early Empire army to one of Late Antiquity. The topics proposed by Villeneuve, James and Gnoli were of great interest to the organizers of the congress although pertaining to the second and third centuries. Tepper’s paper focuses on the history of a military camp in Israel, from its construction in the second century A.D. to its abandonment, probably in Late Antiquity. Speidel’s paper deals with the history of the army in Cappadocia from Early Empire to Late Antiquity. Shatzman’s paper represents the first part of a large body of research. It does not specifically focus on the army, but is intended as the platform for his broad work in progress on the history of the Late Roman army in Palaestina. Reddé and Le Bohec were invited to deliver a special paper on the Late Roman army in Egypt and Africa. Finally, although not presenting at the conference, Pellegrini has provided a contribution on a military unit acting in Italy, supposed by some scholars to have originated from the Near Eastern army.

iii

iv

Prolegomeni ad un congresso: considerazioni sull’esercito romano dell’alto Impero Giovanni Brizzi Università di Bologna

1

La relazione introduttiva ad un convegno risponde di solito, per necessità o per convenzione, soprattutto alle esigenze di una sintesi preliminare; e così sarà, inevitabilmente, anche oggi. Chi vi parla ha trattato in prevalenza, nel suo iter di storico militare antico, d’armamento, di tattiche, di mentalità; ma ha prudentemente fermato i suoi studî proprio sul limitare dell’epoca che interessa i partecipanti a questo Colloquio, contentandosi di toccare soprattutto i primi tre secoli dell’era nostra. È forse per questo che l’amico Ariel mi ha invitato a tracciare qui, in una sorta di premessa al tema del Colloquio stesso, un profilo di massima dell’esercito romano in età altoimperiale.

decidere anche sacralmente l’esito delle guerre. Ciò avviene, a parere di molti, perché ad opporsi all’Urbe non sono più Stati organicamente costituiti, ma popolazioni barbare, e perciò sentite da un lato come prive di un’identità vera e propria, dall’altro forse incapaci davvero di accettare il verdetto di una battaglia campale; sicché vengono a mancare, anche in guerra, le convenzioni ritenute necessarie in passato a formalizzare apertura, svolgimento e chiusura delle ostilità. A mio avviso, tuttavia, ciò è vero soltanto in parte. Questa prassi sembra, infatti, non essere stata sempre del tutto ignota, per esempio, a quelle genti germaniche che costituirono il principale nemico centroeuropeo di Roma: vi era stato un tempo in cui, nella piana del Po, Boiorix, il re dei Cimbri, poteva non solo sfidare Caio Mario a battaglia, ma invitarlo a fissare con lui la data e il luogo 2 di uno scontro evidentemente considerato decisivo. Ora però questo tempo era passato; e, se è vero che forse difficilmente i barbari avrebbero accettato ancora di rimettersi ad un simile verdetto, è vero altresì che i Romani consideravano ormai ogni realtà come “either res 3 Romana or res nullius”. Credo che pochi concetti possano evidenziare questo stato di cose come la nozione di limes, almeno nel significato oggi più comunemente attribuito alla parola. Poco familiare alle popolazioni limitrofe all’impero, dedite talvolta addirittura a forme di seminomadismo, l’idea di un confine rigido verso il barbaricum era stata ormai accantonata quasi ovunque dai Romani stessi, i quali avevano finito per intenderla abitualmente non nella forma di una linea ideale tracciata sul terreno; ma, in modo assai più pragmatico, come una fascia dalla profondità variabile il cui rispetto era affidato, secondo i rapporti di forza esistenti in loco, all’energia o alla discrezione dei singoli governatori provinciali. Ciò sia a causa dell’ormai consolidato principio della pax Romana; sia del fatto che, mancando ogni accordo con una controparte politica riconosciuta, veniva a mancare sostanzialmente anche ogni rispetto delle regole. Un simile vacuum non poteva, naturalmente, non estendersi ai canoni della guerra in generale.

Troppo vasto per poter essere esaurito nello spazio di una semplice relazione, un tema simile esige inevitabilmente delle scelte, chiede, cioè, di essere inquadrato in via preliminare e circoscritto entro limiti precisi che in qualche modo lo definiscano; ma queste scelte debbono privilegiare comunque, secondo me, il requisito essenziale della riflessione e dello sguardo complessivo. Riprenderò dunque in questa sede un tema già trattato, la funzionalità d’insieme dell’esercito altoimperiale, cercando di arricchirlo, per quanto possibile, con elementi di novità; ma curando soprattutto di fornire i dati necessarî al confronto con la struttura tardoantica che costituisce il tema di questo colloquio. Proviamo dunque a partire da una constatazione solo in apparenza sorprendente: nei confronti dei nemici esterni l’età altoimperiale è caratterizzata dall’assenza quasi assoluta di grandi battaglie campali. Se contro le popolazioni del centro Europa non mancano certo conflitti anche di vaste proporzioni, questi assumono talvolta, da una parte e dall’altra, la veste inquietante dell’invasione, più spesso vivono la prassi limitata della reciproca incursione o della scorreria rapida, della razzia o della rappresaglia a devastare il territorio nemico, ma conoscono in ogni caso soprattutto le forme elementari della guerriglia e dell’imboscata; e, anche quando danno vita a fatti d’arme di maggiore rilievo, questi non trascendono però quasi mai il livello dello scontro disorganico e senza regole e mancano per lo più dei caratteri che avevano connotato invece le ‘batailles rangées’ – come felicemente le chiamano i Francesi –, le battaglie campali dei secoli precedenti, i iusta proelia dei Romani, quei confronti cioè secondo crismi e liturgie prestabilite che, per il carattere agonale riconosciuto loro (la definizione è di Johann Huizinga), erano state chiamate spesso, in passato, a

Non può essere questa, tuttavia, la sola risposta. Esiste, infatti, un settore almeno delle frontiere imperiali lungo il 1 Basti, qui, ricordare l’affermazione di Polibio (XXXV.1) secondo cui, al tempo suo, «in Grecia e in Asia una sola battaglia o, più raramente due, erano sufficienti per decidere l’esito di una guerra...». Per un commento cfr. Walbank 1979: 640-641. 2 Plut., Mar. 25.4-5. 3 Così Alföldi 1952: 5.

1

THE LATE ROMAN ARMY IN THE EAST 9

quale il limes segue davvero una linea netta e precisa; e la definizione degli spazî tra poteri i quali non solo si rispettano e si temono, ma evidentemente si riconoscono in via ufficiale, è precisa al punto che, per sottolinearla, gli incontri diplomatici tra i rispettivi rappresentanti avvengono talvolta su isolotti situati al centro del fiume 4 che separa il territorîo delle opposte potenze. Il fiume in questione è, ovviamente, l’Eufrate e il potere alternativo a Roma è quello dell’impero partico. Eppure finanche tra questi nemici le fonti non ricordano quasi mai autentiche battaglie campali. I grandi episodî risolutivi appaiono 5 singolarmente rari se non assenti; e se, tra Augusto e i primi anni di Marco Aurelio non accade mai che i Parti penetrino nello spazio romano, i iusta proelia mancano anche nel caso opposto, quando cioè sono i Romani a invadere il territorio nemico, e persino quando – cosa che, a partire dall’età di Traiano, accade per ben tre volte in meno di un secolo – le armate imperiali si spingono ad assalire e distruggere le capitali d’Occidente del regno 6 7 arsacide. Evidente in occasione di tutti questi conflitti, tale atteggiamento assume proporzioni addirittura sconcertanti al tempo dell’Optimus Princeps. Per esplicita testimonianza di Cassio Dione, infatti. le forze romane dilagano in territorio partico senza incontrare alcuna vera resistenza; e intere regioni si arrendono senza 8 combattere, al punto che l’imperatore, sorpreso e preoccupato per l’inerzia del nemico, reputa addirittura necessario mantener vigili le sue truppe con periodiche

esercitazioni ed allarmi simulati, evidentemente allo scopo di evitar loro pericolosi rilassamenti. Più volte lo stesso Dione sottolinea come spazi enormi vengano traversati dalle legioni nella più totale “assenza di 10 nemici”: e ciò fino all’occupazione della Babilonia e addirittura fino alla presa di Ctesifonte, che Osroes abbandona in gran fretta, lasciando nelle mani del vincitore una figlia e soprattutto il trono d’oro, simbolo della regalità arsacide. Che vi sia stata, durante questa guerra, una vera e propria controffensiva partica 11 personalmente non lo credo affatto; e dunque, poiché su una cosa almeno penso si possa convenir tutti senza eccezione – e cioè sulla superiore mobilità di un’armata come quella orientale, composta essenzialmente di cavalieri – è evidentemente proprio questa forza, che potrebbe costringere il nemico a battersi e non lo fa, ad evitare il confronto, mostrando con ciò una ritrosia che pare senz’altro venata di paura.

9

Cass.Dio LXVIII.23.2. LXVIII.26.42. 11 Non manca, naturalmente, chi è di parere del tutto contrario: tra gli altri Dierauer 1868: 154, nota 4; Petersen 1968: 214; Lepper 1948: 188. Anche di recente qualcuno (Bennett 1997: 201) si è detto convinto che sia stata la controffensiva arsacide in Mesopotamia a permettere lo scoppio della vastissima rivolta ebraica; mentre altri (come Cizek1994: 376-385, part. 384) hanno supposto che i primi a sollevarsi siano stati i Giudei della Partia e che la loro azione abbia poi spinto ad insorgere quei correligionarî i quali vivevano ormai da tempo dispersi nelle provincie orientali dell’impero romano. Queste ipotesi, tuttavia, sembrano da scartare entrambe: benché, infatti, l’interdipendenza fra i diversi teatri dell’insurrezione sia ammessa sostanzialmente da tutti (cfr., tra gli altri, Pucci 1981: 98-99 e 104-119; Bruun 1992: 99, n. 3; Firpo 1999: 72), sembra certo che a muoversi per prime siano state proprio le comunità ebraiche che abitavano le regioni al di qua dell’Eufrate, grazie anche al fatto che se ne erano momentaneamente allontanati i presidî; mentre la diaspora babilonese non si sollevò che in un secondo momento, quando la guerriglia divampata nelle retrovie costrinse le forze romane ad allentare la morsa sulle terre di recente conquista. Non vi è motivo di dubitare dell’asserto di Eusebio (HE IV.2.1) secondo cui il focolaio iniziale della rivolta va cercato in Cirenaica durante il XVIII anno di Traiano, vale a dire nel 115/116; anche perché il dato sembra confermato da alcuni documenti papiracei, i quali parlano di disordini già in corso qui e in Egitto prima del 117 (cfr. Fuks 1953: 131-158); prima cioè, che – chiunque sia stato a riaccendervi la lotta – cominciassero ad agitarsi le regioni della bassa Mesopotamia. Firpo 2001: 233-245, propone uno stimolante collegamento con il sisma che devastò Antiochia, avanzando l’ipotesi che, letto dagli estremisti ebraici in termini apocalittici, tale evento sia stato tra i fattori scatenanti della rivolta. Quanto a me, della realtà di una ‘controffensiva’ partica continuo a dubitare: non solo, infatti, non vedo di quali nuove energie questa avrebbe potuto alimentarsi (nulla, nella situazione interna del regno arsacide, era veramente cambiato rispetto all’inizio della guerra...), ma mi sembra addirittura che il comportamento attendistico abbia finito per configurarsi, da parte degli Arsacidi, come una sorta di opzione strategica obbligata, che fu adottata di nuovo circa un secolo dopo. Al di là del fatto che la mia opinione trovi un parziale conforto nel parere di autorevoli studiosi (come lo Henderson 1949: 127 s.; secondo cui l’azione di Sanatrukes e di Parthamaspates andrebbe addirittura datata ad un periodo anteriore al 116/117), una constatazione resta, comunque, inevitabile malgrado tutto: qualunque risposta si voglia dare al quesito iniziale, non fu la ‘controffensiva’ partica a favorire l’insurrezione ebraica, dovunque essa sia cominciata, ma furono i Parti a giovarsi di quest’ultima per tentar di allontanare i Romani dalle loro terre. Del problema, ad ogni modo, mi sono già occupato altrove (cfr. Brizzi 2002: 167 ss.; Brizzi, in corso di stampa); e non è mia intenzione ritornarvi qui. 10

4 Per quanto riguarda l’incontro di Caio Cesare con gli ambasciatori partici: Cass.Dio LV.10a; Hier., ad a.2017. Cfr. Iust. XLI.1.1: Parthi penes quos velut divisione orbis cum Romanis facta nunc Orientis imperium est; sulla spartizione del mondo anche Strabo XI.515. 5 Forse il solo evento militare di autentico rilievo nel corso di questi conflitti fu la battaglia cosiddetta di Dura Europos, vinta da Avidio Cassio durante la guerra partica di Lucio Vero. Di questo fatto d’arme restano, purtroppo, solo i brevissimi accenni trasmessi da Luciano (hist. conscr. 19.20; 28): certo, come ammette l’autore stesso, non si può liquidare uno scontro simile “in neppure sette parole”, ed è poco credibile che i Parti abbiano perduto oltre 70 mila dei loro contro due soli morti (e nove feriti) romani; ma dovette trattarsi senz’altro di una grande battaglia (e di una strepitosa vittoria romana). Altri combattimenti, anch’essi vittoriosi, sostennero, nel corso di questa stessa guerra, ancora Avidio Cassio a Circesium, sull’Eufrate, e, dopo la presa di Nisibi, forse P. Marzio Vero lungo il corso del Tigri; mentre, durante il conflitto combattuto da Traiano uno scontro forse più impegnativo di altri fu quello che portò alla disfatta delle forze di Sanatrukes. Tutti questi, però, sembrano episodî in qualche modo minori; e, per quanto possa apparire strano, ne sappiamo ancor meno che della battaglia di Dura Europos. 6 Questa è un’osservazione che ho già avanzato più volte. Per un’analisi più dettagliata degli eventi rinvio qui ad alcuni miei lavori passati (cui sarà inevitabile fare, anche in seguito, costante riferimento): Brizzi 1981: 191-193; Brizzi 1983b: 25-27 e soprattutto Brizzi 2002: 165-171. 7 Tranne, forse, che in quello combattuto sotto Marco Aurelio. V. supra, nota 5. La differenza rispetto alle altre due guerre consiste nel fatto che questa era stata cominciata dai Parti stessi; i quali, quindi, si sentirono probabilmente impegnati a combatterla fino in fondo. Quanto invece alle campagne di Settimio Severo, se durante la prima di esse «non si registra apparentemente opposizione di sorta alla penetrazione romana nel territorio partico», la seconda non dura, tra l’imbarco a Brindisi e la presa di Ctesifonte, «neppure sei mesi» (così Calderini 1949: 59 e 70). 8 Il termine ¢maceˆ ritorna più volte in Cassio Dione – LXVIII.18.3b; 19.2 –; e compare – a 19.4 – l’espressione n…chn...¥naimon, vittoria senza spargimento di sangue.

2

GIOVANNI BRIZZI: PROLEGOMENI AD UN CONGRESSO

che l’esercito di Roma abbia adeguato in fretta le sue strutture, annullando di fatto molto rapidamente la minaccia militare partica; ciò attraverso un processo il quale, secondo me, era già in gran parte compiuto nel momento in cui si generalizzarono l’introduzione e l’impiego della lorica segmentata e del pilum pesante. Proprio questi strumenti, infatti, contribuirono in larga misura a mettere le legioni in grado di fronteggiare con 15 successo le cavallerie orientali. Mi sia concesso, qui, di soffermarmi alquanto su quest’ultimo punto, poiché esso sta alla base di tutto il mio ragionamento ulteriore. Formulata oltre un ventennio fa, la mia ipotesi non ha avuto finora, debbo ammetterlo, molta fortuna; ma ciò anche a causa, credo di poterlo dire, dell’atteggiamento aprioristico e forse non sempre scientificamente corretto di un settore almeno della critica. Mentre infatti uno dei presupposti avanzati allora – quello secondo cui la lorica segmentata, preziosa ovunque, dovette offrire il contributo più rilevante proprio sul fronte dell’Eufrate, come protezione efficace contro le frecce degli 16 hippotoxotai – è stato confutato spesso; l’altro, quello relativo all’impiego prevalente del pilum pesante come arma decisiva contro le cavallerie corazzate – e soprattutto contro i catafratti partici o sarmatici – è stato, viceversa, ripreso più volte, senza però che il mio nome 17 sia stato citato quasi mai.

Proprio questa potrebbe, dunque, configurarsi come una significativa ed importante concausa a spiegare l’assenza, in età altoimperiale, di quelli che abbiamo definito, con Huizinga, gli scontri agonali: quello romano era, forse, un esercito al quale tutti i nemici si guardavano ormai bene dall’offrirli. E tuttavia se al fatto che le armate dell’Urbe incutessero un salutare spavento si può senz’altro credere quando si parla dei barbari del centro Europa – i quali, frazionati in una miriade di tribù spesso in lotta tra loro, rimasero a lungo incapaci di minacciare seriamente l’impero – la spiegazione lascia invece perplessi quando è riferita ad un organismo militare da sempre ritenuto fra i più temibili del mondo antico. Se, infatti, i moderni si sono dichiarati spesso convinti che, potendo esplicare appieno «la classica combinazione di potenza di tiro e 12 forza d’urto», la simbiosi tra cavalleria pesante ed 13 arcieri a cavallo «revolutionised ancient warfare»; questa opinione sembra, cosa ancor più singolare, essere stata in certo qual modo recepita dai Romani stessi. Su di essi la disfatta di Carrhae ha avuto un effetto traumatico forte e duraturo al punto di conferire all’evento «il valore simbolico di scontro tra Occidente e Oriente» e di dare origine ad una produzione scritta «che si è poi rinnovata con puntuale ricorrenza» in occasione di ogni conflitto contro gli Arsacidi.14 Per tutta l’età altoimperiale resta, nondimeno, inconfutabile un fatto. Se sono stati i Romani ad essere sconfitti nel primo scontro campale sostenuto con i Parti, sono stati tuttavia sempre loro, in seguito, ad assumere l’iniziativa, invadendo ripetutamente, tra gli inizî del II e quelli del III secolo d.C., le terre del nemico orientale e provocandone le armate a battaglia. Certo, secondo un atteggiamento topico assai frequente, l’interesse per il nemico e l’innata tendenza romana ad esaltarlo dovettero generare anche in questo caso un’eco ricorrente nella letteratura di Roma imperiale, certo dovettero dar vita ad una sorta, se così si può dire, di vezzo culturale. Credo però che questo fenomeno sia stato capace di produrre soprattutto testi retorici o celebrativi e forse qualche trattato etnografico, simile in fondo, anche come impostazione, alla Germania di Tacito; e non, se non in misura assai ridotta, opere tecniche volte alla reale soluzione militare del problema partico o trattazioni storiche vere e proprie. Queste ultime comunque, quando sono sopravvissute, sembrano – si tratti di Cassio Dione o di Cornelio Frontone – mostrarci, sia pur di riflesso, il problema come assai meno grave di quanto comunemente si immagini.

Riconsideriamo dunque, quest’idea anche alla luce delle conoscenze più recenti. Delle due minacce portate dalle cavallerie arsacidi quella costituita dai catafratti era forse la meno grave: la loro ridotta autonomia e la loro scarsa 18 19 manovrabilità, colte già dalle fonti antiche, sono state ulteriormente sottolineate da alcuni studiosi moderni; i quali sono giunti alla conclusione che «la loro terribilità» consistesse «forse...più nell’aspetto che nelle reali 20 possibilità di utilizzazione». Sicuramente inferiori alle 21 cavallerie leggere, i catafratti potevano essere messi in 15

Cfr. Brizzi 1981: 177-201; Brizzi 1983b: 9-30. Cfr., ad esempio, Bishop-Coulston 1993: 85. 17 E benché, si noti, i due presupposti siano formulati insieme, l’uno a complemento dell’altro, nel contesto del medesimo articolo: cfr. Brizzi 1981: 183-184, 195-195, 198 s. 18 Alcuni dei primi lavori relativi alla cavalleria catafratta sono in lingua russa: dell’evoluzione di questo corpo e di quello degli arcieri montati prima della battaglia di Carrhae parla, per esempio, Pugachenkova 1966: 27-43. Cfr. anche: Chazanov 1968: 185-191. Su di essi si vedano, inoltre, Rubin 1955: 264-283; Eadie 1967: 161-173; Gamber 1968: 14 ss.; Bivar 1972: 273-291; Speidel 1984: 151-156; Coulston 1986: 60-68, 71-75; Hyland 1990: 148-156; Mielczarek 1995; Mielczarek 1999: 101105; Harl 1996: 601-627. 19 Dovuta al peso e all’ingombro delle corazze a protezione del cavallo e del cavaliere Iust. XLI.2.10: ceterum intolerandi forent, si quanta his est impetus vis, tanta et perseverantia esset. Cfr. Plut., Crass. 18; Heliod., Aethiopica, IX.15.6. 20 Così Gabba 1974: 32. Del tutto analogo è il commento che Cipolla 19992: 8 riserva all’omologa formazione del Medio Evo: «molto pittoresca, ma poco maneggevole». 21 Credo che quanto mai opportuno sia, a questo proposito, un parallelo con le cavallerie medievali. L’Europa, forse soprattutto quella tedesca, faceva affidamento soprattutto sulla cavalleria pesante, gravemente inferiore rispetto alle cavallerie leggere orientali, mongole o turche che fossero: ancora il Montecuccoli (Aforismi, lib. III, cap. VII, p. 236) 16

Solo apparente, la contraddizione interna a questo quadro può in realtà essere risolta semplicemente ammettendo 12

Così Luttwak 1981: 65. Così Raschke 1978: 821 nota 727. Sia pure ormai datato (Rostovtzeff 1935: 162-163; Debevoise 1938: 93; e in certa misura, malgrado l’opportuno scetticismo sulla reale forza dei catafratti, persino Gabba 1974: 9 ss.; in particolare p. 14), il topos che vuole i Parti militarmente superiori ai Romani almeno sul piano tattico sembra, presso la moderna critica, assai difficile da sfatare. 14 Così Gabba 1974: 14 e nota 16; p. 10 e nota 10; 9-11. 13

3

THE LATE ROMAN ARMY IN THE EAST grave difficoltà anche da formazioni organizzate di fanteria. Poiché il loro impiego, oltre ad essere limitato alla pianura, si riduceva per forza ad una sola, breve carica, sia pur travolgente, al termine della quale occorreva far rifiatare i cavalli, i catafratti divenivano vulnerabili all’attacco di una fanteria organizzata ed armata come quella romana, purché venissero assaliti prima di poter prendere lo slancio o subito dopo averlo esaurito. Lo conferma secondo me oltre ogni dubbio un passo di Tacito, sia pur riferito all’analoga realtà sarmatica: quasi irresistibili quando caricano (ubi per turmas advenere vix ulla acies obstiterit...), i catafratti sono invece incapaci di battersi se vengono attaccati da fermi e soprattutto se vengono scavalcati (...nihil ad pedestrem pugnam tam ignavum...): in questo caso, infatti, è lo stesso tegimen, ferreis laminis aut praeduro corio consertum, che si rivela impetu hostium 22 provolutis,...inhabile ad resurgendum. Si deve dunque concludere che le cavallerie corazzate, a disagio non appena perdono la loro mobilità, possono esser contrastate sia adottando uno schieramento di fanti 23 compatto e profondo, impossibile da sfondare, sia disponendo veli sottili di truppe seguiti a breve distanza 24 da corpi di rincalzo più consistenti: qualora si ostinino ad attaccare schieramenti dell’uno o dell’altro tipo, una volta esaurita la loro forza d’urto, i catafratti, ormai fermi e incapaci di girarsi rapidamente, potranno essere a loro volta attaccati e distrutti.

quella dell’animale al galoppo, il nuovo tipo di giavellotto avrà, su cavalieri sprovvisti di staffe che caricano in ranghi serrati, effetti devastanti. Capace di trapassare le corazze con esito letale e, comunque, di disarcionare il cavaliere con la semplice forza dell’impatto, il pilum pesante può ottenere risultati persino superiori se colpisce il cavallo: cadendo, infatti, l’animale trascina di solito con sé anche i destrieri che seguono, impacciati dalle gualdrappe e gravati dal peso dei loro cavalieri, e incapaci perciò di superare un ostacolo imprevisto. Si può ipotizzare che, sia pure soltanto da una distanza molto ravvicinata, una pioggia di pila pesanti fosse addirittura in grado di arrestare le 25 cariche della cavalleria corazzata. Dando dunque per acquisita (o almeno per consolidata a sufficienza...) la prima parte della mia teoria, quella relativa cioè all’impiego del pilum pesante, vorrei qui insistere ancora un poco su natura e funzioni del secondo tra gli strumenti in esame. Per la nascita della lorica segmentata – attribuita a lungo, secondo la datazione più comunemente accolta, all’età di Claudio – si era proposta di recente una datazione diversa, rialzandola al regno di Tiberio e supponendo che essa costituisse l’evoluzione di un’armatura di tipo gladiatorio messa in campo durante 26 l’insurrezione di Floro e Sacroviro. Per quanto riguarda la sua diffusione, viceversa, ci si era costantemente 27 attenuti al parere, a lungo dominante tra gli studiosi, secondo cui questo tipo di corazza, di uso comune lungo i limites europei, era invece sostanzialmente ignoto proprio sul fronte orientale. A questo categorico asserto si era creduto di trovare un sostegno inoppugnabile nella constatazione che i frammenti di lorica segmentata venivano tutti dalle regioni dell’Europa settentrionale. Chi, come me, cercava di affermare il contrario ribatteva, certo, che l’ormai secolare tradizione nel campo degli studî sul limes e l’enorme sviluppo della ricerca archeologica all’interno di alcuni paesi europei potevano almeno in parte spiegare la concentrazione dei reperti in 28 queste regioni soltanto; ma doveva poi arrendersi di fronte all’apparente evidenza del dato proposto.

Solo uno schieramento vi è nel quale la fanteria romana risulta immediatamente vulnerabile: quello per coorti, che espone all’urto delle masse corazzate schieramenti profondi abbastanza da costituire, se mi si passa il termine, un bersaglio pagante, ma non tanto da risultare indigesto, poiché può essere travolto e spazzato via con il semplice peso della carica. Si tratta, però, proprio della formazione adottata più spesso dalle armate imperiali; sicché quanto mai opportuna risulta l’introduzione del pilum pesante. Dotato di una forza di penetrazione enormemente accresciuta e resa estrema dalla somma delle velocità contrapposte, quella dell’arma stessa e

Ora, tuttavia, entrambi questi asserti si sono definitivamente rivelati privi di fondamento. Mentre 29 infatti i rinvenimenti da Kalkriese costringono a far risalire senz’altro l’adozione della lorica segmentata da parte delle truppe romane all’età di Augusto, frammenti di corazze di questo tipo sono stati trovati di recente 30 anche fuori dall’Europa, in Marocco, ma soprattutto in

esaltava il vantaggio del Turco, derivante dall’ «unico e proprio suo modo di battagliare, avendo egli cavalli più veloci e più agili de’ nostri e meno aggravati degli arnesi di briglie, selle e armature». Sulla superiorità della cavalleria leggera, di quella mongola in particolare cfr. Moryson 1903: 35, 47; Oman 1924: 346; Oman 1937: 758-759; Sinor 1956: 45-46. Come i tardi epigoni europei, anche l’alta aristocrazia arsacide sacrificò forse «tattica e strategia al sogno impossibile di colpire il nemico rimanendo invulnerabile» (la frase è di Lot 1946: 429) . 22 Il passo (Tac., Hist. I.79) ricorda come una schiera di Rossolani forte di ben 9 mila uomini, penetrata in Mesia, venga sorpresa dai fanti di Roma; i quali, agili pur nelle loro armature, attaccano con gladio e pila questi cavalieri corazzati, in tutto simili a quelli partici, e li distruggono agevolmente. 23 È il suggerimento offerto da diverse fonti, come Erodiano (IV.14.7) o Maurizio (Strat. XI.2, p.259). 24 Una tattica del genere fu adottata nel 272 da Aureliano a Immae contro i cavalieri palmireni di Zabdas (Zos., Historia nova I.50:3-4) e da Costantino contro i lancieri corazzati di Massenzio alla battaglia di Torino, nel 312 d.C. (Paneg.Lat. X.22-24): cfr. Gabba 1974: 28-29, note 65, 67 e 68. Va ricordato tuttavia che, in questi scontri mancava, ai catafratti, la copertura, indispensabile, degli arcieri a cavallo.

25 Oltre a quanto citato supra, alla nota , si veda anche: Brizzi 2002: 160-162, 172-174; Brizzi 2003: 142. 26 Così Bishop and Coulston 1993: 85. 27 Fors’anche per l’auctoritas della quale godeva colui che per primo ne aveva proposto una ricostruzione plausibile: cfr. Russell Robinson 1975: 9. Cfr. anche, a mero titolo di esempio, Simkins 1979: 17. 28 Cfr. Brizzi 2002: 173. 29 Cfr., tra gli altri, Franzius 1995: 69-88; Schlüter 1999: 125-159. 30 Cfr. Thomas 2003: 1.

4

GIOVANNI BRIZZI: PROLEGOMENI AD UN CONGRESSO

simile ad esempio alla misericordia o allo sfondagiaco, i cosiddetti ‘pugnali assassini’ del Rinascimento italiano; solo armi simili, infatti, capaci di agire su un anello alla volta, riescono a forzare la coesione di una cotta di maglia. Si tratta, tuttavia, di uno strumento e addirittura di un concetto che paiono del tutto ignoti alle genti del centro Europa, equipaggiate spesso con spade lunghe e pesanti, ma dalla punta smussata. I Romani non ignoravano certo questa abitudine dei barbari di colpire caesim piuttosto che punctim; eppure – fors’anche perché contavano sia sulla loro collaudata eccellenza nella lotta corpo a corpo, sia sul sostegno reciproco abitualmente assicurato loro dai compagni di linea all’interno degli schieramenti – non si erano mai curati, finora, di modificare un armamento difensivo che risaliva almeno all’età delle guerre puniche; e ciò malgrado avessero incontrato da secoli avversarî i quali, come i Celti o i Germani, erano provvisti di asce e spade adatte a sferrare soprattutto fendenti di taglio. Ove ci si attenga alla spiegazione proposta, ci si scontra dunque con una singolare aporia, cronologica forse più ancora che logica. E non solo: non si può ignorare neppure il fatto che per attutire i colpi delle armi da botta, i soli – ripeto – veramente pericolosi, sarebbe comunque bastato semplicemente aggiungere al giaco di maglia una coppia di spallacci rigidi. La soluzione prospettata, quindi, è comunque solo parziale, poiché non chiarisce quale funzione dovessero avere le lame metalliche poste a protezione del busto.

Israele, in contesti archeologici relativi alla prima grande 31 rivolta giudaica. È naturalmente questo secondo il punto che più ci interessa: caduto ormai, in presenza dei nuovi reperti, il principale rilievo mosso alla mia teoria, si possono infatti riconsiderarne presupposti e argomentazione. A tale proposito sorprende, in particolare, il fatto che il mio punto di vista sembri non esser stato minimamente compreso dai critici; tanto più che, in realtà, esso non si discosta poi molto dal loro, o almeno dall’ipotesi 32 originaria cui affermano di aderire. Neppure a me, infatti, sfugge come il principale problema posto dalla lorica hamata – quella cotta di maglia cui la corazza di nuovo tipo si affianca senza sostituirla – nasca in realtà dalla sua natura flessibile. A causa di questo carattere, si 33 è detto, qualsiasi urto spingerà i suoi anelli nelle carni del soldato con una forza proporzionale alla violenza dell’urto stesso; sicché la pelle, i tessuti e le ossa sottostanti ne riceveranno, comunque, un danno di qualche entità. Al contrario, grazie alla sua struttura rigida, la lorica segmentata proteggerà il fante romano 34 anche da questo effetto. Ma tale spiegazione, per quanto intelligente e persuasiva, mi sembra tuttavia solo parziale, poiché ignora alcuni particolari estremamente significativi. Ai suoi sostenitori sembra in primo luogo essere sfuggito il fatto che, almeno a distanza ravvicinata, le sole offese veramente pericolose per chi indossi una lorica hamata vengono dai colpi di mazza e dai fendenti inferti con asce o spade pesanti: se infatti, in sé, la pressione degli anelli rappresenta poco più che un fastidio, cui si può ovviare semplicemente frapponendo un’imbottitura spessa e morbida, per esempio un farsetto, tra l’armatura e il corpo, l’integrità fisica del soldato rimarrà invece ben più esposta alla forza del trauma. La minaccia più grave è dunque quella portata dalle armi che, con un termine tecnico moderno, sogliono definirsi da botta: anche se non riusciranno in alcun modo a lacerare o a squarciare la trama degli anelli metallici, capace di ripartire l’energia del colpo su una superficie molto ampia, una forte percossa o un fendente energico potranno infatti, se aiutati dal peso dell’arma, fratturare un osso o provocare una lesione interna invalidante e talvolta mortale.

Per comprendere fino in fondo la funzionalità della nuova armatura è infatti necessario, secondo me, chiarire anche quest’ultimo dettaglio. Certo, la corazza a segmenti, assai più corta della lorica hamata, aveva un peso molto minore di quella; e tale pregio incise indubbiamente non poco nel determinarne la fortuna presso le fanterie imperiali. Ma essa doveva presentare altresì vantaggi di natura diversa; e in particolare doveva offrire una protezione molto maggiore contro un tipo di arma offensiva che non abbiamo ancora considerato. In realtà vi è stato chi ha colto in parte questo problema, e, giungendo a conclusioni analoghe alle mie, ha 35 affermato che la cotta metallica «n’empêche ni les traumatismes, ni même les blessures ouvertes dans le cas d’un trait acéré ou très violent»; ma nessuno, che io sappia, si è sbilanciato a definire l’arma capace di ottenere un simile risultato.

Al contrario, è solo perforando le maglie che una stoccata diritta può ferire od uccidere chi indossi un corpetto metallico; altrimenti ogni affondo, anche il più violento avrà, per la forza limitata del braccio umano, effetti sostanzialmente modesti. Per riuscire fatale un colpo di punta deve dunque essere inferto con un’arma acuminata,

Natura e caratteri di essa si possono invece, a parer mio, desumere facilmente proprio dalle considerazioni che abbiamo proposto sopra. Mosse dall’energia cinetica impressa loro dagli archi da guerra, le frecce migliori, terminanti con l’aguzza cuspide poligonata che gli 36 Anglosassoni chiamano «bodkin head», portano un

31

Come Gamla – ant. Gamala –, Masada, Zippori – ant. Sepphoris –: cfr. Syon 2002: 134-153 (la notizia del rinvenimento è a p. 145. Brevi cenni in Arma 6, 1994: 16; e 7, 1995: 8); Stiebel 2003: 215-244. Ulteriori riferimenti, anche a testi tuttora inediti, in Cowan 2004: 1-3. 32 Quello di Williams 1977: 77; cfr. Bishop and Coulston 1993: 85. 33 Ibid. 34 Che il Williams 1977 ha probabilmente estrapolato per analogia dagli studî degli specialisti sugli armamenti medievali.

35

Così Feugère 1993: 129 (il quale, peraltro, non si cura in alcun modo di citare i miei lavori, tutti anteriori al suo...). 36 Secondo il Taybugha «le più affidabili tra le punte di freccia sono quelle a sezione triangolare o quadrata. Esse sono usate in battaglia e per trapassare ogni tipo di corazza»: cfr. Latham-Paterson 1970: 25-26.

5

THE LATE ROMAN ARMY IN THE EAST 43

colpo di natura in tutto simile a quello della misericordia 37 o dello sfondagiaco medievali, ma assai più potente; e riescono così a forare il farsetto di maglia, squarciandone la trama. Più pericoloso di qualsiasi fendente è dunque, per il Romano, l’impatto di un dardo di questo tipo, quando almeno esso venga scagliato da un arco potente: come il longbow, l’arma lunga inglese, sarà in grado, secoli dopo, di perforare la cotta dei cavalieri francesi, così si deve concludere che le frecce dei Parti non solo spingessero dolorosamente gli anelli della lorica hamata nella carne dei nemici, ma addirittura riuscissero a 38 lacerarli con esito mortale.

sostenuto da più parti, «a bow designed for use on foot is likely to be more powerful than a horse bow», è almeno verosimile che una corazza capace di resistere alla più poderosa tra le armi del primo tipo, il longbow inglese appunto, offrisse a maggior ragione una difesa efficace anche contro l’arco composto partico. Come le armature del XV secolo, la lorica segmentata doveva dunque esser vulnerabile solo a quadrelli corti e massicci, molto più pesanti delle semplici frecce, e scagliati in antico dalla potenza delle macchine da guerra; e tuttavia ulteriori esperimenti di re-enactment dimostrano che persino questi dardi – pur certo capaci, con la semplice forza d’urto, di produrre traumi «disabling or killing the victim» – avrebbero faticato a 44 perforare una corazza a segmenti. E dunque, se è vero senz’altro che una presenza tanto massiccia di questo strumento nell’Occidente romano non può spiegarsi solo con la casualità dei rinvenimenti o con l’ipotesi di un suo uso generalizzato solo durante la normale attività di guarnigione, si potrà d’altronde escludere, d’ora in avanti, anche l’ipotesi contraria; che, cioè, le forze dell’Oriente romano ne abbiano ignorato l’esistenza. Si potrà, a maggior ragione, escludere altresì – come già suggeriva la logica – che l’adozione del pilum pesante e della corazza a lame da parte dei Romani rappresenti solo il frutto di un concorso di circostanze.

Al contrario, «la cuirasse segmentée est une véritable armure: elle permet de supporter sans dommage des 39 coups beaucoup plus violents». Solo la necessità di proteggersi contro simili colpi può dunque, secondo me, spiegare appieno la necessità di avvolgere il busto con gli scomodi segmenta metallici. Le frecce scagliate dall’arco partico – necessariamente lunghe, onde permettere al braccio di tendere completamente l’arma; ma al tempo 40 stesso sottili e leggere per non ridurne troppo la portata – saranno fisicamente inefficaci contro una corazza a piastra di idonea consistenza. Una lorica segmentata che pesa intorno ai 9 chilogrammi ha lame spesse oltre un millimetro (forse addirittura mm 1.25...); e lo spessore è raddoppiato dal loro sovrapporsi su quasi tutto il tronco del soldato. Urtando contro questa superficie resistente e compatta, è verosimile dunque che le frecce vibrino al momento dell’urto, rimbalzando fino a cadere, o che il loro fusto si spezzi.

Con simili conclusioni sembra singolarmente accordarsi, del resto, anche il dato di alcune fonti letterarie. Il problema di trovare contra equitem Parthum...ullam armaturam... era stato avvertito già all’indomani di 45 Carrhae; e, qualunque sia stato il progetto originario che portò alla creazione della lorica segmentata, fu proprio rispetto alla minaccia partica che l’armatura di nuova concezione si prestò a dare nei fatti una risposta affidabile, o al momento stesso della sua messa in linea, già sul finire del regno di Augusto; o, comunque, poco tempo dopo, quando se ne dovettero fatalmente intuire potenzialità ed efficacia. Ad ogni modo tali doti sembrano esser state senz’altro note ed apprezzate da tempo agli inizî del II secolo: sottolineando come i soldati di Traiano, avvezzi alle terribili ferite causate dalle falci dei Daci, spregino ormai le punture inflitte dalle frecce 46 partiche, Frontone non fa infatti, a mio avviso, solo della facile propaganda, ma sottolinea un fatto reale: gli

Formulata più volte da chi scrive e precisata nel corso 41 degli anni, anche questa ipotesi è venuta, io credo, confermandosi col tempo. Forse non è giusto concedere una fiducia incondizionata alla sperimentazione moderna dei cosiddetti ‘re-enactors’, ma non è permesso nemmeno dubitare della serietà degli esperimenti di alcuni tra loro; ed è quindi assai significativo il fatto che prove di tiro condotte con repliche del longbow inglese contro una copia di lorica segmentata abbiano visto le frecce rimbalzare spesso o il loro fusto infrangersi dopo 42 una penetrazione assolutamente superficiale. Il confronto può essere solo indiretto; ma se, come si è

37

Sulle armi medievali capaci di perforare una corazza cfr. EdgePaddock 1988; 35; 91. 38 Assolutamente categorica è, in proposito, la testimonianza di Plutarco a proposito della battaglia di Carrhae: Plut., Crass., 24-25. Benché l’autore greco accenni a punte ritorte o provviste di alette, sembra più probabile che le punte da guerra impiegate contro bersagli corazzati fossero, appunto, quelle a sezione triangolare o quadrata. 39 Così Feugère 1993: 129. 40 Come ricorda Plinio (nat.hist.XVI, 159-161), nel mondo romano le frecce erano ricavate da un particolare tipo di canna (l’harundo donax); e, per quanto fosse forse più robusto, il fusto dei dardi partici non doveva differire di molto da questo. 41 Oltre a quanto citato supra cfr. Brizzi 2003: 138 ss. 42 3 Cfr. Hardy 1992 : 236; cfr. Massey July 1994: 36-39.

43

Cfr., p. es., Coulston 1985: 291. Alla distanza di 50 metri il dardo di una moderna carroballista è riuscito a forare la piastra superiore della lorica, «piercing 1.25 mm thick steel hoops», ma ha solamente intaccato, sia pur quasi traversandola, quella inferiore: cfr. Wilkins - Morgan 2000: 93. La morte o l’invalidità sarebbero derivate non da una ferita, ma dalla profonda imbutitura prodotta nella lorica in conseguenza del colpo. 45 La frase è di Cicerone, in una lettera a Papirio Paeto scritta pochi anni dopo Carrhae (seconda metà del mese di febbraio 50 a.C.): ad Fam. IX.25.1. 46 Fronto, Princ.Hist. 9:...in bello profectus est cum cognitis militibus hostem Parthum contemnentibus, sagittarum ictus post ingentia Dacorum falcibus inlata volnera despicatui habentibus. 44

6

GIOVANNI BRIZZI: PROLEGOMENI AD UN CONGRESSO prima del III secolo, poco più che un esperimento,52 agli equites legionis si affiancano ora alcuni nuovi corpi di cavalleria pesante: per esempio i contarii, che impiegano il contus, la lunga lancia tenuta con entrambe mani, e sono perciò privi sia di scudo, sia di corazza.53 Come ricorda la Tattica di Arriano,54 proprio Adriano stabilisce infine che alcuni reparti indigeni a cavallo, siano essi Parti od Armeni, Celti o Sarmati, conservino armi ed equipaggiamento caratteristici della nazione d’origine e ne mantenegano i metodi di combattimento: tra le unità leggere entrano così definitivamente in linea in questo stesso momento, reclutati in aree come la Cirenaica e la Numidia, Creta ed il Levante, sia i cosiddetti numeri, le formazioni indigene tra cui eccellono i Mauri, i formidabili squadroni berberi che si sono distinti già sotto Traiano,55 sia gli equites sagittarii, arcieri a cavallo che imitano quelli arsacidi.56

arcieri a cavallo e i loro dardi fanno ora assai meno paura che non in passato. E tuttavia, se pure ne rappresenta forse il momento centrale, l’adozione del pilum pesante e della lorica segmentata non esaurisce affatto il processo di mutamento interno all’esercito di Roma; un processo il quale, impostato come si è visto forse immediatamente dopo la disfatta di Carrhae, conosce una vasta serie di modifiche successive, che giungono senza interruzione almeno all’età di Adriano. Fino dal tempo di Antonio le armate imperiali cominciano ad incorporare, in funzione antipartica, forti nuclei di frombolieri balearici e cretesi, i cui proiettili sono, da un lato, mortali anche per i 47 catafratti, hanno, dall’altro, una gittata almeno pari a 48 49 quella dell’arco partico; e questo schermo, incaricato di proteggere sul campo le grandi formazioni di fanteria pesante, i Romani continuano poi costantemente a potenziarlo e addensarlo, aggiungendovi nuclei di lanciatori di giavellotto e di arcieri cretesi, siriaci o genericamente levantini, dotati a loro volta, questi ultimi, di armi forse più potenti rispetto all’arco stesso impiegato 50 dagli hippotoxotai. Entrano in linea inoltre, forse a partire dall’età di Traiano, nuove e potenti macchine mobili da getto, le cosiddette carroballistae: le quali, vera artiglieria da campagna ante litteram, sono le prime a poter essere schierate e impiegate direttamente sul campo di battaglia.51 Sempre durante l’età antonina, infine, si compie l’ultimo passo: l’esercito romano conosce un forte processo di potenziamento e specializzazione delle forze montate, accresciute con tipologie nuove e assai rafforzate numericamente. Se l’introduzione di reparti catafratti sembra esser stata,

Di fronte alla crescente efficienza delle forze imperiali le possibilità tattiche dell’armata partica vengono affievolendosi sempre più. Ora, per esempio, la cavalleria pesante, di per sé impacciata e poco manovrabile da sempre, risulta ancora più esposta alle nuove armi da getto romane; e, formata com’è dagli esponenti dell’alta aristocrazia arsacide, pare forse troppo preziosa per essere sacrificata in attacchi che rischierebbero di riuscirle fatali. Una frase di Erodiano57 autorizza a sospettare che essa venga, di fatto, ritirata dai campi di battaglia del Levante; e che, tranne rare eccezioni, ne resti sostanzialmente lontana per circa un secolo. Come si è detto, tuttavia, le possibilità dell’esercito arsacide dipendono dall’azione combinata delle sue componenti a cavallo: è solo la loro sinergia a metterle in condizione di sostenere uno scontro campale vero e proprio e persino, in certi casi, di logorare schieramenti chiusi di fanteria. Quella compiuta dai Parti è dunque una scelta strategica effettiva; e si spiega semplicemente con la loro nuova, gravissima condizione di inferiorità tattica. Davanti alle truppe romane che penetrano in forza nel loro territorio, le armate partiche si limitano dunque di solito, profittando dell’unico vantaggio rimasto, la superiore mobilità delle loro cavallerie, a ripiegare per evitare un contatto prolungato, che significherebbe l’annientamento; e barattano così lo spazio, che non manca, con il tempo necessario ad organizzare una

47 Lo attesta, per esempio, Cassio Dione (XLIX.20.3) quando ricorda l’efficacia di quest’arma contro i cavalieri corazzati di Pacoro da parte delle forze di Ventidio Basso (38 a.C.). 48 Secondo Vegezio (Epit., II.23) arcieri e frombolieri avevano armi a lunga gittata: a ben 600 piedi di distanza giungeva, ad esempio, il tiro del fustibalus, una grande fionda attaccata ad un bastone (cfr. Thouvenot 1996: 911). Ma anche le semplici frombole colpivano assai lontano e con notevole efficacia. La portata di tiro dei frombolieri rodii era, in età greca, doppia rispetto a quella dei loro omologhi persiani e pari a quella degli archi migliori (Xen., Anab. III.3.10); una caratteristica, questa, che sembra essersi conservata anche in età romana. Già durante le sfortunate spedizioni di Antonio, infatti, gli hippotoxotai partici avevano perduto la capacità, decisiva a Carrhae, di colpire indisturbati; e avevano dovuto pagare un pesante tributo al tiro di sbarramento dei frombolieri: Plut., Ant. 41.6-7; 42.1; 45.1. 49 Di «a screen to protect the close-order legionary infantry from enemy missiles» parla precisamente il Coulston 1985: 293. 50 E provviste, evidentemente, di gittata maggiore. Sulla superiore potenza dell’arco in dotazione agli arcieri a piedi v. supra, nota 43. 51 Veget., Epit. II.25. Ogni legione disponeva di cinquanta (o cinquantacinque) macchine di questo tipo. Di dimensioni ridotte, ma molto potente, questo particolare lanciadardi poteva essere mantenuto costantemente in assetto da battaglia e seguiva le legioni a bordo di carri trainati da cavalli o da muli (v. Colonna Traiana, Scena XL): il formato più piccolo era stato reso possibile dallo stesso carattere rivoluzionario dell’arma, che per la prima volta aveva le componenti principali — apparato propulsivo e braccia — interamente in metallo; cosa che aveva aumentato di molto sua potenza. Da ultimo Wilkins 1995: 5-59. Cfr. anche Marsden 1969: 164-167, 188-190, pl. 9-13; Le Bohec 1989: 147. Sulla filosofia d’impiego delle macchine da guerra, e in particolare delle carroballistae, Brizzi 1983a: 64 ss. (con precedente bibliografia).

52

La prima ala di catafratti ad essere documentata (CIL XI 5632) risale proprio all’età adrianea: cfr. Gabba 1974: 32; Dixon-Southern 1992: 76. 53 Creati anch’essi nell’età di Traiano o Adriano, forse in previsione di conflitti con i Sarmati: cfr. Gabba 1974: 33; Coulston 1986: 69; DixonSouthern 1992: 75-76. La loro efficacia doveva esser rilevante, paragonabile forse a quella dei lancieri del XIX secolo. 54 Arrian., Tact., 44. Cfr. Gabba 1974: 39. 55 È noto il ruolo svolto da Lusio Quieto durante le guerre dacica e partica di Traiano, prima alla testa di quelle genti maure di cui egli era un capo (HA, Hadr., 5.8), poi al comando di un’armata intera: Syme 1958: 242. Cfr. Gabba 1974: 40-41. 56 Cfr. Coulston 1985: 288-289; 293-294; Dixon-Southern 1992: 77. 57 Herodian. III.5.7-9. La sola eccezione, d’altronde plausibile, sembra essere costituita dalla guerra combattuta sotto Marco Aurelio; quando, avendo preso l’iniziativa essi stessi, i Parti vi si impegnarono veramente a fondo,

7

THE LATE ROMAN ARMY IN THE EAST qualche difesa. È semmai solo la cavalleria leggera che ha il compito di ritardare l’avanzata del nemico.

Carrhae era riuscito ai Parti, riesce costantemente, in seguito, a Roma.

Priva dell’appoggio ravvicinato dei cavalieri corazzati, anche questa componente possiede però, quando deve attaccare forti contingenti misti di truppe nemiche, opzioni tattiche assai limitate. Gli hippotoxotai hanno infatti ormai definitivamente perduto la loro preziosa capacità stand off – la facoltà di colpire, cioè, restando oltre la portata della reazione avversaria che li aveva resi invulnerabili sul campo di Carrhae –: per giungere a tiro delle fanterie imperiali essi dovranno superare ora, come si è detto, un forte sbarramento costituito da arcieri, frombolieri e macchine da getto, con risultati deludenti oltretutto, viste le nuove armature di cui i nemici dispongono. Oltre che poco redditizia e molto costosa in termini di vite, un’azione troppo insistita rischierà poi di riuscir loro fatale. Non potendo infatti contare più su un’azione raccordata con i catafratti, le formazioni leggere partiche dovranno infatti affidare la loro salvezza all’estrema mobilità, poiché, una volta sfiancati i cavalli in un contatto prolungato, rischierebbero di essere raggiunte e annientate dalle unità di cavalieri che scortano ormai abitualmente le grandi formazioni nemiche di fanteria, cavalieri che sono spesso molto più 58 abili di loro nel combattimento corpo a corpo.

Di fronte ad un uditorio composto di specialisti tanto insigni non oso, naturalmente, addentrarmi in un tema che conosco assai meno: le strutture e il funzionamento dell’esercito tardoantico. Mi limiterò dunque a trarre qui alcune conclusioni d’ordine generale, fondate inoltre soprattutto sull’analogia. Si è affermato spesso che la scelta del tardo impero di affidarsi sempre più a corpi di cavalieri obbedisse a motivazioni di ordine tattico e strategico insieme: cercasse cioè da un lato l’efficienza sul campo, puntasse dall’altro ad una mobilità superiore per rispondere alle nuove esigenze della difesa. Ebbene, mi sia permesso, almeno in parte, di dissentire: quanto all’efficienza tout court, la superiorità tattica delle legioni sulle cavallerie, pesanti e leggere, come ‘nocciolo duro’ di un’armata da campagna è, io credo, emersa in modo inconfutabile da quanto siamo venuti dicendo finora. Questa superiorità degli eserciti altoimperiali concerne poi per buona parte anche il settore della tecnologia. Si ripete spesso che le fanterie tardoantiche erano a loro volta corazzate, e sono senz’altro pronto a crederlo; ma lo era una parte soltanto dei loro effettivi, e andavano diffondendosi sempre più le armature di tipo leggero, di solito a lamelle o a scaglie, la cui efficacia protettiva non era neppur lontanamente paragonabile a quella della 60 lorica segmentata. Della sopravvivenza di questa corazza in seno agli eserciti tardoantichi ancora si discute. Se l’ultima rappresentazione per così dire ufficiale di essa 61 è quella che compare sull’arco di Settimio Severo, esemplari di lorica a segmenti si sono riconosciuti sulle stele successive di alcuni soldati: quella, ad esempio, di M. Aurelio Alessiano, conservata al Museo Nazionale Archeologico di Atene, o quella di un legionario da Alba 62 Iulia, in Dacia. Così si tende ora a pensare che l’impiego di quest’arma sia continuato fino alla metà almeno del III secolo; ma si sia fatto via via più ridotto, limitandosi infine esclusivamente alle truppe di prima 63 linea. Di qualità non meno inferiore erano gli elmi. Anche per essi il semplificarsi del processo di fabbricazione — evidentissimo, per esempio, nella tipologia cosiddetta ‘ad arco’, come pure in tutte le altre, Spangenhelme o Gardenhelme che siano –, che dovrebbe consentire tempi minori di lavorazione, si associa al 64 rarefarsi anziché al moltiplicarsi dei reperti; e ciò

Per tornare dunque al punto donde siamo partiti, agli Arsacidi non rimangono molte risorse; e tra queste non figura, se non come extrema ratio, lo scontro campale. Anche negli spazî del teatro armeno e di quello mesopotamico si assiste così al rarefarsi dei grandi fatti d’arme: contro piccole unità isolate le forze partiche potranno sempre ricorrere alla guerriglia o all’agguato, mentre le formazioni più massicce saranno almeno parzialmente vulnerabili con rapide incursioni, necessariamente seguite però da una pronta ritirata e chiuse semmai dal colpo mortale sferrato durante la fuga, dalla ‘freccia del Parto’ divenuta proverbiale nelle fonti romane.59 In certo qual modo obbligata, una simile strategia lascia però le armate nemiche libere di condurre a termine quelle operazioni poliorcetiche in cui esse eccellono; e, per contro, la progressiva occupazione delle città finisce fatalmente, a sua volta, per restringere ancor più e rendere precarî glî spazî di manovra (e di sopravvivenza...) necessarî alle cavallerie partiche. Tra due strutture militari – che, giova dirlo, restano tuttora sostanzialmente incompatibili (o, come si direbbe oggi, asimmetriche...) – vince quella che giunge ad imporre all’altra i proprî metodi di combattimento: ciò che a

60

Come è stato ricordato da Feugère: 129. Cfr. Brizzi 2002: 190. Cfr. Coulston 1995: 13-17. 63 Cfr., tra gli altri, Cowan 2000: 27, 41, 63. 64 Per questo fenomeno è stata proposta una spiegazione che, francamente, mi pare quanto mai cervellotica (e che ha lasciato perplessi anche altri: si veda, per esempio, Perassi 2002: 31 nota 29): quella del Feugère 1993: 143-144, secondo cui sulla diminuita quantità di questo particolare tipo di reperti in età tardoantica avrebbe in qualche modo inciso l’abbandono di pratiche rituali come quella di consacrare gli elmi gettandoli nei fiumi. Al di là del fatto che tale consuetudine pare di per sé poco attestata (e comunque non tale da incidere in modo significativo sulla frequenza dei ritrovamenti), altre (v. infra) sembrano essere, in realtà, le cause di questo fenomeno. 61 62

58

Cfr. Brizzi 1981: 196; Brizzi, in corso di stampa. Sulla predisposizione alla guerra per imboscate: Chapot 1907, 45. Tale caratteristica è forse all’origine «de l’accusation de perfidie» rivolta dai Romani al nemico orientale (ibid., p. 46). Sulla ‘freccia del Parto’: Plut., Crass. 24; Verg., Georg. III.31; Herodian. III.7. Pur se “there is nothing inherently ‘Parthian’ about the ‘Parthian Shot’, poiché l’abilità di saettare rivolti all’indietro è tipica di tutti gli arcieri montati, dagli Sciti in poi (Coulston 1985: 292, con bibliografia), è legittimo il dubbio che il colpo scagliato durante la fuga abbia costituito a lungo la principale, se non unica, risorsa delle cavallerie leggiere arsacidi, tanto da finire per diventarne il tratto distintivo: Brizzi 1981: 196. 59

8

GIOVANNI BRIZZI: PROLEGOMENI AD UN CONGRESSO

meno di lui vede nella legione una specie di ‘magico 68 toccasana’ per l’Impero, e persino a Tacito.

induce da un lato a pensare ad una ‘rivoluzione’ nei processi produttivi dovuta alla specializzazione sempre 65 minore e al forte regresso tecnologico delle maestranze, spinge dall’altro a prestar fede alla testimonianza di 66 Vegezio quando attesta la progressiva diminuzione, in seno all’esercito, di coloro che continuavano a portare un armamento difensivo di qualsiasi tipo. Persino in un settore che, viceversa, conosce durante l’età tardoantica qualche sviluppo, vale a dir quello delle macchine da guerra, la filosofia d’impiego evidenzia un significativo ripiegarsi su se stesse delle dottrine tattiche e strategiche 67 di Roma.

Per i problemi militari posti dalla crisi gli antichi avevano vagheggiato, a dire il vero, anche una soluzione diversa, e ciò prima ancora che al loro studio si dedicasse lo stesso Vegezio: quella proposta dall’Anonymus de rebus bellicis. Ma, di fronte al dramma che andava profilandosi, sarebbe servita veramente a qualcosa l’idea portante dell’autore, rispetto alla quale le macchine da guerra illustrate nell’opera erano soltanto il mezzo; sarebbe servito a qualcosa, cioè, il progetto di ricostituire la potenza dell’impero fondandola sulle risorse di una tecnologia ancora assolutamente primitiva? Certo, Vegezio è stato considerato un inguaribile laudator temporis acti; e vi è chi lo ha accusato «di sciorinare soluzioni impossibili» e di sciupare «nomi venerandi che ormai sono l’ombra di se stessi...» («il nome venerando è, 69 naturalmente, la legione»). Certo, a Vegezio altri hanno opposto proprio l’Anonymus de rebus bellicis. Ma, a sua 70 volta, lo spirito «merveilleusement inventif» di costui non ha prodotto, in fondo, se non strumenti di dubbia efficacia; sicché altri studiosi ancora lo hanno bollato 71 come «ein verrückter Projektenmaker».

Per la sua struttura composita, l’esercito dell’Alto Impero mi ricorda la Grande Armée di Napoleone: un insieme estremamente equilibrato, e dunque assai funzionale, di popoli e forze diverse e complementari – ausiliarî, cavalleria, artiglieria, genio etc. –, abitualmente impiegate al meglio delle loro possibilità in operazioni combinate di teatro; ma al cui interno erano comunque le fanterie ad essere numericamente preponderanti. Ove si consideri in qualche modo legittimo il parallelo, questo risolve ogni nostro dubbio anche sulla mobilità delle legioni stesse: quel requisito che nessuno ha mai negato ai fanti di Napoleone, vale a dire la capacità di spostarsi fulmineamente per raggiungere i teatri di guerra più lontani, non si potrà, io credo, negarlo neppure alle truppe di Roma. Altri sono, semmai, i limiti dell’esercito altoimperiale; e di natura principalmente strategica, legati ai risvolti politici di un comando pericolosissimo a delegarsi, al numero degli effettivi, scarso rispetto alla mole dei compiti, alla progressiva disaffezione per la milizia che riduce gravemente le possibilità di reclutamento. Ma di ciò non vi è tempo di parlare qui. Certo volta a mascherare in qualche modo un’efficienza ormai in parte ridotta, la preponderanza, funzionale e di rango anche se non numerica, delle cavallerie all’interno della nuova struttura tiene dietro, secondo me, alla trasformazione ormai inarrestabile di un impero al cui interno viene affermandosi la tendenza assolutista e teocratica dei sovrani e sembrano prevalere, di conseguenza, i modelli politici e culturali dell’Oriente; un Oriente il quale, non a caso, proclama da sempre la superiorità della cavalleria sulla fanteria.

Tutti questi sono certamente giudizî in qualche modo fondati; ma a mio avviso sono anche un poco ingenerosi verso entrambi i nostri autori. In realtà, di fronte alla crisi – ahimé, irreversibile – dell’Impero, le diverse voci dibattono la questione, ancor oggi di attualità, se sia meglio affidarsi ad un esercito di leva – forse tuttora la soluzione migliore, purché si possa contare su un senso civico paragonabile a quello delle poleis classiche – o se convenga piuttosto ricorrere ad una forza di professione, appoggiata se non integralmente sostituita dalle soluzioni – miracolistiche in questo caso, assai più che miracolose – della tecnologia. Tutti indifferentemente però, sia l’anonimo de rebus bellicis, che invoca «la frusta di 72 Catone» (come qualcuno ha definito la riforma sociale da lui auspicata per creare una società meno ingiusta e risvegliare così lo spirito civico dei Romani); sia il nostalgico nazionalista Vegezio; sia persino il ben più affidabile Ammiano, che vede nell’esercito l’espressione del populus, avvertono, benché forse in modo non sempre cosciente, il senso della crisi profondissima in corso. Come l’esercito della Rivoluzione francese, figlio della leva in massa, anche quello romano altoimperiale, pur divenuto una forza professionista, continua almeno idealmente a vivere del rapporto genetico esistente da sempre, in Roma, tra soldato e cittadino; e quando quest’ultimo scompare, trasformato in suddito da un potere imperiale sempre più assoluto, è la natura stessa della struttura militare che, inesorabilmente, cambia. L’affievolirsi dell’orgoglio civico e la scomparsa del

Non è un caso, dunque, neppure che l’efficienza delle legioni altoimperiali venga, in seguito, costantemente rimpianta; un fatto che non è dovuto solo ad una malintesa nostalgia per il buon tempo andato. Nel quadro della letteratura latina il punto di vista di Vegezio, nazionalista e partigiano di una antiqua consuetudo che egli vorrebbe ad ogni costo ristabilita, si rifà a precursori illustri, ad Ammiano Marcellino, per esempio, che non

65

68

È la soluzione proposta da James 1986: 131-134; James 1988: 271273. 66 Veget., Epit. I.20. 67 Sia permesso, qui, rinviare semplicemente alle conclusioni proposte in un precedente lavoro di chi scrive: Brizzi 1983a: 67 ss.

Così Gabba 1968: 89 s. Così Mazzarino 1959: 52. 70 Così Piganiol 1947: 200. 71 Così Seek 1894: c. 2325. 72 Giuffré 1974: 115. 69

9

THE LATE ROMAN ARMY IN THE EAST senso serviano del munus, il servizio da rendersi allo Stato, minano le fondamenta stesse della società classica; sicché l’ideale del cittadino oplita – il medesimo, in fondo, che anima da sempre il legionario – non riesce a sopravvivere alla crisi. Come è stato esemplarmente 73 detto, nella sempre minor importanza della fanteria, «base tradizionale fino al III secolo dell’esercito romano, si riflette malinconicamente il decadere delle strutture cittadine dell’impero, vale a dire dell’elemento caratteristico della civiltà classica, e, per contro, l’emergenza di un differente tipo di cultura e di vita».

(BAR International Series 297), Oxford: 60-68; 7175. Coulston J.C.N. 1995 The sculpture of an armoured figure at Alba Iulia, Romania. In Arma 7: 13-17. Cowan R. 2000 Imperial Roman legionary AD 161-284, London. Cowan R.H. 2004 Brin Mawr Classical Review 49/2: 1-3. Debevoise N.C. 1938 A political history of Parthia. Chicago. Dierauer J. 1868 Beiträge zu einer kritischen Geschichte Trajans. In Büdinger M. (ed.), Untersuchungen zur römischen Kaisergeschichte, I, Leipzig. Dixon P. - Southern K.R. 1992 The Roman Cavalry, London. Eadie J.W. 1967 The development of mailed cavalry. In JRS 57: 161-173. Edge D. - Paddock M. 1988 Arms and armour of the medieval knight, London. Feugère M. 1993 Les armes des Romains, Paris. Firpo G. 1999 Le rivolte giudaiche, Roma-Bari. Firpo G. 2001 Il terremoto di Antiochia del 115 d.C.: echi di una interpretazione apocalittica di Cassio Dione 68,24,2? In Lewin A. (ed.), Gli Ebrei nell’impero romano. Saggi vari, Firenze: 233-245. Franzius G. 1995 Die römischen Funde aus Kalkriese, 1987-95. In JRMES 6: 69-88. Fuks A. 1953 The Jewish revolt in Egypt. In Aegyptus 33: 131-158. Gabba E. 1968 Considerazioni sugli ordinamenti militari del tardo Impero. In Ordinamenti militari in occidente nell’Alto Medioevo. Settimane di Studî del Centro italiano sull’Alto Medioevo, XV, 1, Spoleto: 89 ss. Gabba E. 1974 Sulle influenze reciproche degli ordinamenti militari dei Parti e dei Romani. In Gabba E. (ed.), Per la storia dell’esercito romano in età imperiale, Bologna: 7-42. Gamber O. 1968 Kataphrakten, Clibanarier, Normannenreiter. In Jahrbuch der Kunsthistorischen Sammlungen 64: 14 ss. Giuffré V. 1974 La letteratura de re militari. Appunti per una storia degli ordinamenti militari, Napoli. 3 Hardy R. 1992 Longbow, Yeovil. James S. 1986 Evidence from Dura Europos for the origins of the late Roman helmets. In Syria 63: 131134. James S. 1988 The fabricae: state arms factories of the later Roman empire. In Military equipment and the identity of the Roman soldier. Proceedings of the fourth military equipment conference, (BAR International Series 394), Oxford: 271-273. Harl O. 1996 Die Kataphraktarier im römischen Heer: Panegyrik und Realität. In JRGZ 43/2: 601-627. Henderson 1949 Review of Lepper’s Trajan Parthian war, Oxford 1948. In JRS 39: 127 s. Hyland A. 1990 Equus. The horse in the Roman World, London. Latham J.D. – Paterson W.F. 1970 Saracen archery, London.

Bibliography Alföldi A. 1952 The moral barrier on Rhine and Danube. In Birley E. (ed.), The Congress of Roman Frontier Studies 1949, Durham: 1-16. Bennett J. 1997 Trajan, Optimus Princeps: a life and times, London. Bishop M.C. – Coulston J.C.N. 1993 Roman military equipment from the Punic wars to the fall of Rome, London. Bivar A.D.H. 1972 Cavalry tactics and equipment on the Euphrates. In Dumbarton Oaks Papers 36: 273291. Brizzi G. 1981 L’armamento legionario dall’età giulioclaudia e le guerre partiche. In CS 18/2: 177-201. Brizzi G. 1983a Il trattato de rebus bellicis e l’impiego delle artiglierie in età tardoantica. In Studi militari romani, Bologna: 49-76. Brizzi G. 1983b Note sulla battaglia di Carre. In Studi militari romani, Bologna: 9-30. Brizzi G. 2002 Il guerriero, l’oplita, il legionario. Gli eserciti nel mondo classico, Bologna. Brizzi G. 2003 L’età di Adriano: armamento e tattiche. In Le Bohec Y. (ed.), Le discours d’Hadrien à l’armée d’Afrique. Exercitatio, Paris: 133-147. Brizzi G. La guerra partica, in corso di stampa. Bruun C. 1992 The spurious ‘expeditio Iudeae’ under Trajan. In ZPE 93: 99-106. Calderini A. 1949 I Severi e la crisi dell’impero nel III secolo, Bologna. Chapot V. 1907 La frontière de l’Euphrate de Pompée à la conquête arabe, Paris. Chazanov A.M. 1968 Katafraktarii i ikh rol’istorii voennogo iskusstva. In VDI 1: 185-191. Cipolla C.M. 19992 Vele e cannoni, trad.it., Bologna. Cizek E. 1994 A’ propos de la guerre parthique de Trajan. In Latomus 53/2: 376-385. Coulston J.C.N. 1985 Roman archery equipment. In Bishop M.C. (ed.), The production and distribution of Roman military equipment. Proceedings of the Second Roman Military Equipment Research Seminar, (BAR International Series 275), Oxford: 220-377. Coulston J.C.N. 1986 Roman, Parthian and Sassanid tactical developments. In Freeman P. and Kennedy D. (ed.), The defence of the Roman and Byzantine East, 73

Gabba 1974: 34.

10

GIOVANNI BRIZZI: PROLEGOMENI AD UN CONGRESSO

Speidel M. 1984 Catafractarii, clibanarii and the rise of the later Roman mailed cavalry. In Epigraphica Anatolica 4: 151-156. Stiebel G.D. 2003 The militaria from Herodium. In Bottini G.C. - Di Segni L. and Chrupcala L.D. (ed.), One land-many cultures. Archaeological studies in honour of S. Loffreda, (Studium Biblicum Franciscanum Collectio Maior 41), Jerusalem: 215244. Syon D. 2002 Gamla, city of refuge: The first Jewish revolt. Archaeology, history and ideology, LondonNew York. Thomas M.D. 2003 Lorica segmentata. Vol. II: A catalogue of finds. (JRMES Monograph 2), Duns: Armatura. Thouvenot R. 1966 Le soldat romain au combat. In Mél. J.Carcopino, Paris: 905-916. Walbank F.W. 1979 A historical commentary on Polybius, III, Oxford. Wilkins A. 1995 Reconstructing the cheiroballistra. In van Driel-Murray C. (ed.), Roman Military Equipment: Experiment and Reality = JRMES 6: 5-59. Wilkins A. - Morgan L. 2000 Scorpio and Cheiroballistra, In Croom A.T. and Griffith W.B. (ed.), Proceedings of the Twelfth international military equipment conference, South Shield 1999, JRMES 11: 77-101 Williams A.R. 1977 Roman arms and armour: a technical note. In JAS 4: 77.

Le Bohec Y. 1989 L’armée romaine, Paris. Lepper F. 1948 Trajan’s Parthian war, Oxford 1948 = repr. Westport, Conn. 1979. Lot F. 1946 L’art militaire et les armées au Moyen Age, Paris. Luttwak E.N. 1976 La grande strategia dell’impero romano, trad.it., Milano 1981 (Baltimore-London). Marsden E.W. 1969 Greek and Roman artillery. Historical development, Oxford. Massey D. 1994 Experiments in archery. In Military Illustrated July: 36-39. Mazzarino S. 1959 La fine del mondo antico, Milano. Mielczarek M. 1995 Cataphracti and Clibanarii. Studies on the Heavy Armoured cavalry of the ancient world, Lodz. Mielczarek M. 1999 Cataphracts: a Parthian element in the Seleucid art of war. In Ancient Iran: 101-105. Moryson F. 1903 Itinerary. Hughes Ch. (ed.), London. Oman A. 1924 A history of the art of war in the Middle ages, Boston-New York. Oman A. 1937 A history of the art of war in the Sixteenth century, New York. Perassi C. 2002 Gli elmi tardoantichi (IV-VI sec.) alla luce delle fonti letterarie, numismatiche e archeologiche: alcune considerazioni. In Lusuardi Siena S. - Perassi C. and Facchinetti G. (ed.), Miles Romanus. Dal Po al Danubio nel Tardoantico. Atti del Convegno internazionale di studî. PordenoneConcordia Sagittaria, 17-19 marzo 2000, a cura di M. Buora, Pordenone. Petersen L. 1968 Lusius Quietus, ein Reitergeneral Trajans aus Mauretanien. In Altertum 14: 211-217. Piganiol A. 1947 L’Empire chrétien, Paris. Pucci M. 1981 La rivolta ebraica al tempo di Traiano, Pisa. Pugachenkova G.A. 1966 O pantisirnom vooruzhenii parfianskogo i baktriiskogo voinstva. In VDI 2: 27-43. Raschke M.G. 1978 New studies in the Roman commerce with the East. In ANRW II.9.2. Rostovtzeff M. 1935 Dura and the problem of Parthian. In YClS 5: 162-163. Rubin B. 1955 Die Entstehung der Kataphraktenreiterei im Lichte der Chorezmischen Ausgrabungen. In Historia 4: 264-283. Russell Robinson H. 1975 The armour of imperial Rome, Oxford. Schlüter W. 1999 The battle of the Teutoburg forest: archaeological research at Kalkriese near Osnabrück. In Creighton J.D. and Wilson R.J.A. (ed.), with contrib.. by Krausse D. et al.), Roman Germany: studies in cultural interaction, (JRA Suppl. series 32), Portsmouth: 125-159. Seek O. 1894, Anonymi. n.3, PW I, 1 (1894), c.2325. Simkins M. 1979 The Roman army from Hadrian to Constantine, London. Sinor D. 1956 Les relations entre les Mongols et l’Europe jusqu’à la mort d’Arghoun et de Bela IV. In Cahiers d’Histoire Mondiale 3: 45-46.

11

L’armée romaine en mer Rouge et autour de la mer Rouge aux IIème et IIIème siècles apr. J.-C.: à propos de deux inscriptions latines découvertes sur l’archipel Farasan Francois Villeneuve Université Paris 1 (Panthéon–Sorbonne)

J’ai eu l’occasion de publier récemment avec deux collègues, puis seul, une inscription latine découverte sur l’île Farasan al-Kubra en mer Rouge1 (fig. 1). Mais il faut parler maintenant des inscriptions latines de Farasan.2 En effet, quinze jours avant ce colloque tenu à Potenza et Matera, une mission aux îles Farasan, aimablement autorisée par le Vice-Ministère des antiquités et des musées du Royaume d’Arabie Saoudite, m’a permis, en compagnie de Mme Laila Nehmé et de Mr. Thamer alMalki, et guidé par M. Ibrahim Miftah, de visiter les sites antiques de la moitié ouest de Farasan al-Kubra et de découvrir une seconde inscription latine, très fragmentaire. Par la suite, en décembre 2006, une seconde mission m’a permis d’achever l’inventaire des sites (ouest de Farasan al-Kubra, visite des autres îles) – qui peut être considéré à peu près complet à présent (fig. 3) – mais sans découverte d’autres textes latins.

Egypte n’ont rien apporté de substantiellement neuf sur le dispositif de l’armée romaine3. Rome entre évidemment en contact avec la mer Rouge au moment de l’annexion de l’Egypte, qui lui donne le contrôle du littoral depuis Clysma jusqu’à Berenike. Presque immédiatement après, en 26-25 ou 25-24 av. J.C., a lieu l’expédition militaire du préfet d’Egypte Aelius Gallus vers l’Arabia Felix. L’expédition est conduite principalement par voie de terre, depuis le territoire du royaume nabatéen et son port méridional sur la côte arabique, Leuke Kôme. Elle témoigne évidemment, à cette date, de velléités de reconnaissance et d’expansion très loin au sud (Arabie et Ethiopie, dit Strabon, XVI.4.22), pour le contrôle de l’approvisionnement en encens et myrrhe, peut-être pour le contrôle de la route de l’Inde. Son échec, catastrophique pour le corps expéditionnaire, pourrait avoir détourné définitivement Rome d’une telle idée, et en tout cas de prendre le contrôle des routes caravanières arabiques et de l’Arabia Felix. L’armée de Gallus échoue devant Marib et remonte par voie de terre, en soixante jours de marche, vers le nord jusqu’à une localité nommée par Strabon (XVI.4.24) Egra Kôme, située en territoire nabatéen mais au bord de la mer Rouge: ce n’est donc pas la ville de Hegra (située à l’intérieur des terres); le savant saoudien A. Ghabban pense avoir identifié les vestiges de cette Egra Kôme au Ra’s Kurkuma, à 45 km au sud d’al-Wajh (fig. 1), plus précisément sur le site d’al-Qusayr, où se trouvent les restes d’un temple nabatéen et un petit port localement appelé Karakuma.4 De cette Egra Kôme, les forces restantes de Gallus font voile, en onze jours, vers le port égyptien de Myos Hormos.

Cette communication sort un peu des limites du colloque: elle concerne une période antérieure, le IIème siècle principalement; et une zone périphérique, parce que les nouveaux documents présentés ici proviennent du sud de la mer Rouge méridionale, qui ne peut pas être considérée, proprio sensu, comme «Proche-Orient»: Egypte, Syrie, ou Mésopotamie antiques. Ces documents, nouveaux et intéressants, ne sont pas dépourvus, on va le voir, de liens avec l’Egypte et avec la Syrie. Ils sont problématiques en bien des points, de la lecture des pierres à l’interprétation historique et géographique des textes: il paraît donc souhaitable de les faire connaître à des publics savants de spécialistes variés. Il est heureux que le Pr. Ariel Lewin en accepte la publication commentée dans les Actes de ce colloque. 1. Les données classiquement disponibles sur la politique militaire romaine en mer Rouge (fig. 1 & 2)

Malgré cet échec, il n’est pas certain que l’expédition d’Aelius Gallus soit la dernière menée par Rome en direction de l’actuel Yémen. Des témoignages assez obscurs du Périple de la mer Erythrée et de Pline l’Ancien laissent entrevoir une autre opération romaine de reconnaissance en mer Rouge (Golfe arabique: Pline, II.169) et Arabie (id., VI.141.160) et une énigmatique destruction d’Aden (Eudaimôn Arabia: Périple, 26: 8): le tout correspondant peut-être à un unique épisode, une expédition dirigée par Caius César, en 1 apr. J.-C.

Ces données ne sont pas très nombreuses et beaucoup d’entre elles sont problématiques. Mais leur nombre augmente ces dernières années grâce aux fouilles conduites à Berenike sur la côte égyptienne de la mer Rouge et aux surveys du désert oriental de l’Egypte; en revanche les fouilles maintenant achevées à Aqaba / Aila en Jordanie et à Qusayr al-qadim / Myos Hormos en

1 2

3 Hormis, en ce qui concerne Aila, la confirmation de l’installation de la Xème légion Fretensis, au IVème siècle, et sa présence dans cette ville au Vème siècle et peut-être au-delà: v. p. ex. Parker S.T. 2005: 415. 4 Ghabban 1994.

Villeneuve et al. 2004a & b; Villeneuve 2004. Villeneuve 2006; Villeneuve s.p.

13

THE LATE ROMAN ARMY IN THE EAST

Fig. 1: carte de la mer Rouge, toponymie actuelle (F. Villeneuve, S. Vatteoni, 2003) 14

FRANCOIS VILLENEUVE: L’ARMÉE ROMAINE EN MER ROUGE

Fig. 2: carte de la mer Rouge, toponymie antique (F. Villeneuve, S. Vatteoni, 2005) 15

THE LATE ROMAN ARMY IN THE EAST En tout état de cause, le seul témoignage épigraphique romain découvert jusqu’ici au sud de la latitude de Hegra, du côté arabique de la mer Rouge, l’a été au Yémen. C’est l’inscription gréco-latine fragmentaire de Baraqish dans le royaume de Saba, publiée par Costa 1977. Elle comporte le nom, et peut-être seulement le nom, en deux lignes en latin puis en une ligne en grec, d’un Publius Corne[lius]… / eques M…. Il y a discussion parmi les savants sur la nature de l’inscription, épitaphe ou dédicace; sur sa date;5 enfin sur la fonction de ce Cornelius «eques»: chevalier, cavalier (M = Maurus, selon Costa 1977), ou portant le cognomen Eques.6 C’est donc un document bien difficile à utiliser. La découverte toute récente à Sirwah, tout près de Baraqish, d’une inscription bilingue sabéo-nabatéenne – la première connue – datée de 6 av. J.-C., une dédicace monumentale de bâtiments de culte faite par le Nabatéen Taymw fils de Kwshay,7 semble pouvoir confirmer l’idée admise par la majorité des savants depuis longtemps: après l’expédition de Gallus, les caravanes nabatéennes continuent leurs activités et prospèrent, entre Hejaz et Saba, sans signe aucun d’une activité romaine sur le continent.

Au Ier siècle, au total, le dispositif romain: contrôle les quelques ports égyptiens du désert oriental, principalement Myos Hormos et Berenike; se heurte à la piraterie pour la navigation commerciale vers l’océan Indien et tente de protéger les navires en embarquant des archers; ne contrôle absolument rien sur la côte arabique, dangereuse, hostile et mal connue, à l’exception du port, assez «nordique», de Leuke Kôme, port au minimum amical, au mieux contrôlé par une garnison romaine; et n’a nullement pris pied sur le continent arabique, où circulent, loin de la côte, les caravanes nabatéennes et qatabanites. L’annexion en 106 du royaume nabatéen, transformé en province romaine d’Arabie, avec une longue côte (fig. 2) qui va du Sinaï à la côte du Hejaz – probablement à la latitude de Hegra, mais rien n’est assuré quant à la « frontière » sur la côte pour l’instant – peut avoir été suivie d’une politique de contrôle maritime plus organisé. Mais le seul témoignage en ce sens est celui d’Eutrope,8 donc tardif, sur la création d’une classis Maris Rubri par Trajan, ut per eam fines Indiae vastaret. Mais il n’y a par ailleurs pas la moindre donnée pour corroborer l’existence de cette flotte. Où serait son port d’attache? Certainement pas à Aila / Aqaba, où les fouilles récentes9 n’ont rien montré en ce sens; au demeurant, le régime des vents en mer Rouge ne rendait pas intéressante la localisation d’un port important à l’extrémité nord du Golfe d’Aqaba, puisque dans la moitié nord de la mer Rouge les vents soufflent toute l’année du nord vers le sud, rendant délicate toute navigation de retour sudnord.10 En Egypte, les données disponibles sur Clysma, Myos Hormos = Qusayr al-qadim11 et Berenike12 ne permettent pas de penser à de véritables ports militaires. Sur la partie septentrionale de la côte d’Arabie Saoudite, les recherches ne sont qu’embryonnaires, et l’on n’en est encore qu’à essayer de localiser les ports anciens et à les identifier à des sites antiques.

Qu’en est-il sur la côte et en mer? Le Périple de la mer Erythrée ne dit quasiment rien de presque toute la côte arabique jusqu’aux îles de Katakekaumene et Camaran, très au sud (fig. 2) sauf que cette côte est infestée de récifs et de pirates (Périple: 20). Il fait cependant mention – assez au nord probablement – du port de Leuke Kôme et d’un centurion / ekatontarches/os qui est établi là, chargé de protéger un paraleptes, percepteur de la douane des 25% (Périple: 19). Ici se pose classiquement un triple problème non résolu: la date de l’information, c’est-à-dire la date de la rédaction du Périple, probablement vers 50 apr. J.-C.; la localisation de Leuke Kôme, pour laquelle trois zones échelonnées sur 600 km de côte ont été proposées (fig. 2); l’appartenance de ce «centurion», de sa troupe et de ce poste de douane: nabatéens, ou romains? Normalement nabatéens, avant l’annexion romaine de l’Arabie; mais peut-être romains, si l’on songe qu’une tête de pont douanière et policière liée à Berenike ou à Myos Hormos peut fort bien avoir été imposée au roi nabatéen par l’amicus romain.

En revanche, les connaissances ont beaucoup progressé ces dernières années sur l’organisation de la côte égyptienne et la protection de l’arrière-pays et des itinéraires entre Coptos sur le Nil et les ports maritimes. Nous savons maintenant que ce réseau de voies et de fortins s’est beaucoup développé au IIème siècle, y compris dans la seconde moitié de ce siècle.13 Il est clair que cet alourdissement du contrôle militaire, qui allait de pair avec un strict contrôle douanier – bien documenté dans les années 150 par le fameux papyrus «de

Il y a donc là aussi bien des incertitudes. En revanche, une chose est certaine: sensiblement à la même époque, nous trouvons chez Pline (VI.101) la mention importante que la flotte du commerce romain qui partait annuellement des ports égyptiens vers l’Inde devait être accompagnée de cohortes d’archers (cohortibus sagittariorum), à cause de la détestable piraterie des Arabes, basés sur des îles.

8

Breviarium, VIII.3.2. Parker 2005, avec la bibliographie antérieure sur les fouilles de ce savant à Aqaba. 10 Facey 2004. 11 Peacock - Blue 2006. 12 Sidebotham - Wendrich W. 1998; 2001-2002. 13 Cuvigny & al. 2003; Reddé 2002. Plus précisément, la route Coptos– Myos Hormos est densément protégée jusqu’au milieu du IIème siècle, mais les fortins y sont ensuite abandonnés; en revanche, la route Coptos–Berenikè reste densément protégée jusqu’au début du IIIème siècle. 9

5

Bowersock 1983 y voit l’épitaphe d’un militaire de l’expédition de Gallus, mort sur place; Costa 1977 y reconnaît une dédicace du IVème siècle apr. J.-C. d’après des critères paléographiques. 6 Demougin 1980. 7 Information aimablement fournie par N. Nebes, 15.09.2004; par M. C.A. Macdonald sur la présentation de ce document par N. Nebes au Seminar for Arabian Studies, London, juillet 2005. Photographie publiée dans Gunter 2005: 39.

16

FRANCOIS VILLENEUVE: L’ARMÉE ROMAINE EN MER ROUGE

Fig. 3: carte de l’archipel Farasan, sites antiques (F. Villeneuve & L. Nehmé, L. Aubry & J. Louvet, 2007).

17

THE LATE ROMAN ARMY IN THE EAST Mouziris»14 – avait pour fonction principale, et de loin, la protection des arrivées et départs du commerce «de l’Inde» lato sensu.

Desanges 1999: 354-355 (nn. 78-83) et de Robin 1989: 154-155: cela doit se situer entre la fin du IIème siècle et les années 220. À cette époque, donc, les Romains sont absents de tout le bassin méridional de la mer Rouge et de ses côtes occidentale et orientale.

Les inscriptions de Berenike, palmyréniennes notamment, confirment cette surveillance portuaire et de l’arrièrepays, mons Berenicidis, à la fin du IIème et au IIIème siècle, par des corps auxiliaires.Ainsi, à une date comprise entre 180 et 212, une dédicace à Yarhibôl est faite «sous Aemilius Celer, préfet du Mons Berenicidis et de l'ala Herculana», et sous un décurion de cette même aile.15 Une autre dédicace, d’un archer palmyrénien, date de 215.16

2. Les données nouvelles des îles Farasan (Fig. 3-10) Ces îles (fig. 1 et 3) sont l’archipel le plus important de la mer Rouge, à 500 km environ au nord du Détroit de Bâb al-Mandeb, à une cinquantaine de milles de la côte arabique. Cet archipel a un symétrique du côté africain, les îles Dahlak, proches du port antique axoumite d’Adoulis.

En revanche, sur la protection de la navigation en mer, au IIème et au IIIème siècles, nous ne savons plus rien. Certes, nous pouvons imaginer que ce dispositif s’est dilué au IIIème siècle, puisque les indicateurs disponibles montrent tous un gros déclin en volume du commerce entre Rome et l’Inde au IIIème siècle, avant une forte reprise au IVème.

Les Farasan (fig. 3) ont deux îles de grande taille: respectivement Farasan al-Kubra (la-Grande) au sud, longue de plus de 50 km; Farasan al-Sughra (la-Petite), plus communément appelée Sajid, au nord. L’archipel est en outre constitué d’une poignée de petites îles avec quelques traces de peuplement ancien (comme Disan et Qummah), et d’une multitude d’îlots et récifs. La grande île a un bon port naturel: l’étroit goulet situé au milieu de la côte Est, en direction du continent. Le relief est extrêmement plat sur toutes les îles, dont le sol est en général constitué d’une croûte rocheuse ou d’accumulations de sable. Pourtant, la grande île, et à un moindre degré l’île de Sajid, possèdent des ressources naturelles appréciables. On y trouve un bon nombre de puits d’eau douce, dont deux sont clairement antiques, à Qusar (sud-ouest de la grande île) et, 2 km plus au sud, à al-Muharraq; une palmeraie d’importance dans le même secteur et d’autres plus réduites près de Sayr et de Khutub; de bonne poches de terres argileuses propices à une petite agriculture; de vastes zones de buisson où prospère un gibier très abondant (gazelles et oiseaux); d’excellentes possibilités de pêche, notamment dans le golfe à très basses eaux situé entre Qurayat et Gharrein, où viennent s’échouer une fois par an des bancs considérables de poissons-perroquets (harid). Toutes ces ressources permettent de comprendre la présence sur ces îles d’une trentaine de petits sites antiques (fig. 4) et islamiques, localisés principalement dans le tiers sud-est de la grande île, le plus dense centre de peuplement; à l’extrémité nord-ouest de la même île; à Khutub sur l’île de Sajid; et ponctuellement, en bordure de mer, sur les îles de Disan et de Qummah.Ces sites sont des villages en habitat groupé, pourvus d’une nécropole distante de quelques dizaines de mètres (Qusar, Farasan, Gharrein, Qurayat, Byut al-Sawaqilah Sayr, Khutub, Disan); des nébuleuses de petits hameaux proches les uns des autres, avec nécropoles également (Wadi Matar, Wadi Jazan); deux ou trois fortins (Qalaat Luqman; Qalaat Farasan, peut-être seulement d’époque ottomane; Jabal al-Kniseh, restes très détruits d’un site antique sur une petite éminence); enfin un site à graffites sémitiques rupestres, près de Hillat el-Mishraf.

Faut-il aussi penser que ce commerce était déjà affaibli au IIème siècle (d’où, logiquement, l’absence d’une activité militaire substantielle qui soit connue)? Cela me paraîtrait un peu étrange du point de vue économique, et cela ne correspond pas bien avec ce que l’on observe dans le désert oriental de l’Egypte durant tout le IIème siècle. Il paraîtrait curieux aussi qu’il n’y ait pas eu, sous les Antonins, poursuite ou reprise des tentatives de contrôle ou de prospection très méridionales commanditées jadis par Auguste (vers l’Arabia Felix) ou par Néron (vers Méroé). Il reste à mentionner un document encore sur l’histoire militaire de la mer Rouge, au IIème ou au IIIème siècle probablement, mais un document où les Romains sont cette fois totalement absents, contrairement à ce qui est souvent écrit:17 c’est la fameuse inscription d’Adoulis (localisation fig. 2), vue et copiée au VIème siècle sous l’empereur Justin par Cosmas Indicopleustès.18 Dans ce long texte, un souverain anonyme, du port d’Adoulis et peut-être d’Axoum, recense ses victoires, en l’an 27 de son règne. Il indique notamment qu’il a traversé la mer Rouge, porté la guerre chez les Arabitai et les Kinaidokolpitai et soumis leurs rois à un tribut, rétabli la paix dans la circulation terrestre et maritime, et guerroyé depuis Leuke Kôme au nord jusqu’au territoire des Sabéens au sud. Il est manifeste que l’on se trouve alors dans une période où les Romains sont tout à fait absents des secteurs situés au sud de leurs frontières. Les dates les plus diverses pour cette inscription et ces événements ont été avancées, du Ier jusqu’au IVème siècle. Je me range pour ma part aux conclusions très argumentées de 14

P. Vindob. G 40.822: Rathbone 2000, avec bibliographie antérieure. Dijkstra - Verhhogt 1999. 16 Sidebotham - Wendrich 1998: 93. 17 En dernier lieu Bukharin 2006. 18 Topographie chrétienne II 61-63 = OGIS 199. Bernand 2000: 32-45, n°277, avec bibliographie complète. 15

Rien dans tout ce matériel ne laissait supposer autre chose qu’une modeste occupation indigène par des pêcheurs, 18

FRANCOIS VILLENEUVE: L’ARMÉE ROMAINE EN MER ROUGE

Fig. 4: exemple de maison antique du village déserté de Byut al-Sawaqilah (F. Villeneuve).

Fig. 5: la tombe de l’ancien cimetière à l’est de la bourgade de Farasan où l’inscription latin n°1 était remployée — remplacée aujourd’hui par une plaque de ciment gris (L. Nehmé).

19

THE LATE ROMAN ARMY IN THE EAST chasseurs et agriculteurs, de tradition sud-arabique puis islamique, traditions confirmées par une dizaine d’inscriptions sud-arabiques et une inscription arabe médiévale.19

correspondant à de petites hauteurs, conviendraient bien pour un fortin romain. L’état du texte est médiocre et variable (fig. 6 et 7). À l’exception de la ligne 1, tout le tiers droit de l’inscription est extrêmement usé. En outre, les lignes 5 à 7, gravées avec moins de soin que les lignes 1 à 4, sont aussi plus abîmées. La proposition de lecture qui suit (comme la copie fig. 6) est donc sûre pour les lignes 1à 3 et les deux premières lettres de la ligne 4 (titulature impériale dont le caractère stéréotypé permet de pallier les quelques incertitudes possibles de lecture), bonne pour l’ensemble de la ligne 4 et pour les deux tiers gauche des lignes 5 à 7, douteuse pour le reste. Ma proposition actuelle est la suivante21:

La découverte d’une, puis deux, inscriptions latines, par le collectionneur local I. Miftah, portée à la connaissance de quelques visiteurs des îles, puis à la mienne et que j’ai examinées en avril 2005, a donc constitué une surprise tout à fait imprévisible.20 La première inscription latine, complète, est gravée dans une pierre locale, calcaire corallien. Le texte, 7 lignes, tient dans un cartouche à queues d’aronde. La pierre est de petite taille (74 cm sur 42, épaisseur 8 à 14 cm): elle provient donc d’un petit monument. Elle a été découverte remployée dans une tombe du cimetière ancien situé immédiatement à l’est de la bourgade de Farasan (fig. 3 et 5). Les tombes de ce cimetière sont en grande partie faites de blocs anciens réutilisés, de facture identique à la pierre portant l’inscription. Ces tombes ont donc été bâties en démontant le bâtiment auquel appartenait l’inscription. Le cimetière est situé dans une dépression, où n’apparaît aucune trace de mur antique.Trois hypothèses de localisation (fig. 3, cartouche à droite) sont possibles pour le bâtiment antique, qui ne peut être distant de plus que quelques centaines de mètres: a) le centre de la localité actuelle de Farasan, probablement installé sur un site antique; b) la petite éminence où se trouve le fortin ottoman tardif de Qalaat Farasan, d’où l’on a une bonne vue sur le port et la côte (à environ 5 km à l’est) et sur l’ensemble de la partie orientale de l’île, mais où aucune trace antique n’est apparente; c) la colline de Jabal alKniseh (aucune trace d’église, en dépit du toponyme arabe), où se trouvent des tessons antiques et de petits vestiges antiques extrêmement confus, mais aucun reste de murs de bâtiment. Nous allons voir que l’inscription a un caractère militaire: les deux dernière hypothèses,

Imp(eratore) Caes(are) Tito Ael(io) Hadr(iano) Antonino Aug(usto) Pio Pont(ifice) Maxim(o) trib(unicia) pot(estate) VII co(n)s(uli) III, P(atri) P(atriae), vexill(atio) Leg(ionis) II Tr(aianae) Fortis et auxil(ia) eius castrensesq(ue) sub praef(ecto) Ferresani portus et Pont(i?) Hercul(is) fec(erunt) et d[ed(icaverunt)] Sous l’empereur César Titus Aelius Hadrien Antonin, Auguste, Pieux, grand pontife, dans sa septième puissance tribunicienne, trois fois consul, Père de la Patrie, la vexillatio de la IIème légion Traiana Fortis, ses troupes auxiliaires et les castrenses sous les ordres du préfet du portus (?) de Ferresan et de la mer (?) d’Hercule, ont fait et dédié (ce monument).

Quelques certitudes sont acquises. L’empereur22 est Antonin le Pieux, la date est 144 apr. J.-C. Le bâtiment, probablement un fort, est le fruit du travail d’une vexillation (et ses troupes auxiliaires, et d’autres encore) de la IInde Légion Traiana Fortis, la légion basée à Nicopolis près d’Alexandrie: ce détachement est donc venu d’Egypte. Les travaux (et certainement la détachement) sont placés sous les ordres d’un préfet (et non d’un praepositus: le F de PRAEF est clair, ligne 6), qui est, si je lis bien, préfet de Ferresan (peut-être du portus de Ferresan) et du PONT HERCVL. Le toponyme Ferresan (ou Ferresanum?) est encore une certitude. C’est évidemment une variante antique de Farasan. Son

19 Avant les découvertes de l’érudit local M. Ibrahim Miftah puis nos prospections, une petite partie du patrimoine archéologique des Farasan était connue grâce au survey de Zarins & al. 1981: 26-27, où se trouve publiée une photo d’une inscription sud-arabique fragmentaire, de Qusar; les autres inscriptions sud-arabiques et l’inscription arabe ont été découvertes ultérieurement. Seule exception au caractère apparemment local du matériel céramique, Zarins signalait deux tessons nabatéens peints, à Gharrein: ces relations avec le monde nabatéen paraissent à présent confirmées par deux monnaies probablement nabatéennes que j’ai ramassées en surface au village antique de pêcheurs de Disan. Il faut attendre le VIème siècle apr. J.-C. pour trouver deux attestations littéraires de voyages et d’activités militaires où les Farasan jouent un rôle. D’une part, Nonnosos, ambassadeur de Justinien auprès des Ethiopiens et des Himyarites, passe par «Pharsan», où il découvre des hommes de très petite taille, noirs de peau, très velus, vêtus d’un simple pagne, paisibles et craintifs, mangeurs de poissons et coquillages (Photios, Bibliothèque: I.3.21-39, ed. Henry R. 1959. Je remercie J. Desanges de m’avoir signalé ce passage). D’autre part, dans les manuscrits grecs du Martyre de Saint Arethas et de ses compagnons, «Pharsan» contribue, pour sept bateaux, à la constitution de la flotte axoumite destinée à combattre contre Himyar (Detoraki 2000: 1, 298299; 2, 368). 20 Les deux pierres ici présentées sont conservées dans la cour de la maison de M. I. Miftah dans la bourgade de Farasan.

21 Elle est le résultat, outre mon travail personnel, de discussions avec de nombreux savants; notamment avec P.-L. Gatier et M. Sartre, qui, lors du présent colloque, m’ont mis sur la piste de la lecture QSVBPRAEF au début de la ligne 6. P.-L. Gatier lisait QVISVBPRAEF (les lettres VI ne me paraissent pas possible, après réexamen de la pierre en décembre 2006), M. Sartre Q..VBPRAEF. 22 Les abréviations n’autorisent pas à restituer avec certitude dans la titulature le datif (dédicace à l’empereur) ou l’ablatif (datation par le règne et les fonctions de l’empereur). Contrairement à mes propositions de restitution antérieures, je me rallie ici à l’ablatif, sur l’argument de mon ami F. Bérard (comm. orale aux Rencontres franco-italiennes d’épigraphie, Rome, septembre 2004), qui considère que le choix — arbitraire — du datif impose la conclusion que le bâtiment est en quelque façon un monument du culte impérial. Je ne partage pas en revanche l’argument de P. Le Roux (communications orales), qui considère que le datif est exclu, la pierre n’étant pas une base honorifique: une dédicace à l’empereur ne se trouve pas seulement sur une base honorifique. Le doute entre ablatif et datif persistera; il n’a pas d’incidence sur l’interprétation historique.

20

FRANCOIS VILLENEUVE: L’ARMÉE ROMAINE EN MER ROUGE

Fig. 6: photographie de l’inscription latine n°1 (L. Nehmé). (probablement ici Marin de Tyr, début du IIème siècle) n’avaient pas connaissance. PONT HERCVL représente une des difficultés du texte: ce PONT est–il un ponti (pontus, mer) ou un pontis (pons, «pont»)?24 Les deux développements présentent des inconvénients, mais les plus déterminants excluent pons, qui ne veut jamais dire «passage maritime», ni «détroit» (et rien sur le terrain ne permet de penser à un pont antique entre les îles de l’archipel Farasan elles-mêmes – encore moins, bien sûr, vers les côtes arabique et africaine, distantes de plusieurs dizaines de milles). Donc Pontus Herculis, la mer d’Hercule. Cette lecture est surprenante pour trois raisons. Pontus est d’emploi rare en dehors de la langue poétique ou emphatique et en dehors du cas du Pont Euxin pour désigner une mer, mare. Le Pontus Herculis est un hapax parfait dans la toponymie antique. Enfin, fort peu d’explications satisfaisantes peuvent être avancées25 pour comprendre un appellation herculéenne d’une partie méridionale de la mer Rouge: l’hypothèse la plus plausible est aussi la plus simple: les «Colonnes d’Hercule» du Détroit de Gibraltar auraient eu un « symétrique » oriental et méridional au Détroit de Bab al-Mandeb. Cependant, ce détroit est connu dans l’Antiquité sous d’autres noms, et le seul argument textuel que l’on puisse invoquer à l’appui de cette

Fig. 7: copie de l’inscription latine n°1 (F. Villeneuve). existence dans l’Antiquité n’est pas une surprise absolue: nous avons vu (n. 19) qu’il était attesté en grec sous la forme Pharsan au VIème siècle apr. J.-C. Néanmoins, c’est de loin son occurrence la plus ancienne: ce toponyme n’apparaît jamais dans une source à caractère géographique de l’époque hellénistique ou romaine impériale, pas même chez Ptolémée,23 à peu près contemporain de notre inscription, et qui est pourtant très détaillé sur les îles de la mer Rouge: on peut en déduire que l’opération militaire romaine dont témoigne l’inscription a permis la découverte et l’usage du toponyme local, dont les sources de Ptolémée

24

Pour les raisons expliquées dans Villeneuve 2006 contre Bukharin 2006, il ne peut pas s’agir d’un pontificis (pontifex, pontife). M. Bukharin a prétendu comprendre que le préfet de Ferresan était aussi un «pontife d’Hercule», chose à tous égards impensable. 25 Examen détaillé de cette question dans Villeneuve et al. 2004b.

23

Chez Ptolémée VI.7.43, l’archipel Farasan semble être appelé l’île des Hierakes, «Faucons».

21

THE LATE ROMAN ARMY IN THE EAST hypothèse est très vague et ténu.26 De plus, en revenant au sens historique de l’inscription: si la mer d’Hercule est ainsi appelée par référence à des Colonnes d’Hercule-Bab el-Mandeb, il faut considérer que le préfet de Ferresan avait un territoire, ou était destiné à contrôler un territoire étendu jusqu’à ce détroit, donc encore 500 km au sud de Farasan: c’est s’aventurer un peu loin des données assurées du texte.

civile de la localité»?29 Et qu’est-ce qui a conduit à mentionner ces humbles gens dans une inscription à caractère officiel?

Or les incertitudes ne manquent pas par ailleurs. Elles concernent avant tout la fin de la ligne 5, très difficile à déchiffrer, où il me semble pouvoir, après ETAVXILIAEIVS (qui est clair) lire CASTR[E]N[S]E[S]: castrenses,27 et le début de la ligne 6, où je pense qu’on peut lire: QSVB ou Q.SVB juste avant PRAEF: il n’y a pas la place pour Q..SVB (qui sub ou –que sub ): donc vexillatio legionis II Traianae Fortis / et auxilia eius castrenses/q(ue) sub praefecto. Cette proposition, au plus près de ce que je vois sur la pierre, a un sens et elle est grammaticalement correcte: je m’y tiens pour l’instant.28 Elle n’a rien cependant pour plaire à un familier de l’épigraphie latine, militaire en particulier. L’usage du suffixe -que pour corréler auxilia et castrenses alors que vexillatio et auxilia sont corrélés par et surprend, et plus encore l’abréviation peu canonique de -que en Q ainsi que le rejet de ce -q(ue) en début de ligne alors que le lapicide aurait pu s’arranger pour le placer juste après catrenses en fin de ligne 5. Mais tout cela peut s’expliquer: la différence et / -que peut se comprendre par la volonté de ne pas mettre sur le même plan les soldats (légionnaires et auxiliaires) et les recrutés locaux; les aberrations que sont l’abréviation du -que et son rejet en début de ligne ne surprennent pas de la part d’un lapicide visiblement maladroit qui a bâclé la graphie de ses deux dernières lignes. Quant aux castrenses, s’ils ne sont pas pas un hapax – le mot est commun dans une Antiquité plus tardive, avec un sens bien différent de celui que nous lui supposons ici – leur occurrence dans le vocabulaire militaire du Haut-Empire paraît être un unicum, ce qui peut conduire à rejeter cette lecture: mais je n’en vois pas de plus sûre sur la pierre. Faut-il lui donner le sens que je lui propose ici: les gens du camp, c’est-à-dire une main d’œuvre locale bien distincte des légionnaires et des auxiliaires? ou bien celui, guère différent, de « population

Fig. 8: photographie de l’inscription latine n°2, provenant de Qusar (L. Nehmé). Aussi étonnante au moins est l’absence du nom du préfet. Cette absence est pourtant assurée, dès lors que ce qui précède PRAEF (de façon sûre CASTR... ) ne se laisse pas lire comme un nom de personne30 et que ce qui suit ne peut pas l’être.31 Il faut donc expliquer cette absence, peu conforme aux usages de l’affichage militaire. La seule explication est à mon avis que le titulaire de la préfecture n’est pas encore nommé – si cette préfecture est nouvelle –, ou que l’on se trouve dans une période d’interim entre deux préfets. Quant au portus suggéré en fin de ligne 6, seulement le P et éventuellement le haut du O en sont lisibles avec un minimum de certitude: il ne faut donc pas échafauder de raisonnements sur la qualification de portus (simple port, ou poste douanier) attribuée à Ferresan, qui n’est pas avérée, mais seulement plausible.

29 C’est le sens que donnerait J.-M. Carrié (Cuvigny - Wagner 1988: corrigenda, n. 14) à l’équivalent grec kastrèsianoi s’il le trouvait dans la documentation papyrologique égyptienne du IVème siècle.J’observe en outre que castrenses est de construction identique à portarenses, membres d’ «une formation paramilitaire assimilable à la militia civile, recrutée à un niveau social inférieur à celui des soldats» (Carrié 2004). 30 Contrairement à ce que j’avais avancé dans Villeneuve et al. 2004a; 2004b, où je pensais, à tort, avant d’avoir vu la pierre, pouvoir lire: Castricius P(ublii) f(ilius) Aprinus praef(ectus). 31 Les lettres FERR, qui suivent immédiatement PRAEF, sont absolument sûres. Si l’on refuse de suivre ma lecture FERRESANI suivi de l’hypothétique PORTVS, on peut imaginer lire en cette seconde moitié de la ligne 6 le gentilice et le cognomen du préfet. Mais 1) après examen des corpus, je ne vois pas quelle association gentilice-cognomen pourrait convenir, en commençant par FERR et avec ce nombre de lettres approximativement; 2) la suite, au début de la ligne 7, est indéniablement ETPONTHERCVL: qui ne peut être la fin d’un nom, et qui impose que ce qui commence par FERR soit, comme PONT HERCVL un élément définissant la compétence du préfet.

26 Strabon, II.5.6 (C 171), stèlai d’Héraklès «en Inde»: ce qui peut désigner une bien vaste zone, de l’Afrique orientale à l’Inde moderne. Je remercie J. Desanges pour cette référence. 27 On ne peut pas lire en revanche castrensia: auxilia eius castrensia, qui autoriserait à continuer par: q(ui) sub praef(ecto) etc. M. Reddé m’indique au demeurant (communication orale13.05.2005) qu’on ne connaît pas d’auxilia castrensia. 28 D’un strict point de vue de lecture, une autre solution, suggérée par P. Le Roux (communications orales) serait possible pour ce passage précis: CASTRAPOSVER / VNT (il est vrai que sous certaines lumières on peut lire ce second A et ce O, mais la ligne serait un peu longue), donc vexillatio (etc.) et auxilia eius castra posuerunt. Mais il faudrait alors continuer le texte comme suit, en supposant grammaticalement un point: Praefectus Ferresani portus et Ponti Herculis fecit et dedicavit. C’est grammaticalement tout à fait possible, mais la structure d’ensemble de l’inscription, en deux phrases disjointes, serait alors bien étrange.

22

FRANCOIS VILLENEUVE: L’ARMÉE ROMAINE EN MER ROUGE La seconde inscription latine32 est très fragmentaire (fig. 8). Il s’agissait d’une pierre (calcaire corallien de nouveau) beaucoup grande que pour la précédente. Le fragment conservé (76 cm sur 50, et 18 à 25 cm d’épaisseur) montre qu’il ne s’agissait pas d’une simple plaque, comme pour l’inscription 1, mais d’un bloc d’assise. La hauteur des lettres est de 9 cm (contre 3,5 à 2,5 cm pour l’inscription 1). Le fragment conservé montre la présence d’un cartouche à bord mouluré, à queues d’aronde. Le fragment correspond à l’angle droit bas du cartouche; il laisse voir une bonne moitié inférieure de la queue d’aronde de droite; le texte conservé correspond à la fin de l’avant-dernière et de la dernière lignes. Sauf changement des hauteurs de ligne, le texte occupait quatre lignes, dans un cartouche de 59 centimètres de hauteur (2 pieds); il n’y a pas de moyen sûr pour restituer la longueur, mais si l’on pense à un rapport longueur / hauteur de 3/2 on peut supposer un cartouche long d’environ 90 cm. Nous avons donc affaire à une inscription nettement plus monumentale que la précédente: probablement affichée sur un bâtiment de plus grande taille. Le fragment a été découvert, au sol, sur le site des ruines antiques (et village récent déserté) de Qusar (fig. 3), soit à deux heures de marche du lieu de trouvaille de l’inscription 1. Il ne s’agit donc pas du même établissement. Qusar a dû être une localité antique (village?) d’une certaine importance, au centre du gros cap qui termine Farasan al-Kubra au sud-est et à proximité de la plus grande palmeraie de l’archipel: les maisons traditionnelles remploient beaucoup de blocs d’assise et tambours de colonne (fig. 9), quelques fragments d’inscriptions sud-arabiques (fig. 10), quelques fragments de décor sculpté sommaire; des fondations de murs antiques affleurent souvent; un puits antique est préservé. À proximité de ce puits se trouve une vaste zone couverte de fourrés d’épineux impénétrables, qui s’expliquent probablement par la présence au sol et dans le sous-sol de ruines plus denses qu’ailleurs: on aperçoit là notamment des chapiteaux sommaires de petit module. C’est en bordure de ce fourré que le fragment d’inscription a été découvert. Je suppose donc qu’il appartenait à un monument assez vaste qui se trouvait là, portant un décor au moins rudimentaire.

Fig. 9: site de Qusar, tambours de colonne remployés dans une baie de maison moderne abandonnée (L. Nehmé).

À la différence de l’inscription 1, il n’y a guère de difficulté de lecture pour ce qui est conservé: ……FERR /…..PRPR à coup sûr.

Fig. 10: site de Qusar, inscription sud-arabique, retaillée pour former une margelle de puits (?) puis réutilisée en arc couvrant une porte de maison moderne abandonnée (L. Nehmé).

De plus il me paraît certain qu’on peut lire un I (la seule autre possibilité est un T) avant FERR, et probable que juste avant ce I on aperçoit le bas d’une lettre qui semble être un V, ou un N.

32

Jusqu’ici non publiée, sauf une mention rapide dans Villeneuve et al. 2004: 421, n.7.

23

THE LATE ROMAN ARMY IN THE EAST Je lis donc:......VIFERR /.......PRPR ou......NIFERR /.......PRPR, ce qui autorise par exemple les reconstructions suivantes pour le texte: ........... ........... ..... VIFERR ...... PRPR

La restitution en dernière ligne de sub (prénom abrégé + gentilice + cognomen) leg(ato) Aug(usti) pr(o) pr(aetore) suppose une tabula ansata assez allongée, en raison de la place nécessaire pour le nom du légat; mais cela dépend évidemment de la longueur de ce nom, de la présence du prénom, etc. Au demeurant, rien ne nous contraint ici: c’est seulement à titre d’hypothèse plausible que j’ai avancé une longueur de l’ordre de 90 cm pour le cartouche.

........... ........... [...................... vexill(atio) leg(ionis)] VI Ferr(atae) [sub. ...... ..... leg(ato) Aug(usti)] pr(o) pr(aetore) ...........[Ce monument …?] ........ [a été érigé …?]

Si l’on voulait absolument raccourcir cette ligne pour éviter d’allonger ce cartouche, en «économisant» les six lettres du LEG AVG, on pourrait penser à pr(aeses) pr(ovinciae) mais praeses me semble un peu précoce sur une inscription qui a des chances de dater du IIème siècle comme l’inscription n°1.Ou – pourquoi pas? à pr(aepositus) pr(aetenturae) – fonction étudiée dans la communication de T. Gnoli ici même: une praetentura, théâtre d’opérations avancées, définirait très bien la zone des Farasan: mais je considère que la fonction est un peu rare pour être abrégée en PRPR, et je ne pense pas que praetentura soit attesté en épigraphie avant le IIIème siècle.34

…par le détachement de] la VIème légion Ferrata [sous l’autorité d’untel], gouverneur de la province Des variantes multiples existent, de lecture et de développement pour l’avant-dernière ligne, de développement pour la dernière, et l’on peut bien sûr combiner ces variantes de plusieurs façons. Pour l’avant-dernière ligne, si on lit: ..... NIFERR on peut développer: [........................ et......................sig]nifer(i duo) [un tel et untel], les deux porte-enseignes

Donc un légat d’Auguste propréteur. Donc la garnison des îles Farasan est rattachée, à l’époque de cette inscription n°2, à l’autorité d’un légat de province: et donc pas à l’Egypte (j’imagine mal que PRPR = praefectus provinciae, hypothèse présentée ci-dessus, pour la bonne forme: on aurait praefectus Aegypti). Cette province ne peut donc être que l’Arabia.

En outre, qu’on lise N ou I, on peut aussi songer, en parallèle à la lecture du toponyme Ferresan dans l’inscription 1: [...........................................................]i Ferr(esani)

Cela est-il compatible avec la VIème Légion Ferrata, s’il s’agit bien d’elle? Certes, une vexillation de cette légion peut être venue de très loin. Mais il est plus plausible qu’on ait utilisé un détachement de cette légion quand elle était présente dans la provincia Arabia. Les choses sont malheureusement un peu floues en ce qui concerne les affectations de la VIA Ferrata à l’époque qui nous intéresse: elle est basée en Syrie jusque dans les années 110; brièvement en Arabie à la fin des années 110; enfin durablement dans le nord de la Judée comme l’a montré la communication de Y. Tepper au présent colloque. Y. Tepper pense qu’elle est se trouva en Judée à partir de 120 apr. J.-C. environ. H.M. Cotton,35 en revanche, ne

ou bien: [...........................................................]i Ferr[esani.......................................................PRPR Pour la dernière ligne, il y a embarras du choix pour développer PRPR: [sub. ...... ...........] pr(aeside) pr(ovinciae) [sub. ...... .....] pr(aeposito) pr(aetenturae) [sub. ...... .....] pr(aefecto) pr(ovinciae) Rien de tout cela n’est impossible, mais il faut bien entendu proposer avant tout la solution la plus simple, correspondant à un automatisme de lecture. PRPR évoque immédiatement legatus Augusti pro praetore. I FERR ou VI FERR dans un contexte fort probablement militaire évoque immédiatement la Legio VI Ferrata. C’est donc sur cette proposition, rédigée ci-dessus, que je propose de travailler.33

34 Praetentura est attesté depuis pseudo-Hygin (sous Trajan), mais dans un sens technique interne au camp; les attestations épigraphiques dans un sens de «zone d’opérations avancées» ne sont pas connues avant les milliaires sévériens de Maurétanie; ni en Orient avant P. Doura 64 A (221 apr. J.-C.), puis le P. Euphr. 3-4 (252-256 apr. J.-C.), puis l’inscription d’Azraq sous Aurélien (praetensio: appartenance à une opération avancée). Feissel - Gascou 1995: 94-107; Bauzou 1996; Christol - Lenoir 2001. 35 Cotton 2000; actualisation de Ritterling 1925:1587-1595. Egalement Sartre 2001: 525, 598-599. Il n’y a aucun doute que cette légion a d’abord été basée dans la province de Syrie, sans doute à Raphanée. Puis en Arabie, après la IIIA Cyrenaica (remplaçant celle–ci quand elle quitta l’Arabie pour quelques années, avant d’y revenir, sans doute en 127). Puis en Judée (ou peut–être déjà Syrie–Palestine), où elle avait son camp dans la vallée de Jezréel, près de l’ancienne Megiddo et du village de Caparcotna, un emplacement qui prendra le nom de Legio (d’où le moderne Lejjûn). Mais H.M. Cotton est sceptique vis–à–vis du

33

Il est évidemment tentant de développer FERR en Ferr/esanum, toponyme – et même ressort de préfecture – attesté dans l’autre inscription, mais cela suppose une césure mal venue et allonge encore la dernière ligne, à moins de supposer Ferr(esanum) abrégé, ce qui est localement concevable, comme l’a signalé M. Reddé dans son intervention à ce colloque après ma communication. L’hypothèse sig]nifer(i duo) n’a pour elle que d’être techniquement justifiée.

24

FRANCOIS VILLENEUVE: L’ARMÉE ROMAINE EN MER ROUGE tient pas pour acquis qu’elle ait quitté l’Arabie pour la Judée avant la révolte de Bar–Kokhba en 132-135. Donc, de deux choses l’une:

chasse destinées à fournir à Rome des animaux exotiques.37 Un second groupe de questions est lié à la durée de la présence militaire romaine à Farasan et à la date de 144. L’inscription n°1 est bien datée, 144 apr. J.-C. L’inscription n°2 ne l’est pas. Nous pourrions considérer par principe que sa date doit être à peu près la même. Cette hypothèse est cependant peu plausible, pour des raisons administratives: dans l’inscription n°2 les Farasan paraissent rattachées à l’autorité du gouverneur de l’Arabie, dans l’inscription n°1 elles sont sous l’autorité d’un préfet et le corps expéditionnaire vient d’Egypte. Dès lors, il n’y a que deux solutions: soit l’inscription n°2 est antérieure à l’inscription n°1, éventuellement d’une vingtaine d’années, si nous admettons qu’elle mentionne la Légion VI Ferrata, que cette légion doit encore être dans la provincia Arabia à ce moment, et si nous considérons qu’elle n’y est plus à partir d’environ 123 apr. J.-C., ce qui nous place vers 120 apr. J.-C.; soit elle est postérieure, ce qui est peu probable, mais non rigoureusement impossible. On voit qu’il y a trop d’incertitudes sur la date de cette inscription n°2: nous ne pouvons pas nous prononcer sur la durée de la garnison romaine de Farasan, ni avant 144 ni après 144.

ou bien notre inscription date, comme la première, des années 140,36 et elle contribue à retarder la date la plus communément admise pour l’installation de la VIA Ferrata près de Meggido; ou bien on tient pour sûr que la VIA Ferrata quitte l’Arabie vers 120 – après quoi il est peu pensable qu’on ait constitué une vexillatio de cette légion pour l’envoyer fort loin de la Palestine: dans ce cas il est vraisemblable que la présence militaire romaine aux Farasan n’ait pas commencé sous Antonin, mais ait déjà existé dans les premières années du principat d’Hadrien. 3. Les questions que suscitent ces données nouvelles La première question est celle du motif: pourquoi une présence militaire romaine aussi loin en mer Rouge, aussi éloignée des bases connues de l’armée romaine? rappelons (fig. 2) que les frontières des provinces romaines sur les deux côtes, égyptienne et arabique, sont à plus de 1000 km au nord, et que Berenike, le principal et le plus méridional des ports romains connus en mer Rouge, est à environ trois semaines de navigation de Farasan, par vent favorable. Il n’y a pour moi aucun doute que l’explication doit être liée d’une façon ou d’une autre au commerce de l’Inde (au sens large): il peut s’agir à Farasan d’une base permanente pour assurer la protection – contre la piraterie – de la flotte annuelle du commerce vers l’Inde; ou simplement d’un poste de relais destiné à fournir d’un point d’escale de secours sûr; ou encore, il peut y avoir là un poste de contrôle douanier destiné à lutter contre la contrebande qui chercherait à éviter la taxe douanière de 25%, la tetarte. Il n’y a en revanche aucune raison, dans les sources, de penser à une idée ou à des préparatifs d’annexion du royaume d’Axoum ou des royaumes de Saba, Qataban et Himyar. Je ne suis pas convaincu par des explications beaucoup plus ponctuelles, comme par exemple des expéditions de

Pouvons-nous en revanche dire quelque chose sur cette date de 144? Du point de vue de l’histoire de l’Abyssinie et des royaumes sud–arabiques, c’est malheureusement une période tout à fait obscure. Mais c’est assurément, tout de même, une période antérieure aux conquêtes axoumites en Arabie du Sud (vers 220),38 et probablement antérieure aussi aux conquêtes du roi anonyme d’Adoulis sur la côte de l’Arabia deserta, vers 200 (v. ci-dessus). Cela donne un repère: les visées romaines bénéficient au milieu du IIème siècle d’une faiblesse des états constitués, en Afrique et en Arabie Heureuse. 37

Une explication par la chasse aux animaux exotiques est suggérée par mon ami T. Bauzou et par P. Le Roux (comm. orales). Ce dernier me signale le passage où Dion Cassius (LXXV.14.3) dit que Plautien, préfet du prétoire de Septime Sévère, envoya des centurions pour s’emparer, dans les îles de la Mer Erythrée, des chevaux consacrés au Soleil et semblable à des tigres, donc peut-être des zèbres. L’épisode est postérieur d’un demi-siècle à la date de notre inscription latine n°1, mais P. Le Roux observe qu’Antonin le Pieux s’est attaché longuement à préparer les cérémonies du neuvième centenaire de la fondation de Rome (147 apr. J.-C.): il n’est pas impensable que trois ans avant cette date une expédition ait été envoyée pour s’emparer d’animaux extraordinaires destinées aux parades et jeux de l’amphithéâtre à Rome. Une telle expédition est possible. Je suis cependant sceptique quant à leur valeur explicative pour nos deux textes. D’une part, les îles de la mer Rouge en général, les Farasan en particulier, ne me semblent pas susceptibles d’avoir fourni les animaux très remarquables dont le public romain était friand, comme les girafes ou les rhinocéros; quant aux zèbres (?) dont semble parler Dion Cassius pour l’époque sévériennne, il faut évidemment relever que les îles de le mer Erythrée qu’il mentionne ne se trouvent pas nécessairement en mer Rouge mais plus probablement au-delà de la mer Rouge. D’autre part, une expédition de chasse, même majeure, n’aurait pas nécessité, sans doute, l’installation, peut-être durable si nos deux inscriptions sont séparées par plusieurs années, d’une garnison à Farasan, et encore moins la création d’une préfecture territoriale. 38 Robin 1989.

consensus qui règne au sujet d’une arrivée définitive de la VI A Ferrata à Caparcotna vers 120 apr. J.-C ( M. Sartre: 123 apr. J.-C.; Y. Tepper, The camp of the Legio VI A Ferrata at Legio, dans le présent volume). Il établit qu’elle est la légion d’Arabie juste avant 119. Il est d’accord sur le fait qu’il y a bien une légion qui a son camp à Caparcotna dès 120 apr. J.-C., mais il n’est pas sûr qu’il s’agisse bien déjà de la VI A Ferrata, considérant qu’il n’y a pas de source explicite pour dire cela. Par conséquent, il n’est pas clair pour lui si la VI A Ferrata arrive en Judée avant ou après la révolte de Bar Kokhba en 132-135. H.M. Cotton considère que sa présence n’est certaine en Palestine que vers 160 apr. J.-C. 36 Ce peut être une hypothèse de travail, mais je ne la fais pas a priori pour ma part.Ce n’est pas parce que deux inscriptions militaires latines sont trouvées sur une île qu’elle datent du même moment – même si ces inscriptions ne sont apparemment que deux et même si ces îles sont les Farasan. Nous l’avons vu, les sites sont différents, distants de 8 km, les bâtiments sont différents. Le contenu même des textes paraît militer pour des dates différentes: corps de troupes de légions distinctes, attaches avec l’Arabie d’un côté, avec l’Egypte de l’autre.

25

THE LATE ROMAN ARMY IN THE EAST Du point de vue de l’histoire militaire romaine en Orient, l’année 144 correspond à un moment où les tensions du début du règne d’Antonin, avec les Parthes, sont apaisées: on peut donc penser que c’est un moment où l’empire a les mains plus libres pour s’occuper militairement d’affaires non guerrières, telles que protéger de la piraterie le commerce avec l’Inde.

semaines de navigation des frontières. Il conviendra d’essayer de repérer les possibilités, qui sont limitées: seules des îles ayant beaucoup d’eau douce et de la nourriture (pêche, fourrage ou gibier) peuvent convenir. Ces vestiges romains de Farasan ouvrent des questions intéressantes, plus qu’ils n’apportent de certitudes. Ils ont en tout cas le mérite de faire sortir de l’ombre la politique militaire romaine en mer Rouge et alentour au IIème siècle.

Or justement, c’est l’époque dont date le fameux papyrus «de Mouziris»,39 contrat de prêt concernant une cargaison de nard, d’ivoire et d’étoffes de l’Inde destinée à la Méditerranée, datable des alentours de 150 apr. J.-C. À lui seul ce document montre que le commerce entre l’Inde et Rome n’avait nullement décliné au milieu du IIème siècle. D’après les calculs de D. Rathbone, la cargaison du bateau concerné valait 7 millions de deniers; s’il y avait 100 bateaux comme celui-là qui faisaient annuellement le voyage de l’Inde, leur cargaison représentait la valeur de toute la production agricole de l’Egypte.

Bibliographie Bauzou T. 1996 La praetensio de Bostra à Doumata (elJowf). In Syria 73: 23-35. Bernand E. 2000 Recueil des inscriptions de l’Ethiopie des périodes pré-axoumite et axoumite, Paris. Bowersock G. 1983 Roman Arabia, Cambridge, Mass. Bukharin M. 2006 Romans in the Southern Red Sea. In Arabia 3: 135-140. Carrié J.-M. 2004 Portarenses (douaniers), soldats et annones dans les archives de Douch, oasis major». In Balmelle C., Chevalier P., Ripoll G., Mélanges d’Antiquité tardive. Studiola in honorem Noël Duval, Turnhout: 261-274. Christol M. - Lenoir M. 2001 Qasr el-Azraq et la reconquête de l’Orient par Aurélien. In Syria 78: 163178. Costa P. 1977 A Latin-Greek inscription from the Jawf of the Yemen. In Proceedings of the Seminar for Arabian Studies 7: 69-72. Cotton H.M. 2000 The Legio VI Ferrata. In Le Bohec Y. (ed.), Les légions de Rome sous le Haut–Empire, Lyon: 351-357. Cuvigny H. - Wagner G. 1988 Les ostraca grecs de Douch II, Le Caire. Cuvigny H. - Robin C. 1996 Des Kinaidokolpites dans un ostracon grec du désert oriental (Egypte). In Topoi 6: 697-720. Cuvigny H. et al. (ed.) 2003 La route de Myos Hormos. L’armée romaine dans le désert oriental d’Egypte, 2 vol., Le Caire. Demougin S. 1980 Eques, un surnom bien romain. In Annali del seminario di studi del mondo classico. Sezione di archeologia e storia antica 2: 157-169. Desanges J. 1999 Toujours Afrique apporte fait nouveau. Scripta minora, Paris. Detoraki M. 2000 Le martyre grec de s. Arethas et de ses compagnon (BHG 166), 2 vol., Paris: thèse de l’Université de Paris IV. Dijkstra M. - Verhhogt A. 1999 The Greek-Palmyrene Inscription. In Sidebotham S.E. - Wendrich W. (ed.), Berenike 97. Report of the 1997 Excavations at Berenike ans the Survey of the Egyptian Desert, including Excavations at Shenshef, Leiden: 207-218. Facey W. 2004 The Red Sea: the Wind Regime and Location of Ports. In Lunde P. - Porter A. (ed.), Trade and Travels in the Red Sea Region. Proceedings of

L’intérêt de Rome à protéger cette navigation contre la piraterie, et à protéger sa douane de la tetarte contre la contrebande était donc évident. Sommes-nous documentés sur ces pirates et contrebandiers en mer Rouge? Non, sauf par le témoignage de Pline au Ier siècle. Mais il n’y avait pas de raison pour que la piraterie eût diminué au IIème siècle. On peut soupçonner qu’elle était en partie l’œuvre des Kinaidokolpitai (v. carte fig. 2), ce groupe de tribus dont le territoire s’étendait de Leukê Kômê jusqu’au nord de Gizân et que le roi anonyme d’Adoulis (v. ci-dessus) alla mettre au pas, sans doute un demi-siècle plus tard, une fois les Romains partis probablement. Au milieu du IIème siècle, ces Kinaidokolpitai traversaient parfois la mer Rouge: vers 150 toujours, un ostracon trouvé au fort romain de Maximianon entre Coptos et Myos Hormos mentionne une dépêche, portée par deux soldats, relative à des Chinedokolpitôn, qui sont peut-être des brigands.40 On peut donc supposer une grosse opération de police en mer Rouge. En 143 ou 144, l’Eloge de Rome d’Aelius Aristide (§ 70) mentionne incidemment un moment de folie (kakodaimonia) des riverains de la mer Rouge, visiblement maté: cela pourrait être une allusion à notre opération de répression et de protection, dont les Farasan sont la base ou l’une des bases géographiques. Reste justement la question de l’extension géographique de l’opération, ou de la conquête si c’est une chose durable. Sur ce plan, nous en sommes réduits aux conjectures. Rien n’interdit de penser à l’existence de bases militaires plus méridionales, jusqu’au Bâb alMandeb par exemple. Plus au nord, il est très probable à mon avis qu’il y a dû y avoir des positions de relais, et sans doute sur des îles: il est difficile d’imaginer une ou deux vexillations abandonnées seules, à 1000 km et trois 39 40

Rathbone 2000. Cuvigny - Robin 1996.

26

FRANCOIS VILLENEUVE: L’ARMÉE ROMAINE EN MER ROUGE Red Sea Project I (British Museum, October 2002), Oxford: 7-17. Feissel D. - Gascou J. 1995 Documents d’archives romains inédits du Moyen-Euphrate (IIIème siècle après Jésus-Christ). In Journal des savants 1995: 65119. Ghabban A 1994 Akrakumi, mina’ al-Hijr: qissat iktishaf mina’ Akra (en arabe: Akrakumi, port de Hegra: histoire de la découverte du port d’Akra). In Kinda 1: 21-25. Gunter A.C. 2005 (ed.), Caravan Kingdoms. Yemen and the Ancient Incense Trade, Washington, DC. Henry R. 1959 Photius Bibliothèque. T. 1, Paris. Parker S.T. 2005 Supplying the Roman Army on the Arabian Frontier. In Visy Z. (ed.), Limes XIX. Proceedings of the XIXth International Congress on Roman Frontier Studies (Pécs): 415-425. Peacock D. - Blue L. 2006 Myos Hormos-Quseir alQadim: Roman and Islamic Ports on the Red Sea. Vol. 1, Survey and Excavations 1999-2003, Oxford. Rathbone D. 2000 The «Muziris» papyrus (SB XVIII 13167): financing Roman Trade with India. Alexandrian Studies II in Honour of Mostafa elAbbadi. Bulletin de la Société Archéologique d’Alexandrie 46: 39-50. Reddé M. 2002 La présence militaire romaine dans le désert oriental. In Topoi. Supplément 3: 385-394. Ritterling E. 1925 Legio. In Pauli C. - Wissowa G. (ed.), Realencyclopädie der classischen Altertumswissenschaft, XII: 1186-1837. Robin C. 1989 La première intervention abyssine en Arabie méridionale (de 200 à 270 de l’ère chrétienne environ). Proceedings of the Eighth International Conference of Ethiopian Studies, 1984, vol. 2, AddisAbaba: 147-162. Sartre M. 2001 D’Alexandre à Zénobie. Histoire du Levant antique, Paris. Sidebotham S.E. - Wendrich W. 1998 Berenike: Archaeological Fieldwork at a Ptolemaic-Roman Port on the Red Sea Coast of Egypt: 1994-1998. In Sahara 10: 85-96. Sidebotham S.E. - Wendrich W. 2001-2002 Berenike: Archaeological Fieldwork at a Ptolemaic-Roman Port on the Red Sea Coast of Egypt: 1999-2001. In Sahara 13: 23-50. Villeneuve F. 2004 Une inscription latine sur l'archipel Farasân, Arabie Séoudite, sud de la mer Rouge. In Comptes rendus d’ l’Académie des inscriptions et Belles-Lettres (paru 2006): 419-429. Villeneuve F. et al. 2004a C. Phillips, W. Facey, F. Villeneuve. A Latin Inscription from South Arabia. In Proceedings of the Seminar for Arabian Studies 34: 239-250. Villeneuve F. et al. 2004b C. Phillips, W. Facey, F. Villeneuve. Une inscription latine de l’archipel Farasân (sud de la mer Rouge) et son contexte archéologique et historique. In Arabia 2: 143-190 et 229-232.

Villeneuve F. 2006 Farasan Latin inscriptions and Bukharin’s ideas: no pontifex Herculis! and other comments. In Arabia 3, 2005-2006: 289-296. Villeneuve F. s.p. Naqshan askariyan latiniyan muktashafan fi jaza’ir Farasan (janubi al-bahr alahmar, al-mamlakah al-arabiyah al-saudiyah): Ruma, al-Iskandiriyah, al-Batra wa tijarat al-Sharq fî-l-qarn al-thani baad al-milad (en arabe: Deux inscriptions militaires latines découvertes aux îles Farasan – mer Rouge méridionale, Arabie Saoudite: Rome, Alexandrie, Pétra et le commerce oriental au IIème s. apr. J.-C.) Actes du Congrès La ville dans le monde arabe, d’après les données archéologiques, Sakaka, 5-7 décembre 2005, 2007, s.p. Zarins J. et al. 1981 The Second Preliminary Report on the Southwestern Province. In Atlal 5: 9-42. Figures copyright Nanterre, France

27

UMR

ArScAn,

Paris1-Paris X-CNRS:

New Light on the Roman Military Base at Dura-Europos: Interim Report on a Pilot Season of Fieldwork in 2005 Simon James School of Archaeology and Ancient History, University of Leicester

resulted in the interment of remarkable testimonies, archaeological, representational and textual, of the city and its life; the dry microenvironments of the site, and the lack of later disturbance or superimposition of further occupation depositions, preserved these remains nearperfectly, and made them accessible to large-scale excavation and rediscovery after the site was identified in 1920.3 Not the least of the remains identified were those of the intramural base of the Roman garrison, occupying a substantial quarter of the town in its later years.

Introduction As an archaeologist in what was at Potenza, and is in the present volume, a gathering primarily of ancient historians, unsurprisingly I am inclined to emphasise the potential and importance of material evidence for our understanding of the Roman military. Perhaps nowhere is this better exemplified than in the case of Dura-Europos, Syria. Although the textual evidence from the city has become central to Roman military studies (e.g. providing as much as two-thirds of the documents assembled by Fink in his study of Roman military papyri),1 elucidation of the history of this Roman-garrisoned city has been archaeology-led; almost all the known texts relating to Dura were recovered from the site by archaeological fieldwork, which also of course produced spectacular and evocative material remains, from arms and armour2 to military installations, fortifications and siegeworks.

The Dura base is of great interest and importance, both to the understanding of the city in its last decades, and also to wider Roman military studies. It remains the only extensively explored example of an urban Roman military base in the eastern provinces dating to the Principate (albeit the later Principate: the best comparator is the later Tetrarchic base at Palmyra).4 Textual evidence reveals that urban basing of troops, whether intramural or immediately adjacent to cities, was routine in the Roman East, in contrast to the custom-built ‘forts’ and ‘fortresses’ typical of the northern European frontiers. Vastly more is known about European military stations than those of the East, a further reason for the importance of the Dura example. Its significance is even greater when the richness of other components of the Durene evidence is taken into consideration; the relatively extensively explored archaeological remains of the base are complemented not only by a valuable military epigraphic record, but also by the famous military papyri, which tell us much about the composition and activities of the garrison.5

What follows is, I hope, an illustration of the potential value of material evidence in its own right, as an independent method of enquiry into the study of the Roman military, and other aspects of life in the ancient world. This presents the early stages of a study of site topography, structures, spaces and their relationships, examined on the ground. I believe that the preliminary results already offer a significant new contribution to our understanding of an important Roman military site. The conclusions offered also draw on the content of archaeologically recovered texts, and previous interpretations of them made by ancient historians; and I hope that they will in turn be of interest and use to scholars working primarily on textual material. In this, research on Dura as a whole provides an excellent example of the potential synergy between textual and material studies, in contexts where, as at Dura, we have relative riches of both.

Not the least of the reasons for undertaking new work on the base are that the Yale excavations were only partially and rather inadequately published, and that many decades of wider military scholarship have been undertaken since digging was halted by lack of funds in 1937. It has now also become apparent that significant work within the military base area was undertaken which is not only unpublished, but for which no documentation appears to survive;6 a notable example is the identification of the excavated foundation of what seems to be a ceremonial

The site and its significance Dura-Europos is a site of great archaeological importance, not because it was exceptionally prominent in antiquity – this Hellenistic military colony grew in Parthian times to be a prosperous centre of economic and administrative activity for the middle Euphrates region – but because after its occupation by Rome from the AD160s, and its destruction by Sasanian Persia in the 250s, it was never reoccupied on any substantial scale. The preparations for the final, fatal siege of the city 1 2

3 For a general account of the discovery and exploration of Dura, see Hopkins 1979. For details of the initial identification of the site, see Breasted 1924. 4 For an overview see Pollard 2000: 43-44, 78, 298-300. 5 The Final Report on the inscriptions was never written, although the material mostly appears in the nine Preliminary Reports; see also Frye et al. 1955. For the papyri, see Welles et al. 1959. 6 Site recording was at time sketchy, and some documentation was subsequently lost, e.g. the notebooks from the final, tenth season: Matheson 1993.

Fink 1971. James 2004.

29

THE LATE ROMAN ARMY IN THE EAST

Fig. 1: Dura-Europos, showing principal regions, installations and buildings (modified from James 2004 fig. 4). 30

SIMON JAMES: NEW LIGHT ON THE ROMAN MILITARY BASE AT DURA-EUROPOS arch within the base, described below. New fieldwork, then, is timely.

the published texts of the Preliminary Reports or the Final Reports.

The new field project Between 5 and 20 April 2005, the writer conducted an initial re-examination of the archaeological remains of the Roman military base sited in the north-western7 part of the city (Fig. 1). This was intended as a reconnaissance and pilot study for a projected multi-year field programme. The opportunity and resources to conduct this initial fieldwork, as part of the activities of the Franco-Syrian Dura mission, were generously provided by M. Pierre Leriche, French Co-Director of the expedition. The 2005 project partly comprised simple visual inspection of the remains of structures visible on the surface, but also involved a series of limited interventions, in the form of clearance of vegetation, cleaning up of previously excavated areas, and some new, small-scale excavations.

An unpublished study of the base does exist, in the form of a Yale Masters thesis written by Elizabeth Holzworth.10 However, this is relatively brief and contains very little beyond the published reports. The existing evidence for the base and the garrison feature prominently in works by C. Bradford Welles11 and especially Edward Dąbrowa.12 They are also very important to Nigel Pollard’s wider study of soldiercivilian interaction in Roman Syria.13 More recently, limited renewed field study has been conducted, by Maurice Lenoir and Corinne Licoppe on the principia,14 and by Mathilde Gelin who, for her PhD thesis on mud brick construction at the city, closely examined the principia, ‘Palace of the Dux’ and not least the southern perimeter wall of the base, of which she re-excavated a short section where it was best exposed in block K515 (Fig. 8).

Location and extent of the base

Chronology

The Roman military took over a substantial sector in the northern part of the city, which became the main (but by no means only) area of accommodation for the soldiers of the garrison, and the location of their command, administrative and most service facilities (principia, bath, amphitheatre: Figs 2 & 3). The military base was physically separated from the rest of the city, in part by the wall running across blocks K7, K5, K3 and K1, with the amphitheatre forming its southeastern corner, and the E3 bath part of its eastern side. The so-called ‘Palace of the Dux Ripae’ in X3/5, the Dolicheneum in X7, also lay within the requisitioned, imperially-controlled area of the city, but appear to have been outside, and of status different from, the military base proper (infra).

Most of the visible remains of the military base appear to belong to a short period of time, at least in archaeological terms. The city was annexed by the Romans around AD 165, and its garrison was apparently already substantial by 200; consequently it probably already occupied significant parts of Dura before the base as we now see it was planned and imposed on the pre-existing streets, blocks, houses and temples of the city. A key finding of past research, and especially of Gelin’s work, has been to confirm that key components of the military base, notably the principia and the southern boundary wall, were all built as a single conception which included the so-called ‘Palace of the Dux Ripae’, in a concerted campaign over a few years, especially 211212.16

Previous work Excavations of the structures of the base were recorded piecemeal in the Preliminary Reports; the existence of a distinct, bounded military quarter in the city only became apparent during the course of the campaigns (Roman soldiers also being widely attested in other parts of the town as well as, of course, at the Palmyrene Gate and in the towers along the city walls). Important highlights include the publication in Preliminary Report V of the so-called ‘praetorium’ (principia) and the adjacent Temple of Azzanathkona which appears to have been converted into the headquarters of cohors XX Palmyrenorum, along with a plan of the heart of the base8 (Fig. 4). No overall plan of the base was attempted beyond the Detweiler plan9 used as the basis for Fig. 3. There was also little attempt at an overview of the base in

So far as we currently know, the military quarter appears only to have undergone minor changes during the subsequent 40 years. The following describes the base on the assumption that the principal features described were all in use together, as they probably were from the 210s to the early 250s. (Incidentally, such an early date for the ‘Palace’ indicates that it originally had some other purpose than headquarters and residence for the dux ripae, apparently a regional military command only created some decades later. It remains to debate its original purpose and the history of its actual use, but I suspect that it was in 10

Who later married Frank Gilliam: Holzworth 1939. Welles 1951. Dąbrowa 1981. 13 Pollard 2000. 14 Lenoir and Licoppe 2005. 15 Gelin 2000: 308-311. 16 Gelin 2000: 311. 11 12

7 The following text follows the convention that site North is on the line of streets A-J. 8 Rostovtzeff 1934: 131-237 and pl. 3. 9 Published with Preliminary Report IX.iii (Rostovtzeff et al. 1952).

31

THE LATE ROMAN ARMY IN THE EAST

Fig. 2: The northern part of the city based on the Detweiler plan published with Preliminary Report IX.iii, with addition of block and street identifiers. (Detweiler plan courtesy of Yale University Art Gallery Dura-Europos Collection.)

Fig. 3: The limits, road system, access points, and identified structures of the military base in relation to the adjacent region X and surrounding topography. (Based on the Detweiler plan, courtesy of Yale University Art Gallery DuraEuropos Collection.) 32

SIMON JAMES: NEW LIGHT ON THE ROMAN MILITARY BASE AT DURA-EUROPOS practice multifunctional; while it may have served as a main or temporary residence for various officials, from visiting provincial governors to figures such as the praepositus praetenturae,17 it was surely also de facto an official mansio, providing fitting accommodation for diplomatic envoys in transit to or from Ctesiphon,18 and probably emperors on campaign).

In the event, the season also resulted in the cleaning, partial (re)excavation and recording of other features in the military base (notably the ‘arch’ foundations and adjacent features, discussed below).

Aims and objectives of the 2005 fieldwork

Conditions on site

The aims of the season were to try to understand:

Little if any backfilling of excavated areas was conducted by Yale. This has the advantage that it is still possible to inspect directly most of the recorded features, but also has the regrettable consequence that their long-term exposure has led to substantial erosion of the archaeological remains by wind and rain, biological factors (root action, burrowing mammals and insects, especially massive ant-nests) and human intervention ranging from the boots of tourists and the tyres of vehicles to the deliberate actions of vandals.

1. 2.

projected line towards the amphitheatre in F3? These issues were to be pursued by excavation .

how the base was physically fitted into the fabric of the Parthian-era city the rationale for its layout and conformation

These were to be achieved by: 1.

2.

3.

4.

5.

Visual inspection of the exposed remains in conjunction with the published plans and records, to check the veracity of the recording and to try to make overall sense of the camp layout. This focused especially on seeking to understand how the base was bounded, and accessed within the town Selective cleaning and recording of an apparently important, excavated but unpublished feature: the gate through the southern mud-brick camp wall on B St between blocks K5 and K3 Examination of the best exposed section of the camp wall immediately to the west of the B St gate, where it runs through the south rooms of the northeast corner house of K5; here Gelin had re-excavated the wall, and her section was re-examined, not least to familiarize the writer with the wall’s construction in particular, and an introduction to excavating mud brick in general Ancillary to 3., the K5 house, internal to the base but for which no drawing is currently known other than the tiny one on the Detweiler city plan, was to be examined and planned, with the assistance of a Total Station. Investigation of the eastward extent of the camp wall, which in the Detweiler plan (the only apparently accurate published drawing)19 appears to reach, and partly to cross, D St. The top of the wall, and the spoil tips from its excavation, are apparent across K1 as far as D St. Did the Detweiler drawing of its partial closure of D St simply mark the end of the excavated area, the end of detectable traces of the wall, or a pedestrian gate into the camp on D St? Was there any evidence of the wall continuing on its

In general, exposed mud brick constructions have dissolved into shapeless mounds, obscuring their original conformation and extent. The djousse20-mortared rubble construction which constitutes the lower part of many walls, has suffered damage, sometimes complete collapse. Such erosion is relatively severe in, for example, block E8, where relatively lightly-built civil housing had been converted into military accommodation by insertion of secondary walls of mud brick; here many of the walls, which stood more than a metre high in the 1930s, have since collapsed. However, elsewhere, for example in E4, barrack/house walls still stands well above head-height, although they, too, are increasingly eroded. Another localised problem is that spoil from the later Yale seasons sometimes obscures features revealed in earlier campaigns (e.g. the south wall of the K5 house, and parts of the north colonnade on 10th St). New features within the military base Visual inspection of structures within the military base resulted in a significant discovery or, more accurately, rediscovery, as it comprised archaeological features evidently excavated during the Yale campaigns, but not published (it is yet to be determined whether unpublished records exist in the Yale archives, although none are known to the curator). 21 The most detailed published plan of the 10th St area (usually seen as equivalent to the via principalis of European military bases in that it is a monumentalised street running laterally from the front of the principia, albeit much reduced and constrained here) is Rep. V Pl. 3

17

See Gnoli in the present volume. P. Dura 60B of AD208, written just a few years before the Palace was built, is a letter from the provincial governor to various station commanders along the Euphrates, including Dura, instructing them to provide the ‘customary hospitality’ for an expected Parthian envoy, a phrase suggesting that such diplomatic travellers passed through places like Dura reasonably frequently. They will also have had substantial retinues requiring accommodation (Welles et al. 1959: 223). 19 The plan published in Rostovtzeff’s Dura-Europos and Its Art is clearly untrustworthy: it omits the B St gate (Rostovtzeff 1938 fig. 6). 18

20 A gypsum-related material incorporating clay and organic components, used as mortar and plaster: Dandrau 1997. 21 Susan Matheson, pers. comm.

33

THE LATE ROMAN ARMY IN THE EAST

Fig. 4: The heart of the military base as presented in Preliminary Report V, Pl. III. NB the colonnaded area west of the ‘praetorium’ (principia) appears to be largely wishful thinking. (Plan courtesy of Yale University Art Gallery DuraEuropos Collection.)

Fig. 5: Detail of Rep. V, Pl. 3 with gate/arch piers and other unpublished details sketched on. The left column position on the north side of 10th Street’s northern colonnade is inferred, as is a hypothetical pier to match that built into the northeast corner of the arch foundation. 34

SIMON JAMES: NEW LIGHT ON THE ROMAN MILITARY BASE AT DURA-EUROPOS

Fig. 6: Unpublished structures at the junction of 10th and F Streets, between the principia and the bathhouse

Fig. 7: The unrecorded foundations of what seems to have been an arched gate spanning the line of F St, but partially blocking 10th St (the ‘via principalis’ of the Roman military base). In the foreground is the north colonnade of 10th St which runs across the line of F St. 35

THE LATE ROMAN ARMY IN THE EAST (Fig. 4). It contains some interesting details, such as two roughly rectangular piers at the east end of the south colonnade, where F St debouches into it from the south, perhaps supporting gates to control access between F St and the heart of the military base; if so, a second ‘wicket’ gate would have been needed to control pedestrians turning left into the south colonnade of the via principalis. However, inspection on the ground reveals that there was a considerably more elaborate structure here than is shown on the plan. Instead of small simple piers, there are two far more substantial, but apparently unrecorded foundations of mortared rubble, aligned either side of the axis of F St, standing partially across 10th St (Figs 5, 6 & 7). Close examination suggests that originally there had indeed been two small piers as shown on the published plan, but that subsequently these were greatly expanded to the north. Dating is unclear, but the visible sequence places construction of the expanded piers subsequent to the building of the colonnade system. The absence of these foundations from the publication is puzzling, as they are substantial and stand some 0.2 m above the stylobate of the adjacent 10th St colonnades; they do not represent remains of earlier structures demolished when the camp was built, but are stratigraphically later than the colonnades. The most probable explanation for their absence from the publication is simply that they were excavated in one of the later seasons; other excavated structures, including whole houses and temples, are visible on the ground, but absent from the published plans for this reason.22 However, it remains curious that these foundations, in the midst of an area excavated relatively early in the Yale/French Academy project, were apparently not discovered and recorded along with the immediately surrounding structures; perhaps they were initially covered by spoil from the baths and 10th St colonnades, piled as elsewhere in the middle of the street (Fig. 10). Who excavated them and when remains a mystery which future work on the archive might illuminate.

unnecessarily massive and elaborate simply to control access. It also half-blocks the width of the so-called via principalis; had its depth had some simple practical reason alone, it could have been built south of the line of the colonnade, actually in F St. This structure was intended to be seen both from F St (albeit with the west pier partly obscured by the corner of E6) and from the principia entrance. The location strongly suggests that it also had a monumental visual character, even though the superstructure above the pier foundations was most likely based around a mud-brick core. All traces of the superstructure were removed in the original excavations, but it is suggested that the plan implies a roofed structure of some kind on top of the piers, most likely in the form of an arch, perhaps constructed simply from timber with stucco decoration. (Other major monumental elements of the Dura base were constructed with such ‘flimsy’ materials: the amphitheatre, and much of the ‘palace of the dux ripae’ are of mudbrick). The symbolic significance of this structure is of course lost, but perhaps it celebrated a victory or imperial visit. Whatever its true character, the discovery of an elaborated structure at this key point within the military base strengthens the case for its possible identification with the guarded ‘groma’ (crossroads) mentioned in the papyri, although the writer believes that an even stronger candidate for this location has now emerged elsewhere; this is discussed in detail below. In the course of examining the piers and their surroundings, some other minor discoveries were also made. On both the published plans the north colonnade is shown running unbroken from the principia to the bath, across the line of F St. Removal in 2005 of Yale spoil, apparently from the excavation of the large piers and which obscured the line of the north colonnade, confirmed that the latter was indeed continuous (Fig. 7). An additional column base visible several metres to the north of the colonnade line but not on the published plan, was also recorded (Fig. 6).

These large piers are each c. 1.8 by c. 6 m, with a carriageway between of c. 4.5 m. One is attached to the northwest corner of the wall defining E4 by the E3 bath on the east, the other to the end of the south colonnade of 10th St along the northern side of E6 on the west. Thus pedestrian access onto the south colonnade was possible in front of the structure, while the gap of c. 3 m between the north ends of the piers and the line of the north colonnade permitted vehicular traffic and animals to turn left after passing between the piers to reach the main carriageway of the ‘via principalis’.

It is apparent that the north portico was constructed according to two different modules, west and east of the line of F St. This road was certainly bridged by the colonnade, and closed to wheeled traffic northwards (the colonnade’s stylobate is continuous), but the odd, wide spacing of the column bases at this point suggests that there was probably still a demarcated routeway, accessible to pedestrians, running across the colonnade on towards the forecourt of the Temple of Azzanathkona to the north.

In plan they are not simple rectangles, but have angled inner corners at the northern ends, and cutouts, probably for doors, set in the north end. The structure most probably functioned as a gate; however, this was not its sole, or perhaps even its primary, rationale. It is

Investigation of the base perimeter The main planned effort of the field season was focused on improving understanding of how the base was bounded, and how it was accessed, from the ‘civil town’. To the north and west it was enclosed by the city walls, through which there was a postern by Tower 1 leading

22

For example, a sanctuary in X9 identified by Pierre Leriche, and an entire undocumented house in block I8: J. Baird, pers comm.

36

SIMON JAMES: NEW LIGHT ON THE ROMAN MILITARY BASE AT DURA-EUROPOS into the southern Wadi. Pierre Leriche23 has also suggested the intriguing possibility of a hithertounsuspected additional gate in the western city wall, between the closely-spaced mud brick towers 25 and 26, at the point where the Ottoman road later crossed the ramparts. This attractive idea is yet to be investigated, but any gate at this point will have been much smaller than, and subsidiary to the great Palmyrene Gate.

southern ranges of rooms belonging to the pair of houses forming the northern end of the block. This was reexamined in some detail in the easternmost, fullyexcavated house (Fig. 8). The conformation of the wall from the evidence of the K5 house The house walls survive up to about 1.5 m high here, and the camp wall, now weathered to a rounded profile, stands to at least a metre, as was clearly visible in 2005 from Gelin’s still-visible reexcavation. It is hypothesised that, when the wall was built (Fig. 9), the rooms through which it was to pass were unroofed, and the mud brick superstructures of the walls demolished, but the rubblebuilt soubassements left intact, perhaps in part because at least on the camp side they were intended for immediate reuse (see below). Stretches of mud brick walling were then constructed within each half-demolished room, and then above the level of the wall-bases were raised as a continuous curtain wall to an unknown height. At least here in K5, there will have been a clear space of only a metre or two between its façade and the roofs of the adjacent housing outside the base. It is suggested that it was raised to a height of maybe 5 m, enough to make it a real obstacle to illicit traffic either from outside or in (Fig. 9). The wall was a substantial practical barrier, if not especially visually impressive. At K5, the reduced rooms of the house on the camp side of the wall were apparently re-roofed, for use as military accommodation. It is noticeable that the sections of wall built across each room were positioned, and if necessary apparently slightly staggered in a series of small offsets or ‘doglegs’, to keep clear of doorways in the existing buildings, because it was intended these should stay in use (Fig. 8). The wall was built flexibly, as a functional barrier; its appearance was immaterial, since most of it was virtually invisible from ground level, squeezed in between close-packed buildings as it was. (Overall the survey showed that the tiny Detweiler plan reproduced in Fig. 2, which shows the wall running across K5 at a slight angle to the street grid, parallel with equally misaligned south walls of the two houses, is wrong; both the house walls and the individual sections of camp wall are aligned on the city grid, at least at this point).

The following account traces the intramural perimeter of the base, from the western city wall eastwards, starting with the most obvious feature, the mud-brick wall which appears to constitute the southern boundary of the base. The southern boundary of the base: the mud-brick ‘camp wall’ from Wall St to B St Barely mentioned in the Preliminary Reports (rating a couple of lines where it is described as 1.65 m wide),24 the mud-brick southern perimeter wall appears, as we have seen, in the Detweiler plan (Fig. 2), apparently running unbroken from the western (‘desert’) wall of the city just south of Tower 21. Its point of origin is confirmed by a site sketch made by Detweiler, preserved at Yale and identified by Gelin.25 The curtain is shown in the Detweiler city plan blocking Wall St, running across block K7, A St, and through K5 to B St where a gate is indicated. It then ran on, again apparently unbroken, across K3, E St and K1, to D St where evidence for it, or at least excavation, stopped. On the ground in 2005, no trace of the camp wall was visible between the western city wall and block K5. It may well have been excavated away by Yale at Wall St, but otherwise may have washed away if left exposed; the immediately-adjacent mud brick stairs up to the wallwalk, near-perfectly preserved when found, are now a low, shapeless mass. (However, in contrast, there is a chunk of upstanding mud-brick construction on the eastern side of Wall St, immediately north of the projected line of the camp wall, which Gelin identifies as part of the inner glacis of the pre-siege preparations: it is not part of the camp wall).26 Traces of the camp wall may still be preserved in what appears to be the surviving foot of the great anti-siege rampart,27 but this will require excavation to investigate.

The B St gate A substantial length of the wall (c. 40 m) is clearly visible right across K5 to B St, fully excavated during the Yale campaigns. Here it was built in lengths between the intact djousse-rubble soubassements (wall-bases) of the

At B St, the wall is interrupted by a gate (Figs 8 and 10). This was excavated by Yale, and the rectangular piers for the gate, which project into the road on each side, are still clearly visible on the surface. These piers comprised a base of unmortared rubble with a mud-brick superstructure; some traces of this existed on the eastern side of the gate, where it was also clear that the pier was built up against the djousse plastering of the wall of the adjacent K3 house. Here the remains stand today only c.200 mm above the level of the B St ground surface; however, the western pier is much better preserved. At its junction with the wall of the K5 house (which at this

23

Pers. comm. Rostovtzeff et al. 1952: 69. Gelin 2000: fig. 175. 26 Gelin 2000: 309. 27 In anticipation of a siege, around 254-255 the Roman garrison strengthened the relatively vulnerable western walls of the city with a sloping external mudbrick glacis, and a great internal embankment. This created a special dry microenvironment which preserved many of Dura’s most famous discoveries of texts, paintings, buildings and artefacts: Leriche 1993: 84. 24 25

37

THE LATE ROMAN ARMY IN THE EAST

Fig. 8: Plan of the fully-excavated house forming the northeast corner of K5, converted to barracks. Running through its southern range of rooms is the camp wall, interrupted by a gate on B St.

Fig. 9: Schematic elevations of the north-eastern house in block K5 (not to exact scale). This shows the suggested sequence of modification of the house by the Roman military after requisitioning of the property. North is to the right, and the north end of the adjacent property, which probably remained in civil hands, is behind the figure at extreme left. 1: the house as taken over, with substantial rooms on the south side of its courtyard. 2: the roofs and the upper, mudbrick parts of the walls of the southern range of rooms are demolished and removed, leaving the mortared rubble soubassements and door frames in place. 3: the new camp boundary wall is constructed inside the room spaces, in sections between the surviving wall-bases. The height of the continuous superstructure is conjectural, but was probably, as shown here, high enough to make it very difficult for anyone to cross to or from the roofs of the adjacent extramural houses. 4: since it seems the military builders were careful to position the new wall with respect to existing doorways inside the new boundary wall, it seems that the intention was to keep the reduced rooms in use, presumably as barracks. They would therefore have been re-roofed, as shown. 38

SIMON JAMES: NEW LIGHT ON THE ROMAN MILITARY BASE AT DURA-EUROPOS

Fig. 10: Looking south down B St, through the gate in the mud-brick wall of the military base, the piers of which are clearly visible projecting into the street on each side. The excavated house in Block K5 is to the right. The line of the western city wall and the Palmyrene Gate are visible in the background. 2001 and 2005 as a strip of grey earth, lacking vegetation, flanked by shallow channels full of grass which presumably represent excavation down each side of the wall. Spoil tips line both sides (visible in the background of Fig. 11). This feature is clearly visible as far as its junction with the line marking the eastern edge of K1 and the western side of D St. The shallow spoil tips continue across the road and a few metres into F7, but the wall, which the Detweiler plan shows continuing across most of the width of the road, was not immediately obvious on the surface. Indeed, it would be surprising if it were, since, to judge from the B St evidence, it is likely that any surviving stratigraphy on the road surface would have been at most 100-200 mm deep before excavation in the 1930s, since when it has been exposed to erosion.

point is double-thickness and surviving some 400 mm high), the pier’s stone foundation is topped with mudbrick surviving to a total height of c.1 m above the road surface. The gate piers leave a carriageway c.3 m wide, adequate for vehicles. Presumably there was a lockable gate at this point, although no traces of door sockets are currently visible. It is not clear whether there was any need for an arch or bridge over the roadway connecting the sections of camp wall on either side; this was not a defensive wall (see discussion infra), and any required surveillance was probably adequate from the towers and wall-walk of the city wall. The camp wall east to D St

Notwithstanding, excavations were undertaken at this, the easternmost point to which the wall has been traced, to try to confirm the apparent partial or complete closure of D St and, by excavating on its projected line beyond the east side of the road, also to investigate whether it continued across F7 towards the apparent SW corner of the base, at the amphitheatre.

To the east of B St, across K3, spoil tips and a shallow depression mark the line of the Yale excavation following the wall, but the wall itself is not visible due to vegetation cover. The Detweiler plan shows the wall crossing and blocking C St, but this cannot be verified as the road line is currently used as a vehicle track. Across K1, what appears to be the top of the wall was clearly visible in 39

THE LATE ROMAN ARMY IN THE EAST

Fig. 11: The two trenches across D St, looking W, with the previously-excavated line of the camp wall, running across K1, visible beyond them. Traces of the wall running across D St may be seen in the floor and sections of both trenches, although the petered out before reaching the edge of F7, marked by the soubassement in the foreground. Excavation inside the adjacent room showed that the wall did not continue across F7, at least on this alignment.

Fig. 12: Plan of the intersection of the wall of the military base with D St.

40

SIMON JAMES: NEW LIGHT ON THE ROMAN MILITARY BASE AT DURA-EUROPOS

Fig. 13: The eleven iron bolt heads in situ on D St. Note traces of wood still in the sockets, most clearly visible in the far left example. Two trenches were opened, one running from the known wall in K1, across the D St boundary, out beyond the middle of the road. The second, separated from the first by a 25 cm baulk, ran on into F7 (Figs 11 and 12).

sockets. Nine of them were closely aligned in a row (Fig. 13). These had lain on the road, with the ends of their wooden shafts (now vanished, but many are attested from Dura) up against the wall of K1.

In the western part of the western trench, poorly preserved but identifiable mud bricks were encountered at less than 10 cm depth, running up to, and apparently just across the line of the edge of the road. The base of the putative camp wall was apparently visible across much of the road as a band of grey earth which was consistent with decayed mud brick of the colour seen and recorded by Gelin in K5. However, it was indistinct and disturbed. The best preserved area, around the junction of the wall line with the edge of D St, was itself somewhat puzzling, and was also affected by disturbance. Here individual mud bricks were visible actually on the line of the street, but some were reddish, not consistent with those used elsewhere in the military camp wall – or at least its initial construction. Perhaps they represent subsequent repairs. However, most interesting was a patch of disturbance running eastwards from the junction of the camp wall with the line at the edge of D St. This comprised several lumps of djousse rubble, some with the impression of reed mats showing they were fragments of ceiling plaster. Just to the north of these was made an important find of 11 iron catapult bolt heads of socketed bodkin type, which had evidently still had their wooden shafts attached when deposited; traces of wood were visible in several

The presence of these bolts, which were evidently a handful of ammunition deliberately placed at this spot, is most likely to be explained as a consequence of the fighting in AD 256; it is very tempting to speculate that they represent evidence of a ‘last stand’ on the perimeter of the military base when the city fell to the Persians! However, their presence at this point, at a level apparently roughly consistent with the road surface, does as least demonstrate that, by 256, the wall across the street had already been demolished, perhaps previously by the Persians if the city was briefly held by them c.253, as Grenet has argued.28 This would provide another reason why the remains of the section across the street recorded by the Yale project have proved so elusive on reexcavation. In the eastern trench, a probable stub of camp wall footing was visible on the western side, and in the western section (Fig. 12). The eastern part of the trench was excavated down to a well-preserved djousse surface at the top of the soubassement of the building at the edge of F7. An area was excavated inside the wall, where the 28

41

Grenet 1988.

THE LATE ROMAN ARMY IN THE EAST

Fig. 14: The eastern military compound at Corbridge, England. The interior is on the right of the picture, with the foundations of temples outside, to the left. Note how close the c.1m-thick compound wall (bold) stands to the adjacent buildings, which it zig-zags around. Alternatively, it may well be that the wall simply stopped at this point, and that from D St to the amphitheatre the base perimeter was defined in some way other than a specially-built wall, as indeed it was on the eastern side (infra).

projected line of the camp wall should run if it continued east of D St towards the amphitheatre. Despite excavation more than 0.3 m below the level of the wall-base, no trace of the camp wall was encountered. The perimeter between D St and the amphitheatre

It also remains unclear whether there was access into the base from the south along E St or F St. For reasons which will become apparent below, it is hypothesised that these roads were blocked off and that, except for some pedestrian access points, direct access into the base was confined to two axes: the south-facing B St entrance, and another facing east on 8 St.

Excavations at D St/F7, then, show that the line of the Roman camp wall did extend at least partially across D St. However, they also demonstrated that it does not run continuously eastwards to the amphitheatre, which, on the basis of its projected line of the southern wall, and on evidence for the line of the eastern side of the base (below), appears to have constituted the SE corner of the military enclosure (Fig. 3). The issue of the line and nature of the boundary across block F7, E St, F5 and F St remains, then, unresolved. It may be that at the D St/F7 line, the wall made a significant change of direction, to take into account the unknown, perhaps very different, layout of housing in F7 and F5. It was argued above that the particular line it took across K5 was determined by the conformation of the specific houses it was driven through, since the military wished to retain them as habitable, if reduced, accommodation. If the housing in the northern part of F7 consisted of one large dwelling like that in the south part of E4, the wall may have made a substantial diversion to the south before continuing eastwards (Fig. 14 shows an actual example of such a zigzagging military cantonment wall from Britain).

The amphitheatre as part of the perimeter In F3 stood the amphitheatre, which appears to have constituted the southeast corner of the base (Fig. 3). There is much to be said about this structure, not least its odd orientation; indeed its long axis appears to be literally oriented. For present purposes it may simply be observed that it filled the northern part of F3, and its walls and controlled entrances regulated access into and out of the very corner of the base. Indeed its location may perhaps partly have been intended to allow regulated ceremonial meetings (gladiatorial, religious) of those living inside the base, with those living outside – whether other military personnel living in the town, or Durene civilians.

42

SIMON JAMES: NEW LIGHT ON THE ROMAN MILITARY BASE AT DURA-EUROPOS On the eastern side of the amphitheatre, G St was closed to all but pedestrian traffic by the wall bounding the northern side of the amphitheatre, which was extended across the line of the road, leaving only a narrow, stepped pedestrian gateway barely a metre wide leading through onto the junction of G St with 8th St (Fig. 3).

of Roman forts (only about 1 m thick). They also lacked the wide band of clear space and ditches normally expected outside forts; indeed, they ran within a metre or so of adjacent buildings, actually zig-zagging around the back of some temples (Fig. 14). The purpose of the Corbridge military compound walls was clearly not defensive. Rather, they were about internal security and surveillance, concerned with the movement of people and materiel in and indeed out of the military area. The same was true of the somewhat makeshift perimeter of the Dura base; defence of the garrison against attack was the job of the city wall. The base enclosure was to prevent unauthorised persons entering, and also to prevent soldiers (and other members of the military community such as soldiers’ families, servants and slaves) absenting themselves without permission. It was also to control access to official installations, perhaps not least the attested, but as-yet unlocated, military granaries.30

This junction marked the second demonstrable vehicleaccessible point of entry into the base. Although reduced in width by the encroachment of the north bounding wall of the amphitheatre, 8th St between the arena and the housing in E4 is still just over 3 m wide, adequate for wagons. No piers for actual gates are visible at this point, but, for reasons which will be explained below, they may not have been needed. The eastern perimeter of the base North of the 8th St entrance, the edge of the base appears to be constituted by the west side of G St, that is the eastern street frontage of E4 (Fig. 3). This contains one doorway onto the street, but otherwise is continuous northwards to the wall of the E3 baths. The E4 housing was linked to the bath by a continuous mortared rubble wall, according to the excavators specially built for the purpose.

Controlled access east of the E3 bath A wall completing the perimeter of the military base, running north from the E3 bath to the city wall, may not have been required, simply because the area was apparently entirely enclosed by other military-controlled and/or official structures, or sheer drops (Fig. 3).

The E3 bath house itself also formed part of a continuous walled barrier on the eastern side of the camp, since it lay across the line of both G St and 10th St (which west of this point forms the nominal via principalis of the base running across the façade of the ‘praetorium’ or principia: Figs 3 & 4). Whether there was any linear barrier connecting the bath with the north wall of the city, to complete the physical bounding of the base, is unknown. However, it may not have been necessary, for two reasons: first is the purpose of the boundary, and second is the way in which the topography and layout of the northern part of the city suggests it would actually have functioned.

To the east of the E3 bath, it is suggested that the north side of 10th St constituted, in effect, a further line of controlled access. The nature of occupation in block E1 is unknown, but it is hypothesised that it, too, was under official control. If so, then, with the effectively continuous frontage of the Dolicheneum and the ‘Palace of the Dux’ to the E, this would leave the major southnorth road H St as the only axis north from 10th St into the northernmost corner of the city. To the south of the Palace, the edge of the city’s central wadi, cut into vertical cliffs by Hellenistic quarrying (P. Leriche, pers. comm.), also prevented access to this area from the rest of the city. It seems that the peninsula of plateau east of the military base proper, comprising the Palace, Dolicheneum and most of the rest of region X, was under at least ‘official’, if not specifically army control; this may have been a zone with a more civil-administrative and governmental purpose, and may have been regarded as functionally distinct from, if spatially contiguous with the military base. This is explored further below.

The purpose of the base perimeter The base lacked any formal, major enclosure wall on its eastern side comparable to that in the south, but was bounded by the closure of streets and doorways, and the construction of buildings across, or modest walls alongside road lines (Fig. 3). Even the more substantial southern camp wall is very unlike the walls of Roman ‘forts’ in Europe, with their external ditch systems. However, it is reminiscent of one of the best western parallels for the Dura base: the pair of military enclosures in the quasi-urban centre at Corbridge, England, just south of Hadrian’s Wall.29 Here two compounds, containing accommodation, workshops and administrative buildings, formed a military enclave somewhat like that at Dura. The compounds were bounded by stone walls, relatively slight by the standards 29

How the base functioned within the wider city Having traced the perimeter of the base, the internal road system can be considered (Fig. 3). Note that the Detweiler plan shows that at least some of the blockedoff north-south roads were also closed along the south side of 8th St, presumably being incorporated into the adjacent housing. This plan helps to explain why the 30

Richmond 1943; Bishop and Dore 1988.

43

Mentioned in P. Dura 106 and 108; Pollard 2000: 46-47.

THE LATE ROMAN ARMY IN THE EAST camp wall was driven through the northern part of several city blocks: it was so that the military could make use of the buildings on both sides of 8th St, which became one of the key roads, if not the principal axis of the base at least in terms of the practicalities of marshalling substantial bodies of troops. Here was a substantial, regular space, close to a gate at each end, into which columns of troops could be led and detailed off to surrounding accommodation, or where they could be assembled ready to lead out of the base in an orderly fashion. It also, of course, provided authority with an axis of surveillance and control of the garrison troops, who, it is suggested, were all or mostly accommodated in the buildings on either side of the street.

direct access southwards to Main St and the Palmyrene Gate, perhaps less for sorties in a military emergency, than for convenience of personnel or supplies arriving by road from the West, or North (if the gate between Towers 25 and 26 did not in fact exist). This route passes by the M7 bath just inside the city gate, and also the ‘caravanserai’ in L4, which (if correctly identified) may well have been used for stabling of military mounts, and for draught and pack animals. The 8th St entrance directly faces the top of road leading up the gentle slope of the wadi from the other, now-lost principal entrance to the city: the River Gate, facing the Parthian/Sasanian frontier and, more prosaically, giving access to pasturage on the Euphrates floodplain.

The extent of military accommodation within the base remains unclear (Fig. 3). All of the pre-existing housing in blocks E8 and E4 was converted to accommodate soldiers, as were the two K5 houses within the base area. The survey of one of the latter conducted in 2005, the fully excavated house forming the northeast corner of K5 (Fig. 8), revealed clear evidence for the adaptation of the house to military use of the same kind as that recorded for E8. Access routes and spaces were reorganised by blocking of some doors, and insertion of secondary rubble-built walls in the same manner encountered in other excavated converted housing, notably in E8. Reduction, but continued use, of the southern range of rooms due to the driving through of the camp wall was discussed above.

These routes will have been the primary concern of ordinary soldiers and supply personnel on routine duties. However, there was also an important display and prestige aspect to the Roman occupation of Dura, which has implications for how the military base, and the adjacent official zone around the ‘Palace of the Dux’, will have been approached and perceived by visitors low and high, including governors, envoys and, probably on occasion, emperors. Important official visitors arriving from the north or west by land would pass through the Palmyrene Gate and along the broad Main St to the arch at the major junction with the north-south-running H St which, like Main St, is exceptionally wide (Fig. 16). Turning left, they would eventually reach another major junction, with 8th St, which, on the left, was apparently widened for some metres (Fig. 2). Turning in here would bring the visitor, after one block, to the east gate of the military base. Proceeding northwards along H St for another block (Fig. 16) would bring him (usually him) to the junction with 10th St: a right turn would lead him to the entrance of the ‘Palace of the Dux’, while continuing straight on would debouch into the far northern corner of the city, behind the Dolicheneum.

It is arguable whether the term ‘barracks’ appropriately describes the somewhat labyrinthine subdivisions of these former private houses, and it also masks the likelihood that some housing within the base may have been retained as larger dwellings for officers. Further, by the time the base was formalised, soldiers could legally marry, and we have direct evidence of a serving soldier living with his family, albeit at one of the Dura garrison’s outposts rather than in the base itself.31 Nevertheless, we may reasonably assume that a substantial area of the housing in the interior was given over to mass accommodation of other ranks, especially younger men who were less likely to have families.32 The evidence of E8 is consistent with this, as are the visible traces of ramshackle additions to the largely-unexcavated housing in J2. It seems likely that the substantial camp wall is itself an indicator that the bulk of the ordinary soldiery was housed here and in the areas immediately west and south, maybe in J4, J6, J8, and the incorporated parts of blocks K south of 8th St.

Those arriving at Dura from the south, or by river, would enter the River Gate, and, if they were on official business, or unable or unwilling to face the undignified scramble up the very steep wadi slope directly to the Main St/H St intersection, they would turn right along the wadi under the shadow of the citadel. This would convey them past the colonnade of the military temple on their right, to the gentle slope ascending to the plateau along a road which would bring them to the H St/8th St junction, as above, facing the entrance to the military base, with the palace on their right.

The logic of the base layout, and especially the location of its two known (and probably only) intramural gates, becomes further apparent if these are considered in relation to the topography of the city as a whole (Fig. 15). The B St entrance becomes comprehensible in terms of 31 32

It is clear from these observations that the H St/8th St junction was a crucial focus within the city, in effect the principal point of public contact between the civil town and the imperial enclave. Here four important axes come together: the road making the gentle ascent from the River Gate; H St leading northwards from the major

P. Dura 46. Phang 2002.

44

SIMON JAMES: NEW LIGHT ON THE ROMAN MILITARY BASE AT DURA-EUROPOS

Fig 15: Access routes to and from the military base.

Fig 16: Suggested main routes for official traffic within the city and suggested identifications of guarded places mentioned in the military papyri (the two portae and the ‘groma’). 45

THE LATE ROMAN ARMY IN THE EAST intersection with Main St, the principal thoroughfare on the plateau to and from the Palmyrene Gate; 8th St leading directly from the nearby eastern entrance of the military base; and the northern continuation on H St accessing the area around the ‘Palace of the Dux’. This crossroads was obviously of great official significance, and is likely to have been controlled by formal guards; it may conceivably have had a system of gates. It seems to me to be an excellent candidate for the location of the ‘groma’ referred to in the military papyri,33 variously located by different authorities.34

base; however, surveillance and control of the movements of the bulk of the soldiers was also surely a factor. Milites could be ill-disciplined and unruly; one of the papyri makes clear that this was sometimes true of the Dura garrison.38 However, there is also abundant evidence for the routine presence of soldiers living, as well as working, outside the base. Roman Dura in its later decades was, in practice, less a civil city with a military enclave, than a city converted into a military base which encapsulated the existing civil community. The soldiers of Dura-Europos were embodied human agents interacting with others – their own dependants, and the wider civil community of the city – not least in physical ways, manipulating material objects in the physical spaces of the town. Soldiers dressed, moved, acted, sounded, and even smelled in characteristic ways.39 We can picture Euphrates the beneficiarius stationed at the Palmyrene Gate using his boot and a cudgel on a recalcitrant camel driver, while nearby the off-duty actuarius Heliodorus supervised a workman painting his portrait on one of the ceiling panels of his shared house; veterans manufacturing and trading in the agora; a soldier’s funeral procession snaking down H St bound for the cemeteries sprawling across the nearby steppe; military slaves sent out to buy jars of wine; soldiers’ smaller children playing with rag dolls and wooden swords in the barrack-houses of E8, while in the courtyards their mothers and older sisters were grinding wheat that had just come up in sacks from the river barges, ready to make bread and military hard-tack for the menfolk’s coming training exercise; their elder brothers learning to repair their fathers’ horse-harness; an archer sitting in the shade of Tower 21 painting the coloured cresting onto the nocks of his arrows before heading off to ask the regimental bowyer if his new bow is ready yet; and always a busy trade in the military brothel in G5. The case of the soldiers in the city of Dura well illustrates the value and interest of studying the toolittle-discussed material conditions of life in the past. As much as the spoken and written discourses preserved in epigraphy and papyri, these actively constituted (and did not merely act as props and stage sets for) the reality of human experience in antiquity, which we all seek to understand and evoke in our work.

The groma was one of three sites in the town which are mentioned in papyri dated to the 240s as guarded by troops of cohors XX Palmyrenorum, the others being the porta praetoriana35 and the porta aquaria.36 It seems overwhelmingly likely that these are the Roman names for the great Palmyrene Gate and its lost equivalent by the river south of the citadel.37 That the Roman garrison should place emphasis on the principal gates through the main city walls makes obvious sense; to the military, the enceinte and the city gates marked the actual and symbolic perimeter of the imperial military station; the camp wall was a mere internal administrative and policing line, which explains the lack of any elaboration to its known intramural entrances. Conclusion The short programme of fieldwork conducted in 2005 has resulted in new preliminary interpretations of the conformation and operation of the Roman military base at Dura, and has helped us understand better how the Roman military fitted itself into, and adapted, the city it had seized in AD165. Clearly, the Roman installations of Dura were not just about accommodating people and functions; they were also about visibly ‘presencing’ imperial power in the city and its region. The fact that, within the city walls, the Roman provincial army saw no need for any very elaborate dividing line or portals between the strictly military enclave and the rest of the city strongly indicates that they regarded the entire intramural space as their own; the base perimeter and the controlled roads and junctions simply differentially regulated the movements of the two communities dwelling in the city. It has long be assumed that the indigenous Europaioi were not normally allowed into the

Acknowledgements I am especially grateful to Pierre Leriche for his generosity in providing me with access to the site as a member of the Franco-Syrian expedition, accommodation in the excellent (and itself historic) Dura dig house, resources to conduct excavation, and not least for his encouragement. The work would also have been very much more difficult without the diligent help of Jen Baird, especially her linguistic abilities in supervising our small excavation crew, and valuable and stimulating

33

Welles et al. 1959: 376-383: P. Dura 107-110. 34 e.g. at 10th and F Streets, or the ‘monumental entrance’ to the principia in the Final Report (Welles et al. 1959: 378) and by Pollard 2000: 46-47. 35 Mentioned in guard-rosters P. Dura 107 and 110. The editors decline to make any specific identification (Welles et al. 1959: 378). Pollard says this ‘probably was the gate near the headquarters building in Block E7’ (Pollard 2000: 46), but this is obscure. 36 Mentioned in guard-rosters P. Dura 107 and 110, which the editors identify as ‘a gate on the river’, further identified as the postern by Tower 2: Welles et al. 1959: 378, 383. 37 Pollard also suggests this identification for the porta aquaria (Pollard 2000: 46, 53), but does not take the additional step of identifying the porta praetoriana with the Palmyrene Gate.

38 39

46

Cf. P. Dura 55, c. AD 220. James 1999; James 2004: 57-66, 256-259.

SIMON JAMES: NEW LIGHT ON THE ROMAN MILITARY BASE AT DURA-EUROPOS M. (ed.), L'Armée Romaine et les Barbares du IIIe au VIIe siècle. Rouen: 83-100. Matheson S.B. 1993 The last season at Dura-Europos, 1936-7. In Scott R. and Scott A. (ed.), Eius Virtutis Studiosi. Classical and Postclassical studies in memory of Frank Edward Brown (1908-1988), Washington DC: 200-215. Phang S.E. 2002 The timing of marriage in the Roman army. In Freeman P. - Bennett J. - Fiema Z.T. and Hoffmann B. (ed.), Limes XVIII: Proceedings of the XVIIIth International Congress of Roman Frontier Studies held in Amman, Jordan (September 2000). BAR IS 1084, Oxford: 873-878. Pollard N. 2000 Soldiers, Cities, & Civilians in Roman Syria, Ann Arbor. Richmond I. 1943 Roman legionaries at Corbridge, their supply base, temples and religious cults. In Archaeologia Aeliana, Series 4 21: 127-224. Rostovtzeff M. (ed.) 1934 The Excavations at DuraEuropos, Preliminary Report on the Fifth Season, 1931-1932, New Haven. Rostovtzeff M. 1938 Dura Europos and its Art, Oxford. Rostovtzeff M. - Bellinger A. - Brown F. and Welles C. (ed.) 1952 The Excavations at Dura-Europos, Preliminary Report on the Ninth Season, 1935-1936, Part 3: The Palace of the Dux Ripae and the Dolicheneum, New Haven. Welles C. 1951 The population of Roman Dura. In Coleman-Norton P.R. (ed.), Studies in Roman Economic and Social History in Honor of Allan Chester Johnson, Princeton: 251-274. Welles C.B. - Fink R.O. and Gilliam J.F. 1959 Excavations at Dura-Europos, Final Report V, Part 1: The Parchments and Papyri, New Haven.

critical discussion of the archaeological evidence; I am also grateful for her critical comments of a draft of the present paper. Additionally I would like to thank the various other colleagues on the project who contributed, including Sebastian for survey work, Nadia for photography, Ghassan for conservation work, and not least to Mohsen, Sleiman, Abdullah and Hamed for their efforts on the excavation. In preparing the present paper, Tommaso Gnoli kindly assisted by giving me advanced sight of his own paper. Finally, but by no means least, I wish to thank Ariel Lewin and Giovanni Brizzi for allowing me to present this work at Potenza, and for the opportunity to include it in the present volume. Bibliography Bishop M.C. and Dore J.N. 1988 Corbridge: Excavations of the Roman fort and town 1947-80. English Heritage Archaeological Report 8. London. Breasted J.H. 1924 The Oriental Forerunners of Byzantine Painting, Chicago. Dąbrowa E. 1981 La garnison romaine a DouraEuropos. Influence du camp sur la vie de la ville et ses consequences. In Cahiers Scientifiques de l'Universite Jagellanne 613: 61-75. Dandrau A. 1997 Gypse, plâtre et djousse. In Leriche P. and Gelin M. (ed.), Doura-Europos Études IV, 19911993, Beirut: 155-157. Fink R.O. 1971 Roman Military Records on Papyrus, Cleveland. Frye R.N. - Gilliam F. - Ingholt and Welles C.B. 1955 Inscriptions from Dura-Europos. Yale Classical Studies 14, 127-213. Gelin M. 2000 Histoire et ubanisme d'une ville à travers son architecture de brique crue: l'example de Doura-Europos, Paris. Grenet F. 1988 Les Sassanides a Doura-Europos (253 ap. J.C.): réexamen du matériel épigraphique iranien du site. In Gatier P.-L. - Helly B. and Rey-Coquais J. P. (ed.), Geographie Historique au Proche-orient, Paris: 133-158. Holzworth E. 1939 The Military Camp at DuraEuropos, Unpublished MA thesis, University of Yale. Hopkins C. 1979 The Discovery of Dura-Europos, New Haven-London. James S.T. 1999 The community of the soldiers: a major identity and centre of power in the Roman empire. In Baker P. - Jundi F. C. S. and Witcher R. (ed.), TRAC 98: Proceedings of the Eighth Annual Theoretical Roman Archaeology Conference, Leicester, 1998, Oxford: 14-25. James S.T. 2004 Excavations at Dura-Europos, Final Report Volume VII, The Arms and Armour, and other Military Equipment, London. Lenoir M. - Licoppe C. 2005 Les principia du camp romaine de Doura-Europos. In Leriche P. and Gelin M. (ed.), Doura-Europos Études V, Damascus: 27-32. Leriche P. 1993 Techniques de guerre sassanides et romains à Doura-Europos. In Vallet F. and Kazanski 47

From Praepositus Pretenturae to Dux Ripae. The Roman ‘Grand Strategy’ on the Middle Euphrates (2nd – 3rd Cent. AD) Tommaso Gnoli Dip. di Storie e Metodi per la Conservazione dei Beni Culturali, Ravenna

were certainly negative for the Persians: The attack against Hҕatra urged the souveraign of that town, Sanatѽrnjq, II to ask Rome for help and Rome immediately sent the cohors IX Maurorum, and in its turn that king spurred the governors of Syria and Mesopotamia to send worried letters to the emperor in Rome.

1. The limes on the Euphrates during the 3rd century Probably no event has ever had more ripercussions on the oriental limes than the ascension to the throne of Ctesiphons by the Sasanids.1 It was no sudden event, but rather more a revolt broken out in FƗrs already eight years before and it grew more and more until it reached such dimensions as to make any Arsacid restauration impossible. Rebels had already been able to defeat twice the army of Artaban I in the open field, when on 28th April 224 the Great King died on the battlefield in Hormizdaghan. It was evident that this had not to be considered as an internal matter in a state that was an adversary of Rome, it was rather an event that was about to influence Roman politics in the Near East heavily, which comes out as glaringly obvious from the programmatic slogan the propaganda by the new iranian dynasty ever and ever repeated: both Herodianus (VI.4.5) and Cassius Dio (LXXX.4) agree in ascribing to ArdašƯr the claiming of those territories that had been once dominated by the Achaemenids.2 That claiming implied the non-recognition by the Persians of all the eastern provinces of Rome. They likely just aimed at the recovery of Mesopotamia and Osrhoene, i.e. those lands that had gone under the shield of Rome by Lucius Verus and Septimius Severus, but anyway I do not share the scepticism of those who maintain that AradašƯr did not aim at the Roman territories from the beginning.3 The clash with Rome and the recovery of the territories lost by the Parthians at the cost of scorching humiliations represented as many qualifying and aggregating features of a true political program being immediatly and actually pursued and attained, as far as possible. Even before he had completely settled his power over the territories previously under Artabanus, ArdašƯr put on a behaviour of evident challenge against Rome. This openly menacing attitude was actually necessary in order to strengthen the Sasanian power over the Arian nobles who rallied to avenge ‘Alexander’s larceny’. The first acts of open hostility against Roman interests go back to 229: Trifling mouvements, as the most important event we know about was an idle attempt against Hҕatra, that was not directly under Rome however.4 The effects of the actions in 239

I think Fergus Millar is right when he maintains that the political frontier represented an ‘information barrier’.5 Even though some considerations by Austin and Rankov limiting Millar’s formulation and underlining how the emperor far away and, so to say, ‘blind’ had his eyes in his provinces by means of the officia of the governors,6 it is evident however that things went wrong on the oriental limes during the twenties of the 3rd century. The excerpt from the LXXX book of the Historiae by Cassius Dio is quite clear about the chain of those events: The situation in Mesopotamia became still more alarming and inspired a more genuine fear in all, not merely the people in Rome, but the rest of mankind as well. For Artaxerxes, a Persian, after conquering the Parthians in three battles and killing their king, Artabanus, made a campaign against Hatra, in the endeavour to capture it as a base for attacking the Romans [thus, after referring about the failure of the attack against the stronghold and of the as much unfortunate attempt against Armenia, he tells on:] He accordingly became a source of fear to us; for he was encamped with a large army so as to threaten not only Mesopotamia but also Syria. [In the end, after referring the slogan by the Sasanian propaganda claiming Persian supremacy over Asia, he concludes:] The danger lies not in the fact that he seems to be of any particular consequence in himself, but rather in the fact that our armies are in such a state that some of the troops are actually joining him and others are refusing to defend themselves (transl. E. Cary).7 The reference to the attack against Hҕatra dates back the beginning of the metus persicus in Rome to 229/230 AD. About this point the passage by Cassio Dio is perfectly homogeneous with the one by Herodian, following whom:

1

Millar 1993: 148: «There can be no doubt that a profound change in the nature of the situation on the eastern frontier was now felt in Rome». 2 For an analysis of similar passages but situated in slightly different contexts: Kettenhofen 1984, Wiesehöfer 1986, Rubin 1998, Kettenhofen 2002. On the importance of the Sasanian ideology for the birth of an Iranian identity see Gnoli 1989. 3 Kettenhofen 1984 and later Sartre 1991 express theyr doubts about it. 4 Dio LXXX.3.2. Mariq 1955, Mariq 1957, Sommer 2003a, Sommer

2003b, Sommer 2005. 5 Passage cited in Millar 1982: 19. 6 Austin and Rankov 1995. 7 Dio LXXX.3-4. On the subject of the lack of discipline among the Romans troops and of the ZW[\¯, about which Cassius Dio will insist next, cfr. in particular Wheeler 1996.

49

THE LATE ROMAN ARMY IN THE EAST In his tenth year [of Severus Alexander, i.e. 230 AD] unexpected letters came from the governors of Syria and Mesopotamia with information that Artaxerxes, king of the Persians, had defeated the Parthians, broken up their eastern kingdom and killed Artabanus, the previous great king who wore the double crown (transl. C.R. Whittaker).8

estimated a triple pincer movement that failed because of the indecision shown by the young emperor, after Herodian. The narration of the expedition given by the author of the Historia Augusta is imaginative,14 but we could maintain that the expedition did not completely fail, although it missed its main aims. Until the 50ties of the 3rd century, before the war against Persia entered its critical phase, the oriental border did not undergo heavy rearrangements but for one exception we very hardly understand. 1993 Fergus Millar expressed all his puzzlement in judging Roman politics in Edessa at that time.15 Actually starting from 212/213 Caracalla abolished the reign of Edessa most probably in connection with his project to organize a Parthian expedition, and he granted the town the statute of colony.16 A phase of interregnum followed, during which the town came again under the control of equestrian officers, following its statute of town belonging to a procuratorian province. The latter phase lasted about twenty-six years, from 213 until 239, even though a very short reign by a certain Abgar ‘the Handsom’ (218/219220/221) shall be taken into account. During all this long period the royal family was represented by a certain Ma W nu pasX grib Y , ‘crown prince’ and it is very interesting that the Ps. Dionysus from Tell MahZ re made a mistake, attributing this person a reign lasting twenty-six years. It is actually certain that Ma W nu never reigned, he always remained a pasX grib Y , because that way he is defined in P2, a Syriac act issued:

Five years had alredy gone by since the battle of Hormizdaghan, and yet the 
   by the governors came to Rome   , unexpected. Herodian insists on the surprise it caused in Rome:  "!#$  % & ( & %021(3 46571( ,98#:6:"4/) ,;% ? ' ) *,+.-/) + @A/)BDCE ,%F-D *G3H (Alexander was badly upset

at the suddenness and unexpectedness of the report that had come).9 The only possible explanation for the Roman inertia that lasted about five years towards the new lords of Iran who were very busy in strenghtening their uncertain and thus vulnerable power is represented by an horrible mistake of underestimation of the opponent, a mistake that left some traces in the above-mentioned passage by Cassius Dio about the vane Persian threats. Such an error should be attributed to the peripheral apparati of the empire because Severus Alexander acted decidedly as soon as he received the news:10 A huge organizing effort with movements of troops from the Danubian front and from Aegypt, as much as from all provinces in the Near East took place. It is probable that this time too, like afterwards, the praetorian fleets from Misenus and Ravenna were mobilized and they should have had the task to ensure supplies to the army and to displace the legio II Parthica that was stationed in Albano Laziale.11 The procurator of Syria Palaestina, C. Furius Sabinius Aquila Timesitheus, received the extraordinay task of exactor reliquorum annonae sacrae expeditionis,12 and extensive requisitions were done in Aegypt and Pamphylia.13 The ambitious plan of attack

In the month of Former Kanun of the year hundred and fifty-two, in the third year of Autokrator Caesar Marcus Antonius Gordianus the Fortunate and Victorious, and in the second year of Aelius Septimius Abgar the king, son of Ma W nu pasX grib Y , son of Abgar the king, who was honoured with the hypateia in Urhoy, in Edessa, the great city, mother of the cities of Bet Nahrin (transl. Drijvers, with arrangements).17

8

Hdn. VI.2.1. There is a problem of textual tradition precisely about the cypher of the year of reign of Severus Alexander. Codices actually read AJILK6M AJNPOEOQ B"NPORQS K6N2RT AJUFV;JA NPS , but Cassola (RAAN 38, 1963: 141143) and Whittaker (1970: 88) rightly refused the transmitted text. The fourteenth year was the last one of the emperor who died in Mainz during the preparations for a Germanic expedition following the defeat in the East and the forced roman triumph. 9 Hdn. VI.2.3. 10 Potter 1987. 11 The presence of the Legio II Parthica in the East is proved by the discovery of some graves of soldiers belonging to this unit in a necropolis in Apamea, seat of this unit adopted also in other occasions (Balty 1987; Balty 1988). The expedition by Gordian III offers on the contrary a good example of how the detachments of the praetorian fleets were employed by means of AE 1956: 10 and maybe AE 1910: 36 with a more uncertain datation: Saxer 1967: 53 nr. 97; Pflaum 1967; Bollini 1966, Reddé 1986: 375. The attribution to C. Iulius Priscus, the brother of Philip the Arab, of the acephalous cursus CIL VI 1638 is mistaken as the hypotesis has been already excluded by means of decisive argumentations by Pflaum 1960: II, 831-839, nr. 324. New and diverging analyses about the latter inscription in Nasti 1997, Gnoli 2000a: 101-112. 12 CIL XIII 1807 = ILS 1330, about which Gnoli 2000b. 13 Requisitions in Aegypt: PSI VII 797, about which Bastianini & Thomas 1977; in Pamphylia AE 1972, nr. 625-628, about which Millar 1993: 149.

Millar’s hesitation still exists. It is not clear why Rome renounced to exert its direct influence on Edessa just during 238/239 and 240/241, when the Persian aggression 14 HA S. Alex., 50-55; about this narration see Rösger 1978. More generally Bertrand-Dagenbach 1990. 15 Millar 1993: 151: “in the meantime strange transformations had been taking place in Edessa”. About the history of Edessa during the 3rd century the recent anthology of Syriac epigraphical and papyrus documents by Drijvers & Healy 1999, whose numeration of the documents I follow here. After a long period of stagnation, during which the classical works by Gutschmidt 1887 and Babelon 1893 were dominating, studies on Edessa have lived a renewal thanks to Drijvers (cfr. in particular Drijvers 1977). The discovery of new documental papyrus materials to be discussed later has inspired a number of contributions so far: Millar 1993; Ross 1993; Gawlikowski 1998; Luther 1999a, Luther 1999b, Ross 2001, about which see Camplani Gnoli 2001, Gnoli 2000a, Sommer 2005. 16 Millar 1990. 17 About the title of pas[ grib\ see Gnoli 2002. About the concept of hypateia see Gnoli 2000a: 67-88.

50

TOMMASO GNOLI: FROM PRAEPOSITUS PRETENTURAE TO DUX RIPAE came true by means of the attack against Dura Europos 
 

   
   




 

  ! "#%$& '

Among these documents the probable mention of a praepositus praetenturae on a very fragmentary text being part of a *+,.-./10 2435-76.6.890 :,.-./ and containing the letters of Postumius Aurelianus, the cohort tribune of 216 AD. (PDura 66 UU) deserves to be underlined. PDura 64 is on the contrary a papyrus constituted of two letters (A and B) coming from a liber epistularum, that probably contained many more letters, by Aurelius Rufinus, proc(urator) Aug(ustorum) n(ostrorum) praep(ositus) praetent(urae) to the tribune of the cohors XX Palmyrenorum [Aurel]ius (?) Iustillus located in Dura Europos. Both are dating back to 221 AD. by means of the consolary mention. Letter A contained the request by Antonius Seleucus, governor of Syria who may quite certainly be identified with Iulius Antonius Seleucus, legate in Moesia Inferior under Elagabal, to compel an imperial libertus in charge of the supply equitibus sive mulionibus qui vexillatione Appadanensi to give wheat ex praedis fiscalibus. In the second letter the request by the governor of Syria regards a librarius of the legio XVI Flavia Firma Antoniniana who had to give somebody satisfaction. The procurator Augusti and praepositus praetenturae asked by the governor, applies to the same cohort tribune in this case too. What emerges from the documents from Dura is that, probably by Septimius Severus but certainly at least by Caracalla, the strip of land the Euphrates flows through to the north of Dura, and more precisely between Dura Europos and the confluence with the Khab( r, was left in charge of a procurator, who was expressedly in charge of the administration of this border territory that took the name of praetentura. The papyri of Dura show quite clearly that this procurator was directly subordinated to the governor of Syria Coele whom he referred to, but he was devoid of any coercitive power. Aurelius Rufinus shall always refer to the cohort tribune located in Dura in order to perform the duties he was given by the governor. As far as one can understand from the matters originating the letters in PDura 64, the questions by Aurelius Rufinus were mainly administrative: the foddering of the cattle of the vexillation located in Appadana ex praedis fiscalibus in one case and an indetermined matter in the second one. Anyway the fact that a librarius was involved makes it quite possible that this letter contained some administrative matter too. Following the editor of PDura, Frank Gilliam, both documents would not allow us to explain some points being essential for us: the relationship between the praepositus praetenturae Aurelius Rufinus and the consularis in Syria Coele Iulius Antonius Seleucus, and the one between Aurelius Rufinus an Iustillus, the cohort tribune in Dura.24 It is anyway possible to maintain that praepositus and tribunus were not inserted in the same command chain. This exchange of letters can be understood as jurisdiction by the praepositus praetenturae on the administration of the territory and particularly on the support to the troop and the managing of the imperial patrimony is taken into account, while he was completely devoid of the ius

that the father of Aelius Septimius Abgar wore a Parthian court title, that Edessa in P2 is named with its semitic name and the title of colony has disappeared, all this shows that Rome might have undergone the new souvreign much more than imposing him, but all these are too week elements.18 What is certain is that the reign of Aelius Septimius Abgar was short. P3, dating back to 242, is dated: In the fifth year of Autokrator Caesar Marcus Antonius Gordianus Eusebes Sebastos, in the consulate of Vettius Atticus and of Laepidus Praetextatus, in the month of Elul of the year five hundred and fifty-three in the former reckoning, in the year thirty of the liberation of Antoniana Edessa the Glorious, Colonia Metropolis Aurelia Alexandria (transl. Drijvers). Everything was again like before. Anyway during the whole reign of Gordianus III Edessa existed, at least formally, even though its government, which was considered as equal to the consulary power, was given to equestrian officiers.19 This interregnum situation lasted until the fifth year of Philip the Arab, 248, following what Jacob of Edessa declares.20 After a short time the situation would change completely. The first conquest of Dura Europos in 252 and then its definitive destruction in 256 would completly change the military structure of the region.21 2. The praepositus praetenturae in the papyri It is not possible to reconstruct here in detail the Roman military structure located at the confluence of the Khab r and the Euphrates. The problem is also complicated because of the possible presence of a good deal of auxiliary units that were often fractioned into smaller units composed of few dozens of men. That dispertion was necessary because of the geographical configuration of the land being unsuited to the support of more consistent garrisons, as D browa22 has recently underlined. This enterprise has recently been undertaken by A. Luther, therefore I refer to those works.23 What I want to stress here is a specific aspect emerging from the study on the papyrus material coming from that region. On three papyri, two among which come from Dura Europos and one from a private archive however from the region of Middle Euphrates, a praepositus praetenturae is mentioned. (

)

18

In PMesopotamia A there is actually no evidence of separation between Rome and Edessa. See Ross 1993. 19 Gnoli 2000a. 20 Iacob. Edess., 281-282 [211] Brooks. 21 About the double conquest of the town see Sartre 2001: 967-970 particularly citing Grenet 1988. 22 Dabrowa 1997. 23 Luther 2002; Luther 2004.

24

51

Gilliam, in Welles et al. 1959: 230.

THE LATE ROMAN ARMY IN THE EAST coercendi, for which he was compelled to ask the cohort tribune (peto compellas ordinatum Augustorum nostrorum libertum, he writes to the tribune). Praetentura during the severian time on the middle Euphrates should be simply understood as an advanced territory, a border zone, the praepositus being devoid of any particular military power - and this corresponds perfectly to the titles he possessed.25


%: 3 *   , *  / ,.-7 3  9/ in the case of Marcellus. It is noteworthy that the expression used to indicate the commands of Iulius Priscus and Pomponius Laetianus,
%: 3 * & 3  *  3 represents a hapax legomenon in our documentation, just as much as our documents would present us with three temporary governors of equestrian rank out of three in a proconsular province, an astonishing coincidence.

We know about a further recurrence of the term praepositus praetenturae referring in my opinion, and against what the editors of the document have maintained about it, to another section of the eastern border. In a legal document dating back to the 3rd century published exactly ten years ago by Denis Feissel and Jean Gascou26 a Roman citizen, Aurelius Abidsautas son of Abidierdas, bouleutes in the town of Neapolis and residing in the villiage of Beth Pur n, along the Euphrates, turns to Iulius Proculus, prefectus and praepositus praetenturae for the effective enforcement of a judgment passed by the procurator of Osrhoene Pomponius Laetianus about an accusation de vi against one of his fellow townsman. This brief document we possess in two well preserved copies is actually very difficult and I have analyzed it elsewhere in a monograph I expressedly devoted to the study of these documents coming from the middle course of the Euphrates.27 Now I wish to repeat synthetically here my conclusions contained in that work of mine diverging from those by the editors of the documents.

The most serious objection that is decisive in my opinion against the Syrian contextaulization of the dossier comes from another document, one of the two Syriac parchments in this documentary group. In P2 at the lines 3-4 we read that the document was compiled ‘in the second year of Aelius Spetimius Abgar the king, son of Ma  nu, pas grib  (crown prince), son of Abgar the king, who was honoured with the hypateia in Urhoy’ (D  LYWS SPT MYWS  BGR MLK  BR M  NU P GRYB  BR MLK  DMYQR BHPT Y  B  RHY). It is evident out of the formula of this dating that the expression determinig the charges of our governors was no simple local variation of temporary senatorial charges, exerted by Iulius Priscus, but that the  *  Pomponius Laetianus and temporarily by Marcellus coincided exactly with the ‘consularship’ the Syriac text attributes Aelius Septimius Abgar, translitterating not by chance the term hypateia into Syriac. The place where this hypateia was exerted was Edessa, thus the institutional framework where our documents shall be located is the procuratorian province of Osrhoene and the hypateiai managed by Roman equites shall all be located in the years of interregnum in Osrhoene between about 241/242 and 248 AD. The advantages emerging from this different contextualization of the documents are manifold and have been described in my monograph. The careers of our people take advantage from it. It is perfectly normal to find members of the ordo equester commanding a procuratorian province and it is not necessary to find out coincidences we know about exclusively from temporary governors in a consular province.

The papyri from the Euphrates were found on the antiquarian market. The problem of their contextualization is therefore open. The editors thought they woluld be able to locate the site recurring in most documents on the basis of some similarities with the toponomastics in the papyri from Dura Europos, the village of Beth Pur n on the left bank of the river Euphrates, to the north of Dura Europos and to the south of Circesium. Such a contextalization of the documents would situate all the administrative and institutional framework recurring in them in the imperial consular province of Syria Coele. This fits very well, in the editors’ opinion, with the titles of the provincial governors appearing in four of these documents, where Iulius Priscus, brother of the emperor Philip the Arab, a certain Marcellus maybe coinciding with Claudius Marcellus who operated in Aegypt under Philip, and Pomponius Laetianus, are mentioned as those who ‘manage the consularity’:
%: 3 * &   3  *  3 or

Let us go back to our documents: PEuphr. 3-4. Following the editors, these two copies of the same text should be dated back between 252 an 256 AD. with a certain preference for the most recent dating. In this case the document would be the latest in the whole archive, as the last document dated or to be dated, PEuphr. 9 dates back to 252. The editors are led to choose such a dating on the basis of very circumstantial points. The former is represented by the career of Pomponius Laetianus representing the terminus post quem. He is actually attested in a papyrus from Dura Europos (PDura 97) on 13th April 251 and in May during the same year as v(ir) e(gregius) proc(urator) Augg(ustorum) nn(ostrorum) in Syria Coele. Following the editors for whom Pomponius Laetianus would stand in for the proconsul of Syria Coele in PEuphr. 3-4, the procuratorship of Syria in PDura 97 should be certainly previous. If we actually would either

25 I wish to make a digression here. I think that what I have just maintained clearly shows my dissent towards some recent interpretations of the term praetentura that have been given in relation to a famous inscription coming from the oasis of Azraq, the analysis of which would not fit for this work. I am convinced that the term is actually quite generical and that we should renounce to give it an unambiguous meaning to the various occurences of the term in different contexts. I am also sure of the identity of meaning between the terms praetentura, praetensio and protensio recurring from time to time in Syria, Arabia, Africa and also in Veneto. 26 Feissel & Gascou 1995. 27 Gnoli 2000a.

52

TOMMASO GNOLI: FROM PRAEPOSITUS PRETENTURAE TO DUX RIPAE exchange the temporary proconsulship of Syria with the procuratorship of Osrhoene, as I think we should, or subordinately a temporary ad hoc charge in the reign of the Abgarids, there is no need to postpone PEuphr. 3-4 in respect with PDura 97. One more point in favour of the dating back to 256 of the text proposed by the editors of our papyrus is twofold. In the above mentioned PEuphr. 9 of 13th June 252 the same person author of the petition PEuphr. 3-4, Abidsautas son of Abidierdos, buys a slave in Beth Pur n simply describing himself as -.2&  3 +./ . In PEuphr 3-4 he describes himself as
-.246 24 *  / 

- 4-76  *-.2 -5 33 B  -.2  5 , and puts the praenomen Aurelius in front of his name. In this case too anyway, there is no need to gather any priority of P Euphr. 9 with respect to PEuphr. 3-4: the praenomen Aurelios points out that the Roman citizenship of Abidsautas was fourty years old, when he bought the slave in Beth Pur n where he used to live, going back to the Constitutio Antoniniana of 212; if he decided to omit his citizenship in PEuphr. 9, then he might have omitted also his belonging to the curia of Neapolis/Appadana, almost irrelevant aspects in the transaction analyzed here. It is obvious on the contrary that he might have underlined his Roman and above all curial citizenship while drawing up a petition to the Roman authorities, which could have conferred him a much better position in the judicial dispute he had enterprised.28 The last dated document, where the toponym Appadana appears, is PEuphr. 1 of 28th August 245, and this shall be considered as the terminus post quem of the petition in PEuphr. 3-4, the terminus ante quem obviously remaining the end of the reign of Edessa in 248. To conclude, PEuphr. 3-4 shall be dated back to a period between AD 245 and 248.

Both praepositi are first-class people possessing very different powers. I do not think that those differences should be attributed to the twenty-years-long gap separating them from one another, i.e. I do not think we observe any institutional evolution in the role of the praepositus praetenturae; I rather think that the difference comes out of the fact that the two charges are done in different although neighbouring sectors of the limes, inserted in disomogeneous administrative contexts. The praetentura Aurelius Rufinus was in charge of was comprised in the territory of the imperial consular province of Syria Coele, it represented the most advanced position of which, while the praetentura of Iulius Proculus, represented the part of the procuratorian province Osrhoene that was along the river Khab( r. The latter type of territories, for which it should even be imprecise to use the term ‘province’, were normally subjected to the superior jurisdiction of the next provinces. Just as much as the province of Judea in its phase of procuratorian government during the 1st century AD was subjected to the imperium by the legate of Syria, so the prefecture of Mesopotamia ‘lent’ so-to-say part of its military and bureaucratic apparatus for the government of the territories in Osrhoene. 3. The dux ripae As is well known, starting from the 40ies of the 3rd century, thus in full contemporaneity with PEuphr. 3-4, the papyri and the inscriptions of Dura Europos present us with four people possessing the status of dux ripae: Licinius Pacatianus in 245 (PDura97, II.21 and 22), Ulpius Tertius in 248 (PDura 97, I.24), Iulius Iulianus in 251 (SEG XVII 770 and SEG XVII 771 = AE 1953, 265 and 266), beyond Domitius Pompeianus, the only one possessing complete titles (
%-1  *   /    / ) in a nondated graffito found in the ‘palace of the Dux Ripae’.30 Notwithstanding the deep exegesis of these documents led many years ago by Frank Gilliam, it is necessary to observe how these documents do not tell practically anything about the powers and functions of the dux as much as they do not explain in anyway when the function was created. This is not because I do not share the conclusions by Gilliam, but because I wish to warn about the fact that his analysis was conducted on the basis of the probable but uncertain analogy of this official with the duces of a later time and attested in different zones. This process can stir up surprises, as the above mentioned case of the praepositus praetenturae demonstrates. What remains is that the dux ripae was the first one among the duces who starting from the end of the 3rd century had regular and permanent commands:

I am sorry if this preliminary digression on PEuphr. 3-4 is maybe too long. Despite the similarity in the titles, as both are praepositi praetenturae, differences between Aurelius Rufinus and Iulius Proculus are evident. Unlike the former who had described himself as a procurator, Iulius Proculus is called praefectus,  .-./ . Thus he was a ‘militaire d’origine équestre de haut rang’.29 While the former was completely devoid of ius coercendi, so much that he had to address to the tribune of Dura in order to execute the orders he received from the governor of Syria, Iulius Proculus receives the petition from a bouleutes from Neapolis just so that he would intervene to compel the accused person to come in front of the court of the governor, i.e. Pomponius Laetianus,
%:  3 * &3  *  3 , who had already received the accusation by Aurelios Abidsautas, but who evidently requested to hear all parts in order to settle the controversy.

like the dux under Diocletian, he was an equestrian commander of garrison troops along a heavily fortified frontier. On the other hand, he was a 28 The ‘exploitation’ of the Roman citizenship by the inhabitants of the provinces has been well explained by Mazzarino 1973, in the case of St. Paul. See more generally Sherwin-White 19732. 29 Feissel & Gascou 1995: 102.

30

53

The graffito is cited by Gilliam 1941: 158.

THE LATE ROMAN ARMY IN THE EAST subordinate of the legate of Syria and commanded only a part of the troops in the province.31

l’armée romaine d’Orient et les raides sassanides du milieu du IIIe siècle. In CRAI: 213-241. Balty J.-C. 1988 Apamea in Syria in the second and third century A.D. In JRS 78: 91-104. Bastianini G. & Thomas J.D. 1977 PSI VII 797. In ZPE 27: 165-177. Bertrand-Dagenbach C. 1990 Alexandre Sévère et l’Histoire auguste. Collection Latomus 208, Bruxelles. Bollini M. 1966 Vibio Seneca. In Studi romagnoli 17: 229-233. Camplani A. & Gnoli T. 2001 Edessa e Roma. A proposito di un libro recente. In MedAnt 4: 61-80. D browa E. 1997 The rivers in the defensive system of Roman Syria (from Augustus to Septimius Severus). In Groenman-van Waateringe W. – van Beek B. L. – Willems W. J. H. – Wynia S. L. (ed.), Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. Oxbow Monographs 91, Oxford: 109-111. Drijvers H.J.W. 1977 Hatra, Palmyra und Edessa. Die Städte der syrisch-mesopotamischen Wüste in politischer und religionsgeschichtlicher Beleuchtung. In ANRW II/8: 799-906. Drijvers H. J. W. & Healey J. F. 1999 The Old Syriac inscriptions of Edessa and Osrhoene. Texts, translations and commentary. Handbuch der Orientalistik 1. Abt. Der Nahe und Mittlere Osten 24, Leiden-Boston-Köln. Feissel D. & Gascou J. 1995 Documents d’archives romains inédits du Moyen Euphrate (IIIe s. après J.C.). 1. Les pétitions (P. Euphr. 1 à 5). In JS: 65-119. Gawlikowski M. 1998 The Last Kings of Edessa, in R. Lavenant (ed.), Symposium Syriacum VII, Uppsala University, Department of Asian and African Languages 11-14 August 1996. Orientalia Christiana Analecta 256, Roma: 421-429. Gilliam J. F. 1941 The Dux Ripae at Dura. In TAPhA 72: 157-175. Gnoli G. 1989 The Idea of Iran: an essay on its origin. Serie orientale Roma 62, Roma. Gnoli T. 2000a Roma, Edessa e Palmira nel III sec. d.C. Problemi istituzionali. Uno studio sui Papiri dell’Eufrate. Monografie di Mediterraneo Antico 1, Pisa-Roma. Gnoli T. 2000b C. Furius Sabinius Aquila Timesitheus. In MedAnt 3: 261-308. Gnoli T. 2002 
  


and Edessa. In Panaino A. - Pettinato G. (ed.), Ideologies ad Intercultural Phenomena. Proceedings of the Third Annual Symposium of the Assyrian and Babylonian Intellectual Heritage Project Held in Chicago, october 27-31, 2000. MELAMMU Symposia. Milano: 79-87. Grenet F. 1988 Les Sassanides à Doura Europos (253 ap. J.-C.) réexamen du matériel épigraphique iranien du site. In Gatier P.-L. – Helly B. – Rey-Coquais J.-P. (ed.), Géographie historique au Proche-Orient (Syrie, Phénicie, Arabie, grecques, romaines, byzantines), Actes de la Table Ronde de Valbonne, 16-18

The analysis by Gilliam shows an official of the army of equestrian rank being in command of a military territory and of the troops located there. The figure of the dux ripae is therefore more analogous to the one of the praefectus praepositus praetenturae Aurelius Rufinus. The simpler administrative structure of the procuratorian province of Osrhoene in comparison to the bigger, more important and more articulated province of Syria had been able to anticipate someway the figure of the dux ripae creating a prefect who possessed a lot of attributions of it, although he preserved older titles.

)

The strip of land to the north of Dura had become a praetentura that was subjected to the civil administration of a procurator, sujected to the command of a vir militaris, an eques of high rank who had the command on the units located in the region. Contemporarily on the Khab r another vir militaris of the same rank had analogous charges. Thus it is possible that the difference in the titles in the region subjected to the command of the two officials, ripa and praetentura, should be simply attributed to the geographical configuration of the land with the steep slope chracterizing the western border of the Euphrates valley. (

Any way the different denomination might be explained, following the failed expedition of Severus Alexander or the disastrous campaign of Gordian III, Roman politics on the border along the Euphrates changed radically. The border was reorganized in a more defensive function, no more open paetentura, but a more and more fortified defensive line. The events would show it was already too late and that Rome should have restructured its defensive tools more profoundly. The fall of Duras Europos first, the destruction of Palmyra afterwards would compel Rome in this sense. Notwithstanding the idea by Luttwak of a Grand Strategy by the Roman empire still has got some adepts, I think that very little of it still resists.32 Rather than great theories or even worse, supposed epoch-making phases in the history of the border, it is time now to go definitively back to the study of the ‘semantics of the limes’. Bibliography Austin N.J.E. & Rankov N.B. 1995 Exploratio: military and political intelligence in the Roman world from the second punic war to the battle of Adrianople, LondonNew York. Babelon E. 1893 Numismatique d’Édesse en Mésopotamie. In Babelon E. (ed.), Mélanges numismatiques, Bd. 2. Paris: 209-296. Balty J.-C. 1987 Apamée (1986): nouvelles données sur 31 32

Gilliam 1941: 174. Luttwak 1976, about it see recently Whittaker 2004.

54

TOMMASO GNOLI: FROM PRAEPOSITUS PRETENTURAE TO DUX RIPAE Potter D. S. 1987 Alexander Severus and Ardashir. In Mesopotamia 22: 147-157. Reddé M. 1986 Mare nostrum. Les infrastructures, le dispositif et l’histoire de la marine militaire sous l’empire romain. BEFAR 260, Rome. Rösger A. 1978 Die Darstellung des Perserfeldzugs des Severus Alexander in der Historia Augusta, in 13. Bonner Historia-Augusta-Colloquium 1975/1976. Antiquitas Reihe 4. Beiträge zur Historia-AugustaForschung Band, Bonn: 167-174. Ross S.K. 1993 The Last King of Edessa: New Evidence from the Middle Euphrates. In ZPE 97: 187206. Ross S.K. 2001 Roman Edessa. Politics and Culture on the Eastern Fringes of the Roman Empire, 114-242 CE, London-New York. Rubin Z. 1998 The Roman Empire in the Res Gestae Divi Saporis – the Mediterranean World in S s nian propaganda. In Electrum 2: 177-185. Sartre M. 2001 D’Alexandre à Zénobie: histoire du Levant antique, IVe siècle avant J.-C.-IIIe siècle après J.-C., [Paris]. Saxer R. 1967 Untersuchungen zu den Vexillationen des Römischen Kaiserheeres von Augustus bis Diokletian. Beihefte der Bonner Jahrbücher / Landschaftsverband Rheinland Rheinisches Landesmuseum Bonn und Verein von Altertumsfreunden im Rheinlande, Köln-Graz. Sherwin-White A.N. 19732 The Roman Citizenship, Oxford 1973 (1939). Sommer M. 2003a Hatra — imperiale und regionale Herrschaft an der Steppengrenze. In Klio 85: 384-398. Sommer M. 2003b Hatra. Geschichte und Kultur einer Karawanenstadt im römisch-parthischen Mesopotamien, Mainz. Sommer M. 2005 Roms orientalische Steppengrenze. Palmyra – Edessa – Dura-Europos – Hatra. Eine Kulturgeschichte von Pompeius bis Diocletian. Oriens et Occidens 9, Stuttgart. Welles C.B. - Fink R.O. – Gilliam J.F. 1959 The parchments and papyri. The Excavations at DuraEuropos final report 5. Inscriptions, parchment, and papyri, Part 1, New Haven. Wheeler E.L. 1996 The laxity of Syrian legions. In Kennedy D.L. (ed.), 18. The Roman army in the East. Journal of Roman archaeology. Supplementary series, Ann Arbor (Mich.): 229-276. Whittaker C.R. 1970 Herodian in two volumes, II. Books V-VIII with an English translation by C. R. Whittaker. LCL 455, Cambridge (Mass.)-London. Whittaker C.R. 2004 Rome and its frontiers. The dynamics of Empire, New York. Wiesehöfer J. 1986 Iranische Ansprüche an Rom auf ehemals achaimenidische Territorien. In AMI 19: 177185.

septembre 1985. Notes et monographies techniques n° 23, Paris: 133-158. Gutschimd von A. 1887 Untersuchungen über die Geschichte des Königreichs Osroëne. Mémoires de l’Académie impériale des sciences de St.-Pétersbourg, VIIe série, St.-Pétersbourg. Kettenhofen E. 1984 Die Einforderung des Achämenidenerbes durch Ardašir: Eine Interpretatio Romana. In OLP 15: 177-190. Kettenhofen E. 2002 Die Einforderung der achaimenidischen Territorien durch di Sasanideneine Bilanz, in S. Kurz (ed.), Yadname-ye Iradj Khalifeh-Soltani. Festschrift Iradj Khalifeh-soltani zum 65. Geburtstag. Berichte aus der Orientalistik, Aachen: 49-75. Luther A. 1999a Die ersten Könige von Osrhoene. In Klio 81: 437-454. Luther A. 1999b Elias von Nisibis und die Chronologie der edessenischen Könige. In Klio 81: 180-198. Luther A. 2002 Römische Militärposten der Severerzeit am Unteren 
b r. In Göttinger Forum für Altertumswissenschaft 5: 1-9. Luther A. 2004 Dura-Europos zwischen Palmyra und den Parthern. Der politischen Status der Region am mittleren Euphrat im 2. Jh. n. Chr. und die Organisation des palmyrenischen Fernhandels. In Rollinger R. – Ulf C. (ed.), Commerce and Monetary Systems in the Ancient World: Means of Transmission and Cultural Interaction. Proceedings of the Fifth Annual Symposium of the Assyrian and Babylonian Intellectual Heritage Project Held in Innsbruck, Austria, October 3rd-8th 2002, Melammu Symposia V. Oriens et Occidens 6, Stuttgart: 327-351. Luttwak E.N. 1976 The grand strategy of the Roman Empire from the first century A.D. to the third, Baltimore 1976, trad. it. Milano 1991. Maricq A. 1955 Hatra de Sanatrouq. In Syria 32: 273288 (= Id., Classica et Orientalia, cit., 3-7). Maricq A. 1957 Les dernières années d’Hatra: L’alliance romaine. In Syria 34: 288-305. Mazzarino S. 1973 L’Impero romano, Roma-Bari. Millar F. 1982 Emperors, Frontiers and Foreign Relations 31 BC-AD 378. In Britannia 13: 1-23. Millar F. 1990 The Roman Coloniae of the Near East: a Study of Cultural Relations. In Solin H. - M. Kajava (ed.), Roman Eastern Policy and Other Studies in Roman History, Proceedings of a Colloquium at Tvärminne, 2-3 October 1987, Commentationes Humanarum Litterarum 91, Helsinki: 7-58. Millar F. 1993 The Roman Near East, 31 B.C.-A.D. 337, Cambridge, Mass. Nasti F. 1997 Il prefetto del pretorio di CIL VI 1638 (= D. 1331) e la sua carriera. In ZPE 117: 281-290. Pflaum H.-G. 1960 Les carrières procuratoriennes équestres sous le haut-empire romain I-III. Bibliothèque archéologique et historique 57, Paris. Pflaum H.-G. 1967 Vibius Seneca, dux vexillationum classis praetoriae Misenatium et Ravennensium. In Studi romagnoli 18: 255-257. 55

The Roman Legionary Camp at Legio, Israel: Results of an Archaeological Survey and Observations on the Roman Military Presence at the Site Yotam Tepper Israel Antiquities Authority (IAA)

Qeni (Fig. 1). Its name preserved the memory of the legionary camp of the Roman period up to modern times.1 This paper focuses on the results of an archaeological survey carried out in the vicinity of Legio, supported by the results of archaeological excavations.2 The paper will present the results of the survey and our conclusions with respect to the location of the legionary camp in this well-defined geographical area that included a Jewish village, a Roman military camp and a RomanByzantine city.3

Abstract The survey of the Legio area, conducted on behalf of Tel Aviv University in 1998-2000, recovered numerous finds related to settlement sites in the region, and enabled us to suggest the locations of a Jewish village, a Roman military camp and a Roman-Byzantine city in a welldefined geographical area, characterized by an abundant water source throughout the year and an important road junction in the northern Land of Israel. The paper presents the finds of the survey, focusing on the Roman military camp at Legio. The location of the camp was determined with the assistance of the survey finds and analysis of the routes of Roman roads and aqueducts, and the distribution of cremation burials. The size of the camp was ascertained on the basis of old maps, aerial photographs and the ceramic finds of the survey. The paper takes into account the results of other surveys and excavations conducted in the area and the discovery of roof tiles with stamps of the legions stationed at the camp site. The summary presents preliminary conclusions, within the limitations of the study, relating to the Roman military presence at Legio.

History of research Although historical-geographic research in the early twentieth century was aware of the existence of Kefar ‘Othnai, a legionary camp and a Roman-Byzantine city in the area of Legio, their precise location was not

1 For the geographical background of the site, a historical overview of the sites in the vicinity of Legio and preliminary results of the survey, see Tepper 2002: 231-242; 2003; 2003a: 29*–31*. For the historical sources on the sites of the region, see also Tsafrir, Di Segni and Green 1994: 170. 2 The archaeological survey at Legio was conducted on behalf of the Department of Classical Studies of Tel Aviv University, with the support of the Institute of Archaeology and the Megiddo excavation expedition of Tel Aviv University. I wish to thank the dedicated survey team, Y. Tepper, A. Caspi, A. Blumenkrantz and M. Shomroni, as well as Profs. B. Isaac, I. Roll, I. Finkelstein, D. Ussishkin, I. Shatzman and M. Gichon, Dr. Y. Shahar, D. Avshalom-Gorni and G. Stiebel for their assistance throughout the project and their constructive comments. I am grateful to Prof. A. Segal, Dr. T. Tsuk and H. Abu Uqsa for permission to use data from their excavations at Legio, and to the Israel Antiquities Authority (IAA) for permission to use archival data and publish them here, as well as preliminary data from excavations and surveys conducted by the IAA in the vicinity of Legio and within Megiddo Prison. My thanks to the committed staff of the Institute of Archaeology of Tel Aviv University and my colleagues at the IAA. Finally, I am most grateful to Prof. A. Lewine, who encouraged me to complete the writing of this study and assisted in publishing it. 3 This study was written after the completion of a comprehensive archaeological survey of the site and the preliminary publication of its results (Tepper 2003) and during the extensive excavations carried out under my direction on behalf of the IAA within Megiddo Prison (20032005). The abundant finds of the survey include architectural finds, inscriptions, roof tiles and small finds including coins and items of military equipment; not all of them can be discussed within the scope of the present article. The above-mentioned excavations will be published in full in the framework of the publications of the IAA. It should be emphasized here that the excavations within the prison exposed the remains of a large Jewish village. On the village’s western margin we uncovered a large residential building that also served as Roman officers’ quarters; in one of its wings was an Early Christian prayer hall (Tepper and Di Segni, in press).

Introduction In the Early Roman period, the village of Kefar ‘Othnai was a located on a ridge south of Nahal Qeni, which runs into the Jezreel Valley south of Tel Megiddo. The site is situated between Samaria and Galilee, and between the coastal plain and the northern valleys of Israel. Its location at a geographical meeting point where there are abundant water sources and next to a junction of ancient roads made this an important site in the Roman period. In the first half of the second century CE, a small Roman fort was erected in a commanding strategic position near the village, and the permanent camp of the Legion VI Ferrata was established on a hill northwest of Nahal Qeni. A network of Roman imperial roads runs to and from the site, and the place is mentioned in historical sources. The site was originally referred to by its Hebrew name, Caparcotani, and later by the name of Legio. During the Roman period, the Jewish village increased in size and incorporated a Samaritan population as well, becoming a large civilian settlement alongside the legionary camp. Apparently in the late third or early fourth century, after the abandonment of the legionary headquarters, the city of Maximianopolis was founded south and west of Tel Megiddo. After the Muslim conquest a village named Lejjun was built on the ruins of the city, spreading all over the southern slopes of Nahal 57

THE LATE ROMAN ARMY IN THE EAST

Fig. 1: General location map showing the sites (no. 1: the Roman army camp and the civilian village; no. 2: the Roman fort; no. 3: the Jewish village) and the Roman roads from Legio. m), with walls built of large field stones and roofed entrance gates on the east and west. Remains of earthen ramparts were preserved on the north and southwest of the fort (Fig. 2). Although the topographical and spatial limitations of the area dictate that the fort could not have housed a Roman military unit larger than a small auxilia,6 many scholars accepted Schumacher’s assessment that

established by the numerous scholars who visited and studied the region.4 Nevertheless, as early as 1908 Schumacher, the first excavator and researcher of Tel Megiddo, identified a Roman military fort on the summit of the ridge south of Nahal Qeni and proposed that this was the legionary base.5 Schumacher exposed the rectangular fort (35 x 55 4

6

Thomsen 1907: 83-85; McElderry 1908: 10-13; Hartman 1910: 169188; Nelson 1913: Dalman 1914: 34-37; Ramsay 1916: 129-131; Ritterling 1925: 1587-1596; Reeg 1989: 361-363. 5 Schumacher (1908: 176, Tafel 1) noted the presence here of a legionary camp, but did not identify it explicitly. Neeman (personal communication) recorded remains of walls and terraces in the same place, as did Illan (1988: 71), who described thick walls at least 37 m long.

The ridge south of Nahal Qeni on which Schumacher identified the fort is currently covered by a forest of pine and cypress, hindering the work of archaeological survey. Few potsherds were collected at the site, only about 40 of which could be identified and dated. Among these, 70% (including lamps) are of the Roman period and the remainder of the Byzantine and Early Muslim periods. This appears to support the dating of the fort described by Schumacher (1908: 188-190) to the Roman period; see Tepper 2003: 160, Fig. 91).

58

YOTAM TEPPER: THE ROMAN LEGIONARY CAMP AT LEGIO, ISRAEL

Fig. 2: Schumacher’s map (1908) of the fort on the summit of the ridge south of Nahal Qeni.

59

THE LATE ROMAN ARMY IN THE EAST

Fig 3: Schumacher’s map (1908), see the aqueduct of ‘Ein el-Qobbi, the upper aqueduct, west to el-Menach Hill and the lower aqueduct east to the hill. this structure headquarters.7

on

the

ridge

was

the

Isaac and Roll9 surveyed the Roman road from Legio to Scythopolis (Beth-Shean), demonstrating by measurement of the milestone stations along the road that it originated on this hill. The “Hill of the Campground,” or el-Manach, as it is called in transcription from Arabic in Mandatory period maps,10 is located northwest of the modern Megiddo Junction and at a junction of dirt roads and footpaths on Schumacher’s map11 (Fig. 3). The traditional Arabic name (the “place of encampment” or “campground”) used for the hill by the inhabitants of the

legionary

On the other hand, from the second half of the twentieth century various scholars have pointed to a hill north of Nahal Qeni as the headquarters of the Legion VI Ferrata.8 7 Thus, for instance, Safrai (1992: 104-105), who entirely overlooks these limitations and states that the central structure described by Schumacher as a fort was the principia of a large military camp whose boundaries were the terrace walls. As a result of earthworks carried out in the area, mostly during the First and Second World Wars, and subsequent forestation, most of the remains on the ridge have been destroyed, and today it is difficult to trace them in the field. 8 Preliminary testing of the area was carried out by Prof. M. Gichon of Tel Aviv University with the intention of conducting archaeological excavations on the hill. The finds included roof tiles with legionary stamps.

9

Isaac and Roll 1982: 34-35, 86; see also Tsuk 1988-9: 92-97; Raban 1999: 104-105, Site 167. 10 Palestina, 1: 20000. Megiddo, S.S: 16-22. 1942. 11 Schumacher 1908: Tafel. Karte des tell el-Mutesellim und elLeddescǀn.

60

YOTAM TEPPER: THE ROMAN LEGIONARY CAMP AT LEGIO, ISRAEL

Fig. 4: The hill of el-Manach in Schumacher’s survey map (1908). In the northern part of the hill is a square enclosure with earthen ramparts on the north and east.

61

THE LATE ROMAN ARMY IN THE EAST the Second Legion was transferred from Egypt after 127 CE.19 From the historical sources it is difficult to establish when the site was abandoned, and the issue is still under debate.20 It has been claimed that the Legion VI Ferrata left Legio for Alexandria during the reign of Alexander Severus21 or during the third century,22 while a fourthcentury date has also been proposed.23 Diocletian (284-305 CE), who instituted changes in the central military command structure and administration, visited the Land of Israel in 286 CE and transferred the Legion X Fretensis from Jerusalem to Aila (Aqaba). Military units consisting of cavalry, archers and camel riders were posted on his command to the lands of the new province and along the eastern limes. He transferred the Fourth Legion to el-Lejjun and the Third Legion to Bostra in Transjordan, and there is no apparent evidence for the presence of the Legion VI Ferrata in the province after the late third century.24 Indeed, the Sixth Legion is absent from the Notitia Dignitatum, and it does not figure in the deployment of Roman military forces in the Land of Israel or in the changes that took place in the province from the fourth century onwards.25

Fig. 5: Aerial photograph (P.S. 680, No. 50680) of the hill of el-Manach, in which the Camp enclosure is visible (H) (courtesy Gichon Collection). area in the early twentieth century,12 as well as its location at a junction of roads and footpaths, support the assumption of researchers of Roman roads that the legionary headquarters was sited on this hill.

The Legionary Camp in maps and aerial photographs

On the basis of historical sources, various scholars have claimed that a second Legion was sent to Judaea in the early part of the reign of Hadrian (117-138 CE),13 or a little earlier, at the end of the reign of Trajan (98-117 CE),14 to reinforce the Legion X Fretensis stationed in Jerusalem. In any case, the stationing of the second Legion in the Land of Israel reflects to Provincia Judaea’s change in status to a consular province.15 Cotton and Eck demonstrated through Latin inscriptions found at Caesarea that the change in status to a consular province did in fact take place during Trajan’s reign, and thus that the second Legion arrived in the province at this time.16 Some scholars have claimed that it was the Legion VI Ferrata that was transferred from Arabia and made its permanent camp at Kefar ‘Othnai.17 On the basis of a milestone inscription, Isaac and Roll proposed that the first Legion that arrived at the site wae the Legion II Traiana, perhaps as early as 117 CE, and that the Legion VI Ferrata was posted there shortly afterwards.18 On the other hand, Cotton believes that the Sixth Legion arrived at the site from Gerasa between 119 and 127 CE, and that

In Schumacher’s map26 the hill of el-Manach is marked as a large triangle, with its apex in the north and its base in the south, bordering Nahal Qeni (Fig. 4). The area of the hill is some 420 dunams (42 hectares) and it is surrounded by dirt roads.27 In the northeast corner of the hill Schumacher marked an extensive area of antiquities bordered on the east, north and northwest by an earthen rampart. A local watercourse has apparently cut into the slope of the hill and eroded the east rampart and part of the west one. The dimensions of the enclosure on the summit of the hill that appears on Schumacher’s map is 250 x 250 m. A square enclosure, including the rampart, of similar dimensions and position is clearly visible on an aerial photograph (Fig. 5).28 During 1945-1946 a large

19

Cotton 2000: 351-357. Cotton 2000 contra Ritterling 1925: 1593. Hirschfeld 1991: 182. 22 Isaac 1990: 433. 23 Safrai 1992: 104 24 Tsafrir 1984: 359-365. 25 Notitia Dignitatum, Ed. Seeck: 72-73; Tsafrir 1984: 362-371. 26 Schumacher 1908: Tafel 1. 27 Schumacher (1908: Tafel 1. Abb. 234-235) describes on the south of the hill a wall 1 m wide (an outer wall?), with a row of sarcophagi along its outer face. This burial ground was located in the survey (Tepper 2003: 109, Fig. 41). The possibility that this was the southern outer wall of a legionary camp covering a large area on the border of the hill of elManach was examined, but in the absence of adequate archaeological data and in view of the evidence of maps and aerial photographs, this possibility was rejected. 28 The photograph (P.S. 680, No. 5068) was taken on December 16th, 1946 (Gichon Collection). Aerial photographs are a standard and valuable tool in research of this kind; see, for example, Fabian 1995: 235-240; Kennedy 2002: 99-110. 20 21

12

Sharoni 1987: 1266. Avi-Yonah 1973: 209-213; Keppie 1973: 859-864. 14 Eck 1984: 55-67; Pflaum 1969: 232-233. 15 In this context, some scholars (e.g. Lifshitz 1960: 109-111) have claimed that the arrival of the Sixth Legion at Legio was a Roman response to the Bar Kochba Revolt, although this proposal has been rejected; see Tsafrir 1984: 350; Oppenheimer 1991: 32. 16 Cotton and Eck 2001: 219-223. 17 Lifshitz 1960: 109-111; Pflaum 1969: 232-233; Avi-Yonah 1973: 209-213; Keppie 1973: 859-864; Dąbrowa 1996: 285. 18 Isaac and Roll 1982: 9; Ibid. 1979: 54-66; 1979a: 149-156; 1998: 198-207; 1998a: 208-210; Rea 1980: 220-221. 13

62

YOTAM TEPPER: THE ROMAN LEGIONARY CAMP AT LEGIO, ISRAEL part of the hill was occupied by a British military camp. Since the 1950s the hill was intensively cultivated, and no architectural remains, apart from the upper west rampart, were found.

built, using Roman technology and engineering skill, between urban or military points of origin (capia viarum).36 Abundant data on the builders and date of the roads are supplied by the inscriptions on the milestones that were erected along them.37

The legionary camp at Legio has never been excavated. However, its area may be estimated on the basis of the results of the survey, and especially the aerial photographs and other measurements, as about 60 dunams (6 hectares). It was thus similar in area to other Roman military camps in this part of the Empire.

The foundation of the legionary camp at Legio determined a new point of origin for the network of Roman roads in the northern Land of Israel, and traditional transport routes were diverted to the site of the camp as a measuring point of distances along the roads. This was the case, for instance, for the road from Caesarea to Pella in Transjordan. The road was built by troops of the Legion X Fretensis in 69 CE, and before the foundation of the camp at Legio its route ran along Nahal Tanninim (Crocodilon Flumen) past Gabae (Gaba Philippi; Tell Abu Shusha) to Scythopolis (Beth-Shean).38 Schumacher discovered a milestone (?) on the eastern margin of Tel Megiddo, which he suggested was the stone from which the Roman roads in the northern Land of Israel were measured.39 Although the point of origin for Roman roads has not been located at Legio, epigraphic evidence has been preserved on milestones on roads leading to and from the camp. Two milestones mentioning the camp at Legio were found along the road from Legio to Scythopolis. On the milestone from the reign of Hadrian (129 CE) the camp is termed “Castron,” while on another stone from the reign of Caracalla, at mile station No. 22 from Scythopolis and mile station No. 2 from Legio, the camp is mentioned as the origin of the road.40 Along the road from Legio to Diocaesarea (Sepphoris), a milestone from the reign of Hadrian was found; it gives the distance from Legio and cites the Legion II Traiana as the road-builders.41 Latin inscriptions giving the distance from Legio are found on Hadrianic milestones at mile vi42 and mile v43 from Legio. Although measurement of distance from the legionary camp at Legio is attested only for the reign of Hadrian, military camps apparently continued to serve as points of origin to measure distance along the roads until the reign of Alexander Severus, after which this role was taken over by the six major cities of the province.44

The camp of the Tenth Legion in Jerusalem, according to a schematic plan of its first stage after the Great Revolt (67 CE), had estimated dimensions of 150 x 250 m, an area of only 37.5 dunams (3.75 hectares). In its second stage up to 285 CE the estimated area of the camp adjacent to the Temple Mount was 31-35 dunams.29 The legionary camp at Tel Shalem, where an inscription of a vexillatio of the Legion VI Ferata was found, is estimated as measuring 180-210 m, or 38 dunams (3.8 hectares).30 In the days of Diocletian the camp at el-Lejjun in Transjordan measured 190 x 240 m or 46.5 dunams (4.65 hectares), judging by the line of its excavated walls.31 Consequently, it seems that Roman military camps in Provincia Judaea32 were smaller than those of Western Europe, where the average area of camps was about 1824 hectares (180-240 dunams).33 It is likely, therefore, that in Judaea the Roman army deployed groups smaller than a full legion.34 Given an average area ratio of 1:4 or 1:3, the legionary camp at Legio could have held only a quarter or a third of a full legion. Thus, although legionaries’ quarters were not excavated or identified, we may estimate that the camp accommodated 2000-2500 legionaries at most.35 The Roman Roads to Legio The building of imperial roads played an important role in the Roman administration. The network of these roads enabled movement of troops in all seasons and efficient communication between institutions of administration and government in times of war and peace, as well as civilian and commercial transport. The roads had an official legal status (via publica) and served both the imperial post and the holders of administrative and military appointments (cursus publicus). The roads were

36

Roll 1994: 21-22; Chevallier 1997: 274-280. Tsafrir 1984: 51. 38 For its route east of Legio, see Isaac and Roll 1982; Hecker 1961: 175-186; for its route on the Manasseh Heights, see Schumacher 1903: 4-10. 39 Schumacher 1908: 8, 169, Tafel 1. The stone is marked on the map but has never been rediscovered. 40 Isaac and Roll 1982: 79-80, 86. 41 Isaac and Roll 1982: 149-156; Hecker 1961; see also note 18 above with references cited there. 42 Avi-Yonah 1946: 97, No. 1. 43 Hecker 1961: 175-186; Lifshitz 1960: 109-111. 44 Isaac proposed that miles were measured from the legionary camp at Legio only during the reign of Hadrian, and not later. However, in other places measurement of distance from military camps is known until the reign of Alexander Severus, for instance on a milestone near Tel Shalem, where another unit of the Sixth Legion was stationed; the inscription gives the number of miles from the legionary camp rather than from Scythopolis (Isaac 1998: 63-66). 37

29

Stiebel 1999: 68-103 and bibliography cited there. Tzori 1950: 53-54; Foerster 1985:139-140. It seems likely that the camp accommodated only about one thousand legionaries. 31 Parker 1987. 32 For more examples see, among others, Hashman 1996: 19-79 (Suweida); Fabian 1995: 235-240 (Beer-Sheba). See also Stiebel 1999: 89-90. 33 See examples in Webster 1969: 187-182: Inchtuthil, 781 x 813 m, ca. 635 dunams; Novaesium, 450 x 600 m, ca. 270 dunams; Caerleon, 425 x 500 m, ca. 212 dunams; Haltern, 388 x 500 m, ca. 194 dunams. For additional examples, see Bohec 1996: 100-111. 34 Tsafrir 1984: 41-47; see also Hirschfeld 1991: 170-183. 35 Shatzman 1983: 266; Coello 1996: 1-2. 30

63

THE LATE ROMAN ARMY IN THE EAST In view of the above, we can identify four stages of Roman road-building in the region of Legio. In the first stage, the road from Caesarea to Scythopolis and on to Pella and Gerasa was built in 69 CE by troops of the Legion X Fretensis; it passed through Gabae rather than Legio. In the second stage, road-building was carried out in the region in 120 CE after the stationing of the Legion II Traiana at Legio. In the third stage, intensive roadbuilding was associated with Hadrian, particularly in advance of his visit to the East in 139-140 CE. At that time the legionary camp at Legio became the center of Roman road-building in the northern Land of Israel, apparently in connection with the stationing of the Legion VI Ferrata at Legio. In the fourth stage, additional roads were built and the importance of the site in the network of Roman roads in the region increased, reaching a peak in the third and fourth centuries; six imperial roads were built to and from the legionary camp, connecting it with major destinations in the northern Land of Israel.45

running south until it forks into two aqueducts (Figs. 3 and 6).

Fig 6: Detail of the aqueduct of ‘Ein Qobbi running south from the spring (Schumacher 1908: Abb. 233).

The Aqueducts leading to Legio It is well known that a Roman military camp required an abundant supply of fresh water for drinking, cooking, hygiene, watering animals, and other purposes.46 Such supplies of water were available in large quantities from Nahal Qeni and its springs near the legionary camp. This fact alone, can explain the situation of the camp at Legio.47 The camp site is located on a low hill in the heart of the hilly area south of Nahal Qeni, where its course from the mountains west of the Jezreel Valley makes a wide bend southwards. This particular situation required water to be conducted to the camp site by aqueducts, whose direction and course contribute to our efforts to locate of the site.

The upper and more westerly of the two (hereafter the “upper aqueduct”) passed along the west side of the hill of el-Manach in a southwesterly direction; 160 m of its length were exposed. The other aqueduct (henceforth the “lower aqueduct”), which passed along the east side of the hill in a southeasterly direction, was preserved for a length of ca. 700 m. The Chicago expedition that excavated at Tel Megiddo exposed 165 m of the “upper aqueduct.”50 Tsuk excavated three sections of this aqueduct, with a total length of ca. 20 m, at the foot of the tel and north of the hill of el-Manach. The aqueduct is both built and rockhewn; with a width of 26-34 cm and a depth of 50 cm. Deposits of travertine are 4 cm thick on the walls and 1 cm thick on the base. The roofing slabs have not survived. Below the travertine was a layer of pink plaster, and below that a layer of gray plaster of a type that does not predate the late second century CE. The foundation of the channel, 15 cm thick, consists of crushed chalk, stone fragments, potsherds and shells.51

Schumacher48 described an aqueduct at Legio49 that conducted water southward from ‘Ein Qobbi, at the foot of Tel Megiddo, towards el-Manach hill. It was exposed about 350 m from the spring for a length of about 100 m,

45

In this context the testimony of Eusebius (Onomasticon, Ed. Klostermann 1904, passim) is significant. He identifies the names of settlements of his period with biblical sites and locates settlements in Galilee by their distance in miles from Legio. Even if we postulate an early date (ca. 300 CE) for this source, and although a Roman military unit is not mentioned at the site named Legio, we have here evidence that the site was an important junction in the road system of the northern Land of Israel (Isaac and Roll 1982: 11; Isaac 1996: 153-167). 46 Von Petrikovits 1975: 105. See Fig 1, above. 47 The output of the springs of Nahal Qeni, as measured in 1970, is 1.53 million cubic meters of water per annum (Tepper 2003: 12-13, 104105). This alone, without counting the output of other springs near the site, exceeds the quantity carried by the urban water supply systems of Diocaesarea (Sepphoris) to the city (Tsuk 1985: 45). 48 Schumacher’s map (Schumacher 1908: Tafel 1) shows several aqueducts (see below). He describes the cascade, including the aqueducts, of the flour mills that functioned at the time of his visit to the site (Schumacher 1908: 166-167). They will not be discussed in detail here (Tepper 2003: 100-104, Figs. 34-39). 49 For the built aqueduct, see Schumacher 1908: 161-162, Abb. 233; its continuation, rock-hewn with roofing slabs, passes the foot of the tel in the burial ground (Schumacher 1908: 168-169, Abb. 250).

To the northwest of el-Manach hill, Schumacher described the pool of “el-Chuwcha,” its western part rock-hewn and its eastern part built of large ashlars. It was rectangular and measured 30 x 90 m.52 It seems likely that this pool received the water from the “upper aqueduct.” In aqueducts of the Roman period, a settling pool for the separation of refuse (piscina) and a distribution pool (castellum) were generally built at the

50 The aqueduct was rock-hewn with roofing slabs; see Lamon 1935: 1, Fig. 1; Guy 1938: Pl. 1. 51 Tsuk 1988-9: 92-97. Today only about 30 m of the aqueduct are exposed at the foot of the tel. For the dating of the plaster of aqueducts, see Porath 1984: 1-16. 52 Schumacher 1908:170, Tafel 1.

64

YOTAM TEPPER: THE ROMAN LEGIONARY CAMP AT LEGIO, ISRAEL boundaries.60 Schumacher identified extensive burial grounds on the hills south of Tel Megiddo, as well as on the margins of el-Manach hill.61 In view of the results of the survey, it is possible to classify the burials in the necropolis of Legio into two main types: cremation62 and inhumation (primary and secondary). Since cremation was common in the Roman legions in the first and second centuries CE, the identification of cremation burials is significant for the location of the legionary camp, on the principle that the presence of legionary burials in a particular place rules it out as the location of the camp.63 Near the Roman fort Schumacher described a rock-hewn tomb. A narrow entrance leads to a square burial chamber containing two ceramic coffins. In one of the walls of the chamber are hewn niches containing urns with a narrow foot; they held remains of human bones.64 West of elManach hill, on the eastern slopes of the hill on which Kibbutz Megiddo is situated, Tsuk excavated a heap of burnt material; in it were two cooking pots that held bones of a young man (Fig. 7).65 Further up the hill, four more cooking pots containing human bones have been found over the years.66 The presence of cremation burials north to the Roman fort and on the slops of Kibbutz Megiddo hill, west and south of el-Manach hill, supports the assumption that the camp was located on this hill.67

point of entry to centers of population; it is probable that the pool at Legio fulfilled one of these functions.53 The members of the British survey team located a large structure containing two columns above Nahal Qeni. They described an arched entrance on its south side leading to a hewn tunnel about 20 m long, partly built and with water flowing in its base.54 Schumacher noted its sanctity and importance to the inhabitants of the village and described the structure from the outside as having a gate from which water flowed; the place was known as ‘Ein el-Sit (‘Enot Qeni).55 Although the structure over the aqueduct and spring was apparently sanctified in the Muslim period,56 the form and dimensions of the stones of the aqueduct attest to an earlier date, apparently the Roman period.57 Near the course of Nahal Qeni, south of el-Manach hill, several aqueducts were located, some stone-built and some constructed from sections of ceramic pipes; they attest to additional exploitation of the water of Nahal Qeni.58 The course and direction of the aqueducts indicate that they terminated at el-Manach hill, the site of the legionary camp. The Necropolis of Legio It is well established that methods of burial and the accompanying gifts attest to the social, cultural and sometimes ethnic affiliations of the deceased.59 Consequently, the distribution of burial types in a region sometimes enables analysis of social processes and changes, as well as the study of ethnic and/or historical

60 For examples from Jerusalem, see Kloner 2002: 93-100; Avni 2002: 101-110. In the border of Gallile and Phoenicia see the different between the burials at Stern and Getzov 2006: 91-123. 61 On Schumacher’s map extensive burial grounds are marked on the slopes of the tel and on the hills to its south and west (Schumacher 1908: Tafel 1). Numerous caves and tombs were described by Guy (1938: 131-138), Lamon and Shipton (1939: 92-97) and Raban (1991: Site 161). Many more burials were identified during the survey of the region of Legio (Tepper 2003: 105-115, Figs. 40-50). 62 Cremation, i.e. the burning of the deceased and burial of the ashes in an urn, was common in the Greek, Roman and Etruscan cultures (Webster 1969: 27; Toynbee 1971: 101). 63 In the Land of Israel cremation burial is found in sites with a Roman military presence, always outside the area of the camp, for instance at Masada (Tsafrir 1984: 143. Note 124), Jerusalem (Kloner 2002: 95-98) and Mamshit (Negev 1971: 124-125). The burials are dated to the first and second centuries. It is well known that during the second century inhumation burial became more popular among legionaries (Webster 1969: 280-281), although at Caesarea cremation burial continues into the third century (Porath 2000: 34). 64 Schumacher 1908: 189-190, Abb: 256a, 287, 289, 291-292. For an additional find of a burial jar containing human bones and Roman artifacts near el-Manach hill, see Tepper 2003: 107. 65 Tsuk 1988-9: 92-97. 66 Archive of Kibbutz Megiddo, IAA Nos. 98-5286; 98-5288. I am most grateful to S. Gillis, reported these finds. See also Tepper 2003: 107108, Fig. 40. 67 In excavations conducted by the IAA on the southwest slope of the Megiddo jail, an additional burial was discovered; my thanks to the IAA anthropologist, Y. Nagar.

53

Reservoirs for the regulation and storage of water are known at the end of the aqueduct at Sepphoris (Tsuk 1989: 104-107), Tiberias (Winogradov 1989: 129-139), Kefar Hanania (Illan 1989: 99-100) and other sites (Amit, Hirschfeld and Patrich 1989). At the place of the ancient pool at Legio only a few large stones survive today, and Kibbutz Megiddo has installed a sewage pool there. 54 Conder and Kitchener 1882: 64-65. 55 Schumacher 1908: 185, Abb 282. Illan examined the structure in 1984. It contained three rooms with openings between them. The walls were plastered, some of them with remains of red decoration. A breach in the western wall led to a hewn tunnel 25 m long, in the base of which was a built aqueduct 20-50 cm wide. Illan assumed that this is the maqam of Ibrahim el-Halil mentioned in Muslim sources, and that the tunnel is earlier in date (Illan 1988: 63-65). 56 Strange 1965: 492-493. 57 The tunnel is 50-70 m wide and ca. 2-2.5 m high. Its upper part is hewn and roofed by diagonally laid rock slabs. The lower part is built of identical stones 60 cm long, laid on their narrow ends to form uniform insets and offsets. On them is gray plaster with diagonal combing. Between the stones the aqueduct is 35 cm deep and 16 cm deep; its total length, in a west-east direction, is 16 m up to the point of flowing (Tepper 2003: 95; 2003a: 29*). A tunnel spring hewn and built in an identical technique and dated to the Herodian period was exposed at ‘En Tsur (Hirschfeld 2000: 301-306); other examples found in the Judean hills include that at Beth Ther, where a Roman inscription was discovered (Ron 1977: 230-250). 58 For the ceramic pipes and other data on the aqueducts, see Tepper 2002: 234, Fig. 10; 2003: 95-99, Fig. 29-30. 59 Toynbee 1971; Tsafrir 1984: 141.

65

THE LATE ROMAN ARMY IN THE EAST Archaeological Excavations and Survey in the Site of the Camp and its Vicinity Salvage excavations conducted on the eastern slopes of el-Manach hill exposed remains of structures with floors of earth and mud, and of paved surfaces. The finds included pottery of the second to fourth centuries CE, roof tiles, ceramic pipes, and Roman coins dating from the first to late third centuries CE.68 An additional excavation on the southern slopes unearthed building remains of a large complex. The excavation exposed a doorpost of an impressive entrance and three building strata of the Roman period, including water supply installations, dated to the second to fourth centuries CE.69 Among the finds of the excavation were roof tiles with stamps of the Legion VI Ferrata and the Legion II Traiana (Fig. 8). Petrographic analysis demonstrated that the tiles were made from local clay, convincing evidence for a legionary workshop at the site.70 These tiles join a larger collection of stamped legionary roof tiles of the Legion VI Ferrata, all collected in the survey of the Legio area or in private collections (Fig. 9).71

Fig. 7: Cremation burials from the slopes of Kibbutz Megiddo (courtesy T. Tsuk).

Of the hundreds of potsherds collected during the survey on the hill of el-Manach, 346 comprised a datable assemblage. Most of the sherds (more than 70%) date from the Roman period (Fig. 10). The remaining sherds are divided as follows: 4.6% from the Persian-Hellenistic periods, ca. 12.5% from the Byzantine period, ca. 10.5% from the Early Muslim period, and the rest from the Mamluk period. Of the finds from the Roman period, the minority dates from the Early Roman period, while the majority dates from the second to third centuries CE. It is interesting that more than 65% of the ceramic finds of the Roman period derived from the southeastern slope of the hill of elManach, a fact that demonstrates the intensive nature of settlement in the Roman period on the hill that we identify as the site of the legionary camp of Legio.

Fig. 8: A roof tile of the Legion II Traiana (courtesy H. Abu Uqsa; photography by H. Smithline).

The finds of the archaeological survey and the salvage excavations on the slope of the hill, and the distribution of Roman pottery over an area of hundreds of dunams on the southern and particularly the eastern part of the hill, apparently testify to a scattered Roman settlement that existed side by side with the legionary camp. This was probably one of the civilian settlements that developed 68 Segal 1999: 48-49. In the excavator’s assessment, this was the entrance to the camp or the paved square of the principia in its center. He dated the assemblage to a late stage in the history of the camp. 69 The excavations, which are not yet published, were directed by H. Abu Uqsa. 70 I am grateful to H. Abu Uqsa for permission to publish this preliminary information on the excavation. The petrographic analysis was carried out by A. Shapiro of the IAA. 71 The tiles are in the collections of Prof. M. Gichon, A. Drori and D. Rubens. All were subjected to petrographic analysis by Prof. Y. Goren of Tel Aviv University and proved to be manufactured from local clay, evidence for a legionary workshop at the site. See also Schumacher 1908: 175; Tepper 2002: Fig. 9; 2003: 64-68, Figs. 11-1 2.

Fig. 9: A roof tile of the Legion VI Ferrata (Tepper 2003). 66

YOTAM TEPPER: THE ROMAN LEGIONARY CAMP AT LEGIO, ISRAEL

Fig. 10: Hellenistic and Roman potsherds from the hill of el-Manach. Hellenistic period: 1. Casserole; 2. Jug; 3. Imported ETS bowl. Roman period: 4-5. “Galilean bowls”; 6-7. Bowls; 8: Mortarium; 9-10. “Kefar Hananya” cooking pots; 11. Juglet; 12. Amphora; 13-15. Storage jars; 16-17. Lamps

67

THE LATE ROMAN ARMY IN THE EAST alongside Roman camps, which were known as vicus when attached to an auxilia or small unit and canabae when attached to a legionary camp.72 The thousands of Roman soldiers who were stationed in a legionary camp represented a large and well-paid consumer force. Consequently, beside each and every military camp sprouted a civilian settlement whose inhabitants supplied the soldiers’ needs on a commercial basis. In these settlements were residential structures and other services, and their status and economic power were dependent on the troops stationed in the vicinity. This was the case in the western and eastern Roman Empire and most likely in Judaea as well.73

examination of the links between Roman military camps and civilian settlements in the eastern Roman Empire. In addition to this, however, the location of the legionary camp on the border of the Jezreel Valley is significant in itself. The Jezreel Valley or “Great Valley” (Mega Pedion), which was defined as the property of the legion (Campus Maximus), contained lands that had long been considered royal estates.75 The special status of these rich agricultural lands and the possibility of exploiting them as an economic and agricultural base for the legion comprised an additional motive for the location of the camp. Indeed, land division (limitatia) was discerned in the area of Legio. In aerial photographs one can make out remnants of long, thin plots that correspond to the routes of Roman roads, evidence of an organized and orderly Roman system that survived in the valley until the nineteenth century.76

The excavations conducted in 2003-2005 in Megiddo Prison exposed finds that have important implications for the civilian settlement alongside the legionary camp. We will note here that at the point of meeting between the Roman fort and the Jewish settlement, evidence was found for a local community among whose members women and Roman military officers are mentioned, and also for an Early Christian community on the margin of the Jewish village of Kefar ‘Othnai.74 This evidence provides an additional tool for our understanding of the area in which a Jewish village and a Roman military camp existed side by side. Moreover, the civilian settlement that developed alongside the Roman camp was apparently the basis of the Roman-Byzantine town of Maximianopolis, founded in the late third century CE or at the latest in the early fourth century.

Our study strengthens the assumption that both the Legion II Traiana and the Legion VI Ferrata were stationed at Legio and supports the assessment mentioned above that the Second Legion was already established at Legio in the early second century CE and was replaced by the Sixth Legion a little later. From the data recovered in the survey it seems likely that the legion remained there during the second and third centuries CE, a conclusion reinforced by studies of other Roman military camp sites in the Land of Israel, whose researchers claim that they were founded and functioned during the reigns of Trajan, Hadrian and Antoninus Pius, in the second and third centuries.77 Legionary activity probably continued at Legio until the late third century CE, the days of Diocletian. In this context it is interesting to note that metal objects and accessories associated with the Roman army and legionary equipment that were recovered in the survey at Legio were dated to the first to fourth centuries CE; they constitute evidence of some Roman military presence at the site even after the abandonment of the legionary headquarters and the camp site.78

The Roman Camp at Legio: preliminary summary of the results of the survey The fact that there was a legionary camp at Legio is indisputable. However, in our study we have proposed that in the area of Legio there were two military installations, a military fort with a Roman legionary camp alongside it. The fort, probably the camp of an auxilia or another small unit, was located on the summit of a ridge south of Nahal Qeni in a commanding strategic position overlooking the surroundings. The camp or the headquarters of a legion, was located on the eastern slopes of the hill of el-Manach. Although its position lacked strategic advantage, it was chosen for two other vital purposes, water and roads. Firstly, its location near an abundant and accessible water source avoided the necessity of supplying water via long aqueducts. Secondly, its location at a major junction of Roman roads emphasizes the crucial role played by the road network in the Roman system, in the eastern Empire as a whole and specifically in the province of Judaea.

It is well known from excavations of Roman legionary camps in Europe and North Africa that their average area was 180-240 dunams (18-24 hectares), and it has been claimed that legionary camps in the Land of Israel would have been similar in size.79 However, as we have demonstrated above by the survey results alone, the camp at Legio was not suitable for the stationing of a full 75

Isaac and Roll 1982: 104-106; Tsafrir, Di Segni and Green 1994: 182. Applebaum 1971: 155; 1989: 163, 70-96; Isaac and Roll 1982: 121, Fig. 6. 77 Tzori 1950: 53-54; Hirschfeld 1991: 170-183; Hashman 1996: 1-19. 78 My thanks to G. Stiebel, who assisted in the identification and dating of the objects; see Tepper 2003: 87-88, Fig. 21-21. The issue of the dating of the Roman legionary presence at the camp site at Legio will be addressed in a separate article. 79 Cagnat 1904: 1061-1064; Daremberg-Saglio 1962: 940-959; Parker 2000: 121-125. Legionary camps in the East are similar in size to that at Legio or even smaller: the camp at Palmyra (Tadmor) was ca. 41.5 dunams in area (Kennedy and Riley 1990: 123, 136-137) and that at Udruh 47 dunams (Kennedy and Riley 1990: 131-133). See also notes 28-33 above. 76

The legionary camp was founded near a Jewish village rather than a town, a fact that invites wider re72

Webster 1969: 203-204; Isaac 1998: 89. Webster 1969: 203-204, 220; Safrai 1992: 110-114; Campbell 1994: 142. 74 Tepper and Di Segni 2006:22-54. 73

68

YOTAM TEPPER: THE ROMAN LEGIONARY CAMP AT LEGIO, ISRAEL legion. Since there is evidence for the stationing of vexillationes of the Legion VI Ferrata near Beth Govrin,80 Sebastia,81 Tel Shalem82 and Tiberias,83 and since inscriptions attest to the participation of troops of this legion in the building of the aqueduct of Caesarea,84 it is possible that the camp at Legio was the permanent base of only some of the legion’s troops, perhaps as a supreme headquarters, like the headquarters of the Legion X Fretensis in Aila.85

Chevallier R. 1997 Les Varies Romaines, Paris. Coello T. 1996 Unit Sizes in the Late Roman Army. BAR Int. Ser. 645. Oxford. Conder R.E. and Kitchener H.H. 1882 The Survey of Western Palestine, Memoirs. Vol. 2. Samaria, London. Cotton H.M. and Eck W. 2001 Governors and their Personnel on Latin Inscriptions from Caesarea Maritima. In The Israel Academy of Sciences and Humanities, Proceedings. Vol. VII. No. 7. Jerusalem: 215-240. Cotton H.M. 2000 The Legio VI Ferrata. In Le Bohec Y. (ed.), The proceedings of: Deuxième congrès de Lyon sur l`armée romaine. Les légions de Rome sous le Haut-Empire, Lyon (17-19 September 1988). Lyon: 351-357. Dąbrowa E. 1996 The commanders of Syrian legions, 1st-3rd c. A.D. In Kennedy D.L. (ed.), The Roman Army in the East. In JRA 18: 277-296. Dalman G. 1914 Jahresbericht des Instituts für das Arbeitsjahr 1913/4. In PJB 10: 3-54. Daremberg C.H. and Saglio E.D.M. 1962 Castra. Dictionnaire des antiquités grecques et romaines. Vol. III. Paris: 940-959. Eck W. 1984 Zurn konsularen status von Iudaea im fruhen 2. Jh. In Bulletin. American Society of Papyrologists 21: 55-67. Fabian P. 1995 The Late-Roman Military Camp at Beer Sheba: A New Discovery. In JRA 14: 235-240. Foerster G. 1985 A Cuirassed Bronze Statue of Hadrian. In ‘Atiqot 27: 139-157. Guy P.L.O. 1938 Megiddo Tombs, Chicago. Hartmann R. 1910 Die Palästina-Route des Itinerarium Burdigalense. In ZDPV 33:169-188. Hashman S. 1996 Roman Army Alignment in the Eastern Slopes of Samaria Dated to the First – Third Century CE, M.A. thesis, Haifa University (Hebrew). Hecker M. 1961 The Roman Road Legio-Zippori. In Bulletin of the Israel Exploration Society 25/3: 175186 (Hebrew). Hirschfeld Y. 2000 Horvat ‘Eleq: Architecture and Stratigraphy. In Hirschfeld Y. (ed.), Ramat Hanadiv Excavations, Final Report of the 1984-1988 Seasons, Jerusalem: 235-370. Hirschfeld Y. 1991 Khirbet Khiraf: a 2nd-c. Fort in the Jordan Valley. In JRA 4: 170-183. Iliffe J.H. 1933 Greek and Latin Inscriptions in the Museum. In QDAP 2 :120-126. Illan Z. 1988 Yesterdays, Tel Aviv (Hebrew). Illan Z. 1989 The Aqueduct to Kefar Hanania. In Amit D. – Hirschfeld Y. and Patrich J. (ed.), The Aqueducts of Ancient Palestine, Jerusalem: 97-100 (Hebrew). Isaac B. 1998 Milestones in Judaea: from Vespasian to Constantine. In Isaac B., The Near East nder Roman Rule, Selected Papers, Leiden-New York-Köln: 4868. Isaac B. and Roll I. 1979 Judaea in the Early Years of Hadrian’s reign. In Latomus 38: 54-66. Isaac B. and Roll I. 1979a Legio II Traiana in Judaea. In ZPE 33: 149-156.

In summary, we should emphasize once again that our conclusions are based primarily on the results of an archaeological survey. The shortcomings of our analysis and the conclusions presented above derive from the well-known limitations of such surveys. The secrets of this important site in the northern Land of Israel still await archaeological excavation and study by advanced techniques under the soil of the Jezreel Valley. Bibliography Historical Sources Eusebius On. Klostermann, E. (ed.). Das Onomastikon der biblischen Ortsnamen, GCS 11 i. Leipzig 1904. Notitia Dignitatum Seek, O. (ed.). Notitia Dignitatum, Minerva 1962. References Amit D. – Hirschfeld Y. and Patrich J. (ed.), 1989. The Aqueducts of Ancient Palestine, Jerusalem (Hebrew). Applebaum S. 1989 Judaea in the Hellenistic and Roman Times. Historical and Archaeological Essays, Leiden-New York-Københaven-Köln. Avi-Yonah M. 1946 Newly Discovered Latin and Greek Inscriptions. In The Quarterly of the Department of Antiquities in Palestine 12: 84-102. Avi-Yonah M. 1973 When did Judea become a Consular Province? In Israel Exploration Journal 23: 209-213. Avni G. 2002 The Northern Boundary of Aelia Capitolina – A View from the Cemeteries. In Faust A. and Baruch E. (ed.), New Studies on Jerusalem, Proceedings of the Eight Conference, December 26, Ramat Gan 2002: 101-110 (Hebrew). Bohec Y.L. 1996 La garnison de Lyon sous le Principat. In Bénard J. et al. (ed). L’armée romaine en Gaule. Paris: 100-111. Cagnat R. 1904 Legio. In Dictionnaire des Antiquités. Vol. 3.2. Paris: 1047-1093. Campbell B. 1994 The Roman Army 31 B.C. – A.D. 337, London. 80

Iliffe 1933: 121. Reisner, Fisher and Lyon 1924: 251. 82 Tzori 1950: 53-54. 83 Avi-Yonah 1946: 91. 84 Lifshitz 1960: 110. 85 Notitia Dignitatum, Ed. Seeck: 73, Line 30. 81

69

THE LATE ROMAN ARMY IN THE EAST Rea J. 1980 The Legio II Trajana in Judaea? In ZPE 38: 220-22. Reeg G. 1989 Die Ortsnamen Israels nach der rabbinischen Literatur, Wiesbaden. Reisner G.A - Fisher C.S. and Lyon D.G. 1924 Harvard Excavations at Samaria, 1908-1910, Cambridge. Ritterling C. 1925 Legio VI Ferrata. In RE 12.2: 15871596. Roll I. 1994 Roman Roads. In Tsafrir Y. - Di Segni L. and Green J. (ed.), Tabula Imperii Romani. IudaeaPalaestina, Jerusalem: 21-22. Safrai Z. 1992 The Roman Army in the Galilee. In Levine L.I. (ed.), The Galilee in Late Antiquity, New York-Jerusalem: 103-114. Schumacher G. 1903 Die ägyptische Hauptstrafsse von der Ebene Saron bis zur Ebene Jesreel. In Mittheilungen und Nachrichten des Deutschen Palaestina Vereins: 4-10. Schumacher G. 1908 Tell el-Mutesellim I, Leipzig. Segal A. 1999 Megiddo Junction. In HA 109: 48-49 (Hebrew). Sharoni A. 1987 The Comprehensive Arabic-Hebrew Dictionary. University of Tel Aviv and Israel Defence Forces Intelligence Corps (Hebrew-Arabic). Shatzman I. 1983 The Roman Army from Augustus till Hadrianus. In Rappaport A. (ed.), The History of the Israel Nation - Judea and Rome – The Jewish Revolts, Jerusalem: 265-282 (Hebrew). Stern E.J. and Getzov N. 2006 Aspects of Phoenician Burial Customs in the Roman Period in the Light of an Excavation near El-Kabri (Kabri). In ‘Atiqot 51: 91-123. Stiebel G.D. 1999 The Whereabouts of the Xth Legion and the Boundaries of Aelia Capitolina. In Faust A. and Baruch E. (ed.), New Studies on Jerusalem, Proceedings of the Fifth Conference, December 23 1999: 86 -103 (Hebrew). Strange G.L. 1965 Palestine under the Moslems. A Description of Syria and the Holy Land from A.D. 650 to 1500, Beirut. Tepper Y. 2002 Lajjun–Legio in Israel: Results of a Survey in and around the Military Camp Area. In Freeman P. - Bennett J. - Fiema Z.T. and Hoffmann B. (ed.), Proceedings of the XVIIIth International Congress of Roman Frontier Studies, Amman, September 2000: 231-242. Tepper Y. 2003 Survey of the Legio Area near Megiddo – Historical and Geographical Research, M.A. thesis, Tel Aviv University (Hebrew). Tepper Y. 2003a Survey of the Legio Region. In HA 115: 29*-31*. Tepper Y. and Di Segni L. 2006 A Christian Prayer Hall of the Third Century C.E. at Kefar ‘Othnai (Legio), Excavations at the Megiddo prison 2005. Jerusalem. Thomsen P. 1907 Loca Sancta, Halle. Toynbee J.M.C.1971 Death and Burial in the Roman World, London. Tsafrir Y. 1984 Eretz Israel from the Destruction of the Second Temple to the Muslim Conquest. Vol. 2: Archaeology and Art, Jerusalem (Hebrew).

Isaac B. and Roll I. 1982 Roman Roads in Judaea I, The Legio-Scythopolis Road. BAR International Series 141, Oxford. Isaac B. and Roll I. 1998 Legio II Traiana in Judaea. In Isaac B., The Near East under Roman Rule, Selected Papers, Leiden-New York-Köln: 198-207. Isaac B. and Roll I. 1998a Legio II Traiana in Judaea – A Reply. In Isaac B., The Near East Under Roman Rule, Selected Papers, Leiden-New York-Köln: 208-210. Isaac B. 1990 The Limits of Empire, The Roman Army in the East, Oxford. Isaac B. 1996 Eusebius and the Geography of Roman Provinces. In Kennedy D.L. (ed.), The Roman Army in the East, Ann Arbor: 153-167. Kennedy D. and Riley D. 1990 Rome’s Desert Frontier from the Air, London. Kennedy D. 2002 Qaryal al-Hadid: A “Lost” Roman Military Site in Northern Jordan. In Levant 34: 99110. Keppie L. 1973 The Legionary Garrison of Judaea under Hadrian. In Latomus 32: 859-864. Kloner A. 2002 Cremation Burials in Jerusalem as Evidence for Population Change. In Faust A. and Baruch E. (ed.), New Studies on Jerusalem, Proceedings of the Eighth Conference, December 26, Ramat Gan 2002: 93-100 (Hebrew). Lamon R.S. 1935 The Megiddo Water System, Chicago. Lamon R.S. and Shipton G.M. 1939 Megiddo I, Seasons of 1925-34, Strata I-V, Chicago. Lifshitz B. 1960 Sur le date du transfert de la Legio VI Ferrata en Palestine. In Latomus 19: 109-111. McElderry R.K. 1908 The Second Legionary Camp in Palestine. In The Classical Quarterly 2: 110-113. Negev A. 1971 The Nabataean Necropolis of Mampsis (Kurnub). In IEG 21: 110-129. Nelson H.H. 1913 The Battle of Megiddo, Chicago. Oppenheimer A. 1991 Galilee in the Mishnaic Period, Jerusalem (Hebrew). Parker S.P. 2000 Roman Legionary Fortresses in the East. In Brewer J.R. (ed.), Roman Fortresses and their Legions, London: 121-138. Parker S.P. (ed.) 1987 The Roman Frontier in Central Jordan – Interim Report on the Limes Arabicus Project 1980-1985, 1-2. BAR Int. Ser. 340, Oxford. Pflaum H.G. 1969 Remarques sur le changement de statut administratif de la province de Judée: A propos d’une inscription récemment découverte à Sidé de Pamphylie. In IEJ 19: 225-233. Porath Y. 1984 Lime Plaster on Aqueducts – A New Chronological Indicator. Leiuchtweiss Institut für Wasserbau der Technischen Universität Braunschweig, Mitteilungen Heft 82: 1-16. Porath Y. 2000 Caesarea – 1994-1999. HA 12: 38-46 (Hebrew). Raban A. 1999 Map of Mishmar Ha-‘Emeq (32). Archaeological Survey of Israel, Jerusalem (HebrewEnglish). Ramsay W.M. 1916 Colonia Caesarea (Pisidian Antioch) in the Augustan Age. In JRS 6: 83-134.

70

YOTAM TEPPER: THE ROMAN LEGIONARY CAMP AT LEGIO, ISRAEL Tsafrir Y. - Di Segni L. and Green J. 1994 Tabula Imperii Romani. Iudaea Palaestina, Jerusalem. Tsuk T. 1985 The Aqueducts to Sepphoris, M.A. thesis. Tel Aviv University (Hebrew). Tsuk T. 1988-9 The Aqueduct to Legio and the Location of the VIth Roman Legion. In Tel Aviv 15-16: 92-97. Tsuk T. 1989 The Aqueducts to Sepphoris. In Amit D. Hirschfeld Y. and Patrich J. (ed.), The Aqueducts of Ancient Palestine, Jerusalem: 101-108 (Hebrew). Tzori N. 1950 An Inscription of the Legio VI Ferrata from the Northern Jordan Valley. In IEJ 21: 53-54. Von Petrikovits H. 1975 Die Innenbauten Römischer Legionslager, Opladen. Webster G. 1969 The Roman Imperial Army of the First and Second Centuries A.D., London. Winogradov Z.S. 1989 The Ancient Aqueduct of Tiberias. In Amit D. - Hirschfeld Y. and Patrich J. (ed.), The Aqueducts of Ancient Palestine, Jerusalem: 123-132 (Hebrew).

71

The development of the Roman forces in northeastern Anatolia New evidence for the history of the exercitus Cappadocicus Michael A. Speidel Bern Universität

The frontier in Anatolia between the Black Sea and the Syrian plain was one of Rome’s most important military borders.1 Its main purpose was to keep Roman territory safe from bellicose neighbours by guarding the Upper Euphrates and the few great invasion routes connecting the East and the West. Naturally, it also served Roman invasion forces as a base for military operations beyond the frontier. With neighbouring Armenia as Rome’s major political concern in the East for four centuries, the imperial army in northeastern Anatolia had an important role to play. Research has steadily improved our knowledge of Rome’s forces in Cappadocia and recently substantial new evidence has come to light (see below).2

32 33 34 35 36 37 38

This list of military units in the Notitia Dignitatum describes the Roman frontier army in Northeast Anatolia and on the eastern shore of the Black Sea in ca AD 394.3 The situation it presents is the result of developments that took place over four centuries of Roman history in the region. The list includes units in Armenia and Pontus, all of which were under the command of the dux Armeniae.4 It is well-known that such lists in the Notitia Dignitatum are not to be understood as complete inventories of the Roman forces and garrison places within the described provinces.5 For these lists never contain information on outposts, on the mobile forces of the field army or on federate troops which may also have been employed in the region. Finally, through error, some sites and units may be missing and other entries may not always reflect the latest developments. Nevertheless, the Notitia Dignitatum contains our most complete list of Rome’s armed forces on the frontier in northeastern Anatolia: thirteen cavalry units, three legions, one cohors equitata and nine infantry cohorts. A number of units can readily be recognized as creations of the late third and fourth centuries. Most of these are known from the Notitia Dignitatum only. What can be said about their history is therefore mainly to be found in their names.

The Late Roman Units For the northern part of this frontier, the Notitia Dignitatum supplies a list of military frontier units and garrison places of the late fourth century AD (Or. 38): 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

Ala prima felix Theodosiana, Pithiae. Cohors prima Theodosiana, Valentia. Cohors Apuleia civium Romanorum, Ysiporto. Cohors prima Lepidiana, Caene-Perembole. Cohors prima Claudia equitata, Sebastopolis. Cohors secunda Valentiana, Ziganne. Cohors, Mochora.

Sub dispositione viri spectabilis ducis Armeniae: Equites sagitarii, Sabbu. Equites sagitarii, Domana. Praefectus legionis quintaedecimae Apollinaris, Satala. Praefectus legionis duodecimae Fulminatae, Melitena. In Ponto: Praefectus legionis primae Ponticae, Trapezunta. Ala Rizena, Aladaleariza. Ala Thedososiana, apud Avaxam. Ala felix Theodosiana, Silvanis. Et quae de minore laterculo emittuntur: Ala prima Augusta Colonorum, Chiaca. Ala Auriana, Dascusa. Ala prima Ulpia Dacorum, Suissa. Ala secunda Gallorum, Aeliana. Ala castello Tablariensi constituta. Ala prima praetoria nuper constituta. Cohors tertia Ulpia miliaria Petraeorum, Metita. Cohors quarta Raetorum, Analiba. Cohors miliaria Bosporiana, Arauraca. Cohors miliaria Germanorum, Sisila. Ala prima Iovia felix, Chaszanenica.

Equites sagittarii. Two units of mounted archers are listed as part of the military forces of the dux Armeniae at Sabbus and at Domana on the Upper Euphrates frontier.6 These regiments of light cavalry were former parts of the great cavalry army of the late third century. They may have been moved to the East by Aurelian in AD 272, or perhaps by Diocletian (AD 284 – 305).7 3 For the date and a recent analysis of the various strands in the section for the dux Armeniae cf. Brennan 1998. 4 For Pontos as a province of its own since the third century see e.g. Hoffmann 1969: 411 with n. 775. Barnes 1982: 216. Marek 2003: 51f. The rubric in Ponto is used only once in the surviving text (15) despite the appearence of further Pontic locations in other parts of the list. In those cases it may have simply been ommitted by a copyist. Cf. Brennan 1998: 40. 5 Zuckermann 1991: 538. Brennan 1998: 37 and 43. 6 ND Or. XXIII.11 and 12. For archaeological remains at Sabbus cf. Mitford 1980: 1189. Idem 1998: 263. 7 For this type of unit in the East cf. Hoffmann 1969: 251. Speidel 1984: 259f. and Lewin 2004: 231ff. with further literature on the subject.

1

Cf. Wheeler 1997: 215f. I am particularly grateful to Werner Eck, Köln and Peter Weiss, Kiel for helpfull comments on an earlier version of this paper and for making their studies available to me even before publication. I also wish to thank Ariel Lewin, Potenza-Matera for usefull suggestions and Alfred Hirt, Basel for his kind help with the collection of some of the material used in this paper. 2 Cf. e.g. Ritterling 1902. Mitford 1980. Speidel 1984. Zuckermann 1991. Wheeler 1997. Brennan 1998. Bennett 2002.

73

THE LATE ROMAN ARMY IN THE EAST only from the Notitia Dignitatum.16 According to their names they were all raised during the reign of Theodosius (AD 379-395), though perhaps at different times as one might otherwise expect their names to follow a more systematic pattern (e.g. prima, secunda etc.). This substancial reinforcement of the army in northeastern Anatolia by newly raised troops is likely to have been (at least in part) a reaction to the great loss of Roman soldiers in the battle at Adrianople in AD 378 as well as to the changing political and military situation on the Armenian frontier during the reign of Theodosius (cf. also below).

Legio I Pontica, created by Diocletian and in existence by AD 288, became the garrison of Trapezus where it was still stationed at the end of the fourth century.8 It may have been set up and trained in Isauria before it was moved north.9 Ala prima Iovia felix, stationed some 40 km south of Trapezus at Chaszanenica on the road to Satala, bears the name of Diocletian’s patron god and is thus a tetrarchic creation.10 Ala Rizena, in the late fourth century, was stationed at Aladaleariza (or Olotoedariza and variant spellings) to the West of Satala on the route to Nicopolis.11 As indicated by its name, this unit had a former association with Rhizaion on the Black Sea. It is neither known from other sources, nor is there currently any other indication that Rhizaion was ever the site of a Roman garrison during the first three centuries. During the first two centuries, however, Rhizaion had a local militia which also took part in Arrian’s march against the Alani in ca. AD 135. Arrian, however, only mentions footsoldiers from this contingent.12 Perhaps, therefore, ala Rizena was a tetrarchic creation, set up in order to fortify the area around Rhizaion. This, at any rate, would go well with its name which echoes similar such creations under Diocletian in the area (legio I Pontica, legiones Armeniacae). The position of ala Rizena at the end of the laterculum maius may indicate that this entry was a fairly recent addition to that register.13 Hence, the unit may have been transferred from Rhizaion (?) to Aladaleariza not long before AD 394.

Ala (castello Tablariensi constituta) and Cohors. The fact that the Notitia Dignitatum lists these two units without a name may indicate that they were very recent creations.17 Together with ala prima praetoria nuper constituta (26), and those units which bear Theodosius’s name they are a clear sign of the significant activity in the region at the end of the fourth century (cf. also below). Thus, of the late fourth century army on the eastern Black Sea and in northeastern Anatolia, as listed by the Notitia Dignitatum, four or five units had been added by the end of the reign of Diocletian, one more by Valens and six by Theodosius who also appears to have moved two other units (ala Rizena, ala prima praetoria nuper constituta) into new garrison places. All known additions came in response to changing historical contexts or great losses in battle. Diocletian (AD 284-305) and the first tetrarchy brought a great many new administrative and military changes to the region. The Roman victory over the Persians in AD 297 led to the peace of Nisibis in AD 298. Roman control in southeastern Anatolia now even included vast areas beyond the Severan Tigris frontier.18 The regiones transtigritanae which then came under Roman rule were probably secured by the new legions I and II Armeniaca. Their garrison places, however, remain unknown.19 The Upper Euphrates continued to be guarded by legio XII Fulminata at Melitene and by legio XV Apollinaris at Satala, as well as by several auxilia. Together with legio I Pontica a second legion may originally have been stationed in the province of Pontus, though this remains uncertain.20 Diocletian’s

Cohors secunda Valentiana, according to its name, was raised in the East by the emperor Valens (AD 364-378).14 It was stationed at Ziganne which appears to be identical with Ziganeos (modern Gudava) on the Black Sea, ca. 60 km south of Colchian Sebastopolis. The cohort was set up as part of the Iberian policy of the emperor Valens in the aftermath of Julian’s defeat by the Persians (AD 363).15 Ala Theodosiana, ala felix Theodosiana, ala prima felix Theodosiana and cohors prima Theodosiana are known Bennett 2002: 307 offers the unconvincing suggestion that these units were the renamed cohorts “III Aug. Cyrenaica eq. sag. and (?IIII) Ituraeorum eq. sag.” 8 CIL III 6746 = ILS 639. AE 1972: 636. Ritterling 1925: 1437. Gilliam 1974: 183-191. ND Or. 38, 16. 9 Zuckermann 1991: 528f. 10 ND Or. XXXVIII.31. 11 ND Or. XXXVIII.17. In the late third century this site belonged to the province of Pontus, as is clear from the find spots of milestones mentioning the governor of that province (listed e.g. in Hoffmann 1969: 411 with n. 775. For an as yet unpublished milestone from the territory of Neoclaudiopolis see Marek 2003: 62 with n. 4. Satala, however, may never have belonged to Pontos). Whether this was still true at the end of the fourth century remains unclear. In any case, the rubric in Ponto (15) is hardly relevant to the location of all three alae in 17-19. Cf. Zuckemann 1991: 529 n. 10. 12 Arrian, Ekt. 7 and 14: oƒ `Rizianoˆ logcofÒroi. 13 Brennan 1998: 47. 14 ND Or. XXXVIII.37. 15 Zuckermann 1991: 534ff. Brennan 1998: 44f.

16

ND Or. XXXVIII.18. 19 and 32. ND Or. XXXVIII.25 and 38. Cf. Eck 1992: 395ff. Brennan 1998: 41. 44. 18 Petr. Patr., fragm. 14. Cf. recently Kuhoff 2001: 164ff. with further literature. 19 The fragmentary inscription CIL III 13630 was taken by Ritterling 1925: 1457 to indicate that Satala was the garrison place of legio secunda Armeniaca during the tetrarchy. The revised reading by Mitford 1997: 147f. no. 12 shows, however, that this inscription, a women’s gravestone, does not mention legio secunda Armeniaca at all. Reconstructions of the tetrarchic disposition of military units based on Ritterlings reading must therefore, at least in part, be discarded: e.g. Hoffmann 1969: 412. Kuhoff 2001: 458 (unaware, it seems, of Mitford’s revised reading). 20 Ritterling 1925: 1351. Ensslin 1942: 63f. Contra Hoffmann 1969: 413 with n. 795. 17

74

MICHAEL A. SPEIDEL: THE DEVELOPMENT OF THE ROMAN FORCES IN NORTHEASTERN ANATOLIA reorganization of the northeastern frontier also led to the arrival of new auxiliary units: ala prima Iovia felix and, possibly, ala Rizena.21

northeastern frontier in the years after AD 378. Perhaps some still existed in ca. AD 394, but the Notitia Dignitatum clearly shows that Theodosius aimed to eventually fill any such gaps.

In AD 363 the emperor Julian (AD 360-363) was defeated by the Persian king Sapur II and Rome had to surrender several territories beyond the Tigris as well as important frontier fortresses and cities to the Persians. With matters thus settled in southeast Anatolia, Persian ambitions shifted north and the rivalry between the two great powers began to focus (as so often before) on Armenia and Iberia. It is no surprise to find this reflected in the list of troops of the Notitia Dignitatum: the emperors Valens (AD 364-378) and Theodosius (AD 379-395) in particular added new frontier troops to the Roman forces in northeastern Anatolia.22 These troop movements (the last ones on the northeastern frontier known from the fourth century) appear to have been undertaken in several steps and were still ongoing in ca. AD 394.23 Although several garrison places of the Notitia Dignitatum remain unlocated the substancial number of new units are clear proof of the continuing importance of this frontier in the years following the partition of Armenia in ca. 387. As a result of this agreement between Rome and Persia in the peace of Ekeleac, the Arsacid ruler, Arsak IV, was left to rule the western, Roman part of the country.24 A new Roman defense line of frontier units separating Roman from Persian Armenia was, however, not established in those years. The defence of western Armenia remained the task of local, Armenian troops.25 Some of the recent movements of Roman troops in the years before ca. 394 may have therefore been intended to increase the stability of the region, particularly after the death of Arsak IV around 390.26 There may, however, have been a second and perhaps even more important reason for the addition of many new units on the northeastern frontier in the years before ca. 394: After the catastrophic losses in the battle at Adrianople the emperor Theodosius first had to regroup and reorganize what was left of the eastern army in order to fight the Gothic invadors. For that purpose he also withdrew frontier units from their garrison places.27 As a result, there must have been gaps in the defense of the

The Older Units The remaining fourteen units listed in the Notitia Dignitatum had joined the Cappadocian army before the late third century. The auxilia appear in a geographical order within each province and category of unit (see the map on p. 87). Thus, the units on the Upper Euphrates frontier are listed from south to north with the alae at Chiaca (21), Dascusa (22), Suissa (23) and (otherwise unknown) Aeliana (24), and the cohorts at Metita (27), Analiba (28), Arauraca (29) and Sisila (30).28 The older cohorts on the Pontic coast appear at sites from west to east: Ysiportus (34), Caene – Parembole (35) and Sebastopolis (36).29 The chapter concerning the dux Armeniae does obviously not list any units which had previously perished or had permanently been transferred to other frontiers of the Roman empire. Also, it does not inform us at what time and under which circumstances any of the fourteen older units joined the army of Cappadocia. In order to recognize such changes in the size and composition of the Roman army on the Upper Euphrates and on the eastern Black Sea we therefore need to turn to other sources. With the discovery of three military diplomas and of several inscriptions, substantial new evidence for the history of these units and of the Cappadocian army has recently come to light. Despite many open questions the size and development of the legionary garrison is well known. During the reign of the emperor Vespasian (AD 69-79) Cappadocia was transformed into a major military province with a garrison of two legions and many auxiliary units, propter adsiduos barbarorum incursus, as Suetonius claims.30 Roman fleets were set up on the coasts of the Anatolian Black Sea, and an immense network of highways along the frontiers and in central and eastern Anatolia was developed.31 Legio XII Fulminata was established at Melitene in AD 70/7132 and in AD 75/76 legio XVI Flavia Firma moved into its permanent camp at Satala.33 Later, probably early in Hadrian’s reign (AD 117-138), legio XVI Flavia Firma was replaced at Satala by legio XV Apollinaris.34 Henceforth the headquarters of the legions XII Fulminata and XV Apollinaris remained at

21 There is no need to assume with Zuckermann 1991: 531-533 that Diocletian reduced direct Roman control on the Pontic coast to a stretch of some 60 km East of Trapezus with Caene Parembole as the eastern most stronghold. See Brennan 42ff. 22 The unlocated Valentia (33) may already have been intended as a garrison place by the emperor Valens. 23 Cf. the anonymous ala and cohort (25 and 38), as well as ala prima praetoria nuper constituta (26). Ala Rizena (17) may also have been moved into ist new garrison place at Aladaleariza only recently (cf. above). 24 Blockley 1987. Idem 1992: 42ff. Wheeler 1999: 215. Greatrex / Lieu 2002: 28ff. 25 Proc., Aed. 3,1,16ff. Nov. Theod. V 3 1 (AD 441) regalium fundorum copiis ac praesidiis munitum. Cf. Zuckermann 1998: 108-112. Greatrex / Lieu 2002: 54. A Roman military command and a defense line was only established under Justinian. 26 For the date of Arsak’s death cf. Hewsen 1978-1979: 116f. 27 Lib., Or. 24,38. For Theodosius’ military measures after Adrianople cf. Hoffmann 1969: 440ff. 460.

28 Cf Brennan 1998: 40. The same order can be observed for the equites sagittarii: Sabbus (11) and Domana (12), but (contra Brennan 1998: 40) not for the legions: Satala (13), Melitene (14), in ponto: Trapezus (15). 29 Sebastopolis is certainly best identified with Colchian Sebastopolis/Dioscurias. Thus Brennan 1998: 42ff. Contra: Zuckermann 1991: 531. 30 Suet., Vesp. 8,4. 31 Fleets: French 1984: 53-60. Marek 2003: 59f. Roads: CIL III 306. 312. ILS 8904. Cf. Mitchell 1993: 124f. 32 Jos. BJ VII.1.2. Mitford 1988: 174. Keppie 2000: 192. 33 Keppie 2000: 192f. Mitford 1997: 140ff. 34 Mitford 1997: 142ff. Wheeler 2000: 293ff.

75

THE LATE ROMAN ARMY IN THE EAST Melitene and Satala.35 During the reign of Antoninus Pius (AD 138-161) a third legion (legio VIIII Hispana) was transferred to Cappadocia, only to be wiped out in a military catastrophe in AD 161.36 Trapezus, a major base of the Roman Black Sea fleet, the classis Pontica, was guarded by a Roman cohort before AD 69, which had been created from a former royal regiment of the king of Pontos.37 Hadrian rebuilt the harbor38 and during the second century both Cappadocian legions appear to have had detatchements stationed at Trapezeus.39 Not until the reign of Diocletian, however, was the fortress at Trapezus the actual headquarters of a legion: the newly raised legio I Pontica (cf. above p. 74).

(3) (4)

¹ œilh tîn Getîn (Ekt. 8) = ala I Ulpia Dacorum. ¹ œilh ¹ Galatik» (Ekt. 9) = ala II Gallorum.

COHORTES: (1) oƒ 'Aplanoˆ pezo… (Ekt. 7). oƒ 'Aplano… (Ekt. 14) = coh. Apula c.R. (2) (oƒ) Bosporanoˆ pezo… (Ekt. 3). oƒ pezo… toxÒtai oƒ tîn Bosporianîn (Ekt. 18) = coh. I Bosporanorum ’ eq. sag. (3) (oƒ) Kurhna‹oi (Ekt. 1). (oƒ tîn) Kurhna…wn (Ekt. 3 and 14). oƒ pezo… toxÒtai oƒ tîn Kurhna…wn (Ekt. 18) = coh. III Augusta Cyrenaica eq. sag. (4) oƒ Keltoˆ ƒppe‹j (Ekt. 2) = coh. I Germanorum ’ eq. (5) oƒ 'Italoˆ (Ekt. 3). oƒ tîn 'Italîn ƒppe‹j (Ekt. 9). oƒ tÁj spe…rhj tÁj 'ItalikÁj pezo… (Ekt. 13) = coh. I Italica ’ volunt. eq. c.R. (6) (oƒ) 'Itura‹oi (Ekt. 1). oƒ pezo… toxÒtai oƒ tîn 'Itura…wn (Ekt. 18) = coh. I Ituraeorum ’ eq. sag. (7) oƒ Nom£dej (Ekt. 3). oƒ pezo… toxÒtai oƒ tîn Nom£dwn (Ekt. 18) = coh. I Numidarum eq. sag. (8) oƒ ƒppotoxÒtai oƒ Petra‹oi (Ekt. 1) = coh. III Ulpia Petraeorum ’ eq. sag. (9) oƒ ¢pÕ tÁj prèthj `RaitikÁj (Ekt. 1) = coh. I Raetorum eq. (10) oƒ tÁj spe…rhj tÁj tet£rthj tîn `Raitîn (Ekt. 1) = coh. IIII Raetorum eq.

It is more difficult to establish the number and history of the auxiliary units of Cappadocia. A well known source for the size and nature of the exercitus Cappadocicus (RIC II 914) before the fourth c. AD is, of course, Arrian’s ‘Ektaksis kat’Alanon’. His description of the military action he took as governor of Cappadocia against the army of the Alani in c. AD 135 provides us with much information about the size and composition of the Roman army of Cappadocia. Apart from references to soldiers of the two Cappadocian legions,40 to the governor’s guards,41 to mounted scouts (exploratores)42 and to contingents of local allies,43 Arrian lists four alae and ten cohortes which fought under his command against the Alani:44. ALAE: (1) oƒ ¢pÕ tÁj œilhj Îtini AÙriano… Ônoma (Ekt. 1) = ala II Ulpia Auriana. (2) oƒ ¢pÕ tÁj œilhj Î Ônoma Kolwno… (Ekt. 1) = ala I Augusta geminae colonorum.

In his ‘Periplus Ponti Euxeni’, the description of his inspection tour of ca. AD 131 of the eastern Black Sea, Arrian also mentions several troops and garrison places on the coast between Trapezus and Sebastopolis/ Dioscourias (which he calls ‘the limit of Roman control’):45

35 XII Fulminata: Dio LXXI.9. ND Or. XXXVIII.14. Mitford 1980: 1186. Bertrandy / Rémy 2000: 254. XV Apollinaris: Dio LV.23.5. ILS 1142. Itin. Ant. 183.5. ND Or. XXXVIII.13. Wheeler 2000: 295ff. Six new tile stamps of legio XV Apollinaris were found at Satala by Dr. M. Hartmann during a survey carried out in 2004. For a report see www.mavors.org/en/projects_3.htm. For legio II Armeniaca alledgedly at Satala see above note 19. 36 Dio LXXI.2.1. Birley 1988: 316-325. Eck 1972: 459-462. 37 Tac., Hist. 3,47. 38 Arrian, Peripl. 16. cf. Dio LXIX.5.3. 39 CIL III 6745. 6747. AE 1975: 783 = AE 1993: 1562. Cf. Mitford 1974: 160ff. Wheeler 2000: 301f. 40 Legio XII Fulminata: ¹ dwdek£th f£lagx (6. 24). oƒ tÁj dwdek£thj f£laggoj pezo… (15). Legio XV Apollinaris: ¹ pentekaidek£th f£lagx (5. 6. 24). ¹ pezik¾ f£lagx ¹ pentekaidek£th (15). Equites legionis: oƒ ¢pÕ tÁj f£laggoj ƒppe‹j (4). oƒ swmatofÚlakej (22). Cf M.P. Speidel at Haensch 1997: 715 n. 32. 41 oƒ ™p…lektoi ƒppe‹j (4.22). koàfoi longcofÒroi (23) Ritterling 1902: 359ff. 365. Speidel 1978: 49f. Haensch 1997: 723f. 42 kataskÒpoi ƒppe‹j (1.11). 43 oƒ te ¢pÕ tÁj mikr©j 'Armen…aj (7), oƒ 'Armšnioi ... toxÒtai (12.29), tÕ ™ke…nwn ƒppikÒn te kaˆ pezikÒn (13), oƒ te ¢pÕ tÁj mikr©j 'Armen…aj sÚmmacoi (14), oƒ Trapezount…wn Ðpl‹tai (7), oƒ Trapezount…wn gumnÁtej (14), oƒ KÒlcoi (7), oƒ `Rizianoˆ logcofÒroi (7.14). These troops were of course not an official part of the Roman army. 44 The essential study of Arrian’s text is that of Ritterling 1902. For the full names of the troops cf. also the evidence quoted below p. XXff.

(1) A garrison of footsoldiers and 20 horsemen at Hissou Limen (3) (2) Five cohorts at Apsaros (6) (3) 400 selected soldiers in a fort at the mouth of the river Phasis (9) (4) soldiers in a fort at Sebastopolis/Dioscourias (10. 17) Arrian does not mention the names of any of these units. The soldiers at the harbor of Hissos were equipped with lances (lÒgcai) and had 20 horsemen assigned to them – a number which Arrian in c. AD 131 judged to be sufficient. The five cohorts at Apsaros were auxiliary units under the command of the praepositus numeri tendentium in Ponto, Absaro, a former high-ranking legionary centurion.46 The name of the detachment at Phasis which Arrian described as tetrakÒsioi stratiîtai ™p…lektoi may be revealed by a tile stamp from the area of the small Justinianic fort at Petra,

45 46

76

Arrian, Peripl. 17. CIL X 1202 = ILS 2660. Speidel 1992: 204f.

MICHAEL A. SPEIDEL: THE DEVELOPMENT OF THE ROMAN FORCES IN NORTHEASTERN ANATOLIA between Apsarus and Phasis: vex(illatio) Fa(siana).47 These soldiers were equipped with artillery and might have belonged to the governor’s pedites singulares, though this remains impossible to prove.48 The camp at Sebastopolis/Dioscurias housed a unit which included horsemen and Arrian’s visit there in ca. AD 131 may have led to some building activities.49 Nearly two decades after Arrian’s tour a new ‘limit of Roman control’ was set up at Pityus, the next harbor along the Black Sea coast. A military camp was built there with the help of soldiers of legio XV Apollinaris. Two fragmentary Latin inscriptions of AD 152 and 223 are proof of the Roman army’s control of the site.50

The third diploma is as yet unpublished and its text is known only from a somewhat faulty transcription of its outer face by an anonymous modern copyist.54 The constitution concerns soldiers of four alae and thirteen55 or fourteen cohortes of the army of Galatia et Cappadocia in AD 101.56 The diploma was issued to a soldier of cohors II[I] Syrorum sagittariorum from Cilician Anazarbos.57 The units as listed in the copyist’s transcription and reproduced by Barbara Pferdehirt are:58 ALAE: (1) Augusta Germaniciana (2) I Claudia Gallorum (3) I Thracum Herculana (4) Gemina colonorum

The New Diplomas

COHORTES: (1) I Italica milliaria (2) I Italica milliaria voluntariorum c.R. (3) I Ituraeorum milliaria (4) I Bosporanorum (5) I Apamenorum (6) I Augusta Cyrenaica (7) I Raetorum (8) I Claudia (9) II Claudia (10) II Hispanorum (11) II Augusta Thracum (12) III Augusta Cyrenaica (13) III Syrorum sag. (14) Augusta c. R.

In many ways, the documents from the provincial archives of the first three centuries from which the constitutions for the military diplomas were drawn up can be understood as forerunners of the late Roman lists from which the Notitia Dignitatum was compiled. It has often been deplored that no military diploma was available to inform us of the auxiliary garrison of Cappadocia. This has now dramatically changed with three military diplomas having come to light for the province of Galatia et Cappadocia from the end of the first and the early second centuries AD.51 The earliest diploma dates to AD 94 and once belonged to a horseman of ala I Thracum Herculana.52 The unit list named three (?) alae and thirteen cohortes. The following names have survived: ALAE: (1) [--- T]hracum Herculana (2) [I Gemina colonorum?] (3) I Augusta Germ‹a›ni[ciana] COHORTES: (1) [I Augusta c]ivium Romanorum (2) [---] (3) I Italica milliaria [voluntariorum Romanorum (4) I [---]

Together these three documents considerably increase our knowledge of the Roman army in Cappadocia. Still, because of some obvious mistakes in the copyist’s transcription of the diploma of AD 101 a number of questions must remain open. Inscriptions, however, supplement the information given by the new diplomas. The careers of equestrian officers, in particular, are a valuable source. They help to bypass the scarcity of documents from Cappadocia for they reveal a remarkable pattern: the majority of equestrian officers commanded more than one unit in eastern Anatolia.59 Taken together

ci]vium

The second diploma ist an exceptional document which, in AD 100, was issued (for unnamed reasons) for only two auxiliary veterans of the Cappadocian army: a former eques of ala [Thracum Herculan]a and a former centurion of cohors I Augus[ta c(ivium) R(omanorum)].53

54

Cf. RGZM p. 18. I owe further information about this diploma to Barbara Pferdehirt, Mainz and to Paul Hoder, Manchester. Its current location seems unknown. 55 Thirteen if cohors I Italica milliaria and cohors I Italica milliaria voluntariorum cR are identical. Cf. below. 56 Unfortunately the line with totals of the alae and cohortes was left out by the copyist. 57 The copyist wrote II Syrorum sagittariorum but no such cohort appears in the list. The numeral III of cohors Syrorum sag. in the unit list, however, must be correct as the units all appear in numerical order. 58 The question mark in II (?) Augusta Thracum in Pferdehirt’s transcription should belong to II Hispanorum, as Barbara Pferdehirt kindly informs me, where the copyist wrote: INHISPANORUM. Surely II Hispanorum is the correct transcription. 59 AE 1897: 123 (PME C 144). AE 1925: 44 (PME C 143). AE 1926: 150 (PME S 25). AE 1933: 270, 269bis (PME C 168). AE 1951: 279 (PME B 14). AE 1974: 226 (PME V 28). AE 1977: 802 (PME E 3). AE 1982: 827 (PME Inc. 40). AE 1991: 1750 (PME H 21a). AE 1993: 1787 (PME S 86). CIL II 2213 (PME P 101). ILS 2741 (PME N 6). 2748 (PME A 31). 8551 (PME A 21). 8864 (PME F 82). 8865 (PME F 109).

47 Speidel 1992: 205ff. Unfortunately, the brick is a stray-find without archaeological context. 48 Speidel, ibid. Cf. also Speidel 1995: 52ff. Contra: Wheeler 2000: 304. 49 Cf. AE 1905, 175 (if correctly restored): -] / [H]ad[rian(us) Aug(ustus) / per Fl(avium) A[rrianum] / leg(atum) [Aug(usti) pr(o) pr(aetore). 50 Kiguradze et al. 1987: 88-92. Speidel 1984: 209-211. Wheeler 2000: 303. Cf. also Lekvinadze 1962: 85 for a brick with the stamp LEG from a nearby outpost found in a 3rd or 4th c. context. 51 For a summary of the development of the Roman administration of Cappadocia see Haensch 1997: 397 an 406f. 52 RGZM 7. 53 Eck / Pangerl 2005: 233-241.

77

THE LATE ROMAN ARMY IN THE EAST with the information from the new diplomas we can now investigate the history of the auxiliary units of the Cappadocian army of the first three centuries AD.60

equestrian careers of the second century and two of its decurions were buried near Iconium.67 As it is still listed in the Notitia Dignitatum it appears to have remained on the northeastern frontier throughout its history.68

The Auxilia of the Cappadocian Army (3) Ala I Ulpia Dacorum was raised by Trajan in the aftermath of his victory over the Dacians in AD 106 or, possibly, already in AD 102.69 It may have come to Cappadocia to fight in Trajan’s Parthian war. The ala was part of Arrian’s expeditionary force in c. 135 and seems to have remained within the exercitus Cappadocicus until the end of the fourth century.70

The Alae (1) Ala II Ulpia Auriana was set up by Trajan (probably for his Dacian wars) from a core of horsemen furnished by the ala I Hispanorum Auriana from Raetia.61 It is not clear when it came to Cappadocia, though Trajan’s Parthian war seems a likely occasion. It took part in Arrian’s operations against the Alani in c. 135 and remained in Cappadocia until the end of the fourth c. AD.62 Its garrison place in the late fourth century was at Dascusa. Earlier, however, it may have been stationed near Zimara.63

(4) Ala I Claudia Gallorum. It comes as a surprise to find this ala among the troops of the Cappadocian army in AD 101. Since the early 1st c. AD ala I Claudia Gallorum (Capitoniana) was stationed in Moesia (inferior).71 Its name also appears in the Lower Moesian diplomas of AD 105 and 111,72 and it has been plausibly restored in a diploma of AD 97 of that province: ... et I Claudi]a Gallorum ...73 It is, however, not on record in the known constitutions for Lower Moesia of AD 92 and 99.74 Perhaps, therefore, ala I Claudia Gallorum temporarily joined the army of Cappadocia after Trajan’s inspection of the troops on the Lower Danube in AD 99, only to return to Moesia inferior before his second Dacian war of AD 105/6.75 Yet, it is possibly of no great significance that ala I Claudia Gallorum is missing from the known unit lists of the diplomas for Moesia inferior in AD 92 and in AD 99, as it seems likely that further parallel constitutions were issued for the army of this province in those years with the names of more alae and cohortes.76 Also, it would be rather surprising to find this unit leaving the Danube exactly during the period when troubles with the Dacian king Decebalus were increasing and the preparations for war had perhaps even begun.77 It is equally surprising to find it back precisely from where it had come already in AD 105.78 Perhaps, therefore, ala I Claudia Gallorum (Capitoniana) never actually left

(2) Ala I Augusta Gemina colonorum was probably raised and stationed in Galatia under Augustus and appears to have remained in northeastern Anatolia throughout the early principate.64 Its name has therefore plausibly been restored in the diploma for Galatia et Cappadocia of 94 by Paul Holder.65 The unit is listed both in the diploma of 101 and as part of Arrian’s expeditionary army of c. AD 135.66 Ala colonorum is also known from two eastern 8868 (PME Inc 30). 9471 (PME A 4). 9503 (PME C 81). Olympia 447 (PME C 136). For the majority of the equestrian officers’ promotions within Cappadocia (including the Syrian sector of the Anatolian limes) cf. also Devijver 1992: 90ff. 60 No reference is made here to the confused works of J. Spaul: Ala2 (1994) and Cohors2. The evidence for and a short history of the auxiliary infantry units of the Imperial Roman Army (2000). The following units were excluded from the list: The name and number of the cohort which set up the building inscription from AD 80/82 found at Pagnik Öreni (Dascusa) by R. Harper (AE 1975: 809) have entirely broken away, as is clear from the photograph in Harper 1974: 108 and Pl. 96.1 (see also Mitford 1974: 172f.). The unit is wrongly recorded as coh II[--- by Mitchell 1993: 119 and by Bennett 2002: 304. The soldier of cohors I Thracum Syriaca who’s unpublished gravestone was found by T.B. Mitford (Mitford 1998: 267) in 1986 not far west of Cengerli came to Cappadocia together with his unit from Moesia superior, probably to fight in Trajan’s Parthian war: AE 1926: 150 = AE 1992: 1689. Cf. Saxer 1967: no. 117 and CIL XVI 46. See now Eck / Pangerl 2005c. His unit was never part of the Cappadocian army. The same must be true, though perhaps during a different Parthian war, for the ‘auxiliary from the Danube’ who’s ‘burial place’ at Vahsin was also recorded by T.B. Mitford (Mitford 1998: 263). For the identification of garrison places and usefull surveys of the archaeological remains see e.g. Bennett 2002. Brennan 1998. Crow 1986. French 1983. Mitford 1977. Mitford 1980. Mitford 1989. Wheeler 1991. Wheeler 1997. Zuckermann 1991 etc., each with further literature. 61 Birley 1988: 382. For ala I Hispanorum Auriana see also Lörincz 2001: 21. 62 Arrian, Ekt. 1. CIL III 6743 (Pingan/Penga, 2nd. c.). AE 1968, 528 = IGLS 21, 49. AE 1996, 1630 (ala II Ulp. Auriana in Cappadocia, 2nd c.). ND Or. XXXVIII.22. Ritterling 1902: 361. 63 Dascusa: ND Or. XXXVIII.22. Zimara: CIL III 6743 with Mitford 1980 1189 and idem 1998: 266. Zimara is not listed in the Notitia Dignitatum as a garrison place. Dascusa, on the other hand, had been the site of a Roman fort since the Flavian period: Cf. Mitford 1974: 172 = AE 1975: 809 with Mitford 1980: 1190. Idem 1998: 262f. Bennett 2002: 304. 64 Speidel 1984: 285f. Mitchell 1993 I: 74. 65 RGZM p. 18f. 66 Arrian, Ekt. 1.

67

AE 1926: 150 (early 2nd c. PME S 25. For the date cf. also Saxer 1967: no. 117). AE 1896: 80 = AE 1977, 802 and IGRR IV 642 (2nd c. PME E 3.). AE 1907: 57 (2nd c.). 1930: 108 (2nd.c.). Ritterling 1902: 361. 68 ND Or. XXXVIII.21. 69 Cf. Eck / Pangerl 2006: 226f. Weiss 2006: 288. Dana / Matei-Popescu 2006: 206 and 209. 70 Arrian, Ekt. 8. CIL VI 1333 = 31633 = ILS 1077 (mid 2nd c. For the date see now Weiss / Speidel 2004: 256.). ND Or. XXXVIII.23. Ritterling 1902: 362. 71 Speidel 1984: 217-221. 409. Cf. RMD IV 209 (AD 78?). MateiPopescu 2004: 185f. 72 CIL XVI 50. RMD IV 222. 73 Weiss 1997: 238 = AE 1997, 1774. Cf. Eck / Pangerl 2005a: 185-192. 74 14. June AD 92: Petolescu / Popescu 2004: 269-273 = AE 2003, 1548. Weiss 2005: 207-217. 14. August AD 99: CIL XVI 45 with RGZM 8 and CIL XVI 44. 75 For Trajan’s inspection of the Danubian army in AD 99 cf. Speidel 2002: 26ff. 76 A further constitution was issued for Moesia inferior in AD 97 (Weiss 2005: 213ff.) and in AD 99 (RMD IV 217). Possible further parallel constitutions for 14. June AD 92 and for 14. August AD 99 are plausibly assumed by Weiss 2005: 209 and n. 11. 77 Speidel 2002: 31f. Cf. also Strobel 1984: 111. 78 CIL XVI 50.

78

MICHAEL A. SPEIDEL: THE DEVELOPMENT OF THE ROMAN FORCES IN NORTHEASTERN ANATOLIA Moesia inferior, and its name in the Cappadocian diploma of 101 is due to a mistake by the modern copyist. If so, ala II Claudia Gallorum may have been meant. This unit is known only from a Lower Moesian diploma of AD 92 where its name appears on both faces of the document.79 If this unit is identical with Cappadocian ala II Gallorum it must have left Lower Moesia soon after AD 92. What little is sofar known of the history of Cappadocian ala II Gallorum goes well with this suggestion (cf. below).80 Yet, neither the Notitia Dignitatum nor the few inscriptions which mention ala II Gallorum use the title Claudia - though this need not be decisive.81 In short, it is temping to recognize ala II (Claudia?) Gallorum in the Cappadocian diploma of 101, but this must currently remain speculation.

AD 135. However, the career of a Roman knight, inscribed on the base of a statue in Ephesos records his promotion from tribunus legionis XV Apollinaris to praefectus alae Augustae Germanicianae in the later 2nd or 3rd c. and therefore suggests that this unit was still in northeastern Anatolia at that time.91 Its later fate is unknown. (7) Ala prima praetoria of the Notitia Dignitatum may be identical with ala I praetoria singularium civium Romanorum which was part of the Lower Pannonian army until around AD 200.92 If so, it will have come to the East during one of the Parthian or Persian wars of the 3rd c.93 Not long before ca. AD 394 the ala appears to have been moved into a new garrison place on the northeastern frontier (nuper constituta).

(5) Ala II Gallorum appears to have been in Cappadocia when Ti. Claudius Helvius Secundus took over command of this regiment in ca. AD 110.82 Its earlier history is difficult to trace (cf. above for a possible identification with ala II Claudia Gallorum). A gravestone from Posala in Galatia suggests that our unit was already in northeastern Anatolia towards the end of the first century.83 Earlier, it may have been stationed on the Lower Danube.84 The unit was certainly in northeastern Anatolia when Q. Baienus Blassianus was its commander, perhaps in the late 20s or early 30s of the 2nd c., as he went on to take the census in Cappadocia, Pontus mediterraneus and in Armenia minor.85 Thus, ala II Gallorum must be Arrian’s ¹ œilh ¹ Galatik» (9).86 Epigraphy and the Notitia Dignitatum confirm its presence in Cappadocia until the late fourth century.87

(8) Ala I Thracum Herculana is now known to have been in Cappadocia in AD 94, 100 and 101. The equestrian career of C. Sappius Flavus who was promoted from the command of this ala to praefectus ripae fluminis Euphratis during the second half of the first c. AD (but before the early 90s) strongly suggests that ala Herculana was in the East at that time, possibly on or near the banks of the Euphrates.94 Yet, the unit is not mentioned in any inscriptions or military diplomas of the Roman province of Syria of the first or early second century.95 In the light of the new Cappadocian diplomas it is therefore best to assume that ala I Thracum Herculana was in Cappadocia when Sappius Flavus took over command. Ala I Thracum Herculana is not yet listed in the Syrian diplomas of AD 12996 but these only record the names of two alae and thus not the complete Syrian cavalry force at that time. The ala was not with Arrian in c. AD 135, as by then it had been permanently moved to Syria.97

(6) Ala Augusta Germaniciana was probably raised and stationed in the East during the reign of Augustus. This is the conclusion to be drawn from its name Germaniciana which is best explained as a special reward for its services to Germanicus during his mission to the East in AD 18/19.88 It is attested as part of the army of Cappadocia et Cilicia under Nero89 and it is also known from the careers of some of its commanders, all from Asia Minor.90 The ala is listed in the diplomas of 94 and 101 but it did not take part in Arrian’s expedition of c.

The Cohorts (1) Cohors I Apamenorum sagittariorum equitata. It is unknown when the cohort joined the army of Galatia/Cappadocia but it appears to have been in

79 Petulescu / Popescu 2004. The authors, however, suggested II Claudia Gallorum to have been a mistake for I Claudia Gallorum (Capitoniana). 80 The otherwise earliest attestation of ala II Gallorum in Cappadocia is AE 1925: 44 with PME 6: 18 (ca. AD 110). 81 Arrian, Ekt. 9 only speaks of ¹ œilh ¹ Galatik». Lower Moesian ala I Claudia Gallorum Capitoniana could also, on occasion, simply be called ala (I) Gallorum (RMD IV 209; AE 1967: 419. IDR 3,4: 326f.). 82 AE 1925: 44. Cf. PME C 143 and PME 6: 18. 83 IGRR III 272 dated to the Flavian period by Holder 1980: 272. That date is, however, not entirely certain. 84 Cf. ILS 8873 from Thessalonica. 85 AE 1972, 70 = AE 1992: 689. PME B 14. 86 Correctly recognized by Ritterling 1902: 362. 87 IGRR IV 964 = ILS 8865 (2nd c.). ND Or. XXXVIII.24. 88 Birley 1988: 378. 89 AE 1914: 128 = AE 1966: 472 (Side). 90 CIL III 6821 & 6831 = ILS 2708. CIL III 6822 = AE 1888: 177 (Pisidian Antioch). AE 1914: 128 = AE 1966: 472 (Side). AE 1972: 592 = IK 13: 837 (Ephesos). AE 1973: 539 (Attaleia).

91

AE 1972: 592 = IK 13: 837 (Ephesos). ND Or. XXXVIII.26. Speidel 1984: 287f. Lörincz 2001: 23. See esp. Eck / Pangerl 2005c: 56ff. and Weiss 2006: 264 for the identity of ala I praetoria. 93 Cf. Speidel 1992: 212ff. Lörincz 2001: 149f. and cf. Hartmann / Speidel 2003: 117f. 94 CIL XII 1357 = ILS 2709 = AE 1992: 1208. Cf. Pflaum 1978: 213ff. no. 5. PME S 8. 95 Diplomas: AD 54: CIL XVI 3. For the diplomas from Domitian to Antoninus see now Weiss 2006: 249ff. See also Eck / Pangerl 2006: 205ff. 96 For these see Eck / Pangerl 2006: 221ff. Cf. also Weiss 2006: 298. 97 This is now proven by a new fragmentary diploma for a soldier of the exercitus Syriacus from AD 129-134: Matei-Popescu forthcoming. Cf. also Weiss 2006: 282. 298. For at least part of its time in Syria this unit was stationed at Palmyra: Speidel 1984: 168f. Gebhardt 2002: 294. It was also part of Lollianus’ cavalry force in c. AD 165: CIL III 600 = ILS 2724. Saxer 1967: no. 64. Kennedy 1997: 69-81. Holder 2003: 116. See now esp. Weiss 2006: 294ff. 92

79

THE LATE ROMAN ARMY IN THE EAST Cappadocia in the early 80s.98 It is attested there in AD 101 and during the early second century,99 but it was not amongst Arrian’s forces against the Alani in c. AD 135. Before AD 144 cohors I Apamenorum was moved to Egypt where it remained for the rest of its history.100

date of his gravestone, it is possible that cohors Augusta was a citizen unit set up in Galatia under Augustus and temporarily stationed in Pontus / Bithynia.110 If this cohors Augusta (civium Romanorum?) is indeed identical with our cohors I Augusta civium Romanorum, it joined the army of Galatia / Cappadocia before AD 94. Its later history is unknown.

(2) Cohors Apula civium Romanorum was raised in Italian Apulia under Augustus and transferred to the East in the early principate.101 Its first garrison place was in or near Alexandria Troas in the province of Asia102 before it was moved south to Side in Pamphylia still early in the first century AD.103 Here it became part of the Galatian army. For the remainder of its history cohors Apula belonged to the army of the north-eastern frontier and was eventually moved into Cappadocia. Although we find it under Arrian’s command in c. 135 AD,104 the cohort is not listed in the new diplomas for Galatia et Cappadocia. In the late fourth century AD (if not already well before) cohors Apula was stationed at Ysiportus = Hissou Limen on the Black Sea where Arrian in c. AD 131 had inspected a garrison of footsoldiers and some horsemen.105

(4) Cohors Aurelia civium Romanorum. The name of this otherwise unknown unit was read on a candelabrum which was found during the excavations at Apsaros on the Black Sea.111 The name Aurelia suggests the reigns of Antoninus Pius or Marcus Aurelius as the time when the cohort was set up.112 The cohort could therefore have joined the Cappadocian army for the bellum Parthicum of AD 161-166.113 It is of course also possible that cohors Aurelia came to the Black Sea at a later date in the second or early third century. Nothing else is known of its history. (5) Cohors I Bosporanorum milliaria equitata sagittaria, raised from the Crimean Bosporani under Augustus,114 belonged to the mid-first century garrison of Galatia.115 It is listed in the new diploma of AD 101 and was in Cappadocia under Arrian’s command in c. AD 135.116 Equestrian careers also show this unit to have belonged to the Cappadocian army of the second century AD and it is still attested there at the end of the fourth century.117 So far, the Notitia Dignitatum is the only source to show the milliary status of cohors I Bosporanorum.

(3) Cohors I Augusta civium Romanorum is listed on the diploma for Galatia / Cappadocia of AD 101 and has plausibly been restored for those of 94 and 100. No other evidence has yet come to light to show that this cohort was part of the Cappadocian army. A cohors Augusta (I) is known to have belonged to the Roman army of Judea since the early principate. It was stationed in the territory of king Agrippa II where its latest known record dates to AD 84 or 89.106 The unit is not listed in the constitutions for the army of Judaea in AD 86, 87 and 90107 and may therefore have left the province by then.108 Yet, it is perhaps more likely to identify the Cappadocian cohort with another cohors Augusta, known from a recently published gravestone of one of its soldiers at Sinope.109 This soldier, C. Octavius C.f. Vel(ina tribu) dom(o) Savatra, was a mil(es) cohort(is) August(ae) et ea[e]dem cohortis sesqueplicarius, thus a Roman citizen recruited from the southern Galatian city of Savatra. He died at some time during the early first century at Sinope while still a soldier. Considering his citizen status and the early

(6) Cohors I Claudia(na?) equitata (milliaria?) is listed in the new diploma for Galatia / Cappadocia of AD 101.118 The early history of this unit is unknown though its name and the fact that it was in Cappadocia suggests that it shared a common origin with its sister unit cohors II Claudia.119 It was not part of Arrian’s forces against the Alani but the career of one of its commanders shows that the cohort was [in Cappa]docia in the second half of 110 The only known member of the Cappadocian cohors Augusta cR, the recepient of the diploma of AD 100, certainly was a Roman citizen: the former centurion Q. Antonius Q.f. T[---]. Cf. Eck / Pangerl 2005: 238. Cf. ala I Augusta Gemina colonorum for another citizen unit set up in sourthen Galatia under Augustus. 111 Mamuladze et al. 2002: 37. The editors transcribed: C. COH AVR C.R. and suggested the first letter to be the abbreviated proper name of the owner. It is, however, perhaps better to understand the first letter as the abbreviation for centurio and to read c(enturionis) coh(ortis) Aur(eliae) c.R. The letters AVR on the candelabrum are certain, as proven by a photograph kindly supplied by L. Aslanishvili. 112 For cohortes Aureliae set up under Antoninus Pius see Weiss / Speidel 2004: 253ff. and 262. 113 Major troop movements had already begun during the last years of the reign of Antnoninus Pius: ILS 1076 and Weiss 2007. 114 ILS 9503. Cf. PME C 81. 115 ILS 9499 = IK 13: 715. Cf. PME A 85. AE 2001: 1918. Speidel 1984: 280 and 300. 116 Arrian, Ekt. 3. 18. Ritterling 1902: 363f. 117 AE 1991: 1749 & 1750 (early 2nd c. Cf. PME H 21a). Olympia 447 (early - mid 2nd c. Cf. PME C 136). ILS 8864 = IGRR IV 1323 (2nd c. Cf. PME F 82). ND Or. XXXVIII.29. 118 For the name Claudiana see below cohors II Claudiana. 119 Speidel 1984: 106f.

98 This conclusion can be drawn from the equestrian career of C. Nasennius Marcellus senior: ILS 2741 = CIL XIV 171. Cf. Holder 2002: 288f. 99 CIL VI 3654 = AE 1974: 226 = AE 1977: 183 (coh(ors) I Apamen(orum) sa(gittariorum) in Cappad(ocia), early 2nd c. PME V 28). ILS 2724 add. (early/mid-2nd.c PME V 17). Alston 1995: 174 and Holder 2003: 116. 100 BGU III 729 (AD 144). P.Lond. II 177 (AD 145). CIL XVI 184 (AD 156/61). RMD III 185 (AD 179). ILS 2724. 2741. etc. and finally: ND Or. XXXI.60. 101 AE 1966: 478, cf. 1977: 7 (1st c. AD). Speidel 1984: 91ff. 279f. 102 IK 53: 34 = AE 1973: 501 = AE 1975: 806 = AE 1978: 790. 103 IK 44: 202 = AE 1966: 478 = InscSide 155. 104 Arrian, Ekt. 7. 14. 105 ND Or. XXXVIII.34. Arrian, Peripl. 3. 106 For the history and records of this cohort see Speidel 1992: 228ff. 107 Cotton / Eck / Isaac 2003: 17-31. 108 Unless the cohort is identical with cohors I Augusta Thracum: Speidel 1992: 230 with n. 16. 109 IK 64: 124.

80

MICHAEL A. SPEIDEL: THE DEVELOPMENT OF THE ROMAN FORCES IN NORTHEASTERN ANATOLIA the second century.120 That officer carried the title tribunus and was in command of cohors I Claudia as his secunda militia. Not long before, in the 160s, our cohort had been the prima militia of a praefectus.121 Therefore, cohors I Claudia equitata appears to have been raised to milliary status after the 160s.122 However, nowhere in our sources is it as yet explicitly called milliara. The cohort remained on the northeastern frontier throughout the third and fourth centuries.123

(9) Cohors III Augusta Cyrenaica sagittariorum equitata was in Moesia during the early principate and may therefore have come to the East together with legio V Macedonica in AD 62.133 Here it appears to have joined the Syrian army before it was moved farther north.134 If so, cohors III Augusta Cyrenaica must have left Syria before A.D. 88 as it is not mentioned in any of the Syrian diplomas of that year.135 Thus, it may have arrived from Syria either in late A.D. 70 together with legio XII Fulminata or in c. AD 75/76 with legio XVI Flavia firma. At any rate, it is now clear that the cohort belonged to the army of Galatia/Cappadocia in AD 101. Arrian in c. AD 135 had under his command a cohort of Kurhna‹oi with horsemen and archers.136 Cohors III Cyrenaica is known to have had archers in its ranks,137 and as cohors I Cyrenaica equitata was probably the garrison of Ancyra in Galatia, the third cohort must be meant. This is confirmed by two 2nd century careers of Roman knights who were promoted from the command of this cohort to a tribunate in Cappadocian legio XII Fulminata at Melitene.138 There are no records of this unit from the 3rd or 4th centuries AD and it is not listed in the Notitia Dignitatum.

(7) Cohors II Claudia(na) belonged to the army of Galatia / Cappadocia in AD 101 as shown by the new diploma of that year. Though this unit is generally referred to as cohors II Claudia it is in two cases called Claudiana.124 If this was indeed the unit’s original name it was called after one of its former commanders and was thus probably raised during the early principate.125 Its name and the fact that it was in Cappadocia suggests that it shared a common origin with its sister unit, cohors I Claudia(na?) equitata (cf. above). Where it was raised and first stationed and when it was moved into Galatia / Cappadocia remains unknown. It was not with Arrian in c. AD 135 even though it was part of the Cappadocian army at that time.126 A third century papyrus letter suggests Apsaros on the Black Sea to have been the garrison place of our cohort.127 Two tile stamps found there, reading COH II [---] and [C]OH II [---] may therefore have been produced by cohors II Claudia.128 If so, the unit was one of the five numeri tendentes in Ponto, Absaro129 all (or most?) of which Arrian appears to have left at their garrison place in c. AD 135.130 The later fate of the cohort is unknown.

(10) Cohors I Germanorum milliaria equitata was with Arrian on his march against the Alani in c. AD 135.139 That is the earliest known record of this unit in Cappadocia. It is unknown when the cohort joined the army of Cappadocia and where its earlier garrison place may have been though it might have come to the East at some time during the early principate.140 Our sources show cohors I Germanorum to have been in Cappadocia and milliary from the second century to the end of the fourth.141

(8) Cohors I Augusta Cyrenaica equitata belonged to the army of northeastern Anatolia during the second century as shown by inscriptions from Ankyra and Ikonion.131 The cohort appears to have been stationed in or near Ancyra and thus must have belonged to the army of Galatia proper.132 That would explain why the cohort is listed on the diploma of 101 but did not take part in Arrian’s march against the Alans in c. AD 135. Neither its earlier history nor its later fate are known.

(11) Cohors I Hispanorum belonged to the mid-first century garrison of Galatia.142 It was moved to Egypt before AD 83 if it is indeed identical with cohors I Hispanorum of the Egyptian military diploma of that year.143 (12) Cohors II Hispanorum equitata civium Romanorum. An inscription from Ancyra and an equestrian career from the reign of the emperor Domitian (AD 81-96) show this cohort to have been part of the army of Galatia/Cappadocia in the second half of the first century

120

CIL IX 2958: tribuno [coh. I Cla]udia equitata [in Cappa]docia. The first Claudian cohort is likely to be meant here, as cohors II Claudia(na) is not known to have had horsemen in its ranks. Ritterling 1902: 364. Alföldy 1968: 197f. no. 105. PME O 2. 121 IScM 2,5,10 = AE 1934: 107 = AE 1940: 101 (Capidava, Moesia Inferior, from the 160s.). Cf. PME I 120. 122 Cichorius 1900: 273. 123 ND Or. XXXVIII.36. 124 IGRR IV 642 with Speidel 1984: 106. ChLA XI 477. 125 Speidel 1984: 101ff. 126 AE 1977: 802. Cf. Speidel 1984: 106. PME E 3. 127 P.Berol.inv. 25050 = (AD 268-270) which gives the unit’s name as c]oh II Claudiana [-. 128 Mamuladze et al. 2002: 33-39. 129 Arrian, Peripl. 6. ILS 2660. 130 Speidel 1992: 204f. 131 Bosch 1967: no. 113 and 115 (Ankara). Ramsay / Ramsay 1928: 183 (Iconium). Cf. Speidel 1984: 281f. 132 Speidel 1984: 300. Mitchell 1993 I: 121.

133

CIL III 647 = ILS 2538. Ritterling 1902: 364 with n. 2. CIL XI 7554 = ILS 9195 = AE 1952: 34. Cf. PME P 111 (under Vespasian). 135 Cf. Weiss 2006: 297f. 136 Arrian, Ekt. 1. 3. 14. 18. 137 CIL XI 7554 = ILS 9195 = AE 1952: 34. 138 AE 1971: 367 (2nd half 2nd c.). ILS 2748 (2nd half 2nd c.). Cf. PME Inc. 40 and A 31. Speidel 1984: 295 n. 39. 139 Arrian, Ekt. 2. Ritterling 1902: 364f. 140 Alföldy 1968: 85. Speidel 1984: 283. 141 ILS 8868 = IGRR III 1420 (1st half 3rd c. PME Inc. 30). ILS 8851 = IGRR I 623 (1st half 3rd c. PME A 21). ND Or. XXXVIII.30. 142 IK 13: 715 (Ephesus). AE 1961: 17 (Olbasa, Pisidia) with Speidel 1984: 281 and 300. 143 CIL XVI 29. Cf. Weiss / Speidel 2004: 259. 134

81

THE LATE ROMAN ARMY IN THE EAST AD.144 This goes well with its appearence in the new diploma of AD 101. The cohort was not part of Arrian’s task force in c. 135 but it is known from equestrian careers to have remained in Cappadocia well into the third century.145 It was, however, no longer in Cappadocia by the time of the Notitia Dignitatum.

Alani in c. AD 135.152 Its later fate is unknown (cf. below cohors milliaria eq. c.R.). (15) Cohors I Ituraeorum milliaria equitata sagittaria.153 It is particularly difficult to identify this unit in our epigraphic sources. A cohors I Ituraeorum is attested in Syria in AD 88 and in AD 93 but does not appear on any of the hitherto known Syrian diplomas of later date.154 If this is the same unit, it was then moved north into Galatia et Cappadocia where it is listed in the diploma of AD 101 now, however, as a milliary cohort.155 From here, cohors I Ituraeorum (or perhaps a large detachment) may have been sent to fight in Trajan’s second Dacian war of AD 105/106, as a cohors I Ituraeorum (though not styled milliaria) is known from Dacian diplomas in AD 109 and 110.156 If so, it then came back to Cappadocia probably for Trajan’s Parthian war.157 Unfortunately, these identifications must remain uncertain. The cohort is next known to have fought against the Alani under Arrian in c. AD 135.158 The later fate of cohors I Ituraeorum milliaria remains unknown.159

(13) Cohors I Italica milliaria is listed on the new Cappadocian diploma of AD 101. It is surprising to find two cohorts of nearly the same name in Cappadocia. Arrian in c. AD 135 speaks of only one Italian cohort which was under the command of an officer named Pulcher.146 Also, there is no other evidence which would confirm beyond any doubt the simultaneous existence of two cohortes Italicae within the Cappadocian army. One might therefore assume a misunderstanding by the modern copyist of the diploma of AD 101. However, an inscription set up in Pisidian Antioch in the mid-first century by T. Volumnius Varro, a Roman knight, records his prima militia as praef(ectus) coh(ortis) [---] Italic(ae).147 All known commanders of cohors I and II Italica c.R. voluntariorum, however, were tribuni.148 Hence, if the title praefectus of the Antioch inscription is not simply due to its early date then there may indeed have been a second cohors I Italica in the Cappadocian army.

(16) Cohors I Lepidiana equitata civium Romanorum bis torquata came from Lower Moesia where it is well attested until AD 127.160 As the army of Lower Moesia of the following years is rather well known it seems likely that cohors I Lepidiana left the Danube not long after AD 127.161 L. Calpurnius Valens, optio of cohors I Lepidiana was buried by his wife in Asia, either at Smyrna or near Magnesia ad Sipylum at some time in the second century.162 Whether this means that the cohort was in Asia for a prolonged stay or whether it was on its way farther east when the optio lost his life remains uncertain. It is clear, however, that the cohort was not among the units of Arian’s field army in ca. AD 135. By AD 199 cohors Lepidiana had become part of the Cappadocian army and was involved in a building project near Melik

(14) Cohors I Italica milliaria voluntariorum equitata civium Romanorum. It is not known when this cohort came to Galatia/Cappadocia but it appears to have already been there in the early 80s of the first century AD.149 It may have been stationed in Syria earlier where its sister unit, cohors II Italica c.R. voluntariorum milliaria is well attested.150 By the new Capadocian diplomas cohors I Italica milliaria voluntariorum cR is now on record as a unit of the army of Galatia / Cappadocia in AD 94 and 101 and equestrian careers show that it was there during the first half of the second century as well.151 Hence, it was this Italian cohors equitata which took part in Arrian’s march against the

152

Arrian, Ekt. 3. 9. 13. Ritterling 1902: 359ff. 365. Cf. Dabrowa 1986: 221-230. Holder 2003: 117 and 142. CIL XVI 35. Weiss 2006: 284f. 297f. During the early to mid-first century AD this unit may have been stationed in Moesia: IGRR III 1139. AE 1926: 80. The cohort would then have been moved back to the East possibly together with legio IV Scythica in AD 56. 155 Evidence for its milliary status also comes from AE 1907: 50 = IGRR I 1462 if this inscription indeed refers to our unit. 156 RMD III 148 (AD 109). CIL XVI 57 (AD 110). See Strobel 1984: 136. Weiss 2006: 284. For the splitting of units which then both kept the original name see Weiss 2006: 278 with examples. 157 Cohors I Ituraeorum is, at any rate, already missing from the Dacian diploma RMD IV 226 = RGZM 16 of AD 114. 158 Arrian, Ekt. 1. 18. Ritterling 1902: 367. 159 See however O. Seek’s suggestion in his edition of the Notitia Dignitatum that cohors I Ituraeorum may have been meant in ND Or. XXVIII.42 (Aegyptus). 160 MacDonald / Mihaylovich 2002: 225ff. with Eck 2005a: 185ff. (AD 97). CIL XVI 45 (AD 99). CIL XVI 58 (AD 111; for the date see RMD IV p. 381). RMD IV 235 (AD 125). RMD IV 241 (AD 127). Cf. Roxan / Eck 1997: 197. Lörincz 2001: 37. Matei-Popescu 2004: 215. 161 CIL XVI 78 (AD 134). CIL XVI 83 (AD 138). RMD IV 253 (AD 138). RMD IV 265 (AD 138/142). RMD III 165 (AD 145). RMD IV 270 (AD 145/146). AE 2001: 2160 (AD 155). RMD I 50 (AD 157). RMD II 111 (AD 161). 162 CIL III 12251 = ILS 2590. Cf. Christol / Drew-Bear 1995: 62. Eck 1997: 190f. n. 22. 153 154

144

Inscription: CIL III 6760. Holder 1980: 312 dates this inscription to the first half of the first century. Cf. also Speidel 1984: 282. Mitchell 1993 I: 121. Equestrian career: CIL II 2213 (PME P 101). Cf. Holder 2003: 117 with n. 56. 145 CIL IX 2649 = ILS 2732 (praef(ecto) coh(ortis) [-] / Hispanor(um) in Cappadoc(ia). 2nd. c. Cf. PME S 36). ILS 8851 = IGRR I 623 (1st half 3rd c. Cf. PME A 21). The same unit is probably also meant in CIL II 2213 (’ eq. late 1st. c. Cf. PME P 101), ILS 9471 (eq. c.R. early 2nd. c. Cf PME A 4) and possibly in CIL XIII 1680 = ILS 1390 (late 2nd/early 3rd c. Cf. PME A 127). CIL IX 5066 may also mention the same cohors II Hispanorum equitata cR, cf. Ritterling 1902: 365. Holder 2003: 142. 146 Arrian, Ekt. 3. 13. Ritterling 1902: 365f. with n. 3. 147 AE 1941: 142. Cf. PME V 130. 148 See PME 6: 92. 149 This conclusion can be drawn from the equestrian career of C. Nasennius Marcellus senior: ILS 2741 = CIL XIV 171 (early 2nd c.), cf. Holder 2002: 288f. 150 Ritterling 1902: 366. Speidel 1992: 224ff. Weiss 2006: 297f. For the unit’s full name cf. also CIL XI 6117: ... trib(uno) coh(ortis) mil(liariae) Italic(ae) volunt(ariorum) quae est in Syria ... 151 ILS 2741 = CIL XIV 171 (early 2nd c.), cf. Holder 2002: 288f. CIL VI 3654 = AE 1974: 226 = AE 1977: 183 (coh(ors) I Ital(ica) ’ volunt(ariorum) cR in Cappad(ocia), 1st half 2nd c.). Cf. PME V 28.

82

MICHAEL A. SPEIDEL: THE DEVELOPMENT OF THE ROMAN FORCES IN NORTHEASTERN ANATOLIA Sherif.163 By the end of the fourth century it was stationed at Caene-Parembole on the coast of the Black Sea, some 60 km east of Trapezus.164

Numidarum with an unknown earlier history came to Lower Pannonia towards the end of the Marcomannic wars.174 However, none of this is necessarily connected with the Cappadocian cohort.

(17) Cohors milliaria equitata civium Romanorum. The early history of this unit is unknown. It was probably already in Cappadocia, though perhaps not yet milliary, when Tib. Claudius Pius came to take command at some time during the early second century.165 It is, however, not listed among the units which made up Arrian’s field army in ca. AD 135. The cohort, now clearly milliaria, is next known from an inscription of AD 177 as part of a detachment which included soldiers from both Cappadocian legions and which was sent to Armenian Kainepolis after 163 in order to stabilize the rule of Armenia’s new king Sohaemus.166 Perhaps the possibility should therefore be considered that this unit is identical with cohors I Italica milliaria voluntariorum civium Romanorum. The Kainepolis detachement probably returned to Cappadocia before the end of Commodus’ reign.167 This unit is not mentioned by any later sources.

(19) Cohors III Ulpia Petraeorum milliaria equitata sagittaria was raised by Trajan in Arabia and soon moved to Cappadocia to fight in his Parthian war.175 It also fought with Arrian against the Alani in c. AD 135176 and remained part of the Cappadocian army until the end of the fourth century.177 (20) Cohors I Raetorum equitata was probably stationed in Roman Syria during the early principate before it came to Moesia, presumably together with Vespasian’s forces on their way to Rome.178 The unit is still attested in Moesia in AD 75 and in AD 92.179 Before the turn of the century, cohors I Raetorum was moved back to the East and became part of the army of Galatia / Cappadocia. It is listed in the new diploma of AD 101 and it took part in Arrian’s march against the Alani in c. 135.180 Before AD 148, however, it was transferred to Asia where its new permanent garrison-place was at Eumenia in Phrygia.181

(18) Cohors I Numidarum (equitata?) sagittaria was in Syria in AD 88.168 As it is not listed in any of the later Syrian diplomas169 it may have been transferred to Cappadocia fairly soon after AD 88. It is, however, missing from the new diploma of AD 101. The cohort took part in Arrian’s march against the Alani in c. AD 135 and it is also known by an equestrian career to have belonged to the second century army of Cappadocia.170 Emil Ritterling concluded from Arrian’s remark oÀ Àppe›w oÀ o¿ke›oi (ekt. 4) that the cohort was partmounted.171 As yet, however, further supporting evidence for this is lacking. The later history of our cohort remains unclear. Recently however, it has become more certain that our unit is not to be identified with cohors I Flavia Numidarum equitata which was in Lower Moesia throughout the first half of the second century until it moved to Lycia / Pamphylia between AD 157 and 166/167.172 Scraps of the papers of a cohors I Nomidarum equitata dating to the first quarter of the third century have been found in Egypt.173 A cohors milliaria

(21) Cohors IIII Raetorum equitata belonged to the army of Moesia superior until the early second c. AD.182 As a part of the contigent from that army, the emperor Trajan included this unit into the expeditionary forces for his Parthian war, as is now shown by a new military diploma of AD 115.183 After the war, cohors IIII Raetorum remained in Cappadocia. It was under Arrian’s command in c. AD 135 and it was still listed in the Notitia Dignitatum.184 (22) Cohors III Syrorum sagittariorum is known only from the new diploma of AD 101 which was made out for this unit included dromedarii which may indicate that it had been stationed for some time in an area with a climate suitable for dromedaries. Hence, if this unit is identical with Cappadocian cohors I Numidarum it had probably been transferred to a province farther south before the early third century. 174 Lörincz 2001: 40. 175 ILS 9471 (early 2nd. c. PME A 4). ILS 9013 = AE 1908: 200 (early 2nd. c. PME A 167). AE 1983: 998 = AE 1993: 1787 = IAM 2: 307: ... (trib(uno) mil(itum) coh(ortis) III Ulp(iae) m(illiariae) / Petraeor(um) electo et retento / ad cens(us) excipiend(os) in partem / provinc(iae) Arm(eniae) item Capp(adociae) c. AD 140/144. PME S 86). Cf. Ritterling 1902: 368. For the raising of the cohortes Petraeorum see Eck / Pangerl 2006: 227f. 176 Arrian, Ekt. 1. 177 ND Or. XXXVIII.27. 178 Overbeck 1981: 274. Speidel 1984: 278f. Matei-Popescu 2004: 222f. 179 RMD I 2 (AD 75). AE 2003, 1548 (AD 92). The unit is, however, missing from the diplomas CIL XVI 22 and RMD IV 208 (AD 78). CIL XVI 39 (AD 93). RMD I 6 (AD 96) etc. 180 Arrian, Ekt. 1. 181 RMD II 100 (AD 148). Ritterling 1902: 369. Ritterling 1927: 31f. Overbeck 1981: 276. Speidel 1984: 278f. Christol / Drew-Bear 1995: 63ff. PME A 23 bis. 182 CIL XVI 39 and AE 1998: 1616 (AD 93). RMD I 6 (AD 96). CIL XVI 46 (AD 100). Ritterling 1902: 369. Strobel 1984: 141f. 183 Eck / Pangerl 2005c. 184 Cf. AE 1991: 1749 & 1750 from the early 2nd. c and AE 1915: 49 = IK 44: 219 (Side, mid 2nd c.). Arr., Ekt. 1. ND Or. XXXVIII.28.

163 AE 1908: 22. Melik Sherif was identified as Carsaga (between Arauracos and Analiba) by Mitford 1980: 1190. Idem 1998: 266. 164 ND Or. XXXVIII.35. Zuckermann 1991: 530. 165 AE 1933: 270 = Habicht 1969: 68ff., no. 30. Cf. Holder 2002: 292. Holder 2003: 117 (prior to 117). 166 ILS 9117 = AE 1910: 161 = Saxer 1967: no. 275. For the date, AD 177, see Piso 1993: 112-116. For the Kainepolis detachement see also Dio LXXI.3. 167 Cf. Ritterling 1925: 1754. Saxer 1967: 93f. Mitford 1980: 1205f. Wheeler 2000: 300f. 168 CIL XVI 35. Weiss 2006: 263f. 169 RMD I 4. RMD I 5. RMD IV 214 = RGZM 6. Weiss 2006: 285. 297f. 170 Arrian, Ekt. 3. 18. IGRR IV 964 = ILS 8865 (cf. PME F 109). Ritterling 1902: 368. 171 Ritterling 1902: 368. The cohorts I and II Flavia Numidarum were, at any rate, both equitatae: cf. only CIL XVI 128. AE 1965: 347. 172 Ritterling 1902: 368. Roxan 1997a: 292f. RMD IV 222 with n. 5 on page 433f. See also RMD I 50 (AD 157). RMD I 67 (AD 166/167). 173 P.Mich. VII 455b = RMR 52. A cohors I Numidarum is otherwise not known from Egypt and it is therefore not necessary to assume that it permanently belonged to the Roman army of that province. However,

83

THE LATE ROMAN ARMY IN THE EAST a soldier of this unit from Cilician Anazarbos. This may point to a garrison place in Asia Minor or in Syria at the time the soldier was recruited (i.e. c. AD 75/76). If, however, our cohort is identical with cohors III sagittariorum, a former garrison place in Syria becomes likely, as this cohort is associated by an equestrian career with legio IV Scythica during the early years of Nero’s reign. After a tribunate in legio IV Scythica and the prefecture of cohors III sagittariorum its commander, a Roman knight from Bithynian Apamea, was appointed to the Senate by Vespasian.185 Hence, cohors III sagittariorum could well have belonged to Vespasian’s forces in the bellum Iudaicum.186 If cohors III sagittariorum is indeed identical with the Cappadocian cohors III Syrorum sagittariorum then it may have been transferred to Cappadocia at some time after AD 75/76 and before AD 88, as the unit does not appear in any of the Syrian diplomas of that or of later years.187 Our cohort may have eventually been stationed at Apsaros, where a tile with the fragmentary stamp [---] SAGI was found.188 Nothing is known of its later fate.

Hispanorum (as well as possibly cohors I and II Claudia) were also early additions to the Roman army in northeastern Anatolia.

(23) Cohors II Augusta Thracum belonged to the army of Galatia / Cappadocia in AD 101. Its earlier history remains unknown. During the second and early third century it is on record as part of the army of Lower Pannonia.189 The earliest attestation there dates to AD 135.190 Cohors II Augusta Thracum may therefore have been moved to the west by Hadrian in the aftermath of Trajan’s Parthian war (perhaps in AD 123).191

The reign of the emperor Trajan (AD 98 – 117) brought further changes to the Cappadocian army. Cohors I Ituraeorum may have been sent to fight in his second Dacian war of AD 105/106. If so, it remained in Dacia until at least AD 110. Cappadocia’s army was, however, throroughly reshaped by Trajan’s Parthian war (AD 114 117). For this war a great field army including many units and detachments from the West was set up and moved into Cappadocia. Part of that army was a detachment of horsemen from Moesia inferior and Dacia. It was led to the East under the command of the equestrian prefect L. Paconius Proculus.193 Cohors I Ituraeorum too will then have returned from Dacia. From Moesia superior cohors IV Raetorum and eight other auxiliary units (or detachments thereof) joined the emperor’s field army.194 Furthermore, a number of new auxiliary units set up by the emperor Trajan also came East to fight in his Parthian war. In AD 114 Trajan joined his army which was awaiting him at Satala, the garrison place of legio XVI Flavia firma and took it into Greater Armenia.195 After the war, parts of Trajan’s field army permanently remained in Cappadocia: ala II Ulpia Auriana, ala I Ulpia Dacorum, cohors (milliaria) equitata cR, cohors I Ituraeorum, cohors III Ulpia Petraeorum, cohors IIII Raetorum.

The Flavian rulers brought along profound changes to the organization of the northeastern frontier. Vespasian upgraded Cappadocia to a consular province with two legions sent from Roman Syria. Our sources, in particular the new diplomas, show that a number of cohorts were also moved from Syria into Cappadocia during the course of the 70s and 80s of the first century AD: cohors I Italica voluntariorum (during the 70s?), cohors III Augusta Cyrenaica (before AD 88), cohors I Ituraeorum and cohors I Numidarum (both shortly after AD 88), and possibly cohors III Syrorum (after AD 75? and before AD 88?). From Moesia inferior cohors I Raetorum was transferred to Cappadocia shortly after AD 92, as was possibly ala II Gallorum. Cohors I Hispanorum, on the other hand, appears to have left the army of northeastern Anatolia and may have been moved to Egypt before AD 83.

Transfers to and from the Cappadocian Army The early history of the exercitus Cappadocicus after the country was transformed into Roman provincial territory in AD 17/18 remains poorly documented. There is no doubt, however, that it received a garrison of auxiliary units.192 Many of these were of eastern origin: Ala I Augusta colonorum and cohors I Augusta (cR) were raised in Galatia under Augustus. Ala Augusta Germaniciana, too, as proven by its name, was an Augustan creation of eastern origin, as was cohors I Bosporanorum, raised from the Crimean Bosporani. Finally, cohors I Apamenorum was set up from the citizens of Bithynian or of Syrian Apameia and joined the Cappadocian army at an unknown date before the Flavian dynasty. Other units were of European origins and were moved in from the West: Cohors Apula, again raised under Augustus, was transferred from Italy to the East during the early principate. The two cohorts I and II

In AD 123 the emperor Hadrian (AD 117 – 138) hurried to the East to settle a Parthian threat. A diplomatic summit with the Parthians was organized on the Euphrates. The Syrian army will have been present there, and possibly the Cappadocian army also sent some troops. The senator Claudius Quartinus had led legionary detachments from Egypt and Judaea to the Euphrates and the emperor came with the Praetorians and the Horse

185 CIL III 335 cf. 6991 & 14188,1 = IK 32: 2 = AE 1982: 860. Cf. Halfmann 1979: 115 no. 18. PME L 50bis. Eck 1981: 242ff. 186 Halfmann 1979: 72. 187 Weiss 2006: 297f. 188 Mamuladze et al. 2002: 35 with fig. 3. Perhaps one might then restore [coh. III] sagi(ttariorum). 189 Lörincz 2001: 43. Roxan 1999: 263. 190 RMD IV 251 = RGZM 27. 191 Lörincz 2001: 87f. 192 Tac., Ann. 13,8. Cf. Tac., Ann. 12,49 and 15,6. For the provincialization of Cappadocia see now Speidel (forthcoming).

193

ILS 2723. Eck / Pangerl 2005c. 195 Dio LXVIII.19.1ff. Cf. Speidel 2002: 36ff. 194

84

MICHAEL A. SPEIDEL: THE DEVELOPMENT OF THE ROMAN FORCES IN NORTHEASTERN ANATOLIA Guards.196 The matter was setteled by diplomacy and Hadrian continued his journey by inspecting the army of Cappadocia and dealing with the affairs of neighbouring allied kingdoms. In particular, he ‘allowed the Armenians to have a king’.197 There are no reports of military action. However, his settlement of the political situation in the Northeast in the aftermath of Trajan’s Parthian war appears to have led to a redistribution of some units of the Cappadocian army. Thus, several units may have left the northeastern Anatolian frontier that year: Ala I Thracum Herculana (to Syria), cohors I Apamenorum (to Egypt), cohors II Thracum (to Lower Pannonia).

Danube were sent to reinforce the exercitus Cappadocicus under the command of an African senator, P. Iulius Geminius Marcianus.205 A large field army was sent to the East and in AD 162 Lucius Verus came to lead the war in person. After the war, cohors Aurelia and possibly some other auxiliary troops remained in Cappadocia to replace the losses. Before the end of the third century we can name only one more addition to the army of Cappadocia: ala I praetoria. Cohors I Augusta cR, cohors I Italica milliaria voluntariorum cR and cohors I Ituraeorum disappear from our list of Cappadocian units during the second century. Permanent transfers to other provinces or losses of the third and fourth century include ala Augusta Germaniciana, cohors Aurelia, cohors II Claudia(na), cohors I and III Augusta Cyrenaica, cohors II Hispanorum, cohors milliaria equitata cR, cohors I Numidarum and cohors III Syrorum.

In AD 129 Hadrian was back in Cappadocia, again inspecting the army and dealing with allied kings.198 King Pharasmanes II of Iberia, in particular, appears to have given Hadrian reason to negotiate. In the end, the king of Iberia received presents and a ‘cohort of fifty men’.199 Cohors I Lepidiana may have been transferred from Lower Moesia to Northern Cappadocia as a result of the emperor’s visit. Pharasmanes, at any rate, was later accused by king Vologaeses of Parthia of having incited the Alans to invade Albania and Media. Parthian presents and Arrian’s march put an end to the Alan incursion in ca. AD 135.200 Not long thereafter, cohors I Raetorum was transferred to the province of Asia.

In his study of Arrian’s expeditionary force, Emil Ritterling concluded that the Cappadocian army suffered few changes between the reign of the emperor Hadrian and the end of the fourth century.206 The Notitia Dignitatum lists fourteen units (including the two legions XII Fulminata and XV Apollinaris) which had been part of the Cappadocian army before the end of the third century. Eleven of these units had fought with Arrian in ca. AD 135 and two others also appear to have been in the province at that time (cohors I Claudia and cohors I Lepidiana). Also, of those units which belonged to the army of Galatia et Cappadocia at the end of the first century, seven or eight survived until the end of the fourth century. Yet, in the light of the new diplomas and of the epigraphic evidence Ritterling’s view requires some modification. For we now also know of at least nineteen units (including legions VIIII Hispana and XVI Flavia firma) which had once been part of the Cappadocian army of the second century but had been transferred or destroyed before the end of the fourth century. Thus, the fate of the Roman army in northeastern Anatolia differed less than hitherto assumed from that of other frontier armies of the Roman empire.207

Since Vologaeses IV had become king of Parthia in AD 148/149, the relations with Rome worsend and towards the end of the reign of Antoninus Pius (AD 138-161) both sides began with major troop movements.201 Detachments possibly including troops from Cappadocia were led into Syria by L. Neratius Proculus, the former commander of legio XVI Flavia at Samosata.202 Substantial reinforcements also came into Cappadocia. The province’s army must even have included a third legion at least during part of Antoninus’ reign (legio VIIII Hispana).203 Cohors Aurelia civium Romanorum may also have been set up and moved to the East at that time (or in the early years of Marus Aurelius). Only a few months after Pius’ death in March AD 161, Vologaeses opened hostilities and entirely destroyed Cappadocia’s third legion.204 That was the beginning of the great Parthian war under the emperors Marcus Aurelius (AD 161 – 180) and Lucius Verus (AD 161 – 169). Immediately detachments from the Roman armies on the

The Size of the Cappadocian Army In the light of the new diplomas and inscriptions we can now better understand the history of the Cappadocian army. Many of the units listed in the new diplomas for Galatia et Cappadocia will have belonged to the eastern Anatolian army long before the constitution of AD 101 was issued. This is certainly true for ala I Augusta colonorum, ala Augusta Germaniciana, cohors Apula, cohors I Augusta, cohors I Bosporanorum, cohors I and II Hispanorum (and possibly cohors I and II Claudia) all

196

CIL VI 1567 = XIV 4473. CIL XIII 1802. Alföldy 1969: 79ff. Birley 1997: 151ff. Gebhardt 2002: 116. 197 HA Hadr. 12,8. 21,11-13. Birley 1997: 150ff., esp. 154. 198 Birley 1997: 224ff. 199 HA Hadr. 17,11. cf. 13,9. For earlier Roman military aid to Iberia see ILS 8795 = IGRR III 133. For the excellent relations shortly after Hadrian’s death cf. HA Pius 9.6 and Fronto, ad Marc. Aurel. 3,2,3. For Pharasmanes’ visit to Rome in AD 141 see Eck / Weiss 2001: 258f. 200 HA Hadr. 17.10ff. Dio LXIX.15. Cf. Braund 1991: 208-219. 201 CIL IX 2457 = ILS 1076 proves that there was far more to the conflict than suggested by HA Pius 9.6 and Marc. 8.6. Cf. also HA Pius 12,7. For the troop movements see CIL IX 2457 = ILS 1076 with Gebhardt 2002: 120ff. and esp. Weiss 2007. 202 CIL IX 2457 = ILS 1076 with Weiss 2007. 203 Eck 1972: 459-462. 204 Dio LXXI.2.1. Birley 1988: 316-325.

205

CIL VIII 7050 = ILS 1102 = Saxer 1967: no. 63. Ritterling 1902: 372. For the development of the Roman army in Syria see now Weiss 2006. 206 207

85

THE LATE ROMAN ARMY IN THE EAST of which had already seen decades of service in the area. These and other units, the Pontica et Galatarum Cappadocumque auxilia of the late 60s of the first century AD made up the Roman forces of Cappadocia Pontusque et quidquid castrorum Armeniis.208 Clearly, not an unconsiderable force.

second century AD are missing from our records is impossible to say. However, in the light of the available evidence they must have been few, if any. One may thus assume that at the turn of the first to the second century AD the army of the province Galatia et Cappadocia included in addition to the two legions at least four alae and perhaps some fifteen or more cohorts. Of the latter at least three (?) were milliary, as is shown by the new diplomas (cohors I Italica milliaria (?), cohors I Italica milliaria voluntariorum equitata cR, cohors I Ituraeorum milliaria equitata sagittaria). Excluding the irregulars, the auxiliary army at that time thus amounted to at least 11'000 men. A fourth (?) milliary cohort was transferred to Cappadocia towards the end of Trajan’s reign (cohors III Ulpia Petraeorum equitata milliaria).

The diploma of AD 94 records 3 (?) alae and thirteen cohorts, the diploma of AD 101 lists four alae and (thirteen or) fourteen cohorts. Arrian mentions four alae and ten cohorts of the exercitus Cappadocicus in c. AD 135. None of these lists, however, is complete. In the case of military diplomas several reasons could have led to names of troops missing from the unit lists. For diplomas, as is well known, list only those units whose soldiers or veterans would receive the legal privileges granted by the imperial constitution recorded in them. Thus, some units of the Cappadocian army may not have been listed simply because they had no soldiers eligible for such legal privileges in AD 94 or 101. Furthermore, it has recently become better known that in rare cases two or even more constitutions for the auxiliary units of one provincial army could be issued on the very same day. In such cases the units were recorded in separate lists for each constitution. These lists were not necessarily of equal length. Such ‘split constitutions’ are now particularly well known for several provinces from the period of the new diplomas for Galatia et Cappadocia.209 Therefore, it is a priori not excluded that parallel constitutions which are now lost named further units of the Cappadocian army in AD 94 and in AD 101. Yet, the likelyhood for this is perhaps rather small.

Arrian’s field army consisted of soldiers from two legions, four alae and ten cohorts of the regular army of Cappadocia. Yet, Arrian’s expeditionary force did not include all units of the exercitus Cappadocicus, as some troops were left behind in order to guard the frontier. The evidence for this comes from Arrian himself as well as from inscriptions. The number of five cohorts at Apsaros alone is a strong indication that more units belonged to the exercitus Cappadocicus under his governorship than he took with him on his march against the Alani, as it seems unlikely that he would have withdrawn the entire Apsaros garrison when danger may have been threatening exactly from that direction. Much rather, Arrian and his field army were operating out of Apsaros where he would then have had one of his main supply stations which therefore could not be stripped of its garrison.210 The five cohorts at Apsaros were auxiliary troops under the command of the praepositus numerorum tendentium in Ponto, Absaro.211 Their strategic role was not confined to the Black Sea coast but clearly also aimed towards Iberia.

Whether or not we need to assume that auxiliary units are missing from the new diplomas and how many missing units there may have been must be judged from other available sources, that is mainly other inscriptions. These clearly show that there must have been more auxiliary units in Galatia et Cappadocia in AD 94 and in AD 101 than listed in the two diplomas: cohors Apula and possibly cohors I Germanorum belonged to the army of the province despite their absence from the new diplomas. Ala II Gallorum was apparently not listed in the diploma of AD 94 though it may have been recorded in the diploma of AD 101. How many more units of the army of Galatia / Cappadocia from the beginning of the

Nor did Arrian strip the entire Upper Euphrates from its garrison. Thus, a detachment from legio XII Fulminata remained at Melitene. Of the cohorts I Germanorum, III Petraeorum, I and IV Raetorum Arrian only had the cavalry with him on his march. Ala II Ulpia Auriana and ala I Augusta Gemina colonorum may also have participated only with detachments. Units which in ca. AD 135 remained at their garrison places were ala Augusta Germaniciana, cohors I and II Claudia, cohors II Hispanorum, cohors I Lepidiana, cohors (milliaria) equitata cR (?) as well as possibly cohors I Augusta cR, and cohors III Syrorum.212 That would bring the exercitus Cappadocicus in ca. AD 135 up to two legions, five alae and fifteen or seventeen cohorts at the least. This would amount to an auxiliary force of around perhaps 12'000 men. Thus, the number of units of the Cappadocian army appears to have increased only slightly since the beginning of the second century despite several transfers

208

Tac. Ann. 15,6 (cf. also 12,49). Hist. 2,6. Examples are for Syria in AD 88, Nov. 7: CIL XVI 35, Weiss 2006: nos. 1 & 2 (3 alae, 17 cohortes) with RMD I 3, AE 1997: 1761 & 1762, and Eck 2003: 229-233 (5 alae, 2 cohortes). For Syria in AD 91, 12. May: RMD I 4 (3 alae, 7 cohortes) with RMD I 5, RMD IV 241 = RGZM 6, Eck / Pangerl 2005b: 108ff. nos. 3 & 4, and Weiss 2006: no. 3 (4 alae, 6 cohortes). Yet further copies are soon to be published by W. Eck and A. Pangerl in Zeitschrift für Papyrologie und Epigraphik. For Moesia inferior in AD 97, 9. Sept.: AE 1997: 1774 (5 alae, 9 cohortes) with MacDonald / Mihaylovich 2002: 225-228 (4 alae, 10 cohortes). Cf. Eck / Pangerl 2005a: 185-192. Moesia inferior AD 99, 14. August: CIL XVI 45 and RGZM 8 (3 alae, 7 cohortes) with CIL XVI 44 (3 alae, 6 cohortes). A third parallel constitution of the same day may be missing: cf. Weiss 2005: 209 with n. 11. Moesia inferior AD 105, Mai 13: CIL XVI 50 (3 alae, 7 cohortes) with RGZM 10 (3 alae, 7 cohortes) and RGZM 11 (3 alae, 7 cohortes). 209

210

Speidel 1992: 205. CIL X 1202 = ILS 2660. 212 Cohors I Augusta Cyrenaica was not with Arrian as it belonged to the army of Galatia which by then had been separated again from Cappadocia. 211

86

MICHAEL A. SPEIDEL: THE DEVELOPMENT OF THE ROMAN FORCES IN NORTHEASTERN ANATOLIA

Map: Approximate location of Roman military sites on the northeastern frontier mentioned above. Unlocated sites are: Aeliana, Auaxa, Silvana, and Valentia. number of alae increased from three or four at the end of the first century to six in the early third century. Also, of the cohorts known to have belonged to the exercitus Cappadocicus in the later second and third centuries only three are not on record as equitatae (cohors Apula, cohors Aurelia, cohors II Claudia(na)). Furthermore, of the perhaps five alae and fifteen or seventeen cohorts of the mid-second century four alae were still in northeastern Anatolia at the end of the fourth century whereas only seven cohorts shared that same fate.213 However, only three cohorts (cohors secunda Valentiniana, cohors prima Theodosiana, and the anonymous cohors) were later added and survived to the end of the fourth century, whereas no less than six alae (ala Rizena, ala prima Iovia felix, ala Theodosiana, ala felix Theodosiana, ala prima felix Theodosiana, the anonymous ala) and two elite cavalry regiments (equites

during these years (cf. above). We still cannot be certain that we now know the names of all units of the Cappadocian army in the mid-second century. It is revealing, hoewever, that at least a third of the auxiliaryunits of Cappadocia were stationed along the eastern coast of the Black Sea. How these numbers changed after AD c. 135 is impossible to determine. Still, we can observe that at least three more cohorts were raised to milliary status during the second half of the second century (cohors I Claudia equitata, cohors I Germanorum equitata and cohors equitata cR). It is perhaps noteworthy that with the exception of cohors I Italica milliaria (the existence of which is perhaps not beyond doubt) all milliary cohorts were equitatae. This goes well with the general impression of a gradual shift towards a larger cavalry force which may also be reflected by the additional transfers to the army of Cappadocia of cohors I Lepidiana equitata (not long after AD 127) and particularly of ala I praetoria (during the earlier third century). Thus, the

213

For the survival rates of the pre-severan auxilia in the Notitia Dignitatum in general see Roxan 1976: 61.

87

THE LATE ROMAN ARMY IN THE EAST sagittarii) of the late-third and fourth centuries are listed in the Notitia Dignitatum.

Dabrowa, E. 1986 Cohortes Ituraeorum. Zeitschrift für Papyrologie und Epigraphik 63: 221-230. Dana, D. / Matei-Popescu, F. Le recrutement des Daces dans l’armée Romaine sous l’empereur Trajan: une esquisse préliminaire. Dacia NS 50, 199-210. Devijver, H. 1992 The Equestrian Officers of the Roman Imperial Army, vol. II. Eck, W. 1972 Zum Ende der legio IX Hispana. Chiron 2: 459-462. Eck, W. 1981 Miscellanea prosopographica. Zeitschrift für Papyrologie und Epigraphik 42: 227-256. Eck, W. 1992 Alam Costia constituerunt. Zum Verständnis einer Militärinschrift aus dem südlichen Negev. Klio 74: 395-400. Eck, W. 1997 Die Verwaltung des Römischen Reiches in der Hohen Kaiserzeit. Ausgewählte und erweiterte Beiträge. 2. Band. Eck, W. / Weiss, P. 2001 Tusidius Campester, cos. suff. unter Antoninus Pius, und die Fasti Ostienses der Jahre 141/142 n. Chr. Zeitschrift für Papyrologie und Epigraphik 134: 251-260. Eck, W. / MacDonald, D. / Pangerl, A. 2002 Neue Diplome für das Heer der Provinz Syrien. Chiron 32: 427-448 Eck, W. 2003 Eine weitere Kopie der domitianischen Bürgerrechtskonstitution für die Truppen Syriens vom 7. Nov. 88. Zeitschrift für Papyrologie und Epigraphik 143: 229-233. Eck, W. 2003a Suffektkonsuln der Jahre 132–134 und Hadrians Rückkehr nach Rom im Jahr 132. Zeitschrift für Papyrologie und Epigraphik 143: 234 - 242. Eck, W. / Pangerl, A. 2005 Eine Bürgerrechtskonstitution für zwei Veteranen des Kappadokischen Heeres. Zeitschrift für Papyrologie und Epigraphik 150: 233241. Eck, W. / Pangerl, A. 2005a Zwei Konstitutionen für die Truppen Niedermösiens vom 9. September 97. Zeitschrift für Papyrologie und Epigraphik 151: 185192. Eck, W. / Pangerl, A. 2005b Neue Militärdiplome für die Provinzen Syria und Iudaea / Syria Palästina. Scripta Classica Israelica 34: 108-118. Eck, W. / Pangerl, A. 2005c Traians Heer im Partherkrieg. Zu einem neuen Diplom aus dem Jahre 115. Chiron 35: 49-67. Eck, W. / Pangerl, A. 2006 Syria unter Domitian und Hadrian: Neue Diplome für die Auxiliartruppen der Provinz. Chiron 36: 205-247. Ensslin W. 1942 Zur Ostpolitik des Kaisers Diokletian. French, D. 1983 New research on the Euphrates frontier: supplementary notes 1 and 2. In: S. Mitchell (ed.), Armies and Frontiers in Roman and Byzantine Anatolia: 71-98. French, D. 1984 Classis Pontica. Epigraphica Anatolica 4: 53-59. Gebhardt, A. 2002 Imperiale Politik und provinziale Entwicklung. Gilliam, J. 1974 A Diocletianic Inscription from Ayasofya and Imperial Military Supernomina.

The result of this development can be found in the list of the Roman frontier units of the dux Armeniae. It is perhaps of some interest to see that at the end of the fourth century all these units were under the command of only one Roman general. The defense line marked by their garrison places was nearly identical with the one formerly under the command of the governors of the province of Cappadocia during most of the first three centuries. The final and permanent end of the traditional Roman control in this region only came in AD 1071. That alone is clear proof of the long-lasting success of the political, administrative and military dispositions of the Roman emperors as well as of the military success of the Roman army.214 Abbreviations and bibliography AE – L’année épigraphique. Alföldy, G. 1968 Die Hilfstruppen der römischen Provinz Germania Inferior. Alföldy, G. 1969 Fasti Hispanienses. Alston, R. 1995 Soldier and Society in Roman Egypt. Bennett, J. 2002 The Cappadocian frontier: from the Julio-Claudians to Hadrian. In Freeman, P. / Bennett, J. / Fiema, Z.T. / Hoffmann, B. (eds.), Limes XVIII: 301-312. Birley, A.R. 1997 Hadrian. The restless emperor. Birley, E. 1988 The Roman Army. Papers 1929-1986. Blockley, R.C. 1987 The division of Armenia between the Romans and the Persians at the End of the Fourth Century A.D. Historia 36: 222-234. Blockley, R.C. 1992 East Roman Foreign Policy. Bosch, E. 1967 Quellen zur Geschichte der Stadt Ankara. Braund, D. 1991 Hadrian and Pharasmanes. Klio 73: 208219. Brennan, P. 1998 The User’s Guide to the Notitia Dignitatum: the case of the Dux Armeniae (ND Or. 38). Antichthon 32 (1988) 34-49. Christol, M. / Drew-Bear, T. 1995 Inscriptions militaires d’Aulutrene et d’Apamée de Phrygie. In: Y. LeBohec (ed.), La Hiérarchie (Rangordnung) de l’armée Romaine sous le Haut-Empire: 57-87. ChLA – R. Marichal / E. Bruckner (eds.) Chartae Latinae Antiquiores. Cichorius, C. 1900 Cohors. RE IV 231 – 356. CIL – Corpus Inscriptionum Latinarum. Cotton, H.M. / Eck, W. / Isaac, B. 2003 A Newly Discovered Governor of Judaea in a Military Diploma from 90 CE. Israel Museum Studies in Archaeology 2: 17-31. Crow, J.G. 1986 A review of the physical remains of the frontiers of Cappadocia. In: P. Freeman / D. Kennedy (eds.) The defence of the Roman and Byzantine east: 76 – 91.

214

For the developments in Roman Mesopotamia cf. Speidel 2007.

88

MICHAEL A. SPEIDEL: THE DEVELOPMENT OF THE ROMAN FORCES IN NORTHEASTERN ANATOLIA R. (eds.), Prinzipat und Kultur im 1. Jahrhundert: 306-310. Lörincz, B. 2001 Die römischen Hilfstruppen in Pannonien während der Prinzipatszeit. Teil I: Die Inschriften. MacDonald, D. / Mihaylovich A., 2002 A New Moesia inferior Diploma of 97. Zeitschrift für Papyrologie und Epigraphik 138: 225-228. Mamuladze, Sh.Kh. / Khalvashi, M.S. / Aslanishvili, L.G., 2002 Rimskie garnisonu Apsara, Vestnik Drevneij Istorii 240.1: 33-39. Matei-Popescu, F. 2004 Trupele auxiliare din Moesia Inferior, Studii si Certari de Istorie veche si Archeologie 52-53, 2001-2002: 173-242. Matei-Popescu, F. (forthcoming) Two Fragments of Roman Military Diplomas discovered on the Territory of the Republic of Moldova. Dacia, N.S., Supplementum I, Ecrits de philologie, d’épigraphie et d’histoire ancienne à la mémoire de D.M. Pippidi. Marek, Chr. 2003 Pontus et Bithynia. Dir römischen Provinzen im Norden Kleinasiens. Mitchell, S. 1993 Anatolia. Land, Men, and Gods in Asia Minor. 2 vols. Mitford T.B. 1974 Some Inscriptions from the Cappadocian Limes. Journal of Roman Studies 64: 160-175. Mitford T.B. 1977 The Euphrates frontier in Cappadocia. In: Studien zu den Militärgrenzen Roms 2: Vorträge des 10. Internationalen Limeskongresses in der Germania Inferior: 501 – 510. Mitford, T.B. 1980 Cappadocia and Armenia Minor: historical setting of the limes. In: H. Temporini (ed.), Aufstieg und Niedergang der Römischen Welt II.7.2: 1169 – 1228. Mitford T.B. 1997 The inscriptions of Satala (Armenia Minor). Zeitschrift für Papyrologie und Epigraphik 115: 137-167. Mitford T.B. 1998 The Roman Frontier on the Upper Euphrates. In R. Matthews (ed.), Ancient Anatolia: 254 – 272. Olympia – W. Dittenberger / K. Purgold, Die Inschriften von Olympia (1896). Overbeck, B. 1981 Das ertse Militärdiplom aus der Provinz Asia. Chiron 11: 265-276. Petolescu, C.C. / Popescu, A.-T. 2004 Ein neues Militärdiplom für die Provinz Moesia inferior, Zeitschrift für Papyrologie und Epigraphik 148: 269273. Pflaum, H.-G. 1978 Les fastes de la province de Narbonnaise. Piso, I. 1993 Fasti Provinciae Daciae I. Die senatorischen Amtsträger. PME – H. Devijver, Prosopographia militiarum equestrium quae fuerunt ab Augusto ad Gallienum (1987-2001). Ramsay, W.M. / Ramsay, A.M. 1928 Roman Garrisons and Soldiers in Asia Minor Journal of Roman Studies 18: 181-190. RGZM – Pferdehirt, B. 2004 Römische Militärdiplome und Entlassungsurkunden in der Sammlung des

Zeitschrift für Papyrologie und Epigraphik 15: 183191. Greatrex, G. / Lieu, S.N.C. 2002 The Roman Eastern Frontier and the Persian Wars. Part II. AD 363-630. Habicht, Ch. 1969 Die Inschriften des Asklepieions. In: Die Altertümer von Pergamon VIII, 3. Haensch, R. 1997 Capita provinciarum. Statthaltersitze und Provinzialverwaltung in der römischen Kaiserzeit. Halfmann, H. 1979 Die Senatoren aus dem östlichen Teil des Imperium Romanum bis zum Ende des 2. Jh. n.Chr. Harper, R.P 1974 Pagnik Ögreni excavations 1971. In: S. Pekman (ed.) Keban Project Publications 1:4. 1971 activities: 107-108. Hartmann, M. / Speidel, M.A. 2003 The Roman Army at Zeugma: results of new research. In Early, R. / Crowther, C. / Nardi, R. / Önal, M. / Abadie, C. / Darmon, J.-P. / Hartmann, M / Speidel, M.A. Zeugma: Interim Reports. Hewsen, R.H. 1978-1979 The Successors of Tiridates the Great: A Contribution to the History of Armenia in the Fourth Century. Revue des études arméniennes 13: 99-126. Holder, P. 1980 Studies in the Auxilia of the Roman Army from Augustus to Trajan. Holder, P. 2002 Two commanders of ala Phrygum. Zeitschrift für Papyrologie und Epigraphik 140: 287295. Holder, P. 2003 Auxiliary deployment in the reign of Hadrian. In J.J. Wilkes (ed), Documenting the Roman Army: 101-145. IGRR – R. Cagnat, Inscriptiones Graecae ad res Romanas pertinentes (1911 – 1927). IK – Inschriften griechischer Städte aus Kleinasien. ILS – H. Dessau, Inscriptiones Latinae Selectae (18921916). InscSide – G.E. Bean, Side Kitabeleri. The Inscriptions of Side (1965). IScM – Inscriptiones Scythiae Minoris Graecae et Latinae Kennedy, D. 1997 The special command of M. Valerius Lollianus. Elektrum 1: 69-81. Kiguradze, N.Sh. / Lordkipanidze, G.A. / Todua, T.T. 1987 Kleijma XV légiona iz Pitsundskogo gorodishcha. Vestnik Drevneij Istorii 2: 88-92. Kuhoff, W. 2001 Diokletian und die Epoche der Tetrarchie. Lekvinadze, V.A. 1962 Pontiyskiy Limes. Vestnik Drevneij Istorii 108.2, 85 Lewin, A. 2004 Limitanei and comitatenses in the Near East from Diocletian to Valens. In Y. Le Bohec / C. Wolff (eds.), L’armée romaine de Dioclétien à Valentinien Ier: 227-236. Lordkipanidze, G. / Braund, D. 1991 Recent Work at Pityus (Pitsunda/Bischvinta, USSR), In: V.A. Maxfield / M.J. Dobson (eds.), Roman Frontier Studies: 335-336. Lordkipandize, G. 1995 Forschungen in Pityus (PizundaBitschwinta), in Kühnert, B. / Riedel, V. Gordesiani, 89

THE LATE ROMAN ARMY IN THE EAST Römisch-Germanischen Zentralmuseums. Ritterling, E. 1902 Zur Erklärung von Arrians Ektaxis. Wiener Studien 24: 359-372. Ritterling, E. 1925 Legio, RE XII 1186 – 1829. Ritterling, E. 1927 Military Forces in the Senatorial Provinces, Journal of Roman Studies 17: 28-32. RMD I – Roxan, M.M. Roman Military Diplomas 19541977 (1978). RMD II – Roxan, M.M. Roman Military Diplomas 19781984 (1985). RMD III – Roxan, M.M. Roman Military Diplomas 1985-1993 (1994). RMD IV - Roxan, M.M. / Holder, P. Roman Military Diplomas IV (2003). RMR – R. Fink, Roman Military Records on Papyrus (1971). Roxan, M.M. 1976: Pre-Severan Auxilia Named in the Notitia Dignitatum. In: R. Goodburn / P. Bartholomew (eds.), Aspects of the Notitia Dignitatum: 59-79. Roxan, M.M. / Eck, W. 1997 A Diploma of Moesia Inferior: 125 Iun. 1. Zeitschrift für Papyrologie und Epigraphik 116: 193–203. Roxan, M.M. 1997a An Auxiliary/Fleetdiploma of Moesia Inferior: 127 August 20. Zeitschrift für Papyrologie und Epigraphik 118: 287-299. Roxan, M.M. 1999 Two Complete Diplomas of Pannonia Inferior: 19 May 135 and 7 Aug. 143, Zeitschrift für Papyrologie und Epigraphik 127: 249-273. Saxer, R. 1967 Untersuchungen zu den Vexillationen des römischen Kaiserheeres von Augustus bis Diokletian. Speidel, M.A. 1995 Ferox: Legionary Commander or Governor? A note on Tab.Vindol. II 154. In: R. FreiStolba / M.A. Speidel (eds.), Römische Inschriften – Neufunde, Neulesungen und Neuinterpretationen. Festschrift für Hans Lieb: 43-54 Speidel, M.A. 2002 Bellicosissimus princeps. In Nünnerich-Asmus, A. (ed.), Traian. Ein Kaiser der Superlative am Beginn einer Umbruchszeit?: 23-40. Speidel, M.A. 2005 Early Roman Rule in Commagene. Scripta Classica Israelica 24: 85-100. Speidel, M.A. 2007 Ein Bollwerk für Syrien. Septimius Severus und die Provinzordnung Nordmesopotamiens im dritten Jahrhundert. Chiron 37 (forthcoming). Speidel, M.A. (forthcoming) Cappadocia. Vom Königreich zur Provinz. In I. Piso (ed.), Die römischen Provinzen. Begriff und Gründung. Speidel, M.P. 1978 Guards of the Roman Army. Speidel, M.P. 1984 Roman Army Studies I. Speidel, M.P. 1992 Roman Army Studies II. Strobel, K. 1984 Untersuchungen zu den DakerkriegenTrajans. Weiss, P. 1997 Neue Militärdiplome. Zeitschrift für Papyrologie und Epigraphik 117: 227-268. Weiss, P. / Speidel, M.P. 2004 Das erste Militärdiplom für Arabia. Zeitschrift für Papyrologie und Epigraphik 150: 253-264. Weiss, P. 2005 Neue Militärdiplome Domitians und Nervas für Soldaten Niedermösiens, Zeitschrift für Papyrologie und Epigraphik 153: 207-217.

Weiss, P. 2006 Die Auxilien des syrischen Heeres von Domitian bis Antoninus Pius. Eine Zwischenbilanz nach den neuen Diplomen. Chiron 36: 249-298. Weiss, P. 2007 Militärdiplome und Reichsgeschichte: Der Konsulat des L. Neratius Proculus und die Vorgeschichte des Partherkriegs und Marc Aurel und Lucius Verus. In: R. Haensch / J. Heinrichs (eds.): Der Alltag der römischen Administration in der Hohen Kaiserzeit. Akten des Kolloquiums zu Ehren von Prof. Dr. Werner Eck (forthcoming). Wheeler, E. 1991 Rethinking the Upper Euphrates Frontier: Where was the western border of Armenia. In: V.A. Maxfield / M.J. Dobson (eds.), Roman Frontier Studies 1989: 505-511. Wheeler, E. 1999 From Pityus to Zeugma: The Northern Sector of the Eastern Frontier 1983-1996. In: N. Gudea (ed.), Roman Frontier Studies XVII/1997: 215 – 229. Wheeler, E. 2000 Legio XV Apollinaris: from Carnuntum to Satala – and beyond. In: Y. LeBohec (ed.), Les légions de Rome sous le Haut-Empire: 259 – 308. Zuckermann, C. 1991 The early Byzantine Strongholds in Eastern Pontus. Travaux et Mémoires 11: 527-553. Zuckermann, C. 1998 Sur le dispositif frontalier en Arménie, le limes et son évolution, sous le BasEmpire. Historia 47: 108-128.

90

The Nabataeaan – Roman Negev in the Third Century CE Tali Erickson-Gini Israel Antiquities Authority

Over the past few decades, archaeological research carried out by a number of researchers in the central and southern Negev and the Arava Valley in southern Israel is steadily revealing evidence that the 3rd c. CE was a definitive turning point in the fortunes of the inhabitants in that period. It is well known that the Roman Empire underwent a series of political, military and economic convulsions in the 3rd c. CE. However, from an archaeological perspective, specific evidence dating to this period is often sparse and the effects of these crises are generally unclear on a regional level.

transitory and the first permanent settlements in the Negev Highlands were not established until the later 1st c. BCE at sites formerly used by them such as Oboda, Elusa and Nessana. The establishment of the first settlements in the Negev Highlands coincides with rapid and rather dramatic developments in Nabataean culture and their economy. These developments include the beginning of the production of unguents and their ceramic containers at Petra, the monumental development of their capital at Petra and the establishment of a new, direct route between Moyat Awad and Oboda by way of the Ramon Crater northwest of the Arava Valley (Johnson 1987: 53; Schmid 2001: 374; Erickson-Gini and Israel 2003: 10*). By the middle of the 1st c. CE at least three new settlements were established in the Negev Highlands along secondary routes at Sobota (Shivta), Rehovot in the Negev and Mampsis (Mamshit).

In the Negev, architectural and other material remains have been uncovered that shed light on developments in the 3rd c. at a local and a regional level. On a broader historical scale, recent research has brought developments in the region in the first half of the first millennium CE into higher resolution, making it possible to either confirm or adjust a number of conceptions concerning the cultural and economic continuity of the Nabataeans communities in the Negev.

The region was annexed by Rome in 106 CE and became part of the Provincia Arabia along with southern Transjordan and the Sinai. Inscriptions found at Oboda are dated using the date “of the eparchy”, i.e., the establishment of the Provincia Arabia, soon after the annexation (Negev 1977: 643). Bostra was designated the administrative center of the new province and the headquarters of the legio III Cyrenaica. Inscriptions found at Mampsis indicate the presence of Roman military personnel there in the 2nd c. century following the annexation. These include a centurion of the legio III Cyrenaica and an official of the Cohors I Augusta Thracum (Negev 1969: 9). At Shar Ramon, an important station on the Petra – Gaza road, an inscription referring to the Cohors VI Hispanorum was found that may also date to the 2nd – early 3rd c. (Figueras 1992: 178).2 Petra remained an important center in this period and received a number of honorific titles such as the ‘Metropolis of Arabia’ in 114 CE and ‘Hadriana Petra Metropolis’ in 131 during a visit by the Emperor Hadrian (Fiema 2002: 61). One of the most significant acts of the Roman administration following the annexation was the construction of the Via Nova Traiana between the Red Sea port of Aila (Aqaba) and Syria through Transjordan. In the Negev, the Petra – Gaza road continued in heavy use throughout the 2nd century (Cohen 1982: 246).

Nabataean Presence in the Negev Highlands The Nabataeans began to cross the Negev in order to market incense and other exotic goods in the second half of the 1st millennium BCE. In the period following the establishment of the Ptolemaic control in Egypt and Palestine in the late 4th c. BCE, the Nabataeans were already based in Petra and the surrounding region in what is now southern Jordan (Diod. II.48.4-5). Their activity as traders in this period in the Negev was probably seasonal: the Nabataeans obtained frankincense and myrrh from traders from South Arabia and transported these goods across the central Arava and Negev Highlands down to the coastal plain of southern Palestine to the emporium of Gaza. In the 3rd c. BCE the Nabataeans established a chain of small forts and caravan stations in the Negev along the original Petra – Gaza road, the Darb es-Sultan at Moyat Awad and ‘En Rahel in the central Arava and at ‘En Ziq.1 Stations were also established at Oboda (Avdat), Horvat Ma’agora and Elusa (Halutza) and along secondary routes leading to the Dead Sea and across the Negev and northern Sinai. The nature of the Nabataean presence in this period in the Negev was apparently 1 The site of Moyat Awad, which is the first Nabataean fort on the western side of the Arava Valley northwest of Wadi Musa, has been identified as the site of Moa on the Madeba mosaic map from the 6th c. CE. However, recent studies on the finds from Moyat Awad have revealed that the site was not occupied anytime in the Byzantine period, precluding the possibility of its identification as the Moa of the Madeba Map. The actual site of Mo'a is probably that of Bir Madhkur on the eastern side of the Arava Valley as proposed by Alt (1935). A recent survey of the castellum of Bir Madhkur revealed artifacts dated to the Nabataean through Byzantine periods (Smith 2005: 58).

2 Inscriptions referring to this unit were found in 1st c. CE contexts in Syria at el-Asiri and from 212 CE at Qasr el-Hallabat in Jordan (Kennedy 2000: 45).

91

THE LATE ROMAN ARMY IN THE EAST excavations carried out by Negev and Cohen in a 2nd 3rd c. caravansary, located next to the pottery workshop on the eastern edge of the site, revealed some late, debased Nabataean painted ware dated to that period, together with coins of Philip (244-249) and Claudius Gothicus (269-270) underlying a destruction layer with 4th c. coins, (Cohen 1980: 46).6

Oboda The site of Oboda was among the earliest Nabataean stations in the Negev and it appears to have been permanently settled in the second half of the 1st c. BCE. According to historical sources it was named after the deified Nabataean king Obodas, probably Obodas II who apparently ruled between 62 and 58 BCE (Fiema and Jones 1990: 245; Negev 1997: 3). Oboda is the only site in the Negev with a well defined acropolis and architectural remains from temples in its precinct, including the temple of Zeus Obodas. The discovery of an inscription in Nabataean Aramaic script dated to 8/7 BCE indicates that the acropolis was constructed before the end of the 1st millennium BCE.3 The excavator of the acropolis, Avraham Negev, reported that it was constructed in two phases: the late 1st c. BCE - 1st c. CE and in the second half of the 3rd c. CE (Negev 1997: 53). Oboda was occupied continuously from the late 1st c. BCE until its destruction in a local earthquake in the early 7th c. CE (Fabian 1996; Korjenkov and Mazor 1999).4 Inscriptions dating to the 3rd c. CE were found by Negev in earlier excavations and important pottery assemblages from the 3rd c. have been discovered by the writer in recent excavations in 1999-2000. The inscriptions found and published by Negev include a number of dedicatory inscriptions, found on the acropolis in the area where a temple of Zeus Obodas was apparently rebuilt or rededicated in the mid-3rd c. CE, epitaphs discovered in the en-Nusra burial cave and a dated inscription found on the tower on the southern side of the town (Negev 1981: 11-27). One dedicatory inscription from the acropolis dates to 267/8 CE (Negev 1981: 12).5 Epitaphs of women found in the en-Nusra burial cave at the site include one dated to the year 136 (=241 CE) and a second to the year 160 (=265 CE). A lintel stone found in the ruins of the tower excavated by Negev contains the following lines inscribed in the tabula ansata of the stone: “In good luck! Zeus Oboda, give help to Eirenaios who built this tower in good omen, in the year 188, with (the assistance of) the builder Wailos from Petra and Eutiches”, (Negev 1981: 27). The date in the inscription is the year 188 (=293/294 CE). Although the corpus of inscriptions dated to the 3rd c. CE found and published by Negev from Oboda is impressive, Negev never published any pottery assemblages dated to the 2nd or 3rd c. that may have been found together with the inscriptions. Joint

Excavations carried out in the area just outside the Byzantine period town wall of Oboda in 1999-2000 by the writer uncovered a structure occupied in the 2nd to early 3rd c. CE. One small room in this structure identified as a kitchen pantry under a stairwell, contained dozens of in situ ceramic vessels that were apparently abandoned together with cuts of meat hanging from above the shelves and at least two imported glass vessels. A jar that originally contained a liquid substance, probably wine, was found on the floor of the pantry with the plaster stopper still intact in the neck of the vessel. This assemblage, together with other vessels found in the structure, form one of the most complete assemblages found in southern Israel from the early 3rd c. CE, the other being that found in the fort at Moyat Awad from the same period. Complete painted Nabataean fine ware bowls found in the pantry are examples of the latest phase in the production of painted fine wares from Petra (Erickson-Gini 2004: Fig. 2.1-3). Similarly, a complete example of the latest type of ceramic unguentaria was found in the pantry (Erickson-Gini 2004: Fig. 2.38). Unguentaria produced at Petra went out of production sometime between 225-250 CE with the collapse of the international trade and the cessation in the production of perfumed oils at Petra (Johnson 1987: 64-66). In other parts of the building pottery from the latest phase its occupation, and identical to the types found in the pantry, were uncovered. Coins found under the floors of the structure date to the late 2nd - early 3rd c. CE (EricksonGini 2004: 511-515). The excavation of this building revealed unusual details: the abandonment of the pantry complete with foodstuffs and an intentional fill over most of the rooms before the area was reoccupied in the later part of the 3rd c. or early 4th c. Since structural damage to the building can be ruled out as a cause of the abandonment of the building, the possibility remains that it was abandoned for several decades due to the ravages of epidemic.7 6 No final publication has ever been completed from these excavations. The results referred to here belong to preliminary reports. 7 Ancient sources refer to the spread of major plagues in the second and third centuries CE, particularly in the time of Marcus Aurelius (the Antonine plague, 165-180 CE) which apparently had wide reaching effects on regional economies (Duncan-Jones 1996) and the ‘plague of Cyprian’ in the mid-third century which may have been responsible for the death of Claudius Gothicus who died in 270 CE. In regards to the third century, this may have actually been an epidemic of measles or smallpox, since the symptoms described by Cyprian, the bishop of Carthage, in De Mortalitate resemble these diseases (Kohn 1998: 251). A later source, Zosimus, recorded the outbreak of severe pestilence during the reigns of Gallus and Gallienus (Zos. I.26, I.37). In the Roman world pernicious diseases were believed to have their source in the South or the East and historical research suggests the strongest vectors of epidemic disease may be attributed to the movement of army units

3 The dedicatory inscription dated to the second year of the Nabataean king, Aretas IV (Negev 1961: 129-130). 4 Negev contended that the site was destroyed in the middle of the 1st c. CE and was quickly reoccupied by the Nabataeans (Negev 1977: 659). However, subsequent excavations at Oboda and other Nabataean sites in the Negev and Arava have not indicated a break in occupation. It is possible that Negev discovered, and misdated, evidence of destruction of 1st c. CE buildings damaged in a earthquake that appears to have affected a wide area in southern Israel and southern Jordan sometime in the early 2nd c. CE (see Korjenkov and Erickson-Gini 2003). 5 “Year 162. Let be remembered Soaidos”. According to Negev, assuming that this dates from the first year of the Provincia Arabia, the beginning of Roman rule in the area, it translates to 267/268 CE (Negev 1981: 12).

92

TALI ERICKSON-GINI: THE NABATAEAAN – ROMAN NEGEV IN THE THIRD CENTURY CE At the end of the 3rd c. the area was once again reoccupied and several private dwellings were constructed around and over the remains of the building described above. These structures are of a lower quality of construction and have very irregular plans. Each has a small courtyard that was used for cooking and other domestic purposes. In most cases one unusual feature of construction was evident: several of the courtyards were built at a higher level than the surrounding rooms and many of the floors of some of these rooms were quarried into the surrounding bedrock. This type of construction has not been found in the region in buildings dated to the Nabataean / Roman or Late Byzantine periods. The residential quarter appears to have been constructed next to a tower located east of the acropolis. Although it has yet to be excavated this tower appears to be identical to the tower constructed on the southern perimeter of the town dated to 293/294 CE (see above). A narrow alley way links the dwellings and leads towards the tower. The earliest coin found in this complex was that of Diocletian dated to 297 CE (Erickson-Gini 2004: 520). The residential quarter was occupied for approximately one hundred years.8

three sides of the camp that may have been used as stables. New excavations carried out in the camp by Fabian and Erickson-Gini in 1999 on behalf of the Israel Antiquities Authority uncovered evidence that the camp was constructed in the late 3rd c. CE (Erickson-Gini 2002).10 The Petra – Gaza Road The route between the site of Moyat Awad and Oboda by way of the Ramon Crater was established by the Nabataeans in the late 1st c. BCE and continued to be a main artery of international trade through the Negev until the first half of the 3rd c. CE. A series of sites are located along the road and are dated to that period at (from southeast to northwest): Har Masa, Qasra, Mezad Neqarot, Shar Ramon, Mezad Ma’ale Mahmal and Mezad Grafon.11 With the exception of Mezad Grafon, these sites were excavated by Cohen in the early 1980s (Cohen 2000: 80-89). The excavations are currently being prepared for final publication and edited by the writer following the demise of the excavator, R. Cohen, and the previous editor, Y. Hirschfeld in 2006.12 The site of Moyat Awad produced an impressive amount of pottery. Close examination of this assemblage shows that the greatest amount was found in the small Hellenistic Nabataean fort, measuring 17 x 17 m. and dates from the latest phase of its occupation that ended in the first half of the 3rd c. CE. In this phase the fort appears to have been used for storage and mercantile purposes: an olive press was found in one of the rooms and a copious amount of whole pottery vessels were found throughout the other rooms (Cohen 2000: 78). A recent examination of the finds by the writer suggests that perfume production took place in the fort in its latest phase. Probes carried out in the terraced fields located in the immediate vicinity of the site by Bruins uncovered pottery sherds from the 2nd CE (Bruins, p.c.). The site was supplied with spring water and aqueducts channeled water to the fields, an open pool and a bath house (with frescos) located inside the Nabataean caravansary built below the fort (Cohen 2000: 76-79). The latest coins found at the site together with other material evidence were those of Septimus Severus

The construction of the southern tower, the residential quarter and the bath house9 located on the plain below and west of the town, appears to have been connected to an increase in the presence of Roman forces at the site and in the general area in the late 3rd c. CE. In this period the Tenth Legion was transferred from Jerusalem to Aila and the Provincia Arabia was divided and its southern territory, including the Negev, Transjordan and Sinai, became part of the province of Palestine. It was in the context of these changes that a large military camp was constructed at Oboda in the late 3rd c. CE. This camp was built on the plateau northeast of the acropolis. It measures approximately 100 x 100 m. and contained eight barrack blocks with sixty-eight rooms. Towers were constructed at each corner as well as four interval towers and two guard towers flanking the main gate located on the eastern side of the camp. A principia was built along the western wall facing the entrance to the camp. A series of casemate rooms were constructed along the inner wall of and trade contacts (Duncan-Jones 1996: 114, 135). In this respect, the vulnerability of communities such as Oboda located along major trade routes such as Petra – Gaza Road, through which both military units and merchants passed on a regular basis, cannot be overemphasized. 8 Towards the end of the 4th c. or early 5th c. CE the complex was destroyed in what was probably a local earthquake and this part of the site, along with all the area in the eastern part of the site, was abandoned. In the later Byzantine period the town was reconstructed between the acropolis and the southern tower and a wall was constructed around the upper part of the town and the cave complex on the western slope of the site. 9 The bath house was excavated by Negev who dated it generally to the Byzantine period (Negev 1997: 176). Subsequent excavations carried out in 1993 by the writer in the area of the well and the pools that supplied water to the bath house uncovered coins from the 1st half of the 4th c. CE (the earliest coins found in the area) and continued to be used after renovations, possibly in the 5th c. (Erickson-Gini, forthcoming). It is notable that the original arches in the bath house are identical in construction and style to those found in the southern tower of the site, dated to 293/294 CE (see Negev 1997: Photo 87).

10

The date of the camp has long been a point of controversy among researchers. In 1975-1976 a few rooms were excavated by Negev and Cohen. Negev immediately proclaimed that camp to be Nabataean and dated it to the 1st c. CE, a date disputed by Cohen (Negev 1977: 622624; Cohen 1982: 245). In the recent excavations in 1999 Fabian proposed that the camp was constructed in wake of the annexation of Nabataea in 106 CE and abandoned during the Bar Kokhba Revolt (132-135 CE), (Fabian 2001). However, only one coin from that period was found in mixed contexts in the camp while several coins of Diocletian were found in several parts of the camp under floors and at floor level. A detailed discussion of the problems in dating the camp may be found in my article (Erickson-Gini 2002). 11 The term ‘mezad’ is Hebrew for ‘fort’. The term ma’ale is Hebrew for ‘ascent’. 12 The writer is currently preparing the report of the ceramic vessels found at the sites of Moyat Awad (Moa), Qasra, Har Masa, Mezad Neqarot, Shar Ramon, Mezad Ma’ale Mahmal and Horvat Ma’agora for that publication.

93

THE LATE ROMAN ARMY IN THE EAST (192-211) and Elagabalus (218-222).13 The large assemblage of pottery found in the fort appears to have been abandoned and left untouched, similar to the assemblage found in the pantry at Oboda from the same period. The site was never reoccupied although two or three scattered examples of Early Islamic vessels and coins were found at the site that were deposited when a small settlement was established a few kilometers away in the 8th c. CE.

(Cohen 2000: 86). Unlike Moyat Awad, the structure was partially reoccupied in the late 3rd through 4th c. CE. In this latest phase, a coin of Gallienus (253-268 CE) was discovered over bedrock below a plaster floor in one of the rooms and variety of African red slipped wares, lamps and other vessels dated to 4th c. were found in several other rooms. The next stop along the Petra – Gaza road in the direction of Oboda, Mezad Ma’ale Mahmal, appears to have a similar history. A small fort, measuring 6.5 x 7 m, was built at the top of the pass above the Ramon Crater sometime in the later half of the 2nd c. CE. The earlier Nabataean structure was excavated by the writer in 2004. The Nabataean structure appears to have been constructed in the middle of the 1st c. CE and was occupied until its destruction by earthquake in the early 2nd c. CE (Erickson-Gini, f.c.). The fort was constructed next to the ruins of the earlier building. It appears to have been occupied until the first half of the 3rd c. CE. Two coins of Gallienus16 (253-268 CE) were found in the fort as well as some examples of African red slipped bowls and other vessels dated to the 4th c. CE. This evidence provides a strong indication that the fort, like Shar Ramon, was reoccupied in the late 3rd through 4th c. CE.

The smaller site of Mezad Neqarot is located in Nahal Neqarot leading between Har Masa and Shar Ramon. At least four structures have been found at the site: the first is an irregularly shaped building (measuring 12 x 17 m.) with rooms built around a central courtyard situated on the north bank of the wadi (Cohen 2000: 83). Material evidence found in this structure indicates that it was occupied from the late 1st c. BCE until the early 3rd c. CE.14 A tower and adjoining courtyard (measuring 7.5 x 8 m) was constructed 20 meters northwest of the earlier building, probably in the second half of the 2nd c. CE (similar to the fort at Qasra and Mezad Ma’ale Mahmal).15 The latest material found in the tower fort includes whole pottery vessels dated to the first half of the 3rd c. CE and coins of Elagabalus (218-222 CE). The site appears to have been abandoned from that date although the water cistern continued to be frequented by travelers in later periods.

The discovery of milestones with painted inscriptions dated to the late 3rd - early 4th c. in the southern Arava (see below) has prompted the suggestion by Ben-David that the series of milestones located between Shar Ramon and Ma’ale Mahmal and on the Nafha Plain on the road to Oboda may also date to the same period. A survey carried out in the 1960s by Meshel and Tsafrir focused attention on these milestones, which at the time were considered to be Nabataean (Meshel and Tsafrir 1974: 5). Meshel and Tsafrir reported that at least 2 milestones were located north of Oboda on the road to Horvat Ma’agora (Meshel and Tsafrir 1975: n. 1). These milestones have since been collected and can be viewed in the visitors’ center at Oboda.17

The other large site along this leg of the Petra – Gaza road is that of Shar Ramon, located next to the spring of Saharonim on the southern edge of the Ramon Crater. This was the site of a Nabataean caravansary, probably constructed in the middle of the 1st c. CE, that remained in use until the first half of the 3rd c. CE. The structure, similar to that of Moyat Awad, measures 42 x 42 m. (Cohen 2000: 85-87). In the second phase of occupation, dating from 106 CE until the 1st half of the 3rd c. CE, a large oven was constructed in one of the rooms and a number of plastered tubs provided washing facilities for travelers. Like Moyat Awad, the largest assemblage of pottery found at the site dates to its abandonment in the first half of the 3rd c. CE and the latest coins were those of Commodus (177-192) and Caracalla (198-217),

At Horvat Ma’agora a square fort with projecting corner towers measuring 21 x 21 m. was discovered overlying earlier Hellenistic and Early Roman Nabataean structures. The fort was dated by the excavator to the Early Byzantine period (Cohen 1985: XIV, n. 46). The dating of the fort is substantiated by the discovery of late 3rd and 4th c. coins found in the middens outside the fort by the excavation team (Y. Israel, p.c.). These include coins of Probus (276-282 CE), Numerian (283-283) and Constantius II (337-361). In view of the substantial evidence of Roman army activity at Oboda and the surrounding area, it is probable that the same army units that constructed the camp at Oboda in the Diocletianic

13

Several other coins dated generally to the early 3rd c. CE were found at the site. Only one Byzantine period coin (of Theodosius I, 379-395) was found. However, a thorough examination of the ceramic assemblage by the writer while preparing the final pottery report for the site revealed a total absence of Byzantine period pottery from any of the excavation areas. 14 According to Cohen, the early building was abandoned in order supply building stone for the tower (Cohen 2000: 82). However, a recent examination of the pottery found in this building by the writer has revealed pottery from the latest phase of occupation of the site. In addition, the building stone found in the early building is of a lower quality than that found in the fort, a further indication that the early building remained occupied beside the fort. 15 An intact water built water cistern with three arches is located approximately half a kilometer away in one of the smaller channels leading into Nahal Neqarot. The latest material evidence found in the fort date to the first half of the 3rd c. CE. The fourth structure is a small watch tower (3 x 4 m.).

16

These coins were reported by Cohen (Cohen 2000: 89). Notably, all the milestones discovered in the area of the Ramon Crater and Oboda appeared to have inscriptions painted in red like the Diocletianic milestones discovered in the Arava. Due to their faded condition, none have yet been deciphered.

17

94

TALI ERICKSON-GINI: THE NABATAEAAN – ROMAN NEGEV IN THE THIRD CENTURY CE period were also responsible for the construction of the fort at Ma’agora north of Oboda and the reoccupation of sites south of Oboda at Mezad Ma’ale Mahmal and Shar Ramon as well as erection of milestones between Oboda and those sites as recently suggested by Ben David (Ben David 2005: 47).

Horvat Hazaza This is the site of an unusual building, probably a Nabataean temple, located on the road between Oboda and Mampsis near modern Kibbutz Sde Boqer. Excavations were carried out by Cohen in 1971 revealing a building with two wings (22 x 51 m. in size), (Cohen 1981: 42-45). New excavations were carried out at the site in 2001 by the writer. The building was constructed in the mid-1st c. CE and is of a very high quality of construction. The south wing of the structure appears to have been abandoned in wake of structural damage, probably by earthquake, sometime in the early 2nd c. CE. The north wing was renovated and the structure remained in use until sometime in the 3rd c. CE. In the latest phase of occupation the building appears to have been used by pastoralists and some of the rooms were turned into sheepfolds. The paucity of coins and rather limited use of the structure (as revealed in recent excavations inside the building and in the middens) contradicts Cohen’s claim that the site served as an important station (Cohen 1992: 1053).20 In the Byzantine period the building was unoccupied and it was used as a source of building stone for terraces built in the wadi below the site.

Mampsis Mampsis belongs to the later group of Nabataean towns that were constructed in a second wave of settlement activity in the Negev in the 1st c. CE.18 The early settlement at Mampsis was made up of rather rude domestic dwellings, some of which were uncovered in excavations carried out by Negev and later by EricksonGini in 1993-1994 (Negev 1988a: 26; Erickson-Gini 1999: 11-13). Public structures such as the caravanserai and Building XIV (a structure located on the hilltop and assumed to be a fortress) were of a considerably higher quality construction. In this period, the town, like other settlements in the Negev, lacked a defensive walled (Negev 1988b: 4, 27). Following the Roman annexation of Nabataea in 106 CE much of the early settlement was obliterated in order to make way for the construction of large, private villas. The largest of these, Building XII, boasted large open courtyards, a covered stable, a lavatory, a second story and rooms with frescos dated to the 2nd-3rd c. CE (Negev 1988a: 162). During the excavation of the building a hoard of coins was found in a large, bronze jar buried in the floor of the staircase tower. The hoard consisted of 10,800 coins, primarily tetradrachms minted during the reigns of Septimius Severus (193-211), Geta (211-212), Caracalla (211-217) and Elagabalus (218-222); (Negev 1988a: 145). The villas constructed in the 2nd c. CE continued to be used throughout the 3rd c. and well into the Late Byzantine when the site appears to have been deserted.

Horvat Bor, Migdal Tsafit and Mezad Tamar Military sites on the roads leading to Mampsis appear to have been constructed in the late 3rd - early 4th c. CE. These include Horvat Bor where a small fort with corner towers was built along the road leading towards Mezad Yeruham and Oboda. This site, which was partially destroyed when the British constructed the modern Dimona – Yeruham highway, was excavated by Israel and Erickson-Gini in 2001 (Israel and Erickson-Gini 2001-2002). A line of small towers and a large fort were constructed or reoccupied along the road between Mampsis and the Dead Sea along Ma’ale Peres. One of these towers, Migdal Tsafit, was excavated by Gichon in 1971. This tower measures 6.15 x 6.15 m. and is located between Mampsis and Mezad Tamar. According to Gichon, the tower was constructed in the Trajanic period and reoccupied in the late 3rd c. CE (Gichon 1971: 28).21 Mezad Tamar, located above the Peres Ascent, was a substantial fort, measuring 38 x 38 m., with projecting

Additional excavations at the site revealed that the town wall was built during the reign of Diocletian in the late 3rd or very early 4th c. CE (Negev 1988b: 27). Negev’s dating of the fortifications was confirmed in excavations carried out by the writer in the eastern corner of the site in 1993. In this area a previously unknown building, Building XXV, was discovered east of the British police building and just inside the town wall and east tower. This building appears to have been established in the early 2nd c. CE and underwent renovations in the later part of the 2nd c. CE. It continued to be occupied and in the early 4th c. CE a service wing was added to the eastern side of the building and a wall was constructed between one of the rooms of this wing and the town wall (Erickson-Gini 1999: 16-18, Pl.7.4). Building XXV appears to have been destroyed in the massive earthquake of 363 CE.19

building, and part of the town wall in the eastern corner and all of this area was abandoned. The churches at the site were probably constructed after the 363 earthquake and a second face of the wall, which Negev assumed was built simultaneously with the first face (Negev 1988b: 12), may have in fact been added in order to shore up the original wall after that event. 20 Unpublished notes by Cohen report that a coin of Hadrian minted in Ashkelon was found inside the small cistern in the building. In the 2001 excavations two Nabataean coins were found in the middens outside the building. These three coins are the only coins that have been found at the site. 21 Two or three unexcavated towers are located along the highway between Mampsis and the Dead Sea at Rotem Junction and along the Peres Ascent. Quarried steps can be seen on the opposite side of a hill with one of these towers above the turnoff to the Small Crater.

18

According to Negev, the earliest coins found at the site are those of Aretas IV (9 BC-40 CE), Malichus II (40-70 CE) and Rabel II (70-106 CE); (Negev 1988a: 27, 75-76). 19 The remains of the building were stripped of building material along with Building XIV, located on the opposite side of the British police

95

THE LATE ROMAN ARMY IN THE EAST corner towers. Gichon’s excavations revealed that the fort was built in the late 3rd or early 4th c. CE but he changed his dating to the Nabataean period on the basis of similarities in the masonry with that of the army camp at Oboda, a suggestion which the writer finds unacceptable (Gichon 1976: 91; see Erickson-Gini 2002: 117).

Tsafir, located 6 km. northwest of Mezad Sayif, Horvat Tsafir, located on the ascent, Mezad Tsafir located at the top of the ascent and Mezad Yorqeam located between Mezad Tsafir and Mampsis (Cohen 2000: 100-103). Along part of the pass wide steps were quarried into the bedrock similar to those found in other parts of the country from the same period (Harel 1959: 175-179; Kloner 1998: 123-127). This line of forts appears to have been constructed in conjunction with the construction at ‘En Hazeva in the late 3rd c. CE and they continued to be occupied as late as the early 6th c. CE.

Mezad ‘En Hazeva and the Scorpions Ascent The site of ‘En Hazeva is probably that of Thamara described by Eusebius as located “One day from Mampsis on the road from Hebron to Aela, where there is now a garrison,” (On. 8). According to the excavators, the site was occupied as early as the 10th c. BCE and continued to be occupied in the 9th-8th c. BCE. It was reoccupied by the Nabataeans in the later part of the 1st millennium BCE (Cohen and Israel 1995: 112).22 An examination of the coins and pottery found at the site have revealed that the site was occupied from the 1st- early 3rd c. CE when it was apparently abandoned (Y. Israel, p.c.).23 Towards the end of the 3rd c. CE new construction was carried out on part of the earlier Iron Age fortress. A fort with four projecting corner towers, measuring 46 x 46 m., was constructed on the tell and what appears to be a cavalry camp and adjoining bath house (measuring 38 x 51 m.) was constructed on the plain south of the fort. The camp was originally accessed by way of an entrance with two arches facing the east. An underground vault, tentatively identified as a treasury vault by the writer, was constructed directly in front of the entranceway and east of the principia (Erickson-Gini 2004: Fig. 1.59). The construction at the site in this phase was related to the military buildup in the general area and may have also been related to the mining activities southeast of the site at ancient Phaeno (Feinan) that were in use in that period as reported by Eusebius (HE VIII.13.5, Mart. Pal. 7.4; 13.1-3; On. 115). The site appears to have been heavily damaged by the earthquake of 363 CE after which the fort, the camp and the adjoining bath house were reconstructed. The plan of the camp was then revised and, in its second phase, the building was accessed from the north and additional rooms and stables were added to the structure. The underground vault was left in a state of collapse with the exception of the southeast corner where a smaller vault was constructed, which took up less than a quarter of the original area.

To the southwest of ‘En Hazeva, on the way to the mining site of Feinan, are located the remains of a small fort dated by the presence of Late Roman (Early Byzantine) coins (Porat 1978: 50). This was a fort with four projecting corner towers and casemate rooms situated around an open courtyard. Sites in the Southern Negev The site of Beer Menuha is located opposite Petra and Wadi Musa on the western side of the Arava Valley. According to the excavator, it was originally occupied in the Hellenistic period by the Nabataeans and reoccupied by them in the late 1st c. BCE or early 1st c. CE (Cohen 1983: 68-69). A tower built in the 2nd c. CE above the original building was occupied until the late 2nd – early 3rd c. CE when the site was abandoned. At Horvat Dafit, located approximately 20 km. north of ancient Aila (modern Aqaba) in the western Arava Valley, the Nabataeans appear to have constructed a caravansary in the 1st c. CE. In the latest phase of the site, now dated to the late 2nd - early 3rd c. CE, a tower measuring 5 x 6 m. and adjoining courtyard were constructed on the site (Cohen 1984: 16-17; B.J. Dolinka, p.c.). A detailed examination of the pottery and other material evidence from the site, originally excavated by Cohen and currently being published by Dolinka, shows that it was abandoned sometime in the early 3rd c. CE. In the Diocletianic period a fort with projecting corner towers was constructed at the oasis of Yotvata in the southern Arava Valley, originally considered to be the ancient site of Ad Dianam of the Peutinger Map. This fort, measuring 40 x 40 m. was built from mud brick on a stone foundation. The fort was excavated by Meshel. A Tetarchic inscription was subsequently discovered at the site with the name Costia (Meshel and Roll 1987: 257262; Meshel 1989).24 Three groups of Roman milestones with 8-10 milestones in each group bearing painted inscriptions, dated to 284-324 CE, were found north of the site by Avner. The inscriptions refer to distances from “Osia” prompting the suggestion that “Osia” may be identified as the “Costia” of the inscription from the fort (Avner 1995: 120).

A line of small forts was constructed in the same period along the road between ‘En Hazeva and the Ma’ale Tsafir or Scorpions Ascent route leading towards Mampis. These include (from southeast to northwest): Mezad Sayif, located 8 km. northwest of ‘En Hazeva, Rogem 22

An examination of the pottery at the site by the writer revealed a very small amount of Hellenistic Nabataean pottery sherds. The site was substantially re-occupied by the Nabataeans at the end of the 1st c. BCE or early 1st c. CE. 23 A similar gap in the dates of coins has been indicated at Nessana where the excavators reported coins up until the reign of Septimius Severus (197/198) and from the second half of the 3rd c. (Colt 1962: 16; Bellinger 1962: 70).

24 Excavations in the fort have been recently resumed by J. Magness G. Davies and U. Avner 2004.

96

TALI ERICKSON-GINI: THE NABATAEAAN – ROMAN NEGEV IN THE THIRD CENTURY CE operating on the Euphrates where the Persian forces were concentrated (ibid, 70, 129). Thus, the possibility that Roman military forces in the Provincia Arabia were not redeployed until the conquest of Palmyra in 272 CE may account for the apparent gap in the occupation of Roman military sites such as ‘En Hazeva in the mid-third century. In this light the hoard of 10,800 coins found at Mampsis described here above may be explained. The nature of the hoard, made up of silver coins and particularly silver tetradrachms, points to the possibility that it could have been used for military payments or a donativa. Donatives were paid by emperors to Roman troops in silver or gold until the later third century when this practice declined (Harl 1996: 222). The historian Herodian reported that Severus Alexander granted a donativa to Roman forces both before and after his Persian campaign (Herodian VI.4.1; VI.6.4).

Conclusions A review of the material evidence found in the Negev from the third century CE reveals two important trends: the first is the apparent abandonment of a large number of sites in the first half of the 3rd c. CE; the second is the construction of military installations in throughout the Negev in the late 3rd c. CE during the reign of Diocletian. The first trend appears to be the directly connected with the decline of international trade through the Negev in the first half of the 3rd c. CE. This is most obvious along the Petra – Gaza road between and including Moyat Awad and Oboda. Notably, during the same period a decline in long distance trade through the Red Sea ports of Egypt has also been indicated (Sartre 2001: 845; Sidebotham 1991: 34). The important trade station at Medain Salih (Hegra) in the northern Hedjaz, guarded by Roman military personnel, was also abandoned in the early 3rd c. (Young 2001: 127). The decline of international trade through the Negev is paralleled with the cessation of production of Nabataean unguentaria at Petra used in the manufacture of perfumed oils. The latest vessels of this type have been dated by Johnson to the period between 225 and 250 CE (Johnson 1987: 64-66). This type of vessel disappears in this period and is no longer found in the next phase of ceramic assemblages dated to the late 3rd and 4th centuries CE. Similarly, Petra and other cities in southern and central Transjordan stopped issuing city coins during or immediately following the reign of Elagabalus (218-222), (Kindler 1983: 78; Fiema 1991: 114). The cessation of coinage from these sites remains unexplained and predates the cessation of mints in other Eastern cities by several decades.

The decline of the Petra – Gaza road had an adverse effect on the inhabitants in the Central Negev and this can be seen in the decline in the quality of construction and pottery production as well as the cessation of imported vessels. The decline did not necessarily mean that the inhabitants abandoned the Negev towns. Inscriptions found at Oboda dating to the middle of the 3rd c. CE indicate that the town continued to be inhabited and functioned in this period. Similarly, archaeological excavations carried out at Mampsis have revealed indications that it was occupied continuously throughout the period. Dated inscriptions found in the form of graffiti on bedrock outcroppings and boulders in the Negev and Sinai in Thamudic, Safaitic, Nabataean and Greek scripts appear to end around 270 CE (Negev 1991: 213).

In some cases there is indirect evidence of the possibility that epidemics may be responsible for the temporary or permanent abandonment of buildings or whole sites in the Negev in the first half of the 3rd c.25 This is particularly evident at Oboda where an entire structure including the kitchen pantry and foodstuffs were left untouched and eventually covered over. The abandonment of military sites in this period may have also been a result of the redeployment of military personnel. The abrupt cessation of coins found at these sites from the time of Elagabalus (218-222) may be an indication that Roman army units from the Provincia Arabia took part in the invasion led by Severus Alexander around 230 CE. One historical source, Moses of Chorene, reported that forces deployed in the invasion were raised from Egypt to the Black Sea and “from the desert” (Ensslin 1965: 128). The invasion against the Persian army under Ardashir apparently resulted in significant losses to the Roman right column

A turning point occurred in the last decade of the 3rd c. CE with the substantial build up of Roman military forces and the construction of numerous camps, forts and towers along key roads in the Negev and Arava Valley as well as the fortification of Mampsis. According to the Onomasticon of Eusebius, in this period the Tenth Legion Fretensis, was transferred from Jerusalem to the Red Sea port of Aila and an impressive list of Roman garrions (phrouria) were “posted beside the desert” including Bersabee (Beersheva), Chermala, Thamara (‘En Hazeva) on the road between Hebron and Aila, Zoara, Arielda (Gharandal), Arnonas (Wadi Mujib) and Mephaath (Kastron Mefaa at Umm er-Rasas), (On. 11, 42, 50, 118, 129; Millar 1996: 188). John Malalas, a later source, refers to the establishment of forts along the limites ‘from Egypt up to the Persian fronteir’ manned by limitanei (border troops), (Jo. Mal. Chron. XII). This buildup in military installations and personnel is confirmed by Ammianus Marcellinus who served in the Roman army on the eastern frontier in the mid-4th c. and described Arabia as filled with strong camps and forts (Amm. Marc. XIV.8.13). Another historian, Zosimus, attributed the buildup to Diocletian (Zos. II.34). Archaeological investigations in Jordan have revealed the presence of military installations from this period including the legionary camp of Lejjun, the base of the legio IV Martia

25

A comparison of the in situ ceramic vessels discovered in early 3rd c. CE contexts at Oboda and Moyat Awad with the assemblage found in the Tomb 2 in the North Ridge Tomb at Petra suggests the possibility that the thirty-six individuals buried there may have been interred in early 3rd c. CE rather than in the 1st c. CE as proposed by the excavators (Bikai and Perry 2001). Perry suggests that the multiple concurrent burials in Tomb 2 may point to an epidemic resulting in a large number of deaths. (Perry 2002: 268-269).

97

THE LATE ROMAN ARMY IN THE EAST Ben David H. 2005 The Paved Roman Road from Petra to the Arava. In Cathedra 116: 31-48. (Hebrew). Burns T.S. and Eadie J.W. (ed.) 2001 Urban Centers and Rural Contexts in Late Antiquity, East Lansing. Cohen R. 1980 The Excavations in ‘Avdat 1977. In Qadmoniot 49-50: 44-46. (Hebrew). Cohen R. 1981 Archaeological Survey of Israel, Map of Sede Boqer – East (168) 13-03, Jerusalem. Cohen R. 1982 New Light on the Date of the PetraGaza Road. In Biblical Archaeologist 45: 240-247. Cohen R. 1983 Mesad Be'er Menuha. In Excavations and Surveys in Israel 2: 68-69. Cohen R. 1984 Horvat Dafit. In Excavations and Surveys in Israel 3: 16-17. Cohen R. 1985 Archaeological Survey of Israel, Map of Sede Boqer – West (167) 12-03, Jerusalem. Cohen R. 1992 Sites in the Negev Highlands in the Hellenistic, Roman and Byzantine Periods. In E. Stern: 1053-1063. (Hebrew). Cohen R. 2000 The Ancient Roads from Petra to Gaza in Light of New Discoveries. In The Incense Routes in Orion and Goren: 54-113. (Hebrew). Cohen R. and Israel Y. 1996a 'En Hatzeva 1990-1994. In Excavations and Surveys in Israel 15: 110-116. Colt H.D. (ed.) 1962 Excavations at Nessana I, London. Duncan-Jones R.P. 1996 The Impact of the Antonine Plague. In JRA 9: 108-136. Ensslin W. 1965 Chapter IV: Sassanid Persia. Section VI. The Persian Wars with Rome. In Cambridge Ancient History Vol. XII: 126-137. Erickson-Gini T. 1999 Mampsis: A Nabataean Roman Settlement in the Central Negev Highlands in Light of the Ceramic and Architectural Evidence Found in Archaeological Excavations During 1993 1994, Unpublished M.A. dissertation, Tel Aviv University. Erickson-Gini T. 2002 Nabataean or Roman? Reconsidering the Date of the Camp at Avdat in Light of Recent Excavations. In Freeman et al.: 113-130. Erickson-Gini T. 2004 Crisis and Renewal – Settlement in the Central Negev in the Third and Fourth Centuries CE with an Emphasis on the Finds from Recent Excavations in Mampsis, Oboda and Mezad ‘En Hazeva, Unpublished Ph.D. dissertation, The Hebrew University in Jerusalem. Erickson-Gini T. and Israel Y. 2003 Recent Advances in the Research of the Nabataean and Roman Negev. In Rosenthal-Heginbottom: 9*-14* (English), 11-14 (Hebrew). Fabian P. 1996 Evidence of Earthquake Destruction in the Archaeological Record – The Case of Ancient Avdat. In Big Cities World Conference on Natural Disaster Mitigation in Conjunction with the Tenth International Seminar on Earthquake Prognostics. Abstracts, Jan. 5-10, 1996, Cairo, Egypt: 25. Fabian P. 2001 New Excavations in the Roman Army Camp at Avdat. In Proceedings of the Twenty-Seventh Israel Archaeological Congress, April 2-3, 2001. Bar Ilan University: 18.

at Bethoro listed in the Notitia Dignitatum (ND 37, 22) and two nearby forts at Khirbet el-Fityan and Rujm Beni Yasser (Parker and Richard 1987) as well as the camp of Da’ajaniya located south of the Wadi el-Hasa to name only a few examples. This military buildup in the Diocletianic period has left an indelible mark on the landscape of the Negev and Arava regions in the form of fortifications and fortified roads extending through the region into southern Jordan and the Red Sea. The subsequent effect of the influx of military personnel and installations on the local economy is apparent in the florescence of settlement and agricultural production in the Central Negev from the 4th century and well into the Late Byzantine period. Bibliography Alt, A. (1935) ‘Aus der Araba II: Romische Kastelle und Strasen,’ ZDPV 58: 1-59. Amm. Marc. Ammianus Marcellinus, Rerum gestarum libri, ed. J. Rolfe, London 1950-1952. Amm. Marc. The Later Roman Empire A.D. (354-378), transl. W. Hamilton, London 1986. Bikai, P.M. and Perry, M.A. (2001) Petra North Ridge Project Tombs 1 and 2: Preliminary Report. BASOR 104: 59-78. Eus. Eusebius Caesariensis episcopus. Eus. – Chron. Chronicon, ed. R. Helm, Die Chronik des Hieronymus, GCS 47, Berlin 1956. Eus. – HE Historia Ecclesiastica, ed. E. Schwartz, Die Kirchengeschichte, GCS 9 i-ii, Leipzig 19031908. Eus. – Mart. Pal. De martyribus Palaestinae, ed. E. Schwartz, Über die Martyrer in Palastina, GCS 9 ii, Leipzig 1908: 907-950. Eus. On. The Onomasticon: By Eusebius of Caesarea, transl. G.S.P. Freeman-Grenville, Jerusalem 2003. Diod. Diodorus of Sicily (1st c. B.C.E.), Bibliotheca historica, ed. C.H. Oldfather, London 1960-1967. Diod. Diodorus of Sicily, 12 volumes. transl. R.M. Geer. Cambridge, Mass. 1954. Herodion Books V-VII. transl. C.R. Whittaker, Cambridge Mass. 1970. Jo. Mal. Joannes Malalas, Chronographia, ed. L. Dindorf, CSHB 13, Bonn 1831. ND Or. Notitia Dignitatum: Notitia Orientis, ed. O. Seeck, 1876, Frankfurt a.M. 1962. Zos. Zosimus, Historia Nova. transl. J. Buchanan and H.T. Davis, San Antonio 1967. Avner U. 1995 The Southern Arava – Roman Milestones. In Hadshot Arkheologiyot 104: 120-121. (Hebrew). Avner U. - Davies G. and Magness J. 2004 The Roman Fort at Yotvata: Interim Report (2003). In JRA 17: 405-412. Begley V. and de Puma R. (ed.) 1991 Rome and India: The Ancient Sea Trade, Madison, Wisconsin. Bellinger A.R. 1962 Coins. In Colt: 70-75. 98

TALI ERICKSON-GINI: THE NABATAEAAN – ROMAN NEGEV IN THE THIRD CENTURY CE Meshel Z. and Roll I. 1987 A Fort and Inscription from the Time of Diocletian at Yotvata. In Eretz Israel 19: 248-265. (Hebrew). Meshel Z. and Tsafrir Y. 1974 The Nabataean Road from ‘Avdat to Sha’ar Ramon. In Palestine Exploration Quarterly 106: 103-118. Meshel Z. and Tsafrir Y. 1975 The Nabataean Road from ‘Avdat to Sha’ar Ramon (concluded). In Palestine Exploration Quarterly 108: 1-21. Millar F. 1996 The Roman Near East: 31 BC – AD 337, Harvard University Press, Cambridge, Mass., London. Negev A. 1961 Nabatean Inscriptions from ‘Avdat (Oboda). In Israel Exploration Journal 11: 127-138. Negev A. 1969 The Chronology of the Middle Nabatean Period. In Palestine Exploration Quarterly 101: 5-14. Negev A. 1977 The Nabateans and the Provincia Arabia. In Temporini and Haase: 520-686. Negev A. 1981 The Greek Inscriptions from the Negev, (Studium Biblicum Franciscanum. Collection Minor N. 25), Jerusalem. Negev A. 1988a The Architecture of Mampsis, Final Report. Vol. I. The Middle and Late Nabatean Periods, (Qedem 26), Jerusalem. Negev A. 1988b The Architecture of Mampsis, Final Report. Vol. II Late Roman and Byzantine Periods, (Qedem 27), Jerusalem. Negev A. 1991 Personal Names in the Nabatean Realm, (Qedem 32), Jerusalem. Negev A. 1997 The Architecture of Oboda, Final Report, (Qedem 36), Jerusalem. Parker S.T. (ed.) 1987 The Roman Frontier in Central Jordan: Interim Report on the Limes Arabicus Project, 1980-1985, (BAR International Series, 340), Oxford. Parker S.T. and Richard S. 1987 The Roman Castellum of Khirbet el-Fityan. In Parker: 429-446. Perry, M. A. (2002) ‘Life and Death in Nabataea: The North Ridge Tombs and Nabataean Burial Practices,’ Near Eastern Archaeology 65: 4. Pp. 265-270. Porat Y. 1978 En Yahav. In Hadshot Arkheologiyot 61-62: 50-51. (Hebrew). Orion E. and Goren A. (ed.), 2000 The Incense Roads, Midreshet Sde Boker. (Hebrew). Roll I. 1989 A Latin Imperial Inscription from the Time of Diocletian Found at Yotvata. In Israel Exploration Journal 39 (3-4): 239-260. Rosenthal-Heginbottom R. (ed.) 2003 The Nabateans in the Negev, Reuben and Edith Hecht Museum, University of Haifa, Haifa. Sartre M. 2001 D’Alexandre a Zenobie, (Histoire du Levant antique), Fayard. Schmid S.G. 2001 Chapter 11. The Nabataeans: Travellers between Lifestyles. In MacDonald et al.: 367-427. Sidebotham S.E. 1991 Ports of the Red Sea and the Arabia-India Trade. In Begley and de Puma: 12-38.

Fiema Z.T. 1991 Economics, Administration and Demography of Late Roman and Byzantine Southern Transjordan, Ph.D. Dissertation, The University of Utah. Fiema Z.T. 2001 Byzantine Petra – A Reassessment. Urban Centers and Rural Contexts in Late Antiquity. In Burns and Eadie: 111-131. Fiema Z.T. 2002 From the Annexation to Aaron: Petra in Roman and Byzantine Times. In Frosen and Fiema: 60-73. Fiema Z.T. and Jones R.N. 1990 The Nabataean KingList Revised: Further Observations on the Second Nabataean Inscription from Tell esh-Shuqafiya, Egypt. In Annual of the Department of Antiquities of Jordan 34: 239-248. Figueras P. 1992 .The Roman Worship of Athena-Allatin in the Decapolis and the Negev. In ARAM 4:1&2: 173-183. Freeman P.W.M. – Bennett J. – Fiema Z.T. and Hoffmann B. (ed.) 2002 Limes XVIII. Proceedings of the XVIIIth International Congress of Roman Frontier Studies held in Amman, Jordan (September 2000). (BAR Int. Ser. 1084, Vol. I). Oxford. Frosen J. and Fiema Z.T. (ed.), 2002 Petra – A City Forgotten and Rediscovered, Helsinki. Gichon M. 1971 Migdal Tsafit. In Hadshot Arkheologiyot 11: 28 (Hebrew). Gichon M. 1976 Excavations at Mezad Tamar – “Tamara” 1973-1975, Preliminary Report. In Sonderdruck Aus Dem Saalburg-Jahrbuch 33: 80-95. Harel M. 1959 The Roman Road at Ma’aleh ‘Aqrabbim. Israel. In Exploration Journal 9: 175-179. Harl K.W. 1996 Coinage in the Roman Economy, 300 B.C. to A.D. 700, Baltimore-London. Israel Y. and Erickson-Gini T. 2001-2002 Mesad Horvat Bor. In Bulletin of the Anglo-Israel Archaeological Society 19-20: 7-17. Johnson D.J. 1987 Nabataean Trade: Intensification and Culture Change, Ph.D. Dissertation, The University of Utah. Ann Arbor. Kennedy D.L. 2000 The Roman Army in Jordan, London. Kindler A. 1983 The Coinage of Bostra, Warminster. Kloner A. 1998 Stepped Roads in Roman Palestine. In ARAM 8: 111-137. Kohn G.C. 1998 Encyclopedia of Plague and Pestilence, Herfordshire. Korjenkov A.M. and Erickson-Gini T. 2003 The Seismic Origin of the Destruction of the Nabataean Forts of Ein Erga and Ein Rahel, Arava Valley, Israel. In Archaeologischer Anzeiger 2: 39-50. Korjenkov A.M. and Mazor E. 1999 Seismogenic Origin of Ancient Avdat Ruins, Negev Desert, Israel. In Natural Hazards 18: 193-226. MacDonald B. – Adams R. and Bienkowski P. (ed.) 2001 The Archaeology of Jordan, Stafford, England. Meshel Z. 1989 A Fort at Yotvata from the Time of Diocletian. In Israel Exploration Journal 39: 228238.

99

THE LATE ROMAN ARMY IN THE EAST Smith, A.M. II (2005) ‘Bir Madhkur Project: A Preliminary Report on Recent Fieldwork,’ BASOR 340: 5-75. Stern E. (ed.) 1992 The New Encyclopedia of Archaeological Excavations in the Holy Land. Vols. 1-4, Jerusalem. (Hebrew). Temporini M. and Haase W. (ed.) 1977 Aufstieg und Niedergang er Romischen Welt. II.6, Berlin-New York. Young G.K. 2001 Rome’s Eastern Trade. International Commerce and Imperial Policy 31 BC – AD 305, London-New York.

100

The Paved Road from Petra to the ‘Arabah – Commercial Nabataean or Military Roman? Chaim Ben-David Jordan Valley College, Zemach

I shall not deal in this paper with the question of the exact route of the incense road in the vicinity of Petra and toward the west to the Negev7. Here only the topographical and archaeological data concerning the constructed road descending from Petra to the southwest will be dealt with. This road was furnished in my view in the late Roman period. I shall limit myself to one remark on the character of the incense Road: throughout the vast geographical space trodden by Nabataean caravans, not a single consructed track, presumably related to this road was ever reported. In addition, the Negev road section between ‘Avdat and Mǀyet AwƗd, especially the constructed road and the milestones along the road8 have not been proved to have been built during the Nabataean period (before 106 CE). These are in my view Roman, though the track was no doubt in use by the Nabataean trade as a camel path.

Numerous tours and researches have been conduced during the last one hundred years in the region close to Petra in order to follow the caravan routes which have made their way from Petra to the Mediterranean ports. This paper will focus on the description and analysis of what is, in my view, the only paved road in this area.1 Remains of this road can be recognized in parallel to the dirt-road that exists Petra westward, beyond the southern flanks of Jabal HƗrnjn, and from there along the camelpath called Naqb er-Ruba’Ư, which descends to the south in the direction of WƗdƯ KhušeƯba and afterwards in a westerly direction toward the ‘Arabah. To the length of approximately 10 km there are convincing remains of a built road in the width of 4 m, the customary width of main roads in mountainous regions during the Roman period. A basic tenet in many investigations is that one of the principal routes of the incense trade in antiquity arrived from the East to Petra and thence to Gaza by way of the ’Arabah valley and the Negev. This route, often referred to as the “Incense Road” or the “Frankincense Road” , was investigated in the course of the twentieth century, mainly in the Negev highlands.2 The section between the ’Arabah and Petra has been less intensively studied3 than the sections through the Negev and it has been generally proposed that the route crossed the ‘Arabah from Mǀyat ‘AwƗd to the BƯr Madhknjr Station and from there by way of WƗdƯ MnjsƗ to Petra.4

The Topography of the Roads from Petra to the ‘Arabah Anyone who seeks to reach the ‘Arabah from the Edom mountain plateau must cut through first the easily passed sand-stone cliffs and then, in a very steep slope, through the igneous rock layer. Most of the ancient roads from the mountains of Edom to the ‘Arabah, passed through natural gaps in order to avoid confrontation with the steep descent of the igneous rocks. The larger gaps in the ‘Wall’ of the mountains of Edom are found to the northwest and southwest of Petra. In the north there are two such gaps: 1) in the limestone ridge south of W. Hanzirah there is no cliff at all. This path is known as Naqeb Dahal, and today a dirt-road passes there; 2) W. Dana, cutting-through a geological fault’ allows a comfortable descent toward Feinan. Today a dirt-road which is not in use (incorporated in the Nature Reserve) can be found there. To the south of Petra there is only one natural gap: the limestone ridge of Dilagha—Naqeb Mas’njdah. Today this pass is an asphalt road descending to the ‘Arabah.

New suggestions regarding the track of the incense road from the ‘Arabah to Petra have been put forward in recent years.5 In addition, there were scholars who argued that this road by-passed Petra itself.6 However, the above suggestions did not pertain to a constructed Roman road but to paths in various levels of arrangement and comfort. None discussed a section of a constructed at least 4 m wide road, as usual in the Roman period remains. The absence of such findings in the vicinity of Petra is all the more striking when compared to the considerably long sections of such roads discovered in the Negev which were always ascribed to the Gaza Petra road.

It is noteworthy that immediately to the west of Petra the igneous rocks form a continuous cliff stretching from the area of Wadi Barwas in the north to Jabal HƗrnjn in the south. Therefore in any departure from Petra toward the west, one is forced to pass the igneous rocks cliff, and because of the dry waterfalls one cannot descend through

1

This paper is based on many hiking tours made in the Petra area in the years 1995-2003. Other participants were G. Peli, A. Shmida, A. Kloner, Z. Meshel, I. Haviv, E. Raz, I. Roll, S. Ravek, R. Porat and A. Izdarechet. 2 Frank 1934: 270-276; Alt 1935; Kirk 1938: 231-234; Negev 1968; Meshel and Tsafrir 1974, 1975; Cohen 1982. 3 Zayadine 1985, 1992. 4 Roll 1999; Parker 2000: 368; Kennedy 2004: 213. 5 Kloner 1996: 127-134. 6 Mildenberg 1995, 1996; Goren 2000.

7

See now the most important discoveries of Andrew Smith (2005, sites 24-25) of new Nabataean carvanserais in the ‘Arabah and his suggestion of a new route that connected Petra and the Negev 8 Meshel and Tsafrir 1974, 1975.

101

THE LATE ROMAN ARMY IN THE EAST

Fig. 1 The limestone ridge to the south of Jabal Harun where the paved road is descending to the ’Araba the wƗdƯs unless he uses snap-link ropes.9 Nevertheless, numerous trail lines are drawn from Petra westward, some harder to pass than others, however, not a single one of them produced the remains of a Roman Road. Of the above trails one of the most important is Naqeb Namala,10 originally a camel track, along which a dirtroad was furnished in the 1980’ and in 2006 it was paved with asphalt. The Nabataean11 and Greek12 inscriptions in its upper part, the fort at its bottom13 and the fact that it uses good water sources in the ‘Arabah, all point to the prominence of this passage in comparison to the other tracks in the area. It seems that Naqeb Namala was an important camel route leading from Petra to the Negev.14

Naqeb er-Ruba’Ư are found at 1,150 asl. Only to the south of Jabal HƗrnjn the igneous rocks wall is broken, as a result of a geological fault, and a north-south lime-stone ridge descends there from the altitude of 1,100 to 600 m to end in the creek of WƗdƯ KhušeƯba(fig 1). On this moderate ridge the Petra-’Arabah Roman road was constructed. Naqeb er-Ruba’Ư and the Remains of the Roman Road Over the last few hundred years a moderate camel naqeb, known as Naqeb er-Ruba’Ư ascends the limestone ridge to the south of Jabal HƗrnjn. Most of the travelers and researches that reached Petra from the direction of the southern ‘Arabah used this route.15 They failed, however, to refer to the Roman road-works accompanying it. Nevertheless, William Libbey and Franklin Hoskins16, who passed on the road in 1902 wrote: “the road leading out of Petra at this point is called the Gaza road, because after two hours it swings round south of Mount Hor …This road was once a Roman road as is plainly seen at

Petra is situated at an altitude of 900 m above sea level, and the western range of the mountains to its west (mentioned above) is at an altitude of 1,100-1,200 m. Therefore every road to the west first has to moderately ascend – the tops of Naqebs Namala, Garon-’Aqeb, and 9

Haviv 2000: 174-176. Haviv 2000: 154, 163. 11 Roche and Zayadine 1999. 12 Zayadine 1993: 92-94; Sartre 1993: 60-61. 13 Frank 1934: 228. 14 Ravek 2003: 58-69. 10

15

Laborde 1838: 158; Lindsay 1838: 34; Olin 1843: 457-458; Wilson 1847: 139; Bartlett 1862: 139; Palmer 1871: 50-51; Kitchener 1884: 211-212; Hart 1885: 255-257. 16 Libbey and Hoskins 1905: 239.

102

CHAIM BEN-DAVID: THE PAVED ROAD FROM PETRA TO THE ‘ARABAH a score of points.” Major Claude Jarvis,17 who ascended Naqeb er-Ruba’Ư on his way to Petra’ wrote an article on the road to Petra from the west the subtitle of which was: “A Forgotten Roman Highway.” According to his descriptions his company drove vehicles from Kuntilla in Sinai through the ‘Arabah and from there up WƗdƯ Khnjšeiba as far as they could proceed. From there, mounted on camels and horses they took what Jarvis called a bedouin camel trail. After some distance they noticed that the road is wider and in sections at the foot of Jabal HƗrnjn they could see the built up margins of the Roman road and at least three destroyed structures along the way. Jarvis believed that the road was originally Nabataean and that the Romans only improved and reinforced it. The Jordanian scholar Fawzi Zayadine18 toured in the beginning of the 1980’ along Naqeb er-Ruba’Ư and remarked that it is the most comfortable camel trail that ascends to Petra. He has identified in this route the track of Sultan Baybars’ who passed Petra in 1276 CE on his way from Egypt to Karak, since Baybars’ itinerary includes the name Naqeb er-Ruba’Ư and the Tomb of HƗrnjn. Zayadine19 also emphasized the constructed sections of Naqeb er-Ruba’Ư, which was in his view part of the main road from Petra to Sinai and Egypt. In their article “The incense Roads between Petra and Bir Madhknjr,” Chen Katz and Ezra Orion20 pointed out that constructed sections of a road are observable between Petra and Ras en-Naqeb, at the south foot of Jabal HƗrnjn. The southernmost section they describe as follows: “most of the course is flattened and to its sides margin-stones were lined up, similar to the constructed sections in the Negev. The width of the road in the better preserved sections is about 4 m.” A Finnish expedition working in Jabal HƗrnjn and its vicinity in the last years.21 Its members recorded and even excavated parts of the Roman road at the foot of Jabal HƗrnjn . In their view this road, which exhibited constructed sections, is part of the route from the ‘Arabah via W, KhušeƯba to Petra. In the Finnish publications22 there is a detailed description of a 2 km long section where the road reaches its maximum width of 4.20 m. Two trenches across the track where excavated, where the margin-stones were clearly seen, as well as several small structures along the road which they interpreted as watch-towers. According to their report the pottery found in those structures is Nabataean/Roman. In their view the path was in use already in the Nabataean Period23 though the road works must probably date to later periods.

Fig. 2 The road carved in the sandstone near Petra Manfred Lindner, who surveyed and researched the region of Petra for 25 years, published a article24 on the system of the ancient trails ascending from the area of Qasr Umm Ratam on the bank of the lower part of W. MnjsƗ, toward Petra . Having discussed in detail seven passages between Petra and Umm Ratam, Lindner wrote “One is at a loss to define any one of the enumerated passages as ‘the road between Petra and Umm Ratam’. At different times with different animals, different loads, different people, one or another passage may have been chosen.” This author subscribes to the above definition, having ascended and descended the above paths in person. In the conclusion to his paper Lindner wrote that the passages between Petra and Umm Ratam were not the “main traffic routes” in antiquity; these passed, according to him, in the Petra-Sabra-KhušeƯba road and along Naqeb er- Ruba’Ư to the ‘Arabah. Remains of the Road from Petra to WƗdƯ KhušeƯbah Section A: From Petra to the Crossing of WƗdƯ Naqeb at the Foot of Jabal HƗrnjn In this section the road ascends from the town pf Petra, at 900 m asl, to 1,050 m at the foot of Jabal HƗrnjn. Today there is a dirt road along the section’s entire length, and this new track destroyed carved and constructed elements along the Roman road that had been visible in the past.

17

Jarvis 1940. 18 Zayadine 1985. 19 Zayadine 1992. 20 Katz and Orion 2000: 128-129. 21 Frosen et al. 2003. 22 Frosen et al. 2001: 389-391; Hertell 2002. 23 This section is the upper part of the “new” Nabataean route found by Andrew Smith, see note 7.

24

103

Lindner et al. 2000.

THE LATE ROMAN ARMY IN THE EAST Section B: from the end of the modern road at the foot of Jabal HƗrnjn, map reference UTM 7310/3555 to map reference UTM 7398/3550 In this section the road continues to ascend to the altitude of 1,150 m asl, to a spot which is marked on the maps as Ras en-Naqeb. This is the road-segment mapped and recorded by the Finnish expedition. It shows clear remains of a road’ 4 m wide, limited along considerable sections by two retaining walls each 0.60 m thick. At map reference 7297/3554, at a distance of ca. 300 m west of the road, there is square structure (20 x 20 m) with a lot of Nabataean shards spread around it (Lindner 1989: 86). The structure is situated on the head of a promontory and dominated the descent of Naqeb Must’agilin that forks out in the direction of Qasr Umm Ratam. Some 200 m to the south, on the edge of the cliff (map reference 7297/3552), there is a tower (8 x 8 m) which dominated efficiently the top of Naqeb er-Ruba’ and all the area on the west.25 Section C: Map reference 7398/3550 through 7296/3536 In this section the road descends southward to an altitude of 880 m asl, and this is the section that is parallel to the camel-trail of Naqeb er-Ruba’Ư (=Naqeb KhušeƯba in some of the modern descriptions). The course of the wide road and its constructed margins are very prominent in this section, the present camel-trail uses generally only half of the ancient width and in some cases deviates from it by a few meters. At 7296/3539 there is a tower26 (10 x 8 m) (fig. 4), and on a lower level another structure (tower?) situated at the head of another path descending

Fig 3 The road carved in the sandstone near Petra However, even today ancient carved segments of the road can be recognized (figs. 2-3) particularly in the sandstone area near the ‘Snake Monument’.

Fig 4 A tower along the paved road 25 26

104

Lindner 1989: 86. Lindner 2003: 70-71.

CHAIM BEN-DAVID: THE PAVED ROAD FROM PETRA TO THE ‘ARABAH

Fig 5 The retaining wall of the road preserved to the height of ca. 4

Fig 6 The remains of the ancient Roman road and the modern naqeb near Abu Khusheiba

105

THE LATE ROMAN ARMY IN THE EAST km toward Risha (on the Dead Sea-’Aqabah Road) at 250 m asl. This easy approach allows gaining access from the ‘Arabah to the foot of the mountains of Edom, at 700 m asl and only from there to begin the scent (for the purpose of comparison the bottom of Naqeb Namala Is at 400 m asl and the bottom of Naqeb Sleisel is at 450 m). The comfortable approach is added to the benefit of the limestone ridge, that the moderate ascent of Naqeb erRuba’Ư takes as an infrastructure. In the south of AbnjKhušeƯba Valley and close to the track of the road a fortified Roman/Nabataean structure (18 x 22 m) was reported in the 1960’.28 The Dating of the Road The discovery of the Roman road that is not parallel to the camel-trail may clarify a fact which is strange at a first glance – the Roman road completely ignores AbnjKhušeƯba, a site that was described in previous research29 as a Nabataean settlement and a road-station at the foot of Naqeb er-Ruba’Ư. This site is found at the foot of the camel-trail leading from the bottom of WƗdƯ KhušeƯbah to Naqeb er-Ruba’Ư, a path that its lower part is a steep camel naqeb with no sign of a wide road (fig 6). As mentioned above Jarvis30 pointed out that the beginning of the ascent in Naqeb er-Ruba’Ư from WƗdƯ KhušeƯbah is in a narrow camel path and only higher up he noticed that the road is wider. Fig 7 The route of the paved road from Petra to the 'Arabah

According to Lindner’s report the shards collected at Abnj-KhušeƯbah are Nabataeans and Roman from the first through the third centuries CE, and the only coin found is of the Nabataean king Rabel II (70-106 CE). The fact that the Roman road descending from Petra does not lead, in its lower part, to Abnj-KhušeƯbah, indicates that it is later than this site, not earlier than the beginning of the fourth or the very end of the third Century CE.

Westward.27 At 7296/3538 above a creek, there is a retaining wall of the road preserved to the height of ca. 4 m which shows ten building courses (fig. 5). Section D: 7296/3536 through 7289/3553 In this section the road turns westward and descends from 880 m, first moderately and then steeply to the streambed of WƗdƯ KhušeƯba at 700 m asl. This is a section of the Roman road which deviates from the present trail of Naqeb er-Ruba’Ư and that is perhaps why it was not described thus far. In the vicinity of 7295/3555 the lower retaining wall of the road is well preserved for the length of tens of meters and it is visible even from afar. The road continues to descend very steeply, a phenomenon not known from comparable Roman roads. Probably because of this steepness the road became a gully for torrents of water, so much so that it is impossible for camels to pass; hence this section is not in use today

It is possible that this paved road is connected with the extensive military operation that took place in our region at the end of the third century CE’ in the days of the emperor Diocletian.31 The transferal of the 10th legion’s camp to Aila-’Aqabah , the construction of the Roman camp at Udruh, close to Petra,32 the construction of the Roman fort in Yotveta,33 and the placing of the inscribed milestones found near Yahel in the ‘Arabah.34 The above are the more prominent signs of this activity. The construction of the Roman road along Naqeb erRuba’Ư toward Abnj-KhušeƯbah Valley indicates the southward direction of the road (fig.7). It does not lead west and definitely not northwest toward ‘Avdat and Gaza. Therefore it seems that it was furnished to connect

This steep and eroded section of the Roman road is the westernmost part recognized today in the field. From here the track continues at the bottom of WƗdƯ KhušeƯba and can be driven in cars. After some 2 km inside the WƗdƯ the road emerges to Abnj-KhušeƯba Valley at the altitude of 600 m asl. From there is a comfortable route (today mostly on asphalt) that descends moderately for ca. 15

28

Kind 1965: 70-71. Lindner 1992, 2003: 66-70. Jarvis 1940: 145. 31 Parker 2000: 372-379; Roll 1989; Lewin 2002. 32 Kennedy 2004: 178-180. 33 Meshel 1989; Roll 1989. 34 Avner 1996; Isaac 1998: 71-75; Kennedy 2004: 205-206. 29 30

27 Here starts Naqeb Um Qamar the “new” Nabataean route found by Andrew Smith, see note 7.

106

CHAIM BEN-DAVID: THE PAVED ROAD FROM PETRA TO THE ‘ARABAH

Fig 8 The paved road in the Araba discovered by Andrew Smith

Fig 9 A milestone along the ‘incence road’ in the Negev 107

THE LATE ROMAN ARMY IN THE EAST to the region of Petra (the vertical ascent is only 300 m). Along the ascent there are the remains of a square tower (5 x 5 m) and at the top a square water cistern. At the very end of the third century CE the southern parts of the Provincia Arabia were transferred to Palaestina, as a consequence of which Petra became the capital of Palaestina Tertia. At the same time the 10th Legion was moved from Jerusalem-Aelia Capitolina to Aila on the coast of the Red Sea. It appears to me that this is the most appropriate time for the construction of a Roman military road that was meant to connect Petra with Aila. This road took the moderately descending limestone ridge to the south of Jabal HƗrnjn, and descended from Petra in a southwest direction through WƗdƯ KhnjšeƯbah to the ‘Arabah. This road heading to Aila was probably constructed by Diocletian as part of the vast military activity in the region, in particular the transferal of the 10th Legion to Aila. One cannot rule out the possibility that the constructed sections with stone margins and the milestones which were found between ‘Avdat and Sha’ar Ramon (fig. 9) also belong to Roman military activity,37 and that their later phase38 may date to the days of Diocletian. These sections were apparently aimed to connect the Negev towns (Elusa and ‘Avdat) to Aila. According to the results of the excavations in the military camp at ‘Avdat,39 it was constructed at the end of the third or beginning of the fourth century CE.40 This perfectly suits my suggestion.

Fig 10 Roman military roads leading to Aila Petra to Aila – the location of the 10th Legion from the days of Emperor Diocletian onwards. In the southeast ‘Arabah south of the Roman fortress of Gharandal, sections of ‘paved’ Roman road in a north-south direction were discovered35 (fig. 8), and close to the fort of qa’ esSa’ƯdƯn a quarry of milestones was located.36 It is highly probable that these sections form also a part of the PetraAila Road .

According to the present suggestion, the constructed road between ‘Avdat and Mǀyet ‘AwƗd, which in the Nabataean period were merely part of a camel-track to Petra, did not continue as a Roman road eastward to Petra’ but turned south, to the area of Yahel in the ‘Arabah, where the Diocletianic milestones were found, and further south to Yotvata, were an inscription of that emperor was located. That road led to Aila, the headquarters of the 10th Legion, not to Petra (fig. 10).

This connection between Petra and Aila is ca. 130 km in topography that ascends to 1,150 m asl and from there downward to the ‘Arabah. Compare this to the only possible preexisting Roman road – connection to the Via Nova Trajana and descent to Aila. The latter road is only 10 km longer but it forced the traveler to ascend to the edge of the Šarah Mountains at 1600 m asl. The site of Abnj-KhušeƯbah was a road-station along the Nabataean trail that preceded the Roman road that ascended to Naqeb er-Ruba’Ư, or, and this is more plausible in my view, to Naqeb ZaqrƗ. The latter is a camel naqeb which preserved construction remains along a path 2-3 m wide, which moderately ascends through WƗdƯ ZaqrƗ toward Ras-ZaqrƗ at an altitude of 1000 m asl. It is one of the most comfortable passages that ascend

37

In this context it is noteworthy an inscription mentioning the Roman military unit cohors VI Hispanorum was found at Sha’ar Ramon. See: Figueras 1992: 176-179; Isaac 1998: 157. 38 Meshel and Tsafrir (1974: 108) in their survey noted that on the plain of nafha ‘the road divides into two closely running branches, one is of later construction and crosses and recrosses the other’. In most of the milestone stations more then one milestone was found. 39 Erickson-Gini 2002. 40 Her conclusion that the military camp functioned only a short time (ibid., 118) may explain why the constructed road is incomplete and the milestones are missing from Sha’ar Ramon eastward to the ‘Arabah. One may point out that Peter Fabian, co-director of the excavations at the military camp at ‘Avdat, believes the camp was established and was active during the second century CE (Fabian 2005).

35

Smith et al. 1997: 59-60. Lately (2007) hikers found similar stretches of a paved road in a sandy area in Wadi Sik east of Gharandal. These stretches seem to be modern so more research has to be done on the paved road in the ‘Arabah. 36 Smith et al. 1997: 60 site 134.

108

CHAIM BEN-DAVID: THE PAVED ROAD FROM PETRA TO THE ‘ARABAH Studies held in Amman, Jordan. BAR IS 1084: 91101. Libbey W. and Hoskins F.E. 1905 The Jordan Valley and Petra, New York. Lindner M. 1989 South Jordan survey. In D. HomesFredericq and J.B. Hennsey (ed.), The Archaeology of Jordan II: Field reports, Leuven: 84-90. Lindner M. 1992 Abu Khusheiba – A newly described Nabataean settlement and caravan station between Wadi Arabah and Petra. In Studies in the History and Archaeology of Jordan 4: 263-266. Lindner M. 2003 Uber Petra hinaus - Archaologische Erkundungen im Sudlichen Jordanien., Rahden. Lindner M., Hubner U. and Hubl J. 2000 Nabataean and Roman presence between Petra and Wadi Arabah survey expedition 1997/98: Umm Ratam. In ADAJ 44: 535-567. Lindsay A. W. C. 1838 Letters on Egypt, Edom and the Holy Land II, London. Meshel Z. 1989 A fort at Yotvata from the time of Diocletian. In IEJ 39: 228-238. Meshel Z. and Tsafrir Y. 1974 The Nabataean road from ‘Avdat to Sha’ar-Ramon’. In PEQ 106: 103-118. Meshel Z. and Tsafrir Y. 1975 The Nabataean road from ‘Avdat to Sha’ar-Ramon’. In PEQ 107: 3-21. Mildenberg L. 1995 Petra on the frankincense – road? In Transeuphratene 9: 69-72. Mildenberg L 1996 Petra on the frankincense road? – again. In ARAM 8: 55-65. Negev A. 1968 The Dating of the Petra-Gaza road. In PEQ 98: 89-98. Olin S. 1843 Travels in Egypt, Arabia Petraea and the Holy Land, New York. Palmer E.H. 1871 The desert of the Tih and the country of Moab. In PEQ 4: 3-73. Parker S. T. 2000 The defense of Palestine and Transjordan from Diocletian to Heraclius. In L.E. Stager, J.A. Greene and M.D. Coogan (ed.), The Archaeology of Jordan and beyond – essays in honor of James A. Sauer, Eisenbrauns: 367-388. Ravek S. 2003 Jordan – along the incense road. In Teva Hadvarim 95: 58-69 (Hebrew). Roche M.J. and F. Zayadine 1996 Exploration Epigraphique et Archeologique au Siq un el ‘Alda au nord de Petra. In Semitica 49: 123-139. Roll I. 1989 A Latin Imperial Inscription from the Time of Diocletian Found at Yotvata. In IEJ 39: 239-260. Roll I. 1999 The Roads in Roman-Byzantine Palestina and Arabia. In M. Piccirillo and E. Alliata (ed.), The Madaba Map Centenary, Jerusalem: 108-113. Sartre M. 1993 Inscriptions de la Jordanie: 4 Petra et la Nabatene, (IGLS XXI) Paris. Smith A.M. 2005 Bir Madhkur Project: A Preliminary Report on Recent Fieldwork. In BASOR 340: 57-75 Smith A.M., Stevens M. and Niemi T.M. 1997 The Southeast Araba Archaeological survey: A preliminary report of the 1994 season. In BASOR 305: 45-71.

Bibliography Alt A. 1935 Aus der Araba II. In ZPDV 58: 1-59. Avner U. 1996 Arabah Valley. In AJA 100: 762-764. Bartlett W.H. 1862 Forty days in the desert: on the tracks of the Israelites, London. Cohen R. 1982 New light on the date of the Petra-Gaza road. In Biblical Archaeologist: 45 240-247. Erickson-Gini T. 2002 Nabataean or Roman? Reconsidering the date of the camp at Avdat in light of recent excavations. In P. Freeman, J. Bennett, Z.T. Fiema and B. Hoffmann (ed.), Limes XVIII – Proceedings of the XVIIIth International Congress of Roman Frontier Studies held in Amman, Jordan. BAR IS 1084: 113-130. Fabian P. 2005 Avdat during the establishment of Provincia Arabia. Ph. D. Thesis. Ben-Guryon University (Hebrew). Figueras P. 1992 The Roman worship of Athena-Allat in the Decapolis and the Negev. In ARAM 4: 173-183. Frank F. 1934 As der Araba I. In ZPDV 57: 191-280. Frosen J. et al. 2001 The 1998-2000 Finnish Harun Project: Special reports. In ADAJ 45: 377-392. Frosen J. et al. 2002 The 2001 Finnish Jabal Harun Project: Preliminary reports. In ADAJ 46: 391-407. Goren A. 2000 Les Nabateens et la route de l’encens. In A. Lemaire (ed.), Les routes du Proche-Orient, Paris: 107-115. Hart H.C. 1885 A naturalist’s journey to Sinai, Petra and South Palestine. In PEQ 18: 231-286. Haviv I. 2000 Trekking and Canyoning in the Jordanian Dead Sea Rift, Desert Breeze Press. Hertell E. 2002 The ancient road in the FJHP survey area. In J. Frösén and Z.T. Fiema (ed.), Petra - A City Forgotten and Rediscovered, Helsinki: 234-240. Isaac B. 1998 The Near East under Roman rule selected papers, Oxford. Jarvis C. S. 1940 To Petra from the west – a forgotten Roman highway. In Antiquity 14: 138-147. Katz C. and Orion E. 2000 The incense Roads between Petra and Bir Madkhur. In E. Orion and A. Goren (ed.), The Incense routes, Sede Boker: 123-129. (Hebrew). Kennedy D. 2004 The Roman Army in Jordan, 2nd rev. edition, London. Kind H. D. 1965 Antike Kupfergewinnung zwischen Rotem unt Totem Meer. In ZPDV 81: 56-73. Kirk G.E., 1938; Archaeological Exploration in the Southern Negev. In PEQ 70: 211-235. Kitchner H.H. 1884 Major Kitchner’s report. In PEQ 17: 202-221. Kloner A. 1996 Stepped roads in Roman Palestine. In ARAM 8: 111-137. Laborde L.de 1838 Journey through Arabia Petraea, London. Lewin A. 2002 Diocletian: Politics and limites in the Near East. In P. Freeman, J. Bennett, Z.T. Fiema and B. Hoffmann (ed.), Limes XVIII – Proceedings of the XVIIIth International Congress of Roman Frontier

109

THE LATE ROMAN ARMY IN THE EAST Wilson J. 1847 The Lands of the Bible visited and described, London. Zayadine F. 1985 Caravan Routes between Egypt and Nabataea and the Voyage of Sultan Baibars to Petra in 1276. In Studies in the History and Archaeology of Jordan, II: 159-174. Zayadine F. 1992 L’espace urbain du grand Petra, les routes et les stations caravanieres. In ADAJ 36: 217240. Zayadine F. 1993 Inscriptions Grecques et Nabateennes au Nord de Petra. In SYRIA 70: 81-94.

110

Be’er Shema – Birsama of the Notitia Dignitatum: A Prolegomenon to the 2006 Excavations Benjamin J. Dolinka W.F. Albright Institute, Jerusalem*

Site Location and Regional Environment*

The vegetation surrounding Be’er Shema is moderate and consists of only a few species. The dominant plant of the area is Artemisium monosperma, but also present is Retuma raetam, Brassica tournefortti, Echiochilon fruticasum, Neurada procumbens, Cyperus capitatus, Iris helenae, Stipagrostis ciliate, Stipagrostis obtusa, Panicum turgidum, Pemisetum divisum, Thymelaea hirsuta, and Lolium multiflorum. The trees in the area are limited to the Acacia raddiana and Acacia tortilis (Waisel et al. 1978; Danin et al. 1975; Danin 1979). The abundance of water near Be’er Shema, combined with enough rainfall to leach most of the soluble salts from the soil profile and together with a relatively low annual evaporation rate of between 1600 and 1700 mm, makes it well suited for both dry agriculture and as a natural pastureland (Dan 1981: 43; Evenari et al. 1982: 33; Levie & Pohoryles 1970).

Horvat Be’er Shema / Birsama (PG 10640 / 07370) is located in the northwest Negev, ca. 25 km westnorthwest of Beersheba and 21 km southeast of Gaza, to the east of the modern Ze’elim-Urim Road and ca. 3.6 km east of the Nahal Besor. As it was situated prominently along the Petra-Gaza Road between Halutza (Elusa) and Gaza, in antiquity it was probably established by the Nabataeans and associated with the trade in aromatics. In the Late Roman and Byzantine periods, Be’er Shema was a large settlement covering an area of more than 40 hectares. It consisted of a village, a Late Roman fortress and at least one Christian church. The regional environment surrounding Be’er Shema consists of relatively flat terrain that is situated ca. 120 m ASL (Gazit 1996: 9). The site falls within the IranoTuranian Phytogeographic Region (Evenari et al. 1982: 9) and is considered a Mildly Arid Climatic Zone (Dan 1981: 18). The site is well watered due to its proximity to the Nahal Besor, the only perennial water source in the Negev lowlands (Gazit 1986: 17-29 passim), and sits atop an aquifer located ca. 25 m below the surface (Gazit & Lender 1993: 274). The average annual rainfall for the area is ca. 180 mm, and the climate is much more temperate than areas to its south, with annual mean temperatures ranging between 18 and 19º C (Rosenan 1970, Dan 1981: 11; Dan 1981a: 213; Evenari et al. 1982: 30, 34).

Ancient Sources Be’er Shema was well known to the authors of classical and late antiquity, and this is most likely due to its regional prominence. The site is first mentioned in the mid-2nd century AD Geography of Claudius Ptolemy, where it appears as ‘Berzamma / ǺȑȡȗĮȝȝĮ’ (V.16.10). Interestingly, these names are not found in the only English translation of this work by Stevenson (1932/1991). This is most likely due to his reliance upon the Germanus (1460) edition of the Geography using the Codex Ebnerianus as a source, which lists the site as ‘Berzaba’ or Beersheba. However, these names do appear in the 1618 translation of Bertio, where Be’er Shema is listed first in the towns of Idumaea. The site is also mentioned in the Notitia Dignitatum, compiled in ca. AD 400, which relates that ‘Birsama’ (ǺȚȡıĮȝȓȢ) was a provincial military base that had a castellum (Fig. 1) garrisoned by the Equites Thamudeni Illyriciani (Or. 34.10, 22).

The soils in the area (Dan & Yaalon 1980: 39-54; Dan 1981, 1981a) fall within the Northern Negev Loess Physiographic Area, and consist of Quaternary and Neocene sediments, mainly aeolian clays and silts covered by the Pleistocene sand of the Gaza Formation. A cross-section of the stratigraphy can be described as follows (Dan 1981: 21): limestone bedrock, over which lies a light brown clay loam with lime concretions; this is covered by a layer of light yellowish-brown loam with lime concretions, above which is a light yellowish-brown loam layer comprised of sandy regosols and quartzic inseptic light brown sand; the topsoil is covered with grasses and herbs, low shrubs and acacia trees.

Another Late Roman source mentioning Be’er Shema is the Codex Theodosianus, a compilation of laws drafted in 438 CE under the emperor Theodosius II. From this we learn the Roman name for the site: castellum Versaminum. In an entry from that document, dated to 23 March AD 409, it appears that the ducal office at Birsama was collecting taxes in-kind from landowners and the border militia contrary to the established law, an act for which that office was threatened with a fine of 100 lbs of gold (VII.4.30). This scenario demonstrates a continuity of the empire-wide administrative reforms implemented by Diocletian, whereby the powers of the provincial governors were separated from that of the military

* I would like to sincerely thank my colleague and friend Dr. Tali Erickson Gini (Israel Antiquities Authority) for her invitation to codirect the 2006 excavations at Be’er Shema, for reviewing a draft of this manuscript and providing useful substantive comments on it. Any and all mistakes, omissions or interpretive flaws are the responsibility of the author.

111

THE LATE ROMAN ARMY IN THE EAST document enumerating major military centres in the region during the reign of emperor Anastasius (AD 491518). Column III of the document lists nine towns in the vicinity of Nessana and the amount of tax assessed to each. Rubin (1990: 30) has suggested that although the top section of Columns I and II is missing, Birsama was probably listed on that part of the document. He bases this assumption on the fact that the site is the only one not listed on those columns that appears in the third column. It is interesting to note that on this papyrus, the tax quota for Birsama (ǺȚȡıĮȝȓȢ) is second highest in amount only to Chermoula and more than three times that assessed for Malaata, suggesting that Birsama had a wealthy tax base. Previous Research Although the Negev was visited by several travellers during the 19th and early-20th centuries, Be’er Shema seems to have escaped their notice. The first modern researcher to discuss the site was the German scholar P. Thomsen (1907: 42, 112). This was done while he was surveying the region and collecting data for his omnibus report on the Roman milestones from the Roman provinces of Palestine, Arabia and Syria (Thomsen 1917). Like the Stevenson cum Germanus version of Ptolemy’s Geography, however, he too mistakenly associated the name of the site with Beersheba. Around the time Thomsen was working in the region, the Czech explorer A. Musil conducted extensive surveys in the area. He was the first to examine the remains at Be’er Shema, known then as Khirbet el-FƗr (Musil 1908: 63-64). He mentions the presence of a church, cisterns, a huge reservoir, and several architectural elements including a capital made out of white marble. In addition, he noted that the site was quite large and the most prominent on the entire plain, covering an area of ca. 400 m (E-W) by 150 m (NS).

Fig. 1 The castellum at Birsama, as listed in Notitia Dignitatum Oriens 34.10 command through the creation of a new office known as the dux, who was responsible for the daily affairs of the region under his direct control. In this case, the dux of the region in which Birsama was situated had clearly overstepped his bounds. The writings of Sozomen, a historian of the Christian church, provide insight about the religious community at Birsama during the late-4th and early-5th centuries AD. In his Ecclesiastical History, written between AD 440 and 443, the site is referred to as ‘Geraris’ – a name assigned to the ducal region in which it was situated. He informs us that a certain Silvanus founded a monastery at Geraris ‘near the great torrent’ (VI.32.8), most likely a reference to the Nahal Besor, and that a man named Zechariah was its head monk (IX.17.4). These passages can be supplemented by the proceedings from the Council of Chalcedon, an ecumenical meeting held in AD 451 that addressed and canonized both the divine and human nature of Jesus. Contained within the list of bishops in attendance is a certain Markianos of Gerar, or ȂĮȡțȚĮȞȩȢ īİȡȐȡȦȞ (Lequien 1740: 663). From these two sources it is clear that Birsama was a thriving diocese with its own bishop and a large monastery headed by a high-ranking monk.

The most comprehensive early research on Be’er Shema was provided by A. Alt (1929 1930, 1931, 1931a) while he was conducting research for his publication on the ‘Römische Kastelle und Straßen’ (Alt 1935) in the Wadi Arava and Negev. He was the first to equate ‘Chirbet elFƗr’ with ancient Birsama, which he referred to as ‘Barsama’ (Fig. 2). In addition, he provided a detailed analysis and synthesis of references to the site ancient authors. He also noted that Barsama formed part of an EW fortified line that ran from the Mediterranean to the Dead Sea, an idea that has been repeated more recently by scholars (Tsafrir 1986: 83-84). Based upon the work of Alt, Avi-Yonah (2002: 148) was able to propose a tentative delimitation for the Saltus Gerariticus.

Further elucidation on Be’er Shema is found in the Descriptio Orbis Romani, a list of all towns in the Byzantine Empire compiled by Georgios Kyprios in ca. AD 600. He makes it clear that during this period, Birsama served as a regional administrative center for the territory of Gerar, or the ȈȐȜIJȦȞ īİȡĮȧIJȚțȩȢ òIJȠȚ ǺĮȡıȐȝȦȞ (Gelzer 1890: 52). Additional information comes from a papyrus document dated to the mid-6th century AD that was found at Nessana during the Colt Expedition’s work at the site in the 1930s. This papyrus, P.Nessana 39:11 (Kraemer 1958: 119-125), is a tax

Some years after Alt, Abel (1938: 179-180, 443) visited Birsama and noted the presence of many pear-shaped (i.e. Byzantine ‘bell-shaped’) cisterns, a large reservoir and several wells. In wake of the research conducted by Alt and Abel, a number of other early surveys and excavations in the region surrounding Be’er Shema took place. However, most of that fieldwork concentrated 112

BENJAMIN J. DOLINKA: BE’ER SHEMA – BIRSAMA OF THE NOTITIA DIGNITATUM several other extant architectural remains within the environs of Be’er Shema. The ceramic and numismatic evidence from both Sites 160 and 161 was Roman and Byzantine, and included a gold coin of Emperor Justin II dated to AD 576/7 minted in Constantinople. In light of the artifactual material he collected, Gazit (1996: 17-18) suggested that many of the sites in his survey area, including Horvat Be’er Shema, were abandoned after the Muslim conquest of the region. This supposition has been convincingly refuted on the basis of a re-examination of the pottery by Magness (2003: 172-174), who has demonstrated the presence of several vessel forms dating to the Early Islamic period illustrated in the report by Gazit. The 1989/1990 Excavations The only archaeological excavations conducted at Be’er Shema were undertaken by the IAA in 1989 and 1990, under the direction of D. Gazit and Y. Lender. Work focused upon the remains of a Byzantine basilica church, which the excavators assigned a date from the late-6th to the mid-7th century AD (Gazit & Lender 1992, 1993). This 21 x 12.4 m building (Fig. 3) was comprised of all of the architectural elements in a typical Byzantine church: a nave, two aisles, a narthex, an apse, a stage (bema), a pulpit (ambo), an altar faced with marble revetment, a baptistery, and a hexagonal reliquary where relics and other sacred items were likely stored. The walls were constructed of well-built ashlars that were covered with white-washed mud plaster on the upper courses, and the roof was likely made up of wooden beams, as indicated by the presence of several iron nails found in situ on the building’s floors.

Fig. 2 Map of Roman castella and roads (Alt 1935: 24), showing the location of ‘Barsama’ in the upper left corner

The most striking feature of the church at Be’er Shema is the mosaic floors that cover the nave, apse, baptistery and stage (bema). These consist of tesserae in many colours depicting zoomorphic and anthropomorphic imagery, as well as geometric patterns and floral motifs (Gazit & Lender 1993: 274-276). Of particular interest is the fact that some of the images of animals are not simply confined to the space within their medallions, but seem to be leaping out from them (Gazit & Lender 1992: 45), an uncommon convention on contemporaneous examples. The mosaics from the middle of the nave at Be’er Shema are characterized by a series of medallions attached to vines that emanate from an amphora, often referred to as the ‘inhabited scroll’ style. This iconography is typical of the so-called ‘Gaza mosaic school’ whose craftsmen were thought to have ornamented both synagogues and churches in the region surrounding Gaza during the 4th6th centuries AD (Avi-Yonah 1981). Among those is the so-called ‘Shellal Church’ (Gazit 1996: 30, Site 8) located not far from Be’er Shema, which was uncovered by soldiers of the Anzac Division in 1917 (Alt 1929: 99), and whose mosaics were removed to Australia shortly thereafter (Trendall 1957). A more recent study on the socalled ‘Gaza mosaic school’ (Hachlili 1987) has determined that the inhabited scroll motif was not the

upon the prehistoric era, particularly the Chalcolithic period (Gazit 1996: 9-10). Since that time Birsama has been mentioned in gazetteers of Roman and Byzantine Palæstina (e.g. Tsafrir et al. 1994: 91), as well as discussions about ancient Christian sites in the region (e.g. Bagatti 1983: 182-183), but these were all simply reiterating the work of Alt. Between 1978 and 1985, Be’er Shema and its environs were surveyed by D. Gazit, on behalf of the Israel Antiquities Authority (hereafter IAA) and as part of the Israel Archaeological Survey for the Map of Urim. Birsama corresponds to his Sites 160 and 161, each of which cover an area of 500 dunams (Gazit 1996: 59). At these sites he noted the presence of architectural elements, pottery, glass and tesserae, and what appeared to be the outline of a theatre. In the northeast sector of Site 160, he recorded the remains of a raised square surface measuring 70 x 70 m that stood 2.5 m higher than its surrounding area and suggested that this likely represented the site of the Late Roman castellum of Birsama mentioned in the Notitia Dignitatum. In addition, a well dating to the Byzantine period that was dug into the aquifer of Nahal Besor was found at Site 160 (Gazit 1994: 24-25). Furthermore, Gazit made note of 113

THE LATE ROMAN ARMY IN THE EAST

Fig. 3 The Byzantine church from Be’er Shema (Gazit & Lender 1992: 44).

114

BENJAMIN J. DOLINKA: BE’ER SHEMA – BIRSAMA OF THE NOTITIA DIGNITATUM product of any specific workshop, but rather a popular style of mosaic for which there was a basic ‘template’ that could be modified by individual mosaicists according to taste. This template was utilized by craftsmen throughout a widespread area and not geographically limited to the environs surrounding Gaza.

The 2006 Season at Horvat Be’er Shema Much like the 1989/1990 excavations at Be’er Shema, the impetus for the 2006 campaign stemmed from unauthorized agricultural work at the site, in which architectural elements from buildings within the Byzantine settlement were unearthed. A preliminary reconnaissance of the site was carried out in April 2006 and the remains of two Byzantine capitals and building debris in one area seemed to suggest the presence of yet another church. In addition, a magnetometry survey was conducted in the part of the site that Gazit suggested was a theatre. Results from that fieldwork revealed the remains of a massive structure within the confines of the survey area. The locale of the Late Roman fort was also investigated and it appears to be of the quadriburgium type, with three of its four corner towers clearly visible.

Perhaps the most significant evidence provided by the excavations at Be’er Shema is the ten Greek inscriptions from the mosaics found within the church (Tzaferis 1996). This collection of commemorative, dedicatory and laudatory inscriptions provides valuable insights into the nature of the Be’er Shema church and to those responsible for its construction and upkeep. First, Inscription 2 informs us that the church was built by the military bodyguards (comites) of Helladios, the governor of the imperial estate of Gerar (‘EȜȜĮįȓȠȣ īİȡȡȐȡȦȞ), and that the name of this place of worship was the Church of the Trinity (ȉȡȚȐįȠȢ įȩȝȠȢ). Inscriptions 9 and 10 make it clear that the patron saint to which the church was dedicated was St. Stephen (†AȖȚȠȢ ȈIJİijȐȞȠȢ), who is given the epithets ‘archdeacon’ (ÐȡȤȚįȚȐțȠȞ) and ‘first martyr’ (ʌȡȠIJȠȝȐȡIJȠȢ). Tzaferis (1996: 82) has even suggested that the reliquaries at the Be’er Shema church contained relics of St. Stephen. In addition, it is stated in Inscription 6 that the Bishop of Gerar at the time of the church’s construction was Makaidonios (ȂĮțĮȚįȩȞȚȠȢ) who is referred to as the ‘chief shepherd’ (ÐȡȤȚʌȠÊȝgȞ). From Inscriptions 2, 4 and 8, we learn that Stephanos was the priest (ʌȡİıȕÌIJeȡȠȢ / Úİȡ±ȠȢ) and guardian (ʌĮȡĮȝȠȞÇqiȠȢ) of the church, and the ‘highest ecclesiastical authority at Birsame’ (Tzaferis 1996: 80). Finally, from Inscription 4, we learn that the financial manager (țİȜĮȡȓIJȘȢ) of the church was named Victor. It is unfortunate that absolutely none of the ten inscriptions found on the mosaics at the Be’er Shema church mention dates. However, based upon palaeography the inscriptions have been assigned to the second half of the 5th century AD (Tzaferis 1996: 84).

Due to budget constraints and the salvage nature of the project, the 2006 excavations were limited to only twenty 5 x 5 m squares, an exceedingly minute portion of the site's overall area, which covers some 50 hectares (123.55 acres). The team was not allowed to conduct fieldwork in the area of the Late Roman fort. Excavations of the 2006 season were directed by Drs. Tali Erickson-Gini (IAA) and Benjamin J. Dolinka (Albright Institute), under a license granted by the IAA (No. A-4812). The Field Supervisor was L. Shilov (IAA), and both funding and workmen were provided by the Farmer’s Union of the Negev. The results of the 2006 field season have done much to clarify the occupational history of the site and have provided important insights into its economy. Excavation has revealed that Be’er Shema was continuously occupied well into the Early Islamic period with subsequent reoccupation in the Mameluk era. This continuity is contrary to what Gazit suggested and is supported by similar evidence cited by Magness regarding other sites in the vicinity.

Taken together, previous research on ancient Birsama (Be’er Shema) has shed a great deal of light on the identification of the site and provided a fairly detailed historical context within which to place the excavations of the 2006 campaign. The excavators of the church place its date in the late-6th to mid-7th centuries AD, dates for which they cite absolutely no tenable archaeological evidence – numismatic or ceramic. The style of the mosaics and the inscriptions on them, together with the ancient literary and ecclesiastical sources, seem to suggest an earlier date for the Birsama church, i.e. sometime in the mid-5th to mid-6th centuries AD. In addition, the mosaics from the church suffered no apparent damage or iconoclasm during the Muslim conquest (Schick 1995: 184, 309), suggesting that it had already been abandoned by that time.

Of perhaps greater significance was the discovery of a massive wine press installation in Area C2 (Figs. 4-5), that is dated by numismatic evidence to the latter part of the reign of Anastasius I, i.e. after AD 512. This wine press was laid out on three levels. The top area consisted of a treading floor (forum vinarum) with a centrallylocated square installation where a wooden corkscrew press was likely placed, and three plaster-lined compartments for storing grapes (tabulata). Below this was a central upper settling tank which drained into two large lower reservoirs (lacus musti) measuring more than 1.5 meters in depth. In Area A, a storehouse filled with wine jars was uncovered. Many of these conformed to Peacock and Williams Class 46, the so-called ‘Palestinian Bag-Shaped Jars’ which date to the 5th and 6th centuries AD (Peacock & Williams 1986: 191-192). In addition, there were also large quantities of the Class 49 ‘Gaza Jars’ as well (Peacock & Williams 1986: 198-199). Be’er Shema therefore represents the southernmost centre yet 115

THE LATE ROMAN ARMY IN THE EAST

Fig. 4 The central upper settling tank and two lower reservoirs from the Byzantine wine press in Area C2 at Be’er Shema uncovered in 2006 (courtesy of Israel Antiquities Authority).

Fig. 5 Detailed view of the southern reservoir (lacus musti) from the Byzantine wine press in Area C2 at Be’er Shema (courtesy of Israel Antiquities Authority). 116

BENJAMIN J. DOLINKA: BE’ER SHEMA – BIRSAMA OF THE NOTITIA DIGNITATUM uncovered for the production of Gaza wine jars. Due to the presence of kiln waste and tester pieces found within the ceramic assemblage from the excavations at Be’er Shema, and also from the survey of Gazit, it is suggested here that these wine jars were produced on-site to contain wine produced in the large press from Area C2. Taken together, this would seem to confirm that the wealthy tax base for Birsama alluded to in the Nessana papyri was provided by the industrial production of wine.

Dan J. 1981 Soil Formations in the Arid Regions of Israel. In J. Dan et al. (ed.), Aridic Soils of Israel. Bet Dagan: 3-50, The Volcanic Center, Special Publication No. 190. Dan J. 1981a Soils of the Sandy Region of the Western Negev. In J. Dan et al.. (ed.), Aridic Soils of Israel. Bet Dagan: 223-238, The Volcanic Center, Special Publication No. 190. Dan J. and Yaalon D.H. 1980 Origin and Distribution of Soils and Landscapes in the Northern Negev (Hebrew). In Studies in the Geography of Israel 11: 31-56. Danin A. 1979 The Vegetation of the Negev. In S. Shmueli and Y. Grados (ed.), The Land of the Negev: Man and Desert, Tel Aviv: 164-176. Danin A. et al. 1975 The Vegetation of the Northern Negev and the Judean Desert of Israel (Hebrew). In Israel Journal of Botany 24: 118-172. Evenari M. et al. 1982 The Negev: The Challenge of a Desert. Cambridge, MA. Gazit D. 1986 Ha-Besor Region (Hebrew), Tel Aviv. Gazit D. 1994 Ha-Besor Region in the Byzantine Period – Man and Environment (Hebrew). In Ariel 100-101: 172-178. Gazit D. 1996 Archaeological Survey of Israel: Map of Urim (125), Jerusalem. Gazit D. and Lender Y. 1992 Horvat Be’er Shema. In ESI 10: 43-45. Gazit D. and Lender Y. 1993 The Church of St. Stephen at Horvat Be’er Shema. In Y. Tsafrir (ed.) Ancient Churches Revealed, Jerusalem: 273-276. Hachlili R. 1987 On the Mosaicists of the ‘School of Gaza’. In Eretz Israel 19: 74 (English summary). Levie E.L. and S. Pohoryles 1970 Agricultural Land Utilization. Map XII/I. In D.H.K. Amiran et al. (ed.), Atlas of Israel, Tel Aviv. Magness J. 2003 The Archaeology of the Early Islamic Settlement in Palestine, Winona Lake, IN. Musil A. 1908 Arabia Petraea II: Edom – Topographischer Reisebericht, Vienna. Rosenan N. 1970 Rainfall. Map IV/2. In D.H.K. Amiran et al. (ed.), Atlas of Israel, Tel Aviv. Rubin R. 1990 The Negev as a Settled Land: Urbanization and Settlement in the Desert in the Byzantine Period (Hebrew), Jerusalem. Schick R. 1995 The Christian Communities of Palestine from Byzantine to Islamic Rule: A Historical and Archaeological Study, (Studies in Late Antiquity and Early Islam 2), Princeton. Thomsen P. 1907 Loca Sancta, Leipzig. Thomsen P. 1917 Der römische Meilensteine der Provinzen Syria, Arabia und Palaestina. In ZDPV 40: 1-103. Trendall A.D. 1957 The Shellal Mosaic and Other Classical Antiquities in the Australian War Memorial, Canberra. Tsafrir Y. 1982 The Provinces of Palestine: Names, Borders and Administration (Hebrew). In Z. Baras et al. (ed.), Eretz Israel from the Destruction of the

Bibliography Ancient Sources Claudius Ptolemy 1618 Theatri Geographiae Veteris Tomus prior in quo Cl. Ptol. Alexandrini Geographiae Libri VIII. P. Bertio ed. Amsterdam 1618. 1991 The Geography. Trans. E.L. Stevenson. London 1991. Originally published New York 1932. Codex Theodosianus The Theodosian Code and Novels and the Sirmondian Constitutions. Trans. C. Pharr, Princeton 1952. Council of Chalcedon = Lequien, Michel. Oriens Christianus. 3 vols. Paris 1740. Georgius Cyprius (Georgios Kyprios). Georgii Cyprii: Descriptio Orbis Romani. Trans. H. Gelzer, Leipzig 1890. Notitia Dignitatum Notitia Dignitatum. Trans. O. Seeck, Berlin 1876. P.Nessana = Nessana Papyri Excavations at Nessana III: Non-Literary Papyri. C.J. Kraemer (ed.), Princeton 1958. Sozomen The Ecclesiastical History of Sozomen, from AD 324 to AD 440. Trans. E. Walford, London 1855. Secondary Literature Abel F.M. 1938 Géographie de la Palestine II, Paris. Alt A. 1929 Barsama. In ZDPV 52: 99-115. Alt A. 1930 Limes Palaestinae. In PJb 26: 43-82. Alt A. 1931 Limes Palaestinae. In PJb 27: 75-84. Alt A. 1931a Beiträge zur Historischen Geographie und Topographie des Negev. In JPOS 11: 204-221. Alt A. 1935 Aus der Araba II: Römische Kastelle und Straßen. In ZDPV 58: 1-78. Avi-Yonah M. 1976 Gazetteer of Roman Palestine. Qedem 5. Jerusalem: Institute of Archaeology, Hebrew University. Avi-Yonah M. 1981 Une École de Mosaïque à Gaza au Sixiême Siècle. In M. Avi-Yonah (ed.), Art in Ancient Palestine: Selected Papers, Jerusalem: 389-395. Avi-Yonah M. 2002 The Holy Land: A Historical Geography from the Persian to the Arab Conquest, 536 BC to AD 640. 4th ed. Revised by A.F. Rainey and Y. Tsafrir, Jerusalem. Bagatti B. 1983 Antichi villaggi cristiani di Giudea e Neghev. SBF Collectio Minor No. 19, Jerusalem.

117

THE LATE ROMAN ARMY IN THE EAST Second Temple to the Muslim Conquest, Volume I. Jerusalem: 350-386. Tsafrir Y. 1986 The Transfer of the Negev, Sinai and Southern Transjordan from Arabia to Palaestina. In IEJ 36: 77-86. Tsafrir Y et. al. 1994 Tabula Imperii Romani: Iudaea, Palaestina (Maps and Gazetteer). Jerusalem. Tzaferis V. 1996 Greek Inscriptions from the Ancient Church at Horvat Be’er Shema. In Eretz Israel 25: 7585. Waisel Y. et al. 1978The Ecology of Vegetation in Eretz Israel, Tel Aviv.

118

Crossing the Negev in Late Roman Times: the Administrative development of Palaestina Tertia Salutaris and of its Imperial Road Network Israel Roll University of Tel Aviv

Following the advent of Caius Aurelius Valerius Diocletianus to supreme power in late November 284, central authority was soon restored all over the Roman Empire, for two decades and more. This impressive recovery, which came after five decades of chronic instability that brought the Roman Empire to the brink of disintegration, was due to a series of fundamental changes in the long-lived but by now clearly overdue Imperial system. These changes were the outcome of a sequence of reforms, initiated by Diocletian and his fellow-rulers with the clear intention to confront the adversities and perils that threatened the stability of the Empire.1

their steady movement and supply in all season and under every weather condition.2 These measures were implemented with special care along the eastern frontier zone of the Roman Empire, in response to two principal threats, which repeatedly menaced and even endangered the Roman dominion in the Orient. From the Middle Euphrates to the north, there was the Persian threat. The highly motivated Sassanian Empire, with its large, well-organized and essentially mobile military force, exercised again and again an aggressive strategy towards Rome. This was the case in the mid-third century, during the reign of Shapur I, and again, at the end of the century, in the times of Narseh. Although, in the year 298 CE, Diocletian’s colleague Galerius won a significant victory over Narseh, which enabled the establishment of a firm Roman hold on northern Mesopotamia, a renewed Sassanian surprise attack, in the shape of a long-range incursion of large mobile cavalry units, could not be ruled out.3

The main reforms included: x x

x

x

x

The adoption of a collegiate system of government of four rulers, conventionally denominated by modern scholars as tetrarchy. The reformation of the taxation pattern (vectigalia) and of the economic system as a whole, which became from now on much more state-integrated on one hand, and very much centralized on the other. The reordering of the provincial framework (ordinatio provinciarum), which included the initiation of a policy of separating military from civil power in frontier provinces, as well as the systematic division of these provinces into smaller units. The reorganization of the Empire’s military power (exercitus), a main aspect of which was the implementation of a large-scale fortification program all along the frontiers, that included the building of city-walls (muri), camps (castra) and forts (castella), and the distribution of a variety of new or reorganized military units along them. The establishment of an elaborate communication system of built and engineered roads (strata) between the strongholds, to secure an efficient command and control of the troops, and to ensure

To confront such a threat, the Tetrarchic defense system in Mesopotamia and Eastern Anatolia was based on fully equipped legions that camped in fortified cities located along the frontier, and on auxiliary, mainly mobile units deployed in camps and forts erected near by the chief watering places and along the main roads that stretched towards the rear. What we have here is a basic idea of a frontier defense in depth, the origin of which goes back to the mid-first century CE. Tacitus (Annals XV.3), when describing the reorganization of the Roman army in the East by Corbulo towards the renewal of the war with the Parthians in 62 CE, is very explicit on the matter. He specifies that Corbulo’s preparations included “the stationing of his remaining legions on the Euphrates bank, the arming of an improvised force of provincials, and the closing of hostile avenues of approach by 2

For the collegiate system of the Tetrarchy and its ideology, see: Chastagnol 1994 and Kolb 1995. For the state-integrated economic system of Diocletian, see: Carrié 1994 and Corcoran 2000: 205-233. For his provincial policy, see the most recent survey of Rees 2004: 24-27. For Diocletian’s fortification program along the frontiers, see the survey of Southern and Dixon 1996: 127-147 and the critical discussion of Reddé 1995. For the road system, see bellow. 3 The main written sources related to the encounters between the Romans and the Sassanians during the Tetrarchy are reproduced, in English translation, in Dodgeon and Lieu 1991: 121-139. For discussion, see: Millar 1993: 174-189 and Isaac 1998: 437-469. For Galerius’ campaign in 298, see: Zuckerman 1994 and Leadbetter 2002. For the territorial gains and the peace-treaty that followed, see Winter 1989 and Blockley 1992: 5-7. For the potential capabilities of the Sassanian elite cavalry, see the most recent survey of Farrokh 2005.

1 The present paper continues, up-dates and, when necessary alters the discussion and interpretation on topics which I elaborated and documented in detail in my article published almost two decades ago (Roll 1989). Therefore, I will lay stress in the present paper on more recent bibliography on these topics, published afterwards. On the empire of the Tetrarchy in general and on the variety of aspects related to Diocletian’s reforms in general, many studies have been published in recent years. The more important among them are: the voluminous overall work of Kuhoff (2001, a new edition of which is in preparation), and the collected studies edited by Demandt et alii 2004, and by Boschung and Eck 2006. Additional up-to-date publications, on specific topics, will be mentioned in the appropriate notes to follow.

119

THE LATE ROMAN ARMY IN THE EAST garrison-posts. Further, as the region is deficient in water, forts were erected to command the springs ...”. (… reliquas legiones pro ripa Euphratis locat, tumultuarium provincialium manum armat, hostilis ingressus praesidiis intercipit. Et quia egena aquarum regio est, castella fontibus imposita …). Apparently, that frontier policy has been reinvigorated and even expanded by Diocletian along the eastern limits of Asia Minor and Northern Syria, as well as further to the south, as we shall see below.4

realized that much more had to be done to provide proper protection to this vast region. Consequently, an impressive fortified frontier system was erected along its eastern limits during the Tetrarchy. Malalas (XII.308) provides an explicit definition of the Tetrarchic frontier system, as well as a short account of its nature: “Diocletian also built fortresses on the limes for Egypt to the Persian borders and stationed limitanei in them, and he appointed duces for each province to be stationed further back from the fortresses, with a large force to ensure their security”. Let alone the term limitanei, the meaning of which in the early fourth century could have been different than later on, Malalas’ testimony is well supported by the archaeological and epigraphic data surveyed or uncovered along the Eastern frontier zone. This data includes a variety of military structures – many of them with Tetrarchic dedicatory inscriptions, as well as long segments of engineered and paved roads – many of them marked with Tetrarchic milestones. The overall picture of all that is of a military frontier system based on the idea of a dynamic defense in depth, across a broad frontier zone, the origins of which go back, as we have seen, to the first century CE.7 This frontier system consisted of several legions, each one of them deployed in a camp located in a central position of a distinct segment of the frontier zone. In front of such a camp and all along the segment, a series of auxiliary, mainly mobile units were distributed, in a running chain of forts and fortresses. The backbone of the system consisted of a main strategic road that extended along a substantial part of the frontier zone, serving as a chief connecting artery between its military installations. In Syria, this artery was

Further to the South, in the vast wilderness that extended between the Euphrates and the Red Sea, the menace was different in nature. The population of this area consisted of a variety of desert tribes, which followed a nomadic way of life. Ammianus (XXII.15.2) informs us that the people under discussion were “Arabs, whom we now call Saracens” (… Arabas, quos Sarracenos nunc appellamus). Ammianus (XIV.4.3) elaborates on them and adds that their life is always on the move and that “they rove continually over wide and extensive tracts, without a home, without fixed abodes, nor laws” (sed errant semper per spatia longe lateque distenta, sine lare, sine sedibus fixis, aut legibus). The Saracens were unable (at least at this stage) to organize in large-scale armed forces, nor to attack cities or to endanger the Roman rule over an entire province. What they could, and constantly did, was to carry out plundering incursions, in small horse or camel-mounted bands, with the basic aim of taking booty. Our basic source on the matter is once again Ammianus (XIV.4.1) “Whenever (the Saracens) have caught sight of any prey from a high place, they seize it with swift swoop, and soon after they have seized it, they fade away” (qui si praedam dispexerint celsius, volatu rapiunt celeri, ac si impetraverint, non immorantur). The incursions of the Saracens were usually considered by the Romans as feats of banditry rather than acts of war. However, their menace was a constant one, which became more intense during the late third century CE and the century to follow.5 Towards the year 290 CE the Saracen menace seem to have reached a degree that forced Diocletian to carry out a punitive campaign against them, a deed which seem to have ended successfully.6 However, the emperor

(captivitatis vinculis Sarracenum) is glorified; see: Nixon and Rodgers 1994: 89. The feat is also alluded to in two other sections of the same text (ibid: 86 and 99), therefore, one can hardly follow the argument of Leadbetter 2004, who tends to dismiss a Diocletianic campaign agaist the Saracens. For the importance of the Panegyrici Latini as a unique source on the Empire of Diocletian and later fourth century rulers, see the recent review of Hostein 2004. 7 Ever since the publication of Edward Luttwak’s Grand Strategy of the Roman Empire (1976, which is an altered version of his PhD Thesis submitted to the Johns Hopkins University in 1975 under the title Force and Diplomacy in Roman Strategies of Imperial Security) the shape and nature of the Roman frontiers has emerged as an issue of “grand debate”, well summarized most recently by Whittaker 2004: 1-27 and 28-49. On one hand, Luttwak’s thesis has been challenged all along, on the conceptual level as well as in its concrete details, by Ben Isaac in his major study The Limits of Empire (1990; rev. ed. 1992). Isaac’s argumentation has been soon followed by a growing number of scholars, and the essays of some of the more prominent among them have been edited by D.L. Kennedy in The Roman Army in the East (1996). Isaac’s views have also been embraced by the scholars who tend to minimize the size of foreign threats to the Empire’s Eastern frontier, mentioned above in note 5, as well as by J.-M. Carrié and S. Janniard in their recent reviews published steadily in Antiquité Tardive from 2000 onwards. On the other hand, Luttwak’s thesis has been followed by Williams 1985: 91-101, and elaborated even much further by Nicasie 1998: 117-184. For a more balanced view see: Millar 1993: 174-189, and notably the two papers of Ariel Lewin published in 1990 and 2002. It should be added that I myself tend to refrain from using the term “grand strategy” because it reflects a concept literally unknown in Antiquity. However, I will try to demonstrate in the present paper that written sources and archaeological data indeed reflect the implementation of an Eastern frontier policy of the kind I do elaborate in my main text, from the Tetrarchic era onwards.

4 For a comprehensive survey of the main Roman military sites along the eastern frontier zone of Asia Minor, see: Gregory II 1997: 9-132; for discussion, see: Gregory I 1995: 212-222. For a detailed analysis of Tacitus’ account on the eastern campaign of Corbulo, see: Gilmartin 1973; for a more recent reassessment, see: Vervaet 2002. 5 For a detailed discussion on Ammianus’ testimony on the Saraceni, see: Matthews 1989: 342-353. For a general survey of the Late Roman sources related to the Saraceni, see the more recent publication of Retsö 2003: 454-525. The menace posed by the Saraceni to Rome’s eastern frontier is strongly emphasized by Parker 1991; 2002; 2006a: 548-551. On the other hand, several scholars tend to minimize, and even nullify the very existence of such a menace, for which see: Isaac 1992: 235249; Fiema 1995; Eadie 1996; Graf 1997; 2002. For a more balanced view, see: Lewin 1990; Millar 1993: 184-189; Butcher 2003: 407-409; Sartre 2005: 358-363. 6 Diocletian’s campaign is mentioned in the Panegyrici Latini XI.5.4, where “the subduing of the Saraceni by the bonds of captivity”

120

ISRAEL ROLL: CROSSING THE NEGEV IN LATE ROMAN TIMES the Strata Diocletiana, which stretched from Sura on the Euphrates to Palmyra and Damascus, and apparently, even further on to the south. Damascus was considered by the Romans as a focal point for their Eastern frontier system and that goes in line with the testimony of Malalas (XII.307-308), about the building of arms factories there by Diocletian, “bearing in mind the incursion of the Saracens”.8 Further to the South, it was the Via Nova that played the role of the frontier’s trunck road. This first-class built and engineered highway extended from Bostra in Southern Syria to Aila on the Red Sea, and connected the variety of military installations distributed all along the frontier zone of the provincia Arabia. Although, the artery’s initial construction goes back to the reign of Trajan (see below), some two dozens of Tetrarchic milestones discovered so far along it indicate that it was extensively repaired during the Tetrarchy.9 It should be clear that the trunck road under discussion was not conceived by the Romans as a linear defensive element of a static defense system. It was regarded, and it actually served, as the chief carrier of military units, of their command and control personnel and of their supply, by which the entire frontier zone of the province could be crossed.10

Many more forts and fortresses were built along the roads of the rear communication network that extended from the trunck-road towards the west, the ultimate destination of which were the main harbors of the Mediterranean Sea. These installations were usually built at characteristic key points of the region, that is, the water sources, the crossroads, the passes and the ascents. The task of their units was to safeguard the traffic along the rear roads, and to secure the Imperial rule over the hinterland crossed by them.12 During the reign of Diocletian, two significant shifts, of administrative and military nature, occurred in the southeastern end of the Roman Eastern frontier zone. The first one consisted of the detachment of the whole southern part of the provincia Arabia in favour of the provincia Palaestina. The zone under discussion included the eastern end of the Sinai plateau, the hill-and mountain-region of the Negev, the Rift valley of the Arava, and the ridge-land of Edom. Consequently, the area of the provincia Palaestina was enlarged substantially. At first glance, that move seems to be in contradiction with the official Tetrarchic policy, to diminish the territorial domain of the provinces. However, the explanation for this peculiar transition seems to lie in the second shift promoted by Diocletian, namely, the transfer of the legio X Fretensis, and possibly, of the legio VI Ferrata as well, from the mainland Palaestina to the southeastern frontier zone. The point is that the governor of the traditional and larger provincia Arabia already had under his command two legions: the legio III Cyrenaica – stationed at Bostra for almost two centuries by now, and the newly-created legio IV Martia, which was most probably assigned to the legionary installation of el-Lejjun in the land Moab. The transfer of the legio X Fretensis from Jerusalem to Aila and the removal of the legio VI Ferrata from Megiddo to Udruh would mean – under the former territorial division – to allocate four legions under the command of one governor, that is, of the governor of Arabia, and to devoid completely, of any legion, the governor of the provincia Palaestina. One may assume then, that the attachment of the southernmost frontier zone to the provincia Palaestina came precisely to avert such an unbalanced situation. Moreover, with this territorial shift, the governor of the provincia Palaestina was charged with direct responsibility over the border zone which protected the land of Edom, from Wadi Hasa to Aila. Consequently, this segment of the fortification system could actually, as well as officially, be considered as a limes Palaestinae.13

In line with the concept of a defense in depth, advanced forts were also erected further to the east, on the main tracks leading beyond the frontier zone, towards the interior of the Arabian desert, which enabled their units to carry out deep reconnaissance incursions of Roman mobile patrols into the nomad lands. Ammianus (XXIII.5.2) defined the land beyond the Imperial frontier interiores limites, and he explicitly attributes the building of the outposts there to Diocletian. One may add that in a Latin inscription found near a crossing point over the Middle Tigris, the outpost to which it probably belonged is denominated “the eyes of the legions” – occuli legionum.11 8 For the military installations of the frontier zone of Syria, see: Gregory II 1997: 133-258. For the specific area between Sura and Palmyra, see: Konrad 1999. For the Strata Diocletiana see the recent summary of Bauzou 2000; see also the paper of Konrad 1996, who forwards evidence on pre-Diocletianic use of the artery. For the role of Damascus and its defence in Roman times, see the recent study of Burns 2003. 9 For the military installations of the provincia Arabia, see the compact and very useful survey of Kennedy 2004. For the northern segment of the Via Nova, see the comprehensive study of Bauzou 1998, with a specific discussion on the Tetrarchic milestones at pp. 250-252. For the middle segment of the road, see: Mac Donald 1996; Parker 2006a: 544545. For the southern segment, from Petra to Aila, see: Graf 1995. 10 In his most recent archaeological survey on the Roman Danube, Wilkes (2005: 148-162) identifies a frontier zone there which, in many aspects is similar to the one described here, but he prefers to define it as a “continuous military cordon”. Breeze and Jilek (2005), in their summary of the papers presented during the 19th Limes Congress in the section dedicated to “how did frontiers actually work”, emphasize the care with which the Roman army went about dealing with its strategic, tactical and local problems, and the detailed control exercised by the military authorities over the entire frontier zone. As for the supply provided to the Roman army on the Arabian frontier Zone, see the interesting evidence discussed most recently by Parker 2005. 11 For the meaning of Ammianus’ interiores limites, see the magisterial paper of Malavolta 1982, and the note of Mayerson 1988. For the inscription that mentions occuli legionum, see: Kennedy 1988.

12 For the nature, various installations and ultimate destination of the rear communication network of the Roman southeastern frontier zone, see: Roll 1999; 2002; 2005a; 2005b. 13 For a more recent discussion on the administrative changes in the southern Levant, see: Barnes 1996: 548-550. For the legionnary camps in that region and their garrisons in Late Antiquity, see the most recent summaries of Kennedy 2004: 154-159 (el-Lejjun), 178-180 (Udruh), 206-209 (Aqaba). For the legio IV Martia, see: Parker 2006a: 555-557. For the transfer of the legioX Fretensis to Aila , see: Dabrowa 2000: 321-322. For the possible transfer of the legio VI Ferrata to Udruh, see:

121

THE LATE ROMAN ARMY IN THE EAST elaborate on the matter here.16 Later on, when the Negev became part of the Nabataean Kingdom, the traffic along many of its potential routes increased substantially, when compared to previous ages (see Map 1). This traffic was first and foremost of commercial nature and the core of its activity was at the city of Petra, the capital of the Nabataeans, as well as their chief burial ground. The area of Petra forms a difficult terrain that consists of rugged cliffs built mainly of a foundation of granite, largely covered with red Nubian sandstone, which is crowned in several places with layers of limestone. The city’s main approach from the East descended along the Wadi Musa and the tortuous and narrow gorge of the Siq. Towards the west, the conditions of the terrain appear to become even harsher, a fact which certainly enhanced the ancient tradition about Petra, of being an almost inaccessible stronghold. Nonetheless, the area can be crossed by a series of passable ascents, which lead from the city towards the Aravah. Two of them, which are more suitable for camel traffic than the others because they slip down along descending limestone spurs, extend to northwest – along Naqb Namala, and to south-west – along Naqb Ruba’i.17 Following the descent along the northwestern artery, the copper mines of Phaenon and the town of Zoara could be reached, without difficulty. From the main road-station of Hazeva, Judaea and its capital Jerusalem could be attained, along the eastern Mount Hebron artery and passing by the towns of Mampsis and Ira. Jerusalem could also be reached along the roads that passed by the fortress of Uza, or by the oasis of Ein Gedi. Following the south-western descent, along Naqb Ruba’i, the Nabatean road installations of Abu Hashabe, Gharandal, Rujum Taba and Dafit ensured the proper needs of traffic towards the port of Aila.18

Following the territorial changes we came to mention, provincia Palaestina became, indeed, a rather large province. Consequently, and in accordance with the established policy of dividing provinces into smaller entities, a first division of its territory occurred, towards the late 350’s. The northern part of its area, down to the Valley of Berosaba, became an administrative unit of its own, under the existing name of provincia Palaestina, while the whole southern region was organized as a new province called Palaestina Salutaris. Towards the end of the fourth century, an additional division occurred, which involved mainly the northern province: its territory was divided into Palaestina Prima and Palaestina Secunda. In the south, the change was essentially nominal: its name changed into PalaestinaTertia Salutaris, as attested to in the recently published P. Petra 1 [Petra is named there as meter kolonion… metropolis tés trité Palaistines Saloutarias].14 Let’s turn our look now at the main roads that crossed Palaestina Tertia Salutaris, which encompassed the eastern strip of the Sinai tableland, the Negev, the Aravah and the land of Edom. This topic, however, requires four preliminary remarks of general nature. First, most of the roads that were engineered and built by the Romans in the area under discussion, followed alignments set out in much earlier times, some of them even at the beginning of the process of urbanization in the Early Bronze Age. Second, the priorities among the potential roads that crossed the area were determined by the central political power that ruled over the region or had strong reasons to get involved in its affairs because of military or economic interests. Third, the area under discussion makes a typical example of a land of transition of international importance, that is, a land bridge between the Mediterranean and the Red Sea, and a passageway for trans-Arabian trade routes, the destination of which were the harbors of the southeastern Mediterranean shore. Fourth, the destinations of the roads in the area itself were the urban centers that flourished there in various times, the military camps that garrisoned significant military units, and the sites of extensive extraction and smelting of copper.15

On the other hand, the journey from Petra to Egypt could have been undertaken along two arteries, which did not include road-installations, but were dotted with successive watering places. One of them extended to the south-west and passed by the sources of Gerez, Kuntilla and Themed, stretching on along the traditional transpeninsular route later known as Darb al-Hajj, until Clysma. Another thoroughfare seemed to have extended from Abu Hashabe straight to the west, by the watering places of Beer Menuha, Beer Ada, Beer Karkom, Bir Maein and Bir Lussan, and then, further on across the

An overview on the routes and transportation in the Bronze and Iron Ages across the region under discussion has been presented by me most recently at the Fifth International Congress on the Archaeology of the Ancient Near East held at Madrid in April 2006. As its content, with a full bibliographical apparatus will be published in the forthcoming Proceedings of the Congress, I will not

16 The paper’s title is “Routes and transportation across the Negev in the Bronze and Iron Ages”. See also the summary of Jasmin 2006. 17 For the centrality of Nabataean Petra, see the papers gathered and introduced by Martha Joukowsky 2001. For the ancient passages that crossed the rough terrain of the Petra area, see: Lindner et alii 2000: 538-545, and Zayadine 1992, who emphasized the importance of the two roads that extended from Petra to the Aravah by Naqb Namala and Naqb Ruba’i. As for the eastern approach to Petra, during recent soundings carried out under the paved street in the Siq, the remains of an earlier gravel road were uncovered, which date prior to the first century BCE; see: Bellwald 2004: 73-75. 18 For a pre-Roman date of the sites distributed along the Eastern Mount Hebron artery, see the recent reassessment of Jodi Magness 2003: 123128. For Nabataean presence south and west of the Dead Sea , see: Fischer, Gichon and Tal 2000: 137-138, and Hirschfeld 2006. For Nabataean presence in the southern Aravah, see: Dolinka 2006.

Kennedy 2004: 179, and Parker 2006a: 553. For the much debated issue of the limes Palaestinae, see the recent overview of Gichon 2002, and the criticism raised by Jodi Magness 2003: 112-129. 14 For the edition, translation and commentary on the P. Petra 1, see: Frösén et alii (ed.) 2002: 23-34. See also the review of Fournet 2003. 15 For the routes and transportation across the region in earlier times, see below. For the merging of the trans-Arabian trade routes with the region’s road network, see: Roll 2005b.

122

ISRAEL ROLL: CROSSING THE NEGEV IN LATE ROMAN TIMES

123

THE LATE ROMAN ARMY IN THE EAST

124

ISRAEL ROLL: CROSSING THE NEGEV IN LATE ROMAN TIMES Central Sinai, until Wadi Tumilat. This could have been the artery which, according to Strabo (XVI.4.2), extended for 5600 stadia from Heroonpolis in Egypt to Petra of the Nabateans and to Babylon “the whole of the journey being in the direction of the Summer sunshine”. Another route mentioned by Strabo (XVI.2.30) is “the overland passage of 1260 stadia (from Gaza) to Aila … the journeys along it being made on camels, through desert and sandy places”. That seem to be the Darb Ghazzeh, which passed by the emerging Nabataean towns of Elusa and Nessane, the road-station of Quseima and the watering places of Bir Lussan, Quraiya, Kuntilla, Bir Mashash, and down to Aqaba.19

should assume that the Roman army and administration continued to employ the existing traffic network of the region, inherited from the Nabateans. That seems to be the case with the western route, that extended from Elusa to Nessana and further to the South, along the watering places of the Darb Ghazzeh, to Aila (with an alternative short-cut from Kuntilla to Yotvata). The problem with this route is that no Imperial road installations of any kind have been identified along it south of Nessana. Therefore, its traditional identification with the Elusa-Aila artery of the Tabula Peutingeriana is rather problematic. On the other side of the Negev and in the northern Aravah, the arteries mentioned above, which connected between the metropolis Petra and the colony and legionary camp of Jerusalem, continued to serve traffic also under the Romans. That was also the case with the route that stretched from Petra to the south-west, along the southern Aravah to Aila. Furthermore, the Nabataean-created artery, that was used by the Romans extensively and, in any case, much more than any other way of the Negev, was clearly the Petra-Gaza road. Archaeological data provided by surveys and excavations carried out at almost all its road installations indicate that this artery continued to serve, intensively, throughout the second and third centuries CE. A tower erected near by the southern exit of Oboda, towards Petra, and dated by its foundation inscription to 293/294 CE, seem to indicate that security was maintained along this specific artery during the Tetrarchy. If we accept a late third-century date for the military camp at Oboda (Avdat), it implies the garrisoning of a full-scale military unit there, which was certainly involved also, if not mainly, with the task of securing traffic along the Petra-Gaza road.21

The most important artery that crossed the Negev in Nabatean times was the one which connected between Petra and Gaza. Actually, this artery made the final segment of the fabulous overland incense road, which began at Timna in Souhtern Arabia and, according to Pliny the Elder (12, 32, 64), was 1487 miles long and included 65 halts for camels. From Petra, the road made its way to the Aravah along Naqb Namala, until the roadstation of Moa. From there, it followed the course of Nahal Neqarot to the road-station of Sha’ar Ramon and then, climbed the ascent of Mahmal towards the tableland of the Central Negev until reaching Oboda. The key points of this section of the road were duly dotted with guarding stations and cisterns. From Oboda the artery continued along the course of Nahal Besor to Elusa and then, to Gaza.20 The emperor Trajan ordered, in the year 106 CE, the acquisition of the Nabataean Kingdom to the Roman Empire and the formation of the provincia Arabia in its place. Road-building in the newly annexed province became a first priority and, soon after the acquisition and under the same emperor, a first-class strategic highway was erected across its entire territory, as we noted above. A series of inscribed milestones found along it indicate that it was built and paved (apervit et stravit) during the years 111-114 CE, under the care of Trajan’s legate Claudius Severus, that it extended from the limits of Syria to the Read Sea (a finibus Syriae usque ad Mare Rubrum), and that its official name was Via Nova. The accomplishment of this enterprise had a direct impact on the communication lines that extended from it to the west and covered the region that later became Palaestina III. From now on, the arteries of this region evolved into a rear road-network of Arabia’s chief highway. On the other hand, we don’t have any parallel epigraphic evidence to attest tangible Imperial road building along the rear roads of the newly built highway. Therefore, we

During the fourth century CE a significant shift occurred in the Imperial priorities concerning the Roman communication network of Palaestina Tertia Salutaris. A new Imperial highway was established, which connected Aelia Capitolina (Jerusalem) with Petra – “the metropolis of the Third Palaestina Salutaris” (see map 2), and with Aila which, according to Eusebius (Onom. 6. 17-19) was located “at the limits of Palaestina in the southern desert and by the Red Sea, whence freight was shipped from Egypt and from India”. This seem to be the road mentioned subsequently by Eusebius (Onom. 8. 6-9) when discussing the village called in his own days Tamara “which is located at day’s journey distant from Mampsis, on the road from Hebron to Aila, where there is now a military garrison”. This Tamara should certainly be identified with Hazeva, which is indeed located at one day’s journey form Mampsis, at the northern end of the Aravah-road to Aila. Near the main spring of Hazeva, a square Roman fortress of the second half of the third century CE has been uncovered, as well as an impressive

19

For a comprehensive study on the Darb al-Hajj, see: Tamari 1982. For the Darb Ghazzeh, see: Meshel 2000: 99-117. 20 For an up-to-date state of research on the Incense Road, see: Roll 2005b: 111-113. For a most recent summary on the Petra-Gaza road, see: Erickson-Gini 2006: 160-163. I have been informed by the authors that more detailed publications on the Petra-Gaza artery and the sites along it will soon be published by C. Ben-David and by Y. Hirschfeld (who has just passed away).

21

For references on the Via Nova, see above note 9. For the NessanaAila road, see: Meshel 2000: 99-117. For the continuity of the Nabataean roads and sites into the Roman period, see: Erickson-Gini 2006: 163-165. For the Diocletianic inscription found at Oboda, see: Negev 1981: 26-27. For a late third century date of the military camp at Oboda, see: Erickson-Gini 2002.

125

THE LATE ROMAN ARMY IN THE EAST bathhouse. In the late third or early fourth century CE, most probably under Diocletian according to the excavator, four projecting corner towers have been added to the fortress, which made of it the largest military structure of its kind in the entire Aravah, until Aila.22

lowered gradually along the southern bank of Nahal Peres, until the plain. From Zoara, the road ascended to the east, towards the Via Nova, along a recently discovered built and paved ascent.24 The main branch of the highway continued from Mampsis to the south-east, until the limits of the eastern Negev’s tableland. There, another fall of altitude had to be overcome, of 640 meters along less than 3 Kilometers, which lowers towards Hazeva in the Northern Aravah. To cope with that, a sophisticated engineered ascent has been built – or rather carved out of the limestone rock, which would be a better definition. The ascent is called today Ma’aleh Aqrabim, after the similar toponym mentioned in the Bible (Num 34. 4; Jos 15. 3; Jud 1. 36), which was translated by Jerome (Comm. in Ezech. 19) ascensus scorpionis. In English the ascent is indeed know today as the Scorpion Pass. It comprises clusters of long and shorter serpentines, with margins carved into the rock. The surface of the road consists of leveled rock as well as of built segments of pavement made of rectangular carved blocks. Along particularly steep sections of the ascent, series of steps were carved out of the rock, to attenuate the gradient’s harshness. Excavations carried out at the four road installations distributed along the ascent indicate that these facilities were built on a uniform plan, using similar materials, probably as a part of a single construction project, in the first half of the fourth century CE. A group of anepigraphic milestones discovered north of the ascent indicates that this thoroughfare was considered by the Imperial authorities an official line of transportation.25

The highway under discussion left Jerusalem towards the south, running along the watershed of the western mountain range of Judea, until Hebron. A series of inscribed milestones found along this northern segment indicate that it served as an Imperial artery already in the second century CE. Then, the road converged upon the military garrison of Chermela, and continued towards the mountain of Amasa, at the southern end of the Judean range, which reaches the altitude of 860 meters above sea level. South of it, the terrain falls sharply in altitude, for some 400 meters along 4 kilometers, until the Valley of Arad is attained. To cope with this fall of terrain, an impressive built and engineered ascent was erected called today Ma’aleh Dragot, which stretched along the watershed of a series of hills that ondulate gradually downwards. Along its markedly steep segments, steps have carefully been built, to attenuate the steep gradient and to make travel easier, for men and for beasts of burden. One should add that these built steps were clearly unfit for wheeled traffic. As I have recently elaborated in another occasion, the making and use of built or rockcarved steps indicate a Late Roman building of the road or, alternatively, a Late Roman repair of it. That method has been embraced when wheeled traffic ceased to serve as a main mean of Imperial transportation, apparently, from the late third or rather from the early fourth century onwards.23

From the Late Roman fortress of Hazeva to the south and along most of the Aravah, no remains of the highway have been preserved but, its existence cannot be a matter of doubt. One may assume that, beyond Hazeva, the road’s course crossed to the eastern side of the Rift Valley, which enabled a direct approach to the copper mines of Phaenon, and further on along the Naqb Namala, to Petra. An even easier approach to Petra could be carried out further to the south, along the newly discovered paved ascent of Naqb Ruba’i. Apparently, this ascent was aimed first and foremost to serve the traffic from Petra towards the south but, given the fact that it was the only available paved road that descended from Petra to the west (according to our present knowledge), it could have served the traffic to the northern Aravah as

In the Valley of Arad, the remains of the Roman highway disappear completely, under a thick layer of loess, but, there is no doubt that its course continued southwards, to the fortress of Malatha. From there, the highway stretched further on to Mampsis, which became a fortified town and a main crossroads of the northeastern Negev, apparently, under Diocletian. From there, an eastern branch of the highway extended straight to the east, passed by the Late Roman quadriburgium called today Mezad Tamar, and descended towards the salty plain of Zoara. That means a fall of altitude of not less than 900 meters. To cope with that, two consecutive ascents have been built and engineered by the Romans: the Ma’aleh Tamar – that consists of a sequence of large descending serpentines along a steep slope, and Ma’aleh Peres – that

24 For the importance of Mampsis from Diocletian onwards, see: Negev 1988: 1-3. For Mezad Tamar, see the preliminary report of Gichon 1976. The two ascents of Ma’aleh Tamar and Ma’aleh Peres have not yet been properly published; see the summary of Ben-David 2002: 159161. The paved ascent of Zoara, on the other hand, has been surveyed and published recently by Ben-David 2002b. 25 A preliminary topographical study on the Ma’aleh Aqrabim has been carried out almost half a century ago, by Harel 1959. Excavations carried out by R. Cohen (1993: 1140-1144) along the military installation distributed along the ascent have shown that they were built as a part of a single construction project, in the first half of the fourth century CE. For the milestones found north of the ascent, see: Roll 1989: 260, n. 101.

22

For the appropriate references on Tamara, see: Tsafrir et alii 1994: 247. For the latest preliminary account on the excavations at Hazeva, see: Cohen and Israel 1996; the finds point towards the site as the best candidate for Tamara hence, one can hardly follow the most recent suggestion by Hirschfeld 2006: 170-171, to identify Tamara with En Tamar located further to the north. 23 The texts of the inscribed milestones uncovered along the JerusalemHebron road have been gathered and published nine decades ago by Thomsen 1917: 81-83; an up-to-date study on this artery is still wanting. For a recent topographical study on the Ma’aleh Dragot based on aerial photos, see: Fixler and Kloner 2004. For my recent discussion on the stepped ascents, the way they were used and their date, see: Roll 2005a.

126

ISRAEL ROLL: CROSSING THE NEGEV IN LATE ROMAN TIMES well. The eastern alignment of the highway in the central Aravah is attested to by segments of a paved road still preserved near and to the south of Gharandal, and by a group of several fractured and non-inscribed milestones uncovered in a limestone quarry near Qa’a Saidiyn. The road also passed by the recently surveyed Roman military installations located at Bir Madkur, Qasr Tayiba, Qa’a Saidyin and Qasr Gharandal.26

The ultimate destination of the highway was certainly Aila, but its remains are still wanting. In sum, the road that stretched for Jerusalem to Petra and to Aila was organized as the chief Imperial artery of traffic in the rear of Palaestina Tertia Salutaris, and it seems to have remained so throughout the whole of Late Antiquity.29

Further to the south and near by the modern settlement of Yahel, three milestations have been uncovered by Uzi Avner and his team of surveyors, along an alignment which implies a bend of the road towards the western side of the Aravah. The milestones have the unusual shape of a stele, or of a straight slab, and most of them bear carved or painted inscriptions dating to the first and second Tetrarchy. Some of the inscriptions mention the name of the caput viae – Osia. One of the inscriptions also provides the distance from it – 12 Roman miles, which brings us precisely to the Roman fortress at Yotvata.27 Let us remember that at this fortress an Imperial Tetrarchic building inscription has been uncovered, which mentions the name of the place – Costia – and also commemorates the garrisoning of a newly created mounted unit there. Apparently, that was the reason for declaring the place an official destination for an Imperial traffic line, officially denominated Osia/Costia in Late Antiquity.28

The data in the two maps has been provided by I. Roll and drawn by Yura Smertenko, on a Digital Terrain Model made by John K. Hall with his kind permission

Note

Bibliography Avner U. - Davies G. and Magness J. 2004 The Roman fort at Yotvata: interim report (2003). In JRA 17: 405412. Avner U. and Roll I. 1997 Southern Arava, Roman milestones. In ESI 16: 135. Barker G. W. et alii 1999 Environment and land-use in the Wadi Faynan, Southern Jordan: the third season of geoarchaeology and landscape archaeology (1998). In Levant 31: 255-292. Barnes T.D. 1996 Emperors, panegyrics, prefects, provinces and palaces (284-317). In JRA 9: 532-552. Bauzou T. 1998 Le secteur nord de la via nova en Arabie de Bostra à Philadelphia. In Humbert, J.-B. and Desreumaux, A. (ed.), Fouilles de Khirbet esSamra en Jordanie, I. Turnhout: 101-255. Bauzou T. 2000 La “Strata Diocletiana”. In Nordiguian L. and Salles J.-F. (ed.), Aux origines de l’archéologie aérienne: A. Poidebard (1878-1955), Beirut: 79-91. Bellwald U. 2004 Streets and hydraulics: the Petra National Trust Siq Project in Petra 1996-1999. The archaeological results. In Bienert H.-D. and Haeser J. (ed.), Men of Dikes and Canals: The Archaeology of Water in the Middle East, Rahden: 73-94. Ben-David C. 2002a The roads from Judea to Zoar and to Moab in the Roman period. In Eshel Y. (ed.), Judea and Samaria Studies 11: 151-164 (Hebrew). Ben-David C. 2002b The Zoar ascent – a newly discovered Roman road connecting Zoar-Safi and the Moabite plateau. In Freeman P. et alii (ed.), Limes XVIII. BAR-IS 1084. Oxford: 103-112. Ben-David C. 2005 The paved Roman road from Petra to the Arabah. In Cathedra 116: 31-48 (Hebrew). Blockley R.C. 1992. East Roman Foreign Policy. Formation and Conduct from Diocletian to Anastasius, Leeds.

26 The arguments in favor of the existence of a north-south Imperial artery along the Aravah in Late Antiquity have been formulated and discussed most recently by Ben Isaac 2006. The importance of Phaenon in Classical times is gradually brought to light during the multi-purpose survey in the Wadi Faynan carried out by Barker et alii 1999. For the centrality of Petra in Late Antiquity, see the comprehensive overview of Fiema 2002. For the paved ascent of Naqb Ruba’i, see: Frösén et alii 2001: 389-391, and the more detailed paper of Ben-David 2005. For the paved segments along the eastern side of the Aravah and the Late Roman sites there, see: Smith, Stevens and Niemi 1997: 59-65 and Smith 2005. 27 A preliminary account on the Late Roman milestones found near Yahel have been presented in a joint paper of Uzi Avner and myself at the XX Limes Congress held in León, Spain, in September 2006. The paper’s title is “Tetrarchic milestones and related roads in the Southern Aravah”, intended to be published, with full bibliographical apparatus, in the Proceedings of the Congress. For a previous short announcement on the discovery, see: Avner and Roll 1997. 28 In my initial publication of the Tetrarchic buildind inscription found at Yotvata (Roll 1989), I suggested to reconstruct the text in line 6 alam c(um) osti{o} constituerunt, on the basis of a series of circumstantial arguments. Nowadays and following the discovery of the Tetrarchic milestones at Yahel mentioning Osia as their caput viae, which is clearly located at Yotvata, I must reconsider the issue. It is clear to me now that line 6 should indeed be read as is: alam Costia constituerunt, as suggested by Eck 1992. On the other hand, my proposed reconstruction of lines 7-9: per providentia(m) Prisci/ pr(a)esidis [provinciae/ Syriae Palaestinae] has been proven to be correct. Two honorific inscriptions found at Caesarea (Burrell 1993; see also: Lehman and Holum 2000: 49 and 51) attest that Priscus was indeed a pr(aeses) prov(inciae) Pal(aestinae). Barnes (1996: 548 and 550) was clearly unaware of Burrell’s publication, when writing three years later that my suggested reconstruction “is unjustified”. One may add that the renewed excavations at the fort of Yotvata carried out in 2003 have not yelded new substantial data on its Tetrarchic phase; see: Avner, Davies and Magness 2004.

29

For the importance of the north-south Imperial highway in the economic relationship between Roman Aila and the Aravah, see: Parker 2006b. For the data on the military aspects of Roman Aila, the monumental marble inscription of Constantinian date included, see: Kennedy 2004: 206-209. The Greek and Roman sources on Aila and its surroundings have been gathered and translated into Hebrew by Tsafrir 1989.

127

THE LATE ROMAN ARMY IN THE EAST Fiema Z.T. 2002 Late-Antique Petra and its hinterland. In Humphrey J.H. (ed.), The Roman and Byzantine Near East, 3, Portsmouth: 191-252. Fischer M. - Gichon M. and Tal O. 2000. En Boqeq. Excavations in a Oasis on the Dead Sea, II. The Officina, Mainz. Fixler Y. and Kloner A. 2004 Mapping and topographical study of Ma’aleh Dragot based on airphotoes. In BarGal Y. et alii (ed.), Erets Israel Studies – Aviel Ron Book, Haifa: 83-98 (Hebrew). Fournet J.-L. 2003 Review on The Petra Papyri, I. In AntTard 11: 398-404. Frösén J. et alii 2001 The 1998-2000 Finish Harun Project: Specialized Reports. In ADAJ 45: 377-392. Frösén J. et alii (ed.), 2002 The Petra Papyri, I. ACOR, Amman. Gichon M. 1976 Excavations at Mezad Tamar/”Tamara” 1973-75. Preliminary Report. In Saalburg-Jahrbuch 33: 80-94. Gichon M. 2002 45 years of research on the limes Palaestinae - the findings and their assessment in the light of the criticism raised (C 1st – C 4th). In Freeman P. et alii (ed.), Limes XVIII. BAR-IS 1084, Oxford: 185-206. Gilmartin K. 1973 Corbulo’s campaign in the East: an analysis of Tacitus’ account. In Historia 22: 583-626. Graf D.F. 1995 The Via Nova Traiana in Arabia Petraea. In: Humphrey, J. H. ed. The Roman and Byzantine Near East; Some Recent Archaeological Research, Ann Arbor: 241-267. Graf D.F. 1997 The Via Militaris and the Limes Arabicus. In Groenman van Waatering W. et alii (ed.), Roman Frontier Studies 1995, Oxford: 123-133. Graf D.F. 2002 Nomads and the Arabian frontier: the epigraphic perspective. In Freeman P. et alii (ed.), Limes XVIII. BAR-IS 1084, Oxford: 153-160. Gregory S. 1995-97 Roman Military Architecture on the Eastern Frontier. 3 Vols. Amsterdam. Harel M. 1959 The Roman road at Ma’aleh Aqrabbim (“Scorpions’ Ascent”). In IEJ 9: 175-179. Hirschfeld Y. 2006 The Nabataean presence south of the Dead Sea: new evidence. In Bienkowski P. and Galor K. (ed.), Crossing the Rift, Oxford: 167-190. Hostein A. 2004 Le Corpus de Panegyrici Latini dans deux ouvrages récents. In AntTard 12: 373-385. Isaac B. 1992 The Limits of Empire. The Roman Army in the East. 2nd ed. Oxford. Isaac B. 1998 The Near East under Roman Rule. Selected Papers, Leiden. Isaac B. 2006 Roman organization in the Arabah in the fourth century AD. In Bienkowski P. and Galor K. (ed.), Crossing the Rift, Oxford: 215-221. Jasmin M. 2006 The emergence and first development of the Arabian trade across the Wadi Arabah. In Bienkowski P. and Galor K. (ed.), Crossing the Rift, Oxford: 143-150. Joukowsky M.S. 2001 Nabataean Petra. In BASOR 324: 1-4. Kennedy D. 1988 A lost Latin inscription from the banks of the Tigris. In ZPE 73: 101-103.

Boschung D. and Eck W. 2006 Die Tetrarchie, Wiesbaden. Breeze D. and Jilek S. 2005 Strategy, tactics, operation: how did frontiers actually work? In Visy Z. (ed.), Limes XIX. Pécs: 141-146. Burns R.A. 2003 The defence of Roman Damascus. In Dąbrowa E. (ed.), The Roman Near East and Armenia. Electrum 7. Krakow: 47-69. Burrell B. 1993 Two inscribed columns from Caesarea Maritima. In ZPE 99: 287-295. Butcher K. 2003 Roman Syria and the Near East, London. Carrié J.-M. 1994 Dioclétien et la fiscalité. In AntTard 2: 33-64. Carrié J.-M. and Janniard S. 2000 L’Armée romaine tardive dans quelques travaux récents. In AntTard 8: 321-341. Chastagnol A. 1994 L’Évolution politique du règne de Dioclétien. In AntTard 2: 23-31. Cohen R. 1993 Hellenistic, Roman and Byzantine sites in the Negev hills. In NEAEHL 3: 1135-1145. Cohen R. and Israel Y. 1996 En Hatzeva 1990-1994. In ESI 15: 110-116. Corcoran S. 2000 The Empire of the Tetrarchs: Imperial Pronouncements and Government AD 284-324. Rev. ed. Oxford. Dąbrowa E. 2000 Legio X Fretensis. In Le Bohec Y. (ed.), Les légions de Rome sous le Haut-Empire, Lyon: 317-325. Demandt A. et alii (ed.) 2004 Diokletian und die Tetrarchie, Berlin-New York. Dodgeon M.H. and Lieu S.N.C. 1991 The Roman Eastern Frontier and the Persian Wars (AD 226-363): A Documentary History, London. Dolinka B.J. 2006 The Rujum Taba Archaeological Project (RTAP): results of the 2001 survey and reconnaissance. In Bienkowski P. and Galor K. (ed.), Crossing the Rift, Oxford: 195-214. Eadie J. 1996 The transformation of the eastern frontier, 260-305. In Mathisen R.W. and Sivan H.S. (ed.), Shifting Frontiers in Late Antiquity, Aldershot: 72-81. Eck W. 1992 Alam Costia constituerunt. In Klio 74: 395400. Erickson-Gini T. 2002 Nabataean or Roman? Reconsidering the date of the camp at Avdat in light of recent excavations. In Freeman P. et alii (ed.), Limes XVIII. BAR-IS 1084, Oxford: 113-130. Erickson-Gini T. 2006 “Down to the Sea”: Nabataean colonization in the Negev highlands. In Bienkowski P. and Galor K. (ed.), Crossing the Rift, Oxford: 157166. Farrokh K. 2005 Sassanian Elite Cavalry AD 226-642. Osprey, Oxford. Fiema Z.T. 1995 Military architecture and the defence “system” of Roman – Byzantine southern Jordan: a critical appraisal of current interpretations. In Studies in the History and Archaeology of Jordan, V, Amman: 261-269.

128

ISRAEL ROLL: CROSSING THE NEGEV IN LATE ROMAN TIMES Nicasie M. J. 1998 Twilight of Empire: The Roman Army from the Reign of Diocletian until the Battle of Adrianople, Amsterdam. Nixon C.E.V. and Rodgers B.S. 1994 In Praise of Later Roman Emperors. The Panegyrici Latini, Berkeley. Parker S.T. 1991 The Nature of Rome’s Arabian Frontier. In Maxfield V.A. and Dobson M.J. (ed.), Roman Frontier Studies 1989, Exeter: 498-504. Parker S.T. 2002 The Roman frontier in Jordan: an overview. In Freeman P. et alii (ed.), Limes XVIII, BAR-IS 1084, Oxford: 77-84. Parker S.T. 2005. Supplying the Roman army on the Arabian frontier. In Visy Z. (ed.), Limes XIX, Pécs: 415-425. Parker S.T. 2006a The Roman Frontier in Central Jordan. Final Report of the Limes Arabicus Project 1980-1989. Dumbarton Oaks Byzantine Studies, 2 vols, Cambridge MA. Parker S.T. 2006b Roman Aila and the Wadi Arabah: an economic relationship. In Bienkowski P. and Galor K. (ed.), Crossing the Rift. Oxford: 223-230. Reddé M. 1995 Dioclétien et les fortifications militaries de l’Antiquité tardive: quelques considérations de méthode. In AntTard 3: 91-124. Rees R. 2004 Diocletian and the Tetrarchy, Edinburgh. Retsö J. 2003 The Arabs in Antiquity: Their History from the Assyrians to the Umayyads, London-New York. Roll I. 1989 A Latin Imperial inscription from the time of Diocletian found at Yotvata. In IEJ 39: 239-260. Roll I. 1999 The roads in Roman-Byzantine Palaestina and Arabia. In Piccirillo M. and Alliata E. (ed.), The Madaba Map Centenary 1897-1997, Jerusalem: 109113. Roll I. 2002 Crossing the Rift Valley: the connecting arteries between the road networks of Judaea/Palaestina and Arabia. In Freeman P. et alii (ed.), Limes XVIII. BAR-IS 1084, Oxford: 215-230. Roll I. 2005a Terrestrial transportation of the Roman army in the East: the case of Judaea/Palaestina and its road network. In Visy Z. (ed.), Limes XIX, Pécs: 749-761. Roll I. 2005b Imperial roads across and trade routes beyond the Roman provinces of Judaea/Palaestina and Arabia: the state of research. Tel Aviv 32: 107118. Sartre M. 2005 The Middle East under Rome, Cambridge MA-London. Smith A.M. - Stevens M. and Niemi T.M. 1997 The Southeast Arabah Archaeological Survey: A preliminary report of the 1994 season. In BASOR 305: 45-71. Smith A.M. 2005 Pathways, roadways and highways: networks of communication and exchange in Wadi Araba. In Near Eastern Archaeology 68: 180-189. Southern P. and Dixon K.R. 1996 The Late Roman Army. London. Tamari S. 1982 Darb al Hajj in Sinai: an historicalarchaeological study. Rome.

Kennedy D. (ed.) 1996 The Roman Army in the East, Ann Arbor. Kennedy D. 2004 The Roman Army in Jordan. 2nd ed. London. Kolb F. 1995 Chronologie und Ideologie der Tetrarchie. In AntTard 3: 21-31. Konrad M. 1996 Fruehkaiserzeitliche Befestigungen an der Strata Diocletiana. In Damaszener Mitteilungen 9: 163-180. Konrad M. 1999 Research on the Roman and Early Byzantine frontier in North Syria. In JRA 12: 392410. Kuhoff W. 2001 Diokletian und die Epoche der Tetrarchie: das roemische Reich zwischen Krisebewaeltigung und Neuaufbau (284-313 n. Chr.), Frankfurt am Main. Leadbetter B. 2002 Galerius and the eastern frontier. In Freeman P. et alii (ed.), Limes XVIII. BAR-IS 1084, Oxford: 85-89. Leadbetter B. 2004 Trade, frontiers and the Limes Palaestinae. In Studies in the History and Archaeology of Jordan, VIII, Amman: 247-251. Lehmann C. M. and Holum K.G. 2000 The Greek and Latin Inscriptions of Caesarea Maritima. ASOR, Boston. Lewin A. 1990. Dall’Eufrate al Mar Rosso: Diocleziano, l’esercito e i confini tardoantichi. In Athenaeum 78: 141-165. Lewin A. 2002 Diocletian: politics and limites in the Near East. In Freeman P. et alii (ed.), Limes XVIII. BAR-IS 1084, Oxford: 91-101. Lindner M. et alii 2000 Nabataean and Roman presence between Petra and Wadi Arabah survey expedition 1997/98: Umm Ratam. In ADAJ 44: 535-564. Luttwak E.N. 1976 The Grand Strategy of the Roman Empire from the First Century AD to the Third, Baltimore-London. Mac Donald B. 1996 The route of the Via Nova Traiana immediately south of Wadi al-Hasa. In PEQ 128: 12-15. Magness J. 2003 The Archaeology of the Early Islamic Settlement in Palestine, Winona Lake. Malavolta M. 1982 Interiores limites (nota ad Amm. Marc. XXIII, 5, 1-2). In Ottava miscellanea greca e romana, Roma: 587-610. Matthews J. 1989 The Roman Empire of Ammianus, Baltimore. Mayerson P. 1988 A note on the Roman Limes: “Inner” versus “Outer”. In IEJ 38: 181-183. Meshel Z. 2000 Sinai: Excavations and Studies. BARIS 876, Oxford. Millar F. 1993 The Roman Near East, 31 BC – AD 337, Cambridge MA-London. Negev A. 1981 The Greek Inscriptions from the Negev, Jerusalem. Negev A. 1988 The Architecture of Mampsis. Final Report, II: The Late Roman and Byzantine Periods. Qedem 27, Jerusalem.

129

THE LATE ROMAN ARMY IN THE EAST Thomsen, P. 1917. Die roemischen Meilensteine der Provinzen Syria, Arabia und Palaestina. In ZDPV 40: 1-103. Tsafrir Y. 1989 Aila (Eilat) and its Gulf in Greek and Roman sources. In Cathedra 53: 149-192 (Hebrew). Tsafrir Y. et alii 1994 Tabula Imperii Romani – Iudaea/Palaestina: Maps and Gazetteer, Jerusalem. Vervaet F.J. 2002 Caesenius Sospes, the Neronian wars in Armenia and Tacitus’ view on the problem of Roman foreign policy in the East: a reassessment. In Mediterraneo antico 5: 283-318. Whittaker C.R. 2004 Rome and its Frontiers: The Dynamics of Empire, London-New York. Wilkes J.J. 2005 The Roman Danube: an archaeological survey. In JRS 95: 124-225. Williams S. 1985 Diocletian and the Roman Recovery, London-New York. Winter E. 1989. On the regulation of the eastern frontier of the Roman Empire in 298. In French D.H. and Lightfoot C.S. (ed.), The Eastern Frontier of the Roman Empire. BAR-IS 553, Oxford: 555-571. Zayadine F. 1992 L’Espace urbain du grand Pétra, les routes et les stations caravanières. In ADAJ 36: 217239. Zuckermann C. 1994 Les campagnes des tétrarques, 296298. In AntTard 2: 65-70.

130

From the Constructive Technique Used in the Military Architecture of the Limes Arabicus to its Conservation and Valorisation Giuseppe Claudio Infranca Istituto Superiore per le Tecniche di Conservazione dei Beni Culturali e dell’Ambiente ‘A. de Stefano’, Rome

When the desertification processes begun in the Azraq plain, and the environment transformed from green savannah, full of water, to an arid steppe, was probably when the Roman castle of Aseikhim was built.

The colour variation was unique in its chromatic variation, from east to north the black of the basaltic rocks shading in the whitish of the liken patina that thousands and thousands of stones were covered with since a long time. From south to west on the other hand, the yellowish calcareous hills painted in red by the multitudes of red flints and punctuated everywhere by spots of white colour. They were huge rounded boulders that only waited the pining moment of when the great white limestone would split in a uniform way because of the great temperature variations to then become small pebbles at the mercy of both wind and sun.

The Roman castle, situated along the limes Arabicus, was built on the highest top of that eastern strip of land of the Roman province called Arabia. The conical shaped hill that dominated the underlying plain and overlooked the sinuous wadi Aseikhim that reached the arid clearings of the Arab desert, and advanced slow but inexorable. Septimius Severus ruled at this time. The Emperor, who was of Arab origin, before turning his attention to the eastern part of his empire, had to consolidate his power in Rome. The limes Arabicus had already been built along the Via Nova Traiana. The province Arabia, where the oriental front had been created, had been under the Roman direct control for a long time, and the army had already built the first military structures of the new alignment of the limes Arabicus. The reason that induced the Romans to make such a shift in the alignment of the limes Arabicus was a strictly economic one, linked to the control of the principal silk, incense and spices trade routes from the East that arrived to Rome’s Empire. It was thus of increasing importance the construction of a new limes which integrated the pre-existing one crossing it with the Strada Diocleziana, a road built for this new occurrence.

The plain below the Aseikhim hill had the same geomorphological characteristics of the top. It was, in fact, constituted by black basalt while the declivity where yellowish in colour because of the calcareous conformation mixed to the purple red brought by the flint horizons, that run down to the plain with their shrill jingling. Because of this chromatic variation, the hill was named Aseikhim, which in Arabic means strong change of colours from black to reddish. Then as today the arid landscape of Aseikhim is unchanged. The Arabic desert around the Aseikhim hill is, today, full of life and rich in archaeological and historical heritage, such as the Bedouin settlements, the rounded enclosures for domesticated animals and the huge traps, constituted by dry walls in basaltic stone, of various shapes and sizes with their kilometric protuberances. Once upon a time, these traps were used by the lords of the desert for hunting, so much as to extinguish any presence of gazelles, foxes, ostrich, and other wild animals that once used to live in this area.

So that Qasr Aseikhim together with Qasr Azraq and the other military installations of that area became a necessity both as an act of territorial expansion, and for a grater control of the principal commercial routes from the East. The choice of the site, in an elevated position, guaranteed control over the territory and vigilance of the limes as commercial border for the transit of both goods and people.

The arid environment where the Roman castle of Aseikhim was built must have been surprisingly live with both trade of goods and peoples who crossed the frontier on a daily basis, and on a vegetation and faunal level for the rich and blooming environment.

The conical shaped hill of Aseikhim,1 today, as at the time of the ancient limes Arabicus, has an open view in all directions of the compass card.

Thus, since the limes was both frontier, and access to the opulent Roman Empire, it represented an extremely important geographical area from a political and environmental point of view, in which man had a fundamental role as builder and planner of military and urban systems of exceptional cultural and historical value.

The landscape changed in both colours and geological aspects. From the layers of cooled volcanic lava, which had changed into a vast heap of black basalt stones, to the calcareous plain overlaid by a stratum of flints, where the wadi Aseikhim crossed the territory with its winding path.

1

The Aseikhim hill is 641m high, s.l.m.

131

THE LATE ROMAN ARMY IN THE EAST In the area near Qasr Aseikhim the limes Arabicus was organised with an initial line of sighting towers and water supplies (wells or water sources). In the vicinity of the defence lines built by the Parts at first and then by the Persians there was a second line with blockhouses, that thanks to their extraordinary position represented the focal points to guarantee an initial barrier founded on the apportion of Arab auxiliary troops. The castle of Azraq represented the focal point of the borderline system in which the Roman legion was detached. Lastly, the Strata Diocletiana and the urban settlements closed the borderline system.

From the analysis of the raised walls it is evident that:

It is know, from historical sources, that the fortifications built during the 3rd century AD were occupied by military garrisons until the 5th century AD perhaps because the great earthquake that was registered between 498 and 552 AD. Furthermore, there is certainty that only the military settlements of Lejjun, Fayan, and Yasser remained to defend the limes orientalis during this last period.

b)

a)

c)

In particular, the Roman castle of Aseikhim, which was under the military control of the larger castle of Azraq, where the III Cyrenaica2 was detached, guaranteed safety and protection to the limes. Qasr Aseikhim together with Qasr Uweinid and Qasr Jawa were perfectly aligned in the complex military system of defence of the limes to guarantee the first barricade, possibly with auxiliary militias of Arab origin.

d)

As a general identification, the construction technique used at Qasr Aseikhim is the pseudo-isodomus one, and has many similarities with the IV manner of the polygonal structure that was thus named by Giuseppe Lugli, who had classified it because of the horizontal stratification, so as to be able to consider it an imperfect opus-quadratum. However, it differs from it because of the rectangular shape of the façade.

The construction technique of the Roman castle of Aseikhim The castle was certainly built after having conveniently levelled the rocky terrain on the summit, and at the same time quarrying stone from it. The construction material for the castle had been found by the Roman constructors in the quarry that had been opened near the settlement, and that functioned as fabrica where today traces of manufacture and extraction of the large blocks used in the construction of the fortress are still visible.

The other evaluation aspect for this masonry is the manner in which it was constructed, and not so much the epoch, because the characteristic of the pseudo-isodomus technique or of the polygonal one are those of using different approach techniques that will bring to the same result or to a difference in details for the processing modalities.

The remains of the castle, today, appear as a heap of debris on which stand out majestic raised walls. The masonry structure of these raised walls appears, at fist sight, to be built with the pseudo-isodomus technique, the façade of squared basaltic stone. However, after a the most accurate analysis, the conviction has been reached that, because of some details of no trifle importance, this is an autonomous novelty and a new construction technique compared to those used in the past. Thus, in the limes Arabicus, borderland region of the Roman Empire a new building technique developed in an original way, so strong in its architectural characteristics as to reach modern times with popular traditions of great cultural and scientific value.

2

The construction system of the wall is made of an external facing of large square blocks on the façade and by an internal facing of similar blocks to those used on the outside. Between the two, there is a filling of debris chips from the processing and extraction of the basaltic stone that constitutes the masonry. Only in the rooms to the eastern side, the filling is mixed earth and stone flakes. This indicates a different and successive construction, that can be traced back to the Byzantine times; The processing of the construction materials (basaltic stones) has been carried out by expert personnel who placed them with great workmanship so that they survived centuries of attacks and natural degrade; To this construction technique some extraordinary construction novelty of outstanding architectural and structural value were introduced, such as the ashlars masonry arch, and the flat ceilings built with stone blocks as replacement for wooden beams, which were difficult to find in that desert geographical location; The construction materials where always only found on the construction site for the settlement, quarrying the stone directly on location.

In the construction details of the structure at hand the different choices can be seen, which the workers faced during the work phases carried out by various teams in which all have similar capabilities and are able to work autonomously and thus with skill. Even though founded on the plan that lay the base of the military installations such as the opus quadratum, there was one single building team where the figure of the master-architect was of fundamental importance, and upon whom everything depended, thus carrying out a proto-industrial action different in both construction and work protocols. Moreover, the pseudo-isodomus technique or the polygonal one gave the fortress a greater sturdiness and

Bowersock 1983: 98.

132

GIUSEPPE CLAUDIO INFRANCA: FROM THE CONSTRUCTIVE TECHNIQUE USED IN THE MILITARY solidity that the opus quadratum and this was one of the strategic choices that the Romans wanted to give their castles that other than being in control of the western frontier were also settlements for the defence of enemy attacks. Already under the reigns of Claudius and Trajan, it is clear that at first there is a late return to the polygonal technique in the construction of the road and of the aqueduct’s bridges.3 This helped make the structure they were building both grandiose and solid. Perhaps it was through this construction technique that the new and ingenious way to build the structures of the eastern border of the Empire was introduced by the architects of Trajan’s military engineer corps, during the construction of the Via Nova Traiana in the area of the limes Arabicus. Back to the masonry of Qasr Aseikhim, it falls into the typology of the pseudo-isodomus technique, because of the stratification and the quality of the basaltic rock. However, it would be best to understand why the Roman castle of Aseikhim is comparable to the pseudo-isodomus technique, and not the opus quadratum. This is because of a series of precise references first of all the meaning given by the ancient Romans to the quadratus – “any quadrangular shape (square, rectangle, and trapezium) also means cubit, which means it not only has a value for the surface, but also of volume”4 – it is, thus, for this indisputable reason that we have accepted the walls of Aseikhim as pseudo-isodomus technique. Moreover, Vitruvius clarifies that: “When the Greeks move away from the opus quadratum they place the stones (of either rock or hard stone) in layers and carry on as with bricks forming alternate rows and thus they reach the best steadiness quality for eternity. This kind of firmness can be obtained in two ways: one of these is called isodomus, the other pseudo-isodomus. The isodomus one is that which presents rows of equal thickness; the pseudoisodomus one, on the other had, presents the blocks of stone placed at unequal distance, but have different heights. Both these techniques are equally solid.

In particular, the masonry technique found along the limes Arabicus presents some differences in the heights of the rows, always precisely horizontal, while the length of the blocks vary. The complete absence of mortar, or of a binder, becomes extraordinary in the positioning and wedging that each block had in common with the other blocks of the row. Investigating further, it appears that the Roman military architects were able to pull from the two precedent building techniques the best construction elements that they then re-elaborated in the new manner where the architectural quality and great solidity of the structure represented an absolute building novelty in the creation of the ashlars masonry arch and of the flat stone ceilings. This way a new and articulated technique was created which was called pseudo-isodomus and that because of the special architectonic and structural connotations we link to the original construction manner sought by the Roman military architects, calling it Roman. The pseudo-isodomus technique Roman manner created and developed exclusively along the limes Arabicus caused the walls of the castella to look powerful and majestic to the enemies that incautiously approached them. Furthermore, the use of the pseudo-isodomus technique Roman manner in the 3rd century AD gave new vigour to the ancient construction techniques of the Greeks, which were at the base of the new applications that were also used in medieval times. Another characteristic of this new building technique invented by the military architects of the limes Arabicus was the use of linking and wedging between two wall facings placed with an alternate rhythm in the row which gave a formal look and a structural solidity of great technologic interest, which will certainly be studied during the conservation phase.

Plinius also confirmed this version: “The Greeks build their walls with hard stones, or flints, in layers, just as they do with bricks. This kind of structure they call isodomus, when, on the other hand, the blocks (coria) are of different height, they call it pseudo-isodomus. A third construction technique is the emplecton…”

During the analysis of the raised wall structures, various restorations and alterations have been noticed (of the southern and western external sides, and the inner court northern and southern sides) which are presumed to have been carried out during the long lifetime of the remains of Qasr Aseikhim and referable to the Byzantine period. Time when it is historically known that various conservation works were carried out in many fortifications of the limes Arabicus. The stratigraphic situation is quite articulated by numerous variables formal to the original technique linked to the dimensions of the single stone components recovered or revived from quarry activity and thus the height of the rows that in some wall fractions have a slightly undulated appearance. In these cases, the use of the recovered material has been recorded as the work of the Byzantines who in the fifth

Thus, it appeared natural that the masonry structure of Aseikhim was not opus quadratum, and neither was it polygonal technique nor exclusively pseudo-isodomus. The construction technique to be found in the limes Arabicus held similarities with both the polygonal structure technique and the pseudo-isodomus technique of the Greeks, but studying the technological details, it was comparable to the pseudo-isodomus technique for the form and to the polygonal one for the joint structure.

3 4

Lugli 1957. Lugli 1957: 174.

133

THE LATE ROMAN ARMY IN THE EAST and sixth centuries AD lacking lime carried out the restorations of the defence structures5 this way.

The survey of the foundations of the castle of Aseikhim were already carried out in the first archaeological mission in 2000, when a deep hole was discovered 4m deep dug by clandestine people under the ashlars masonry arch of the central room on the southern side. The pit was dug to the side of the foundation, which was constituted by irregular stones that filled the foundation trench to complete the bed trimming of the castle, placed directly upon the bedrock.

The construction material of the Roman castle has been found and quarried on top of the Aseikhim hill itself from which the flat space was obtained to house the fortress itself. The material was thus detached from the bedrock on the south-western side of the hill, a few yards away from the castle. The architectonic elements of the castle were worked with levers or hammered wedges taking advantage of the fractures and cavities that the rock featured naturally. Once extracted the rock material this was rough-hewed with maul blows to take out protrusions and leaving the outer face perfectly smooth. The blocks, once ready for the collocation, were dragged or rolled by use of posts until their rest place, showing in this operation the art of composing simple closing up arrangements where both simplicity and power characterise the assemblage. The meeting between blocks was functional to the alignment directrix of the blocks themselves, while within the wall the filling served as further wedging between the two wall facings, which were in turn jammed with the use of the diatonics that rhythmically weave the walls apparatus creating a solid system.

Thus, the walls were held in place by their weight. Qasr Aseikhim lacked adequate angular props in the walls, as they were not necessary because there were good joints; in fact, there were few points of contact and the depth of the stones to the inside of the wall barely touched the other wall facing. Furthermore, the imitation of the opus quadratum was evident, the perfection of which, however, is not reached, both because the volumetric irregularity of the blocks is preferred, and because the construction material is the hard basalt difficult to cut with precision. The blocks were aligned with long faces slightly undulated, and every now and again there is some flaking to be found. The cutting of the blocks was carried out on four sides almost parallel to each other, the jointures were vertical, and every now and again, they present a slight obliquity corrected by the insertion of stone tassels in the resulting spaces.

The assemblage of the different blocks is the demonstration of the architectonic greatness of the Roman builders who managed with simple closing up combinations to perfectly adhere the new blocks to the precedent ones.

Moreover, a few clutch blocks have been found, with the teeth-cuts especially in the blocks already in place, to improve the cohesion between near blocks. This teethshaped cut could have been made with the bevel square or with a lead lamina to reconnect within the wall under construction with the corresponding corner; principally in the south-western corner of the castle and in the proximity of the openings to the inner court. The use of the mentioned tool, which was used to conveniently square the blocks, that had to be superimposed and tallied, gives us further proof of the technical abilities of the workers during the construction of the building, and the opportunity to square perfectly the façade of the blocks of Qasr Aseikhim.8

Probably the ancient building site had to include digging up plans of the work level by use of inclined wooden castles and by low wooden trolleys, pulled by winches.6 The castle’s foundations rest in large part on the basaltic bedrock, which presented itself as compact and adapt to house the insertion of the blocks of the raised walls that were, thus, inserted and engraved between the natural rocky summits to create a single solid structure. The other side of the foundation,7 that is not aligned to the natural rock’s flat surface, had been completed with a step made of blocks of basalt, and a trench foundation constituted by gravel that was quarry waste, upon which the first row of raised blocks was placed. The foundation trench with quarry waste gravel allowed to compact the alignment of the castle on the top of the hill. The remaining parts of the foundation are consistent thanks to the artificial step positioned to the natural bedrock of the hill’s summit on top of which the blocks of the first row were placed. On the first row the second one was then positioned and successively the others, to build the magnificent walls today ruins of the ancient fortress.

The cubic volume of the blocks varied as their depth varied, while the disposition was always according to the length respecting the height of the row. The walls are slightly convex because the dressing of the façades starts from the side of the blocks and extenuates towards the centre. The plan shape of Qasr Aseikhim is quadrangular even though its sides are uneven.9 The irregularities of the

5

8

Ravegnani 1983: 57. Lugli 1957: 75. 7 North-eastern side of the castle.

Lugli 1957: 75. The sides of the castle measure in metric: on the southern and western side 23.60m, on the northern one 24m.10, and on the eastern side 23m.

6

9

134

GIUSEPPE CLAUDIO INFRANCA: FROM THE CONSTRUCTIVE TECHNIQUE USED IN THE MILITARY layout are a characteristic of the pseudo-isodomus technique.

were restored to better face the frequent sieges of the Persians.12

The castle has only one entrance on the southern side, which gives access to a room with two openings, one opens on the inner court and the other on an inner room. From the court, there was access to the glacis and patrol communication trench through a brickwork staircase. Furthermore, it is probable that there also were some rooms on the first floor. On the contrary, it is certain that in the proximity of the corners on the southern access side, there were two rooms,10 but this has still to be proven.

Coming back to the level ceilings of Qasr Aseikhim their original composition shows the technologic workmanship set up on the arch, unfortunately inglobated in the wall on which they rest, joined oblong beams made of basaltic stone placed on projections that are somewhat similar to the pseudo-domes of the Mycenaean ‘tholos’. This very characteristic is different from any other earlier construction technique transforming the pseudo-isodomus Roman manner into a unique construction technique of great originality so much as to develop autonomous technical progress of deep technological importance. The technological innovation suggested by the Roman military architects, re-elaborating the pseudo-isodomus technique, allowed to build large structures in geographical areas where wood was scarce and stone was the only was construction resource. This is how the Romans were able to build their fortresses in an unexceptionable way with only the construction materials found in situ and without any further need for transport notwithstanding the difficult arid environmental conditions, such as those of Aseikhim.

The stratigraphic units of the collapses once they have been studied for the conservation intervention, will allow to better understanding the true consistency of the Roman castle of Aseikhim. From the study carried out up until now of the collapses on the surface, it has been noticed that these have been caused by human action and definitely not by a seismic even as was reported by some historical sources.11 However, it cannot be excluded that the earthquake that involved the region between 498 and 502 AD could have caused the definitive abandon of the castle on the part of the troops dislocate superficial in Qasr Aseikhim. This hypothesis will be better verified when the collapses will be removed and the underlying ones will be studied, which all the ones relative to the structures of the second floor. We wait the discovery of the reasons for the abandon and the exact dimension and shape of the fortresses, where the conservation works linked to the study and stratigraphic analysis of the masonry and collapses will start, together with the actual archaeological research.

Finally, from the relieves carried out it is demonstrated that all measurements correspond to the mathematical standards applied by the Roman architects and that, thus, the castle was built by using a plan, or with a construction model typical of Roman architecture. Around the castle run two perimeter walls built by use of the polygonal technique of very poor craftsmanship, which suggests an earlier date for these walls that are not coeval to the Roman fort.

Ten rooms have been found on the ground floor, and they all have rectangular shape.

The perimeter walls, on lower heights than the donjon have two guard stationings drawn by the terrain’s orography itself, and were used to keep under control the principal transit route which run parallel to the wadi Aseikhim. The principal door of this fortification is on the south-western side and is of the scea type. (See aerial photo). Aureil Slein, who flew over the castle during his unusual and adventurous survey on the traces of the Silk Route, noticed that both perimeter walls, and both surrounds13 had been built later than the original construction of the Roman fort and had possibly been built by Arab auxiliary troops.

Only in three out of these ten rooms the weaving of the stone ceiling was not possible to detect because the collapses would not allow a correct reading of the architectonic spaces (see plan). The ceilings have been built by using ashlars masonry arches, usually placed on the long sides of the rooms, only on the eastern side there are some ashlars masonry arches on the short sides of the inner rooms. The eastern side probably suffered extensive restorations during its lifetime, this is clear from the different joint technique of the blocks and the different composition of the wall filling, stuffed with earth and small ill placed stones. It is thought that this restoration was carried out under Justinian, when many fortresses of the limes Arabicus

Stein, moreover, conjectured that the Arab troops had, also, put up their tents close to the castle, to live in. Stein does not, however, mention the numerous stemming that 12

This information has been reported by the historian Procopius and by the Anonymous. 13 The two large surrounds discovered by Stein have been investigated during the 1st archaeological mission in Aseikhim and has been suggested that it would be more logical for them to be neither hydraulic stemming nor tent camps, but for their square shape they were used to keep horses and dromedaries belonging to the troops detached in Aseikhim.

10 Thus, the original aspect probably had to be that of a fortress with two angular towers if not with four towers on all the corners. However, this still has to be proven by future restoration works to the monument! 11 Amiran 1950.

135

THE LATE ROMAN ARMY IN THE EAST were to be found outside of the defence walls, even though they are clearly visible in the picture he took during his trip, and that had, presumably, been built contemporarily to the surrounds in a later historical time, much closer than we can imagine. Only archaeological research will give us an answer to such questions!

links of the inner walls of the castle are functional to sustaining the ceilings and the blocks placed within the masonries. This solution was obtained by placing small pebbles as earthing up, which also allowed the draining of the water that infiltrated with the rains. The building site works, therefore, had to proceed only thanks to two groups of workers,16 they worked in opposite directions to reach the corner, and from here they went back to erect another row of blocks and so on. For this reason the joining of the two sides and of the corners is necessary, until the suture of the four building sites through a keystone, that it has in common with the other three sides and two corners, the thickness of the blocks and the internal static setting. Where the stone material resulted lacking, small triangular pads were used, in order to obviate to the inconvenience.

The conservative intervention at Qasr Aseikhim We have ascertained how the origin of this construction typology derives from the pseudo-isodomus technique and the polygonal one, as Giuseppe Lugli states.14 The Romans, during the Imperial period, build public works such as bridges, aqueducts and castles with the intention of making them sturdy and grandiose similar to the ancient megalithic masonries. In such cases, the material used was the limestone, as in the fortifications of the southern limes Arabicus, while in the northern castles of the limes Arabicus they were built with basalt, only because it was the principal material to be found and the most durable stone material, easily found, in the region.

The external surfaces were worked with a tooth chisel to make them perfectly smooth, while the faces where cut with maul and chisel. The wall facing was aligned perpendicularly, except where it was necessary to calculate the gradient necessary for the tapering of the wall. The workers in this specific case, demonstrated the perfect knowledge of the work tools and construction techniques. Lastly, in the doors, some useful expedients were used, such as creating false arches in order to avoid the lateral push of the blocks and to have only the compression one (vertical).

The walls of the castle were constituted by a double masonry facing with a dry stone technique interior and on the exterior with the pseudo-isodomus technique in Roman manner, composed by regular rows. The masonry structure tends to create all possible closing up combinations of these huge basaltic blocks in order to adhere the various stones which constituted the wall facing. When this was not possible the empty spaces where filled with small stones, avoiding thus the roughhewing during their positioning. The walls of the castle did in fact hold up because of their weight, and they were supported by both good joints and disjointments.

Quite probably, at the time, about thirty workers participated to the construction of the castle, using for the connection of all the quarried blocks between six months to one year, while for the extraction and processing of the blocks themselves, the time needed could vary between one to two years. Thus, the structure could be completed in about two years, quarrying all the materials from the site chosen for the construction of the castle and placing them according to the plan.

The construction work of the castle appears on each side, to have been carried out by two workers groups that starting from the common centre then moved off in opposite directions till they met with the other group at the corners, which had been pre-determined by the use of tools15 and building rules that were needed to skilfully construct the structure. For this reason the sides do not measure up, but are slightly off.

The construction techniques used by the Romans at Qasr Aseikhim are still used in Jordan for the construction of dwellings and Bedouin villages close to the Jordanian desert, which often substitute the typical tent village. The modern construction of the Bedouin villages by use of the pseudo-isodomus technique, as it was taught by the Romans to the indigenous populations of the limes Arabicus in Imperial times, is still carried out with construction material found in situ and with minimal tools. In particular the tools used by these local workers are: the quarry pickaxe,17 the small pick also defined “for better of for worse”,18 the dented hack hammer that has between 12 to 27 teeth,19 the uphand vertical tracing tail

The principle at the base of the choice of this construction style was that of holding up the walls thanks to the weight and reciprocal contrast of the cut blocks installed. While the leaning of the walls was on top of an embankment, specific foundations were created there where the bedrock was not strong enough, or where the incline of the hill required a filling up. Thus, the wall is destined to hold its own weight and to resist lateral solicitations and tractions. For these reasons its thickness has been measured, with our unit of measure, to ca 90 cm, because this was placed according to the vertical gravitation and to the constant push of the ground. The

16 The workers that were used could be the army engineers , or the army could give out by contract the construction of the military settlements to local builders according to necessity and construction schedule. 17 In Arabic bik or q’att’a. 18 In Arabic harara. 19 In Arabic shahuta.

14

Lugli 1957: 100. The tools used to fix the corners where the stake is that with the plan bob alignment allowed the skilful construction of the walls. 15

136

GIUSEPPE CLAUDIO INFRANCA: FROM THE CONSTRUCTIVE TECHNIQUE USED IN THE MILITARY mallet,20 and the uphand horizontal tracing tail mallet,21 the bush hammer,22 the curved mallet23 with various kinds of chisel,24 and the iron mitre square for the checking of the corners.25

where to extinguish the lime, in the case in which it will be needed for specific consolidation interventions in the elevated wall structures. Finally, it is useful to specify that the start of the works will have to coincide with the mildest season avoiding both the hottest period and the seasonal rains that in the desert are particularly strong, even though they only last for a few days in which the temperature lowers noticeably.

For this reason it is easy to find skilled labour for a possible reconstruction of the castle, after carefully studying the stratigraphic units of the collapses, which Qasr Aseikhim suffered through time because of men, who reduced one of the most astonishing cultural resources of the region to ruins.

The reconstruction It must be precised, however, that even though a skilled labour exists, the vigilant presence of an architect/conserver is indispensable, because the attitude of these workers is certainly not inclined to an attentive and respectful conservation. The habit of these workers is that of using any stone for building purposes and thus the original blocks could be overhauled if care is not taken for the precise reconstruction of the fallen blocks with meticulous dedication and on site research, having perfect knowledge of the ancient construction modalities.

Before starting the conservation of the Roman castle of Aseikhim, the various collapses caused by the devastating hand of man will have to be accurately analysed. These collapses produced a conspicuous mass of ruins near the elevations and within the rooms. It will be of fundamental importance to establish, by the stratigraphic analysis of the collapses, the direction and involvement of the different heaps of fallen masonries involved. This will be useful to establish both the modalities for the recomposition of the walls and the dynamics and time lapses that caused the collapse itself.

The conservation building site The erection of the conservation building site at Qasr Aseikhim will be placed close up to the castle, placing the crane on a side of the castle, possibly near an ample space where the basalt blocks may be placed, ready for reappointment and possibly divided by construction characteristics (protrusions, shelves, wall facings, ceilings, beams, arch wedges, and so on).

Thank to the ability and experience that some local workers still posses for the construction of masonries using the pseudo-isodomus Roman manner technique, the conservation intervention here proposed is feasible. In order to facilitate it, it will have to be supported by the presence on site of a crane, with which the reconstruction will be carried out in quite a short time compared with the large mass of big stones that will have to be moved back in place. It is for this reason that it was possible to propose a project for conservative intervention and in particular for the reconstruction of the ancient fallen wall structures, even though their state of abandon is noticeable.

This assemblage of fallen blocks in an area close to the crane will only happen after the various stratigraphic units of the collapses, which the castle suffered during its lifetime, will have been studied. Furthermore, in order to have a building site functional to the needs of the reconstruction conservative intervention here proposed, a storing area for the tools and conservation materials will be needed. Moreover, another one will be needed to store the wood necessary for the safety of the walls structures of the castle, and also for a place where to carry out the necessary carpentry for the conservation (centres, footbridges, supports, etc.). It will also be important to predispose an area for the preparation for the beginning of the conservation works, constituted by a pit26 for the mixing of the earth,27 possibly close to the cistern. It would also be correct to predispose an area for the storage of straw28 and a pit

The studies carried out by us of the Roman castle of Aseikhim have attentively analysed all archaeological remains that are still on site and, in the end, resulted in the filling out of an abacus29 of all the original architectonic elements of the castle. the abacus will become, in the actuation phase of the conservation, the most useful instrument that will allow the realization of the reconstruction intervention. Already in this analytic phase, it was possible to establish that all the architectural elements that composed the castle are to be found on site, and thanks to this extremely important verification, it is possible to proceed to the planning of the reconstruction. One of the first steps will be that of carrying out the stratigraphic analysis of the collapses and thus proceed to free the rooms from the debris placing the retrieved blocks in an area close to that of their replacing. As

20

In Arabic chakung. In Arabic dabura. 22 In Arabic mutaba. 23 In Arabic mitraka. 24 In Arabic tumbor or izmil. 25 In Arabic zanya. 26 In Arabic hufra or bufar, khandaq. 27 In Arabic trap. 28 In Arabic kuta. 21

29

The abacus here following has been geographically rendered by Serena Marino, who has also produced the graphic representations of the relieves and of the project under the guidance of Giuseppe Claudio Infranca, planner put in charge by the Department of Antiquities of Jordan.

137

THE LATE ROMAN ARMY IN THE EAST previously mentioned the work will have to start from the central room to then move towards the corners of each wall facing. This operation will have to be carried out by use of a crane, numbering the blocks according to the stratigraphy of the collapse.

gradient that have been taken care of by inserting a basaltic stone block as a step (in the north-western corner) or a trench foundation with shapeless gravel (central part of the western side). Thus, the original foundations were built by use of rock outcrops and recompacting there where the differences in gradient made impervious the fabric alignment. The foundation, in all its length, has an offset depending from the underlying alignment surface to the first row that is wider than the thickness of the wall facing. Furthermore, it has been noticed that where the foundation was constructed by use of trench and excavation this reaches the most reliable layers resting on the rock both for a greater consistency to the structures and to support the estimated loads.

Considering the large number of collapses to be found in the castle of Aseikhim, the masonry structures are not in such a bad static condition. There is very little cracking in the elevated walls, with the exception of the southwestern corner of the castle. Surely, the exceptional state of the foundations has guaranteed a good stability so much as to be able to support the re-composition forecast by the planned conservative reconstruction. In fact, the destructions of the masonries have been perpetrated, by men, through time and they have only caused the collapse of the areas directly involved in the dismantlement. The breakages in the elevated masonries have been caused by collapses that inclined the summit of the structure, and by the removal of the stones by hand of clandestine people. While the underlying walls and the structures close to the collapses have resisted thanks to the thickness of the walls, and to the stability of the masonry built with great skill and technological perfection.

In the various cases the foundations, in their entirety guarantee, stability to the superior structures to the elevations, also in the hypothesis of a reconstruction of the collapses thanks to the reconstruction conservation intervention here proposed. The conclusion already reached during the first archaeological mission in 200031 by the study of the trial trenches is that the foundations of the castle are in an excellent state both because of their manufacture and their size and stability, thus there is no need to proceed with further trial trenches to check or strengthen the foundations structure. From the verification of the monument’s stability, the reconstructive conservative intervention could be planned.

On the other hand, when the flat ceilings were demolished the underlying masonries behaved in a different way according to the loads and drop dynamics. In particular, the ceilings of the rooms to the corners of the castle features, in the south-eastern side the shattering of the wall apparatus because of the collapse of the ceiling. While on the south-western side the masonries perfectly survived the collapse of what are probably two ceilings.30 Then the condensation of the blocks and the greatness of the loads seriously tried the remaining elevated walls throwing out of balance the wall that faces to the west close to the corner, thus throwing it out of plumb. This serious breakage will have to be either recomposed or blocked, as soon as the room opposite the south-western corner will have been emptied of the mass of construction materials, following the methodology of the analysis of the stratigraphic units of the collapses. Finally, as far as the room to the north-western corner is concerned, the collapse of the ceiling has dismantled the walls, thus reducing the elevation. This corner possibly had two storeys and thus, when it collapsed the large loads supported by these walls greatly damaged all underlying masonries, which had, moreover, already been badly restored in Byzantine times. There are no certainties for the north-eastern room that has been emptied by clandestine people, and the collapsed materials have been scattered around the area.

The wall structures The masonries of Qasr Aseikhim have been built with a double facing using the pseudo-isodomus technique squared with the inner dry wall technique, they are made of regular rows with great use of stone wedges to regularise the facings. The filling in the walls of Aseikhim was done principally with earth and stone splinters with the function of link between the two wall facings. A peculiar characteristic was the care with which the Romans had perfectly placed the inner dry wall that in the end determined the outstanding construction of the wall structure with skilled workmanship. The choice of earth with binding abilities, together with volcanic lava and stone flakes placed as joint in the wall structure had the function of link as it is for the looping elements, that are very frequent in the castle of Aseikhim. The double facing walls of the Roman fortress of Aseikhim have equal thickness, with inner dry wall composed of basalt stone flakes, earth and in some cases lava material. Contrarily on the eastern side of the castle, where some restoration work has been carried out, probably in Byzantine times, different construction techniques have been used such as filling with small pebbles instead of

The foundations The foundations rest directly on the natural basaltic bedrock with the exception of some differences in

31

The trial trench carried out during the 1st archaeological mission in Qasr Aseikhim demonstrated, thanks to a clandestine dig, the consistency of the foundations under the ashlars masonry arch of the central room.

30 The presence of a room on the second storey is supposed for the south western corner.

138

GIUSEPPE CLAUDIO INFRANCA: FROM THE CONSTRUCTIVE TECHNIQUE USED IN THE MILITARY earth without any real joint consistency. Thus, the consolidations and wall reconstructions of Byzantine times are perfectly easy to locate in the castle and as cited by the historians of the time could have caused the first collapses from which the successive vandalisms, on the part of clandestine people looking for hidden treasures, results. All of Aseikhim’s masonries, both in the outer and inner facings, have the same dimensions and the blocks have been summarily squared on the façades, while they are roughly-hewed in wedges towards the inside of the wall to better connect the two facings by being closed up and drowned in earth, basaltic flakes and lava material. Both in the outer and inner facings the stone rows are horizontally aligned with great precision. The ancient builders had taken care to cross the vertical joints to the superior blocks, placed horizontally. In some cases, it has been noticed that the elements of the two wall screens touch, this constitutes a greater guarantee of stability.

from which the threads will be fixed to check their perfect reconstruction. As the ancient builders, the conservers will also have to carry out the reconstruction works by replacing the blocks from the centre towards the corners or those parts of the original walls that are still present in elevation. Thus, the conservation intervention will have to be carried out with the irreplaceable aid of a crane and with two teams composed by two restorers and two master masons the latter supported by auxiliary personnel for the better installation of the heavy fallen blocks. Considering that, in Jordan, competent workers are easy to find it will be useful to have expert personnel that have a great familiarity with the basaltic stone material, to avoid accidents and non-fulfilments during the actuation phases of the restoration. The two teams will have to be placed on both wall facings, inside-outside or inside-inside, then the two workers groups will move towards the corners, or towards the wall fragments still standing. The joint phase of the last block within the row will be very delicate because this is the one that will have to endure the possible transformations that the wall has undergone each time there was a collapse or a reappointment in situ. Another delicate operation will be the restoration of the original masonries in their natural bed, necessary to tally the scarf of the oblong blocks with the squared blocks, defined ‘short’. This restoration will allow to strengthen the whole construction system that in origin was based on this very expedient. The wall will be filled, as it was originally, with volcanic pumice, earth and stones of various dimensions according to the real requirements of the joints that will be constituted one at a time. In some cases, mortar might also be used in concomitance to the joint.

The facing was rigorously flat, as the surface. This was required for a better uniform distribution of the stress. Furthermore, in Qasr Aseikhim, it was possible to see that the blocks were laid upon the same side they rested in the quarry, thus respecting the horizontal orientation of the natural layers. On the other hand, the narrow, long blocks used for the ceilings, the angular and the scarf ashlars were quarried on different quarry layers. The quarry from which they were extracted was to be found near the castle and there is evidence of the extraction activity, which was carried out for the monument (see photo). The walls of the Roman castle develop in a regular fashion alternating to the square rows rectangular ones circa half their size. The portals, on the other hand, were built with larger ashlars while the underlying wall portions were made of oblong blocks prone to discharge the combined compressive and bending stress stresses from the walls. The whole system of apertures turned out to have been built with greater workmanship and care, and the different elements that constituted them are perfectly scarfed between themselves. Particular are some flat arch lintels32 of the castle which possessed a double function of access and slight light source, as there seem to be no windows or light openings from the survey carried out of the remains that have reached us. The system of the portals, other than having walls increased by openings, also had monolithic entrances such as the stones used in the corners of the monument.

A convenient support to work at certain heights will be the three resting points33 trestle used, by the Bedouins populations, as scaffolding for the construction of their dwellings. The use of this trestle is still to this day vastly spread, and consists of three resting points: two on the ground and one to the wall. When the three resting points are stable, it is possible to rest a wood-floor ancillary scaffold on two trestles. This system will present a few limitations during the conservation intervention when in correspondence of the ashlars masonry arch. In order to avoid inconveniences, and for the safety of the workers and of the remains themselves it will be important to place some central supports to the arches (it will be explained in more depth here following) on which it will be possible to rest the above mentioned trestles. In some cases, trestles with four resting points may be used. Another advantage of the three resting points trestle, that it is useful to mention here, is that it is functional to reduce the need for scaffolding allowing a greater

The reconstruction conservation intervention will have to be supported by a constant alignment with the use of compasses and by plan bobbing the wall facings. The realignment of the fallen rows will thus be carried out with a layout constituted by stakes fixed to the masonries 32

33

Peculiar is the lintel of the access to the inner courtyard of the castle.

139

In Arabic hasan.

THE LATE ROMAN ARMY IN THE EAST of small dimension.35 This way, the completion of the support will aid the layout of the room and of the ruins that obstruct it, in order to prepare it for the conservative intervention.

manoeuvrability in the presence of ashlars masonry arches that have survived still standing. The ashlars masonry arches In the castle of Aseikhim, it is obvious that the ashlars masonry arches are the principal holding structure for the flat stone ceilings. They are only scarfed for a few centimetres by use of larger sized ashlars in the inner screen walls. Only in the eastern side of the castle the ashlars masonry arches are also scarfed on the external wall.

The level ceilings of the Roman castle The ceilings of Qasr Aseikhim are level, resting on the perimetrical walls of the rooms on one or more of the ashlars masonry arches according the dimensions of the room. The covering of the castle’s rooms were built with a non-thrusting system resting on the thrusting structures (arches). Thus, the level ceilings were constituted by basaltic stone beams sustained by the ashlars masonry arches on the one side, and on the other they rested on shelves or protrusions, made of more of the same material used for the construction of the castle, and joined on the top of the room wall facings. In some cases traces have been found of two or more shelf or protrusion rows that allowed the oblong beams to cover longer spans maintaining their maximum length. This expedient that the Roman builders adopted from the Mycenaean pseudodomes or possibly from the pseudo-isodomus technique, was useful to guarantee, in the arid regions, where wood was scarce, the construction of great architectures by only using the construction materials that could be found on site, without the burden of transport. In Qasr Aseikhim itself, there is evidence of this absolute autonomy of construction materials all to be found on site.

The ashlars masonry arches both for the completion of the superior structures or for their replacing need wooden or stone central supports. Jordan’s Bedouin populations have also used ashlars masonry arches for their dwellings and thus for the restoration or the reconstruction intervention it will be good to use some of the central supports that they have experimented with,34 which are: - the inverted V central support constituted by wooden elements placed so as to make a triangle, and having for a base the trampling surface of the castle of the adequate dimension of the original size of the arch. The shape of the arch will have to be constructed by use of a resting surface made from a mixture of earth, straw and stone fragments. On this central support will then successively be placed, by use of the dry stone technique, the stone ashlars found in situ at parallel facings and a wedge, starting symmetrically from the ends to the positioning of the keystone. The shoulder of the ashlars masonry arch, in the initial phase of the replacing, has provisional support function for the false-arch; - the wooden buttresses central support made by load bearing beams of conspicuous thickness placed parallel. The above mentioned central support will have a stability increase through a lighter structure of transversal stiffening and of bracing linked with parallel buttresses. The heads of the buttresses will be liked with wooden elements, which sustain the false ground camber hidden in the corner. Lastly, a false wall will be constructed that will be used to hold the laying of the arches’ ashlars. This false wall will have to be built with lime, straw and volcanic tufaceous rocks, to be found on site. It was anticipated that the realization times of the centre for the re-composition of the arch might be of three days for the support structure and two to three weeks for the reconstruction of the whole arch. Finally, a week to remove the support after the intervention; - The stone wall central support . This support could be useful in the case that the arch is elevated and has to be secured for the replacing of the top floor. The necessary condition will have to be that, within the room, the ruins caused by the collapse be of little quantity and the stones

34

Upon the level ceilings of the castle of Aseikhim there was a beaten-earth surface mixed with small aggregates from quarry and processing debris that was condensed with care. This beaten earth surface is still to this day visible in the two covered rooms that are to be found over the main entrance and the central room of the northern side. In the case in which, during the conservative intervention, some oblong shaped architectonic elements of the castle should irreparably brake, they might be conserved in the same way as sculptures. On the other hand, if many blocks belonging to the ceiling were to be lost the cover could be realized by substituting the basaltic stone blocks with wooden beams, the reversibility of which is well known in the practice of archaeological restoration. The wooden beams, in substitution of the original covering elements could coincide with the use of the arboreal essences of juniper, poplar and oak, as they often happens to be used by the Bedouin populations for the construction of their dwelling’s coverings, for waterproofing. In the case of the use of wood to cover some of the level ceilings of the castle of Aseikhim, it will also be convenient to provide a wood floor above the beams to then house the above described untilled earth upon which to place a mixture of loam earth and crushed straw. This last element will be prepared directly on top 35 A similar case is that of the central room of the northern section of the castle.

Marino - Lodino 1999.

140

GIUSEPPE CLAUDIO INFRANCA: FROM THE CONSTRUCTIVE TECHNIQUE USED IN THE MILITARY of the wooden covering, as the Bedouin populations are used to doing for insulation and waterproofing. Once the restoration of the castle will be complete, it will be correct to proceed to create the wall crest to close the wall apparatuses so that the deteriorating factors, that are very intense in the Jordanian desert, would not assail the monument. Furthermore, the patrol communication trench or the stone pavement ceilings of the castle will have to be made serviceable. The safety of the visitors will be guaranteed with a simple wooden barrier, that because of its reversibility would not impugn the original arrangement of the castle top, to this day unknown, and that it is hoped will be clarified during the restoration works of Qasr Aseikhim. The architectural coverings When the intervention will be completed, it will be wise to restore both the beaten earth coverings and pavements using a tool made of a flat stone and fitted with a wooden handle. During the first archaeological mission in Aseikhim the beaten earth pavement of the inner courtyard was found to be constituted by a lime untilled earth and an over layer of very fine earth. The lime in this case had been used with the function of sealing against water infiltrations. It will be necessary to predispose a lime plaster coating, of very fine aggregates and earth, in the inner rooms of the castle to guarantee the minimum standards for the liveability of the monument, that will have to be recogitated according to both its use and the new functions it might have after the restoration. Bibliography Amiran D.H.K. 1950 A Revised Earthquake Catalogue for Palestine’. In Israel Exploration Journal 1: 223246. Bowersock G.W. 1983 Roman Arabia. Cambridge Mass. Lugli G. 1957 La tecnica edilizia Romana, vol. I, Roma. Marino L. - Lodino M. 1999 La casa tradizionale nei villaggi della Giordania, Verona. Ravegnani G. 1983 Castelli e città fortificate nel VI secolo, Ravenna.

141

Military Architecture of the Limes Arabicus: Strategy, Commercial Routes, Quotidian Life Maysoun Al-Khouri Director of Istituto Superiore per le Tecniche di Conservazione dei Beni Culturali e dell’Ambiente “A. De Stefano”, Lungotevere Pietra Papa N. 61, 00146 - Roma. Tel. 06 55301117

We started our study of the limes arabicus from Qasr Usaykhim (Fig. 1) after undertaking five archaeological missions to the site of wadi Usaykhim.1 (Fig. 2)

headquarters was in Bosra, while the main settlements reached by the Roman road built by Tajan were Filadelfia (Amman), Gerasa (Jerash), Petra and Aqaba.

It was remarked that, after Rome strengthened its power on the eastern front, the desert became a geographical component that acted as a buffer zone between the bellicose oriental populations and the coastal area that comprised Palestine, Syria and Phoenicia. At the centre of the Arabic desert, there were the green valleys of the Tigris and Euphrates that run towards the Indian Ocean.

Along this defensive system, a great concentration of legions was gathered. Fortifications, cities and towers were built in clusters with every now and again a large open spaces which gave the impression of freedom; but just there the checks were more rigid and the Roman troops better disengaged to launch lethal attacks in accordance to the in depth strategy they professed.

The Roman army hypothesised several scenarios of likely military invasions from the east that were reduced to a sole possible direction for the assault, the southeast.

The limes includes large civilian areas, where there was a direct involvement of the populations. This way, in time, some of the Roman forts became wonderful towns. This urbanisation process was strongly facilitated by the very military authorities that sought a live limes arabicus full of social and commercial activities, which were extremely important on the eastern front.

At this point, the desert was fortified so that the enemy would always show the left side, as all other directions had been covered by the Roman military installations. For a time Palmyra actively collaborated to this project by controlling the commercial and military exchanges of the eastern front. Later, the rebellion in Rome and its destruction, placed the Roman troops in great difficulty; they managed to fill the space left by the caravan city. Trajan, in view of this project, had already annexed the Nabatean reign, reducing it to a Roman province called Arabia. He, however, did not complete the conquest of Mesopotamia, other fundamental region of the eastern game board.

The limes arabicus was vastly strengthened by the Emperor Septimius Severus both with the annex of Nisibi and Singara, thus taking back the frontier on the High Tigris, and with the construction of the first castles in the region of modern Azraq, the completion of which was carried out between the end of the second century AD and the first part of the third century AD, to be definitely completed in the fourth century AD. The construction of the Strata Diocletiana that had been contemplated as a new alignment, to the east of the via Nova Traiana, defined the oriental game board.

The limes arabicus (Fig. 3) was thus modulated upon military settlements both fixed and movable that together with the pre-existing cities and those of new foundation created a borderline area essential for the security of the eastern front.

The Strata Diocletiana allowed a larger manoeuvre margin to the troops and an increased steadiness to the fortifications, which had been founded to respond to the attacks of the Sassanid Persians’ new monarchy, who had replaced the Arsacid Parts, who had had more peaceful traditions, with a cruel revolution between 212 and 227 AD.

The oriental borderline organization protected the entire inner route system of communication functional to commerce, movement of troops and rigid administrative and customs control carried out by the army of Rome. For this reason in a large area between the Mediterranean, the Red sea and the plateau near Petra a series of movable and fixed settlements were built. The remaining part of the limes arabicus was linked with the neighbouring Nabatean Reign that, from Petra, reached Bosra with the via Nova Traiana that parallelly run as a divide. The

The Persian’s new nationalistic power was soon revealed to be fanatic as far as religion was concerned, and to be militarily untameable. Soon the Roman Empire was faced with the “crisis of the third century” because of the converging aggressions of the Germans on the northeastern border.

1

Qasr Usaykhim is located on the north eastern side of Jordan near Qasr Azraq. Five archaeological excavations conducted there by the Istituto Superiore per le Tecniche di Conservazione dei Beni Culturali e dell’Ambiente “A. De Stefano” – Italy with the Department of Antiquities of Jordan.

During this tormented historical period a new great danger to the Roman Empire was constituted by the nomads, who in small disorganised groups created a state

143

THE LATE ROMAN ARMY IN THE EAST

Fig. 1: The remains of Qasr Usykhim

Fig. 2: The site of Wadi Usaykhim

144

MAYSOUN AL-KHOURI: MILITARY ARCHITECTURE OF THE LIMES ARABICUS

Fig. 3: Limes arabicus sites and roads

145

THE LATE ROMAN ARMY IN THE EAST

Fig. 4: Pseudo-isodomo technique

146

MAYSOUN AL-KHOURI: MILITARY ARCHITECTURE OF THE LIMES ARABICUS

many types of fortifications, also comprising the single defensive wall.

of continual chaos along the limes arabicus, situation that added new difficulties to those the Roman army was already faced with.

The limes arabicus was not a military area uprooted from the other regions of the empire, but was characterised, apart from the military settlements, by urban settlements and villages, which rendered the neighbouring country live and prolific with initiatives and economical, commercial, and social activities that were encouraged by the army itself.

The limes arabicus was thus re-designed. Along the principal routes the “punctuated defences” were built; these were large camps or castra, each of these camps housed a Roman legion with victuals, supplies and weapon factories; and tens of castella. At this point, it must be recalled that the Romans, though they fortified their strongholds, were never ones to conduct a defence war, on the contrary the attitude in war was one of attack. Rare are the occasions when they used pitched fortifications, and never as defence posts.

In the agricultural estates the productions where of cereal, the growth of which was encouraged by the environment, integrated by vineyards and olive cultivations, and on occasion the rearing of cattle and sheep. All of these rural activities are attested in the beautiful ancient mosaics that today enrich both Jordan and Syria. The great nomadic populations that lived beyond the limes arabicus favoured animal husbandry to agriculture. The dromedary, for example, was never systematically used by the nomads until recent times, so much so that there still are numerous herds of wild dromedaries to be found. On the other hand, the semi-nomads, who lived on both sides of the Roman border, added to the cultivation of agricultural products the herding of both sheep and goat.

The military arms that regulated this activity was that of the Engineers (sappers), that showed how the Roman army was dedicated to living in complete autarchy. This service was precisely structured, and controlled with what can be considered true skill. Firstly, the sappers took care to guarantee adequate shelter and safety to the troops during military reconnaissance missions with the creation of a camp. Soon the advance of the sappers, in Roman army life, became of great strategic, territorial, political, economic and social importance.

The Roman army imposes itself in the limes arabicus as a tool for economic development, while to fully carry out the role it had at its disposal large materials and human resources with a heavy expense of financial means.2 The army of Rome tried to get a return from these expenditures by fully entering the commercial trades that happened on a daily basis in the limes arabicus, but never managed to enter it completely.

The man in charge of defining the collocation of an army camp during a patrol was the metator who had the task of finding a suitable area where the troops, tired from their daily advancement in enemy territory, could find rest. The librator, on the other hand, had the task of making sure of the horizontality of the levelling of the camp; he also gave advice on the construction of canals or helped the artillery in the placement of the war machines. That of the mensor was the other important role, he was the one to decide the disposition of the rooms, who delimitated the surfaces and the areas intended for the soldiers, furthermore, he often replaced the architect. He was the one who would define and carry out the construction of the military settlements, and of the war machines.

In fact, the limes arabicus, as an area densely populated by the army, also guaranteed the protection of the civilian community that inhabited the neighbouring towns, which were equipped by the Roman army with towers for sighting, of amazing means of communication to gather the troops in the event of an enemy attack, and, of course, the extraordinary road system. The limes arabicus surely was one of the places where there were the better road systems because these had been planned and built for the transit of precious goods from the East, which for centuries enriched Rome. Moreover, it was a land where the simple existence of urban social life was enough to demonstrate the efficiency of the safety measures imposed by the Roman army, as it was essential to the survival of those settlements from the Persian armies and the hoards of nomads from the desert. It would not have happened unless the area had been fitted with a defence system and great arms contingents.3

Within the settlements, apart from the standard procedures of the army system, there also was a workshop, called fabrica, that simultaneously produced arms, armour and construction materials for the fortified buildings. The fabrica was under the direct control of the magister, who was in turn helped by both an optio and a doctor. Often in the limes arabicus the position of the fabrica coincided with the presence of a quarry, and always near the settlement to be built. The picking of the area where the Roman soldier could carry out both his official and his non-official duties was always made so as to render life as easy as possible, after all, the Roman world had them all used to such comforts.

The role of the Roman soldier actually was extraordinary. The soldiers were used, apart from marching over long distances, to build, during the daily advancements, roads,

The Roman sappers destined to the limes arabicus were, thus, the ones to build their monumental, these grouped

2 3

147

Le Bohec 1992: 302. Luttwark 1997:110.

THE LATE ROMAN ARMY IN THE EAST bridges and encampments all of which were necessary to guarantee maximum security. Thus, one of the places where the Roman soldier could better move, and show he’s valour as conqueror to its highest degree, was the construction of the roads, primarily to be carried out within enemy lines or in their vicinity, along all frontiers.

Arabs, deployed in the blockhouses of small dimensions, guaranteed to the borderline system the maximum control of the tactic and strategic dynamics that allowed the eastern border to last longer than any other border of the Roman Empire. It has been observed that some fortifications were set up in a drawn back position to the line of communication of the limes arabicus.

As far as these roads are concerned, one must not imagine a slabbed road, but a flat layer of beaten earth, where competently displayed signs indicated the direction to follow in the arid savannah or the Arab desert. The roads, at the time were only laid with gravel in the proximity of the borderline cities sought by the Roman army that controlled its urban expansion.

The limes arabicus was further completed by a system of towers to guard and liaison, and to defend the water systems, it was placed close to the enemy lines. This system of towers was an open barrier that signalled the enemies’ movements.

The principal roads of the limes arabicus, as we have had the opportunity to mention, were initially the via Nova Traiana and later the Strada Diocletiana. From these great arteries the secondary roads unwound, which linked the military settlements distributed over the borderlands, to the towns, the towers, and the principal road crossings of the great commercial routes from the east to the Mediterranean coasts.

This new strategic dislocation sought by Septimius Severus replaced the pre-existing from the second century AD that had develop on the concept of “forward defence”, having a buffer zone constituted by some vassal reigns or countries at the same time independent and refractory. Coming back to the limes arabicus we here describe the typologies of the military installations created between the third and fourth century AD (Fig. 5):

The great roads built along the limes arabicus were fortified and under strict control of the army. Thus, while these run parallel to the border, the other secondary routes of the limes stretched straight across the border and beyond. The key role they played in the in the composition of the limes arabicus was unavoidable. The defence structures of the fortified routes were constructed in a different way according to the topographic conditions or the military tactics: with walls, palisades, and ditches; apart from the fixed settlements that were the castella, forts, and blockhouses that were conveniently connected, as was the tower system that tightly controlled the borderline structure.4

1. The castra and the large castella that housed first the legions of Rome and those of Byzantium after the fall of the Western Empire: The castra housed the entire Roman legion. Up to now only two of these great military settlements have been found, even though it is know from historical sources of other sites which would have housed other legions with the same characteristics, as in the case of Aquaba with the X Fretentis. The two castra mentioned are those of elLejjun, which has an extension of 5.50 ha, and of Udruh with an extension of 6.00 ha, both have a rectangular conformation.

Along these arteries the engineers of the Roman army upon reaching a watercourse or a wadi proceeded with the construction of a bridge that, because of the lack of wood in the Roman province of Arabia, were preferably built with stone utilizing the pseudo-isodomus technique. The use of the polygonal structure for the construction of the walls and the bridges of the via Nova Traiana probably was propaedeutic to the pseudo-isodomus technique used for the construction of the fortifications and towers of the limes arabicus, as the lapse between one construction technique and the other is quite brief.

The so-called large castella which housed a whole legion or a large part of one, they were nothing but castra that were a military variant because of their conformation in the shape of a castella. The dimension of the large castella vary between 1 to 2 ha, they come in two versions according to their planimetry: a) quadrangular, Umm as Rasas and Da’janlya; b) rectangular, al-Hadid and Humayma.

The large castra or the large castella, focal points of the borderline, were surrounded by the other less important fortifications where the Roman troops were deployed to protect the limes arabicus.

Amongst these large castella there is that of Umm elJimal that varies slightly from the quadrangular form because of orographic reasons, but must be assimilated to this typology.

From the analysis carried out along the limes arabicus, it appeared that the auxiliary troops, mainly composed by

4

Causetti 1939: 43.

148

MAYSOUN AL-KHOURI: MILITARY ARCHITECTURE OF THE LIMES ARABICUS

Fig. 5: Typology of military installations on Limes arabicus constituted the garrison (presidium) of the castellum. The castellum was in fact a military community that did not have the status of colony or municipium. To the quadrangular typology belonged the castles of Umm elWalid, Qasr el-Baij, and Jawa, on the other hand those of Khirbet el-Qirana, el-Quweira, Qasr eth-Thuraiya, Qasr el Hallabat, Khan ez-Zafaran and Muhattet el Haj, present a number of towers for sighting and defence. Finally, the castles with trapezoidal shape are: Qasr Uweinid, Umm el Jimal, and Qasr el-Kithara. The dimensions of these castles vary between 1500 and 2000 m2.

2. The castella where the Roman troops were permanently based: The castella housed some troops of the Roman legions and as the large castella had a quadrangular or rectangular shape. They differed only by the dimension that varied between 3500 and 6200 m2. The known castella of the limes arabicus are: Qasr Azraq, Deir el Kahf, Khirbet el-Fityan. 3. Cohort castella in which it is presumed that both Roman and auxiliary troops were housed:

4. The blockhouses exclusively for auxiliary troops:

The cohort castella5 applied the length of ca 40m to the sides of the square, measure that corresponded to the width of the front of a cohort. A cohort was a unit that 5

The blockhouses housed the auxiliary troops and their dimensions were always inferior to 1000 m2.

Causetti 1939: 12.

149

THE LATE ROMAN ARMY IN THE EAST different from those proposed by some archaeologists who had carried out the initial studies on the subject. The divisions made by the archaeologists only took into account some formal aspects that have no extrinsic reason for being in common with the true peculiarities of the military settlements of the limes arabicus. It is for this reason that strong doubts arose both at a territorial and architectural levels of the original tactic dislocation of the military settlements.

The blockhouses were: Qasr Aseikhim (Fig. 6), Qasr el‘Al that has a quadrangular shape, Khirbet ez-Zona also quadrangular but with towers in the corners. Rujm Beni Yasser was of trapezoidal shape. A separate matter is the case of Rujim es-Sadaqa. 5. The turris or the burgus: The Turris positioned along the limes arabicus were observation posts, for liaison and transmission of fire or smoke signals, for the defence of water sources and advanced defence. As far as their architectonic forms are concerned they were built with a quadrangular or circular plan. Along the limes arabicus the towers that have been found are the exact same number of the ones that were once positioned in front of the enemy lines. Proof of this is the discovery, made during the surveys carried out of the territory around Qasr Aseikhim, of two towers belonging to the limes arabicus. One has a rounded shape, it garrisoned a water source; the other, with a quadrangular plan, was used for signalling. These new discoveries, which were made during the archaeological Missions of the Istituto Superiore per le tecniche di conservazione dei Beni culturali e dell’ambiente “A. De Stefano” are the sign that other surveys must be carried out to complete the exploration of the limes arabicus.

The archaeological discoveries along the limes arabicus to this day have not yet given a comprehensive picture of the military architectures built by the Romans. This also because no fortifications belonging to either Parts nor Persian, who the historical sources mentioned had opposed themselves to the Roman limes, have been found. It has been remarked that superficial surveys have been favoured to accurate and punctual archaeological research. Furthermore, it has been disclosed that during the restorations of some of these ancient masonries easy integrations have been used, irreversibly polluting the originality of the monuments with modern materials impugning a future reading of the monuments themselves. The new subdivision, is exclusively linked to the original aspects of the functions, typologies and architectural volumeters, to allow the reading of the tactic dislocation that the Roman army had been able to impose in many centuries of domination over the borderlands of the limes arabicus.

The burgus, on the other hand, were created separately from the military plan, designated for the enforcement of public law, this was a village with an embryo of municipality. However, this small civilian community had small pertinence with the limes arabicus, because no trace of any such thing has been found. On the other hand, there are some examples of burgus speculatorius, which was a small stationing, that ensured a street-police service and/or vigilance of the public order in the proximity of the villages of the limes arabicus.

It is from these considerations drawn form the subdivision of the military settlements of the limes arabicus that we learned with wonder that the mix of simplicity and technical preparation was always at the base of every military structure built by the Romans. The perimeter wall, for example, constitutes the most important part of the fortification. This was funded on the model of the plain fortresses that had preceded Roman architecture. The choice for the pseudo-isodomus technique for these constructions is not casual.

Finally, some examples of station have also been found in the limes arabicus, these were military settlements made up of a few graduates or war veterans that carried out the task of fighting brigandage, very frequent in the oriental borderlands.

As far as the planimetry is concerned, the enclosure usually was of quadrangular shape and, at times, rectangular, often with rounder corners. The mathematical proportions at the base of the military settlements of rectangular shape were almost constantly between 3:2 and 5:4. In the fortifications, the interruptions of the enclosures for the access doors were always minimal, as these are weak points in case of attack. On the other hand, in the large military settlements where the legions of Rome were dislocated the doors were different in order to show the courage of the Roman soldier, even though the Roman architects created towers of different manufacture and dimensions in order to protect them.

This subdivision does not only concern the military aspects and the architectonic dimension, but also matters concerning the strategic dislocation in the territory, as the schematism here introduced is only used to clarify and understand the limes arabicus in its original conception. It is well known that the evolution and rhythm of the huge Roman frontier was not of a single kind, but that it was quite flexible in the different regions in which it had been promoted and where it had developed. To this purpose, we have studied this classification recurring to the volumetric, typological and functional architectonic aspects. Thus, this new classification of the military installations of the limes arabicus is quite

150

MAYSOUN AL-KHOURI: MILITARY ARCHITECTURE OF THE LIMES ARABICUS

Fig. 6: Plan of Qasr Usaykhim Luttwak E.N. 1997 La grande strategia dell’impero romano, Milano. Parker S.T. Romans and Saracens, Winona Lake 1986. Parker S.T. (ed.) 1987, The Roman Frontier in Central Jordan, Oxford 1987.

To that purpose, the towers, together with the ramparts, were the principal elements of the settled defence where the various war machines for the defence of the military settlements of the limes arabicus were installed. Bibliography Al – Khouri M. (ed.) 2003 Il limes arabicus, Roma. Causetti E. 1939 Fortificazioni e macchine belliche. In Civiltà romana 11. Isaac B. 1990 The Limits of Empire. The Roman Army in the East, Oxford. Kennedy D. 1982 Archaeological Exploration on the Roman Frontier in North-East Jordan, Oxford. Le Bohec Y. 1992 L’esercito romano, Roma.

151

Economic Conditions, Security Problems and the Deployment of the Army in Later Roman Palestine. Part I: Economy and Population Israel Shatzman The Hebrew University of Jerusalem The present paper is the first part of a study that aims to review the economic conditions prevailing in Palestine in the Later Roman Empire,1 to examine the security problems of that region, as known from the available written sources and can be learnt from physical remains, to present and discuss the deployment and installations of the Roman army units and, finally, to analyse the correlation between these three factors. More specifically, the questions probed in this study are the following: if Palestine prospered economically and demographically in the Later Roman Empire, is it an indication that the available sources tend to magnify the gravity of the security problems, thus distorting the true, real conditions under which people worked, travelled, traded, manufactured and marketed their products, managed their social activities, in short lived? Or, rather, should that prosperity be construed as a result of the success of the Roman authorities to maintain law and order internally and to protect the country from external threats and inroads? Needless to say, that could be only one factor, presumably necessary but not sufficient, among others that contributed to a flourishing and viable economy. If, conversely, it can be shown that fluctuation, or even continued deterioration over a long time, combined with shrinking of the population, marked the Palestinian economy, should that be regarded as a vindication of the reliability of those sources that are interpreted to indicate the prevailing of formidable problems of insecurity, as well as a proof of the failure of the army to cope with those problems? Another pertinent and baffling question is whether the Roman authorities were more concerned with the danger of inroads and invasions posed by external enemies rather than with internal factors that threatened to disrupt and cause havoc to the peaceful economic and social activities of the population of Palestine.2

The topics mentioned above are not new and have long been discussed and debated by not a few scholars. My justification for presenting yet another study of the subject is threefold. First, in my opinion erroneous conceptions and misguided approaches to the available evidence, both literary and archaeological, linger on and bedevil many publications, old and most recent, dealing with Later Roman Palestine. The term limes Palaestinae is a good example: not only its nature and basic characteristics are still disputed, but also not a few scholars ignore the problems and use it in a simplistic manner, thus endowing it with an arbitrary meaning; in consequence wrong analyses and conclusions about the Roman government’s approach to and perception of the economic conditions, security problems and military situation in Palestine are met in their publications. Second, the limitations and diverse nature of the sources of evidence are often not recognized and taken in consideration. Third, the need to examine the economic conditions, to the extent that they are known, vis à vis the security problems and the deployment of the army units and vice versa, is usually neglected. The treatment of these problems and topics in this study is not meant to be and cannot be exhaustive but is detailed enough to present my approach to and interpretation of them. Introductory Remarks and Observations

1 Chronologically the terms Later Roman Palestine and Later Roman Empire are used in this paper to refer to the period from Diocletian to the early 7th century. Many scholars call it, sometimes starting with Constantine’s defeat of Licinius and foundation of Constantinopole (324) or his death (337), Byzantine, or early Byzantine period. For Israeli archaeologists, the Roman period in Palestine, or Eretz Israel for that matter, extends from 37 (or 63) BCE to 324 CE, subdivided into Early Roman period (37 BCE – 132 CE) and Late Roman period (132 CE – 324 CE); the Byzantine period spans the years 324 CE to 638 CE, that is, to the Arab conquest. See, e.g., Stern 1993: 4, 1529; Tsafrir 1996: 280 n.5. In referring to such defined periods, they are given in quotation marks. 2 The term Palestine needs clarification for its meaning has changed over time and, for various reasons, can mean different things as used by different people. The considerations involved in the preparation of the maps of Roman Palestine and Judaea - Palaestina published, respectively, by Avi-Yonah (1940: 1-5) and Tsafrir, Di Segni and Green

153

It is widely held that the economy and population of Palestine reached their peak in the Later Roman Empire; that was the most prosperous period of Palestine before modern times.3 The demographic growth and economic prosperity of the country is all the more remarkable and striking if it is true, as not a few scholars have argued, that Palestine witnessed a widespread and profound crisis in the 3rd c., political, economic and social.4 According to these scholars, Palestine, like other provinces, was severely affected by the constant wars conducted by and the recurrent defeats of the Roman armies in their (1994: vii-ix) may serve as good examples. For the purpose of this study Palestine is mainly used to refer to the territories included in the provinces Palaestina Prima, Palaestina Secunda and Palaestina Tertia, for those territories formed one military command, probably as early as Diocletian. However, for obvious reasons, the discussion, particularly of the economic conditions, is not confined to those territories, and thus includes, for example, the region of western Galilee and the northern Golan that belonged to the province of Phoenicia; cf. Dauphin 1998: 3940. 3 See, e.g., Avi-Yonah 1958; Broshi 1979; 2001; Dan 1985: 293-318; Safrai 1994: 436-458; Tsafrir 1996; Hirschfeld 1997; Bar 2004b. 4 See, e.g., Avi-Yonah 1976b: 89-108; Levine 1982; Sperber 1991: 178179; Schäfer 1997: 170-175.

THE LATE ROMAN ARMY IN THE EAST engagements against foreign enemies, the Germans in Europe and the Sassanians in the East, who invaded the empire time and again, as well as by the collapse of the central Roman government, characterized by incessant rebellions of generals and rapid turnover of emperors. In consequence, the population of Palestine suffered from forced labour, requisitioning of goods and livestock, billeting of soldiers in towns, heavy taxation, inflation and return to exchange in kind, which resulted in reduction in the number of settlements, abandonment of land, migration of rural inhabitants into the cities without being able to find employment, a drop in food production, widespread poverty and demographic decline.

increasing debasement of the coinage and inflation in the third century,10 such requisitioning had destructive potential for many,11 both those who found their living in agriculture (particularly small farmers and tenants) and trade and industry. Now, the reduction of deductions from the salary of soldiers to cover expenses for victuals had perhaps started as early as Hadrian and it is doubtful if such stoppages were not cancelled altogether under Septimius Severus.12 According to one view, the latter instituted a tax in kind for the purpose of supplying the military subsistence allowance, the so-called annona militaris, which could be quite plausible given the supposed cancellation of the stoppages and the growing inflation.13 However, the available evidence rather indicates, as several scholars have persuasively argued, that it was part of the regular annona that was given over to the soldiers, and it is also well to bear in mind that the state habitually resorted to compulsory supplementary supplies whenever the need arose; these exactions were increasingly practised during the 3rd century and were the origin of the annona militaris of the 4th century.14 Whatever be the right interpretation, the end result for those who were required to deliver the annona was the same. Once the state stopped deducting from the salary of the soldiers the value of the victuals supplied to them and, in addition, began to pay part of the salary in kind, it had to requisition the food in order not to incur extra expenditure. One way or another, the civilian producers will have suffered losses.

Things, however, are not as simple and clear as that. To begin with, the 3rd c. crisis. In a recent article D. Bar has denied the very existence of such a crisis in Palestine; “on the contrary,” he claims, “this general period saw a growth in population and in the number of settlements and a certain economic prosperity.”5 According to Bar, that prosperity had started earlier, when the Jewish population resigned itself to the Roman rule, following the failure of the Bar Kokhba revolt, and continued uninterruptedly through the 6th c. the main reason for this “unique phenomenon” is related to the fact that Palestine enjoyed a period of nearly 500 years “of calm, in which wars, plagues, and earthquakes were scarce”.6 Given these two contradictory views, a short discussion of the relevant evidence and the arguments raised is called for.7 No major external invasions or fighting between contenders for the imperial throne took place in Palestine in the 3rd c., except the conquest of the country by Zabdas, Zenobia’s general, in 270. Because of lack of sources, it cannot be determined whether or not the short Palmyrene rule and the regaining of the control of the province by Aurelian caused extensive damages, economic, financial or otherwise.8 However, the recurrent wars in the East in the 3rd c., that is, those of Caracalla, Alexander Severus, Gordian III, Philip the Arab, Valerian, Aurelian and Carus – preceded by the civil war between Pescennius Niger and Septimius Severus and the latter’s campaigns against Parthia and Hatra in Mesopotamia in the 190’s – might have strained the resources of the population of Palestine; it too, and not only that of the provinces directly involved, was presumably required to make contributions to the logistical needs of the army during those wars. Given the kinds of foodstuff, livestock and commodities needed, as well as the practice of levying taxes in kind9 and the ever-

The above are general considerations, but one would like to have more specific evidence for concluding that Palestine was indeed affected by those factors in the 3rd century, and there is some such evidence. That the Roman army of Judaea/Palestina was called upon to send forces to participate in the wars Rome conducted outside the province is attested as early as Trajan, when a detachment (vexillatio) of the Legio X Fretensis took part in the Parthian war.15 A detachment of the Legio VI Ferrata served in in Numidia in the mid- 140’s,16 and a detachment of the Legio X Fretensis took part in the wars of Marcus Aurelius against the Quadi and the

5

Bar 2002, cited at p. 43. Bar 2004b, cited at p.7. 7 For the traditional view of the crisis of the 3rd century see, e.g., Rostovtzeff 1957: 393-501 (with a disputable interpretation of the causes); Alföldi 1985: 157-185. For a short account of the crisis which suggests that it was less severe than has been thought see Cameron 1993: 2-11. 8 On the Palmyrene conquest and Aurelian’s war against Zenobia see Watson 1999: 59-82. 9 On taxation in kind see, e.g., Neesen 1980: 104-116; Brunt 1988: 324325, 531-532; Duncan-Jones 1990: 187-198. According to Hopkins 6

154

(2000: 255-257), most taxes raised in kind, as well as the bulk of the rents of rich aristocrats, were transformed into money in the 1st and 2nd centuries CE. 10 See Callu 1969 and 1975; Jones 1974: 187-227; Crawford 1975 and 1978; Corbier 1978. 11 On requisitioning of corn, livestock, goods and transport etc. see Van Berchem 1937 164-187; Rostovtzeff 1957: 423-424 and 721-723 nn. 4546; Mitchell 1976. 12 Speidel 1992: 97-98; Neesen 1980: 109, 248 n.6; Brunt 1988: 209210. 13 Van Berchem 1937 and 1978; Develin 1971. 14 See Rostovtzeff 1957: 484 and 743 n.42; Millar 1964: 152 n.2; Carrier 1978 and 1986: 462-463; Corbier 1978: 294-296; Neesen 1980: 111, 248 n.6, 249 f. nn 5-6; Brunt 1988: 344. 15 Dessau, ILS 2727. It is extremely unlikely that the entire legion served in that war as is implied in Dabrowa 1993: 15, but see Dabrowa 2000: 320. 16 Saxer 1967: 30-31.

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY Marcommani.17 A detachment of the Legio X Fretensis probably took part in the Parthian war of Caracalla, and years later a detachment of the same legion participated in the wars in Gaul under Gallienus.18 A special force of the army of Palestine, armed with maces and clubs in addition to their other armour, fought with Aurelian against the Palmyrenian army in the battle at Emesa.19 Septimius Severus stayed for a while in Palestine in 199 after the failure of the 2nd siege of Hatra,20 and Caracalla surely passed through the province on his visit to Alexandria in 215, and again when he returned for the Parthian war.21 After the final subjugation of Palmyra in 272 Aurelian passed no doubt through Palestine on his way to Egypt to quel the Alexandrians.22 Diocletian stayed at Tiberias for about 2 months in the summer of 286.23 Finally, there is a more specific piece of evidence. An inscription that records the career of Timesitheus, mainly known as the Praefectus Praetorio and father-inlaw of Gordian III, mentions that, while serving as procurator of the province Syria Palaestina, one of his tasks was to collect the outstanding annonae requisitioned in connection with a sacra expeditio, which must be understood as a reference to the Persian war of Alexander Severus.24 This particular evidence confirms the above mentioned supposition that the population of Palestine, too, was required to make contributions to the logistical needs of the army during the Persian wars of the 3rd century. As said before, the logistical needs included not only various items of foodstuff, notably wheat and barley, but also livestock and other means of transportation and various goods, and not infrequently people were forced to transport the requisitioned materials and livestock themselves (prosecutio annonae); this could be ruinous.25 How much onerous these requisitions and the army movements and the visits of the emperors were one cannot assess, but surely they must have disrupted to some extent the economic life of the country.

connection with the annona (arnona in Jewish sources);26 debasement of coins (also called inflation) with which was associated rising of prices;27 forced labour (angaria in Jewish sources), which included requisition of transport;28 billeting;29 soil exhaustion, failure of crops, abandonment of and flight from the land, which, with other factors such as over-taxation, brought about a severe decline of the Palestinian agriculture;30 the country was also afflicted by drought, famine and pestilence.31 The trouble with the Talmudic sources, early or late does not make difference, is that they usually reflect subjective opinions, nostalgic views and dogmatic or schematic notions, are prone to exaggeration and imprecision, and have come down to us in manuscripts which go back not to the original sayings and accounts but to late redactions.32 Three examples will suffice to illustrate the problematical nature of such stories for the extraction of precise, reliable information: Talmud Yerushalmi, Pe’a 7.3. Said R. Yohanan: The late fruits (i.e. the worst of them) that we used to eat in our youth were better than the peach that we would eat in our old age. For in his day (i.e. Yohanan) the world was changed (i.e. for the worst). Talmud Yerushalmi Demai 1.1. Said R. La: In the beginning grapes were plentiful and the grape-shells were not considered of importance. However, now the grapes are not plentiful, grape-shells are considered of importance.33 Talmud Yerushalmi Ta’anit 3.4. There was a pestilence in Sepphoris. [But the pestilence] did not enter the market-place, for R. Hanina dwelt in it. And the Sepphoreans [murmered] saying: what is this old man [doing] amongst us [living] in health, he and his neighbours, while the city [‘s condition] is going and worsening…One time they had to make a fast, and the

Sayings of Jewish Sages and accounts in rabbinic or Talmudic sources provide varied stories and rulings that stemmed from or were connected with conditions and developments in Palestine under Roman rule. As interpreted by many scholars, the relevant evidence culled from these sources seemingly demonstrate the prevailing of the phenomena characteristic of the general crisis of the 3rd c.: ruinous taxation, notably in 17

26

Dabrowa 1993: 17 and 2000: 321. Dabrowa 1993: 18 and 2000: 321. Zosimus 1.52. 20 Dio 75.13,1; SHA Severus 17.1. 21 Dio 77.22,1 – 23,4. 22 Zosimus 1.61. 23 Barnes 1982: 50-51. 24 ILS 1330: proc. prov. Syriae Palaestinae ibi exactori reliquor. annon. sacrae expeditionis. On C. Furius Sabinius Aquila Timesitheus see A. Stein in PIR III: 228-229 (no.581). 25 Cf. Cod.Theod.viii.5.7: Paravedorum exactio patrimonia multorum evertit. Although of a later date (354), the evidence is valid for earlier times as well. See Rostovtzeff 1957: 484-485 and 743-745 nn. 44-45; Mitchell 1976 (with earlier literature). 18 19

See Avi-Yonah 1976b: 95-102; Hadas-Lebel 1984; Isaac 1990: 282291. 27 See Avi-Yonah 1976b: 102-104; Sperber 1966, 1970: 17-22, and 1991: 38-40, 68-90, 128-144, 178-179 and passim. For an analysis in economic terms of conflicting versions of the Tannaitic rule of purchase (“silver acquires gold” and “gold acquires silver”), which in fact reflect developments in the Roman monetary system, see Kleiman 1973. Kleiman, a Professor of Economics at the Hebrew University of Jerusalem, corrects Avi-Yonah’s earlier treatment of this evidence (in the Hebrew edition of this book), but apparently accepts his view concerning the inflation and the economic crisis of the 3rd century (1973: 49 n.3). In a postscript he contests Sperber’s conclusion (1991: 77 [the first edition was published in 1974]) that people “tended to see in the gold coinage…the real stable standard”. 28 See Avi-Yonah 1976b: 94; Sperber 1969; Isaac 1990: 291-297. 29 See Avi-Yonah 1976b: 94-95; Isaac 1990: 297-304. 30 See Sperber 1972a, 1972b and 1978: 11-69, 102-118; Levine 1982: 127-130. 31 See Sperber 1974 and 1978: 70-99; Hamel 1990: 103-108; Levine 1999. 32 For a brief account of the rabbinic literature see Safrai 1994: 459-462; on its limitations and the methodology employed to use it in historical research: 34-38. 33 The translation, with the explanations in the round brackets, follows that of Sperber 1972b: 230-231.

155

THE LATE ROMAN ARMY IN THE EAST rain did not come down. R. Joshua [b. Levy] made a fast in the South, and rain came down.34

Unfortunately the information supplied by the Talmudic testimonies does not enlighten us to know and understand how and to what extent the Palestinian population coped with the adverse situations. Such basic, important questions as whether settlements expanded, shrank or were abandoned, whether the population increased or decreased, whether trade grew or declined cannot be resolved on the basis of the literary sources. In contrast, the finds obtained in archaeological surveys and excavations, mainly in the last four decades or so, are more helpful, although they too are not devoid of difficulties.

R. Yohanan, the foremost Sage of 3rd c. Palestine, died in 279, and R. ‘Ilai (here R. La) was his disciple. Their sayings have been construed to reflect the deterioration of agricultural conditions in Palestine in the second and third thirds of the 3rd c., and perhaps there is something to such an interpretation. However, they surely reflect nostalgic sentiments and subjective attitudes, and the information they contain is imprecise and can hardly contribute to real understanding of the causes, extent and effects of agricultural developments in the period under discussion. The tales about R. Hanina b. Hama and R. Joshua b. Levi, heads of schools in Sepphoris and Lydda, respectively, reflect events that occurred sometime between 220 and 240.35 The problem with such tales about drought, pestilence, famine and the irregularity of rains is that they span the period from the late 1st century CE, or even earlier, to late antiquity, which puts in question D. Sperber’s inference that the situation in these respects was particularly severe in the 3rd c.36 Likewise the Talmudic evidence concerning forced labour and billeting cannot show that the Palestinian population was more oppressed by these two munera in the 3rd c. than earlier or later. There is indeed some evidence about rising of prices, debasement of coins and inflation, abandonment of land and migration to towns, but again the data are limited in number and their significance unclear.37 It would be wrong, however, to dismiss altogether as useless the Talmudic testimonies concerning the economic and social conditions in Palestine in the 3rd century.38 Combined with the evidence available in other sources on the multifarious problems and hardships that disrupted life in many parts of the empire in that period, sometimes to extreme degree, they do suggest that the above delineated factors troubled Palestine as well.39 34

The translation, with the supplements in square brackets, follows that of Sperber 1974: 276. According to the rest of the tale, rain came down finally also in Sepphoris. 35 Sperber 1974: 277. 36 Sperber 1974: 272: “Irregularity of the seasons and successive droughts, followed by famine and pestilence, took their tool of both the population and the land, contributing a tragic element to the bleak and somber drama of the third century”. For selected examples of drought in Palestine (1st c. BCE – 4th c. CE) see Gichon 2002: 187, and a fuller list Klein 1986 Appendix 5: 40-48. On drought and famine as recurrent phenomenona in Palestine in the 3rd, 4th and the 6th c. see Dauphin 1998: 508-509 and 510-511. On famine and subsistence crises in antiquity see Garnsey 1988. Food crisis, or subsistence crisis, was frequent, famine rare. On these two different phenomena in world history see Garnsey 1998: 272-292. 37 See Strobel 1989; Safrai 1994: 433-434. For the debate about the value of rabbinic literature for historical research see Neusner 1999, a polemical review of Safrai 1994, and Safrai 1999, a sensible response and a useful exposition of the limitations of this genre of literature and the ways to treat it in a historical study. For my own critical review of Safrai 1994 see Shatzman 1995. 38 Contra the impression one gets from Bar’s treatment of these sources (2002: 44-46). 39 Security problems, both internal (brigandage etc.) and external (nomadic inroads), will be discussed in part two of this study.

156

Twenty years ago Z. Safrai published an article (in Hebrew) in which he claimed that a slow process of population increase was at work in Palestine during the ‘Roman-Byzantine’ period, aiming to examine the influence of that process on the agricultural and economic structure of the country. That claim was based on the findings retrieved in various archaeological surveys, and two figures presented graphically the demographic growth. The graphs of the first figure present the findings discovered in six regions: 1. Upper Galilee; 2. Beth Shean Valley; 3. Area of Issachar (W of the southern part of the Sea of Galilee); 4. Lower Galilee; 5. Transjordan; 6. Carmel. The second figure consists of eleven graphs presenting the findings in eleven regions: 1. Judaea; 2. The Judaean desert; 3. The region of Benjamin and Ephraim; 4. Samaria; 5. The Golan; 6. Samaria stage 2; 7. Schechem area, (combined findings); 8. Ramat Menasseh; 9. Area of Athlith, on the coast of the Carmel; 10. Southern Sharon; 11. Area of Daliya.40 According to Safrai, “there is one clear-cut conclusion from all the surveys: there was a drastic increase in settlements in the Roman-Byzantine period.”41, that is, from some time in the 2nd c. to the late 6th c.. The increase in settlements indicates demographic multiplication and economic growth, “and they both lasted for a rather long period of time (from 200 to 400 years)”. For Safrai, the Talmudic testimonies do reflect real economic difficulties, particularly of the rural sector of the population, and he explains away the apparent incongruity with the archaeological data by asserting that: “All the signs seem to indicate that it was population increase which resulted in economic stress and it was this stress which in the final analysis brought about economic growth”.42 Safrai also holds that the number of cities increased continuously from the late 2nd c. reaching “over-urbanization” in the 5th and 6th c., which too points to steady and continuous demographic and economic growth.43 In short, without saying it in so many words, there was no crisis in the 3rd century. Critical reservations concerning these findings 40

Safrai 1986. Safrai 1994: 436-458 is an English version of the article, with the same figures. Safrai’s sources of information are not made sufficiently clear. 41 Safrai 1986: 28 and 1994: 440. 42 Safrai 1986: 47-48 and 1994: 457-458. For my critical comment on this assertion see Shatzman 1995: 186-187; cf. Dauphin 1998: 520. 43 Safrai 1986: 39-40 and 1994: 454-455. A graph, based on “civil and episcopal lists”, presents this growth.

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY were carried out.52 Subsequently the archaeological survey became a major tool of research and many Ph.D. dissertations were based on field surveys. Noteworthy in particular is the Archaeological Survey of Israel project, organized in accordance with the division of the entire area of Israel into squares of 10 kmX10 km. In the course of the survey of each such unit, sites and their findings are recorded, measured and photographed, and potsherds on the surface are collected. The survey is published by Israel Antiquities Authority (formerly Israel Department of Antiquities and Museums) according to the maps of the Ordinance Survey of Israel at a scale of 1:20000. Each publication consists, inter alia, of a geographical account of the area, an archaeological overview, index of sites by period, a description of the findings of all the sites surveyed, including photos and drawings, bibliography, and the relevant 1:20000 map.53 Thus far publications of some 50 of these so-called maps have appeared, the earliest one in 1978 and the most recent in 2005, that is, about 18% of the units. In survey-based studies, the beginning, duration and abandonment of the settling of the sites, as well as their size, are determined almost exclusively on the number and types of the potsherds collected, helped occasionally by coinage finds and, quite rarely, by glass finds. One example may illustrate the tremendous progress achieved since the first explorations made in the 19th c.: the famous Survey of Western Palestine (published 1881-1883) recorded 48 sites in one area of the NW part of Samaria whereas a meticulous survey of the same area that was conducted in the 80’s of the last century yielded 261 sites.54 C. Dauphin’s recent comprehensive study of the demography and ethnic-religious composition of the population of Palestine in the ‘Byzantine’ period is based on the findings that have been retrieved in the excavations and surveys of the country since the 19th c. The entries of the 2930 sites included in her catalogue and organized in 19 unequal geographic units detail the findings and give references to all publications relating to each of the sites described.55 Described as the birth of the discipline of archaeological demography,56 it is not devoid of some intrinsic drawbacks. The catalogue is indeed very informative but it provides details concerning the occupation of the sites only in the ‘Byzantine’ period and, as is avowed by Dauphin herself,57 the entries do not provide, more often than not, evidence that can establish the chronological development of the sites within this period of some 3 centuries. This shortcoming of the

and their interpretation are presented below, but in the meanwhile it should be noted that according to the graphs drawn by Safrai himself, the regions of the Beth Shean valley and the Carmel witness stagnation that lasted from some time in the 2nd c. to the late 6th c., and that of the Judaean desert shows some decline in those four centuries. According to Bar, the results of archaeological surveys “allow us to observe with some confidence that the period from about the mid-2nd to the early 4th c. was one of the most prosperous in the history of Palestine…most settlements display unbroken continuity into the peak Byzantine period”.44 To confirm this statement he cites four examples. Lachish: 25 sites in the Hellenistic period versus 103 sites in the ‘Late Roman’ period;45 Apollonia: expansion of settlement into previously uninhabited areas in the agricultural hinterland of this coastal city during the 3rd c;46 the NE slopes of Menasseh hills: 35 sites in the ‘Early Roman’ period versus 70 sites in the ‘Late Roman’ period;47 and Urim (some 20 km. W of Beer Sheva): 11 sites in the ‘Early Roman’ period versus 59 sites in the ‘Late Roman’ period.48 Bar, however, admits that some archaeological surveys reveal a decline in settlement in some cases and gives two examples: the area of Hanita and Achziv in western Galilee witnessed a decline in the ‘Late Roman’ period, which the researchers relate to the general economic crisis;49 the southern part of the Carmel shows indications of a decline in the course of the 3rd c.50 These two examples as well as the cases of the Beth Shean valley, the Carmel and the Judaean desert, which are revealed in Safrai’s graphs (above), may suggest that the economic and demographic conditions in Palestine in the periods under discussion differed considerably in different regions and places. Several more examples indicate indeed the prevailing of some variety rather than uniformity of conditions, but first an account of the archaeological surveys that have been conducted in Israel need to be set forth. Archaeological Surveys in Israel: Contributions and Limitations The origins of the archaeological exploration of Palestine go back to the 19th c., and to-day scholars are indebted to accounts and observations made by the pioneers in this field of research, as well as to various works accomplished during the time of the British Mandate.51 Following the establishment of Israel some surveys were conducted in the 50’s and 60’s of the 20th century, and in the wake of the Six-Day War of 1967 hasty archaeological surveys of Judaea, Samaria and the Golan

52

See Kochavi 1972. Note that the photos and drawings appear only in the Hebrew section, but it is easy to locate them thanks to the lists of illustrations with site numbers that are given in the English section. 54 Zertal 1996: 19. 55 Dauphin 1998: vol.3. 56 See the title of Dauphin 1999. 57 Dauphin 1987: 7* and 1998: 53. Note that at an early stage of her work Dauphin expected to be able to present 6 maps for each region: “Roman’, “Roman/Byzantine’, 3 ‘Byzantines’, and Early Arab (1987: 5*-6*). In the event, only 2 regions (the Judaea desert and the Golan) have 4 maps each: ‘Roman’, ‘Late Roman’, ‘Byzantine’, and Early Arab (1998: Figs 27-34). 53

44

Bar 2002: 50. See Dagan 1992: 18. 46 See Roll and Ayalon 1993: 141-151. 47 Zertal 1996: 88-93. 48 Gazit 1996: 15*-16*. 49 Frankel and Getzov 1997: 34*-36. 50 Gibson, Kingsley - Clarke1999: 104 and 108. 51 See, e.g., Dauphin 1998: 41-44. 45

157

THE LATE ROMAN ARMY IN THE EAST

Published Maps of the Archaeological Survey of Israel 1,2 Akhziv and Hanita 22 Haifa – West 23 Haifa – East 24 Shefar'am 26 'Atlit 27 Yagur 28 Nahalal 30 Dor 31 Daliya 32 Mishmar Ha-'Emeq 41 Mt.Tavor 45 'En Dor 46 Gazit 48 Benjamina 52,53 Mikhmoret and Hadera 54 Ma'anit 69 Herzliyya 78 Rosh Ha-'Ayin 80 Lod 83,83/1,83/2,83/12 Land of Benjamin 84 Ashdod 87,88 Niztanim 91 Ziqim 98 Lakhish 101 Jerusalem – Northwest 102 Jerusalem – Northeast 104 Nes Harim 105,106 Jerusalem – South 108/2 Herodium 109 Amatzia 109/7 Deir Mar Saba 125 Urim 139 Nahal Yattir 144 Tel Malhata 166 Shivta 167 Sedé-Boqer – West 168 Sedé-Boqer – East 196 Har Nafha 198 Har Hamran – Southwest 199 Har Hamran – Southeast 200 Mizpé Ramon – Southwest 203 Har Ramon 204 Makhtesh Ramon 225 Har Saggi – Northeast

158

(1997) (2003) (1983) (2003) (1978) (2004) (1982) (2005) (1981) (1999) (1998) (1998) (1991) (2005) (2000) (1990) (1998) (1994) (1997) (1993) (2005) (2005) (2004) (1992) (2003) (2001) (2004) (2000) (1984) (2007) (1994) (1996) (1991) (2003) (2004) (1985) (1981) (1990) (1986) (1993) (1991) (1999) (1994) (1992)

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY evidence holds true for almost all surveys. Indeed, all these researches suffer, some to a greater measure than others, from various limitations and deficiencies, both at the field work stage and the subsequent analysis, interpretation and publication of the results.58

impossible to discern developments, including the beginning and end of habitation, within those periods.61 Thirdly, many publications do not provide information about the number and typology of the potsherds collected for each site, merely stating the periods (Hellenistic, ‘Roman’, ‘Byzantine’, etc.); this holds true for almost all the maps of the Archaeological Survey of Israel.62 Fourthly, many publications merely note the number of the sites settled in each period, without giving any estimate of the size of the inhabited areas. Fifthly, only quite late did scholars become aware of the importance of the estimation of the size of sites chronologically, but in not a few cases they simply gave a maximal size in accordance with the area of the modern village built over the ancient site. Now, it is well to bear in mind that the methods of estimation of the size of a multi-period site depend to a large degree on incomputable variables; no satisfactory mathematical formula has been devised to calculate the different factors and to extrapolate the size of a site chronologically. Besides, testing excavations of surveyed sites have revealed in some cases significant discrepancies between the potsherds collected on the surface and those unearthed.63 Aware of these limitations, more and more scholars have preferred, in recent years, to estimate the size of sites in categories rather than in precise figures. Finally, quantitative changes in the potsherds surveyed or unearthed need not always reflect demographic growth or decline for other factors may have affected such fluctuations, for instance changing trends in the use of pottery, variation in pottery supply, difficulties in access to centres of pottery manufacture, etc.64

To take first the survey itself, one may note several obstacles to and deficiencies in work: in some areas overgrown vegetation impeded the collection of potsherds; in many cases modern villages built over and around ancient sites made difficult the identification of the old core, the collection of material and, sometimes, the estimation of the size of the ancient site; terracing and intensive agricultural work over the periods damaged and sometimes obliterated, partially or completely, the traces of ancient sites. In addition, some surveys were carried out superficially or partially, that is, those conducted mainly from motor vehicles or limited to selected areas. Speaking more generally, surveys differed from one another in the amount of time (per person) invested in surveying the same size of area. These and other factors obviously have affected to a certain degree the quality of the results attained in the various surveys; one adverse result is that in not a few sites the meager number of potsherds makes futile any attempt to date their occupation periods.59 But perhaps even more serious are flaws and limitations involved at the second stage of the study. The typological-chronological characteristics of ceramics were only gradually recognized, notably of the so-called ‘Late Roman’ and ‘Byzantine’ periods, and hence the vast majority of the publications until the very recent years are deficient in this respect.60 Secondly, because of the crude division of the occupation of sites, based on the pottery types, into long periods of time (‘Roman’, or even ‘Early’ and ‘Late Roman’, and ‘Byzantine’ periods), it is 58 On these problems see Dauphin 1987: 6*-8* and 1998: 48-54; Safrai 1994: 438-440; Tsafrir 1996: 271-273; Finkelstein, Lederman Bunimovitz 1997: 11-22; Ben-David 1999: 15-23 (Hebrew); Lapin 2001: 40-44; Magness 2003: 5-7; Leibner 2004: 49-53 (Hebrew). On some topics, notably the method to estimate the size of a site in a given period, there is no consensus. On problems, methodological and others, and biases in surveys in the Mediterranean countries see Barker 1991 and Vallat 1991. 59 For example: the survey of the Lower Golan by the Israel Antiquity Authority yielded 961 potsherds versus the 5631 collected by BenDavid (Ben-David 1999: 23); the survey of the area of Mizpe Ramon in the 1980’s recorded 353 sites versus the 27 noted by N. Glueck earlier (Haiman 1991: 13*). For sites lacking indicative finds see, e.g., Kochavi - Beit-Arie 1991: 16*; Gophna - Ayalon 1998: 16*; Dagan 1992: 28*. See also below on the Negev’s sites. 60 Telling of the poor state of the research of the pottery of these centuries is that a recent publication (Magness 1993) has soon become the standard work for parallels. One reason for my radical critique of M. Gichon’s archaeological research of the limes Palaestinae was his reliance on potsherds collected in surveys conducted in the 1950’s and 1960’s for dating the sites he ascribed to it. It was clear that the potsherds he relied upon for dating the occupation periods of the sites were not sufficiently diagnostic. See Shatzman 1983: 137 ff. with the relevant notes, esp.155 n.66, and 1991: 245-246 with nn. 99-100. This was later admitted by Gichon himself; his attempt to defend the dates he assigned to the sites is not persuasive and will be treated on another occasion (Gichon 1991: 318 and 2002: 185 n.2).

159

Notwithstanding the above mentioned limitations and flaws, it would be unwise to ignore the findings retrieved in archaeological surveys; although not precise, they might indicate trends in habitation in given regions along periods of time. Thanks to advancement in knowledge and refinement in techniques of work and analysis, some of the more recent surveys yield more reliable results and help to check and correct errors and faultly reached conclusions of earlier ones.65 Three surveys of different regions that were conducted with careful attention to the methodological problems may serve as examples.

61 “We never know whether a site was occupied throughout the duration of a given period or only part of it” (Finkelstein, Lederman - Bunimovitz 1997: 21). 62 An exception is Finkelstein and Magen 1993, an archaeological survey of the region between Jerusalem and Ramallah, an area slightly exceeding 500 sq km. Although said to be published in accordance with the guidelines and 1:20000 maps of the Archaeological Survey of Israel, it gives, for each site, information about the number of the potsherds and their breakdown (in percentage) into periods. This essential information is included in the English, shorter section of the book. 63 See Leibner 2004: 69-71. 64 See Millett 1991. 65 For example, methodological faults of the surveys of the 1960’s are noted in Finkelstein, Lederman-Bunimovitz 1997: 6; they add that “reexamination of the sherds collected in southern Samaria during the 1968 survey indicates that some 40% [sic!] of the sites ascribed to the Iron I had, in fact, yielded no clear evidence of occupation at that time”.

THE LATE ROMAN ARMY IN THE EAST A survey of Southern Samaria, an area of ca. 1050 sq. km. between Ramallah and Shechem, was conducted for seven years, October 1980 to December 1987. A twovolume book of this survey appeared in 1997. In part one of this publication the researchers who took part in the project give a clear, detailed exposition of the methods of the field survey and data recording and of the methods of pottery recording and analysis.66 For example, dating was based on diagnostic potsherds only, and a detailed descriptive list of the pottery types according to periods is given.67 Unfortunately the periodization is rather crude, namely, only ‘Roman’ and ‘Byzantine’ periods; presumably it was impossible, on the basis of the pottery typology, to apply a division into shorter sub-periods, not even to ‘Early’ and ‘Later Roman’ periods, but it should be also mentioned that this research embraced all periods from the Chalcolithic through the Early Ottoman and did not concentrate on the ‘Roman’ and ‘Byzantine’ periods. It is also noteworthy that some potsherds could not be classified into one of two consecutive periods and are marked, those relevant here, ‘Roman’/’Byzantine’ and ‘Byzantine’/Umayyad. The size of single-period sites was assessed according to the extent of the ruins and, in their absence, according to the dispersal of the main pottery concentration. A reasoned discussion of the factors that have been taken into account in the estimation of the size of multi-period sites is given, rejecting any mathematical formulae. Two factors were mainly taken into account in assessing the size of a site in the different periods of occupation: the overall size of the site and the percentage of the potsherds of each period in the overall number of potsherds collected; the full size of the site was assigned to the period mainly represented by the potsherds. The size of sites was estimated in nine categories: 0.1 – 0.3 ha. (=1 – 3 dunams), 0.4 – 1 ha. (4-10 dunams, and so on), 1.1 – 2 ha., 2.1 – 3 ha., 3.1 – 4 ha., 4.1 – 5 ha., 5.1 – 7 ha., 7.1 – 9 ha., 9.1 – 11 ha. The following table summarizes the number of the sites and their aggregate areas in the two periods.68 Number of sites ‘Roman’ Period ‘Byzantine’ Period

242 (225 with estimated size) 331 (313 with estimated size)

Minimal aggregate area

Maximal aggregate area

Mean aggregate area

58.1ha. (= 581 dunams)

132.3 ha. (= 1323 dunams)

95.2ha .(= 952 dunams)

108.9 ha. (= . 1089 dunams)

233.7 ha. ( = 2337 dunams)

171.3 ha. (= 1713 dunams)

sites with estimated size is 88, that is 35%. The increase in the aggregate area inhabited, however, is much higher; minimal: 50.8 ha., that is, 87%; maximal: 101.4 ha., that is, 77%; mean: 76.1 ha., that is, 80%. One advantage of this research, notwithstanding the weaknesses and flaws mentioned above, is that it indicates, with high degree of probability, that the population of Palestine in the rural region surveyed witnessed a very substantial increase in the ‘Byzantine’ period. On the other hand, a disadvantage of the periodization system taken in this research is that it cannot point out the beginning of the demographic growth; nor does it help to know whether a peak was reached very early at the start of this period, followed by stable conditions, or rather the growth was slow, moderate and steady, or whether fluctuations occurred during the three centuries of the period. Yet, the 114 sites that are assigned to a period called ‘Roman/Byzantine’, based on what are termed “unclassified Roman or Byzantine sherds”,69 may well suggest that the rise in the number of the sites began in the ‘Later Roman’ period. If so, this would detract from the reliability of the Talmudic sayings supposed to indicate the prevailing of adverse demographic and economic conditions in the 3rd c; it would also follow that the difference in the number of sites between the ‘Roman’ and ‘Byzantine’ periods was far less radical than what the figures cited above in the table seem to show. More precise and instructive results have been gained by an archaeological survey, in the framework of a Ph.D. dissertation, of the Lower Golan, an area of 215 sq. km., extending E of the upper part of the Jordan River and NE of the Sea of Galilee. The researcher, H. Ben-David, conducted a thorough survey of 45 sites, aimed at collecting 100 diagnostic potsherds, predominately rims, as a sample for each site; in the event, he was able to retrieve this quantity in the vast majority of the sites, except two that yielded each only about 50 diagnostic potsherds. On the basis of their typological characteristics, Ben-David has been able to classify and divide this collection of potsherds into seven sub-periods: Hellenistic (200 – 50 BCE), ‘Early Roman’ (50 BCE – 100 CE), ‘Middle Roman’ (100 - 250 CE), ‘Late Roman’ (250 – 350), ‘Early Byzantine’ (350 – 450 CE), ‘Middle Byzantine’ (450 – 550), and ‘Late Byzantine’ (550 – 650).70 The estimation of the maximal size of the sites, varying from 0.3 ha. of the smallest to 3.5 ha. of the

It comes out that the number of all the sites listed in the ‘Byzantine’ period surpassed that of the ‘Roman’ period by 89, an increase of 36%. The increase of the number of 66

Finkelstein, Lederman - Bunimovitz 1997: 11-39. Note, however, that although the pottery plates present only diagnostic potsherds, they do not “portray all periods of occupation, all types retrieved, or the quantitative aspect of the assemblage” (ibid.: 19). 68 For the lists see Finkelstine, Lederman - Bunimovitz 1997: 915-927. No size is given for 17 sites of the ‘Roman’ period and for 18 of the ‘Byzantine’ period. 67

160

69 See the definition and lists of these sites in Finkelstein, Lederman Bunimovitz 1997: 19 and 919-921. 70 Ben-David 1999. Ben-David’s database includes, in addition to the 5631 shards retrieved in his own survey, some 1000 potsherds collected in pervious surveys of the sites of the Lower Golan. The typologicalchronological analysis of the potsherds was mainly based on finds of a particular group of plain wares widely found in the Galilee and the Golan and known as Kefar Hananya wares (Adan-Bayewitz 1993), of which 5130 rims were collected at all but one of the sites, and on a family of imported fine wares; 622 rims of this group of Later Roman Red Ware were found in 66% of the sites surveyed (Ben-David 1999: 148-149 and 169-175).

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY Period Number of Sites Aggregate area

‘Early R’. 25

‘Middle R’. 38

‘Late R’ 38

‘Early B’. 29

‘Middle B’. 26

‘Late B’. 14

28.2 ha.

34.2 ha.

55.2 ha.

53.5 ha.

22.8 ha.

7.9 ha.

Table 2 Years

0-50

Sites Total Area

33 28.1 ha.

50100 35 21.9 ha.

100150 37 27.6 ha.

150200 39 35.8 ha.

200250 40 35.6 ha.

250300 39 44.2 ha.

300350 34 50 ha.

350400 29 49.3 ha.

400450 29 53.5 Ha.

450500 29 38.4 ha.

500550 29 22.8 ha.

600650 20 7.6 ha.

Table 3 largest, was made in accordance with the dispersal of the potsherds and the confines of remains of the ancient buildings, as well as by analysis of air photos at a scale of 1:1000 of each site. The maximal size was ascribed to the sub-period with the largest quantity of potsherds retrieved, and two main factors were taken into account in calculating the inhabited area of a site in each of the other sub-periods: the percentage of the potsherds (compared to those of the period with the largest quantity) and the length of the sub-period. The aggregate area of all the sites inhabited in each sub-period, in addition to the number of the sites, is quite interesting and instructive, as Table 2 above shows.71

dating could not be precise and a deviation of up to 50 years had to be taken into account, he has divided the period 0 - 650 into 50-year units, calculating the number of all the sites inhabited and the aggregate area settled in each of these units. Table 3 above summarizes the data presented by him.72 Although not fully precise,73 the data registered in the table demonstrate the deficiencies of a study that divides 650 years into 6 sub-periods, let alone one that distinguishes only 2 periods in this long span of time. They also show, even more clearly than the former table, that the number of sites settled is somewhat deficient in indicating the size of settled-areas and the supposedly concomitant demographic conditions. On the positive side, one may note that the growth of the inhabited area in the Lower Golan began in the early 2nd c.; the level reached in the mid-2nd c. remained stable for the next 100 years and rose again in the next half of century; the peak gained around 300 CE was maintained during the 4th c., culminating in the 1st half of the 5th c. A sharp decline marks the 2nd half of the 5th c. continuing in the next 150 years, apparently with accelerated pace. Obviously this outline of the settlement history and demography of the Lower Golan partly corrects but also partly corroborates the one drawn on the basis of the former table. It is particularly instructive to note that no decline is registered in the 1st half of the 3rd c. and that the 2nd half is marked by a substantial increase.

It is well to bear in mind that not all the sites were settled continuously throughout the various sub-periods, and this dynamic character of the settlement history in this region may be summarized as follows: 18 sites were inhabited continuously during all the 3 sub-’Roman’ periods and in at least the first 2 of the ‘Byzantine’ sub-periods; 16 sites were inhabited during some of the sub-’Roman’ periods but not in the ‘Byzantine’ period; 11 sites were first settled in the ‘Middle’ or ‘Late Roman’ periods and their occupation lasted throughout the ‘Byzantine’ sub-periods. The table does not reveal these developments, but it does show that the number of the sites settled in any given consecutive periods might be misleading as an index of the changes in the aggregate inhabited areas and in the demographic conditions that occurred during those periods. Also, according to this table, the inhabited area of the Lower Golan reached its peak in the mid-3rd c. and remained stable to the mid-5th.C, followed by a sharp decline in the next century and even more sharper in the next one.

The recent Ph.D. dissertation of U. Leibner is mainly based on an archaeological survey of the eastern Lower Galilee, an area of about 285 sq. km, extending from the southern slopes of Upper Galilee to the Sepphoris Tiberias road in the south and from the basin of the Sea of Galilee in the east to the margins of the wide Galilaean valleys in the west. Methodologically it resembles Ben-

The above depiction of the settlement history of the Lower Golan needs some revision and modification in view of a closer study of the diagnostic potsherds, which is claimed by Ben-David to have enabled him to estimate the beginning and end of occupation of those 27 sites that did not exist throughout all the sub-periods. Since the

72

Ben-David 1999: 224-226 and 232-234. I have some reservations, but this is not the place to discuss them. In the mean time note that table 8, registering the size of each site in each of the ‘Roman’ and ‘Byzantine’ periods, presents 44 sites only. On the other hand, the study offers details concerning 48 and not only 45 sites, as is well-illustrated in the maps accompanying the discussion of the settlement history of the region. See Ben-David 1999: 229, 238, 240, 242, 245, 249, and 252. 73

71

Ben-David 1999: 225-229. The figure for the area settled in the ‘Early Roman’ period includes Gamala and another site excluded by BenDavid.

161

THE LATE ROMAN ARMY IN THE EAST David’s research, but with some significant differences. In addition to the Hellenistic period (300 – 50 BCE), the 7 subsequent centuries studied in this research are divided into 6 sub-periods: ‘Early Roman’ (50 BCE – 135 CE), ‘Middle Roman’ (135 – 250), ‘Late Roman’ (250 – 350), ‘Early Byzantine’ (350 – 450), ‘Middle Byzantine’ (450 – 550), and ‘Late Byzantine’ (550 – 650). A thorough survey of 46 sites have yielded, on the average, 204 identified potsherds per site, a sample considered by Leibner as a necessary condition for a chronologicallybased study of the settlement history of the sites. More than 100 diagnostic potsherds apiece have been retrieved in 39 sites, more than 60 in 4 sites, and somewhat less than 40 in 3 sites. For the typological-chronological analysis of the 9419 potsherds, all rims, retrieved in the survey, Leibner has utilized the results of many previous pottery studies, both in Israel and abroad; the abovementioned periodization is based on these studies. He provides a detailed descriptive list of the pottery types he has identified in this database, 15 for the Hellenistic period and 60 more divided between the 6 other subperiods. Shallow excavations were conducted in 3 sites to test the reliability of the findings made in the sufrace survey. To test the assumption that the 46 sites surveyed, all known from previous surveys and studies, represents reliably the settlement history of the region, a systematic survey of a 9 sq. km area was conducted, and indeed no new sites have been discovered. The estimation of the maximal size of the sites, varying from 0.05 ha. of the smallest to 9 ha. of the largest, was made in accordance with the dispersal of the potsherds and the confines of remains of the ancient buildings, as well as by analysis of air photos, the same method used by Ben-David. Given certain potential distortions inherent to the database, no overall fixed formula was applied to determine the size of all the sites in each of the sub-periods.74 Instead the size of the sites was estimated in 6 categories: 0.05 – 0.3 ha. (=0.5 – 3 dunams), 0.4 – 1 ha. (4-10 dunams, and so on), 1.1 – 2 ha., 2.1 – 3 ha., 3.1 – 4 ha., 4.1 – 6 ha., 6.1 – 9 ha. The size of the sites in the different sub-periods was estimated according to these categories, taking into account the specific variables characteristic of each site.75 Leibner points out that while the earliest time of occupation and the exact size of a site in a given subperiod cannot always be determined with sufficient clarity, the time of the abandonment of sites is made very clear by the pottery findings.

1. The extent of the inhabited area and size of population reached peak during the 2nd. C. CE and prevailed through the early 3rd. c. 2. The abandonment of two sites close to the mid-3rd c. marks the beginning of a sharp decline that reached its lowest level during the 4th c. Of the 35 sites that were inhabited in the ‘Middle Roman’ sub-period, 20 were abandoned during the ‘Late Roman’ sub-period and the beginning of the ‘Early Byzantine’ sub-period; also, a decrease was noticed in some of the remaining sites. 3. A relative settlement stability marks the 5th c. and the beginning of the 6th c.; two sites were even reoccupied. 4. 3 sites were abandoned during the 1st half of the 6th c., followed by a decline that reached nadir during the ‘Late Byzantine’ period, when only 13 sites remained settled. I have presented a somewhat detailed, certainly not exhaustive, account of the methodology, assumptions and criteria of work employed in these three researches, both in the field work and the analysis and interpretation of the data retrieved, in order to illustrate the problems encountered and the biases involved in archaeological surveys. These studies, like several others that have been conducted in Israel, may be considered as the more advanced in their scientific approach to the archaeological survey as a research tool.76 Yet, although discerning and defining the inherent problems and employing sagacious techniques to overcome the biases and reduce the deficiencies, they too cannot eliminate all potential errors and produce the precise picture of the settlement and demographic history of the regions surveyed. As said before, the less advanced and rigorous in their methodological techniques and approach studies, more often than not the older, suffer from many more and varied faults and limitations. Hence, all these shortcomings should be recognized and the conclusions have to be taken with proper and, sometimes, strong reservations. Archaeological Surveys and the Settlement History of Palestine North of the Negev In the following I present 4 studies, based mainly on archaeological surveys, and 19 maps of the Archaeological Survey of Israel, which are, together with the 3 studies discussed above, relevant to the settlement history of Palestine in the periods under discussion. The order of the presentation is from the north to the south.

The main conclusions of this research, relevant to the present article, are the following:

74 One such distortion is the over-presentation of later periods, particularly the ‘Late Roman’ and ‘Early Byzantine’, as has been shown by comparing the collected samples of the shallow excavations with those of the surface survey. Other variables that could influence differently the representation of the sub-periods in each site include types of soil, topographical conditions, the extent of erosion, and the length of the sub-period. 75 Leibner 2004: 71-73. Leibner states that his intimate knowledge of the sites, thanks to recurrent visits and lengthy work in them, has helped him to estimate the variation of the sizes.

162

Mt. Hermon is a fringe region, extending over approximately 1800 sq. km. Only a small fraction of the region, an area of some 120 sq. km., was surveyed in the 76 Perhaps it is not superfluous to mention that controversies concerning theories of interpretations of archaeological finds, as well as methods of surveys and their advantages and disadvantages prevail. For the arid zones of the Middle East see, e.g., Finkelstein 1995: 15-27.

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY years following the Six-Day War of 1967. The finds retrieved in the 64 sites that have been noticed and surveyed, including those unearthed in the excavations of several of the sites (mainly farmsteads, small villages and seasonal dwelling places), suggest that settlement-wise and demographically this area witnessed its best times during the 2nd – 4th c. The chronological limits of these relatively better economic conditions are, however, quite vague, but it is significant that only one site yielded a number of potsherds which may be ascribed to the ‘Byzantine’ period.77 The northern Golan, a region extending over more than 800 sq. km S of Mt. Hermon, was surveyed repeatedly by several researchers after 1967; several sites of the region were also excavated. Of the sites registered and researched in these surveys and excavations, 51 have yielded finds belonging in the ‘Early Roman’ period, 69 in the ‘Late Roman’ period, and 40 in the ‘Byzantine’ period. The settlement of the region apparently culminated in the ‘Late Roman’ period, but the great reduction in the number of the sites in the ‘Byzantine’ period need not necessarily indicate a proportional population decline, for several sites were exceptionally large in that period; some demographic decrease did take place, however. It is noteworthy that the northern part of the region, the Odem area, was first settled in the ‘Late Roman’ period, probably because of the rough, basaltic terrain, severe climatic conditions and shortage of stable water sources. It has been suggested that people settled there under a growing demographic pressure in adjacent areas, and that this was facilitated by the construction, sometime in the 2nd c, of the Roman road Banyas – Damascus, which passed through this area. However, it seems that the area was more sparsely settled in the ‘Byzantine’ period.78

Sites/ Period

Number of sites

New sites

Sites deserted at end of previous period

Hellenistic

106

55

31

51

‘Roman’

170

98

34

72

‘Byzantine’

194

72

48

122

Early Arab c.640c.110080

13

181

13

Sites continuing from previous period

The authors comment that the sites abandoned were situated particularly in the eastern area in the Hellenistic/’Roman’ transition and in the western in the ‘Roman’/’Byzantine’ transition. The steady rise in the number of the sites continuing from period to period, except the last one, is noteworthy. One should also note the decrease in the number of new sites and increase in the number of sites deserted in the ‘Roman’/’Byzantine’ transition, which seem to reflect a loss of momentum in the demographic growth. Indeed, the largest increase in the number of sites (64) occurred in the “Roman’ period, which is 2.6 times more than that in the ‘Byzantine’ period (24). Unfortunately, since the authors provide the maximal size of the sites only, without estimates of their sizes according to the different periods, the aggregate settled area of the region in each period is unknown, which reduces the ability to assess the demographic fluctuations. No such ambiguity may be claimed regarding the staggering number of the sites deserted at the end of the ‘Byzantine’ period. Yet, it strains credulity that all those 181 sites which are not attested by diagnostic potsherds or otherwise in the Early Arab period were abandoned within a very short span of time right at the end of the ‘Byzantine’ period. It is rather possible that a substantial number of them had been in decline or abandoned well before the end of the ‘Byzantine’ period. This may well hold good for other cases as well.

The settlement history of Upper Galilee, a region of about 850 sq. km, is well presented, as much as the available evidence makes it possible, in a recent study.79 Mainly based on the authors’ archeological survey of a part of the region (310 sq. km.) and data recorded in previous surveys for the whole region, it gives a descriptive list of 398 sites. The results gained by analysis of the data retrieved are summarized in a table entitled ‘Number of Sites – Continuity and Interruption’. The section relevant to the present article is reproduced here.

77

Dar 1993: 26, 78, 200-240 (pottery) and passim. The one ‘Byzantine’ potsherd found in Kafr Dura is hardly significant. Dar’s statement that the ‘“Hermon settlements flourished and were prosperous’’ (ib.: 78), is patently a subjective judgment. To include Mt. Hermon in a discussion of settlement growth in “late Antiquity” is quite misleading (contra Bar 2004a: 79). 78 Hartal 1989 131-132 and 2001: 2-3, 143-145, with maps 3-6 in vol.2. There are some incongruities between the figures given in the text and the number of the sites marked on the maps. For an inscribed milestone associated with the road and dated to 162 see Urman 1985: 109-110 and 133 n.54. 79 Frankel, Getzov, Aviam and Degani 2001.

163

A. Zertal has published so far 4 volumes of his survey of the Menasseh hill country, that is, the region that extends from the Jordan river to about 20 km W of Shechem and from the valleys of Esdraelon and Beth Shean in the north to a twisting line S of Shechem, reaching close to 20 km to the south of that city in the eastern and western sections of the region.81 Zertal has decided to divide the whole region into 5 units, and each volume deals with one of them: the Shechem syncline in the central part of the region, from Jenin (Ginae) to Shechem, an area of about 400 sq km (I); the eastern part, including the desert fringes bordering the Jordan valley, an area of about 540 sq km (II); the NW part, from Nahal ‘Iron to Nahal Shechem, an area of about 420 sq km (III); and the Jordan valley, including the adjacent hills on the west and 80 The term Early Moslem, or Early Islamic, is sometimes used instead of Early Arab. 81 Zertal 1992, 1996 and 2005; Zertal - Mirkam 2000.

THE LATE ROMAN ARMY IN THE EAST Carmel, the Esdraelon valley, Lower Galilee and the valley of Beth Shean.82

the Sartaba area, altogether an area of about 250 sq km (IV). The units differ in their geographical and climatic conditions, with the eastern parts characterized by low precipitation and barren land. The descriptive lists include, inter alia, the size of the sites as well as the number of potsherds found and their chronological distribution, but no attempt has been made to estimate the size of the sites by periods. The retrieved results of each of the 4 units (I-IV) are summarized in the following table: I

II

III

IV

Total number

Hellenistic

66

24

60

24

216

‘Early Roman’

62

35

72

19

188

‘Late Roman’

75

‘Byzantine’ Early Arab

133 70

64 143 55

84 97 47

125 91 23

Period/ Map

‘Early Roman’

‘Roman’

‘Byzantine’

Early Arab

17

32

12

3

47

55

3

4

51

120

16

‘I’ 45; ‘II’ 3887

35

‘Late Roman’

Haifa-East83 84

Shefar’am

3

85

Yagur

Nahalal l86

352

31

7089

6

Mishmar HaHa-’Emeq90

112

106

49

Har Tavor91

30

49

2

13

25

10

12

37

-

363

539

133

‘En Dor 93

Gazit

94

/Total

195

50

Daliya88

92

464

31

The general rise and decrease of the number of sites, in the ‘Byzantine’ and Early Arab periods respectively, is conspicuous. But one should also notice the variations: an increase by almost double or more in 3 maps and by treble in 1 map in the ‘Byzantine’ period; total, or close to, disappearance of sites in 3 maps in the Early Arab period. Conversely, the increase in one map (Shefar’am) is quite moderate, and a slight decrease in two maps (Nahalal and Mishmar Ha-’Emeq) marks the transition

Generally speaking, diversity rather than uniformity characterizes the settlement history of these 4 units. More specifically, one may note a decrease in 2 units and an increase in the other 2 units in the ‘Early Roman’ period compared to the Hellenistic one and an increase in 3 units and a decrease in one unit in the ‘Byzantine’ period compared to the ‘Late Roman’ period. Likewise, the rate of increase and decrease in the number of sites registered by period differs, in several cases considerably, from one unit to another. It is noteworthy that the number of sites reached peak in unit IV in the ‘Late roman’ period, decreasing in the ‘Byzantine’ period, and that the increase in unit III from the ‘Late Roman’ period to the ‘Byzantine’ period is really low in comparison with units I and II. Continuous increase from the ‘Early Roman’ through the ‘Byzantine’ period marks units I, II and III and substantial decrease in the Early Arab period marks all the units, but with varying degrees.

82 Only some of the maps provide the division into ‘Early Roman’ and ‘Late Roman’ periods. The sites listed include not only settlements but also rock-hewn burial caves, various installations, etc. Sites dated only tentatively to a given period are included. 83 Ronen - Olamy 1983: v, viii and maps 6-8. Map 8 shows 12 sites, including 4 located in hills, as settled in the Arabic period, which is in incongruity with the statement (on p. v) that no remains from the Early Arabic period were found in the hills. 84 Ronen - Olamy 1983: v, viii and maps 6-8. Map 8 shows 12 sites, including 4 located in hills, as settled in the Arabic period, which is in incongruity with the statement (on p. v) that no remains from the Early Arabic period were found in the hills. 85 Olamy, Sender - Oren 2004: 10*-11*, 17*-18* and maps 3-4. 86 Raban 1982: vi-viii and maps 4-5. The ‘Roman’ period is divided here into ‘Roman I’, ‘Roman II’ and ‘Roman III’ sub-periods, but no explanation is given for this division. After checking some of the entries I have decided to amalgamate the sites listed under ‘Roman I’ and ‘Roman II’ sub-periods into one list, that of the ‘Early Roman’ subperiod, and those listed under ‘Roman III’ are registered here in the ‘Late Roman’ sub-period. 87 No explanation is given in Raban 1982 for the division of the ‘Byzantine’ period into ‘Byzantine I’ and Byzantine II’ sub-periods. 88 Olamy 1981: v, maps 4-5 and pp.13-14 of the Hebrew version. 89 No explanation is given for listing these 70 (or 71) sites in the ‘Roman-Byzantine’ period. Here they are all listed in the ‘Byzantine’ period, but many of them may have been occupied in the ‘Late Roman’ sub-period as well. 90 Raban 1999: 22*-23*, 31* and maps 5-6. 91 See Gal 1998: 15*, 19*, and maps 4-5. 92 See Gal 1998: 46*, 49* and maps 4-5. 93 See Gal 1991: 13*, 15* and maps 4-5. 94 The maximal number of sites of the map of Nahalal has been taken into account for calculating the total number of sites in each of the ‘Roman’, ‘Byzantine’ and Early Arab periods.

I proceed with maps of the Archaeological Survey of Israel. The term site is used by the authors of these maps to refer not only to inhabited areas like farmsteads, hamlets, villages and other settlements, but also to various agricultural, pastoral and manufacturing installations, for example wine- and olive-presses, threshing floors, cisterns, dams, water conduits, rock-cut reservoirs, terraced-fields, animal pens etc. Therefore, the growth and decline in the number of sites reflect, to some extent and generally speaking, intensification and slow down, respectively, in the economic activity, and not only demographic changes. The following table presents the number of sites recorded occupied, by period, according to data compiled from 9 maps. These maps stretch over various parts of Mt.

164

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY from the ‘Roman’ to the ‘Byzantine’ period. Indeed it seems that the settlement peak in the latter two maps was reached in the ‘Late Roman’ period, followed possibly by slow decline during the ‘Byzantine’ period. The decrease in the number of sites in these two maps, particularly in the case of Nahalal, in the transition to the Early Arab period is the least in comparison to the other maps.

Ziqim: ‘Roman’: 93; ‘Early Roman’: 15; ‘Late Roman’: 44; ‘Byzantine’: 111; Early Arab: 87.98 It is very probable that the increase in the number of sites in the ‘Byzantine’ period was even slighter than these data show; possibly the peak had been reached already in the ‘Late Roman’ period, followed by some stability.99 The very moderate decline in the number of sites manifest in the transition to the Early Arab period is noteworthy.

The archaeological survey of the region extending from N of Jerusalem to Ramallah, conforming basically with the guidelines of the Archaeological Survey of Israel, includes four consecutive 10x10 sq. km. units, that is 400 sq. km. The combined results are presented in the following table, with the units listed from west to east.95 Period

Map 15-14

Map 16-14

Map 17-14

Map 18-14

Total

‘Roman’

38

51

48

29

166

‘Byzantine’

43

56

71

38

208

Early Arab

27

24

16

19

86

Rosh Ha-’Ayin: with 17 sites listed as ‘Roman’, 37 as ‘Roman-Byzantine’, 77 as ‘Byzantine’, and 26 as Early Arab, the data suggest that the settlement expansion in this area had begun in the ‘Later Roman’ period and culminated in the ‘Byzantine’ period, which was followed by decline in the transition to the Early Arab period.100 Lod: ‘Roman’: 45; ‘Byzantine-Roman’ 43; ‘Byzantine’: 106; Early Arab: 26.101 Analysis of the finds indicates that this area witnessed a substantial increase in settlement during the ‘Roman’ period reaching apex some time in the ‘Byzantine’ period. Whether or not the low ebb evident in the Early Arab period had its beginnings in the previous period cannot be ascertained.

The variation between the units in the rate of growth and decline between the periods is noteworthy, with 2 maps (15-14, 17-14) exhibiting the opposite extremes in the rate of increase from the ‘Roman’ to the ‘Byzantine’ period and the decrease from the ‘Byzantine’ to the Early Arab period. Compared to the preceding table, both the increase and decrease are much less sharp, and the increase in two maps (15-14, 16-14) is very moderate; given the known deficiencies of surveys, it may be misleading.

Lachish: ‘Roman’: 103; ‘Byzantine’: 158; Early Arab: 36.102 A settlement expansion marked this area in the ‘Roman’ period, probably with acceleration in the 3rd c. The peak reached in the ‘Byzantine’ period was followed by a sharp decline in the transition to the Early Arab period. The coastal maps are marked by a much lower rate of increase and decrease in the number of sites compared to the inland maps. Taken together the first show a less than 15% increase in the ‘Byzantine’ period compared to the ‘Roman’ period and a decrease of about 33% in the Early Arab period. In contrast, the rates of increase and decrease of the inland maps are more than 80% and about 75%, respectively. These differences may suggest that the inland sites differed substantially in their economic basis and structure from the coastal sites, but this point cannot be pursued here.

Next are given summaries of the data presented in 6 maps of the Archaeological Survey of Israel; these maps extend from about 10 km. N. of Tel Aviv in the north to about 10 km. S. of Qiriyat Gat in the south, 3 on the coast and 3 inland. I start with the first group, continuing with the latter, from north to south in both cases. Hertzliya: with 30 sites listed in the ‘Roman’ period, 43 in the ‘Byzantine’ period and 18 in the Early Arab period, this map is characterized by a relatively moderate settlement increase and decrease in the transitions between these periods. 96

The Negev The Negev plays a major role in discussions of the security problems faced and military and strategic measures taken by the Roman government. Hence, it is

Ashdod: ‘Early Roman’: 33; ‘Late Roman’: 61; ‘Byzantine’: 75; Early Arab: 40; 4 sites listed in the ‘Roman’ period cannot be dated more precisely.97 The biggest increase in sites occurred during the ‘Late Roman’ period, and the total number of sites culminated in the ‘Byzantine’ period; the decrease in the transition to the Early Arabic period is relatively moderate. 95 The data have been culled from the figures and the enumeration of the sites in Finkelstein - Magen 1993: 94-95, 131, 263-264, 337-338. They do not include a great number of graves, quarries, cisterns, rock-hewn wine-presses, watch-booths, pens, etc, which cannot be dated and hence are listed separately. 96 Gophna - Ayalon 1998: 12*-13*, 17* and maps 4-5. 97 Berman, Barda - Stark 2005: 12*-13*, 16* and maps 4-5.

165

98 Berman, Stark - Barda 2004: 14*-15*, 18* and maps 4-5. The ‘Roman’ period is said to include sites with no precise chronological identification. Since, however, 3 sites (7, 55, and 196) are listed in the ‘Late Roman’ period and 1 site in both the ‘Early Roman’ and ‘Late Roman’ periods, I have detracted them from the list of the sites of the ‘Roman’ period. 99 If about 66 of the 93 sites listed in the ‘Roman’ period are dated in the ‘Late Roman’ period, which is reasonable enough, peak was reached in that period. 100 Kochavi - Beit-Arie 1991: 10*, 15*-16* and maps 2-4. 101 Gophna - Beirt-Arie 1997: 12*, 19* and maps 4-6. 102 Dagan 1992: 18*, 27*-28* and maps 7-9.

THE LATE ROMAN ARMY IN THE EAST useful and instructive to examine in detail the demographic and economic developments that took place in this region from the High Empire to the Early Arab period, beginning with a table that summarizes the data retrieved in 3 maps that are located in the northern section of the Negev, 2 on the same longitude as that of Beer-Sheva and 1 stretching 10 km. to the south. It is well to bear in mind that, given the inconsistent precipitation, averaging 200 mm. annually, this semi-arid region was barely fit for dry farming. Map/ Period ‘Early Roman’ ‘Late Roman’ ‘Roman’ ‘Byzantine’ Early Moslem Undated sites

‘Urim103

Yattir104

identified as ‘Byzantine’ was either not diagnostic or could not be located. The pottery she examined from the rest of the sites contradicts Govrin’s conclusion that the ‘Byzantine’ settlement “came to a slow, gradual process of desertion and destruction”.107 The results of her examination of the material are tabulated as follows (in Magness’ table ‘Byz’ refers to the 5th/6th to 7th c, and ‘EI’ to the 8th to 9th c.):108 Byz

Tel Malhata105

17 59

4

Prob Byz

EI 3 (incl 1 fortress)

Large sites

3

Farmsteads

4

2

5

30 89 13

4 67 4

Structures

6

5

3

120 19

Watchbooths

3

2

1

8

149

23

Forts

2

2

Cisterns

2

2

Other

1

1

Several comments are in place. First, the number of sites peaked in all 3 maps in the ‘Byzantine’ period, surpassing that of the former period by two, three and, at least, eight times, respectively. Second, it is important to note the significant rise that occurred in ‘Urim in the ‘Late Roman’ period during which the number of sites more than trebled. Third, Tel Malhata is quite an exceptional case in respect of the drastic change, growth and decrease, in the number of sites between the ‘Roman’ – ‘Byzantine’ – Early Arab periods. Fourth, given the number of undated sites because of lack of indicative finds, the settlement history may have differed to some extent from the one indicated by the dated sites in the case of both Yattir and Tel Malhata. Now, in a critical review of the map of ‘Urim, J. Magness has examined the dating of 9 sites provided with illustrated diagnostic pottery: 7 campsites, 1 farm and a group of 7 structures with a cistern and pottery concentrations. The original dating of these sites in the ‘Byzantine’ period has been corrected by her to 5th and 6th c. through the Early Islamic period.106 She has also reviewed all sites identified in the map of Yattir by Y. Govrin as ‘Byzantine’ and/or Early Moslem, or were undated but have ‘Byzantine’ or Early Moslem pottery that she could find in the storerooms of the Israel Antiquities authority. The sites include large settlements, forts and fortresses, farmsteads, isolated or small groups of structures, watchbooths, pens, agricultural terraces, dams, and cisterns. According to Magness, the pottery from 28 sites Govrin

Prob EI

Byz + Byz and/ EI or EI

3

2 1

5

If this is so, one may well agree with Magness’ conclusion: “There is thus an approximately equal number of Byzantine and early Islamic settlements and farmsteads, as well as some that were occupied during both periods. The only marked decline appears to be in the number of (small) forts, watch-booths, and cisterns”. I may add that if she is right, the significant increase in settlement in the area of Yattir probably occurred in the 5th/6th c., and not right at the beginning of the ‘Byzantine’ period, and that the seemingly sharp decline suggested by the survey’s data in fact was rather moderate and limited. To-date 10 maps of the central and southern Highlands of the Negev have been published. The physical and climatic conditions of this region are much harsher than those of the northern part of the Negev; thus, for example, the average annual precipitation is as low as 50100 mm. in some areas.109 The following is a summary of the number of sites recorded by period in each of these maps.

107

Govrin 1991: 20*. Magness 2003: chap.2, esp. 72-73. For a review of the settlement profile of Yattir map see Hirschfeld 1997: 51-53 with Fig. 38. Hirschfeld claims that, to judge by the ceramic finds, 67 of the sites are primarily ‘Byzantine’, but he does not analyze the finds nor does he offer a precise dating of the occupation of the sites within the ‘Byzantine’ period. 109 The authors of all the maps, referred to in the following notes, provide short geographical and ecological accounts of these areas. See also Bruins 1986: 9-32; Shereshevski 1991: 8-17; Avni 1996: 6-7. For the controversial topic of climatic changes see Finkelstein 1995: 31-35, with the literature there cited. For methods of water conservation in the Negev see R. Rubin 1988. 108

103

Gazit 1996: 15*-18*, 22* and maps 5-7. It is curious that the distinction between ‘Early Roman’ and ‘Late Roman’ sites (15*-16*) is not made clear in the index of sites listed by period (22*). 104 Govrin 1991: 19*-20*, 23* and maps 4-6. Govrin’s interpretation of the military aspects of the settlement history of the area seems to me very dubious. 105 Beit-Arie 2003: 13*-14*, 16* and maps 3-4. 106 Magness 2003: 171-173.

166

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY Shivta: ‘Early Roman’ (Nabataean): 45; ‘Late Roman’: 90; ‘Byzantine’: 181; Early Arab: 4; undated: 118.110 Sede-Boqer - W.: Nabataean: 11; ‘Roman’: 6; ‘Byzantine’: 44; Early Arab: 6.111 Sede-Boqer – E.: Nabataean: 6; ‘Roman’: 3; ‘Byzantine’: 45; Early Arab: 8.112 Har Nafha: Nabataean: 42; ‘Roman’: 41; ‘Byzantine’: 233; Early Arab: 12; undated: 147.113 Har Hamran – SW.: Nabataean: 11; ‘Roman’ 5; ‘Byzantine’-Early Arab: 118; undated: 153.114 Har Hamran – SE.: Nabataean: 26; ‘Roman’: 14; ‘Byzantine’-Early Arab: 107; undated: 170.115 Mizpe Ramon: Nabataean: 35; ‘Roman’: 17; ‘Byzantine’Early Arab: 112; undated: 157.116 Har Ramon: Nabataean: 3; ‘Roman’: 7; ‘Byzantine’: 27; Early Arab: 39; undated: 114.117 Makhtesh Ramon: Nabataean: 10; Nabataean-’Roman’: 2; ‘Roman’: 3; ‘Roman’-Byzantine’: 24; ‘Byzantine’: 95; Early Arab: 19; undated: 92.118 Har Saggi: Nabataean: 10; ‘Roman’: 15; ‘Byzantine’Early Arab: 95; undated: 121.119

precisely a substantial number of sites because of the lack of reliably chronological finds, some of the authors define them, as the case may be, ‘Nabataean’-’Roman’, ‘Roman-Byzantine’ and ‘Byzantine’-Early Arab. No less conspicuous and troubling is the great number of sites that have not yielded datable finds in 8 of the maps; in 5 maps the number of these surpasses the number of all sites dated to the Nabataeaen through the Early Arab periods.123 Any assessment of the demographic and economic developments that took place in the Negev is hampered because of these uncertainties. I’ll come back to this problem. The population of the Negev in the Later Roman Empire consisted of settlers and nomadic or semi-nomadic population. Typical of the former group are the so-called cities of the Negev (below), and settlements that are usually classified into 3 categories: large (15 to 60 structures), medium (5 to 15 structures) and small (1-5 structures); sometimes they are termed villages, hamlets and farmhouses, respectively. Many structures of these settlements are characterized by ashlar masonry, carved architectural elements, and arches and flat slabs indicative of stone roofs, thus resembling the highest quality dwellings in the so-called cities. Run-off agriculture is widely held to have been the basis of life for the population of these settlements, supported by animal husbandry. Farming and herding activities are particularly manifest in terraced wadis, dams, stone walls, water conduits, cisterns, threshing floors, silos, wine- and olive-presses, watch-booths, and animal pens - sometimes adjacent to or even within the settlements themselves.124 These agricultural settlements extend in the area S of the Beer Sheva valley, notably the environs of the so-called cities of the Negev, gradually dwindling towards the south with the Ramon Crater apparently marking their southernmost limit.125 Campsites, animal pens, cisterns, rock shelters, middens, and culting installations testify to the seasonal places of abode of the second group, the semi-nomadic population who mainly roamed south of the region of the permanent settlements, extending as far as 70 km S of the Ramon Crater. It is interesting that in some cases they succeeded in developing terraced fields in wadis, with the necessary agricultural installations for farming the land.126

Several remarks and comments are called for. Because the Nabataeans ruled the Negev until the annexation of their kingdom by Trajan, the authors of these maps use the term ‘Nabataen’ period, that is, c. 200 BCE – 106 CE. The dating of the sites to this period is based primarily on findings of typical Nabataean pottery, sometimes of ‘Early Roman’ wares, but in some cases (Har Nafha, Har Hamran – SW), the dating is tentative, or dubious, because of the lack of such findings.120 In 4 cases (Har Hamran – SW and SE, Mizpe Ramon and Har Ramon) the term ‘Roman’ is explained to refer to the ‘Late Roman’ period.121 In the other cases, it may sensibly be assumed that this is the chronological meaning of the term, but the authors of the maps do not clarify this matter. Indeed, dating is a major problem, as is admitted explicitly by almost all authors. For example, the dwindling of settlement in the Har Nafha area with the transition to the Early Arab period, which is suggested by the number of the sites listed by period, is said to be “misleading, since it is difficult, on the basis of pottery, to distinguish between the end of the Byzantine and the beginning of the Early Arab period”.122 Unable to date 110

Baumgarten 2004: 14*-21*, 248-25 and maps 3-4 Cohen 1985: vii, xiii-xiv, xx and maps 4-5. 112 Cohen 1981: vii, xi, xvi and maps 4-5. 113 Lender 1990: xxiii-xxv, xxix-xxx and maps 4-5. 114 Haiman 1986: 19*-22*, 33*-34* and maps 5-6. 115 Haiman 1993: 15*-17*, 21*-22* and maps 5-6. 116 Haiman 1991: 21*-24, 26* and maps 4-5. 117 Haiman 1999: 11*-12*, 18* and maps 4-5. 118 Rosen 1994: 18*-21*, 27* and map 3. 119 Avni 1992: 17*-20*, 24* and maps 4-5. 120 Lender 1990: xxiii; Haiman 1993: 15*. A. Negev introduces confusion in using, sometimes (e.g. 1988: 24), the term “Late Nabatean period” to refer to the 2nd c. and the 1st half of the 3rd c. 121 Haiman 1986: 19*-20* and 28* n.37; 1991: 21*; 1993: 15*; 1999: 11*. 122 Lender 1990: xxiii. According to Haiman (1995: 31), “most of the 233 sites identified as Byzantine do not include any indications of that period; and 150 of them can be identified as farms and installations 111

167

Several settlements of the permanent population stand out because of their, relatively, good state of preservation and size: Haluza, Sa’adon, Mamshit, Rehovot-in-the-Negev,

identical to those of the Negev Highlands dated to to the seventh and eighth centuries c.e.”. See also Avni 1992: 13* and 16*; Haiman 1986: 16*; 1991: 23* and 1993: 16*; Rosen 1994: 19*. 123 It is surprising that in 2 maps, Sede-Boqer - W and – E, all sites are defined by period, and not even one is regarded doubtful. 124 See, e.g., Haiman 1986: 20*-23*; 1995: 34-35; Rubin 1990: 132-144 (Hebrew); 1991: 201-202; 1996: 52-55; 1997: 270-272; Avni 1996: 1923; Hirschfeld 1997: 53-60; Baumgarten 2004: 14*-20*. 125 Haiman 1995: 30-31; Magness 2003: 131. 126 Rosen 1994: 20*-21*; Haiman 1995: 31-32; Avni 1996: 24-32.

THE LATE ROMAN ARMY IN THE EAST Shivta, Nizana and ‘Avdat.127 Erroneously all of them are often called cities, a misnomer that falsely awards them an official municipal status and conveys misleading notions about their economy and layout. True, Haluza was a city (polis), and so was Mamshit, at least in the 6th c., whose identification with Mampsis is generally accepted.128 The rest, however, were large villages129, with no clear orderly plan, like other villages of Palestine;130 indeed Shivta is explicitly named a village in a 5th c. source and Nessana is referred to as a village in several of the papyri found in the site.131 Many scholars hold that in the Later Roman Empire agriculture and animal husbandry formed the economic mainstay of the majority of the population of these large settlements, probably with the exception of Mamshit,132 and in this respect they did not differ from the population of the other settlements. The overall plans and extent of their built-up areas are fairly known thanks to various surveys, excavations and aerial photos of the clearly visible remains. The following is a short account of their development.133

constructed, and the theater continued to be in use at least to the 2nd half of the 5th c. In the Medaba map Haluza is depicted as a large city surrounded by towers, with large buildings atop of which are red roofs. Several of the Nessana papyri show that it retained its status as a district city at least to the end of the 7th c.134 The estimated builtup area, 39 ha., reflects the conditions obtained when Haluza, an un-walled city, was at the height of its prosperity, that is, 4th-6th c. By comparison with other cities of Palestine, it was a medium-size city.135

Haluza: attested as a Nabataean site, apparently a way station on the route from Petra to Gaza, by the 2nd half of the 3rd. c. BCE. The development of the site is evidenced by, inter alia, a theater that was built in the 1st half of the 1st c. CE. It is believed that the network of streets was laid at the beginning of the ‘Late Roman’ period; its economic well being under Nabataean rule continued after the Roman takeover, reaching peak in the 4th c. Named as a polis of Idumaea by Ptolemy in the 2nd c., it certainly had that status in the Later Roman Empire; its territory included Nizana and Shivta and presumably the other large settlements, except Mampsis. It served sometimes as a center of activity for imperial officials. It had a school of rhetoric in the mid-4th c. and a bishop by the late 4th c., a cathedral and 3 churches were

Mamshit: first mentioned, as a polis of Idumaea, by Ptolemy in the 2nd c., it was located on the road from Hebron to Aila by Eusebius (Mampsis), and its status as a polis is recorded by 6th c. sources. Finds retrieved in excavations indicate it was a Nabataean settlement by the early 1st c. CE, with a caravanserai (45X25 m) which was still in use in the ‘Byzantine’ period, a fort and several towers, one internal and 6 peripheral. It continued to develop successfully under Roman rule and various building activities and reconstruction works took place during the ‘Later Roman’ and ‘Byzantine’ periods. Considerable damage was caused by the earthquake of 363. It had 2 churches, dated, variously, to the 2nd half of the 4th c. and the 5th c., a bath-house and water-supply system (dated to the 2nd c.), and several more public buildings whose functions are not clear. Excellent construction, not obtained in the other settlements, characterizes both private dwellings and public buildings. The Nabataean internal tower was dismantled to give room for the construction of a new building in the 2nd c., and over the Nabataean fort was later built the earlier church. Mamshit was surrounded by a wall, dated to the “late Byzantine period” by S. Applebaum, but according to A. Negev it was built under Diocletian or the early 4th c. However, the ten 4th c. coins that were found under the city-wall and other coins indicate that a date in the 5th or the early 6th c. is to be preferred. A wide gate flanked by 2 massive towers represents Mampsis on the Medaba map, and pottery and other findings, retrieved in the

Sa’adon: probably mentioned in one of the Nessana papyri (soudanon), but not in any literary source, it is located about 9 km SW of Haluza and 4 km. SE of Rehovot-in-the-Negev, away from main roads. Densely built, with 2 or 3 churches, it had no orderly plan. Pottery findings and 1 coin, all collected in a surface survey, indicate the occupation of the site, whose estimated size is 2.5 ha., during the 5th to 7th c. There are no finds to suggest that the site was settled in the High Empire, but only excavations can ascertain that this was really so.136

127

These are the official names awarded to these sites by the Governmental Commission of Names. The Arabic and the ancient Greek names of the sites are, respectively, as follows: Khalasa/Elusa, Kh. Sa’adi/Soudanon (?), Kurnub/Mampsis, Kh. Ruheibeh (the ancient Greek name is unknown), Isbeita (or a-Sbeita)/Sobata, Auja elHafir/Nessana, and ‘Abda/ Eboda. 128 For Haluza see Dan 1982; Mayerson 1994: 194-203; Tsafrir, Di Segni - Green 1994: 119. For Mampsis see Hierocles, Synecdemos: 721.8; Georgius Cyprius, Descriptio Orbis Romani: 1049; Tsafrir, Di Segni - Green 1994: 177. Doubts about the identification, based on the supposed abandonment of Mamshit in the mid-6th c. (Foss 1995: 226 n.19), are superfluous for the site was still inhabited in the 7th c. (below). 129 Shereshevski 1991: 216-217; Foss 1995: 225-226 (criticizes Shereshevski for his use of the term ‘urban settlement’); Hirschfeld 1997: 38 n.21 and 56-59; Lewin 2002: 321-324. 130 Shereshevski 1991: 136-149; Hirschfeld 1997: 60-64 and 2003 (Shivta). 131 Shivta: Nilus, Narrationes: VII; Abel 1935; Nessana: Kraemer 1958: 47 (no.16), 53 (no.17), 56 (no.18), 78 (no.24), 85 (no.27). 132 See, e.g., Mayerson 1960; Mazor 1981 (wine-presses); Rubin Shereshevski 1988: 54; Rubin 1990: 73-104; Shereshevski 1991: 198200; Hirschfeld 2003. 133 Many points that are problematic and controversial cannot be dealt here, and only a few of them are mentioned. On the contribution and limitations of the archaeological evidence for the study of urban life in the 6th-8th c. cf. Russell 1986.

168

134

Ptol. 5.16,10; Tsafrir, Di Segni - Green 1994: 119; Negev 1993a; Shereshevski 1991: 82-90; Mayerson 1994:197-203; Segal 1995: 89-91 (theatre); Magness 2003: 191; Saidel - Christopherson 2005. Against the view that Haluza was the capital of Palaestina Tertia/Salutaris see Fiema 2001a: 425-426, 2001b: 112-113, and 2002b: 192 and 213. 135 Shereshevski 1991: 204 and 247; Broshi 2001: 100-101. Estimates of 80ha. (Negev), or even 60 ha. (Hirschfeld 1997: 39), would make it a large-size city. 136 Kraemer 1958: 230 (no.79, l.47); Rubin - Shereshevski 1988; Shereshevski 1991: 90-93.

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY c.141 The built-up area of this un-walled settlement is estimated at 8-11.5 ha.142

recent excavations, show that the city was still inhabited in the 7th c, and was not abandoned or destroyed by the Arabs as was maintained by A. Negev.137 Mamshit is the smallest of the large settlements of the Negev, about 4 ha., just like Joppa.138

Nizana: located at a major cross-road, it is not mentioned in any ancient literary source and its ancient name is known thanks to the discovery of the collection of papyri on the acropolis. Inhabited by the 3rd.c. BCE, it subsequently witnessed economic growth under Nabataean rule. The Roman takeover of the Nabataean kingdom ushered in a period of decline which lasted to the 4th c. The construction of a fort on the northern peak of the acropolis in the late 4th c., a date corroborated in the recent excavations of the site, marks a period of continuous development of the settlement, both in the private and public sectors, which went on to the early 8th c. Several churches were constructed (the oldest in the 1st half of the 5th c.): 2 on the acropolis and at least 2 in the lower, eastern part of the settlement. The settlement had 2 caravanserais and a hospice. The water supply system included 4 cisterns (on the acropolis) and several wells in the lower parts. Among the late buildings of the settlement are included a monastery, recently discovered, that was built on the northern slope of the acropolis in the late 6th c., the St. Mary’s church that was built on the SW peak of the acropolis in the early 7th c., and the very large church (45X35.6m), unearthed in the recent excavations in the lower part, that was built not before the late 7th c. These and various other finds show that as from about the early 5th c. the settlement was vigorously growing, reaching economic and demographic peak during the 6th-7th c.; deterioration began after the mid8th c., although occupation lasted to about the mid-9th c.143 The built-up area of Nessana, both the acropolis and the lower part, is estimated at 17 ha.144

Rehovot-in-the-Negev: not known from any literary source, it was located about 10 km SW of Haluza on a secondary road leading to Nizana. Nabataean habitation in the site, apparently a way station, is attested by pottery finds of the early 1st c. CE, but whether the inhabitants lived in tents or permanent dwellings is not known. The little available evidence suggests that the site continued to be somewhat inhabited in the ‘Later Roman’ period. As from about the mid-4th c. the settlement developed extensively reaching the height of its growth in the 5th and 6th c. It is supposed that the densely built area of the settlement spread out in all directions from the original centre, evidently with no plan. It had a bath-house and 4 churches - 2 within the built-up area and 2 outside it; the earliest is dated around 400 CE. A large building (30x30 m) equipped with a stable was perhaps a caravanserai, possibly a residence of a rich person. The supposed decline and demise of the settlement during the 7th c. has been recently challenged; the various extant finds rather suggest that the site was still inhabited until at least the mid 8th-c.139 The built-up area of this un-walled settlement is estimated at 10-12 ha.140 Shivta: located away from the main roads of the Negev and mentioned only in a late literary source (Nilus’ Narrations) and the Nessana papyri, it was inhabited by the Nabataeans by the 1st c. CE. That there existed any substantial settlement under the Nabataean rule is doubtful and at any rate no structural remains have been identified positively (pace A. Negev). Although this seems to hold good for the ‘Late Roman’ period as well, the intensive agricultural activity that has been revealed in the immediate vicinity in this period may suggest that the settlement witnessed some constructional growth simultaneously, a possibility which would gain corroboration if certain cisterns could be dated to this period. The extant, visible structures, the best preserved of the 7 large settlements, date from the ‘Byzantine’ period; whether they were gradually constructed over a long period from some time in the 4th c. or were rapidly built in the late 4th c. is difficult to determine. There were 3 churches, one of them (the North Church) was first built some time after the mid-4th c., and the settlement had a caravanserai and a good water-supply system, including public cisterns and water reservoirs. It remained occupied and active after the Arab conquest, probably to the 9th

‘Avdat: situated on the major trade road from Petra to Gaza, first mentioned, in a literary source, by Ptolemy in the 2nd c. CE and then in the Peutinger Table and in the Nessana Papyri. Firm evidence of Nabataen occupation of the site, including a religious compound on the western part of the acropolis, dates to the second half of the 1st c. BCE. Building activity on the site shortly after the Roman takeover is evidenced by inscriptions. In addition, a substantial number of buildings were built E and NE of the acropolis, extending as far as the military camp (below), in 3 phases: 1st c., 2nd c., and from about the mid-3rd to early 4th c., and a residential quarter was developed SE of the acropolis in the 3rd c. They were damaged by the earthquake of 363; according to Fabian, Korjenko and Mazor they collapsed on that occasion but according to Erickson-Gini their destruction was caused

137

Ptol. 5.16,10; Tsafrir, Di Segni - Green 1994: 177; Applebaum 1956; Negev 1988: passim, esp.: 2, 7-8, 24 and 74 1993b; Shereshevski 1991: 20-36; Israeli 1994; Erickson-Gini 1996, 1999 and 2004: 234-244; Kloner 1998: 10-19 (water-supply system); Magness 2003: 188-190. 138 Shereshevski 1991: 22; Broshi 2001: 101. 139 Tsafrir, Patrich - Rosenthal-Heginbottom 1988; Shereshevski 1991: 94-102; Tsafrir - Holum 1993; Magness 2003: 191-194. 140 Hirschfeld 1997: 39.

169

141 Tsafrir, Di Segni - Green 1994: 234; Margalit 1987 (the Northern Church); Segal 1988; Shereshevski 1991: 61-82; Negev 1993c; Tsuk 2002 (water supply); Magness 2003: 185-187; Hirschfeld 2003; Baumgarten 2004: 16*-17* (agricultural development). 142 Hirschfeld 1997: 39. 143 Colt 1962; Shereshevski 1991: 49-60; Negev 1993d; Magness 2003: 183-185 Urman 2004 (a detailed, preliminary report of the recent excavations). 144 Shereshevski 1991: 51 and 243 n.93. Cf. Hirschfeld 1997: 39.

THE LATE ROMAN ARMY IN THE EAST by an earthquake that occurred in the late 4th or early 5th c. Temples dedicated to Zeus-Oboda, Aphrodite and Apis testify to changes made in the Nabataean religious compound in the 2nd half of the 3rd c., to be replaced by 2 churches that were built on the western part of the acropolis, the earliest probably in the late 4th/early 5th c. Some time after the earthquake of 363 a new residential quarter developed on the slopes W and N of the acropolis. The settlement had probably a caravanserai, located S of the military camp, which was in use from the 1st to the 4th c., a bath-house with a well, located in the plain W of the site, and a large number of cisterns; several winepresses within and close by to the settled area testify to a major source of income of the population. The occupation of the settlement ended in the early 7th c., around 630 CE., when it was destroyed by a local earthquake, and not by the Arabs as was maintained by A. Negev.145 According to the most recent estimate, the built-up area of ‘Avdat in the ‘Byzantine’ period stretched over 7 ha.146 The occupation of a nearby military camp (100X100m), located 300 m NE of the acropolis, was short-lived, but the date is disputed; early 2nd to, at most, early 3rd c. CE according to Fabian, late 3rd c. according to Erickson-Gini.; on the whole, the findings and arguments presented by Fabian seem to me more convincing.147 In the late 3rd c. 2 towers were constructed at the southern and northern ends of the residential quarter S of the acropolis; 2 additional towers were integrated into the southern church on the acropolis. The settlement had a fortress (61X39/41m, with no soldiers’ quarters within), situated on the eastern part of the acropolis, and a wall that partly encircled it, from NE to SW, in several sections integrating with the external walls of the dwellings on the periphery of the settlement to form a continuum. Dating the construction of the fortress and the encircling wall is problematic; both the dates suggested, early 4th c. and late 4th/early 5th c., face difficulties which cannot be discussed here.148

b.

c.

d.

e.

Taking into account the construction activities and the types of buildings and installations built from the 1st c. CE through the Early Moslem period, as well as pottery evidence not discussed above, which reflect the economy and demography of these settlements, the following observations may be made, some more secure than others. a. Haluza, Mamshit, ‘Avdat and probably Nizana enjoyed active economy in the late 1st c., while Rehovot-in-the-Negev and Sa’adon (if it existed at 145 Tsafrir, Di Segni - Green 1994: 114-115; Shereshevski 1991: 36-48; Negev 1993e, 1996 and 1997; Korjenkov - Mazor 1999; Erickson-Gini 2002: 113 and 2004: 224-227 and 245-265; Magness 2003: 187-188. 146 Hirschfeld 1997: 39; according to Shereshevski (1991: 204 and 261 n.219) 8.5 ha. The size of the built-up area in the ‘Late Roman’ period did not differ much according to the reconstruction of the settlement history of the site by Fabian 2005: 30 (fig.c-1) and 33-35. 147 Fabian 2005; Erickson-Gini 2002 and 2004: 247-253, respectively, both rejecting Negev’s view that it was a 1st c. CE Nabataean camp. 148 Shereshevski 1991: 43-47; Negev 1997: 89-105; Erickson-Gini 2002: 119.

170

that time) lingered far behind; the situation of Shivta is not clear. After the Roman takeover there was no arrest in economic activity but, rather, continuation, perhaps even some growth, at Haluza, Mamshit and particularly ‘Avdat; in contrast, Nizana suffered decline and no improvement can be noticed at Rehovot-in-the-Negev, not to mention Sa’adon. Shivta’s economic growth may well be inferred from the agricultural development evidenced in its immediate vicinity. All the settlements stood at the height of their economic and demographic expansion for some span of time during the Later Roman Empire. While the inception of that expansion, its culmination and duration, possibly with some fluctuation, varied, they all witnessed flourishing times in the 5th-6th c. No drastic decline set in prior to the 7th c., and, except at ‘Avdat, the occupation of the settlements lasted, with various degrees of habitation, economic activity and level of life, to the 8th c. and in some cases even later. The recent study by Erickson-Gini of the type and quality of the pottery assemblage from ‘Avdat, Mamshit and ‘En Hazeva strongly suggests that the international long distance trade of luxury goods that passed through the central Negev had decreased drastically by the mid-3rd c. and after some time apparently terminated; her findings thus support the view that the importance of Petra dwindled about the mid- 3rd c. because of changes in that trade. The ceramic assemblage of the 4th c. attests active interregional trade of agricultural produce and in particular the production of wine.149

The developments that took place in these 7 settlements have implications for the history of the other sites of the region. Sufficient evidence, literary, epigraphic and archaeological, indicates that it was the international trade carried along the main roads, notably the Petra – Gaza road with its branches, that was the main source of income for the population of the few permanent settlements in the 1st to early 3rd c. It seems that agricultural farming was limited, merely a subsidiary resource for those settlements in those times. Basically they were way stations, not large in size but with relatively affluent communities, Haluza standing out in its advancement; their impact on the nomadic or seminomadic population of the region will have been limited. However, with the waning of the international trade during the 2nd third of the 3rd c., agriculture and animal husbandry developed substantially, apparently with variation in pace and time in different places, to become a successful economic enterprise that made possible the growth and formation of those large settlements of the region in the Later Roman Empire. Seen from this angle of view, although biased to some extent, the amazing 149 Erickson-Gini 2004: 409-413; Fiema 2001b: 112; Lewin 2002: 342343. See also now Erickson-Gini in the present volume.

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY increase in the number of the sites recorded for the Nabataean and ‘Roman’ periods vis-a-vis that for the ‘Byzantine’ period, in the above referred to 10 maps, can make sense. In other words, it is very likely that the growth of the other sites was concomitant with, and perhaps in part followed shortly after, that of the large settlements, which means that the origins of this development may well have started during the 3rd c. and went on in the 4th c., becoming a general phenomenon in the 5th c. That holds good for the permanent villages, hamlets and farms characterized by the finely constructed buildings defined above, which are mostly located in the environs of the 7 large settlements, decreasing and then disappearing as one moves away and S of them.

third of the 3rd. c; apparently that was basically due to the decline of the consuming markets outside Palestine, itself a result of the general political and military chaotic conditions that prevailed in many of the European and east Mediterranean provinces of the Roman empire at that time. Nevertheless, Haluza, Mamshit and ‘Avdat, 3 of the prominent settlements that had profited from that trade, do not seem to have sustained critical damage; indeed public and domestic buildings constructed in the 2nd half of the 3rd c. show that ‘Avdat was economically strong enough to undertake the expenses involved. It is likely that there, as well as at Shivta, the change into economy mainly based on agricultural farming, particularly wine production, had started by that time, and it spread progressively from some time in the 1st half of the 4th c. to become dominant in the northern and central Negev in the 5th and 6th c., extending, to a limited extent, even to the southern part of the Negev Highlands. Now, if the population of the Later Roman Empire, both in the permanent settlements and seasonal sites of the Negev, surpassed by many times that of the Higher Empire, as is generally held, it is important to examine two topics: the sources of living of the population and the causes that brought about the demographic growth and the development of agricultural farming based on sophisticated run-off water catchment systems.

The dating of the settlements with simple structures, which are situated in the areas S of the environs of the 7 large settlements, is disputed. M. Haiman has surveyed and identified 80 farms of this type, estimating their number in the Negev Highlands at 300. He dates them to the 7th-8th c., and since they were not, in his opinion, cost effective, he construes their founding as an enterprise initiated by the Umayyad rulers to settle the nomadic and semi-nomadic population of the region.150 According to G. Avni, they are dated by pottery to the late 6th-8th c.; their rise was a spontaneous response of the nomadic and semi-nomadic population who lost opportunities for trade, escort services and hired labour when the large settlements declined sharply in the transitional late ‘Byzantine’-Early Moslem period; that population established those farms, as well as seasonal settlements, while adopting the techniques of run-off agriculture and short-range transhumance, instead of long-range migration, as means of livelihood.151 Flaws in both interpretations have been pointed out by J. Magness, who has suggested corrections to the dating of many sites; in a number of cases she has exploited reports of relevant excavations. Generally speaking, she dates the inception of some sites to the 5th c., with occupation going on through the 7th c.; the founding of many other sites she dates after the Arab conquest of Palestine, with occupation continuing even to the 9th c. In other words, it was a process that lasted for more than 3 centuries.152 However, she herself admits that it is impossible to date precisely the establishment of most of the sites under discussion on the basis of the available pottery, mainly retrieved in surface surveys, for the lack of diagnostic types, that is, a foundation date is as much possible in the late ‘Byzantine’ period as at the beginning of the Early Moslem period.153

The first topic is associated with the controversial question concerning the size of the population, which can be discussed here only selectively. In default of statistical data, estimation is the only recourse. In one study, M. Broshi’s method is mainly based on a coefficient of 400500 persons per hectare for walled, densely populated cities. Accordingly, the population of the large settlements of the Negev, not including Sa’adon, is estimated at 24500, to which he adds about 5500 for the inhabitants of the other villages and farms and for the nomadic or semi-nomadic population.154 R. Rubin has criticized his calculations, but his arguments are only partially persuasive.155 First, pace Rubin, the factor of water-supply and consumption ought to be taken into account in calculating the population of built-up settlements; in the specific case he cites, Shivta, Broshi’s estimate (2000) has been rather corroborated by Hirschfeld’s recent independent estimate (2200), arrived at by a thorough study of the dwelling units of the site.156 Second, it is true that Broshi did not take into account Sa’adon and several other sites, notably that of Mezad Yeroham,157 and that he most probably underestimated 154

As was said above, the international trade in luxury goods that moved through the Negev was disrupted in the 2nd. 150

Haiman 1995: 34-45. Avni 1992: 19*-20* and 1995: 83-87. Magness 2003: chap.6 153 Magness 2003: 136. Note also that according to Hirschfeld (1997: 35-6 n.10 and 55 n.79) the finds Haiman uses for dating the sites in the Early Moslem period represent the last phase of the occupation of the sites and not the beginning. 151 152

171

Broshi 2001: 86-109, esp.96-98; 22000 (including Sa’adon) according to Shereshevski 1991: 213. 155 Rubin 1990: 61-62. 156 Hirschfeld 2003: 402-403. 157 Cohen 1993: 1136-1137. Note that the total size of the site, about 25 ha., reflects the combined areas occupied in the Nabataean, ‘Roman’, and the ‘Byzantine’ (5th-6th c.) periods. The built-up area in the ‘Byzantine’ period, the peak of the development of the site, did not include all areas occupied earlier, notably area C. Hirschfeld states (1997: 57; cf. Dauphin 1998: 996) that, according to R. Cohen, the site stretched over an area of 10 ha., which is what Cohen had said in his first, short report of his excavations (Cohen 1966).

THE LATE ROMAN ARMY IN THE EAST the number of the semi-nomadic population, but the alternative Rubin prefers, Avi-Yona’s estimate of 80000 persons, cannot stand scrutiny. Even if one follows G. Avni’s estimate of the population of the sites located between Shivta and Nizana in the north and the Ramon Crater in the south, 10500-12600 persons,158 adding both the residents of the large settlements and the nomadic settlements to the south, the entire population of the Negev in the Later Roman Empire would have been about 45000. However, the population coefficient used by Avni, 500-600 persons per 100 sq km., is more than twice higher than that calculated by I. Finkelstein for the ‘Byzantine’ period, namely 350-400 persons per 200 sq. km, which seems more persuasive.159 Indeed, a 35000 estimate accords Avi-Yona’s view that between one-third and one-half of the inhabitants of the Negev lived in rural and nomadic settlements in the ‘Byzantin’ period.160

the pilgrimage movement must have been limited; c) even in the settlements situated on pilgrim routes only part of the population will have lived on or benefited from this kind of tourism.164 Secondly, by “service in the army” Finkelstein refers to the limitanei, envisaged as farmers-soldiers who allegedly received plots of land and other subventions from the state.165 This is an erroneous, out-dated notion of the limitanei and their functioning within the Roman army in the Later Roman Empire; the fallacy of this notion has already been exposed by several scholars and will be explained in detail in the second part of this study. At best, the Roman forces stationed in the Negev may have purchased some of their needs from the local population, not a substantial source of income. Thirdly, that the Roman government subsidized the population is sheer speculation, not backed by any evidence, neither explicit nor implicit, or compelling arguments. Here is a list of certain actions employed by the Roman authorities that may be used to test the plausibility of Bruin’s and Finkelstein’s suggestions. In the past Rome established colonies for strategic reasons and allocated lands to discharged soldiers or needy citizens because of political and social considerations; it never extended its support beyond the initial stage of the act of colonization. The settlers had to cope on their own with their economic problems.166 Roman emperors supported at times cities, mainly when struck by natural disasters, like earthquakes and droughts, pressed by financial distress or involved in exceptional undertakings.167 Their beneficia, usually granted in response to specific requests and intended to show their liberality and enhance their public image, included gifts of money and corn, remissions of fiscal debts, temporary immunity of taxes, grants of right to bona vacantia, etc.; such material benefactions were never given on a permanent basis.168 A system of subsistence payments to children, mainly confined to Italian cities, was introduced by Nerva or Trajan and lasted to the early 3rdc. The scale was apparently limited, and the purpose is debated: a means to ease Rome’s manpower needs, poor-relief or enhancement of the Emperor’s prestige.169 Free food allowances (bread, oil, wine and meat) were distributed regularly to the population of Rome and Constantinople in the Later Roman Empire; only a few other cities (Puteoli and Tarracina in Italy, Alexandria, and perhaps Carthage) were singled out to benefit from a somewhat

But such a figure entails some difficulties. According to the data and calculations presented by H.J. Bruins, the cereal produce of the run-off farming of the Negev could have fed, given the occurrence of drought years, probably once every 3-4 years, seed requirements for sowing and areas allocated to horticulture, no more than 8000 people. Hence, Bruins infers, the population of the Negev had to rely on food supply from the outside, and the imperial authorities must have subsidized the local communities.161 I. Finkelstein has arrived to a somewhat akin conclusion. The economic system prevailing in the maps of Sede-Boqer in the ‘Byzantine’ period, a combination of cereal cultivation, horticulture, and herding, can support, according to his calculations, a population of 350–400 people. He then claims that the population of the ‘Byzantine’ sites recorded in those 2 maps may be estimated at about 550 people, and hence the population needed external resources, which came by way of investments of the imperial government, pilgrimage to Sinai, and service in the army.162 If Bruins and Finkelstein are right, the higher one estimates the total population of the Negev, the greater number of people would have depended on external resources. The trouble with the premises about the external sources is that they are partly exaggerated and partly unfounded. Pilgrimage indeed contributed to the economy of Haluza, Rehovot-in-the-Negev and Nizana, situated on the road to Sinai, but its importance for the Negev population in general should not be exaggerated: a) that was only one of the three routes by which pilgrims travelled to Mt. Sinai;163 b) given their geographical distribution, what the majority of the permanent and seasonal sites derived from

164

158

Avni 1996: 72-73. Finkelstein 1995: 59. 160 Avi-Yonah 1964: 120. 161 Bruins 1986: 194-201 and 1994: 308. 162 Finkelstein 1995: 50-63. 163 Mayerson 1994: 173-182, 183-196 and 232-242. Conflicting opinions are pronounced by Mayerson concerning the question what route pilgrims mostly used to travel to Mt. Sinai, namely via Clysma or Aila (178-180), or via Haluza, Rehovot- in-the-Negev and Nizana (237). 159

172

Mayerosn’s claim that the growth of the so-called towns of the Negev was due to the advent of Christianity and the spread of monasticism in Sinai (1994: 236-237) is contradicted by his own statement (1994: 47) that the church had hardly secured a foothold in the region in the late 4th c. I deal with this question below. 165 Finkelstein’s authority for this topic is R. Rubin (1990: 33-42, 175180, 187, etc.). 166 See, e.g., Salmon 1969; Keppie 1983; Mann 1983. 167 For example, Augustus and Tiberius gave generous financial and other aid to many cities of Asia Minor which were severely hit by earthquakes. See Magie 1950: 469 and 499-500 with the relevant notes. 168 Millar 1977: 420-434. On emperors’ largesses see also MacMullan 1962. 169 Griffin 2000: 115-117; Purcell 2000: 430-432.

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY similar treatment because of their role in supplying the two capital cities. Except in special circumstances, all the rest of the cities of the empire had to fall back on their own resources.170 Cash handouts to the plebs frumentaria of Rome and to the army on specific events constituted a hallmark of the imperial policy.171 Emperors resorted to subsidies to foreign tribes and rulers as a strategic means, but not to ethnic groups or to the population of a certain locality or region within the empire.172 Finally, foreign tribes, sometimes former enemies, were invited or forced to settle on lands allocated to them by the emperors, but again, once settled no more any material aid was bestowed on them.173 The following patterns of action emerge: permanent cash subventions and/or food allowances to privileged groups, namely, the plebs frumentaria of Rome and Constantinople - and, on a limited scale, of a few other cities – and the army; occasional benefactions to cities; periodic subsidies to foreign rulers and tribes; subsistence payments to children, limited in time; allocation of land to settlers (citizens, veterans, and foreign tribes), a one time action. Both Bruins’ and Finkelstein’ solutions to bridge the supposed gap between the food carrying capacity (by cereal cultivation, horticulture and herding) of the Negev and the food requirements of its population in the Later Roman Empire, implying a shortage of about 70-80% and 25% respectively, run counter to these patterns of action, that is, to Roman policy. Without positive evidence, it is unacceptable that the Roman government went out of its way and initiated a new policy, namely, a steady provision of the population of the Negev – a rural, unprivileged population in a remote province - with food or money which lasted for some three centuries.

public areas and a further reduction by half is made because of the sparse water supply. A substantial reduction on that ground is indeed justified,176 and the resulted estimate is congruous, as noted above, with that of Hirschfeld, who has used a different method of calculation. But surely the sparse water supply factor should also be taken into consideration in the estimations of the other settlements, perhaps on a lesser scale, although here one treads on a precarious ground. In short, Broshi’s estimates for the large settlements are probably higher, perhaps to a significant degree, than the real size of the population, and it is well to bear in mind that he himself emphasizes time and again that his estimates are the maximal possible.177 It is instructive that his calculation of the population of the cities of Palestine north of the Negev is based on a 300 coefficient per hectare.178 In contrast, his figures for the rest of the sites and the semi-nomadic population are, as explained above, rather low. My last example is Bruins minimalist estimate of the grain-growing capacity of the Negev. On the one hand, it would be wise to take his estimate with a grain of salt in view of the unknown variables involved.179 On the other hand, he concluded that the population of the Negev had to depend on imperial subsidies, but without examining the carrying-capacity of horticulture and animal husbandry. To sum up: cereal farming, horticulture, and animal husbandry constituted the mainstay of the economy, and the export of wine and, to some extent, of oil was presumably highly profitable.180 Given the tremendous development of the large settlements, it is clear that there existed a labour force of, probably, moderate size who found their living in the building industry, as others found it in road services to pilgrims. Some, not significant gains were obtained thanks to the purchase power of the civil bureaucracy and the army. With the extra income accrued by all these activities and sources the population could afford to bear the costs for purchases of goods and, when needed, food imported from beyond the region, and it stands to reason that it had learned by experience to leave reserves for drought years. All in all, given the uncertainties involved in all the calculations, a 30-40000 estimate for the total population (20-25000 for the large settlements and 10-15000 for all the rest), seems both reasonable and compatible with the various resources of the region in the Later Roman Empire; the population need not have depended on steady support by the government to develop and thrive.

Now, all these estimations are bedevilled by various sorts of uncertainties and assumptions, and hence may be challenged.174 Finkelstein is aware that his method for calculating the size of the population of the 2 maps of Sede-Boqer, multiplying the number of dwelling units by a family-size coefficient of 4.3, is fraught with pitfalls, and yet he has decided, too optimistically as it seems to me, to ignore them, thus arriving at his rather high estimate of the population of the ‘Byzantine’ period.175 To take another example, Broshi’s method, based on a coefficient of 400-500 persons per hectare to calculate the number of people living in a densely populated walled city. The estimate for Shivta, 2000, is reached after a 12% deduction from the built-up area (11.5 ha.) is made for 170

Rickman 1980; Sirks 1991; Garnsey - Whittaker 1998: 328-332. Campbell 1984: 165-171 and 182-190; Duncan-Jones 1990: 115-116 and 1994 39-41 and 78-85. 172 Hendy 1985: 261-263; Duncan-Jones 1994: 43-45. 173 De Ste Croix 1981: 509-517. 174 For problems in estimations of the size of ancient populations see Zorn 1994. 175 Finkelstein 1995: 60 and 63. I may add that it is rather disturbing to realize that no figures, let alone a detailed enumeration, are provided concerning the number of isolated houses and dwelling units in central sites that have been counted to reach the 550 persons estimate. Also, one may feel sceptical about the assumption that animal husbandry was “relatively limited”. 171

173

176

Cf. Baly 1935: 179, who is cited by Broshi (2001: 95) on this point. Some scholars prefer a much lower coefficient, 250 per ha., and for a discussion see Faust 2005: 109-110. 178 Broshi 2001: 102 with n. 16. 179 According to Rubin (1990: 186), the variables involved in attempts to estimate the agricultural carrying capacity of the Negev are unknown and, hence, it is futile to make such estimations 180 On the wine industry and trade of Palestine, including the Negev, see Mazor 1981; Mayerson 1994: 250-255 and 347-351; Kingsley 2001; Fabian - Goren 2002. 177

THE LATE ROMAN ARMY IN THE EAST Now to the second question. There is no evidence or clue in the written sources to enlighten us on the causes and process that led to the efflorescence of the Negev in the Later Roman Empire. One school of thought associates it with the prevailing of a wetter climate in the ‘Roman’ and ‘Byzantine’ periods.181 Several other scholars have suggested, using evidence observed in a variety of studies, that a process of desiccation began ca. 100 CE and continued through the ‘Byzantine’ period.182 According to one view, there may have occurred slight climatic amelioration during the 4th-8th c. that cannot be discerned by the present scientific techniques; that amelioration can help to explain the expansion of the runoff agriculture to the arid areas of the Negev.183 A different approach has been recently enunciated by Y.Hirschfeld; although claiming that there is evidence to show that a substantial increase in precipitation took place in the Near East in the early 4th c., he admits that other factors, such as security, organization and technology, were more important in creating the economic prosperity of Palestine, notably the Negev, in the Later Roman Empire.184 The truth of the matter is that at present one faces, in the relevant scientific studies, not a few ambiguous and contradictory data and conclusions concerning the climatic conditions that prevailed in Palestine, particularly the Negev, in the Later Roman Empire. For example, according A. Frumkin, in the years 100 BCE-400 CE the Dead Sea level, which is generally considered to reflect wetter or drier periods (as the case might be), was relatively high (except during the 1st c. CE and the beginning of the 2nd c. CE), with peak levels at about -368 m. In contrast, the Dead Sea level during the years 400-900 CE was significantly lower, possibly varying between -436 and -390 m.185 A.S. Issar maintains, however, that the period 200 BCE-700 CE was mostly cold and humid, except during the transition period of 200-300 CE which was warm; a warm and dry period began about 700 CE.186 There is thus disagreement between the findings relied upon by these two scholars concerning the climatic conditions of the Negev in 3 spans of year: 1st c. and beginning of 2nd c. CE, 3rd c., and 400-700 CE. Furthermore, as reconstructed by both these scholars the climatic history of the Negev during these periods stands in contrast with its settlement history, thus undermining the assumption of close inter-

relation between climatic changes and the rise and decline of cultures in arid or semi-arid regions. In this case Issar’s claim that climate was the main factor responsible for the settling of the desert by man is belied by his own findings, for if he were right the Negev should have been densely settled from 200 BCE through 200 CE, which did not happen. In brief, amelioration of climatic conditions need not have always resulted in the development of sophisticated agricultural systems. As for the Later Roman Empire, the available archaeological finds (density of settlement, spread of terraced cultivation into the southern Negev Highlands, and remains of agricultural installations and a variety of crops) testify primarily to the initiative, enterprise, skill and diligence of the inhabitants of the Negev as the decisive factor. In other words, human deeds, not climatic conditions were responsible for the rise, spread and long duration of the run-off agricultural systems.187 The question that arises is the following: what kind of population lived in the Negev in the Later Roman Empire? More specifically: in what respects did it differ from that of the region in former times? R. Rubin has presented the thesis that under the influence of the Roman army, the Roman administration and the Christian church the Negev witnessed a process of Romanization in the Later Roman Empire, which entailed the use of the Roman imperial law and of the Roman calendar; the introduction of certain architectural-cultural elements, notably mosaic pavements, tile roofs, the bathhouse and the basilical church; and, indirectly, the spread of the Greek language and wine- and oil- production – characteristics common to the Mediterranean culture. Sedentarization and the development of the sophisticated run-off systems, with cereal, vine and olive as the main crops, formed part of the Romanization process, whose agents were soldiers, administrators and priests.188 It was not an internal process, for the local population merely acted under the impact of the foreign influence, adopting Roman ideas, methods, technologies and lifestyle, and adapting accordingly their own traditions. Mayerson’s view of the urbanization of the Negev as due to the advent of Christianity and spread of monasticism is in line with this interpretation of the evidence.189 Now, Rubin envisages this process to have occurred with a limited amount of immigration, but on this point D. Bar seems to disagree for he claims that at least part of this settlement process resulted from immigration from other parts of Palestine. He is of the opinion that, under the pressure of the intensive agricultural development and the concomitant demographic growth that took place in the

181 Issar - Govrin 1991; Issar, Govrin, Geyeh, Wakshal - Wolf 1992; Issar - Makover-Levin 1995. A major problem of these studies, although not the only one, is that they regard the 1st-6th c. as a uniform climatic period, thereby undermining the theory that correlates mechanically or deterministically human settlement with climatic conditions. Notwithstanding the reservations I noted above, the settlement history of the Negev in those 6 odd centuries shows, generally speaking, a marked difference between the ‘Roman’ and ‘Byzantine’ periods. 182 Klein 1982 and 1986 (n.b. chronological sheet no.2); Frumkin, Magaritz, Carmi - Zak 1991; Bruins 1986: 189 and 1994; Rosen 2000: 54-56. 183 Thus Avni 1996: 68-69. 184 Hirschfeld 2004. Some of his references (e.g. Frumkin 1997) do not support his assertion that the period 4th to 6th was a humid one in the Levant. 185 Frumkin 1997: 244. 186 Issar 1998: 123-124.

187

Cf. Rubin 1989 and 1995. Rubin 1990: 69, 72 and 189, 1996, 1997a and 1997b. Cf. Rosen 2000: 56 (“The essence of the classical period ‘Green Revolution’ in the Negev was the transplantation of a Mediterranean-zone agricultural complex into the arid zone… the expansion of the Mediterranean zone into the desert, in terms of culture, society and subsistence, in fact proved a remarkably stable phenomenon, enduring for at least half of a millennium”. 189 Mayerson 1994: 237. 188

174

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY more stable and fertile districts of Palestine in the Later Roman Empire, people were driven to immigrate and settle in the “frontiers” of the country, that is, zones which, because of adverse geographical and/or climatic conditions, were sparsely settled and where towns, villages and farmhouse were scarcely found during most periods. Such “frontier” zones included Mt. Hermon, the Odem area in the northern Golan, the swamps of the Esdraelon valley, Mt. Carmel, the eastern desert parts of Samaria and Judaea, the Negev, etc. Rome provided the political context within which the transformation of the “frontiers’ could take place, but it did not attempt to guide its development or to encourage the settlement process.190

material dependence on that community. In short, the growth and material expansion of the settlements was not due to the influence of the Church and the Christianization of the population of the Negev, even if one takes into account the contribution of pilgrimage to the economy of the region, which had its limits, and the involvement of the clergy in the social life of the Christian community. There is no denying the fact that the bath-houses and basilical churches constitute good examples of the implantation of foreign elements, Roman or Christian, onto the large settlements and the lifestyle of their communities. Still, it is instructive that most, if not all, of the churches in the Negev are marked by irregularities in their layouts. Even D. Chen, who assumes that architects and masons were familiar with Roman architectural theory, has to admit that these deviations from the theoretical norms suggest that local masons were employed in their building rather than well trained architects from other regions of Palestine.195 But it seems that this was not an exceptional case, limited to the Negev, but rather the usual practice, in contrast to Roman architecture that was marked by careful planning and meticulous construction. Y. Tsafrir argues that most planners of synagogues, churches, and city streets in Palestine and other provinces lacked profound architectural knowledge. They worked with a basic program and according to the local tradition of building and the conditions of the site: “In the Byzantine period people built with a free hand, to some degree a sloppy hand, a kind of comfortable disorder…many of the practical decisions, such as the division of the inner space into a nave and aisles, the dimensions of the doorways, and the like, were decided by the builders in the course of construction, on the basis of conditions created on the site, or by convenience”.196 If so, the churches are not a simple case of import of a foreign element into the material culture of the Negev.

Although insightful and containing much that can be substantiated, these interpretations and judgements, which are not all congruous among themselves, are vulnerable to criticism and need correction and significant modifications. Let me start with Christianity, which in the 6th c. could boast of having at least 20 churches of fine quality, with mosaic pavements, frescoes and imported marble, and several monasteries.191 The churches were indeed the most conspicuous and impressive public buildings of the 7 large settlements. However, it is important to point out that Christianity came to the region only close to the mid-4th, as may be learnt from Jerome’s biography of St. Hilarion (ca. 300370). On his visit to Haluza he found there a pagan community devoted to the cult of Venus; the inhabitants were so much impressed by this famous worker of miracles that they entreated him to found a church.192 At that time, Haluza was a polis of a well-established status, with, inter alia, a theatre dated from the 1st c. CE and a school of rhetoric. It has been plausibly suggested that it had a bishop in late 4th c., although the earliest unequivocal evidence dates from 431 (the Council of Ephesus).193 At any rate, settlement growth clearly preceded the Church in Haluza, the leading city of the Negev, which holds good for Mampsis, Rehovot-in-the Negev, Nizana and ‘Avdat as well (above), and it is more probable than not that the 2 other large settlements (Sa’adon and Shivta) developed and expanded earlier than the establishment of the churches or, at most, simultaneously. One may also note offerings made for various purposes to the church and monastery of St. Sergius in Nizana that are recorded in two 7th c. documents included in the Nessana papyri.194 Many of the donors were local inhabitants of Nizana, but the rest were people from 13 other settlements of the Negev, including Shivta, Haluza and Beer-Sheva. It is true that these and other documents testify to the devotion of the Christian community of the Negev to the Church and its institutions, but at the same time they demonstrate its

The large settlements of the Negev differed from one another in several respects (topography, density, the extent of empty areas, fortifications, etc.), but they all had one thing in common, namely, they did not resemble typically Roman cities in their layout. None of them were built according to pre-prepared plans, and none of them have the over-all appearance of a Hippodamic or orthogonal city, perhaps with the exception, to some extent, of Haluza. Whereas in Roman cities the orthogonal scheme, including well-defined areas for public buildings and activities, was generally implemented even in the face of topographical difficulties, in these settlements the local terrain determined street alignment and building orientation. In the few areas where orthogonal plan was executed it was easy to do so, but on the whole there was no hierarchy of streets and no division into definite quarters. It has been

190

Bar 2004a and cf. 2004b. For details see the account of the 7 large settlements (above) and the literature there cited. 192 Mayerson 1994: 197-198. 193 Mayerson 1994: 48. 194 Kraemrer 1958: 227-234 (nos. 79 and 80). 191

195 196

175

Chen 1981, 1985 and 1990. Tsafrir 1990, cited at p.536.

THE LATE ROMAN ARMY IN THE EAST claimed that they resembled Jewish towns and differed substantially from the eastern Mediterranean GraecoRoman cities.197 Generally this is probably true, although one may note that Scythopolis, a fine exmple of Roman urbanism in the East, lacked the orthogonal plan, explained by the excavators of the city as an “exception that was dictated by the topography of the site”.198 Furthermore, a detailed examination of the plans and constructional features of the private dwelling persuasively shows that the “the Byzantine domestic architecture preserved the characteristics of the Nabatean era, although on a lower standard. A definite continuation between the two periods may be discerned”.199

Negev slopes were left bare or even cleared and terraces were built across wadi beds in order to catch as much as can run-off. Thus, the run-off agricultural system in the Negev was at most a local adaptation of a foreign method, but could have equally been a purely indigenous development. Fifth, it is wrong to regard the limitanei as farmers-soldiers who played major role in promoting the run-off farming system and typically Mediterranean crops; these were regular soldiers to all intents and puposes, certainly in the 4th c., by which time run-off agriculture and wine industry were in existence in various areas in the Negev. These regular soldiers cannot have been associated with any agricultural enterprises.

Rubin’s thesis that sedentarization and the development of the sophisticated run-off farming systems, with cereal, vine and olive as the main crops, formed part of a Romanization process, whose agents were soldiers, administrators and priests, may be questioned on several grounds. First, if sedentarization is attested by building construction in the large settlements, as is maintained by Rubin, we have now evidence that both public and private buildings were constructed at ‘Avdat, Mamshit and, probably, Haluza in the 2nd and 3rd c., that is, prior to the ‘Byzantine’ period, during which, according to Rubin, the Romanization process took place. Second, as explained above, it is quite possible, even likely, that some of the sites defined ‘Byzantine’ or undated had their origins in the ‘Late Roman’ or ‘Roman’ period, which would point at an early process of sedentarization. Third, farmsteads and areas typical of the run-off agricultural system at the vicinity of Shivta, ‘Avdat and, apparently, several other sites, have been dated in the ‘Later Roman’ period. Fourth, There is good evidence that run-off agriculture, let alone simple one, was practised in the Negev as early as the Chalcolithic period and the Bronze Age, and particularly in the Iron Age.200 In other words, there is no need to assume that the local people of the Negev must have depended heavily on external expertise to develop the run-off agricultural system, and certainly, as will be presently shown, it was not due to mass immigration of settlers from other regions of Palestine that the system was developed in the Negev.201 As is recognized by Rubin himself,202 terracing in the humid regions of Palestine aimed at preventing soil erosion in mountainous areas, whereas in the arid zones of the

I give one more case that is instructive for the question of external, material-economic influence. It has been recently found out that a ceramic workshop at Haluza produced a type of jar similar to the Gaza and Ashkelon wine jars from the second half of the 4th to the 6th c.; so far these so-called Elusa jars were found in ‘Avdat, Mamshit and Haluza.203 Evidently those responsible for the production of the Elusa jar were familiar with the Gaza and Ashkelon wine jars and wanted to imitate them,204 but they need not have been entrepreuners from the southern coastal plain of Palestine who opened their business in Haluza. It is more probable than not that some local people initiated the business intending to exploit the good name of the coastal wine-jars in order to market the locally produced wine for local consumption. In other words, the material and various other elements of the Mediterranean/Roman culture were not necessarily introduced, let alone imposed, by foreign ‘agents’, but developed by local people, i.e., it was an internal process. Now to the demographic and ethnic profile of the population: local or foreign? Aside from general impressions and inferences one can gain from the not too many available literary sources, there are now available results reached in two different researches: the onomasticon of the population of the Negev and an anthropological study of skeletal remains from the cemetery located at Reheovot-in-the-Negev. In the latter,205 92 well preserved skeletons were excavated from randomly selected 5 sections of the cemetery. The researcher Y. Nagar carried out varied examinations of the bones and has made age and sex estimations and reconstructed the demographic profile and morphological characteristics of the local population, including the frequencies of 22 epigenetic traits of a proven genetic basis. The morphological characteristics were compared with those of four other populations (total sample size 510 individuals), namely two ‘Byzantine’ populations

197

Shereshevski 1991: 136-149. Tsafrir - Foerster 1997: 91. 199 Shereshevski 1991: 149-155, cited at 155. 200 Bruins 1986: 36-37 and 67-87. Haiman (1989: 184-185 and 2003: 74) accepts that simple agriculture was practised in the Iron Age, claiming that the run-off system, with dams and terraced wadis, belongs in later periods. However, Bruins’ excavations have shown the operation of the advanced system in the Iron Age, at least at Hurvat Halukim. 201 Hence, Haiman’s theory that the settlement system of the Negev Highland in the Iron Age was due to a population migration from northern Palestine (2003: 75), favourably reviewed by Dauphin (2003: 70-71) but with perceptive criticism, cannot be adduced as relevant. 202 Rubin 1997: 271. For criticism of these theories see at length Lewin 2002. 198

203

Fabian - Goren 2002. According to the authors (ibidem: 148), in stratigraphic contexts the jar was found in ‘Avdat “in late Roman layers”; if so, the production of this type of jar may well have begun in the 1st half of the 4th c., or even earlier. 204 For finds of Gaza jars in the Negev see, e.g., Erickson-Gini 2004: 490-491 (Mamshit), and 529-530, 555, 559, 561-562 and 571-572 (‘Avdat). 205 Nagar 1999.

176

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY from various sites in central and northern Israel, presentday Bedouins from Sinai and the Negev, and Indians from the collection of the Medical School of Tel Aviv University (used as a reference population). The findings and conclusions are interesting and instructive for various purposes, but I’ll single out only those particularly relevant to this article. The population of the settlement was typical of a regular civilian population in terms of the representation of infants, children and adults of both sexes, although with a slight under-representation of infants. Life expectancy was about 26 years at birth and 30 years at the age of 10, with average death for adults at 41 (44 for males and 39 for females). These and other findings invalidate the theory that the local population included a large foreign element, namely pilgrims and monks who came from far and wide; the demographic increase, reflected in the expansion of the built-up area of the site in the ‘Byzantine’ period, resulted from immigration of inhabitants from the periphery in the early stages and mainly from natural, internal population growth.206 Although one may discern some weaknesses and lack of clarity in Nagar’s study and generalizations,207 the main conclusion is persuasive: the growth of the population in Rehovot-in-the-Negev was not due to massive Christian immigration (pilgrims and monks) or influx of settlers from other regions of Palestine. In other words, the expansion of the site in the ‘Byzantine’ period had mostly to do with changes within the local population of the Negev.

‘Byzantine’ period in view of the issue of the durability of pastoral nomadic culture. The fundamental question is whether or not nomadic populations leave remains which can be detected by meticulous surveys. One school of thought affirms that this is the case and, hence, maintains that an area without remains in a given period was not inhabited by a nomadic population at that time. Other scholars hold that not all nomads of the desert leave remains traceable archaeologically and, hence, it is wrong to assert that absence of archaeological record indicates a demographic vacuum.209 The latter’s view seems to me to rely on more persuasive empirical examples and arguments, particularly so as time and again new discoveries show the limitations of archaeological surveys. Accordingly, it is quite likely that the settlement growth witnessed in the Negev in the Later Roman Empire is due to a considerable extent to sedentarization of nomadic people that had not left remains in the ‘Roman’ period. Such a development accords closely with Nagar’s conclusion concerning the growth of Rehovot-in-the-Negev and A. Negev’s study of the onomasticon of the Negev, which both indicate the dominance and continuation of the indigenous population in the sites of the Negev from Nabataean times and the Roman takeover of the region to the Later Roman Empire. This sedentarization was, generally speaking, a gradual process, and it can help to understand the expansion of the 7 large settlements as well as the proliferation of the ‘Byzantine’ sites recorded in the surveys in general.

Nagar’s conclusion concerning the identity of the population of Rehovot-in-the-Negev is congruous to a great measure with Negev’s study of the personal names recorded in the inscriptions found in the Negev (Haluza, ‘Avdat and Nizana) and in the Nessana papyri. Almost all the names dated up to 350 CE are pure Nabataean names. The list covering the years 350-650 CE has a large proportion of Nabataean names, but includes names pertaining to Arabian tribes who presumably infiltrated the Negev at some time. There is also a group of Biblical, Greek, and Roman names, which all but 3 are dated in the 6th and 7th c.; their bearers may well have been people of Nabataean stock who had changed their names under the impact of Romanization and Christianization. At any rate, the original Nabataean ethnic element clearly remained prominent to the 7th c., in which new Arab names appeared in the aftermath of the Arab conquest.208

To conclude this discussion of the Negev: the waning of the international trade in luxury goods, notably spices and perfumes, during the 2nd third of the 3rd c. marked a crucial turning point in the economy of the region. Naturally those who were hard hit by the loss of their sources of income will have been the people most interested in finding new means of livelihood. It stands to reason that they took the initiative to introduce or promote the new agricultural systems, and it does not matter whether they acquired the knowledge and expertise needed for such enterprises from regions beyond the Negev or developed the systems on their own. What is particularly important and instructive for the purpose of this paper is that farmsteads and areas typical of the run-off agricultural system have been dated in the ‘Later Roman’ period, which not only shows a chronological connection between the decline of the international trade and the gradual rise of the run-off agriculture and horticulture, but also strongly suggests that it was a casual connection. Furthermore, it is clear that the initial stages of the growth of the agriculture that was to prevail and become the mainstay of the economy of the Negev in the Later Roman Empire had preceded the administrative and military reforms of Diocletian. This puts in doubt the widely held interpretation that

It is worthwhile now to look again at the tremendous increase in the number of sites recorded in the 206

Nagar 1999: 118-119. I give 2 examples: the life table of the population is reconstructed on the assumption that the population was a stable one, but then this very assumption is presented as one of the main results of the research, namely a stable population with growth rate of about 0.5% per year (1999: VIII-IX and English abstract); Nagar does not clarify what support he has for his statement that the population of Rehovot-in-the Negev is part of a continuum beginning with the Nabataeans. 208 Negev 1991: 3 and 130-152. That the infiltration of Arabian tribes began in the 1st c. CE does not follow from the dating of the nonNabataean Arabian names in the period 350-650 (pace Negev). 207

209 See, for the first position, Rosen 1992, and for the opposite view Finkelstein - Perevolotsky 1990 and Finkelstein 1995: 22-30, all dealing particularly with the Negev, with ample literature.

177

THE LATE ROMAN ARMY IN THE EAST local resources the number will increase.212 Had it been written in Greek, the letter would have had polis, boule and bouleutai instead of civitas, curia and decuriones, respectively. At any rate, given the state-imposed duties the cities had to shoulder and the central role decuriones, often called curiales, played in meeting the government demands, as well as in undertaking to provide for various local needs, the liturgical system, an emperor had to assess the adequacy of the economic and demographic capability of a community before granting it the status of a city.213 A curia of 50 members was perhaps the minimum required, but according to Libanius 600 members was the standard in Syrian cities in the 2nd half of the 4th c.214 The increased burdens imposed on the curials, not an economically homogenous order, in the Later Roman Empire resulted in attempts to secure immunity, the richest by promotion to equestrian or senatorial ranks, those with modest means by enrollment in the civil service and the humblest by joining the army. Imperial legislation to prohibit these practices met limited success, and it was especially difficult to prevent desertion to bishoprics, the priesthood and monasteries. All in all the curial order witnessed gradual decline from the 4th c. to the 6th c, and the eventual demise of the curial government, which was taken over by notables government,215 has been interpreted as evidence for the decline of the traditional city, although this is disputed.216 It is reasonable to assume that these processes occurred in Palestine as well, but my aim here is not to deal with the question whether or not the composition and social structure of the governing institutions of the city in Palestine witnessed such transformation in the Later Roman Empire that would justify deeming it a radical change in the nature and essence of the city.217 Rather, the purpose is to outline the economic and demographic conditions of the Palestinian cities in the period under discussion, which, as was shown above, constituted a major factor in an emperor’s decision to grant the status of city at the beginning of the period.

ascribes the efflorescence of the Negev to those reforms and questions the usual perception by which they are assessed.210 It is no less important to call attention, already at this stage, to the phenomenon of unfortified sites (farmsteads, hamlets, villages, and most of the large settlements) widespread all over the Negev both in the ‘Later Roman’ period and the Later Roman Empire. This phenomenon as well as the existence of the advanced type of agriculture prior to Diocletian’s time is suggestive for the understanding of the security problems of the region and the deployment of the Roman army in the country, and their implications will be dealt in the 2nd part of this study. Finally, it is worthwhile to call attention here to relevant archaeological explorations in Jordan. Generally speaking, archaeological surveys of the countryside there have lagged behind those carried out in Israel and are bedevilled by the same methodological and other limitations. In the area S of R. Zered (W. el-Hesa), that is, the region included in the territory of Palaestina Tertia, there is a marked decrease in the recorded sites attributed to the “Roman’ period, as compared with the Nabataean period, a slight increase in the 4th and 5th c., and a very sharp decline in the 6th and 7th c. The results retrieved in the surveys of the area extending from R. Zered to R. Yarmuk show an increase of sites attributed to the ‘Roman’ period and a further, larger number of ‘Byzantine’ sites.211 With all due caution needed because of the limited reliability of these data and the uncertainties involved in their interpretation, it seems that the settlement conditions in the region E of the ‘Arava differed considerably from those that prevailed in the Negev; this difference will have to be taken into consideration. Cities A Latin inscription of the late 3rd/early 4th c. was found in Pisidia at the site of the ancient village Tymandus. The letter, forwarded to one Lepidus, probably the governor of the province, is a reply by an emperor to the request of the local people, granting the village the status of civitas. To be able to carry out the functions other cities are entitled to do, they will have to elect magistrates, aediles, quaestors and other posts. But first the governor is instructed to appoint 50 decuriones, town-councillors, that is, to establish the city’s council, curia. The Tymandeni had promised that a sufficiency of decuriones would be forthcoming locally, which carried weight in the emperor’s decision to grant the request, expressing his hope that by the favour of the gods and the growth of the

In the 6th c. there existed some 40 cities in Palestine, including those situated in Palaestina Tertia E of the ‘Arava valley.218 Many of these were old, renowned cities and most of them are known or may be assumed to have had city status at the early 3rd c. A few are recorded for the first time as cities in 6th c. sources, and I begin with Bitulion, which should not be confused with Bethelea or 212 ILS 6090; Jones 1964: 720; Millar 1977: 395-396. Magie’s suggestion to date the inscription at the time of Antoninus Pius (1950: 1502 n.24) rests on flimsy evidence. 213 See, e.g., Jones 1940: 138-144; Garnsey and Saller 1987: 28-34. 214 Libanius Or.2.33; Jones 1964:724-725. 215 ‘Notables’ is a modern term; they are referred to by such words as primates/proteuontes or possessores/ktetores, but the sources do not provide any legal definition and no precise term is employed consistently to refer to them. 216 See Jones 1964: 737-763; Liebeschuetz 2001: 104-136; contra Whittow 1990. Laniado 2002 is a meticulous re-examination of the available sources. 217 See Holum 1996, who is critical of Whittow’s interpretation. 218 Jones 1971: 546-547 (tables XXXIX-ILI).

210 E.g. Tsafrir 1996. But see Lewin 2002; 2007: 468-473; 478-480 arguing an organic growth of the Negev and the importance of the production of its wine. 211 Graf 2001: 220-223, with the literature there cited; cf. Parker 1999: 143. Note, however, the concentration of sites in the environs of Phaeno, an area known for its copper mines where condemned people were sent to work. See McQuitty 1988 and cf Millar 1984: 137-143.

178

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY Bittylius, a village located N of Gaza that was the native place of the historian Sozomenus.219 It is recorded as early as the decree of the Egyptian priests in honour of Ptolemy IV for his victory at the battle at Raphia in 217 BCE.220 It is then mentioned by Jerome in the late 4th c., in the biography of Hilarion, by Sozomenus and Julius Honorius in the 5th c., and is recorded on the border with Egypt in the map of Medaba in the 6th c. According to Theodosius, whose work on the Holy Land is dated at 530, it was situated 12 miles SW of Raphia, which accords its location in the map of Medaba.; according to Cyril of Scythopolis the site was located close to the sea shore, at a distance of 90 miles from Jerusalem (thus also Paul of Elusa). By the late 5th/early 6th c. it had a bishop, Theognius, and its bishop participated in the council of Jerusalem in 536. Both Hierocles and Georgius of Cyprus, writing their works in the 6th c. but probably deriving much of their information from official documents of the 5th c., include Bitulion in their lists of the cities of Palaetina Prima.221 It has been persuasively suggested to identify Bitulion with Tell esh-Sheikh, situated on the sea shore close to 3 km W of Sheikh Zuweid (about 16 km SW of Raphia). In his 2 month exploration of the site in 1913, J. Clédat found remains of several structures (one with an impressive mosaic pavement depicting Greek mythological themes) belonging (so he thought) to a fortress, a bath-house, a great number of other buildings, perhaps including some stores, pieces of marble statues, inscriptions in Greek, and a necropolis.222 In his view, these and many other structures he noticed along the way from El-Arish to Raphia date from the 2nd to the 4th c. The area of Sheikh Zuweid was surveyed in 1957 and the dates of the pottery collected range from the Hellenistic to the Early Arab periods; they include large amphorae from Africa Proconsularis and Tripolitania dated to the 3rd and, possibly, early 4th c.223 Now, the appointment of a bishop and the grant of city status by the late 5th c. would suggest that Bitulion had significantly expanded by that time, presumably thanks to its location close to the main road leading from Palestine to Egypt and the possession of an anchorage; of course, it could have always benefited from those advantages. However, one has to take into consideration that Paul of Elusa and Cyril of Scythopolis regarded it as a small city; perhaps other, unknown factors, and not only physical-demographic

growth, worked for the elevation of the status of Bitulion.224 A relevant case is that of Maiuma, the harbour of Gaza, which was granted by Constantine the status of polis, with the name Constantia, as a reward for its conversion to Christianity. Precisely for that reason Julian reversed Constantine’s decision in response to the appeal of the people of Gaza, although Maiuma retained its own bishop, clergy and festivals.225 But one should not conclude that the grant of polis status had no implications concerning the demographic-economic condition of Gaza-Maiuma. In a letter he sent to the vicarius of Asiana, Constantine recorded his assent to grant city status to the people of Orcistus (on the borders of Phrygia and Galatia) because of two main considerations: first, Orcistus had a large number of town-councillors and numerous citizens; second, the inhabitants were devoted to Christianity.226 In other words, the adherence of the people of Orcistus to Christianity was important, but a well-to-do class that could man the local curia and a large population formed a major consideration in Constantine’s decision to grant the village of Orcistus city status. This probably holds good for Gaza-Maiuma, as well for other cities. Bitulion, too, would not have been granted city status if it had not satisfied the criterion of minimal adequacy of population and people of substantial means. Let me now briefly review several cities listed by Hierocles and Georgius, mainly those that were granted the rank of a city in the Later Roman Empire and those that are little known from other sources. Sycomazon has been identified with Kh. Suq Mazen, a site situated about 17 km NE of Raphia and 2.5 km E of the main coastal road. Possibly mentioned in the Nessana papyri (Eamazon), it was a bishopric by 451, but is not recorded in any pre-5th c. sources. Sycomazon is most probably the Suq Mazay of the Aramaic translation of Numbers (34:15), a settlement on the SW border of Eretz Israel, and the name alludes to its having a market of some importance.227 Diocletianopolis is listed in Palaestina Prima and was possibly a bishopric by 359. To judge by Hierocles’ order of the cities, which is somewhat confused, it was situated S of Azotus and N of Gaza. Most scholars accept its identification with Saraphia, which is recorded as a city situated not far from AscalonMaiuma by Antoninus of Placentia and was a bishopric in

219 Confusion: Avi-Yonah 1976a: 43 (lumping together references to the 2 different sites); Jones 1971: 280 and 464 n. 73 (not realizing that Bethelea, a village in the territory of Gaza, cannot be identified with the city of Bitulion, situated on the border with Egypt). For Betheleah, apparently to be identified with modern Beit Lahia, see Alt 1926 241242; Abel 1940: 224-225; Figueras 1987-9: 122, and references in Tsafrir, Di Segni - Green 1994: 81-82. 220 Alt 1926: 239-240; Abel 1939: 227-228. 221 For references see Tsafrir, Di Segni - Green 1994:91. On the works of Hierocles and Georgius of Cyprus see Jones 1971: 514-521. 222 Clédat 1915. The researcher was unable to provide a plan of the fortress or to trace the line of the encircling wall, and had to admit that it perhaps had no towers. The structures ascribed to the fortress stretch over an area of about 115X75 m (1915: 21-22 with Fig.3). 223 Aharoni - Ben-Arieh 1974. On the amphorae in this region see also Clédat 1915: 18.

224

The suggestion that Bitylion extended to include an inland section, that is, Tell Jenin which is located at about 2.5 km SE of Tell eshSheikh (Aharoni - Ben-Arieh 1974), is disputed. According to Figueras (1987-9), Tell Jenin was the site of Boutaphis, recorded in the papyri of Theophanes who passed by this area ca.317-323, which is commonly identified with Bethaphu of Eusebius, a village 14 miles W of Raphia. It is possible, however, that Boutaphis/Bethaphu was annexed to Bitulion when the latter was granted the status of city; at any rate, later sources contain no mention of Boutaphis or Bethaphu. See also Barag 1973. 225 Eusebius, Vita Constantini 4.38-39; Sozomenus, Historia Ecclestiastica 5.3, 6-8; Glucker 1987: 43. 226 Millar 1977: 410. 227 For references see Tsafrir, Di Segni - Green 1994: 238. It is marked as a market site in Safrai 1994: 255, but with no discussion.

179

THE LATE ROMAN ARMY IN THE EAST 536. Saraphia, generally located at Kh. Esh Sheraf, was situated a few km S of Ascalon. Obviously, Diocletianopolis received its name and was granted city status under the tetrarchy.228 The settlement of Azotus Paralius, as distinct from inland Azotus (Azotus Mesogaeus), is documented as early as the 8th c. BCE, but it is difficult to find sure references to it prior to the Later Roman Empire, possibly with the exception of Ptolemy. Eusebius’ references clearly relate to inland Azotus, and it is noteworthy that for him it was a famous but small city.229 Azotus Paralius is attested as a bishopric in the 6th c. and is prominently marked on the Medaba plan. The results of the archaeological excavations of Azotus Mesogaeus that suggest its decline in the ‘Byzantine’ period support the view that Azotus Paralius was the more important city in that period, but there is no substantial evidence to confirm it.230 Ono was situated at a distance of 3 miles from Lod/Lydda (=Diospolis) according to a Talmudic tradition (5 miles according to another tradition). An Oxyrhynchus papyrus dated 291 records a town-councillour of Ono who came there to redeem Jewish slaves; his position is evidence to its having received city status by that time. Previously it was subject to Lod, which had been granted city status, more precisely colonia, by Septimius Severus in 199. The elevation of Ono to the rank of a city may well have been quite recent and, in any case, was subsequent to that of Lod.231 Maximianopolis, situated close to the Jewish village of Kefar ‘Otnai and the camp of the legion VI Ferrata, on the western border of the Esdraelon valley, was a bishopric by 325 (the Council of Nicaea). Obviously, it was granted the rank of a city under the tetrarchy. In addition to a necropolis, an oval-shaped theatre (71X52 m), which could also function as an amphitheatre, has been located but has never been excavated.232 Helenopolis was apparently established by Constantine and named in honour of his mother Helena. Situated in Palaestina Secunda, like Maximianopolis, it is supposed to be located at Kafr Kama, where remains of a 6th c. church and other buildings were excavated. It has been suggested that its territory included Mt. Tabor.233 Zoara, situated at the south end of the Dead Sea, is listed in the cities of Palaestina Tertia, and was a bishopric in the 5th-6th c. It is mentioned by Josephus, in the documents from the Babatha archive and many other sources, but is not recorded as a city before the 6th c.234 Augostopolis, also listed in Palaestina Tertia, had a

bishop by 431 (Council of Ephesus). Avi-Yonah suggested locating it at Tafila, close to 30 km SE of the Dead Sea,235 but thanks to the recently found Petra papyri it has been quite persuasively identified with Udruh, situated about 10 km E of Petra, as was suggested by F.M. Abel.236 If so, it is a settlement that was mentioned by Ptolemy in the 2nd c. (adrou) and probably by Stephanus of Byzantium (adara) and is recorded in the Beer Sheba Edict (adroa) as the town with the highest assessed amount among the settlements of S Jordan.237 The legionary fortress at this site was perhaps constructed about 300 and it seems that a civilian settlement soon developed as well. Presumably the name Augustopolis was given to it with the grant of city status.238 Arindela, also a bishopric, is persuasively located at Gharandal, some 12 km SE of Tafila.239 Georgius of Cyprus includes Beer Sheva in the list of the cities of Palaestina Tertia; that it had that status is not recorded by any other source. Jones suggested that in the past it had been under the jurisdiction of Eleutheropolis because it used the Eleutheropolitan era, and there is some evidence that seems to support his suggestion..240 Aila (Elath), listed by Georgius of Cyprus, was a bishopric by 325 (the Council of Nicaea). It is mentioned by many and various sources, but is not recorded as a city in any of them.241 To sum up: about a dozen of the cities listed by Hierocles and/or Georgius of Cyprus were granted their city status during the Later Roman Empire, which is an increase of some 30% in the number of the cities of Palestine. As explained above, various factors could be and were involved in the decisions of emperors to grant the rank of a city, and yet some minimal population and sufficiency of affluent citizens were required. Hence, this significant increase in the number of the cities is indicative of a certain degree of economic and demographic growth in the urban sector of Palestine. Although important by itself, such a conclusion is somewhat general and may be misleading, particularly in view of the fact that some of the old cities could have suffered from stagnation or decline, as is evident, for example, in the case of Azotus Mesogaeus. In other words, it is only when we have specific, detailed evidence on the fate of the individual cities that we can determine positively whether they flourished, declined or were in a state of stagnation. In several cases such evidence is available and can contribute to arrive at more refined, well-based and 235

Avi-Yonah 1976a: 33. Abel 1967, 2: 178; Fiema 2002: 209-210. 237 Ptol.Geog.5.17,5; Steph.Byz.18.18; Alt 1921: 8-10. The date of the Beer Sheva Edict is disputed 238 The identification of Augustopolis, probably a city by 431, with Udruh implies that the legionary base there had been established earlier than is assumed by Isaac 1995: 140-141. For the fortress, with suggested dating at 300, and the civilian settlement that developed within and around it see Kennedy 2000: 168-170. 239 Avi-Yonah 1976a: 31. 240 Jones 1971: 279. For references see Tsafrir, Di Segni - Green 1994: 75 and for a discussion see Di Segni 2004: 42-43, who does not explain why she infers that Beer Sheva did not have the rank of a city and was included in the territory of Haluza. 241 See Tsafrir, Di Segni - Green 1994: 59-60; Dolinka 2003.

228

236

Alt 1931; Tsafrir, Di Segni - Green 1994:112 (Diocletianopolis) and 222-223 (Saraphia). 229 Eus. On. 20.18: polichne episemos. Jerome’s insigne oppidum is a close translation, and there is no reason to think that his phrase means a big city and that he refers to Azotus Paralius. 230 Ptol.5.16,2; Dothan 1993: 102; Tsafrir, di Segni - Green 1994: 72. 231 See Schwartz 1991: 109 and 118 n. 91; Tsafrir, Di Segni - Green 1994: 198. 232 Tsafrir, Di Segni - Green 1994: 170; Segal 1995: 52-53 (the theatre); Tepper 2002 (a recent survey of the area). 233 Abel 1967, 2: 245 and 347; Tsafrir, Di Segni - Green 1994: 142. It is not clear why Avi-Yonah (1976a: 64) calls it “area”, but see also AviYonah 1966: 123. 234 Tsafrir, Di Segni - Green 1994: 263.

180

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY instructive conclusions concerning the demography and economy of the cities of Palestine in the Later Roman Empire, which in recent publications is considered as a period of great prosperity for them.242 As this paper has already outgrown an appropriate size, I’ll limit my discussion to 5 cities.

periods. Farms and villages surveyed in the environs and underwater finds off the shore indicate that Dor thrived as a regional centre for the agricultural hinterland and functioned as a port both in the ‘Roman’ and ‘Byzantine’ periods. A structure that has been identified as an installation for producing purple dye was in operation in the 2nd-3rd c. and the 6th c. Although the episcopal church survived until ca. 800, the authors conjecture that the city went into decline after the Arab conquest of Palestine, noting the abandonment of the adjacent farmouses during the 2nd half of the 6th c. at the latest.245

Dor: In his Onomasticon Eusebius described Dor/Dora as “now deserted”, which Jerome followed faithfully in his Latin version; also, according to the latter, the pilgrim Paula was amazed by “the ruins of Dor, a city once very powerful”.243 Such statements have seemingly been corroborated by the results of the yearly excavations of Tel Dor that E. Stern conducted since 1980. The findings show that Dor flourished in the ‘Roman’ period, but the city witnessed sharp decline by the 2nd quarter of the 3rd c., although some structures on the summit of the mound continued to be occupied in the ‘Byzantine’ period.244 However, thanks to a remarkable study of all the available sources (descriptions of 19th c. explorers and travelers; excavation reports, documents, maps and photos of archaeologists of the Mandatory Department of Antiquities; aerial photographs of the area, etc.), C. Dauphine and S. Gibson have been able to correct and modify persuasively that verdict of the settlement history of Dor in the ‘Byzantine’ period. The upper part of Tel Dor was indeed only sparsely inhabited in the ‘Byzantine’ period, which reasonably explains the statements made by Eusebius and Jerome. But in addition to the settlement on the mound, the traditional site of the city from the Middle Bronze Age IIA through the ‘Roman’ period, the lower parts of the mound began to be settled as early as the Persian period. Prominent elements of this lower city have been detected E and N of the mound; it seems that it was established as a pre-planned settlement during the ‘Early Roman’ period as an extension of the settlement on the mound. The upper city stretched over an area of about 6 ha., while the lower city eventually expanded to 6.8 ha. The known main elements of the bustling ‘Byzantine’ lower city include the following: encircling fortification wall whose date of construction cannot be ascertained, that is, ‘Roman’ or ‘Byzantine’; street network with a colonnaded cardo maximus (at least 10 m. wide), a principal east-west decumanus, and paved streets; various houses and structures, some of high level building quality; a theatre (with a diameter of about 60 m.), dating from the 2nd or 3rd c., which may have continued to be used for administrative functions in the ‘Byzantine’ period; an episcopal, large basilical church, whose first stage of construction occurred in the early 4th c.; water supply system consisting of an aqueduct that was probably constructed in the ‘Roman’ period but still functioned in the ‘Byzantine’ period, water cisterns and reservoirs. Structural remains testify to the existence of extra-mural suburbs at Dor both in the ‘Roman’ and ‘Byzantine’

Several comments are called for following Dauphin and Gibson’s impressive analysis and interpretation of the diverse sources. The expansion of the lower city in the ‘Early Roman’ period shows that Dor peaked demographically and, understandably, economically about the 2nd. c., when both its lower and upper sections were populated, with a total settled area of about 15 ha.246 Now, they accept that the sharp dwindling of the settlement on the mound began by the 2nd quarter of the 3rd c. and that by the time of Eusebius, and even more so at the time of Jerome and Paula, it had been largely abandoned. But the abandonment of the upper city is puzzling and the two scholars have not tried to explain it. Stern’s explanation, namely that Dor suffered because of the rise of Caesarea, is untanable for various reasons, notably the fact that the city reached the zenith of its expansion in the 2nd c., that is, after about 2 centuries of coexistence with Caesarea.247 At any rate, unless one posits that the population density in the lower city was gradually doubled concomitantly with the desertion of the upper city, which is extremely unlikely, we have to draw the conclusion that the population of Dor declined almost by half during the 3rd-4th c. Finally, since Dor’s agricultural hinterland had already become depopulated during the 2nd half of the 6th c., as has been shown by Gibson’s survey, the impact of that process must have been felt in the city itself; in other words, the decline of ‘Bysantine’ Dor, a city much smaller than ‘Roman’ Dor, had started by that time and should not be ascribed to the Arab conquest. Hippus (Susitha): established as a city under the Seleucids in the 2nd c. BCE, conquered by the Hasmonaeans, liberated from Jewish rule by Pompey, bestowed by Augustus upon Herod, and set free again 245

Dauphin - Gibson 1994-95. For a 38 m. long section of the aqueduct, dated probably not earlier than the 3rd c., see ‘Ad 1999, and for the water-supply system Berg, Sharon - Zilberstein 2002; Peleg 2002. For the theatre see also Segal 1995: 50-51 and for the survey of the hinterland of Dor Gibson, Kingsley - Clarke 1999. 246 Berg, Sharon - Zilberstein 2002: 155, an estimate only slightly higher than that of Dauphine and Gibson. 247 In the 2nd edition of his book Stern reiterates his view that Dor declined because of the rise of Caesarea and that it was a deserted city in the ‘Byzantine’ period (2000: 263-264 and 319-322). In the chapter he added to the original book he discussed at length (374-389) various finds of the ‘Roman’ period retrieved in the years 1993-1999, ignoring altogether the impressive expansion and existence of the lower city in the 2nd c.(but contrast Berg, Sharon - Zilberstein 2002: 155-157) as well as Dauphin and Gibson’s study, which is quite odd.

242

See, e.g., Walmsley 1996. Eus. Onom. 78: 9-13; Jerome Epist. 108. 244 Stern – Sharon 1995: 68 and next note. 243

181

THE LATE ROMAN ARMY IN THE EAST made of imported marble.250 The population of the ‘Byzantine’ period robbed the ‘Roman’ buildings to construct new buildings, notably the 4 churches known to-date, but also a variety of buildings along the decumanus maximus. Noteworthy are also shops and workshops that encroached upon the N and E sides of the Forum in the 6th-8th c.; constructed in a very haphazard manner, they even extended over the paved area. The water-supply system consisted of several aqueducts, cisterns and large water reservoirs, one on the S side of the Forum. The water source of the earliest aqueduct, dated to the Hellenistic period, was located at a distance of 4 km. from the city. Later, in the ‘Roman’ period, several more aqueducts were built, drawing water from as far as 25 km. from the city. All these were channeled together to feed an inverted siphon constructed upon the saddle E of the city, from which Hippus got its main supply of water. The 2 bath-houses of the city testify to the success of this system, an instructive demonstration of the technological and financial resources of the city authorities in the ‘Early Roman’ period. Finally, the conjectured existence of a theatre at the NW part of the city has yet to be proven by the spade.

after the latter’s death, Hippus is listed as one of the cities of the Decapolis by Pliny and mentioned as a city of Palestine by Eusebius. A bishopric by 359, it is listed as a city of Palaestina Secunda by Hierocles. These bare facts, known from literary sources and the city coins, mark the main stages in the political history of Hippus; for its economic and demographic history one has to rely entirely on the results gained in archaeological researches.248 The site of Hippus was explored by G. Schumacher in the late 19th c., and by C. Epstein and several other Israeli archaeolosits who carried out salvage excavations in the mid-20th c.; continuous yearly excavations started in 2000, headed by A. Segal. Situated on a plateau atop of a mount, about 2 km E of the Sea of Galilee, the city stretched over an area of ca. 8.7 ha. The fortification system consisted of an encircling wall 1550 m. long, 2 fortified gates, a few towers, artillery emplacements, and a ditch situated on the saddle E. of the city, at a major entrance to the city. It would appear that the fortifications were constructed at the early 1st c. BCE and were apparently maintained to the Early Arab period. The pre-planned settlement was built in the ‘Early Roman’ period; a grid system, with streets running at right angles to one another and forming the familiar insulae, characterizes the city plan. The colonnaded decumanus maximus, 4.2 m. wide and stretching for a length close to 600 m., was paved with rectangular flagstones; it has been calculated that the number of the columns on both sides of the street could have amounted to about 500. The Forum (42 X 42 m.), built in the 2nd. c. CE and situated across the decumanus maximus at the centre of the city, was paved with rectangular flagstones and was enclosed by colonnades on the eastern and northerns sides and by an open temple and a decorative gate-like structure on the western side. The remains of a Hellenistic compound were discovered N of the Forum. On the site of the Hellenistic temple that had been destroyed a new temple was built, possibly already in the 1st c. CE, which was dismantled much later to be replaced by a Christian church constructed in the late 5th/early 6th c. Close to this area was set up, in the ‘Byzantine’ period, an industrial complex that included wine- and oil-presses, which functioned to the 8th c. To judge by what is known about the Forum and the decumanus maximus and by the numerous architectural remains of various other buildings, the 2nd and 3rd c. were the main period of the efflorence of the city. One illuminative indication of the affluence of the citizenbody of the city is supplied by the large number of red and gray granite columns that were brought from Aswan and the Troad respectively,249 not to mention other numerous architectural elements (cornices, capitals, etc.)

The question that arises is whether the economic efflorescence of Hippus in the 2nd and 3rd c. lasted through the Later Roman Empire. The encroachment of stores and workshops upon public areas indicates the weakening of municipal control, perhaps already in the 5th c., which is also suggested by the plundering of ‘Roman’ buildings. These phenomena may be construed as evidence for the decline in the economic vigour and resources of the city and its population. It is possible, however, to interpret them as changes in the socialcultural life of the community, a proof that with the Christianization of the local population the values and ideals of the Classical city had been abandoned. Now, the economy, and flourishing, of Hippus was no doubt based on its links with its agricultural hinterland; the city territory extended over a large part of the southern Golan, an area that may be estimated at about 370 square km. Unfortunately, there is no study of the rural history of this district comparable to that of the Lower Golan, the adjacent NW district that was studied by Ben-David (above). There the growth of the settled area began in the early 2nd c. reaching culmination in the 4th c., or early 5th c., followed by a sharp decline in the 2nd half of the 5th c. that continued to the 7th c. The Northern Golan, the district extending NE of the territory of Hippos, witnessed a sharp reduction in the number of the sites settled in the ‘Byzantine’ period. The proximity of of these 2 districts to the territory of Hippus makes it reasonable to assume that it, too, underwent such a process. Altogether it seems, therefore, that Hippus witnessed not only a social-cultural change but also a process of economic stagnation or decline from sometime in the Later Roman Empire, perhaps the 5th c, although it

248

See Epstein 1993; Segal A., Mlynarczyk, Burdajewicz, Schuler M. Eisenberg M. 2004 and 2005; Tsafrir, Di Segni - Green 1993: 147 (references); Eisenberg 2003 (fortifications). For the water-supply system see Meshel Z., Fahlbusch H., Tsuk T. - Peleg Y. 1996; BenDavid 2002; Tsuk, Peleg, Fahlbuch - Meshel 2002. 249 Williams-Thorpe - Henty 2000: 164, 166 and 168; all the 18 columns studied were re-used for the building of a 5th c. church, the so-called cathedral of the city.

250 Fischer’s account of the use of marble in Hippus (1999: 61, 65, 68 and 260) needs now updating.

182

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY managed to survive quite well the Arab conquest until its destruction by the earthquake of 749.

in the early 6th c., but the nature of this renovation and the date of the original wall are not known. The fact that it encircled a huge area, about 135 ha., more than any other city of Palestine, suggests that it was constructed at a time when the settlement had outgrown by far the ‘Roman’ city, perhaps the early 5th c. When the city reached the peak of its expansion, late 5th - early 6th c., the total settled area, including that extending beyond the encircling wall, stretched over some 155 ha. Still, without further excavations, we cannot be sure that the whole area enclosed by the walls was densely settled. Examples of old buildings that were left in ruins or were destroyed in this period are the old basilica, the odeon that was destroyed in the early 6th c. and probably pagan temples. The magnificent nymphaeum, the portico with a decorative pool, colonnaded streets, a bath-house and the theatre serve as examples of buildings and monuments that were restored or preserved; also noteworthy is the hippodrome that was transformed into an amphitheatre. New buildings include the agora, churches and monasteries, the Palladius Street (over an older one), a bath complex, a commercial plaza (the Sigma, built in part on the site of the destroyed odeon), private dwellings and quarters, notably those on the slopes of the tel. Some of these as well as shops and private commercial and industrial installations that encroached on streets, public spaces and onto former monumental structures testify to the gradual transformation of the urban landscape and to profound social, cultural and religious changes. The continuous and at times very extensive building activity, both in the public and private spheres, also testifies to the economic strength and financial ability of the city and and its population. Very little specific evidence is available concerning the sources of money: in one case a donation by the Emperor Anastasius I for the construction of a large basilica (60X28.7 m), with an elaborate roofing system, and in another the financing by a notable of a portico of a bath-house; presumably the majority of the public buildings were financed by public money and of course private construction was based on private resources. However, by the mid-6th c the times of ambitious, grand building projects were over; whatever term is used to describe the conditions prevailing in the late 6th c. and early 7th c., stagnation, decline or economic and demographic contraction, it appears that the city’s prosperity had ended by that time. This change in the fortunes of Scythopolis is often and reasonably ascribed to damages, immediate or of long term effects, caused by the Samaritan revolt of 529, the plague of 541/2, earthquakes, the Persian conquest (about 614) and the Arab conquest of 635, but it is likely that other factors as well contributed to it. Although Scythopolis, Baysan in Arabic, remained a living city under the Umayyad rule, the gradual process of decline continued until its destruction by the disastrous earthquake of 749.

Scythopolis/Beth Shean: the written and archaeological sources about the population and settlement history of this city from the 1st through the 8th c. are abundant; particularly important are the material and epigraphic finds unearthed in the excavations conducted in the last 2 decades of the 20th c. The following account presents merely the salient characteristics of the main stages in the development of the city essential for this study.251 In the 2nd c., Scythopolis exhibited the urban landscape of the Greco-Roman city of the East at its best, although because of topographical restraints there were no decumanus maximus and cardo maximus and it was possible to implement the familiar orthogonal pattern with the usual rectangular insulae - only in several sections of the civic center. Paved with dressed basalt blocks, the wide streets were flanked by colonnades and covered roofs. The public buildings the city could boast of included a fine theatre adjacent to it was a cult complex with a central building, a group of altars and a nymphaeum, a basilica, a hippodrome, an odeon, a magnificent nymphaeum, temples, bath-houses, plazas, several other impressive monuments and 2 freestanding gates (W and NE), all richly decorated and many having various elements made of marble;252 some of these, like the cult complex, the basilica and the theatre, were built in the 1st c. and rebuilt in the 2nd c. The available epigraphic evidence indicates that this expensive building activity was financed both by public money and private donations. To what extent that public building went on after the 2nd c. is not clear, and it has been suggested that a slow down occurred during the adverse conditions of the mid-3rd c. At any rate, the maximal size of the area of the city in the 3rd c. has been estimated at 70-80 ha. Scythopolis was transformed to a large extent during the period from the later 4th to early 6th c. as a result of several processes: demographic growth, expansion of the settled area, building and rebuilding of new and old public monuments and commercial buildings, desertion or deliberate destruction of old buildings and the Christianization of the pagan population. The severe earthquake of 363 that destroyed many of the old buildings gave occasion for the introduction of changes in the urban landscape of the city, in part related to the rising power of Christianity. An outward expression of the extensive expansion of the city was a 2.5-3 m thick wall, about 4.8 km long, that incorporated the 2 freestanding gates and may have had, altogether, 6 gates. Two inscriptions (522/3 and 525/6) record its renovation 251

This summary account is mainly based on Tsafrir - Foerster 1997 and Mazor - Bar-Nathan 1998; for references see Tsafrir, Di Segni - Green 1993: 223-225. I cannot deal here with several bothering matters and questions that are still unsolved. 252 For the use and provenance of granaite columns in Scythopolis (Aswan, Troad and Kozak Dag) see Williams-Thorpe - Henty 2000: 164-166 and 169, and for marble Fischer 1999: passim, esp.49-50, 6163 and 74-79.

To sum up: to judge by what is known about the urban development of Scythopolis during the 1st – 8th c., the city witnessed 2 periods of particularly vigorous growth and prosperity: late 1st c. – early 3rd c.; later 4th to early 183

THE LATE ROMAN ARMY IN THE EAST 6th c. The conditions prevailing in the span of years between these 2 periods are not sufficiently known, but it seems that the 3rd c. crisis did not cause any significant decline, if at all, and evidently not irreversible. In contrast, the changing face of the city in the last 2 centuries of its existence, characterized by the invasion of private commercial and industrial installations into public spaces and the narrowing of the streets, as well as by the deterioration or remaining in ruins of old public monuments, went hand in hand with loss of municipal control and economic and demographic decline, ending in a natural disaster of which the city was not able to recover. Scythopolis was the regional centre of a fertile, rich agricultural area, it was also a manufacture centre famous for its flax and garment industry and it benefited from its becoming the capital of Palaestina Secunda in the late 4th c. The interconnection between these 3 factors and the prosperity, or decline, of the city, is more complex than straightforward, not the least because various other factors have to be taken into account, but this goes beyond the scope of this paper.

and to a network of 7 public roads, gradually constructed in the 1st-2nd c., that facilitated land communication with all the regions and cities of Palestine;255 it was the administrative centre of the Roman province and later an ecclesiastical centre; and it had a large fertile agricultural hinterland.256 The port of the city functioned to the end of the Roman rule, although the inner harbour and the breakwater, seriously damaged as early as the late 1st c., had been silted and submerged, respectively, by the early 4th c., and the middle and outer harbours needed a thorough, extensive reconstruction that was carried out by Anastasius I (491-518).257 In Hadrian’s time, a second channel was added to the original one, dating probably from Herod’s time,258 on the high level aqueduct that supplied Caesarea with water from the Shuni springs and from another group of springs at ‘Ammiqam, about 7 km NNE and 14 km. NE of the city respectively.259 The new channel more than doubled the quantity of the water supplied and it stands to reason that it was constructed to meet the needs of the growing population of the city. The low level aqueduct was constructed in the 1st half of the 4th c.260 and was fed from an artificial lake that was created by the damming of Crocodile River (Nahal Taninim), about 4 km N of the city. Again the water supply may have been doubled, although due to the deterioration of the older aqueduct the actual quantity it delivered must have decreased. Thus in 385 Flavius Florentius the governor of the province reconstructed one section of the High level aqueduct,261 and in the 530’s Stephanus the governor of Palaestina Prima had that aqueduct cleared, for which he was praised by Choricius of Gaza in an encomium dated to 534/6.262 At any rate, the grand project of the 4th c. testifies to a very substantial demographic increase suggesting concomitant expansion of the settled area. The construction of a new encircling wall in the late 4th or early 5th c. points to the same development.263 The area enclosed by the new wall has been estimated at either 95 ha. or 115 ha., and both estimates are curiously claimed to indicate an area two or three times larger than that of the ‘Roman’ city.264 This is an enormous expansion, and it is doubtful that all this area was densely occupied at the time of the construction

Caesarea: from an urban point of view, almost everything that took place in Caesarea under Roman rule from the 1st to the 7th c. can be seen as partly a continuation and expansion and partly alteration or transformation of the city that Herod had planned and built in the years 22-10 BCE. The written and archaeological evidence is enormous and no less copious is the scholarly literature dealing with it; only the bare minimum that is required for this study is presented here.253 Conspicuous in Josephus’ account of Herod’s Caesarea is the harbour, built by the use of advanced construction techniques, as has been confirmed and clarified by modern researces; other monuments and features marking the city from its origins, known partly from Josephus and partly from diggings, include the royal palace, a theatre, an amphitheatre, a temple to Augustus and Roma, erected on an elevated site, the grid pattern characteristic of the layout of the Greco-Roman cities of the East, probably an aqueduct, an elaborate underground sewer system and an encircling wall. Caesarea was the centre of the Roman rule of Palestine from beginning to end, capital of the province of Judaea, Syria Palaestina following the suppression of the Bar Kokhba Revolt, and of Palaestina Prima from the late 4th c. It was made a colony by Vespasian, although whether or not it received veterans of the Roman army is disputed,254 and a metropolis by Alexander Severus, and later on it was a bishopric.

255

Roll 1996. On the territory of Caesarea see Avi-Yonah 1966: 143-145; Levine 1975a: 5-6. 257 Hohlfelder 1992; Oleson 1996; Raban 1996 and 2004. 258 Pace Porath 2002: 125, who dates it to the 1st c. CE, not later than Vespasian 259 For inscriptions attesting the construction of the new channel by legionary detachments under Hadrian, of which some were engraved on the W wall of the channel, see Di Segni 2002: 47-51. 260 Porath 2002: 127. 261 Di Segni 2002: 60-61. 262 See Mayerson 1994: 267-270; Di Segni 2002: 61-62. 263 On this wall, whose date cannot be determined accurately, see Peleg - Reich 1992; Lehman 1995. For a recent discovery of a segment of the Herodian wall SW of the theatre, bearing on the identity of the wall at the N end of the city, see Poraht 2000: 36* with fig. 52. 264 The 95 ha. estimate goes back to Avi-Yonah (Broshi 2001: 101) and is repeated in Holum 2005: 91, with the “two to three times” comparison. For the 115 ha. estimate see Patrich 2001: 80 and 84, with the same comparison. For an estimate of 110-120 ha. see Lehman 1995:121. 256

Three factors constituted the mainstay of the flourishing of Caesarea: it was a commercial centre of the whole country thanks to its harbour, the largest in the country, 253 See Tsafrir, Di Segni - Green 1994: 94-96 (references); Segal 1995: 64-69 (the theatre); Raban - Holum 1996 (collection of papers); Holum, Raban - Patrich 1999 (collection of papers); Fischer 1999: passim (the use of marble); Patrich 2001; Porath 2002 and Peleg 2002 (the watersupply system). 254 See the conflicting opinions of Eck Forthcoming (affimative) and Isaac Forthcoming (negative).

184

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY of the wall. Two points are noteworthy. Various finds and considerations strongly suggest that the expansion was due to a long and gradual process beginning in the 2nd c.265 The Jews, for example, who had been massacred or fled in the First Jewish Revolt later returned to Caesarea and their number kept growing apparently to the 4th c.266 The same holds true for the Samaritans, and it is quite significant that the office of the Roman governor at Caesarea consisted mostly of Samaritans in the early 4th c.267 Second, reaching peak at about the late 4th c., the city continued to flourish to the early 6th c., although as a result of the division of Palestine into 3 provinces (prior to 409) the importance of its position as the administrative centre of the country decreased and later on the official recognition of the bishop of Jerusalem as a Patriarch in the mid-5th c. entailed a diminution in the ecclesiastical status of the see of Caesarea.268

Caesarea were left with no peasants to cultivate their agricultural estates, which understandably fell victim to the rage of the insurgents.273 Although Justinian extended relief, in particular reduction of taxes due for 2 consecutive years,274 that revolt must have been a major factor that contributed to a gradual procees of economic decline. If Caesarea was afflicted by the plague of 541, which is a likely possibility in view of its spread from Pelusium along the Palestinian coast,275 the considerable demographic loss will have generated immediate and enduring detrimental consequences for its economy. In conclusion, Caesarea seems to have steadily expanded, in terms of the area built and settled, size of population and trade, industrial and other economic activities, from the late 1st to at least the 5th c. According to one view, a process of decline had already started in the 5th c.,276 but according to another view Caesarea was still at the prime of its prosperity to the mid-6th c.277 To the present writer it seems that while expansion may well have come to a halt some time in the 5th c., it was not immediately followed by a process of decline; that more probably began in the 2nd third of the 6th c., presumably at a slow pace. The Arab conquest, however, resulted in a speedy demise of the city.

At its apogee, the city retained its original orthogonal plan, but some significant changes in the public landscape took place.269 The pagan temples had been destroyed and churches were constructed, notably the one that was erected on the former site of the Herodian temple to Augustus and Roma. The Herodian hippodrome/ amphitheatre and theatre, both situated close to the sea, and the oval amphitheatre in the NE zone of the city had been demolished by nature and man; that was eventually the fate of the eastern hippodrome, constructed in the 2nd c., although it continued to function presumably to the early 6th c. Despite the temporary deterioration in the maintenance of the aqueducts, which occasioned the construction of wells in some places, a network of terracotta and lead pipes continued to distribute the water supplied by the aqueducts to public and private fountains, bath-houses, latrines and gardens. The discovery of various private houses - notably luxurious mansions, apparently equipped with warehouses, in the area S of the Crusader wall, and of sub-urban villas in the NE zone of the city - attests the existence of very wealthy and wellto-do people.270 Warehouses and granaries, located at the inner harbour and in the area S of the Crusader wall and several other places,271 as well as huge amounts of pottery, amphoras and dolia, evidence thriving importexport trade at this period. Unfortunately, various methodological problems make it difficult to study the finds on the basis of a quantitative, chronological analysis.272 The flourishing of the city population suffered a severe blow by the Samaritan revolt of 529/30. It appears that as a result of the annihilation of a huge number of the Samaritans, the Christian landowners of

Petra: the finds retrieved in recent archaeological excavations and surveys have undermined the old view of an abrupt decline of Petra after the takeover of the Nabataean kingdom by Tajan; they also shed new light on changes and developments that occurred in the Later Roman Empire. Particularly valuable are the papyri found in the Petra Church in 1993 for the completely new evidence they provide concerning the economic and social conditions prevailing at Petra in the 6th c. Although there are many gaps in knowledge and uncertainties, these finds, together with the few available inscriptions, old and new, and some long known, not much informative literary sources, make it possible to present some general outline of the history of Petra under Roman rule.278 Petra continued to be a center of trade and manufacture after it succumbed to direct Roman rule in the early 2nd c. As was explained above, finds from ‘Avdat, Mamshit and ‘En Hazeva indicate that the international long distance trade of luxury goods that passed through the central Negev continued to the early 3rd c., diminishing and then terminating some time later. These data support finds at Petra showing that the city retained its position as

265

273

Cf. Levine 1975a: 8; Patrich 2001: 79-80. See Holum 1998: 163-169 with the literature there cited. 267 Levine 1975b: 107-112; Holum 1998: 169-171; Patrich 2001: 81. 268 On the conflict of the see of Caesarea with Jerusalem see Z. Rubin 1996. 269 On the public landscape of Caesarea in the early 4th c. see Patrich 2002. 270 Holum, Hohlfelder, Bull - Raban 1988: 181-184; Porath 1998: 39-44 and 2000 37*-39*; Patrich 2001: 95-96. 271 Patrich 1996. 272 Blakely 1996.

Procopius, Vita Arcana 11.24-30; Cyril of Scythopolis, Vita Sabae 70. For interpretation see Holum 1998: 171 and 2005: 100. 274 Cyril of Scythopolis, Vita Sabae 72-73. 275 For a recent discussion of this plague see Horden 2005, with the literature there cited. 276 See Levine 1975b: 136-139 and cf. Patrich 2001: 97. 277 Holum: 2005: 91. 278 I follow here mainly Fiema 2002, an admirable comprehensive and detailed analysis and interpretation of the archaeological data and written sources, with ample literature. See also Fiema 2001b, a shorten version, and Fiema 2001a.

266

185

THE LATE ROMAN ARMY IN THE EAST a prosperous metropolis of the international trade in aromatics and spices in that span of years. In the 2nd half of the 3rd c. that trade drastically decreased, with Petra becoming merely an ordinary provincial town. This change in the fortunes of Petra was due to a diminished demand in the targeted markets for the luxury goods exported by the Petraeans, a consequence of the anarchic conditions prevailing since the 230’s. Thereafter, some inter-regional trade was conducted with cities of Palestine and there was some import from the West in the 4th c., but when the disatrous earthquake of 363 occurred it damaged a city already with limited resources. The violent earthquake destroyed both domestic structures, notably at ez-Zantur, and public buildings, such as the Temple of the Winged Lions, the theatre, the Great Temple and those at the colonnaded street, as well as the dams that prevented the Siq from flooding by the winter rains. Most of the buildings remained in ruins and those rebuilt were, generally, only poorly or partially restored. That the population that lived in the city at that time had no interest in reconstructing the magnificant Nabataean monuments is quite understandable, but it is telling of the limited resources of the civic authorities and the declining euergitism of the urban elite that the repair of the installations that protected the Siq from flooding was insufficient and of limited efficiency. No less instructive is the fact that the city wall that was constructed sometime after 363 encompassed an area smaller than the one occupied in the ‘Roman’ period, and even within the enclosed area large parts were abandoned. Indeed during the 5th-7th c. occupation was gradually contracting. The Petra papyri, of which the dated ones cover the period from 537 to 592, mainly document possession of and transactions related to the real estates of some of Petra’s affluent citizens. The picture emerged from these documents reflects a society whose economic mainstay lay almost entirely on agricultural lands, with some urban possessions; information attesting trade or commerce hardly turns up in these papyri. This seems to accord the disappearance from the archaeological record of import finds after the 4th c. It is also doubtful that Petra was a significant centre of industry, considering the fact that the available information is limited to some ceramic production. Petra may have served as the local market of its agricultural hinterland, where the villages and farmhouses - with vine-yards, seed land and gardens known from the papyri could have still enjoyed some degree of prosperity. There are no finds, however, to suggest that it functioned as a centre for inter-regional exchange in goods. One aspect indicative of the changed preferences of the wealthy local elite and authorities is the fact that only churches, five in all, are known to have been constructed at Petra and its immediate environs in the ‘Byzantine’ period, but not even one civic monument. Church construction, even on a magnificent scale, need not imply a flourishing of urban life or demographic growth,279 and in the case of Petra of the 5th and 6th c. all the indications are that it does not. Finally, the abrupt 279

decline of Petra in the wake of the Arab conquest and its disappearance from historical record – there is no mention of it in accounts of the Arab takeover in contrast to other sites of S Jordan (Aila, Udruh, Rabba etc.) suggest that a process of decline, and not only stagnation, had started in the late ‘Byzantine’ period. Four stages may be distinguished in the history of Petra from the time it came under direct Roman rule: a prosperous city thriving thanks to its position as a metropolis of a long distance international trade, a position lost close to the mid-3rd c.; an ordinary provincial city, with limited resources and revenues (mainly accruing thanks to the agricultural hinterland and inter-regional trade) and abandonment of some residential areas – to the earthquake of 363; a period marked by significant contraction in the settled area, failure to repair satisfactorily vital public installations, diminishing or absence of inter-regional trade and industrial activitiy and apparently a gradual process of decline – from the late 4th to the early 7th c.; abrupt decline and desertion of the city following the Arab conquest. The main conclusion to be drawn from this survey of some twenty cities is that there was no one pattern of urban development in the Later Roman Empire. Generalizations to the effect that the cities of Palestine witnessed expansion and prosperity throughout the whole ‘Byzantine’ period misrepresent the reality that was diversed and complex. Rather, to a certain degree, each city had its own course of growth and decline, or stagnation; by simplication several patterns of development of cities may be suggested. a. Cities that flourished and/or were at the zenith of their expansion in the ‘Roman’ period, but had started, during the 3rd c., to witness contraction, stagnation or decline, each at its own time and pace, which continued to the end of the Roman rule in Palestine: Dor, Petra, Azotus Mesogaeus and SamariaSebaste.280 b. Cities that flourished from the 2nd c. or the Severan period to the 2nd half of the 4th c. and then either suffered abrupt, permanent eclipse in the aftermath of of the earthquake of 363, like Antipatris,281 or underwent, through the Early Arab period, a long, gradual process of stagnation or decline, like Hippus. c. Two periods of particularly vigorous growth and prosperity, with the span of years between them marked mainly by routine economic activity, stagnation or decline: Scythopolis: late 1st c.–early 3rd c., later 4th to early 6th c.; Apollonia: the Severan period, 6th to early 7th c.;282 Pella: 1st to late 2nd c., 6th-7th c.283

280

Schürer 1979: 162-164; Avigad 1993: 1307-1310. Roll - Ayalon 1989: 146-148 and 174-175; Beck - Kochavi 1993: 7071. 282 Roll - Ayalon 1989: 38-67 and 1993: 73. 283 Walmsley 1992; Smith 1993: 1179-1180. 281

See Di Segni 1999: 158.

186

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY d. Continuous, gradual growth from the 2nd c. to the Later Roman Empire with expansion and flourishing peaking from late 4th to early 6th c.: Caesarea, Sepphoris, 284 Haluza. e. A modest city in the 2nd–early 4th c., vigorous expansion in the aftermath of the Christianization of the Roman empire, zenith of growth in the 5th-6th c.: Jerusalem.285 f. Cities that flourished in the 4th-6th c. thanks to their position as centres of prosperous agricultural hinterland and export-import trade (unlike Jerusalem that its expansion and revenues were due to pilgrimage, imperial and other benefactions, that is, its status for Christianity): Gaza286 and Ascalon.287 g. cities that apparently did not expand to any significant extent, if at all, after the grant of city status, whether that was in the ‘Late Roman’ period, like Diocletianopolis, Maximilianopolis, Ono and Mamshit (?), or in the 5th-6th c., like Bitulion, Sycomazon, Beer Sheva and Gabae288; such nonactive economy probably characterized also some old cities during the Later Roman Empire, like Joppa and Raphia.

decrease in coin emissions in the 5th c. as compared to the 4th c., which, according to his interpretation, meant a sharp reduction in the number of coins in circulation in the 5th c. He construed this evidence as support and proof of his theory that there occurred a dramatic economic and demographic decline in Palestine in that century.289 In two subsequent studies, his findings were partly confirmed and partly rejected; his interpretation of the numismatic evidence has been shown to be invalid. It is true that the number of coins issued and supplied to Palestine in the 5th, 6th and 7th c. dropped considerably in comparison with the 4th c., but the quantity of the coins in circulation in the 5th c. and the first half of the 6th c. did not drop since old coins remained in use for 100-150 years. As is argued by G. Bijovsky, rather than indicating an economic decline, the monetary currency of the 5th c. is to be understood as a symptom of creative monetary activity.290 In a study based on a database consisting of 15,236 coins from 70 sites, mostly villages but including five cities, H. Gitler and D. Weisburd have confirmed the drop in the number of the coins supplied in the 5th c., but have also shown that in the subsequent two centuries the number remained more or less stable. The exceptional was the 4thc, so they argue, because of the unusual number of coins issued then.291 Gitler and Weisburd do not argue that because the 4th c. was exceptional in coin production, it was also exceptional in economic and demographic growth, and wisely so. Too many issues are involved in an attempt to show a causal correlation between coin production and economic and demographic growth. At any rate, the above survey and discussion do not support such a correlation.

Evidently, many cities expanded, witnessed vigorous economic activity and flourished during the whole length or much of the Later Roman Empire. Not a few cities, however, were already in decline by the late 3rd c. and a few others declined abruptly or set on a course of gradual decline half-way through the period. In addition, several other cities remained stagnant. Such varied development in the urban sector resembles phenomena noticed above in the rural sites; certainly this variety has to be taken into account in any analysis of the economic conditions prevailing in the country and of their correlation with other factors that affected its settlement history in the Later Roman Empire.

Let me start with the coinage evidence. In a study published in 1998, Z. Safrai argued, on the basis of the numismatic evidence he collected, that there was a sharp

Secondly, according to the commonly held view, following Avi-Yonah’s paper on this subject from 1958, Palestine owed its tremendous growth in the ‘Byzantine’ period to the Christianization of the empire. Thanks to its status as the Holy Land, emperors and wealthy Christian believers lavished huge sums of money upon the building of churches, monasteries and other ecclesiastical monuments; the multitudes of pilgrims who arrived in order to visit the holy sites spent their money in the country, gave donations and bought expensive relics. This influx of capital boosted the economy, and with the rise of prosperity the population multiplied.292 This explanation of the prosperity of the country has been recently challenged by D. Bar.293 His first main argument is that, although prosperity and population density reached peak in the ‘Byzantine’ period, the process of

284 Netzer - Weiss 1994; Weiss - Netzer 1997a and 1997b; Weiss Forthcoming. Note that the destruction associated with the earthquake of 363 did not halt the expansion and flourishing of the city. 285 Tsafrir Y., Di Segni L. - Green J. 1994: 145-146 (references and literature); Isaac 1999; Tsafrir 1999a and 1999b; Z. Rubin 1999; Avni 2005. 286 Glucker 1987; Mayerson 1994: 250-255 and 347-351; Di Segni 2004; Ashkenazi 2004. 287 Tsafrir, Di Segni - Green 1994: 110-111; Stager 1993: 110-111; Fuks 2001: 49-96; Dvorjetski 2001. 288 Schürer 1979: 164-165; Mazar 1988; Tsafrir, Di Segni - Green 1994: 126.

289 Safrai 1998. For rejection of this theory see, e.g., Parker 1999: 136137. 290 Bijovsky 2000-2. On the ‘shelf life’ of coins in Palestine hoards see Waner - Safrai 2001, and on the supply of coins in the 6th and 7th c. see Walmsley 1999. I cannot discuss their finds here. 291 Gitler - Weisburd 2005, a particularly valuable study, both on account of the size of the database but mainly the statistical and other techniques used to check the validity of the results obtained. 292 See, e.g., Avi-Yonah 1958: 41-49; Wilken 1992: 178-183; Cameron 1993: 178-179; Parker 1999: 136 and 169; Stemberger 2000: 82-85 (with some qualifications). 293 Bar 2004c.

Conclusion Before presenting my conclusions, two topics need to be dealt with here, perforce in brief: the connection between the coinage evidence and the economy of Palestine; the Christianization of the empire and the economic and demographic growth of the country.

187

THE LATE ROMAN ARMY IN THE EAST economic and demographic growth had started in the ‘Late Roman’ period, as early as the 2nd half of the 2nd c., and hence basically should not be attributed to the Christianization of the empire but to other factors.294 Secondly, the construction of churches had very little to do with the economic growth of the country: the memorial churches that were built thanks to outside donations constitute merely about 10 % of the archaeologically known churches;295 the building of churches during the first hundred years after the takeover of Palestine by Constantine took place mainly in the urban sector and hardly affected the countryside,296 where the vast majority of the population lived; the construction of churches, monasteries and hospices, financed as it was by foreign donors, had limited, short-term influence on the local economy; the Christianization of the countryside was a long process that occurred in the main from the mid-5th onwards, and it was only then that the majority of the community churches of the rural sector were built; the contribution of pilgrims to the local economy was limited, for pilgrims tended to restrict their travelling in the country, which was along well-defined roads, to holy places. In sum: “The building of churches and monasteries was not the reason for economic expansion…, but the reverse”. The hundreds of the ‘Byzantine’ churches “are the best evidence of the inhabitants’ financial cap capabilities”. According to Bar, the main reason for the prosperity of the country “is the fact that from the second century to the sixth century Palestine witnessed a period of calm in which wars, plagues and earthquakes were scarce”.297

necessarily confirm Bar’s claim that the main reason for the prosperity of the country was due to scarcity of wars, plagues and earthquakes. As explained at the beginning of this paper, the topic of (in)security, with which wars and invasions are associated, will be explored in the 2nd part of this study. It remains to see to what extent the generally held view about the flourishing of Palestine in the ‘Byzantine’ empire, including Bar’s claim that Palestine prospered from the ‘Late Roman’ period to the 6th c., can be substantiated. A preliminary remark is in order. The notions ‘flourishing’, ‘prosperity’, ‘stagnation’, ‘decline’, or their like, which abound in the works of archaeologists and historians who deal with Later Roman Palestine, are impressionistic, that is, they do not refer to really quantified data, and the chronological limits attributed to them are, more often than not, anything but precise. It is impossible to determine, for instance, whether the total production and revenues of a certain settlement or a rural area were greater in the 5th c. than in the 4th c. or vice versa. Similarly whether or not the level of prosperity of a city claimed to have flourished in the 4th-6th c. remained constant or varied during this span of years is a question that cannot be resolved confidently. It is therefore well to keep in mind that, given the available evidence and data, I have not been able to eliminate the usual imprecision involved in the use of these terms. The first main conclusion, which has already been proposed but should be highlighted now, is that diversity rather than uniformity marked the settlement and economy history of Palestine in the period under discussion and earlier too. Thus, the developmental diversity of the country is conspicuous in the 3rd c. If we had to rely solely upon written sources, Jewish and nonJewish as well, we would maintain that the whole of Palestine was severely afflicted by unbearable taxation, requisitioning of goods and livestock, billeting of soldiers, debasement of coins and inflation, a phenomenon which is supported by coinage finds, drought, famine, pestilence etc., and as a result witnessed such an economic and demographic crisis that both the countryside and the cities declined sharply. But these sources, as said before, do not reveal how the population reacted to the adverse conditions and whether and to what extent was able to overcome the difficulties. The available archaeological evidence indicates that certain areas and cities did indeed undergo decline or halt of growth and flourishing in the course of the 3rd c., which may well be interpreted as the adverse effects caused by the conditions described in the written sources; these include the area of Hanita-Achziv in W Galilee, the E Lower Galilee, the S coast of the Carmel, Dor, Beth Shean, Pella, Azotus Mesogaeus, Samaria-Sebaste and Petra. It may also be mentioned that three of the large settlements of the Negev (Haluza, Mamshit and ‘Avdat)

Earlier I argued that the contribution of pilgrimage to the economy of the Negev was restricted to a small number of sites along certain roads and that even in those places it was of secondary importance. On this point I find Bar’s argument persuasive, that is, not the whole country but only holy sites, notably Jerusalem and Bethlehem, benefited from the pilgrimage movement. I also concur, in the main, with his other major argument, namely that the outside capital invested in Palestine - lavish donations for the construction of Christian churches and other monuments, purchasing of relics etc. – can have had only a limited influence on the economy of Palestine. For example, if only few churches were built in the 4th c. (only nine are known for certain), the money expended on them cannot have been sufficient to boost economic activity in all the regions of the country, and certainly not in those areas where Jews and Samaritans formed the majority of the population.298 But this does not 294

Cf. Bar 2002 and 2004a. Bar 2004c: 313 and 2003: 406-408. For archaeologically known churches from the 4th to 7th c. see Ovadiah 1970; Ovadiah - De Silva 1981 and 1982; according to Tsafrir, Di Segni - Green 1994: 19, about 390 churches are recorded in their work, and see also the appended map of churches. 296 Cf. Armstrong 1967 (4th c. churches) and 1969 (5th and 6th c. churches). 297 Bar 2004c: 316; cf. 2004b: 7. 298 Churches dated by inscriptions: 4th c. - 1, 5th c. – 9; 6th c. – 38; 7th c. - 3 (Di Segni 1999: 160). For a slightly less extreme chronological 295

division of the churches known archaeologically see Ovadiah 1970: 193, but the 4th c. is the lowest one, with only 9 churches.

188

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY archaeological surveys. They depict the ‘Byzantine’ period as a uniform period, and if we were to accept their data at face value, it would follow that almost the whole country underwent a total settlement transformation at the early 4th c. continuing to flourish to the early 7th c. I have demonstrated that such a miraculous demographic and economic situation did not occur in very many cases. I am also confident that if we had results of more surveys conducted along the methodological principles advocated and practised by H. Ben-David and U. Leibner, we would be able to divide the settlement history of many more areas of Later Roman Palestine into several sub-periods. In other words, more areas (or maps of the Archaeological Surveys of Israel for that matter) would be assigned to one of the developmental patterns suggested above. It is quite possible that with more reliably and precisely dated sites at our disposal, there will be need to modify somewhat the patterns I have suggested, or to assign some of the settlements to the suggested patterns differently, but the obtaining of diverse patterns of development cannot be in doubt.

suffered setback as a result of the termination of the longdistance trade of luxury goods that passed through the Negev, but apparently managed to recuperate, radically reforming the economic basis of their livelihood. Other archaeological finds, however, testify to the expansion or flourishing of settlement in many other areas, e.g., Mt. Hermon, N Golan, Lower Golan, Upper Galilee, Nahalal, Menasseh hill counry, the hinterlands of Apollonia and Lod, Rosh Ha-’Ayin, Ziqim and Lachish. The same holds good for several cities, e.g., Sepphoris, Hippus, Caesarea, Antipatris, Maximianopolis, Ono and Diocletianopolis. Given the available evidence, I suggest that several developmental patterns characterized Later Roman Palestine. The first noteworthy pattern is the beginning of real decline or stagnation, in the later 4th c. or the early 5th c., of areas and cities which had been expanding or flourishing until that time, e.g., Mt. Hermon, N Golan, Lower Golan, Nahalal, Mishmar Ha-’Emeq, the eastern part of the Menasseh hill country, Hippus, Petra and apparently Maximianopolis, Ono and Diocletianopolis. A different developmental pattern marked settlements and cities whose expansion and flourishing had begun at some time in the ‘Roman’ period and lasted to the early 6th c., e.g., Caesarea, Sepphoris, Haluza, ‘Avdat and, possibly, Shivta. A third pattern is suggested by settlements and cities whose expansion began or was renewed in the 4th c. or early 5th c. and lasted to the 6th c., e.g., the hinterland of Dor, Beth Shean, Apollonia, Jerusalem, Azotus Paralius, Rehovot-in-the Negev, Sa’adon, Bitulion and Augustopolis.

The corollary of the prevailing of various, different patterns of settlement development is that it is incorrect to characterize the span of years from the mid-2nd c. through the 6th c. as an uninterrupted, continuous period of prosperity, as does D. Bar. No less misleading is the widely held view that demographic growth and flourishing economy obtained throughout Later Roman Palestine both in space and time. To be sure, it is very likely that in any given sub-period of the ‘Byzantine’ period, say the 50-year units applied by Ben-David, the aggregate settled area of the rural settlements and cities was larger, sometimes even much larger than in that of every sub-period of the ‘Roman’ period. It is also true that a certain number of rural settlements and cities succeeded in maintaining a stable level of economic activity and in not a few cases even a somewhat steady increase, although the rate of growth was most probably quite modest, insignificant by modern standards (see below). Yet, many other settlements and cities grew and decreased, and in addition some sites died out and others came into being. This means not only that a continuous shifting in the settlement profile of the country took place, but also that overall the diverse processes cancelled out the contradictory effects of each other. Put in another way, I propose that it is quite likely that no significant change, increase or decrease, in the aggregate settled area, the size of the population and the gross domestic product of Palestine occurred during the period from some time in the 1st half of the early 4th c. to the later 6th c.

Three more patterns deserve attention. First, many cities and rural settlements witnessed decline in the course of the 6th c., including both those whose decline had long been in progress and those whose decline just set in. To this pattern belong, in addition to those listed in the first pattern noted above, the hinterland of Dor, Caesarea, Samaria, the area S of R. Zered and probably Upper Galilee. A converse pattern marks settlements, areas and cities that flourished from the 5th to the 7th c., e.g., Pella, Apollonia, Ashkelon and Gaza and their rural hinterlands, and the large settlements and many other sites of the Negev. The third pattern is suggested by the radical drop in the number of sites dated to the Early Arab period in all the archaeological surveys. On several occasions I have expressed my doubts concerning the reliability or accuracy of these data, raising the possibility that the settlement decline of not a few of the surveyed areas had already started in the ‘Byzantine’ period. That the dating of the data relating to the the Negev is indeed mistaken, although in the opposite direction, has been argued by several archaeologists claiming either that the sites were and continued to be settled from the 5th/6th c. to the early 8th c., or that many or most of the sites were established in the wake of the Arab conquest. Either way, this would suggest another pattern of settlement history.

The understanding and explanation of the economic and demographic processes that took place in Later Roman Palestine involve many and various questions, both particular and more general. For instance, one needs to consider whether the same factors functioned in all the regions and both in the countryside and the urban sector. If different factors functioned in different regions and

It is important to stress here once more a fundamental weakness or limitation of the vast majority of the 189

THE LATE ROMAN ARMY IN THE EAST areas at the same time, and along the whole period, this too has to be tackled. This entails the question whether different factors of change generated similar results in different regions. In several cases I have postulated a close connection between the economic development of a city or a large settlement (Dor, Ashkelon, Gaza, Shivta) with that of the adjacent rural area, but the study of this topic, comprising as it does a variety of questions,299 may yield interesting and instructive results. If I am right that following the level of expansion achieved in the 1st half of the 4th c. the economy and population size of Palestine as a whole returned to a stationary state, this too has to be explained. Another set of questions concerns the impact made by external versus internal factors upon the demographic and economic situation of Palestine, including the question to what extent trade was internal or directed to foreign markets, which has bearing on the extent of the integration of the country into the economy of the Roman empire. These and other questions that I have not detailed may well be discussed with profit in connection with the topics at issue in the debate over the nature of the ancient economy, which is generally introduced, schematically and quite often inaccurately, as a controversy between the ‘modernist’ and ‘primitivist’ approaches of M.I Rostovtzeff and M.I. Finley respectively.300 There is no place here for such a discussion, only for a short account of the much debated topic of economic growth in the Roman empire.301

with the primary causes being “the political integration of the Mediterranean basin” and “prolonged peace”.304 Hopkin’s detailed causes and characteristics of this growth, notably technical innovations, are to be found among those listed by R.B. Hitchner and R. Saller, who both derive their lists from modern economic theories. The five basic factors of economic growth are the following: (1) trade and specialization, (2) capital investment in production, (3) improved technology, (4) education and training of people, (5) institutional framework favourable to economic activities.305 Hitchner argues that all these factors existed and operated effectively in the early Roman empire, and he enumerates signs and evidence that show, in his view, that economic growth occurred “on a scale sufficient to sustain “the advantages of wealth and luxury” for more than two centuries”.306 Saller emphasizes that over the period from 100 BCE to 200 CE, on which Hopkins focuses, the economic growth of the Roman empire amounted to 25%, which is 0.1% per year; that could be defined as significant from the perceptive of 100 BCE but totally insignificant by modern standards. He argues that some of the factors (2, 3 and 4) barely existed and that the effectiveness of all of them was limited and unsustainable. He would admit that the large potential market of the empire, the long periods of peace in many regions, the relatively low taxes and the political and legal system that protected property rights (and, one may add, reduced expenses), which are all characteristics of the Roman empire’s institutional framework, were capable of promoting economic growth. His conclusion is that “we might imagine a gentle growth curve for the Roman empire”.307

Economic growth may be defined as “the process whereby the wealth of a given community…increases through time; more specifically, the sustained increase in wealth over time (normally per year) measured in the real per capita production of goods and services”; one may add other elements, for instance, increase in consumption and improvement of welfare, but they introduce complications and are misleading and questionable.302 As is emphasized by R. Saller, it is important to distinguish between per capita growth in production and aggregate growth,303 and it is noteworthy that scholars, generally speaking, have dealt with the economic expansion of Later Roman Palestine from the perceptive of aggregate growth. According to K. Hopkins, the Roman empire achieved “modest, though significant economic growth”,

On the whole I find Saller’s arguments more persuasive,308 and yet it seems that his “gentle growth curve” can accommodate the kind of economic growth envisaged by Hitchner. As said before, they agree about the basic factors to be considered in the study of economic growth in the Roman empire, and it is also noteworthy that they are of the opinion that it had halted even before the decline of the 3rd c. Hopkins too accepts that economic growth had stopped by the 3rd c.309 Now, in their examination of the functioning of the causes of economic growth, these and many other scholars deal with the Roman empire as a whole. They refer to certain

299

See, e.g., de Ligt 1990 and 1991; Erdkamp 2001. Rostovtzeff 1957; Finley 1973 (and later editions). For a correction see Saller 2005. The origins of the controversy go back to the 19th century. 301 For three recent, useful collections of papers (with ample literature) concerning the ancient economy, including the topic of economic growth, see Mattingly - Salmon 1999; Scheidel - Von Reden 2002 and Manning - Morris 2005. Useful too, on the Roman econonmy, are Von Freyberg 1989, Lo Cascio 1991: 327-365 and Harris 1993. 302 Millett 2001: 19-27, cited at 19, with an exposition of modern economic theories of causes of growth. The definition of economic growth as “a rise in the average real income per head and a corresponding rise in population” (Hitchner 2005: 207) emphasizes the element of demographic increase but omits the crucial aspect of scale over time. 303 Saller 2005: 228-229. Although sometimes blurred, the distinction is usually recognized or tacitly assumed, e.g., Hopkins 1983: xv-xviii; Lo Cascio 2000: 77-78. 300

304 Hopkins 2002: 219-230, cited at 219. He is also of the opinion that such a growth took place in the Graeco-Roman world from about 1000 BCE to 200 CE (1983: xiv-xxi). See Millett’s discussion of the seven proposed components of this process; he rightly remarks that the “evidence is heavily biased towards the Roman empire in the first and second centuries AD” (2001: 27-31, cited at 31). 305 Saller 2005: 231-232. All these appear, formulated slightly differently, in Hitchner 2005: 208, with the addition of an advantageous physical environment. 306 Hitchner 2005: 208-219, cited at 219. 307 Saller 2005: 232-237. 308 I think Saller’s arguments and conclusion are not contradicted by what M. Corbier says, in her discussion of the 3rd c. economy, about improvement in farming methods and technological advances (2005: 400-406). 309 Hopkins 1980, 1983 and 2002.

190

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY fundamental differences between the western and eastern parts of the empire and allow for provincial or regional variations. Taking into account the diversity of the economic forms, M. Corbier has recently highlighted the “value of a regional or even micro-regional approach to the study of ancient economies”,310 an approach that is appropriate to Palestine.

Abel F.-M. 1967 Géographie de la Palestine, 1-2, Paris. ‘Ad U 1999 Nahsholim. In Excavations and Surveys in Israel 19: 23*-24*. Adan-Bayewitz D. 1993 Common pottery in Roman Galilee: a study of local trade, Ramat-Gan. Aharoni Y. - Ben-Arieh S. 1974 Survey between Raphia and the brook of Egypt. In ‘Atiqot 7: 88-94 (Hebrew, with English summary p.14*). Alföldi G.1985 The social history of Rome, London. Alt D.A. 1921 Die griechischen Inschriften der Palaestina Tertia westlich der Araba, Berlin. Alt D.A. 1926 Bitolion und Bethelea. In Zeitschrift des Deutchen Palästina-Vereins 49: 236-242 and 333335. Alt D.A. 1931 Diocletianopolis und Sariphea. In Zeitschrift des Deutchen Palästina- Vereins 54: 171182. Applebaum S. 1956 Mamshit. In Israel Exploration Journal 6: 262-263. Armstrong G.T, 1967 Imperial church building in the Holy Land in the fourth century. In Biblical Archaeologist 30: 90-102. Armstrong G.T. Fifth and sixth century church building in the Holy Land. In The Greek Orthodox Theological Review 14: 17-30. Ashkenazi Y. 2004 Sophists and priests in late antique Gaza according to Choricius the rhetor. In B. BittonAshkelony - A. Kofsky (ed.), Christian Gaza in late antiquity, Leiden: 195-208. Avigad N. 1993 Samaria (City). In E. Stern (ed.), The new encyclopedia of archaeological excavations in the Holy Land, 4, Jerusalem: 1300-1310. Avi-Yonah M. 1940 Map of Roman Palestine, 2nd ed., Jerusalem. Avi-Yonah 1958 The economy of Byzantine Palestine. Israel Exploration Journal 8: 39-51. Avi-Yonah M. 1964 Essays and studies in the historical geography of Palesatine,Tel Aviv (Hebrew). Avi-Yonah M. 1966 The Holy Land from the Persian to the Arab conquests (536B.C. to A.D. 640): a historical geography, Grand Rapids, Michigan. Avi-Yonah M. 1976a Gazeteer of Roman Palestine, Jerusalem. Avi-Yonah M. 1976b The Jews of Palestine. A political history from the Bar Kokhba war to the Arab conquest, New York. Avni G. 1992 Archaeological survey of Israel: map of Har Saggi northeast (225), Jerusalem. Avni G. 1996 Nomads, farmers, and town dwellers. Pastoralist-sedentist interaction in the Negev highlands, sixth – eighth centuries, Jerusalem. Avni G. 2005 The urban limit of Roman and Byzantine Jerusalem: a view from the necropolis. In Journal of Roman archaeology 18: 373-396. Baly C. 1935 S’baita. In Palestine Exploration Fund Quarterly Statement 65: 171-181. Bar D. 2002 Was there a 3rd-c. economic crisis in Palestine? In J. Humphrey (ed.), The Roman and Byzantine Near East III. Journal of Roman Archaeology Supplementary Series 49: 43-54.

Underlying my review of the settlement and economic history of Later Roman Palestine is the problem of economic growth. To some degree my findings accord with recent evaluations of the economic and demographic crisis of the 3rd c.,311 as well of the economy of the Later Roman Empire,312 particularly that of the East in the 5th6th c.313 However, it is by now clear that a wide range of factors have to be considered in order to be able to understand to what extent the processes that occurred in Palestine were connected to and influenced by its very existence within the Roman empire and local conditions respectively. It is also clear that no one factor, as important as it may have been, was decisive enough to bring about total economic growth, let alone similar economic and demographic results throughout the whole territory and the whole span of years of Later Roman Palestine. This holds true for all the explanations suggested so far as well as for the security factor, as I intend to show and explain in the second part of this study. This does not mean that there is no need of finding out or suggesting an explanation that is comprehensive and flexible enough to take account of the diverse developmental patterns and exceptional cases, as well to clarify the interconnectivity among the various factors. Bibliography Abel F.-M. 1935 Note sur Sbaïta. In Journal of Palestine Oriental Society 15: 7-11. Abel F.-M 1939 Les confines de la Palestine et l’Égypte sous les Ptolémées. In Revue Biblique 48: 207-236 and 530-548. Abel F.-M 1940 Les confines de la Palestine et l’Égypte sous les Ptolémées. In Revue Biblique 55-75 and 224-239.

the de de 49:

310

Corbier 2005: 439; cf. Ward-Perkins 2000: 363-365. See, e.g., Corbier 2005, where it is suggested (407) that the rural areas did not suffer from a real crisis, but that there were problems linked to exceptional events (famines, epidemics, etc.), some local or provincial, others more widespread. Her main conclusion is that the Roman model of economy, with its dominant characteristics (396-397), was basically stable. However, her evaluation of the negative aspects of inflation and of the outcomes of the perseverance of war from Marcus Aurelius onwards (428-430) comes closer to the traditional view. For the western part of the empire see Witschel 1999. 312 On the question of economic growth, or decline, in the agricultural sector in the 4th-5th c. see Whittaker 1998: 277-287. His focus is on the western empire, and he mainly questions whether the old view of the prevalence of decline is well grounded. For a balanced analysis of trade, industry and economic activity in the urban sector see Garnsey Whittaker 1998. 313 See Ward-Perkins 2000, who highlights the intensification of settlement and the high population density (321-324), as well as the prosperity of both the rural and urban sectors (342 and 354-355). 311

191

THE LATE ROMAN ARMY IN THE EAST Bar D. 2003 The christianisation of rural Palestine during late antiquity. In Journal of Ecclesiastical History 51: 401-421. Bar D. 2004a Frontier and periphery in late antique Palesatine In Greek, Roman, and Byzantine Studies 44: 69-92. Bar D. 2004b Geographical implications of population and settlement growth in late antique Palestine. In Journal of Historical Geography 30: 1-10. Bar D. 2004c Population, settlement and economy in Late Roman and Byzantine Palestine (70-641 AD). In Bulletin of the School of Oriental and African Sudies 67: 307-332. Barag D. 1973 The borders of Syria-Palaestina on an inscription from the Raphia area. In Israel Exploration Journal 23: 50-52. Barker G. 1991 Approaches to archaeological survey. In G. Barker - Lloyd J. (ed.), Roman landscapes. Archaeological survey in the Mediterranean region, London: 1-9. Barnes T.D. 1982 The new empire of Diocletian and Constantine, Cambridge, Mass. Baumgarten Y. 2004 Archaeological survey of Israel: map of Shivta (166), Jerusalem. Beck P. - Kochavi M. 1993 Aphek (in Sharon). In E. Stern (ed.), The new encyclopedia of archaeological excavations in the Holy Land, 1, Jerusalem: 62-72. Beit-Arie I. 2003 Archaeological survey of Israel: map of Tel Malhata (144), Jerusalem. Ben-David C. 1999 Settlement in “Lower Golan” during the Hellenistic, Roman and Byzantine periods. Ph.D.Diss., Ramat Gan. (Hebrew). Ben-David C. 2002 The aqueducts of Hippos-Susita. In D. Amit, J. Patrich - Y. Hirschfeld (ed.), The aueducts of Israel. Journal of Roman Archaeology Supplementary Series 46: 199-206. Berg J.E., Sharon I. - Zilberstein B. 2002 Water supply, distribution and disposal in Roman Dor. In D. Amit, J. Patrich - Y. Hirschfeld (ed.), The aqueducts of Israel. Journal of Roman Archaeology Supplementary Series 46: 155-167. Berman A, Barda L. - Stark H. 2005 Archaeological survey of Israel: map of Ashdod (84), Jerusalem. Berman A., Stark H. - Barda L. 2004 Archaeological survey of Israel: map of Ziqim (91), Jerusalem. Bijovsky G. 2000-2 The currency of the fifth century C.E. in Palestine – some reflections in light of the numismatic evidence. In Israel Numismatic Journal 14: 196-210. Blakely J.A. 1996 Toward the study of economics at Caesarea Maritima. In A. Raban - K.G. Holum (ed.), Caesarea maritima: a retrospective after two millennia, Leiden: 327-346. Broshi M. 1979 The population of western Palestine in the Roman-Byzantine period. In Bulletin of the American Schools of Oriental Research 236: 1-10. Broshi M. 2001 Bread. Wine, walls and scrolls, Sheffield: 86-92.

Bruins H.J. 1986 Desert environment and agriculture in the central Negev and Kadesh-Barneaduring historical times, Nijkerk, The Netherlands. Bruins H.J. 1994 Comparative chronology of climate and human history in the southern Levant from the Late Chalcolithic to the Arab period. In O. Bar-Yoseph R.S. Kra (ed.), Late Quaternary chronology and paleoclimates of the easren Mediterranean, Tuscon, Arizona: 301-314. Brunt P.A. 1988 Roman imperial themes, Oxford. Callu J.-P. 1969 La politique monétaire des Empereurs romains de 238 à 311, Paris. Callu J.-P. 1975 Approches numismatiques de l’histoire du IIIe siècle (238 à 311). In H. Temporini - W. Haase (ed.), Aufstieg und Niedergang der römische Welt. II.2, Berlin-New York: 594-613. Cameron A. 1993 The later Roman empire, London. Campbell J.B. 1984 The emperor and the Roman army 31 BC – AD 235, Oxford. Carrier J.-M. 1978 Les finances militaries et le fait monetaire dans l’Empire romain tardif. In G. Vallet (ed.), Les ‘devaluations’ à Rome: époque républicaine et imperiale, Rome: 227-248. Carrier J.-M. 1986 L’esercito: transformazione funzionali ed economie locali. In A. Giardina (ed.), Società romana e impero tardoromano,1, Roma-Bari: 449488. Chen D. 1981 Byzantine architects at work in Mampsis and Sobota, Palaestina Tertia. In Liber Annuus Studii Biblici Franciscani 31: 235-244. Chen D. 1985 Byzantine architects at work in Oboda, Nessana and Rehovot, Palaestina Tertia. In Liber Annuus Studii Biblici Franciscani 35: 291-296. Chen D. 1990 On planning of synagogues and churches in Palaestina: a comparison with Syria and Illyria. In G.C. Bottini, L. Di Segni, E. Alliata (ed.), Christian archaeology in the Holy Land. New discoveries, Jerusalem: 523-533. Clédat J. 1915 Fouilles à Cheikh Zouède. In Annales du Service des Antiquités de l’ Égypte 15: 15- 48. Cohen R. 1966 Mezad Yeruham. In Israel Exploration Journal 16: 123-124. Cohen R. 1981 Archaeological survey of Israel: map of Sede-Boqer – east (168), Jerusalem. Cohen R. 1985 Archaeological survey of Israel: map of Sede-Boqer – west (167), Jerusalem. Cohen R. 1993 Hellenistic, Roman, and Byzantine sites in the Negev hills. In E. Stern (ed.), The new encyclopedia of archaeological excavations in the Holy Land, 3, Jerusalem: 1135-1145. Colt H.D. (ed.), 1962 Excavations at Nessana, I, London. Corbier M 1978 Devaluations et fiscalité. In G. Vallet (ed.), Les ‘devaluations’ à Rome: époque républicaine et imperiale (Rome): 273-309. Corbier M. 2005 Coinage, society and economy. In A.K. Bowman, P. Garnsey - A. Cameron (ed.), The Cambridge Ancient History, 2nd ed. XII, Cambridge: 393-439. Crawford M.H. 1975 Finance, coinage and money from the Severans to Constantine. In H. Temporini 192

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY Dolinka B.J. 2003 Nabataean Aila (Aqaba, Jordan) from a ceramic prospective. BAR IS 1116, Oxford. Dothan M. 1993 Ashdod. In E. Stern (ed.), The new encyclopedia of archaeological excavations in the Holy Land, 1, Jerusalem: 93-102. Duncan-Jones R. 1990 Structure and scale in the Roman economy, Cambridge. Duncan-Jones R. 1994 Money and government in the Roman empire, Cambridge. Dvorjetski, E. 2001 The economic activity and special agricultural production of Ashqelon from the Hellenistic to the Byzantine periods. In A. Sasson, Z. Safrai - N. Sagiv (ed.), Ashkelon: a city on the seashore, Tel Aviv: 99-118 (Hebrew). Eck W. Forthcoming The presence, role, and significance of Latin in the epigraphy and culture of the Roman Near East. In H.M. Cotton, R.G. Hoyland. J.J. Price - D.J. Wasserstein (ed.), From Hellenistic to Islam: cultural and linguistic change in the Roman Near East. Eisenberg M. 2003 The fortifications of Antiochia Hippos (Sussita) in light of the Hellenistic-Roman city fortifications in the Roman east. M.A. Diss, Haifa (Hebrew). Epstein C. 1993 Hippos (Sussita). In E. Stern (ed.), The new encyclopedia of archaeological excavations in the Holy Land, 2, Jerusalem: 634-638. Erdkamp P.P.M. 2001 Beyond the limits of the ‘consumer city’. A model of the urban and rural Economy in the Roman city. In Historia 50: 332-356. Erickson-Gini T. 1996 Mamshit (Mampsis) Excavations and Surveys in Israel 15: 108-110. Erickson-Gini T. 1999 Mamshit (Mampsis) – 1994. Excavations and Surveys in Israel 19: 91*-92*. Erickson-Gini T. 2002 Nabataean or Roman? Reconsidering the date of the camp at Avdat in light of recent excavations. In P. Freeman, J. Bennett, Z.T. Fiema - B. Hoffmann (ed.), Roman Frontier Studies 2000. BAR IS 1084, Oxford: 113-130. Erickson-Gini T. 2004 Crisis and renewal – settlement in the central Negev in the third and fourth centuries C.E., with emphasis on finds from recent excavations in Mampsis, Oboda and Mezad ‘En Hazeva. Ph.D.Diss., Jerusalem. Fabian P. 1995 The late-Roman military camp at Beer Sheba: a new discovery. In J. Humphrey (ed.), The Roman and Byzantine Near East: Recent archaeological research. Journal of Roman Archaeology Supplementary Series 14: 235-240. Fabian P. 2005 Avdat during the establishment of Provincia Arabia: the Roman military camp and its position in the eastern defense system of the empire. Ph.D.Diss. Beer-Sheva, (Hebrew with English abstract). Fabian P. - Goren Y. 2002 A new type of a late Roman storage jar from the Negev. In J.H. Humphrey (ed.), The Roman and Byzantine Near East III. Journal of Roman Archaeology Supplementary Series 49: 145153.

W. Haase (ed.), Aufstieg und Niedergang der römische Welt. II.2 (Berlin - New York): 560-593. Crawford M.H. 1978 Ancient devaluations: a general theory. In G. Vallet (ed.), Les ‘devaluations’ à Rome: époque républicaine et imperiale (Rome): 147-158. Dabrowa E. 1993 Legio X Fretensis, Stuttgart. Dabrowa E. 2000 Legio X Fretensis. In Y. Le bohec - C. Wolff (ed.), Les legions de Rome sous le HautEmpire. Collection du Centre d’Études Romaines et Gallo-Romaines. Nouvelle série, no 20, Lyon: 317325. Dagan Y. 1992 Archaeological survey of Israel: map of Lachish (98), Jerusalem. Dan Y. 1982 Palaestina salutaris (tertia) and its capital. In Israel Exploration Journal 32: 132-137. Dan Y. 1985 The Byzantine rule. In M.D. Herr (ed.), The history of Eretz Israel: the Roman-Byzantine period, Jerusalem: 233-274 (Hebrew). Dar S. 1993 Settlements and cult sites on Mount Hermon, Israel: Ituraean culture in the Hellenistic and Roman periods. BAR IS 589, Oxford. Dauphin C. 1987 Le catalogue des sites byzantines de la Palestine: buts, méthodes et limites d’une étude démographique. In Eretz-Israel 19: 2*-9*. Dauphin C. 1998 La Palestine byzantine: peuplement et populations, 1-3. BAR IS 726, Oxford. Dauphine C. 1999 The Birth of a New Discipline: Archaeological Demography. In Bulletin of the AngloIsrael Archaeological Society 17: 77-91. Dauphin C. 2003 Review article: Rural landscape, settlement archaeology and political ideology. In Bulletin of the Anglo-Israel Archaeological Society: 21: 65-86. Dauphin C. - Gibson S. 1994-95 The Byzantine city of Dor/Dora discovered. In Bulletin of the Anglo-Israel Archaeological Society 14: 9-38. de Ligt L. 1990 and 1991 Demand, supply, distribution: the Roman peasantry between town and countryside: rural monetization and peasant demand. In Münsterische Beiträge zur antiken Handelgeschichte 9: 24-56 and 10: 33-77. de St Croix G.E.M 1981 The class struggle of the ancient Greek world, London. Develin R. 1971 The Army pay rise under Septimius Severus and Caracalla and the question of Annona militaris. In Latomus 30: 687-695. Di Segni L. 1999 Epigraphic documentation on building in the provinces of Palaestina and Arabia, 4th -7th c. In J.H. Humphrey (ed.), The Roman and Byzantine Near East II. Journal of Roman Archaeology Supplementary Series 31: 149-178. Di Segni L. 2002 The water supply of Roman and Byzantine Palestine in literary and epigraphical sources. In D. Amit, J. Patrich - Y. Hirschfeld (ed.), The aqueducts of Israel. Journal of Roman Archaeology Supplementary Series 46: 37-66. Di Segni L. 2004 The territory of Gaza: notes of historical geography. In B. Bitton-Ashkelony - A. Kofsky (ed.), Christian Gaza in late antiquity, Leiden: 41-60. 193

THE LATE ROMAN ARMY IN THE EAST Garnsey P. 1988 Famine and food supply in the GraecoRoman world: responses to risk and crisis, Cambridge. Garnsey P. 1998 Cities, peasants and food in classical antiquity, Cambridge. Garnsey P. - Saller R. 1987 The Roman empire: economy, society and culture, London. Garnsey P. - C.R. Whittaker 1998 Trade, industry and the urban economy. In A. Camron - P. Garnsey (ed.), The Cambridge Ancient History, XIII, Cambridge: 312337. Gazit D. 1996 Archaeological survey of Israel: map of Urim (125), Jerusalem. Gibson S., Kinsley S. - Clarke J. 1999 Town and country in the southern Carmel: report on the Lanscape Archaeology Project at Dor. In Levant 31: 71-121. Gichon M. 1991 When and why did the Romans commence the defence of southern Palestine? In V.A. Maxfield - M.J. Dobson (ed.), Roman Frontier Studies 1989, Exeter: 318-325. Gichon M. 2002 45 years of research on the limes Palaestinae – the findings and their assessment in the light of the criticisms raised (C 1st – C4th). In P. Freeman, J. Bennett, Z.T. Fiema - B. Hoffmann (ed.), Roman Frontier Studies 2000. BAR IS 1084, Oxford: 185-201. Gitler H. - Weisburd D. 2005 Coin finds from villages in Palestine during the late Roman and Byzantine periods (A.D. 383-696/7): a quantitative examination of monetary distributions. In J. Lefort, C. Morrison J.-P. Sodini (ed.), Les villages dans l’empire Byzantine (IVe-XVe siècle), Paris: 539-552. Glucker C.A.M. 1987 The city of Gaza in the Roman and Byzantine periods. BAR IS 325, Oxford. Gophna R. - Ayalon E. 1998 Archaeological survey of Israel: map of Herzliyah (69), Jerusalem. Gophna R. - Beit-Arie I. 1997 Archaeological survey of Israel: map of Lod (80), Jerusalem. Govrin Y. 1991 Archaeological survey of Israel: map of Yattir (139), Jerusalem. Graf D. 2001 Town and countryside in Roman Arabia during late antiquity. In T.S. Burns - J.W. Eadie (ed.), Urban centers and rural contexts in late antiquity, Ann Arbor: 219-240. Griffin M. 2000 Nerva to Trajan. In A.K. Bowman, P. Garnsey - D. Rathbone (ed.), Cambridge ancient history, 2nd ed. XI, Cambridge: 84-131. Hadas-Lebel M. 1984 La fiscalité dans la literature rabbinique jusq’à la fin du iiie siècle. In Revue des Études Juives 143: 5-29. Haiman M. 1986 Archaeological survey of Israel: map of Har Hamran – southwest (198), Jerusalem. Haiman M. 1989 Preliminary report of the western Negev highlands emergency survey. In Israel Exploration journal 39: 173-191. Haiman M. 1991 Archaeological survey of Israel: map of Mizpe Ramon - southwest (200), Jerusalem. Haiman M. 1993 Archaeological survey of Israel: map of Har Hamran – southeast (199), Jerusalem.

Faust A. 2005 The settlement of Jerusalem’s western hill and the city’s status in Iron Age II revisited. In Zeitschrift des Deutcshen Palästina-Vereins 121: 97118. Fiema Z.T. 2001a Historical conclusions. In Z.T. Fiema, C. Kanellopoulos, T. Waliszewski, and R. Schick (ed.), The Petra church, Amman: 425-436. Fiema Z.T. 2001b Byzantine Petra – a reassessment. In T.S. Burns - J.W. Eadie (ed.), Urban centers and rural contexts in late antiquity, Ann Arbor: 111-134. Fiema Z.T 2002 Late-antique Petra and its hinterland: recent research and new interpretations. In J.H. Humphrey (ed.), The Roman and Byzantine Near East III. Journal of Roman Archaeology Supplementary Series 49: 191-253. Figueras P. 1987-9 Bitylion and Boutaphis. Israel – People and Land 5-6: 121-124 (Hebrew, with English summary 16*). Finkelstein I. 1995 Living on the fringe. The archaeology and history of the Negev, Sinai and neighbouring regions in the Bronze and Iron ages, Sheffield. Finkelstein I - Magen Y. 1993 Archaeological survey of the hill country of Benjamin, Jerusalem. Finkelstein I., Lederman Z. - Bunimovitz S. 1997 Highlands of many cultures. The southern Samaria survey: the sites, Tel Aviv. Finkelstein I. - Perevolotsky A. 1990 Processes of sedentarization and nomadization in the history of Sinai and the Negev. In Bulletin of the American Schools of Oriental Research 279: 67-88. Finley M.I. 1973 The ancient economy, Berkeley-Los Angeles. Fischer M.L. 1999 Marble studies: Roman Palestine and the marble trade, Konstanz. Foss, C. 1995 The Near eastern countryside in late antiquity: a review article. In J. Humphrey (ed.), The Roman and Byzantine Near East: Recent archaeological research. Journal of Roman Archaeology Supplementary Series 14: 213-234. Frankel R. - Getzov N. 1997 Archaeological survey of Israel: map of Akhziv and Hanita (1,2), Jerusalem. Frankel R., Getzov N., Aviam M. - Degani A. 2001 Settlement dynamics and regional diversity in ancient Upper Galilee. Archaeological survey of Upper Galilee, Jerusalem. Frumkin A. 1997 The Holocene history of Dead Sea levels. In T.M. Niemi, Z. Ben Avraham - J.R. Gat (ed.), The Dead Sea: the lake and its setting, New York-Oxford: 213-234. Frumkin A., Magaritz M., Carmi I. - Zak I. 1991 The Holocene climatic record of the salt caves of Mount Sedom. In The Holocene 1: 191-200. Fuks G. 2001 A city of many seas: Ashkelon during the Hellenistic and Roman periods, Jerusalem (Hebrew). Gal Z. 1991 Archaeological survey of Israel: map of Gazit (46), Jerusalem. Gal Z. 1998 Archaeological survey of Israel: map of Har Tavor (41); map ‘En Dor (45), Jerusalem

194

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY Holum K.G. 1998 Identity and the late antique city: the case of Caesarea. In H. Lapin (ed.), Religious and ethnic communities in later Roman Palestine, Bethesda, MD: 157-177. Holum K.G. 2005 The classical city in the sixth century: survival and transformation. In M. Maas (ed.), The Cambridge companion to the age of Justinian, Cambridge: 87-112. Holum K.G., Hohlfelder R.L., Bull R.J. - Raban A. 1988 King’s Herod dream: Caesarea on the Sea, New York. Holum K.G., Raban A. - Patrich J. 1999 Caesarea papers II. Journal of Roman Archaeology Supplementary Series 35. Hopkins K. 1980 Taxes and trade in the Roman empire (200 BC-AD 400). In Journal of Roman Studies 70: 101-125. Hopkins K. 1983 Introduction. In P. Garnsey, K. Hopkins and C.R. Whittaker (ed.), Trade in the ancient economy, Berkeley-Los Angeles: i-xxv. Hopkins K. 2000 Rents, taxes, trade and the city of Rome. In E. Lo Cascio (ed.), Mercati permanenti e mercati periodice nel mondo romano, Bari: 253-267 Hopkins K. 2002 Rome, taxes, rents and trade. In Scheidel W. – Von Reden S. (eds), The Ancient Economy, Edinburgh: 191-230. Horden P. 2005 Mediterranean plague in the age of Justinian. In M. Maas (ed.), The Cambridge companion to the age of Justinian, Cambridge: 134160. Isaac B. 1990 The limits of empire. The Roman army in the east, Oxford. Isaac B. 1995 The army in the late Roman east: the Persian wars and the defense of the Byzantine provinces. In A. Cameron (ed.), The Byzantine and early Islamic near east. III States, resources and armies, Princeton: 125-155. Isaac B. 1999 Jerusalem from the Great Revolt to the reign of Constantine 70-312. In Y. Tsafrir - S. Safrai (ed.), The history of Jerusalem: the Roman and Byzantine periods (70-638 CE), Jerusalem (Hebrew): 1-14. Isaac B. Forthcoming. Latin as the language of the empire in cities of the Near East: was Caesarea Maritima a veteran colony or a titular colony? In H.M. Cotton, R.G. Hoyland, J.J. Price - D.J. Wasserstein (ed.), From Hellenistic to Islam: cultural and linguistic change in the Roman Near East. Israeli S. 1994 Mamshit – 1999 Excavations and Surveys in Israel 12: 101-106. Issar A.S. 1998 Climate change and history during the Holocene in the eastern Mediterranean Region. In A.S. Issar - N. Brown (ed.), Water, enivironment and society in times of climatic change, Dordrecht: 113128. Issar A. S. - Govrin Y. 1991 Climatic changes and the desertification of the Negev at the end of the Byzantine period. In Cathedra 61: 67-81 (Hebrew). Issar A.S., Govrin Y., Geyeh M.B., Wakshal E. - Wolf M. 1992 Climate changes during the Upper Holocene

Haiman M. 1995 Agriculture and nomad-state relations in the Negev desert in the Byzantine and early Islamic periods. In Bulletin of the American Schools of Oriental research 297: 29-53. Haiman M. 1999 Archaeological survey of Israel: map of Har Ramon (203), Jerusalem. Haiman M. 2003 The 10th century B.C. settlement of the Negev highlands and Iron Age rural Palestine. In A.M. Maier, S. Dar - S. Safrai (ed.), The rural landscape of ancient Israel. BAR IS 1121, Oxford: 71-90. Hamel G. 1990 Poverty and charity in Roman Palesatine, Berkeley. Harris W.V. 1993 Between archaic and modern: some current problems in the history of the Roman economy. In W.V. Harris (ed.), The inscribed economy. Production and distribution in the Roman empire in the light of instrumentum domesticum. Journal of Roman Archaeology Supplementary Series 6: 11-29. Hartal M. 1989 Northern Golan heights: the archaeological survey as a source of regional history M.A. Diss., Qatzrin (Hebrew). Hartal M. 2001 The material culture of northern Golan in the Hellenistic, Roman and Byzantine Periods. Ph.D. Diss.Jerusalem, Qatzrin (Hebrew). Hendy M. 1985 Studies in the Byzantine monetary economy c. 300-1450, Cambridge. Hirschfeld Y. 1997 Farms and villages in Byzantine Palestine. In Dumbarton Oak Papers 51: 33-71. Hirschfeld Y. 2002 Deir Qal’a and the monasteries of western Samaria. In J.H. Humphrey (ed.), The Roman and Byzantine Near East. III. Journal of Roman Archaeology Supplementary Series 49: 155-189. Hirschfeld Y. 2003 Social aspects of the late-antique village of Shivta. In Journal of Roman Archaeology 16: 395-408. Hirschfeld Y. 2004 A climatic change in the early Byzantine period: some archaeological evidence. In Palestine Exploration Quarterly 136: 133-149. Hirschfeld Y. 2005 The expansion of rural settlement during the fourth-fifth centuries CE in Palestine. In J. Lefort, C. Morrisson - J.-P. Sodini (ed.), Les villages dans L’empire Byzantine (IVe-XVe siècle), Paris: 523537. Hitchner R.B. 2005 “The advantages of wealth and luxury”: the case for economic growth in the Roman empire. In Manning J.G. - Morris I. (ed.), 2005 The ancient economy: evidenc eand models, Stanford: 2007-222. Hohlfelder R.L. 1992 The changing fortunes of Caesarea’s harbors in the Roman period. In R.L. Vann (ed.), Caesarea’s papers: Straton’s Tower, Herod’s Harbour, and Roman and Byzantine Caesarea. Journal of Roman Archaeology Supplementary Series 5: 75-78. Holum K.G. 1996 The survival of the bouleutic class at Caesarea in late antiquity. In A. Raban - K.G. Holum (ed.), Caesarea maritima: a retrospective after two millennia, Leiden: 613- 627. 195

THE LATE ROMAN ARMY IN THE EAST Levine L.I. 1975b Caesarea under Roman rule, Leiden. Levine L.I. 1982 Eretz Israel in the 3rd century. In Z. Baras, S. Safrai, M. Stern - Y. Tsafrir (ed.), Eretz Israel from the destruction of the Second Temple to the Muslim conquest, Jerusalem: 119-143 (Hebrew). Levine L.I. 1999 Who participated in the fast-day ritual in the city square? Communal fasts in third-and fourth-century Palestine. In Cathedra 94: 33-54 (Hebrew). Lewin A. 2002: Il Negev dall’età nabatea all’epoca tardoantica. In Mediterraneo antico 5: 319-375. Lewin A. 2007 The Impact of the Late Roman Army in Palaestina and Arabia. In L. de Blois – E. Lo Cascio (ed.), The Impact of the Roman Army, Leiden: 463480. Liebeschuetz J.H.W.G. 2001 The decline and fall of the Roman city, Oxford. Lo Cascio E. 1991 Forme dell’economia imperiale. In A. Sciavone (ed.), Storia di Roma II.2, Torino: 313365. Lo Cascio E. 2000 The Roman principate: the impact of the organization of the empire on production. In E. Lo Cascio - W.R. Rathbone (ed.), Production and public power in classical antiquity, Cambridge: 7785. MacMullen R. 1962 The emperor’s largesses. In Latomus 21: 159-166. Magie D. 1950 Roman rule in Asia Minor, Princeton. Magness J. 1993 Jerusalem ceramic chronology: circa 200-800, Sheffield. Magness J. 2003 The archaeology of the early Islamic settlement in Palestine, Winona Lake, Ind. Mann J.C. 1983 Legionary recruitment and veteran settlement during the Principate, London. Manning J.G. - Morris I. (ed.), 2005 The ancient economy: evidence and models, Stanford. Margalit S 1987 The north church of Shivta: the discovery of the first church. In Palestine Exploration Quarterly 119: 106-121. Mattingly D.J. - Salmon J. (ed.), 1999 Economies beyond agriculture in the classical world, London. Mayerson P. 1960 The desert agricultural regime of Nessana and the central Negev, London. Mayerson P.1994 Monks, martyrs, soldiers and Saracens, Jerusalem. Mazar B. (ed.), 1988 Geva, Jerusalem (Hebrew). Mazor G. 1981 The wine-presses of the Negev. Qadmoniot 53-54: 51-60 (Hebrew). Mazor G. - Bar-Nathan R. 1998 The Bet Shean excavation project – 1992-1994. In Excavations and surveys in Israel 105: 7-38. Meshel Z., Fahlbusch H., Tsuk T. - Peleg Y. 1996 The water supply-system of Susita, Tel Aviv. Millar F. 1964 A Study of Cassius Dio, Oxford. Millar F. 1977 The emperor in the Roman world (31 BC – AD 337), London. Millar F. 1984 Condemnation to hard labour in the Roman empire, from the Julio-Claudians to Constantine. In Papers of the British School at Rome 39: 124-145.

in Israel. In Israel Journal of Earth Sciences 40: 219223. Issar A.S. - Makover-Levin D. 1995 Flourishing and abandonment of the desert settlements during the Roman-Byzantine period. In Michmanim 8: 17-22 (Hebrew). Jones A.H.M. 1940 The Greek city from Alexander to Justinian, Oxford. Jones A.H.M. 1964 The later Roman Empire 284-602, Oxford. Jones A.H.M. 1971 The cities of the eastern Roman provinces, 2nd ed., Oxford. Jones A.H.M. 1974 The Roman economy. Studies in ancient economic and administrative history, ed.P.A.Brunt, Oxford. Kennedy D. 2000 The Roman army in Jordan, London. Keppie L. 1983 Colonization and veteran settlement in Italy 47-14 B.C., London. Kingsley S. 2001 The economic impact of the Palestinian wine trade in late antiquity. In S. Kingsley - M. Decker (ed.), Economy and exchange in the east Mediterranean during late Antiquity, Oxford: 44-68. Kleiman E. 1973 Bi-metallism in Rabbie’s time. Two variants of the Mishna “gold acquires silver”. In Zion 38: 48-61 (Hebrew). Klein C. 1982 Morphological evidence of the lake level changes, western shore of the Dead Sea. In Israel Journal of Earth Sciences 31: 67-94. Klein C. 1986 Fluctuations of the level of the Dead Sea and climatic fluctuations in Eretz-Israel during historical times. Ph.D. Diss., Jerusalem ( Hebrew). Kloner A. 1998 Water-retaining dams in the northeastern Negev highlands. In Bulletin of the AngloIsrael Archaeological Society 16: 7-32. Kochavi M. (ed.), 1972 Judaea, Samaria and the Golan, Jerusalem (Hebrew). Kochavi M. - Beit-Arie I. 1991 Archaeological survey of Israel: map of Rosh Ha-’Ayin (78), Jerusalem. Korjenkov A.M. - Mazor E. 1999 Seismic origin of ancient Avdat ruins, Negev desert Israel. In Natural Hazards 18: 193-226. Kraemer, C.J. 1958 Excavations at Nessana, III: Nonliterary papyri, Princeton. Laniado A. Recherches sur les notables municipaux dans l’empire protobyzantine, Paris. Lapin H. 2001 Economy, geography, and provincial History in later Roman Palestine, Tübingen. Lehman C.M. 1995 The combined Caesarea expeditions: the excavation of Caesarea’s Byzantine city wall. In Annual of the American Schools of Oriental Research 52 (ed. W.G. Dever): 121-131. Leibner U. 2004 History of the eastern Galilee during the Hellenistic, Roman and Byzantine periods in light of an archaeological survey. Ph.D.Diss. Ramat Gan (Hebrew). Lender Y. 1990 Archaeological survey of Israel: map of Har Nafha (196), Jerusalem. Levine L.I. 1975a Roman Caesarea: an archaeologicaltopographical study, Jerusalem. 196

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY Ovadia A. - De Siva G. 1981 Supplementum to the ‘Corpus of the Byzantium Churches in the Holy Land’ (Part I). In Levant 13: 200-261. Ovadia A. - De Siva G. 1982 Supplementum to the ‘Corpus of the Byzantium Churches in the Holy Land’ (Part II). In Levant 14: 122-170. Parker S.T. 1999 The Byzantine period: an empire’s new Holy Land. In New Eastern Archaeology 62/3: 135180. Patrich J. 1996 Warehouses and granaries in Caesaria Maritima. In A. Raban - K.G. Holum (ed.), Caesarea maritima: a retrospective after two millennia, Leiden: 146-176. Patrich J. 2001 Urban space in Caesarea Maritima, Israel. In T.S. Burns - J.W. Eadie (ed.), Urban centers and rural contexts in late antiquity, Ann Arbor: 77-110. Patrich J. 2002 The martyrs of Caesarea. The urban context. In Liber Annuus Studii Biblici Franciscani 52: 321-346. Peleg Y. 2002a Two aqueducts to Dor. In D. Amit, J. Patrich - Y. Hirschfeld (ed.), The aqueducts of Israel. Journal of Roman Archaeology Supplementary Series 46: 148-154. Peleg Y. 2002b The dams of Caesarea low-level aqueduct. In D. Amit, J. Patrich - Y. Hirschfeld (ed.), The aqueducts of Israel. Journal of Roman Archaeology Supplementary Series 46: 141-147. Peleg M. - Reich R. 1992 Excavation of a segment of the Byzantine city wall of Caesarea. In ‘Atiqot 21: 137171. Porath Y. 1998 The Caesarea excavation project – March 1992-June 1994: expedition of the Antiquities Authority. In Excavations and Surveys in Israel 17: 39-49. Porath Y. 2000 Caesarea – 1994-1999. In Excavations and Surveys of Israel 112: 34*-40*. Porath Y. 2002 The water supply to Caesarea: a reassessment. In D. Amit, J. Patrich - Y. Hirschfeld (ed.), The aqueducts of Israel. Journal of Roman Archaeology Supplementary Series 46: 104-129. Purcell N. 2000 The cities of Italy: an imperial anomaly. In A.K. Bowman, P. Garnsey - D. Rathbone (ed.), Cambridge ancient history, 2nd ed. XI, Cambridge: 423-432. Raban A. 1982 Archaeological survey of Israel: map of Nahlal (28), Jerusalem. Raban A. 1996 The inner harbor basin of Caesarea: archaeological evidence for its gradual demise. In A. Raban - K.G. Holum (ed.), 1996 Caesarea Maritima: a retrospective after two millennia, Leiden: 628-666. Raban A. 1999 Archaeological survey of Israel: map of Mishmar Ha-’Emeq (32), Jerusalem. Raban A. 2004 The history of Caesarea’s harbors. In Qadmoniot 127: 2-22 (Hebrew). Raban A. – Holum K.G. (ed.), 1996 Caesarea Maritima: a retrospective after two millennia, Leiden. Rickman G. 1980 The corn supply of ancient Rome, Oxford.

Millett M. 1991 Pottery: population and supply patterns? In G. Barker - Lloyd J. (ed.), Roman landscapes. Archaeological survey in the Mediterranean region, London: 18-26. Millett P. 2001 Productive to some purpose? The problem of ancient economic growth. In Mattingly D.J. - Salmon J. (ed.), Economies beyond agriculture in the classical world, London: 17-48. Mitchell S. 1976 Requisitioned transport in the Roman empire: a new inscription from Pisidia. In Journal of Roman Studies 66: 106-131. Nagar Y. 1999 The anthropology of Rehovot-in-theNegev population as an example of a large Byzantine settlement in the Negev. Ph.D.Diss., Tel Aviv (Hebrew with English abstract). Neesen L. 1980 Untersuchungen zu den direkten staatsabgaben der römischen Kaiserzeit (27v. Chr. – 284 n.Chr.), Bonn. Negev A. 1988 The architecture of Mampsis: final report, I-II, Jerusalem. Negev A. 1991 Personal names in the Nabataean realm, Jerusalem. Negev A. 1993a Elusa. In E. Stern (ed.), The new encyclopedia of archaeological excavations in the Holy Land, 1, Jerusalem: 379-383. Negev A. 1993b Kurnub. In E. Stern (ed.), The new encyclopedia of archaeological excavations in the Holy Land, 3, Jerusalem: 882-893. Negev A. 1993c Sobata. In E. Stern (ed.), The new encyclopedia of archaeological excavations in the Holy Land, 4, Jerusalem: 1404-1410. Negev A. 1993d Nessana. In E. Stern (ed.), The new encyclopedia of archaeological Excavations in the Holy Land, 3, Jerusalem: 1145-1149. Negev A. 1993e Oboda. In E. Stern (ed.), The new encyclopedia of archaeological excavationsin the Holy Land, 3, Jerusalem: 1155-1165. Negev A. 1996 Oboda: a major Nabatean caravan halt. In Aram 8: 67-87. Negev A. 1997 The architecture of Oboda. Final report, Jerusalem. Netzer E. - Weiss Z. 1994 Zippori, Jerusalem. Neusner J. 1999 Rabbinic sources for historical study: a debate with Zeev Safrai. In J. Neusner - A.J. AveryPeck (ed.), Judaism in late antiquity Part 3 vol.I, Leiden: 123-142 Olamy Y. 1981 Archaeological survey of Israel: Daliya map (31), Jerusalem. Olamy Y. - Gal Z. 2003 Archaeological survey of Israel: map of Shefar’am (24), Jerusalem. Olamy Y., Sender S. - Oren E. 2004 Archaeological survey of Israel: map of Yagur (27), Jerusalem. Oleson J.P. 1996 Evidence for the history of the harbors of Caesarea. In A. Raban - K.G. Holum (ed.), Caesarea maritima: a retrospective after two millennia, Leiden: 359-380. Ovadiah A. 1970 Corpus of the Byzantine churches in the Holy Land, Bonn.

197

THE LATE ROMAN ARMY IN THE EAST structure during the Mishnaic and Talmudic periods. In A. Kasher, A. Oppenheimer - U. Rappaport (ed.), Man and land in Eretz-Israel in antiquity, Jerusalem: 20-56 (Hebrew). Safrai Z. 1994 The economy of Roman Palesatine, London. Safrai Z. 1998 The missing century. Palestine in the fifth century: growth and decline, Leuven. Safrai Z. 1999 Rabbinic sources as historical: a response to Professor Neusner. In J. Neusner - A.J. Avery-Peck (ed.), Judaism in late antiquity, Part 3 vol.I, Leiden: 143-167. Saidel B.A. - Christopherson G.L. 2005 Four days at Khalasa: using aerial photography and GIS analysis to reappraise Woolley and Lawrence’s survey of Byzantine Elusa in the western Negev desert. In Palestine Exploration Quarterly 137: 53-64. Salmon E.T. 1969 Roman colonization under the Republic, London. Saller R. 2005 Framing the debate over growth in the ancient economy. In Manning J.G. - Morris I. (ed.), The ancient economy: evidence and miodels, Stanford: 223-238 (first Published in Scheidel W. Von Reden S. (ed.), The ancient economy [Edinburgh 2002]: 251-269. Saxer R. 1967 Untersuchungen zu den Vexillationen des römischen Kaiserheeres von Augustus bis Diokletian, Köln. Schäfer P. 1997 The history of the Jews in antiquity, Luxembourg. Scheidel W. - Von Reden S. (ed.), 2002 The ancient economy, Edinburgh. Schürer E. 1979 The history of the Jewish people in the age of Jesus Christ (175 B.C.-A.D. 135), 2. Revised and edited by G. Vermes, F. Millar - M. Black, Edinburgh. Segal A. 1988 Architectural decoration in Byzantine Shivta, Negev desert, Israel. BAR IS 1287, Oxford. Segal A. 1995 Theatres in Roman Palestine and provincial Arabia, Leiden. Segal A., Mlynarczyk, Burdajewicz, Schuler M. Eisenberg M. 2004 Hippos – Sussita: fifth season of excavations September-October 2004 and Summary of all five seasons (2000-2004), Haifa. Segal A., Mlynarczyk J., Burdajewicz B., Schuler M. Eisenberg M. 2005 Hippos – Sussita (July 2005), Haifa. Shatzman I. 1983 The beginning of the Roman defensive system in Judaea. In American Journal of Ancient History 8: 130-160. Shatzman I. 1991 The armies of the Hasmonaeans and Herod, Tübingen. Shatzman I. 1995 Review: Z. Safrai, The economy of Roman Palestine, London 1994. In Scripta Classica Israelica 14: 186-190. Shereshevski J. 1991 Byzantine urban settlements in the Negev desert, Beer-Sheva. Sirks A.J.B. 1991 Food for Rome: the legal structure of the transportation and processing of supplies for the

Roll I. 1996 Roman roads to Caesarea Maritima. In A. Raban - K.G. Holum (ed.), Caesarea Maritima: a retrospective after two millennia, Leiden: 549-558. Roll I. - Ayalon E. 1989 Apollonia and southern Sharon, Tel Aviv (Hebrew). Roll I - Ayalon E. Apollonia-Arsuf. 1993 In E. Stern (ed.), The new encyclopedia of archaeological excavations in the Holy Land, 1, Jerusalem: 72-74. Ronen A. - Olamy Y. Archaeological survey of Israel: map of Haifa – east (23), Jerusalem. Rosen S.A. 1992 Nomads in archaeology: a response to Finkelstein and Perevolotsky. Papers of the British Schools of Oriental Research 287: 75-85. Rosen S.A. 1994 Archaeological survey of Israel: map of Makhtesh Ramon (204), Jerusalem. Rosen S.A. 2000 The decline of desert agriculture: a view from the classical period Negev. In G. Barker D. Gilbertson (ed.), The archaeology of Drylands. Living at the Margin, London: 45-62. Rostovtzeff M.I. 1957 The social and economic history of the Roman empire, 2nd ed., Oxford. Rubin R. 1988 Water conservation methods in Israel’s Negev desert in late antiquity. In Journal of Historical Geography 14: 229-244. Rubin R. 1989 The debate over climatic changes in the Negev, fourth-seventh centuries c.e. In Palestine Exploration Quarterly 121: 71-78. Rubin R. 1990 The Negev as a settled land: urbanization and settlement in the desert in the Byzantine period, Jerusalem (Hebrew). Rubin R. 1991 Settlement and agriculture on an ancient desert frontier. In Geographical Review 81: 197-205. Rubin R. 1995 Climate versus the human factor: the rise and fall of Byzantine Negev settlement. In Michmanim 8: 23-32 (Hebrew). Rubin R. 1996 Urbanization, settlement and agriculture in the Negev desert – the impact of the RomanByzantine empire on the frontier. In Zeitschrift des deutschen Palästina-Vereins 112: 49-60. Rubin R. 1997 The Romanization of the Negev, Israel: geographical and cultural changes in the desert frontier in late antiquity. In Journal of Historical Geography 23: 267-283. Rubin R. - Shereshevski J 1988 Sa’adon – an urban settlement of the Byzantine period in the Negev. In Qadmoniot 81-81: 49-54 (Hebrew). Rubin Z. 1996 The see of Caesarea in conflict with Jerusalem from Nicaea (325) to Chalcedon (451). In A. Raban - K.G. Holum (ed.), Caesarea maritima: a retrospective after two millennia, Leiden: 559-574. Rubin Z. 1999 Jerusalem in the Byzantine period – an historical survey. In In Y. Tsafrir - S. Safrai (ed.), The history of Jerusalem: the Roman and Byzantine periods (70-638 CE), Jerusalem: 199-238 (Hebrew). Russell J. 1986 Transformations in early Byzantine urban life: the contribution and limitations of archaeological evidence. In The 17th international Byzantine congress major papers, New Rochelle, NY: 137-154. Safrai Z. 1986 The influence of demographic stratification on the agricultural and economic 198

ISRAEL SHATZMAN: ECONOMIC CONDITIONS, SECURITY PROBLEMS AND THE DEPLOYMENT OF THE ARMY Tsafrir Y. 1999b The topography and archaeology of Jerusalem in the Byzantine period. In Y. Tsafrir - S. Safrai (ed.), The history of Jerusalem: the Roman and Byzantine periods (70-638 CE), Jerusalem: 281-352 (Hebrew). Tsafrir Y., Di Segni L. - Green J. 1994 Tabula Imperii Romani: Judaea – Palaestina, Jerusalem. Tsafrir Y. - Foerster G. 1997 Urbanism at ScythopolisBet Shean in the fourth to seventh centuries. In Dumbarton Oak papers 51: 83-146. Tsafrir Y. - Holum K. 1993 Rehovot-in-the-Negev. In E. Stern (ed.), The new encyclopedia of archaeological excavations in the Holy Land, 4, Jerusalem: 12741277. Tsafrir Y., Patrich J. - Rosenthal-Heginsbottom R. 1988 Excavations at Rehovot-in-the- Negev, I: the northern church, Jerusalem. Tsuk Z. 2002 The water-supply of Shivta in the Byzantine period. In C. Ohlig, Y. Peleg - T. Tsuk (ed.), Cura aquarum in Israel, Siegburg: 65-80. Tsuk T., Peleg Y., Fahlbusch H. - Meshel Z. 2002 A new survey of the aqueducts of Hippos-Susita. In D. Amit, J. Patrich - Y. Hirschfeld (ed.), The aqueducts of Israel. Journal of Roman Archaeology Supplementary Series 46: 207-209. Urman D. 1985 The Golan: a profile of a region during the Roman and Byzantine periods. BAR IS 269, Oxford. Urman D. 2004 Nessana Excavations 1987-1995. In D. Urman (ed.), Nessana: excavations and Studies, I, Beer-Sheva: 1*-118*. Vallat J.-P. 1991 Survey archaeology and rural history – a difficult but productive relationship. In G. Barker Lloyd J. (ed.), Roman landscapes. Archaeological survey in the Mediterranean region, London: 10-17. Van Berchem D. 1937 L’annone militaire dans l’empire romain au IIIe siècle. In Bulletin de la Société Nationale des Antiquaires de France 80: 117-202. Van Berchem 1977 L’annone militaire est-elle un myth? In H. van Effenterre (ed.), Armées et fiscalité dans le monde antique, Paris: 331-336. Von Freyberg H.-U. 1989 Kapitalverkehr und Handel im römischen Kaiserreich (27 v. Chr. – 235 n. Chr.), Freiburg im Breisgau. Walmsley A. 1992 The Byzantine period: Byzantine domestic occupation in areas III and IV. In A.W. McNicoll - J. Hanbury-Tenison (ed.), Pella in Jordan 2. The second interim report of the Joint University of Sydney and College of Wooster excavations at Pella 1982-1985, Sydney: 163-181. Walmsley A. 1996 Byzantine Palestine and Arabia : urban prosperity in late antiquity. In N. Christie - S.T. Loseby (ed.), Towns in transition. Urban evolution in late antiquity and the early Middle Ages, Aldershot: 126-158. Walmsley A. 1999 Coin frequencies in sixth and seventh century Palestine and Arabia: Social and economic implications. In Journal of the Economic and Social History of the Orient 42: 326-350.

imperial distributions in Rome and Constantinople, Amsterdam. Smith R.H. 1993 Pella. In E. Stern (ed.), The new encyclopedia of archaeological excavations in the Holy Land, 3, Jerusalem: 1174-1180. Sperber D. 1966 The inflation in 4th century Palestine. In Archiv Orientalni 34: 54-66. Sperber D. 1969 Angaria in rabbinic literature. In L’Antiquité Classique 38: 164-168. Sperber D. 1970 On social and economic conditions in 3rd century Palestine. In Archiv Orientalni 38: 1-25. Sperber D. 1972a Flight and the Talmudic law of usucupation: a study in the social history of 3rd century Palestine. In Revue Internationale des Droits de l’Antiquité 19: 29-42. Sperber D. 1972b Trends in third century Palestinian agriculture. In Journal of the Economic and Social History of the Orient 15: 227-253. Sperber D. 1974 Drought, famine and pestilence in amoraic Palestine. In Journal of the Economic and Social History of the Orient 17: 272-298. Sperber D. 1978 Roman Palestine 200-400: the land. Crisis and change in agrarian society as reflected in rabbinic sources, Ramat-Gan. Sperber D. 1991 Roman Palestine 200-400: money and prices, 2nd ed., Ramat-Gan. Stager L.E. 1993 Ashkelon. In E. Stern (ed.), The new encyclopedia of archaeological excavations in the Holy Land, 1, Jerusalem: 1103-1112. Stemberger G. 2000 Jews and Christians in the Holy Land: Palestine in the fourth century, Edinburgh. Stern E. (ed) 1993 The new encyclopedia of archaeological excavations in the Holy Land, 1-4, Jerusalem Stern, E. 2000 Dor-ruler of the seas, 2nd ed., Jerusalem. Stern E. - Sharon I. 1995 Tel Dor – 1992/1993. In Excavations and Surveys in Israel 14: 61-71. Strobel K. 1989 Inflation und monetäre Wirtschaftstrukturen im 3.Jh. n.Chr. Zu Daniel Sperbers Bild der wirtschafts-und währungsgeschichtlichen Krise. In Münsterische Beiträge zur Antiken Handelsgeschichte 8: 10-31. Tsafrir Y. 1986 The Transfer of the Negev, Sinai and southern Transjordan from Arabia to Palaestina. In Israel Exploration Journal 36: 77-86. Tsafrir Y. 1990 On the pre-planning of ancient churches and synagogues. A test case: the northern church at Rehovot in the Negev. In G.C. Bottini, L. Di Segni, E. Alliata (ed.), Christian archaeology in the Holy Land. New discoveries, Jerusalem: 534-544. Tsafrir Y. 1996 Some notes on the settlement and demography of Palestine in the Byzantine Period: the archaeological evidence. In J.D. Seger (ed.), Retrieving the past. Essays on archaeological research and Methodology in honor of Gus van Beek, Winnona Lake, Indiana: 269-283. Tsafrir Y. 1999a The topography and archaeology of Aelia Capitolina. In Y. Tsafrir - S. Safrai (ed.), The history of Jerusalem: the Roman and Byzantine periods (70-638 CE), Jerusalem: 115-166 (Hebrew). 199

THE LATE ROMAN ARMY IN THE EAST Waner M. - Safrai Z. 2001 A catalogue of coin hoards and the shelf life of coins in Palestine hoards during the Roman and Byzantine periods. In Liber Annuus Studii Biblici Franciscani 51: 305-325. Ward-Perkins B. 2000 Land, labour and settlement; Specialized production and exchange. In A. Cameron, B. Ward-Perkins - M. Whitby (ed.), The Cambridge Ancient History, XIV, Cambridge: 315-391. Watson A. 1999 Aurelian and the third century, London. Weiss Z. Forthcoming Artistic trends and contact between communities in Sepphoris of late antiquity: recent research. In S. Bagnert, D. Gwynn - D. Hull, Leiden. Weiss Z. - Netzer E. 1997a Architectural development of Sepphoris during the Roman and Byzantine periods. In D.R. Edwards - C.T. McCollough (ed.), Archaeology and the Galilee: texts and contexts in the Greco-Roman and Byzantine periods, Atlanta: 117130. Weiss Z. - Netzer E. 1997b The Hebrew University excavations at Sepphoris. In Qadmoniot 113: 2-21 (Hebrew). Whittaker C.R. 1998 Rural life in the later Roman empire. In A. Camron - P. Garnsey (ed.), The Cambridge Ancient History, XIII, Cambridge: 277311. Whittow M. 1990 Ruling the late Roman and early Byzantine city. In Past and Present 129: 3-29. Wilken R. 1992 The land called holy: Palestine in Christian history and thought, New Haven. Williams-Thorpe O. - Henty M.M. 2000 The sources of Roman granaite columns in Israel. In Levant 32: 155170. Witschel C. 1999 Krise – Rezession – Stagnation? Der Westen des römischen Reiches im 3.Jahrhundert n.Chr., Frankfurt. Zertal A. 1992 The Menasseh hill country survey: the Shechem syncline, Haifa (Hebrew). Zertal A. 1996 The Menasseh hill country survey: the eastern valleys and desert fringes, Tel Aviv (Hebrew). Zertal A. 2005 The Menasseh hill country survey: from Nahal Bezek to Sartaba, Tel Aviv (Hebrew). Zertal A. - Mirkam N. 2000 The Menasseh hill country survey: from Nahal ‘Iron to Nahal Shechem, Tel Aviv (Hebrew). Zorn J.R. 1994 Estimating the population size of ancient settlements: methods, problems, solutions, and a case study. In Bulletin of the American Schools of Oriental research 295: 31-48.

200

Fluctuating Provincial Borders in Mid-4th Century Arabia and Palestine* Joonas Sipilä University of Helsinki

southern tip of the Dead Sea.7 The western border of the province of Arabia ran from the intersection of the Wadi Mnjjib and the Dead Sea up north to Adraa.

1. Introduction The formation of the Tetrarchy in March 2931 heralded the beginning of far-reaching reforms in provincial organisation.2 The old provinces were, in general, divided into smaller entities – “chopped into slices to ensure that the terror was universal”, according to Lactantius.3 These provincial reorganisations also affected the province of Arabia and the adjoining province of Palaestina, both of which underwent changes. More reorganisations were to follow during the fourth century. Indeed, the histories of the provinces of Arabia and Palaestina are intertwined, with numerous divisions and transfers of provincial territory back and forth apparently taking place. In fact, Arabia and Palaestina of the Principate became one geographically reduced province of Arabia and three provinces of Palaestinae by the end of the fourth century.

In the southwest most of the Sinai belonged to the province of Arabia. The northern part of the Sinai was part of the province of Aegyptus. The dividing border began at Clysma (part of Aegyptus) and continued to the northeast until Boutafis ((U[Z­\OX), also known as Beth Tappuah ((MNZG\U³), situated fourteen miles west of Raphia, which was part of Syria Palaestina.8 The ports of Rhinocolura and Pelusium were part of Aegyptus. Thus, provincia Arabia did not border on the Mediterranean Sea. To the east of Arabia there was no further Roman province and consequently no need for the governor to define the limits of his power with regard to another officeholder. Instead the province merged in the East with the desert. The nomadic groups living on the desert edge had no fixed abodes and the Roman administration’s control over them would wax and wane. This means that the provincia Arabia was more or less “open” to the east and southeast, a real frontier zone with no strictly delimited administrative borderline.9

The objective of this paper is to investigate the changes of provincial boundaries that occurred in the region around the mid-4th century as well as to suggest possible motives for the changes. To properly consider the situation, a very brief look is also taken at the previous reorganisations from Diocletian onwards.4 The provincia Arabia5 that Diocletian inherited consisted of the Transjordan, Negev and Sinai with the northern border extending to the ‘lava-lands’ of Hawran and including the Jebel Drnjz.6 The border between southern Palaestina and Arabia west of the Dead Sea ran in the Beer-Sheba valley, approximately at level with the

The province of Syria Palaestina was bound in the east and south by the province of Arabia, whereas in the west it was limited by the Mediterranean. It consisted of the regions of Judea, Idumaea, Samaria, Galilee, Peraea and Gaulanitis.10 The territories of the Syrian cities of Ptolemais, Tyre and Caesarea Philippi marked the northern border of Palaestina.

This article has been carried out within the framework of the "Ancient and Medieval Greek Documents, Archives and Libraries" research centre at the University of Helsinki. The centre is part of the "Centres of Excellence in Research" programme of the Academy of Finland. It is part of a project made possible by funding of the Finnish Cultural Foundation. The conference trip was made possible by a grant from the Niilo Helander Foundation and a Helsinki University’s Chancellor’s travel grant. I would also like to thank especially Dr. Z.T. Fiema and Prof. P. Brennan for their comments. All errors of omission and interpretation are mine. 1 All dates are AD unless otherwise noted. 2 Lactantius De Mort. Pers. 7.2 3 Lactantius De Mort. Pers. 7.4 - “Et ut omnia terrore complerentur, provinciae quoque in frusta concisae”. For the provincial reorganisation of Diocletian in general see e.g. Jones 1964: 42-52 & 1451-1461; Barnes 1982: 195-225. 4 See Sipilä 2004, on which this paper is partly based, for discussion of the course of the provincial reorganisations during the 4th century in Arabia. 5 The term Arabia is used here explicitly to denote the territory of the Roman province of Arabia unless otherwise noted. 6 Brünnow & Domaszewski 1909: 264-270; Negev 1977: 643-645; Kettenhofen 1981: 62-73; Sartre 1982: 54-64; Bowersock 1983: 90-104; Graf 1992: 450.

2. The Diocletianic Reorganisation The major Diocletianic reorganisation to affect Syria Palaestina and Arabia took place around the year 300. At the time, southern Transjordan and the Negev were 7

Tsafrir 1986: 78. Eusebius, Onom. 11.18-20 mentions Bghtavo³ and gives its distance to Raphia as 14 miles. See Barag 1973 for a milestone found near Raphia mentioning the border of Syria Palaestina. BoutÇvir appears in the papyri of Theophanes, see Barag 1973: 52 n.7. The edict of Anastasius (Sartre 1982b: 111-119, #9045 & #9046) is interesting as it shows the dux Palaestinae drawing revenue from the portoria collected at Clysma. The implications of this are not completely clear but do not suggest that Clysma was part of Palaestina in Diocletianic times. 9 Or at least the administrative border is less visible to modern scholars. See e.g. the debate over the status of the northern Hedjaz that has revolved around the Ruwwafa inscription. Bowersock 1983: 96-98; Graf 1988: 180; Lewin 2004: 109. 10 Abel 1967: 164. 8

201

THE LATE ROMAN ARMY IN THE EAST detached from the province of Arabia and incorporated into the province of Palaestina. This change is reflected in Eusebius’ Onomasticon and De Martyribus Palaestinae and also evidenced by inscriptions from Yotvata and the Promontory Palace of Caesarea mentioning praeses Priscus.11

of provinces.14 This very controversial entry has been linked, as in here, to the appearance of Nea Arabia (mÈa ’AqabeÊa, most probably Arabia Nova in Latin) in one Oxyrhunchus papyrus.15 Furthermore, the formation of Arabia Nova is connected with the appearance of Aegyptus Iovia and Herculia (later renamed Mercuriana).

It is proposed that military considerations, specifically related to internal security and the desire to minimise the probability of a revolt, were the primary motives behind the Diocletianic reorganisation of Arabia and Palaestina. This view assumes that the provincial governors and more specifically the praeses of Arabia still had military responsibilities at the time.12 Due to the posting of IV Martia to El-Lejjun and the transfer of the legion X Fretensis to Aila the overall command of three legions would have been concentrated in the hands of the governor of Arabia if the division had not been carried out. As such the decision to detach southern Transjordan and the Negev from Arabia and attach them to Palaestina was a very logical decision reflecting these military concerns, made because of and at the same time as the transfer of the legions. This transfer can also be interpreted as a precaution against a rebellion in Egypt as well as a measure to protect Aila and the trade flowing through it. After the reorganisation the geographically diminished province of Arabia would still have had two legions (in Bostra and el-Lejjun). Also Palaestina retained its legion, although transferred to southern Transjordan, which was incorporated into Palaestina at the same time or slightly earlier. The possible but highly speculative transfer from Caparcotna to Udruh of the unaccounted-for legion VI Ferrata at the same time would have further underlined the imperial motive to detach southern Transjordan from Arabia.13

Whatever was the actual extent and location of this Arabia Nova, it seems to have been an ephemeral creation. The province of Arabia Nova, as well as the provinces of Aegyptus Iovia and Aegyptus Mercuriana, disappeared by 325. This probably ensued right after Constantine’s victory over Licinius. There is no mention of Arabia Nova or of the two Aegypti in the list of participants of the Council of Nicea.16 On the basis of the participant roster, the area of southern Transjordan and the Negev apparently belong to the province of Palaestina. This also receives oblique support from Eusebius, who in his “Commentary on Isaiah”, written sometime between 324 and 337, referred to Petra as a city of superstitious people in Palaestina.17 The provinces of Palaestina and Arabia thus appear to have the extent they had at the end of Diocletian’s reign.18 4. The Mid-4th Century Changes – Letters of Libanius The arrangement of the provinces of Palaestina and Arabia that existed in 325 was not, however, to be final. Further changes occurred around the middle of the fourth century. The conflicting evidence that apparently points to complicated administrative changes derives from the letters of Libanius, the Notitia Dignitatum, as well as from references in the law codes and literary texts. In the voluminous correspondence of Libanius, letters 315, 321 and 334, dated to the years 357/8, have been interpreted as related to the provincial reorganisations of Palaestina and Arabia.19 All of them were addressed to Clematius, the governor of Palaestina.

3. The Enigmatic Arabia Nova A problematic phase in the course of the reorganisations is then encountered in the Laterculus Veronensis, datable to 314-320, which mentions arabia item arabia in the list

In the letter, Ep. 315, written in 357, Libanius recommended the rhetors Eunomus and Eudaemon “from Elusa” (ȑȟ ʧǼȜȠȪıȘȢ) to governor Clematius. The phrase ȑȟ ʧǼȜȠȪıȘȢ has been interpreted to mean that the two men were not only residents of Elusa at the time but also that they made their request from there.20 This interpretation, in turn, has in the past been construed as a proof that Elusa belonged to the jurisdiction of Clematius and was thus part of the province of Palaestina. However, Mayerson has suggested that Libanius may just

11 Eusebius De mart. Pal. vii.2; i.1; Onom. 6.17-23; 36.13-14; 112.8-12; 142.7-8; 144.7-9; 168.8; Roll 1989; Burrell 1993. See Sipilä 2004: 321326. A group of cadaster stones evidences that the northern boundary of Arabia was also slightly extended to the north somewhat earlier (c. 295) to include Leja and the city of Maximianopolis/Shaqqa. See Brünnow & Domaszewski 1909: 270; Sartre 1982a: 66-70; Millar 1993: 194-195 & 543-544; Syon & Hartal 2003. 12 The division between civil and military spheres of command on the provincial level was apparently implemented rather gradually. Praesides are encountered taking care of military concerns still after the formation of the first tetrarchy (e.g. CIL VIII 2572; 9041; 9324; ILS 640). This also seems to have been the case in Arabia. In the year 296, praeses Aurelius Antiochus is credited for renovating a military fort (AE 2000: 1540). Also, praeses Aurelius Asclepiades is attributed for building a fort, the castra praetori mobeni (Qasr Bshir) (AE 1895: 182; CIL III 14149). On the other hand it has been suggested that the fort built by Asclepiades was not a military installation at all but could have functioned as a residence, praetorium, for the peripatetic governors (Whittaker 1994: 139; Isaac 1990: 172-174). For duces during the first tetrarchy, see Lewin 2004: 228-229. 13 Sipilä 2004: 326-331.

14

Lat. Ver. as edited by Barnes 1982: 202-203. P. Oxy. 3574, edited by J. Rea 1983. 16 Mansi 1759a: 693-702. Furthermore, a praefectus Caecilius is attested c. 325, ruling over a unified Aegyptus (Barnes 1982: 151). 17 Eusebius, Comm. in Isaiam, 392. 18 See Sipilä 2004: 331-336. 19 Foerster’s letter enumeration is followed here. It is also important to bear in mind that Libanius did not date his letters but that the letters have been dated by modern scholars and as such the datings can at times be debatable. 20 Seeck 1906: 131; Brünnow & Domaszewski 1909: 277. 15

202

JOONAS SIPILÄ: FLUCTUATING PROVINCIAL BORDERS IN MID-4TH CENTURY ARABIA AND PALESTINE as well have referred to the origo of the rhetors with the phrase ȑȟ ʧǼȜȠȪıȘȢ instead of their current residence.21 In this case, though, the reference is more probably to their current residence as Eudaemon comes originally from Pelusium in Egypt.22 On the other hand, Libanius asks Clematius to favour the rhetors by summoning them to speak in his court. This calls into question the relevance of their domicile. The men apparently resided not very far from Clematius for Libanius to petition him. However, that does not rule out a city nearby in a neighbouring province. Consequently, it is difficult to regard the letter Ep. 315 as unambiguous evidence concerning the identity of the province that Elusa was part of in 357.

were under the jurisdiction of Clematius, and connected with the letter 334, which provides evidence of an actual administrative division. This has led to a postulate that in 357/358, when the letters were written, the existing large province of Palaestina was divided in two and that the southern part, (southern Transjordan and the Negev) was named Palaestina Salutaris (or Secunda).28 A closer examination of letters Ep. 315 and especially of Ep. 321, however, suggests that such a conclusion, while possible, is heavily relying on far-reaching, poorly supported inference and thus uncertain. 5. The Mid-4th Century Changes – Ecclesiastical Sources

The next letter, Ep. 321, dated to 357, is also a petition. It concerns a rhetor called Dynamius. In the letter Libanius beseeches the governor Clematius to show goodwill towards Dynamius, a man “who adorns Petra”.23 In the past this has been interpreted as evidence that Dynamius lived in Petra and furthermore that Petra was part of the province of Palaestina and belonged to the jurisdiction of Clematius.24 However, Mayerson has proposed that Dynamius the rhetor was just expressing his hopes of enjoying Clematius’ hospitality on the way to Petra. Furthermore, the expressed praise of Dynamius as a man who adorns Petra can be interpreted to mean merely that Dynamius was an outstanding rhetor, a primus inter pares among those who could be expected to appear in Petra. The letter does not imply that Petra belonged to the province of Palaestina.25 Indeed, the more unassuming interpretation of Mayerson should be favoured over the one making more far-reaching assumptions and based on inferences from the extremely meagre information provided by Ep. 321.

The ambiguous nature of the letters of Libanius as evidence for the provincial reorganisation of Palaestina is not the only dilemma. There are more open questions connected to the mid-century evidence regarding the provincial status of southern Transjordan and the Negev. The traditional interpretation of Libanius’ letters 334, 321 and 315 that postulates the creation of Palaestina Salutaris/Secunda in 357/358 does not find support in other sources. Indeed, there is no evidence that supports the existence of a province called Palaestina Salutaris or Secunda between c. 340 and c. 390. Instead, the evidence apparently suggests that at least southern Transjordan belonged to Arabia at the time. The rosters of participants of the Councils of Serdica (343) and of Alexandria (362) list one Asterius from Petra, Arabia, as attending both of them.29 Asterius of Arabia is also known from other sources, namely from the writings of Athanasius and Theodoret. Athanasius (c. 293-373) referred to Asterius and called him in the first instance Asterius of Arabia and the second time Asterius of Petra, Arabia.30 Moreover, the works of Athanasius do not reveal any knowledge of the existence of multiple Palaestinae at that time.31 Theodoret also claimed that Asterius of Arabia was present at the Council of Serdica, although he wrote at a later date.32 In addition Epiphanius called Petra the metropolis of Arabia in his Panarion, written c. 379.33

The last of the letters, Ep. 334, is a petition and the one letter that seems to explicitly refer to a provincial division. Libanius wrote this letter to Clematius on behalf of his friend Hieronymus who was a resident of the city of Elusa. Libanius admits that he knows that the letter is useless since Hieronymus is under the jurisdiction of another man due to the “division of the administrative office”.26 Thus, letter 334 explicitly indicates that Elusa is under the jurisdiction of someone else than Clematius, the governor of Palaestina. In addition the statement “even if someone would divide Palaestina further”, indicates that the division had been quite recent. The conclusion is unavoidable – Elusa and most probably the Negev as well were no longer part of Palaestina at the time the letter was written.27

28

Seeck 1906: 135-136; Brünnow & Domaszewski 1909: 277-279; Jones 1964: 1459-1460; Barnes 1982: 214. Contested by Mayerson 1984: 227-228; Mayerson 1987. For Salutaris as a province name see Balty 1980: 467 n. 1. The province name Salutaris continued to be used. See Merkelbach & Stauber 2002: 445, for a later inscription mentioning Palaestina Salutaris (first published in Egan & Bikai 1999). In the Petra Pap. I.1 (inv. 68) from the year 537 Petra is called “metropolis of the Third Palestine Salutaris”. 29 Mansi 1759b: 42, 354. 30 Athanasius, Apol. cont. Ar. 333; Athanasius, Tom. ad Ant. 808. 31 E.g. Athanasius, Apol. cont. Ar. 333 – Acacius of Caesarea and Macarius from Palaestina; Athanasius, Hist. Aria. ad Mon. 725 – list of provinces where the attending bishops came from. 32 Theodoret, Hist. Eccl. II.8.30. 33 Epiphanius, Panarion LI.22.10-11, describing pagan festivities celebrated in Petra. Some credibility to his testimony may be added by the fact that Epiphanius was a native of Eleutheropolis, Palaestina and consequently was personally acquainted with the general region.

The letters of petition, Ep. 315 and Ep. 321, have in the past been interpreted to indicate that both Elusa and Petra 21

Mayerson 1987: 254-255. Kaster 1988: 279-280. Libanius, Ep. 321: …t© joslo³mti tÂm PÈtqam. 24 Seeck 1906: 342; Brünnow & Domaszewski 1909: 277-278. 25 Mayerson 1987: 255. 26 Libanius, Ep. 334: …jatÀ tÂm tolÂm Ðqw±r. 27 Mayerson 1984: 227; Mayerson 1987: 256; Sipilä 2004: 337. 22 23

203

THE LATE ROMAN ARMY IN THE EAST The lists of participants of other church councils do not provide any further relevant information. The list of participants of the Council of Seleucia (359) has only a few referrals to provinces and they do not imply any modifications.34 The list of the Council of Constantinople (381) has more information but does not offer anything regarding the status of southern Transjordan and the Negev except obliquely, by listing only one provincia Palaestina.35

mentions imply that early in the reign of Theodosius not only was there only one province of Palaestina, but also that southern Transjordan was still considered a part of Arabia. 6. The Mid-4th Century Changes – Other Sources Codex Theodosianus contains more clues relevant to the subject. Four imperial edicts, from the years 363, 383, 384 and 385, are addressed to the governor of Palaestina. The first edict is addressed to Leontius, Consularis Palaestinae.40 The other three recipients carry the title of proconsul Palaestinae.41 It is noteworthy that their titles do not specify the “Palaestina” they governed. It could be suggested that this was the case because at the time of the edicts only a single province of Palaestina existed.42 Furthermore, a law preserved in Codex Justinianus tying the agricultural population to land and dating to the year 386 seems to refer to Palaestina in singular form.43

Apart from later emendations, the references to provinces in the lists of the Church Councils could become more problematic if the emerging Church hierarchy and geographical extent of the metropolitan sees did not entirely coincide with the civil provinces. However, there is no reason to suspect such an exception with regard to the mid-4th century Negev and southern Transjordan.36 At any rate, the evidence makes it hard to automatically accept the notion that Petra was part of Palaestina in 357/8 when Libanius wrote Ep. 321.

Novella 103 of Emperor Justinian from the year 536 offers further clarification. In this edict Justinian elevated the status of the governor of Palaestina Prima to proconsul. The reason given for this was that Palaestina had earlier been governed by a proconsul, but the rank had been downgraded to ordinarius when Palaestina, formerly one province, was divided in three.44 This apparently means that the proconsules Palaestinae appearing in Codex Theodosianus and receiving their instructions between 383-5 had governed a single Palaestina, the only province of that name existing at the time.45

In this context it is interesting to note that the nun Egeria, when touring the Biblical sites c. 384, wrote that she travelled from Clysma through a region called Arabia which had earlier been part of Aegyptus.37 It is impossible to know whether she referred to Arabia as a province in her time or earlier, as a geographical area, or as something else. Neither did she indicate any awareness of different Palestines or Arabias in existence. Although it proves very little, it might be suggested that she did not comment on Sinai being in Palaestina because at the time there existed only one large province of Arabia including the Negev and the Sinai.

Novella 103 can be connected with yet another letter of Libanius (Ep. 686), dated to 361 and addressed to Cyrillus, that has been interpreted as relevant to the provincial reorganisations. In this puzzling letter Libanius congratulated Cyrillus for being advanced “from office to

It is also worth noting that Sozomen, writing at the time of Theodosius II (408-50), related in his Church History a story of a dispute early in the reign of Theodosius the Great that the Petrans and the Areopolitans of Arabia had with the people of Raphia and Gaza from Palaestina.38 In a similar vein, Zosimus mentioned how, in 387, a rhetor named Hilarius made a powerful impression on Emperor Theodosius with his oratory. As a consequence, Theodosius granted him rule over Palaestina.39 These

40

Cod. Theod. XII.1.55. Cod. Theod. XI.36.28; XI.30.42; X.16.4. 42 It is worth noting that Not. Dig. lists all the consular provinces in the pars Orientis, with the exception of Dacia Mediterranea, without a qualifying name, whereas the praesidal provinces carry such distinctions when there exists a consular province with the same primary name (Not. Dig. Or. i). 43 Cod. Just. XI.51. – “… [n]eque id Palaestinae provinciae possessoribus suffragetur, sancimus, ut etiam per Palaestinam …”. In singular in Beck’s edition (1837). Hermannus (1887, fratres Kriegelii) in his edition noted (n. 13) that it is in singular in Haloander’s edition but decided for the plural (“Palaestinas”) referring to Cod. Theod. VII.4.30. Krüger (1877) has also plural but notes that Palestine appears in singular in his manuscripts K & Z (Parisiensis 4537; Parisiensis 15423). As such the mention is a weak argument for the existence of only a single Palaestina in 386 but neither is it evidence against that interpretation. 44 Iust. Nov. CIII – “.. utque ex maiore in minorem dignitatum translata est, quum Palaestina quae antea tota una fuit, in tres partes postea divisa fuerit, nec proconsulatus eius manserit sed in hunc magistratum, qui ordinaries vocatur translatus sit.” See Mayerson 1988: 66-70 for a discussion of the unrest in Palaestina in the early 6th century, reflected in this edict. 45 Mayerson 1987: 260; Cod. Theod. XI.36.28; XI.30.42; X.16.4; In addition, St. Jerome in one of his letters written in 385/386 mentioned a proconsul of Palestine (Ep. 108.9). 41

34

Mansi 1759b: 322-323. Mansi 1759b: 568-569. The emerging church organisation followed the civil provinces, a principle laid out in Nicea. Even though Innocent I, (Ep. 24.2), in the beginning of the fifth century, declared that the ecclesiastical organisation should not have to follow the reorganisations of secular provinces, exceptions were not common. See, however, the reservations of Sartre 1982a: 72. For the organisation of the Church see Jones 1964: 874-894. Ecclesiastical concerns might have been reflected in the appearance of provincia Bostron in the council list of Constantinople (list in Mansi 1759b: 569). 37 Itinerarium Egeriae 7.1. On the date of Egeria’s travels, see Devos 1967. 38 Sozomen, Hist. Eccl. VII.15.11. For the dating of Sozomen’s work see Bidez 1960: lxvi. 39 Zosimus, Hist. Nova IV.41. Zosimus defines Hilarius rhetorically as the governor of “all Palaestina”. The tenure of Hilarius is usually dated to 392/3, based on the letters of Libanius (PLRE I 435 ‘Hilarius 8’). The dating is not, however, very secure, and there is no alternative name for 387/388 either. 35 36

204

JOONAS SIPILÄ: FLUCTUATING PROVINCIAL BORDERS IN MID-4TH CENTURY ARABIA AND PALESTINE office, from Palaestina to Palaestina”.46 This has been interpreted, consistent with the postulated view of the reorganisation of 357/8, to mean that Cyrillus was promoted from praeses Palaestinae Salutaris to consularis Palaestinae Primae.47 However, Mayerson has proposed that actually Libanius only congratulated Cyrillus for the elevation of his office to a higher rank: from the praeses Palaestinae to consularis Palaestinae. Incidentally, governor Leontius was addressed with the same title in the edict of 363.48 It seems more likely that the question is indeed of advancement in titles and honours.

In the light of the evidence the suggestion that the administrative change of 357/8, recorded in Libanius’ letter 334, created the province of Palaestina Salutaris (or Secunda or Prima, for that matter) appears unfounded. Yet Libanius’ letter, Ep. 334, clearly indicates that some kind of a reorganisation or an administrative shift had occurred. Consequently, it is suggested that in 357/8, the Negev was detached from Palaestina and attached to the province of Arabia, which then assumed almost its Severan proportions. Admittedly, the reconstruction of the exact course of the reorganisations is hypothetical. None of the evidence is in any way conclusive and many of the sources problematic. However, the proposed scenario of provincial reorganisations at least fits the surviving evidence.

In fact, the works of Libanius cast some doubt on the existence of multiple Palaestinae. Although, as Mayerson has noted, the name “Palaestina” occurs around forty times in the letters and orations of Libanius, he never used any modifying term for it. Neither did he show awareness of the existence of multiple provinces of the same name or of a change in the official name of the province before or after the date of Ep. 334 that alluded to an administrative division.49

8. The Factors behind the Reorganisations While the Diocletianic change seems to have been clearly connected to military considerations, the reasons for the piece-by-piece moves of territory from Palaestina to Arabia that apparently took place c. 340–357/8 are very difficult to fathom in the absence of direct evidence. Indications of the possible causes of the two mid-century reorganisations may be isolated by shortly examining central factors influencing the regions of Palaestina and Arabia. It can be suggested that the reorganisations were undertaken due to a combination of administrative, economic and cultural as well as military reasons.

7. Suggested Course of the Mid-Century Reorganisations To recapitulate, contrary to some earlier views, the province of Palaestina Salutaris/Tertia does not appear to have been created before c. 390. Evidence that seems to be against the existence of Palaestina Salutaris as a province between 357-c.390 includes conciliar lists, both contemporary and later literary mentions, letters of Libanius and references to Palaestina in Codex Theodosianus and Codex Justinianus as well as the indirect evidence contained in Justinian’s Novella 103. On the other hand, there is no evidence that would attest the existence of such a province during the period 357- c. 390. As such the view that a province named Palaestina Salutaris or Secunda was created already in 357/8 is consequently hard to sustain.

A factor significantly contributing to the reorganisations may have been the rising importance of Christianity and the special role of Palestine as Terra Sancta, which brought the hitherto provincial backwater to the centre of attention. By the mid-fourth century Palestine had become the destination of an ever-growing number of pilgrims who made donations and revitalised local economies through their needs for different services.50 The new status as Terra Sancta also channelled new wealth to Palaestina, and to a lesser extent to Arabia, in the form of building projects.51 The bustling pilgrimage movement and issues related to it undoubtedly began to increasingly demand the attention of the governor, thus adding to his workload. The pilgrims also included people of high status who required special consideration from the governor.52 Increased burden on the administration of the province may have also been

Judging from the wealth of evidence apparently placing Petra in Arabia around the middle of the fourth century, it is postulated that Petra and southern Transjordan were detached from Palaestina and incorporated into Arabia. This happened at some time between 325 (Council of Nicaea) and 343 (Council of Serdica). Taking into consideration the presumed date of writing of Eusebius’ “Commentary on Isaiah” (between 325 and 337), which still places Petra in Palaestina, it can be suggested that the proposed change took place around the year 340. The Negev remained part of the province of Palaestina at the time.

50

Hunt 1982: 128-154. The regional emphasis on Palestine proper is evident e.g. in Itin. Burd. Southern Transjordan seems to have received fewer donations than the other regions of Palaestina and Arabia (Fiema 2001: 120). 51 Constantine himself had initiated numerous ambitious building projects in the Holy Land, see Bagatti 1971: 157-190; Hunt 1982: 6-49. Also the secular building projects concentrated mainly on Palestine and, to a lesser extent, to northern Transjordan (Di Segni 1999: 159-163). 52 E.g. Jerome (Ep. 108.9) commenting on the proconsul knowing the family of the Elder Paula. The governors could hardly ignore such visitors.

46

Libanius Ep. 686: .. =G°W_ dÈ ëti eÓr ÐqwÂm Ñn Ðqw±r, eÓr PakaistÊmgm Ñj PakaistÊmgr. 47 Seeck 1906: 112, 388. 48 Mayerson 1987: 258-259. 49 Mayerson 1987: 257.

205

THE LATE ROMAN ARMY IN THE EAST increased the importance of the city of Aila.59 Roads connected the Negev and southern Transjordan, and the trade from Aila travelled overland to Gaza.60 Constant conflicts with Persia increased the importance of alternative trade routes for long-distance trade around the mid-fourth century. Although Aila thrived, the city of Petra seems to have been in decline by the latter half of the fourth century.61 In general, southern Transjordan and the Negev do not appear as prosperous as Palestine and northeastern Arabia, although the mining industry thrived in the region.

imposed by the aftermath of the rebellion of Jews in 351352 that took place in Palestine.53 Against this background, the detachment of first southern Transjordan and then the Negev from the province of Palaestina in order to ease the administrative burden of the governor of Palaestina was a sensible decision. Both were peripheral areas of difficult conditions from the point of view of a governor mostly residing in Caesarea and focusing on the administration of the Holy Land proper. Also the reason behind the advancement of the rank of the governor of Palaestina from praeses to consularis should relate to the importance of Christianity and the rise of importance of the area. It was suitable for reasons of prestige to have the religiously important area governed by a high-status official. At the same time it bolstered the position of the governor, so that he would be better able to deal with high-prestige persons.54 In this light, Ep. 686 of Libanius, sent in 361, is a reflection of an imperial act aimed at the enhancement of the status of the governor acting as the guardian of the Holy Land.

The interests and issues relevant in the areas of the Negev and southern Transjordan were also probably somewhat removed from the issues important in Palestine proper facilitating their detachment. Consequently, it is probable that the economic and geographic cohesion of the Negev and southern Transjordan was a factor that influenced the reorganisations. Administratively the decisions to move the Negev and southern Transjordan from one province to another in relatively quick succession instead of making them a province of their own indicate not only the apparent unity of the areas, but also a readiness of the imperial decisionmakers to experiment with provincial divisions. The areas were distinct but not important enough to merit formation into a province of their own yet by the middle of the fourth century. On the other hand, revenue from Aila and the trade-related traffic to the Mediterranean coast, as well as the mining industry and efflorescence of the Negev, must have gradually rendered growing local importance to the regions. This most probably eventually constituted a factor arguing for separation of these areas from other provinces and their formation as a province of their own.

Considerations related to the cultural and economic affinities of the Negev and the southern Transjordan might also have constituted a factor influencing the detachment of first southern Transjordan and then the Negev from Palaestina.55 Despite being separated by Wadi Araba, the two regions appear to be intimately connected.56 The countryside of both Arabia and Palaestina seems to have been experiencing growth during the fourth century.57 In the south, the Negev experienced development that has never been seen before or since.58 In southern Transjordan, the importance of the Red Sea and the geopolitical realities that favoured that route and also favoured commercial maritime traffic

In the military sense it must be noted that the detachment of southern Transjordan from Palaestina and its attachment to Arabia c. 340 (and later the detachment of the Negev) most probably enlarged the command of dux Arabiae. Unlike in the time of Diocletian, the concentration of a large number of units under the command of one man was, however, not so dangerous anymore by the reign of Constantius II. The civil-military division was more strictly observed and the division seemingly existed also in Arabia in the mid-fourth century.62 If the lists of units in the Notitia Dignitatum are

53

Aur. Vict. De Caes. 42.11; Jerome, Chron. 282; Socr. Hist. Eccl. XI.33; Sozomen Hist.Eccl. IV.7.5.; The presence of magister militum per orientem Ursicinus proves that the rebellion must have been a serious challenge to Roman authority requiring prompt action (Isaac 1998: 453). For Ursicinus, see PLRE I 985 ’Ursicinus 2’. In a letter dated to 363 (Ep. 1418), Libanius notes the efforts of governor Leontius in putting down banditry in Palaestina. 54 Powerful persons could cause problems for the governors, especially if the governors were not of high standing. See e.g. Brown 1992: 29-33; Lendon 1997: 228-233; Cod. Theod. I.16.4. is a symptom of the problem. 55 An interesting, if somewhat later, phenomenon that can be seen as a reflection of the cultural affinity of southern Transjordan and the Negev is the presence of certain architectural features in churches in southern Transjordan and in the Negev as opposed to ones in northern Transjordan that cannot be completely explained by liturgical differences, see Fiema 2001: 120-121. See also Lewin 2002: 361 for the economic and cultural influence of Petra on the Negev area. 56 E.g. the formulas in funerary inscriptions show dissimilarities between the southern and northern parts of Arabia (Sartre 1993, 144). 57 Walmsley 1996: 151; Safrai 1994: 436-453. 58 Foss 1995: 222-223. The countryside of the Negev was dominated by large villages and consequently only one settlement in the Negev claimed city status – the city of Elusa (Rubin 1996: 51); Lewin 2002: 319-375; Hirschfeld 1997: 33-71. See also Sheresevski 1991.

59 Eusebius, Onom. 6.17-19. Aila is mentioned by Eusebius as being a city to which ships sail from India and Egypt. India in this context means southern Arabia and Ethiopia. Part of the increasing importance of Aila had been the decline of the Egyptian ports of the Eastern Desert due to political instability and insecure conditions (Sidebotham 1989: 222-223). 60 The contacts between southern Transjordan and the Negev persisted also later. Petra Pap. I, 2 (inv 4) reveals Petrans having land-holdings west of Wadi Arabah and residing in Gaza. 61 See Fiema 2002: 238-239. 62 The Not. Dig. (Or. xxxvii) lists the governor of Arabia as dux et praeses. This combined post evidently did not exist yet in the midfourth century. Fl. Silvinianus (PLRE I 842), involved in building

206

JOONAS SIPILÄ: FLUCTUATING PROVINCIAL BORDERS IN MID-4TH CENTURY ARABIA AND PALESTINE consulted, it is evident that the command of dux Arabiae, even with the Negev and southern Transjordan, would not have been unduly large when compared to e.g. command of dux Thebaidos in Egypt.63 At the time the units also had lower numerical strengths than during earlier times.64

Bibliography Abel F.-M. 1967 Géographie de la Palestine. II – Géographie Politique. Les Villes. Troisiéme édition, Paris. Athanasius 1884a Apologia contra arianos. Athanasii opera omnia, tomus I. Migne J.-P. (ed.), PG 25. Athanasius 1884b Historia arianorum ad monachos. Athanasii opera omnia, tomus I. Migne J-P. (ed.), PG 25. Athanasius 1887 Tomus ad Antiochenos. Athanasii opera omnia, tomus II. Migne J.-P. (ed.), PG 26. Aurelius Victor 1993 De Caesaribus. Pichlmayer Fr. (ed.), Lipsiae. Bagatti B. 1971 The Church from the Gentiles in Palestine. History and Archaeology. Studium Biblicum Franciscanum. Collectio Minor No 4, Jerusalem. Balty J. 1980 Sur la date de création de la Syria Secunda. In Syria 57: 465-481. Barag D. 1973 The Borders of Syria-Palaestina on an Inscription from the Raphia Area. In IEJ 23: 30-32. Barnes T.D. 1982 The New Empire of Diocletian and Constantine, Cambridge-London. Beck G. (ed.) 1837 Corpus Iuris Civilis. Vol II, Lipsiae. Bidez J. 1960 Einleitung. Sozomenus, Kirchengeschichte. Bidez J. (ed.), GCS 50 Berlin: i-lxvii. Bowersock G.W. 1980 Mavia, Queen of the Saracens. In Eck W. – Galsterer H. and Wolff H. (ed.), Studien zur Antiken Sozialgeschichte. Festschrift Friedrich Vittinghoff. Kölner Historische Abhandlungen 28, Köln: 477-495. Bowersock G.W. 1983 Roman Arabia, Cambridge, Mass. Brennan P. 1998 Divide and Fall. The separation of legionary cavalry and the fragmentation of the Roman Empire. In Hillard T.W. et al. (ed.), Ancient History in Modern University, 2. Grand Rapids: 238-244. Brown P. 1992 Power and Persuasion in Late Antiquity. Towards a Christian Empire, Madison. Brünnow R.E. and Domaszewski A.v. 1909 Die Provincia Arabia. III Band, Strassburg. Burrell B. 1993 Two Inscribed Columns from Caesarea Maritima. In ZPE 99: 287-295. Coello T. 1996 Unit Sizes in the Late Roman Army. BAR IS 645, Oxford. Devos P. 1967 La date du voyage d’Egérie. In Analecta Bollandiana 85: 165-294. Di Segni L. 1999 Epigraphic documentation on building in the provinces of Palaestina and Arabia, 4th - 7th c. In Humphrey J.H. (ed.), The Roman and Byzantine Near East, Vol 2. Some Recent Archaeological Research. JRA, Suppl. Series no. 31, Portsmouth, RI: 149-178. Egan V. and Bikai P.M. 1999 Archaeology in Jordan. In AJA 103: 485-520.

The enlargement of the command of dux Arabiae also gave him easier access to more troops. This might have been desirable due to the continuing hostilities with Persia that might have included raids at least in the northern parts of Arabia.65 There are also some other scattered indications of lesser problems throughout the fourth century – perpetrators being either brigands or small groups of Saracens.66 Even so, the military factor was not as much a determining factor, as it was a facilitating factor regarding the provincial reorganisations. The decision not to form an independent province of southern Transjordan and the Negev yet at this point reflects hesitancy of the imperial authority, as well as readiness to experiment with provincial divisions in search of optimal solutions to administrative issues. Even so, the existing cultural connections between southern Transjordan and the Negev began to be discerned. Shared traditions, environmental conditions and geographical cohesion tied the two areas together. This unity of Southern Transjordan and the Negev was ultimately recognised as they were combined to form a province that was a viable unit both politically and economically. This last reorganisation occurred c. 392 and resulted in the creation of three provinces of Palaestina – Prima, Secunda and Salutaris/Tertia – and in the truncation of the province of Arabia.67 activities of military nature in 348 and 351 is only dux. Aurelius Valerianus is attested as dux at the time of Constantius II around 352353 (IGLS xiii # 9062). On the other hand, Theodorus addressed in 346/352 (Cod.Theod. IX.40.4) carries only the title praeses Arabiae. This suggests that the offices of praeses and dux were separate in the province of Arabia around the mid-fourth century. Flavius Archelaus (PLRE I 101) attested in 349/350 (AE 1933, 171) is most likely only a civilian governor although with clarrisimate rank (see Sartre 1982a, 102-103). See also Kotansky 1991 for evidence of dux Diogenianus, in office sometime in the latter half of the 4th century and possibly also holding praesidal powers. Kotansky suggests a date between 353 and 367. 63 The military units under dux et praeses Arabiae and dux Palaestinae (Not. Dig. Or. xxxvii; Or. xxxiv) combined consist of 3 legions, 32 cavalry units and 16 infantry units whereas the dux Thebaidos (Not. Dig. Or. xxxi) commanded 6 legions (one of them divided between two places), 25 cavalry units and 11 other infantry units (this accepts the correction of Brennan [1998] for Seeck’s lines 30-31 that combines them to read Equites promoti indigenae legionis tertiae Diocletiano, Ambos). 64 At least the legions appear to be significantly smaller than their counterparts during the Early Empire. See the thorough discussion in Coello 1996. See also Elton 1996: 89-101; Nicasie 1998: 67-76. 65 Libanius is aware of the problem when he notes the difficulty of managing Arabia due to raids of barbarians in a letter written in 364 (Ep. 1236). The raids might have been exacerbated by the aftermath of Julian’s Persian war. 66 Mayerson 1989: 71-74; Isaac 1990: 91-93. 67 Relying on St. Jerome’s statement in Hebr.Quest.in Gen. 21.33-34 (written c. 388-392) and the fact that Notitia Dignitatum lists three Palaestinae (Palaestina, Palaestina Secunda and Palaestina Salutaris)

as well as Arabia. This is confirmed by Codex Theodosianus for AD 409 (VII.4.30) although it lists Tertia for the name Salutaris and names the first province Palaestina Prima – see Sipilä 2004: 341-344.

207

THE LATE ROMAN ARMY IN THE EAST Kotansky R. 1991 Magic in the Court of the Governor of Arabia. In ZPE 88: 41-60. Krüger P. 1887 Codex Iustinianus, Berolini. Lactantius 1984 De Mortibus Persecutorum. Creed J.L. (ed.), Oxford. Libanius 1921 Libanii Opera. Vol. X. Epistulae 1 839. Foerster R. (ed.) Lipsiae. Libanius 1922 Libanii Opera. Vol. XI. Epistulae 840 1544. Foerster R. (ed.) Lipsiae. Lendon J.E. 1997 Empire of Honour. The Art of Government in the Roman World, Oxford. Lewin A. 2002 Il Negev dall’età nabatea all’epoca tardoantica. In Mediterraneo Antico 5: 319-375. Lewin A. 2004 Limitanei and comitatenses in the Near East from Diocletian to Valens. In Le Bohec Y. and Wolff C. (ed.), L’armée romaine de Dioclétien à Valentinien Ier, Paris: 227-236. Mansi G.D. (ed.) 1759a Sacrorum Conciliorum Nova et Amplissima Collectio. vol II, Florentiae. Mansi G.D. (ed.) 1759b Sacrorum Conciliorum Nova et Amplissima Collectio. vol III, Florentiae. Mayerson Ph. 1984 “Palaestina” vs. “Arabia” in the Byzantine Sources. In ZPE 56: 223-230. Mayerson Ph. 1987 Libanius and the Administration of Palestine. In ZPE 69: 251-260. Mayerson Ph. 1988 Justinian’s Novel 103 and the Reorganization of Palestine. In BASOR 269: 65-71. Mayerson Ph. 1989 Saracens and Romans: Micro – Macro Relationships. In BASOR 274: 71-79. Merkelbach R. and Stauber J. (ed.) 2002 Steinepigramme aus dem Griechischen Osten. Band 4. Dei Südküste Kleinasiens, Syrien und Palaestina, Leipzig. Millar F. 1993 The Roman Near East 31 BC - AD 337, Cambridge-London. Negev A. 1977 The Nabateans and the Provincia Arabia. In ANRW II.8: 520-686. Nicasie M.J. 1998 Twilight of Empire. The Roman Army from the Reign of Diocletian until the Battle of Adrianople, Gieben. Patrich J. 1995 Church, State and the Transformation of Palestine - The Byzantine Period (324 - 640 CE). In Levy T.E. (ed.), The Archaeology of Society in the Holy Land, London: 470-487. Prinz O. (ed.) 1960 Itinerarium Egeriae. Samlung vulgärlateinischer Texte. Rea J. 1983 Pap. Oxy. 3574. Petition to a praeses Herculiae. The Oxyrhynchus Papyri. Bowman et al. (ed.), vol L: 183-6. Roll I. 1989 A Latin Imperial Inscription from the Time of Diocletian Found at Yotvata. In IEJ 39/3-4: 239260. Rubin R. 1996 Urbanisation, Settlement and Agriculture in the Negev Desert – The Impact of the RomanByzantine Empire on the Frontier. In ZDPV 112/1: 49-60. Safrai Z. 1994 The Economy of Roman Palestine, London. Sartre M. 1982a Trois études sur l’Arabie romaine et byzantine. Collection Latomus vol. 178, Bruxelles. Sartre M. 1982b Bostra. IGLS XIII, Paris.

Elton H. 1996 Warfare in Roman Europe AD 350-425, Oxford. Epiphanius 1922 Panarion. Holl K. (ed.) GCS 31, Leipzig. Eusebius 1890 Eusebii Caesariensis Opera. Vol. IV. Historiae ecclesiasticae, libri I-X. Dindorf W. (ed.) Lipsiae. Eusebius 1904 Eusebius Werke III: Das Onomastikon der Biblischen Ortsnamen. E. Klostermann (ed.) GCS 11. Leipzig. Eusebius 1857 Commentaria in Isaiam. Eusebii Pamphili: Opera omnia quae exstanta, tomus VI. Migne J.-P. (ed.) PG 24. Fiema Z.T. 2001 Appendix A: Southern Jordan and the Negev: Comparative Ecclesiastical Architecture. In Fiema Z.T. - Kanellopoulos C. - Waliszewski T. and Schick R. (ed.), The Petra Church, American Center for Oriental Research Publications 3, Amman: 120121. Fiema Z.T. 2002 Petra and Its Hinterland during the Byzantine Period: New research and interpretations. In Humphrey J. (ed.), Roman and Byzantine Near East 3. Some New Discoveries. JRA, Suppl. series 49, Portsmouth, RI: 191-252. Foss C. 1995 The Near Eastern countryside in late antiquity: a review article. In Humphrey J.H. (ed.), The Roman and Byzantine Near East. Some recent archaeological research. JRA, Suppl. Series 14. Portsmouth, RI: 213-234. Frösén J. - Arjava A. and Lehtinen M. 2002 The Petra Papyri I. American Center of Oriental research publications 4, Amman. Graf D.F. 1988 Qura ǥArabiyya and Provincia Arabia. In Géographie historique au Proche-Orient. Notes et Monographies Techniques 23: 171-211. Graf D.F. 1992 The Syrian Hauran. In JRA 5: 1-25. Hermannus A. (ed.) 1887 Codex Iustinianus. Corpus Iuris Civilis. Vol III. Fratres Kriegelii (ed.) Lipsiae. Hirschfeld Y. 1997 Farms and Villages in Byzantine Palestine. In DOP 51: 33-71. Hunt E.D. 1982 Holy Land Pilgrimage in the Later Roman Empire, Oxford. Innocentius 1845 Epistolae et Decreta. Migne J.-P. (ed.) PL 20. Isaac B. 1990 The Limits of Empire: The Roman Army in the East, Oxford. Isaac B. 1998 The Eastern Frontier. In CAH XIII: 437460. Jerome 1969 Hebraicae questiones in Libro Geneseos. Antin P. (ed.), S. Hieronymi Presbyteri Opera. Pars I. Opera Exegetica 1. Hebraicae questiones in Libro Geneseos etc. CCSL LXXII. Jerome 1877 Epistolae. Migne J.-P. (ed.) PL 22. Jones A.H.M. 1964 The Later Roman Empire 284-602: A Social, Economic, and Administrative Survey. Vols. I-II, Oxford. Kaster R.A. 1988 Guardians of the language: the grammarian and society in Late Antiquity, Berkeley. Kettenhofen E. 1981 Zur Nordgrenze der provincia Arabiae im 3. Jahrhundert n. Chr. In ZDPV 97: 62-73. 208

JOONAS SIPILÄ: FLUCTUATING PROVINCIAL BORDERS IN MID-4TH CENTURY ARABIA AND PALESTINE Sartre M. 1993 Inscriptions de la Jordanie, Tome IV: Pétra et la Nabatène méridionale, du wadi al-Hasa au Golfe de Aqaba. IGLS XXI, Paris. Seeck O. 1906 Die Briefe des Libanius. Texte und Untersuchungen zur Geschichten der Altchristlichen Literatur, Neue Folge 15, Leipzig. Sheresevski J. 1991 Byzantine Urban Settlements in the Negev Desert, Beer-Sheva. Sidebotham S.E. 1989 Ports of the Red Sea and the Arabia-India Trade. In T. Fahd (ed.), L'Arabie préislamique et son environnement historique et culturel: actes du Colloque de Strasbourg 24-27 juin 1987. Travaux du Centre de recherche sur le ProcheOrient et la Grèce antiques / Université des sciences humaines de Strasbourg, 10. Brill: Leiden: 195-224. Sipilä J. 2004 Roman Arabia and the Provincial Reorganisations of the Fourth Century. In Mediterraneo Antico 7: 317-348. Sozomenus 1960 Kirchengeschichte. Bidez J. (ed.) GCS 50, Berlin. Syon D. & Hartal M. 2003 A New Tetrarchic BoundaryStone from the Northern Hula Valley. In Scripta Classica Israelica 22: 233-239. Theodoret 1911 Kirchengeschichte. Parmentier L. (ed.) GCS 19, Leipzig. Tsafrir Y. 1986 The Transfer of Negev, Sinai and Southern Transjordan from Arabia to Palaestina. In IEJ 36,1-2: 77-86. Walmsley A. 1996 Byzantine Palestine and Arabia: Urban Prosperity in Late Antiquity. In Christie N. and Loseby S.T. (ed.), Towns in Transition, Aldershot: 126-158. Whittaker C.R. 1994 Frontiers of the Roman Empire: A Social and Economic Study, Baltimore and London. Zosimus 1887 Historia Nova. Mendelssohn L. (ed.), Lipsiae.

209

Zosimos II.34.1 and ‘The Constantinian Reform’: Using Johannes Lydos to Expose an Insidious Fabrication Peter Brennan University of Sydney

earlier periods, such as the time of Constantine.5 A major artefact for locating a Constantinian stratum is Zosimos II.34, but it is a very questionable marker in such an excavation. All scholars, even Van Berchem, recognise the polemical nature of his treatment of Constantine and its impact on his evaluation of Constantine’s military practice. Most recognise that Zosimos preserves at best a half-truth on the source of comitatenses.6 The facts, too, even in this half-truth, are vulnerable.

We do well to remind ourselves how fragile is our real knowledge of Constantine’s role in the history of the lateRoman army. There is no contemporary literary evidence. Aurelius Victor says that he renewed the military; he offers no details.1 Two 6th century Byzantine authors, Zosimos and John the Lydian, provide such details as exist.2 Both say that Constantine established generals and transferred from the praetorian prefects to them command over the soldiers. John gives very precise figures for the size of Diocletian’s army and a suspiciously general claim that Constantine doubled it when he gained control of the east, the other half of the empire. Victor and Zosimos also say that Constantine abolished the praetorians.3 Zosimos at II.34.1 makes a general comparison between the practices of Diocletian and Constantine. His specific claim, leaving aside his prejudiced evaluation of it, is as follows: “Since, by the foresight of Diocletian, the frontiers of the Roman empire were everywhere studded with cities and forts and towers, in the way I have already said, and the whole army was stationed there, the barbarians had no way through; for everywhere there was an opposing force able to resist their attacks. Constantine destroyed this security by removing the greater part of the army from the frontiers and placing them in cities which did not need help”. This passage, either explicitly or implicitly, shapes much of the analysis of the scatter of documentary and archaeological evidence for the Constantinian army. It is not what it seems.

Few scholars now accept Zosimos’ gross contrast between the practices of Diocletian and Constantine. Evidence, especially archaeological, increasingly shows that both were concerned with the frontiers and their armies; Constantine complemented and strengthened the limital works of his predecessors. Diocletian and his successors already had court armies separate from frontier armies and these were based in interior cities, including, but not restricted to, the new imperial ‘capitals’. The term comitatenses had already appeared by the time of Licinius at the latest. The commander of equites Dalmatae Aquesiani comit(atenses) dedicated a temple at Prutting in Noricum after the death of Galerius in 311 to commemorate a victory won on 27 June, 310. A vexillatio equitum Dalmatarum comit(atensium) Anchialitana had been based at Anchialos on the Black Sea long enough to incorporate that site in its title by the time a soldier named Valerius, who had served in it for 20 years, was buried at Thyatira; on all criteria, his career predates Constantine’s arrival in the east.7 Thus, the term comitatenses, whatever its precise origin and significance at this time, existed for units serving in the Danubian region before Constantine controlled it. Whoever created this term, it was not Constantine. Most likely it was Galerius or Licinius, two military emperors whose roles in the development of the late-Roman army can no longer be retrieved from the sources; they are further hidden by a modern tendency to assign to Constantine most fourth century military innovations before AD 354, when Ammianus Marcellinus’ history provides some better control. Zosimos II.34 is the magnet that attracts to Constantine’s name the military developments of his predecessors and of his sons, in the process obscuring both his achievement and theirs.

Theodor Mommsen’s depiction of the late-Roman army as the joint, evolutionary creation of Diocletian and Constantine held sway until Denis Van Berchem, in a masterly work that took up the lead of Eric von Nischer, argued a revisionist view that became the orthodoxy when taken into the magisterial works of A.H.M. Jones and Dietrich Hoffmann.4 In this view it was Constantine who took the major step in creating comital armies (comitatenses) separate from frontier armies (limitanei, though this is a post-Constantinian term) in the dual structure that is the distinctive feature of the late-Roman army, as set out in the early 5th century text known as the Notitia Dignitatum. That text is a tralatician one, problematical for the nature of the Roman army for its own date(s) and notoriously difficult to excavate for 1

5 From a massive bibliography, see Clemente 1968; Brennan 1996. My forthcoming edition and commentary on the text (to be published by Liverpool University Press in its Translated Texts for Historians series) revisits these problems and the use of this text. 6 For instance, Van Berchem 1952: 115; Jones 1964: 1.100; Hoffmann 1969: 1.186-187. 7 CIL III 5565 & 11771; CIL III 405.

Aur. Victor Caes. 41.12. Zos. II.32-34; Joh. Lyd. de mag. II.10; III.31; III.40; de mens. I.27. 3 Aur. Victor Caes. 40.25; Zos. 2.17. 4 Mommsen 1910; Von Nischer 1923; Van Berchem 1952; Jones 1964; Hoffmann 1969; Seston 1955, in an excellent critique of Van Berchem, gave good reasons for reducing, rather than expanding, the role of Constantine in the evolutionary history of the late Roman army. 2

211

THE LATE ROMAN ARMY IN THE EAST Zosimos’ claim that Constantine removed the greater part of the troops from the frontiers is at best patent rhetorical exaggeration and it is generally recognised as such. If he removed any troops, it was few. Dietrich Hoffmann’s detailed analysis has revealed that Constantine can have detached from frontier armies no more than a very few of the comitatenses listed in the Notitia Dignitatum: at most he can find some units derived from old legions and a few cavalry units, and both are very tentatively and inconclusively argued.8 Most comitatenses had a different origin: some units already served in the court armies of Constantine’s predecessors; some units were created a novo for the comitatenses by Constantine and his successors; some units were formed as seniores and iuniores from existing units (whatever the actual mechanics of their formation, much disputed, no one sees them as detached frontier units). Many units, those classified as pseudocomitatenses, were units detached from frontier armies, but not until after the time of Constantine. It is this later large-scale detaching of units that very likely influenced Zosimos, or, more likely, his late fourth century source, to misinterpret data from the Constantinian era in terms more appropriate to his own time.

troops for a specific purpose into a general Constantinian reform of the comitatenses and evaluated it tendentiously. The seed of truth in John’s material, whatever it is, has been turned into a germ of error by Zosimos and by those scholars who use it to flesh out a ‘Constantinian reform’. There are already scholars who refuse to use Zosimos for Constantine’s practice. Fergus Millar would question every aspect of the changes he attributes to Constantine, at least as they affect the East.11 Benjamin Isaac and Dick Whittaker dismiss him because his facts do not fit the evidence.12 The case for dismissal is even stronger if one can show that these facts derive from a misunderstanding, whether unconscious or deliberate, of his source. Scholars have shown scant interest in pursuing the source of Zosimos’ facts, though passages in John the Lydian’s de Magistratibus are occasionally cited, as if the similarities between the two vindicate Zosimos’ statements.13 But do they? This is what John says: “After Constantine with Fortuna left Rome and the forces which were guarding the bank of the Ister were distributed over Lower Asia by the emperor’s decision, the state was deprived of Scythia and Moesia and their revenues, for the barbarians from over the Ister overran Europa with no one to oppose them. Since those in the east were burdened with immoderate imposts contrary to the former situation, it became necessary that the prefect no longer command the court and the forces under arms (the former being handed over to the so-called magister, the latter being turned over to the recently established generals); he was in future to have the title Prefect of the East, administering the East in addition to Lower Asia and its affairs.”14 In another passage (III.31) John notes again the dispersal over Lower Asia of the forces guarding the Ister and the loss of Scythia and Moesia and their revenues, this time specifically attributing the actions to Constantine and giving a reason, “in fear of tyranny”. After a digression on the Danube, John (at III.33) reiterates the loss of provinces and their revenues and notes a reorganisation of Syria and Palaestina and the need to create a praetorian prefect of the east, together with prefects of Africa, Gaul, Illyricum and Italy. This time there is no mention of a transfer of the prefects’ powers to the generals; the changes are, instead, related to plans for a surprise attack on the Persians and for this Constantine’s own writings are given as the source.

Most scholars, even those uncomfortable with Zosimos’ facts, are understandably reluctant to jettison a source that would leave no fixed point around which to shape a largely shapeless scatter of information. Guy Sabbah, reflecting well the continuing influence of Van Berchem’s book 50 years after its publication, called this text, in his paper at the 2002 Lyon Congrès on the army of this period, “le texte essentiel sur la réforme constantinienne”.9 Scholars ritually note Zosimos’ limitations, and then appeal to him as preserving a seed or kernel of truth or, in the ambiguous metaphor of a recent overview of the late Roman army, “a germ of truth”.10 The seed in Zosimos II.34 is only a seed of truth, if the facts on which it is based are correct - and germane to Constantine’s military practice. This paper aims to show that this is not so and that Zosimos based his generalisation about Constantine’s military practice on information, given more explicitly in John the Lydian, about a specific and temporary withdrawal of troops from lower Danubian frontier armies. It is not clear whether this withdrawal occurred under Licinius or under Constantine, but it was not about the detaching of units to form a permanent field army. It will be shown, further, that Constantine’s reorganisation of the Danubian armies actually transferred comital army units to frontier armies; this implies a different strategy to the one Zosimos implies, indeed, one that continues and develops the practice of his predecessors to strengthen territorial armies. Zosimos has turned a withdrawal of Danubian

John and Zosimos seem to derive their information from a source tradition about prefectural changes (the focus of both authors at this point); they are not following a single, common source, since they differ on the territorial jurisdiction of the prefects. John’s military details are peripheral to the prefectural changes and his connection between the two is obscure at best - thus the military and prefectural details were already linked in an established

8 Hoffmann 1969: 1.186ff., 226ff., 261-262; see below for more detail on possible transfers from Danubian armies. 9 Sabbah 2004: 31 stressing the priority that Van Berchem gave to literary sources; similarly Paschoud 1971: 235. 10 Southern and Dixon 1996: 17; Hoffmann 1969: 187-188 (“ein wahrer Kern”); implicitly, Jones 1964: 1.100; 2.608.

11

Millar 1993: 209-210. Isaac 1990: 162-163; Whittaker 1994: 206-207. 13 Paschoud 1971: 235; Nicasie 1998: 15-16, 39-40. 14 Joh. Lyd., de Mag. II.10, repeated with a few verbal differences at III.40 (text and adapted translation from Bandy1983). 12

212

PETER BRENNAN: ZOSIMOS II.34.1 AND ‘THE CONSTANTINIAN REFORM’

context of a planned Persian war in this later period, but John’s military details are better related to the only withdrawal of troops and subsequent incursion recorded – that is, the events around 324 given in one of the better texts for the period.16

tradition. Zosimos’ military details are even more peripheral to prefectural changes, for the introductory words which attach them to his account of those changes show them as very much an afterthought, one stimulated and developed from a particular detail that he and John have in common: “Constantine did something else, which gave the barbarians unhindered access to the land subject to the Romans”. One is struck by the conceptual correspondence between John’s specific details and Zosimos’ generalisations. Zosimos extracted from common material the detail of barbarians overrunning Danubian frontiers unopposed; he slightly rephrased it in his new introduction as barbarians having unhindered access to Roman lands in general and also generalised the details. He, thus, turned the withdrawal of troops from the Danube into a general policy of withdrawal from frontiers and he turned the placing of them in lower Asia into a general policy of placing withdrawn units in cities not needing help. He then contrasted this generalisation with his view of Diocletian’s policy (preclusive frontier defence), given in an earlier, now lost, part of his work and here reiterated and praised.

According to this source, just prior to the civil war of 324, Goths devastated Thrace and Moesia after breaking through limites neglectos; in fear of Constantine, who was based at Thessalonica, and as a result of some unspecified action on his part, they made peace and returned captives; as a consequence, Licinius complained that his parts had been taken over by Constantine. There is no compelling reason to doubt a Gothic incursion at this time.17 It is a reasonable assumption that the limites were neglected because troops had been withdrawn in preparation for the civil war.18 If so, this was an emperor reinforcing a civil war army in the usual way, not forming a new permanent field army. One would expect the emperor to have been Licinius, for these troops were in his part of the empire; since his court was based at Nicomedia in lower Asia, it is plausible that he concentrated his withdrawn units there “in fear of tyranny” (that is, against an imperial rival), in the words of one of John’s passages.19 E.A. Thompson speculated that it was Constantine who withdrew Licinius’ troops before the civil war.20 This seems improbable in itself and was certainly not the case if the troops were withdrawn to lower Asia, for it was only after the civil war that lower Asia was available to Constantine.

If Zosimos does not base his generalisation about Constantine’s actions on the facts preserved in John, then his view remains the assertion of a prejudiced and tendentious source, an assertion for which one can find no compelling support, but much evidence which runs counter to it. If Zosimos does base it on the material preserved in John, as seems very likely, then it is John, not Zosimos, who provides the only explicit literary details about Constantine’s role in the formation of comitatenses. John, unlike Zosimos, does not put them in this context. Martijn Nicasie has seen the need to address explicitly the significance of John’s material.15 He takes John to refer to Constantine’s withdrawal of units from the lower Danube to form a permanent field army based at Constantinople in lower Asia and thus as corroborating Zosimos’ generalisation. This is not the natural inference to be drawn from John’s details, though their precise historical context remains elusive. Their dating criteria seem to point to two different periods. There is independent evidence for a withdrawal of lower Danubian troops and subsequent barbarian invasion in the context of the civil war of 324, but no evidence for Constantine preparing a war against Persia at this time. On the other hand, there is independent evidence for his preparing a war against Persia in the latter years of his reign, but no evidence for a withdrawal of units from the Danube or barbarian incursions across the Danube at that time. It is, in fact, most likely that John, in characteristic fashion, is telescoping and muddling two sets of changes to the prefecture (which is his real interest), one in the context of the civil war of 324, the other associated with the creation of territorial Caesars in 335 and preparations for a Persian war. One could conjure out of the sparse historical record for the period a scenario of military withdrawal and barbarian invasion on the Danube in the 15

It might be argued that Constantine retained these withdrawn units in an expanded central army after the civil war. Hoffmann considers two groups of comitatenses listed in the Notitia as possible transfers from Danubian provincial armies at this time. First are six legions which must originate in detachments from Danubian legions – Primani and Decimani from the Moesia II legions I Italica and XI Claudia; V Macedonica and XIII (Gemina) from the homonymous legions of Dacia Ripensis; X Gemina and XIV (Gemina) from the homonymous legions of Pannonia I.21 Hoffmann attributed to a single occasion under Constantine the permanent transfer into comital armies of each of these, and other, detachments from long-established legions, almost totally on the basis of Zosimos’ generalisation.22 16

Origo Constantini 21. Barnes 1976: 149-153 sees this Gothic war as a doublet of the Sarmatian war. The latter, which the author never mentions, provided the only sure victory title of this time, but its significance in providing a victory title (which so enraged Licinius that he had coins commemorating it melted down) and a casus belli to ignite the cold war smouldering between the rivals since 321 does not preclude a subsequent Gothic incursion which did not result in a victory title. 18 If, as is persuasively argued in Stefan 2000: 33-54, the two Scythian legions interchanged their garrisons after this war, a total withdrawal of the legions, in the way Origo and John describe for lower Danubian units in general, would have made such an exchange more feasible. 19 Joh. Lyd. de mag. III.31. 20 Thompson 1956: 378-9. 21 ND Or. 6.45-46; 7.39 & 42; 8.38-39. 22 Hoffmann 1969: 1.186-188, 226-232; 2.87 n.150. 17

Nicasie 1998: 39-40.

213

THE LATE ROMAN ARMY IN THE EAST Without Zosimos, there would be little reason to posit a large-scale reform. It patently did not happen in the East, whose long-established legions seem to have provided no permanent detachments for comital armies and it is clearly not what John had in mind. It is certainly possible that Constantine retained detachments from frontier legions in the expeditionary army that was continuously engaged in warfare on the Danube in the next decade. In doing that he need only have been following the practice of his predecessors; so, too, if he had used such detachments to form part of the expeditionary army against Persia late in his reign. The fate of this army on his death might well be a more crucial point, though an equally irretrievable one, in the expansion and history of the comitatenses than anything Constantine did himself.

origin of the numbered Dalmatae in the later court and regional armies is that they were new creations, not Constantinian transfers from Danubian armies. A numbered series of units seems much more appropriate to a set of new creations than to a restructured set of detached units; the latter tend to be identified by site names, as Dalmatae Passerentiacenses and the units of catafractarii with Gallic site name suffixes.27 The upshot of all this is that John the Lydian provides no good evidence that Constantine transferred units from Danubian frontier armies to form or expand a central army. In fact, when one examines Constantine’s actual reorganisation of the Danubian armies, inasmuch as it can be reconstructed from its traces in the Notitia Dignitatum, its essence suggests a rather different strategy. It is now generally agreed that Constantine strengthened Danubian frontier armies by adding two new categories of units, auxilia and cunei equitum.28 The source and status of these units is controversial. There are essentially two views on their source, leaving aside units carrying dynastic names derived from Constantine or his house, almost certainly new creations. One view would see them as a reorganisation of existing units in provincial armies; the other would see them as transfers from other armies. The former is the prevailing view, argued by Van Berchem, specifically in seeing cunei in the Notitia as reconstituted equites, and accepted by Hoffmann. It exists now almost faute de mieux, for the criticisms of a range of scholars have removed one by one most of the planks on which it rests.29 A major argument for reconstitution lies in the overlap in nomenclature of unit types in the two categories, but the failure to recognise a peculiarity in the Notitia has made the overlap look greater than it is.

The second putative group of transfers is a series of numbered Equites Dalmatae listed among comitatenses in the Notitia.23 Hoffmann suggested that they were originally Dalmatae in Danubian provincial armies, transferred into the field army under Constantine.24 Ralf Scharf, noting that there were no Dalmatae in the Notitia in Moesia II or Scythia, refined this speculation to find the numbered Dalmatae specifically in these “missing” Dalmatae; he saw them as units detached by Licinius for his civil war.25 It would seem unlikely that Constantine took into his comitatenses such a large number of units (at least nine) from Licinius’ defeated army. I, too, have noted the anomaly of the “missing” Dalmatae (alongside other cavalry units “missing” if the disposition in other Danubian ducates once extended to the lower Danube), but I argued a different history for them as the source of the Equites Dalmatae Illyriciani in oriental ducates.26 This is not the place to pursue that argument, but I still believe that there are much better grounds for regarding oriental Illyriciani as units transferred from Danubian (i.e. Illyrican) armies than as units settled from the third century field cavalry under Aurelian or Diocletian, the traditional view. The date and context proposed for the transfer was always provisional. The most obvious date was when Galerius returned to the lower Danube to detach reinforcements for his Persian war; he might well have resettled these detached troops as part of a tetrarchic military reorganisation of the east from Egypt to the Persian border. A later date for such a transfer cannot be ruled out, for there is no direct, unequivocal evidence for Illyriciani in the East before the late fourth century, but it is almost inconceivable; it would necessitate a seismic reconfiguration of the history of the late Roman army in the Near East. Regardless of the fate of the “missing” Dalmatae, the simplest and best explanation for the

The Notitia lists equites in each of the ducates from Noricum to Moesia I in a patterned way; each has two units of Promoti, two of Sagittarii and multiple Dalmatae; Valeria and Pannonia I each has a unit of Mauri.30 Moesia I also has two units of Promoti, two of Sagittarii, and three Dalmatae, but they are categorised as cunei.31 Here equites have been reclassified as cunei, with overlapping names, but this represents a bureaucratic, not a substantive, change. When, under Theodosius or Arcadius, Moesia I was transferred from the western to the eastern court, all units in its equites category were reclassified as cunei, for that was the existing classification of its first-listed unit (Constantiaci) and of all other Danubian cavalry units administered from this court; there was probably no longer any great 27

ND Oc. 6.57; Or. 5.34; 6.36; 8.29; most pseudocomitatenses are also identified in this way. 28 Nicasie 1998: 39-40; Lewin 2003: 73-74 with notes 2-4 for a good overview of scholarship on cunei. 29 Van Berchem 1952: 93-100; Hoffmann 1969: 211 et passim. Clemente1968: 134-145 and Soproni 1974: 59-70 give good arguments against reorganisation and conclusively reject one of its cardinal elements, the view that the Equites strand in the Notitia is obsolete. 30 ND Oc. 32.28-38; 33.29-45; 34.16-23, 31.36; there is only one unit with different nomenclature (Flavianenses: Or. 33.45). 31 ND Or. 41.13-19.

23

ND Or. 5.36-37; 6.37; 7.27; Oc. 6.56=7.174. Equites IX Dalmatae (Or. 5.37) implies at least nine units. 24 Hoffmann 1969: 1.261-262. 25 Scharf 2001: 186-193; he offers no history for the other unit-types “missing” from the same provinces, that is, promoti and sagittarii. 26 Brennan 1998: 2.241-244. Lewin 2004: 231-234 noted the argument without rejecting or accepting it. Scharf 2001: 186 cited the article, but did not address its reasons for rejecting the traditional view of the source of Illyriciani in oriental ducates; his proposal for Illyriciani is a variant of the traditional view.

214

PETER BRENNAN: ZOSIMOS II.34.1 AND ‘THE CONSTANTINIAN REFORM’

Their categorisation in the Notitia as legiones not auxilia is no barrier to such a history.36 Secondly cunei. There are two cunei units each of Divitenses, Fortenses and Solenses.37 The three in Dacia have the additional name Dalmatae, so it is likely that all belonged to the Dalmatae cavalry-type. If the full names of the two Dacian cunei equitum Dalmatarum without further name were cunei Dalmatarum Martensium, they would match the auxilia Martensium also in Dacia.38 This would complete the pattern of four auxilia in Danubian frontier armies whose names are replicated in four sets of cunei equitum.

difference between equites and cunei. Another pattern also associates cunei in Moesia I with original equites and thus strengthens the inference that they were part of the original equites disposition. The cunei in Moesia I are listed in the Notitia in topographical order, all except the first-listed unit (Constantiaci), a real cuneus; this topographical order parallels that of the equites category in all other ducates, whereas cunei in the neighbouring ducates (Dacia Ripensis and Pannonia II) are listed nontopographically. The equites category, once the cunei in Moesia I are placed in their correct category as equites, shows little overlap with the cunei. The same unit-types rarely occur in both categories. Cunei in these Danubian ducates include no Sagittarii or Mauri and only one unit of Promoti. Equites in these Danubian ducates include no Scutarii, Stablesiani, Catafractarii, Armigeri or dynastic names, the types appearing as cunei. Only Dalmatae are common to both classifications. There are eight cunei Dalmatarum (five of them in Dacia Ripensis) alongside twenty-five equites Dalmatae. Some of the cunei Dalmatarum are patently not reconstituted Danubian equites Damatae, but units transferred from elsewhere, as has often been recognised. It is not necessary that any of them were reconstituted equites and thus it is best to keep totally separate the two Danubian cavalry categories, equites and cunei.

The source of this group of associated elite infantry and cavalry units, linked by nomenclature, in Danubian armies is best found in parent units in Constantine’s civil war army of 324. Inscriptions reveal that infantry Divitenses were in origin a detachment of the old Norican legion II Italica, apparently once based at Divitia (Deutz, opposite Köln). Several soldiers from infantry Divitenses and cavalry Dalmatae Divitenses and Dalmatae Fortenses left sepulchral monuments in Italy sometime in the early 4th century to judge from their names and other indicators.39 Most regard all these units as having fought for Constantine in the civil war between Maxentius and Constantine in 312, though Hoffmann argued that the cavalry units, but not the infantry units, were in the defeated army of Maxentius. He found them a past history as units of Maximian who had passed to the new Caesar, Severus, and later deserted to Maxentius, and a future history as the beaten soldiers of Maxentius who were sent by Constantine, according to a panegyrist, from the fleshpots of Rome to guard the Rhine and Danube – in this case, rather improbably, to Licinius’ Danubian frontier, rather than to his own.40 One cannot categorically reject this ingenious reconstruction, but a

This alternative view of the source of auxilia and cunei, that is, as transfers from comital armies, has occasionally been suggested.32 The failure to pursue this view might be another legacy of Zosimos in persuading us to look only in one direction for Constantine’s actions. Again nomenclature gives the key. Listed among palatine and comital units in the Notitia, amongst its most senior units (generally recognised as going back to the earliest comital armies), are legions named Divitenses, Fortenses, Solenses and Martenses.33 Two of these names reappear among auxilia and three among cunei equitum in Danubian frontier armies in the Notitia and a case can be made that all four were once represented in both auxilia and cunei. First auxilia. Auxilia Fortensium and Martensium are listed.34 While there are no auxilia Divitensium or auxilia Solensium, infantry units named Divitenses Gallicani and Solenses Gallicani, listed in the Thracian magistral army, could once have been in Danubian frontier armies, as were other units in this list.35

Hoffmann 1969: 1.474f. suggests that these Gallicani came from Gaul with Richomeres to Thrace in 377 (Amm. XXXI.7.4) and were left in the Thracian army, but such broken pairings would be odd for units deriving directly from another field army (Divitenses-Tungricani and Solenses-Martenses were the normal pairings). 36 The unit category auxilia (sometimes auxiliares, sometimes milites) in provincial armies includes unit-types categorised as legiones in field armies. This is a fact, however perplexing. Auxilia is used in this form only for Danubian armies; homonymous infantry units in other western non-field armies are named milites or numeri. It seems, thus, to have been devised purely for use in the Danubian region, perhaps to categorise and distinguish the new elite infantry units there from long established infantry units, both legions and cohorts; cunei would have served a similar purpose for cavalry. All provincial infantry units, of whatever category, when returned to central or regional armies, were categorised as legiones, for auxilia had a different significance in these armies. 37 Dacia has two cunei of Dalmatae Divitenses and a cuneus of Dalmatae Fortenses (ND Or. 42.13, 14, 16); Valeria has a cuneus of Fortenses (ND Oc. 33.28); Scythia and Moesia II each has a cuneus of Solenses (ND Or. 39.13; 40.12). 38 ND Or. 42.17-18 & 26. 39 CIL XI 4787; VI 3637; V 5823; 7000; 7012; AE 1982.258. Hoffmann 1969: 1.117-118; 258- 260. 40 Hoffmann 1969: 1.258-260; Pan. Lat. 9.21.2. This Gallic past of Divitenses seems assured, though its more precise history has been made more problematical by recent epigraphical and archaeological material, see Scheithauer & Wesch-Klein 1990; Carroll-Spillecke 1993 and 1997.

32 Jones 1964: 1.100 writes of units “such as the Dalmatae Divitenses, which have clearly been brought in from Gaul or elsewhere” under Constantine; Hoffmann 1969 occasionally speculates, irresolutely, on a field army origin for cunei (e.g. 1.176-177) or auxilia (e.g. 1.234 Fortenses; 2.154 n. 358); Nicasie 1998: 39 considers, but rejects, the possibility, for it does not fit his assumptions. 33 Divitenses: ND Oc. 5.147; Fortenses: ND Or. 5.45; ND Oc. 5.225 & 255; Solenses: ND Or. 8.34; Martenses: ND Or. 7.40. 34 auxilia Fortensia in Valeria (ND Oc. 33.49); auxilia Martensium in Dacia ND Or. 42.26) 35 ND Or. 8.43 & 50. The Thracian regional army certainly included units detached from the frontier armies, such as Constantini Dafnenses and Balistarii Dafnenses (ND Or. 8.45-6), clearly detached from Constantiniana Dafne. The Divitenses and Solenses are listed in a similar, relatively low place among units in the Notitia. There is a cuneus Solensium in each Thracian ducate (ND Or. 39.13; 40.12.

215

THE LATE ROMAN ARMY IN THE EAST simpler alternative seems possible and thus preferable. All these units, infantry and cavalry, originating in Gallic armies, were part of Constantine’s central army, garrisoned in Italian cities. One can include with them, though unattested in Italy, the infantry and cavalry units associated with the names, Martenses and Solenses. Martenses are attested in a tetrarchic exercitus Gallicanus and it is generally agreed that Martenses and Solenses were a paired creation of a western tetrarchic central army.41 The suffix Gallicani for Solenses and Divitenses in the Thracian magistral list indicates a Gallic past. Hoffmann himself toyed with the idea that some of these Danubian cunei, and indeed the cunei armigerorum as well, were in origin cavalry associated with the likenamed central legions. He let it lapse, because he could not find any cunei Martensium. I have suggested above where they might be found.42 A plausible conclusion is that all these units, deriving from earlier Gallic armies, were in Constantine’s civil war army of 324. It is not a big step to see them as the source of the like-named auxilia and cunei in Constantine’s reconstruction of the military in Danubian regions in the decade following his victory in 324. It is possible that many of the other cuneitypes (scutarii, stablesiani, catafractarii, armigeri), attested as cavalry units in comital armies, were also transferred to Danubian armies under Constantine, but that is pure speculation.

continuation of the long-standing practice whereby oldstyle auxiliary units had lesser privileges than elite units of whatever type and there is no distinction in the privileges of elite units, regardless of the army in which they served. In regard to the discharge of veterans from elite units, the law actually improves the privileges of ripenses by allowing them honesta missio (and access to its privileges) after 20 years service; this puts them on a par with the privileges of comitatenses, for a prior law had required 24 years service before allowing them honesta missio. Whether this prior law was one of Licinius or of Constantine is unclear, but the point is not – Constantine is eliminating a distinction that had occurred in the service conditions of elite units that had been to the detriment of ripenses. It is only in regard to veterans discharged for reasons of age or health that any distinction remains between comitatenses and ripenses: comitatenses are given tax exemptions of 2 capita, regardless of years of service, while ripenses are given 1 caput if discharged between 15 and 24 years and 2 capita after 24 years. Van Berchem found this relatively minor distinction enough to assert that the two types of units were now “inégaux en dignité”, a view that has become widespread.45 It would seem better to emphasise the fact that for all serving soldiers and for all veterans except those discharged on grounds of age or health, it did not matter whether one served in the comitatenses or the ripenses. It is not obvious why transfer from one to the other would be regarded as a degradation any more now than in the time of Diocletian.

There lurks another reason why scholars have made little of such signs that the nucleus of cunei and auxilia in the Constantinian reconstruction were units transferred from comital to provincial armies. Again it is Nicasie who seems explicitly to provide the reasoning, in stating that such a transfer would “have signalled a degradation of the troops concerned”.43 In this view this type of transfer (from field to provincial armies), though almost universally agreed to have occurred under Diocletian, is regarded as a degradation and inconceivable under Constantine. The assumed new factor seems to be the belief that the recent division of elite units into comitatenses and ripenses had created a dual army with unequal privileges and status. The only contemporary reference to the term ripenses is a Constantinian law of 325. Its transmitted text has been often emended, to arguable benefit, and its context remains problematical, but there is little disagreement on its essential content.44 It makes a threefold distinction between comitatenses and ripenses and alares/cohortales in terms of the tax exemptions and discharge arrangements available to each. All serving comitatenses, ripenses and protectores receive the same tax exemptions (4 capita), far more than that accorded alares/cohortales (1 caput). This is a

The history of one unit in this period seems to confirm that there was no great distinction under Constantine between comitatenses and ripenses, for it appears as both comitatensis and as cuneus (I am assuming that cunei were from the beginning classified as ripenses). A unit of Equites Mauri Scutarii, which appears in an Oxyrhynchus papyrus of 6th May 339, includes comitatenses in its title. It is surely one of the two units (or their parent unit) of Equites Mauri Scutarii classified as cunei in the Notitia and listed at Hermopolis and Lycopolis.46 Since units in the category cuneus appear almost exclusively in Danubian provincial armies, except for these two Egyptian units, it is most likely that these units (or their parents) were once in a Danubian army.47 45 Van Berchem 1952: 83-87 recognised that the law made minimal differentiation between comitatenses and ripenses and improved the position of ripenses, but still regarded comitatenses and ripenses as unequal in dignity (p.85). Jones 1964: 97 agreed that in this law “the comitatenses are given superior privileges to the ripenses”. 46 P. Oxy. 4084 – the only reference to the unit that calls it comitatenses. ND Or. 31.23-24 – the omission of Mauri from the latter is a copying error, since papyrological sources over two centuries never omit this element of its name, though they often omit scutarii. Mauri Scutarii were at Hermopolis from 340 to the 6th century and at Lycoplis in 375; the latter still had links to Oxyrhynchus, so probably derives from the unit at Oxyrhynchus in 339. See Lewin 2003 for references and the suggestion that the unit at Oxyrhynchus was the parent of the Notitia units. 47 There are three other cunei listed outside the Danubian region: the one in Britain (ND Oc. 40.54) is a longstanding unit with a different history; the history of the two cunei among comitatenses (ND Or. 7.34;

41

Hoffmann 1969: 1.173-77; 2.64, n.461 on CIL VIII 15551. Hoffmann 1969: 1.176-177. 43 Nicasie 1998: 39; Hoffmann 1969: 1.259 used a similar argument from degradation to argue that only as punishment could Dalmatae Divitenses and Dalmatae Fortenses be the source of homonymous cunei in Danubian armies. 44 CTh. VII.20.4. On this law, see Barnes 2001: 679-680; Campbell 1994: 243-244. 42

216

PETER BRENNAN: ZOSIMOS II.34.1 AND ‘THE CONSTANTINIAN REFORM’

mobile striking force. It is hard to find evidence to back up any elements of such a strategy under Constantine, except for the creation of military commanders at court (and their jurisdiction and importance is debatable). Freed from the assistance of Zosimos’ insidious chains, we can look afresh at the role Constantine and others played in the ongoing history of the late-Roman army – in the Near East, as elsewhere. As an example, I have looked at Constantine’s actual Danubian reorganisation and have found that the nucleus of the cunei and auxilia, the major element in that reorganisation, were units transferred from his civil war army. They acted as elite units, virtual comitatenses in status and operational use, within frontier armies; they were available for detachment to form expeditionary armies in the traditional way or even to be transferred to other provincial commands, as the Mauri Scutarii were transferred to Egypt.

They were transferred to Egypt before 339, via an intermediate presence in an expeditionary army that gave them the status of comitatenses. The most obvious army is the expeditionary one prepared by Constantine for a Persian war; Constantius II did not use it for this purpose, but he might have transferred these units from it to Egypt.48 Such a history of Mauri Scutarii, in Danubian, comital and Egyptian armies in the period 324/339, reveals a continuing flexibility in the use of elite units moving temporarily or permanently between provincial, regional and court commands; the reorganisation of the Danubian armies with units transferred from the west via Constantine’s civil war armies fits the same model.49 It is, thus, no surprise that cunei are still set alongside comitatenses in a law of Constantius II of 347.50 The great distinction between comitatenses and ripenses comes later, its first marker being the creation of the category pseudocomitatenses, by 365 at the latest, for units transferred from provincial armies (ripenses or their equivalent) to central or regional armies. Units still continued to be transferred permanently from provincial to comital armies, but they were no longer made comitatenses, being placed in a separate, probably less privileged, category.51 This represents a major change in mentalité, whose implications for the history of the lateRoman army I hope to pursue elsewhere; its major development lies in the post-Constantinian evolution of the Roman army.

Bibliography Bandy A. 1983 Johannes Lydus, On Powers, or, the Magistracies of the Roman State, American Philosophical Society, Philadelphia. Barnes T.D. 1976 The Victories of Constantine. In ZPE 20: 149-153. Barnes T.D. 2001 Foregrounding the Theodosian Code. In JRA 14: 671-685. Brennan P. 1996 The Notitia Dignitatum. In Les littératures techniques dans l'antiquité Romaine, Foundation Hardt: Entretiens sur l'antiquité classique 42: 147-78. Brennan P. 1998 Divide and Fall. The separation of legionary cavalry and the fragmentation of the Roman Empire. In T. Hillard et al. (ed.), Ancient History in a Modern University. (Grand Rapids): 2: 238-244. Brennan P. 2001 Barbarian Impact on the Late Roman Army. In Tijdschrift voor Geschiednis 114: 261-266 (review of Nicasie 1998). Campbell J.B. 1994 The Roman Army 31BC-AD337. A Sourcebook, Routledge. Carroll-Spillecke M. 1993 Das römische Militärlage Divitia in Köln-Deutz. In KJ 26: 321-444. Carroll-Spillecke M. 1997 The late Roman frontier fort Divitia in Köln-Deutz and its garrisons. In Groenmann-van Waateringe W. et al. (ed.), Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies, Oxford: 143-149. Clemente G. 1968 La “Notitia Dignitatum”, Cagliari. Hoffmann D. 1969 Das spätrömische Bewegungsheer und die Notitia Dignitatum. (Epigraphische Studien 7/1), Düsseldorf. Isaac B. 1990 The Limits of Empire. The Roman Army in the East, Oxford. Le Bohec Y. and Wolff C. (ed.) 2004 L’armée romaine de Dioclétien à Valentinien Ier. Actes du Congrès de Lyon, 12-14 septembre 2002, (Collection du Centre d’Études Romaines et Gallo-Romaines, nouvelle série 26), Lyon.

In sum, I hope to have shown that Zosimos II.34.1, the essential text for the Constantinian reform, holds no truth on Constantine’s actual military practice or policy, He has based his factual generalisation on data found also in closely related passages in John the Lydian. John’s data does not vindicate Zosimos; in fact, it shows where Zosimos’ error lies. John’s details relate to a specific withdrawal of units on the lower Danube in the context of the civil war of 324, not to a planned withdrawal of units as part of a new strategy, even on the Danube, let alone as part of a grand strategy. Zosimos’ prejudiced misinterpretation of the details in his source has fed a powerful myth of Constantine as the creator of a new strategy based on dual field and frontier armies with an enlarged, permanent, privileged central army as its Oc. 6.85) is unknown, but they might derive from Danubian ducates. The term cuneus never occurs as a unit descriptor outside official sources (laws and the Notitia). These observations taken together suggest that cuneus was a term created by Constantine for a specific categorising purpose limited to his Danubian reconstruction. 48 P. Oxy. 4084 shows that the wife and the father-in-law of the soldier had Egyptian names (Aurelia Nonna and Agathos Daimon) – this suggests that the unit had been in Egypt some time before its date of mid-339. 49 For more on this strategy, see Brennan 2001 and Whittaker1994: 203, 207, 22, generously acknowledging material from my unpublished 1972 Cambridge PhD thesis. 50 CTh.V.6.1: universis tam legiones quam vexillationes comitatenses seu cunei. The differences between comitatenses and ripenses had widened by the 370s: CTh.VII.20.8 (AD 372); CTh.VII.13.7 (AD 375). 51 CTh. VIII.1.10 – since actuarii of pseudocomitatenses get lesser payments than those for palatini and comitatenses, it is assumed that the units, too, were differentiated in service conditions; Jones 1964: 126.

217

THE LATE ROMAN ARMY IN THE EAST Lewin A. 2003 The Egyptian cunei. In Tyche 18: 7376. Lewin A. 2004 Limitanei and comitatenses in the Near East from Diocletian to Valens. In Le Bohec and Wolff 2004: 227-236. Millar F. 1993 The Roman Near East 31BC-AD337, Harvard. Mommsen T. 1910 Das römische Militärwesen seit Diocletian. In his Gesammelte Schriften, Berlin: 6. 206-283. Nicasie M. 1998 Twilight of Empire. The Roman Army from the Reign of Diocletian until the battle of Adrianople, Gieben. Paschoud F.1971 Zosime Histoire Nouvelle. Tome 1 Livres I et II, Paris. Sabbah G. 2004 L’armée romaine de Dioclétien à Valentinien Ier. Les sources littéraires. In Le Bohec and Woolf 2004: 31-41. Scharf R. 2001 Equites Dalmatae und cunei Dalmatarum in der Spätantike. In ZPE 135: 185-193. Scheithauer A. - Wesch-Klein G. 1990 Von Köln-Deutz nach Rom? Zur Truppengeschichte der Legio II Italica Divitensium. In ZPE 81: 229-236. Seston W. 1955 Du comitatus de Dioclétien aux comitatenses de Constantin. In Historia 4: 284-296. Soproni S. 1974 Beiträge zur Frage der Liste von Valeria der Notitia Dignitatum. In AAH 26: 59-70. Southern P. & K. Dixon 1996 The Late Roman Army, Routledge. Stefan A. 2000 La legio I Iovia Scythica à Troesmis. Àpropos de CIL III 6174. In Romanité et cité chrétienne. Mélanges en l’honneur d’Yvette Duval, Paris: 33-54. Thompson E.A. 1956 Constantine, Constantius II and the Lower Danube Frontier. In Hermes 84: 378-379. Van Berchem D. 1952 L’armée de Dioclétien et la réforme constantinienne, Paris. Von Nischer E. 1923 The Army Reforms of Diocletian and Constantine and their Modifications up to the time of the Notitia Dignitatum. In JRS 13: 1-55. Whittaker C.R. 1994 Frontiers of the Roman Empire, Johns Hopkins.

218

Two Sieges of Amida (AD 359 and 502-503) and the Experience of Combat in the Late Roman Near East Noel Lenski University of Colorado at Boulder

When John Keegan published his book The Face of Battle in 1976, it began something of a revolution in the study of military history. Turning away from earlier discussions of organization, logistics, strategy and tactics, Keegan focused on the lived experience of front-line soldiers and the way this affected battlefield operations (cf. Holmes 1986). His aim was at once to recreate individual perspectives on combat and to elaborate how these would have affected its process and outcome. Though it took over a decade, Keegan’s approach did not go unnoticed by ancient historians. Beginning with Victor Davis Hanson, the subject has been taken up and vigorously debated by classical Greek historians.1 While trying to establish the details of fighting within the hoplite phalanx, these have pushed and shoved their way like armed combatants around the question of the othismos, its importance, and its function. Ultimately, however, an argument that began very much in the spirit of Keegan’s efforts to reproduce the soldier’s battlefield experience has shifted its focus onto very different questions, above all questions of combat operations at the macro level (See Goldsworthy 1997). Nor have Roman historians kept out of the fray. Among others, Adrian Goldsworthy and Philip Sabin have offered detailed and insightful treatments that have opened a new front in Roman military studies.2 Here again, however, in the absence of adequate ancient battlefield narratives, the debate has often been laid out around conclusions drawn from modern data and models, and has already begun to swing back round to large-scale operational questions.3

of battlefield experience, and, where these are preserved, they tend to be relatively brief and, above all, stylized according to the conventions of ancient historiography (cf. Lendon 1999). More importantly, we have no battle from which multiple accounts from frontline soldiers survive and are thus hard-pressed to offer the sort of dynamic panorama recreated by Keegan. In what follows I hope to show how these problems can be remedied, if not fully overcome, by turning to lateantique reports not of field battles but of sieges. Like Keegan, who concentrated on just three battles, I will keep my focus narrow by examining two sieges, both at Amida, the first in 359 and the second in 502-503. To be sure, our accounts of these do not give us the same level of detail with which Keegan worked nor were they the sorts of daylong battles that he describes. By reading these sources carefully, however, we stand to gain a much better vantage point onto the kind of ground-level, circumstantial information that is necessary to recreate the lived experience of ancient battle from the perspective of those engaged in fighting. To compensate for the absence of multiple perspectives on a single battle, I will supplement with examples from other siege operations reported in our ancient sources. This once again runs the risk of compromising the integrity and specificity of the evidence for the Amida sieges and thereby rendering a pastiche of battles rather than a soldier’s-eye account of a specific battle. Efforts will be made, however, both to limit comparanda generally to late Roman and eastern contexts and to specify the source of each comparison, thus allowing the reader to judge its relevance and value to the two sieges in question.4

It would seem, then, that try as they do to engage the “face of battle” question, ancient historians struggle with the temptation to reduce investigations of battlefield experience to overviews of battlefield operations and to substitute analogies with modern comparanda for eyewitness accounts by the soldiers themselves. The reason for this tendency is, of course, the sources. With very few exceptions we simply lack first-hand accounts

4 The collection of comparisons, though mostly late and eastern, is broad based, but hardly comprehensive. On siege warfare in late antiquity, see Elton 1996: 257-63; Southern and Dixon 1996: 148-67; Haldon 1999: 183-89; Whitby 2000: 310-11; Halsall 2003: 223-27; Lendon 2005: 290-301. On the conduct of ancient sieges more generally, see Marsden 1969; Kern 1999; Campbell and Delf 2003; Campbell and Hook 2005. Far and away the best attested siege in antiquity is that of Jerusalem in A.D. 70, though it is also one of the least representative because of the simultaneous civil war that occurred within the city as the siege was taking place, cf. Price 1992: 115-61; Goldsworthy 1999; Lendon 2005: 233-60. A closer parallel to the sieges narrated here is Iotapata, with its eyewitness testimony by the commander Josephus (BJ III.141-392). Siege archaeology can also help reconstruct the soldier’s experience of battle, see especially James 2004: 31-8; James forthcoming on the siege of Dura Europos in 256; Mitchell 1995: 177-218 on the siege of Cremna in 278; and, for all its flaws, Yadin 1966 on the siege of Masada from AD 73-74.

1

Hanson 2000, originally published in 1989. See also Hanson 1991 and the essays in Lloyd 1996. 2 Goldsworthy 1996: 116-282; 1999; Sabin 2000; cf. Lee 1996; Haldon 1999: 228-33. Already MacMullen 1984, with its focus on the social function of the legion and the social psychology of the soldier, had gone a long way toward investigating this question. Lendon 2005 covers both Greek and Roman warfare in a brilliantly written, if not wholly convincing sweep. 3 Wheeler 1998 and 2001 argues that contemporary historians have relied too heavily and uncritically on modern comparisons to Marshall’s (1947: 50-65) low figures for active participation in combat, which are disputed; he remains skeptical about the value of the “face of battle” question and above all about the use of comparative material to arrive at a composite picture of ancient combat experience. For his version of the Roman experience of field battle see 2004a and 2004b.

219

THE LATE ROMAN ARMY IN THE EAST – both. In times of siege, however, the civilian inhabitants of fortress cities seem to have been well prepared, psychologically, logistically, and militarily, to participate in combat. The importance of these citizen combatants will become clearer below (see below at note 30). Here it suffices to say that a line is better drawn between combatant and non-combatant than soldier and civilian when discussing the experience of combat in a lateRoman fortress city.

Historiographical and Topographical Background Both the siege of 359 and that of 502-503 were undertaken by the Persians in the presence of the Shahanshah himself and both were successful at overrunning the city, though only after a lengthy struggle. More importantly, both are well attested in contemporary, eye-witness sources. In 359, Ammianus Marcellinus, who was serving as a protector in the eastern army under the command of the magister equitum Ursicinus, had taken refuge in Amida in the wake of Shapur II’s invasion of Mesopotamia under the guidance of the Roman deserter Antoninus. Ammianus was present throughout Shapur’s siege of the city and offers a lengthy, detailed, and seemingly accurate account of what he lived through. Moreover, as a self professed quondam miles who played an active role in combat operations and command decisions in 359, his testimony is particularly to be prized.5 Though the 502-503 siege is not reported in eyewitness accounts nor indeed the accounts of soldiers, it is described in three sources close to the event: the contemporary Ps.-Joshua the Stylite lived in nearby Edessa and offers circumstantially credible detail; Procopius of Caesarea seems to have had access to the contemporary history of Eustathius of Epiphaneia, whose narrative climaxed in the Amida siege; and Ps. Zachariah of Mytilene, who wrote an epitome of the Greek Ecclesiastical History of his early fifth-century ‘namesake,’ was himself a monk of Amida and may also have had access to Eustathius’s detailed contemporary account.6

As to the setting, Amida (Turkish DiyarbakÕr) was located on a rock escarpment that created a bend on the right bank of the Tigris River (cf. Amm. XVIII.8.11, XIX.5.4). During the period of growing tensions between Rome and Persia in the 330s, Constantius II deliberately chose the site because of its natural defensibility and good water supply and constructed a fortified city there from the local black basalt.7 After the destruction of Constantius’s enceinte in 359, the curtain walls were rebuilt in the early 370s by Valens and expanded to offer a new home to the inhabitants of Nisibis, who had been expelled from their city when the Persians occupied it in 363.8 It was thus Valens’s walls that were besieged and taken by Kavadh in the engagement of 502-503. After Anastasius regained control of the city from the Persians in 505, he and in turn Justinian are known to have undertaken some rebuilding (Procop. Aed. II.3.27; cf. CIG IV.8898). In their turn, the Arabs conquered the city in 639 and, according to inscriptions still in the walls, also rebuilt portions of the curtain. On balance, however, it is felt that the walls of Amida that still stand today — at 8-12 meters high and 3-5 meters thick, with 4 main gates and 82 towers stretching to a circumference of 5 kilometers — are largely those constructed by Valens.9 In this sense, they give a fairly clear reflection of how the fortifications would have appeared when Kavadh arrived in 502-503. In both sieges not just the walls but also the terrain will have played a serious role, for the rock escarpment that surrounds two thirds of the city rendered the walls that much more resistant to siege engines and undermining. In fact, in neither siege did undermining play an important role, no doubt because of the difficulty of excavating Amida’s basaltic bedrock.10 Moreover, with the steepness of the escarpment on the south and

Given that in 502-503 Amida was defended by its citizenry rather than commissioned soldiers, the search for a soldier’s narrative for this second siege would be futile. This conundrum does, however, draw attention to another issue. The late Roman east had a vast array of both fortresses and fortified cities (Gregory 1997; cf. Lawrence 1983; Lander 1984; Foss and Winfield 1986). The former had military garrisons but no civilians, the latter generally – though not always, as at Amida 502-503 5 Ammianus was also present at a number of the sieges he describes in conjunction with Julian’s Persian invasion of 363. Like all historians, he has had his detractors. Particularly harsh as regards the narrative of the Amida siege is Rosen 1970: 51-68, whose critique is based primarily on bald contrarianism rather than demonstrable fact. The oft noted (e.g. Matthews 1989: 55) geographical gaffe at XVIII.9.2 — where Ammianus reports that Amida borders the Tigris on the south, as opposed to the east — is something of a red herring, for the river bends around the city on east and south. Matthews 1989: 57-66, 287-95 offers the most sympathetic and, to my mind, informed view on Ammianus’s siege narratives. Lightfoot 1989 defends the accuracy of the Amida narrative on yet another crux, and Trombley 1999 examines how Ammianus’s military experience affected his historiography. More on Ammianus’s literary construction of this and other sieges at Crump 1975: 97-113; Sabbah 1978: 579-82; Naudé 1988; den Hengst 1999. 6 On Ps. Joshua see Greatrex 1998: 64-5; Trombley and Watt 2000: xxixxxviii. On Ps. Zachariah and Procopius, see Kugener 1900: 202-3; Haury 1905: xix; Allen 1980: 471-3; 1988: 1-3; Greatrex 1998: 61-66. John of Ephesus (Lives of the Eastern Saints 58 [PO XIX.217-221]) also reports the eyewitness testimony of a monk from the Amidene monastery of John of Urtaye who had survived the siege, though it does not offer insights into military operations.

7

Amm. XVIII.9.1; History of Jacob the Recluse VII.1-5 at Nau (19151917), translated at Dodgeon and Lieu 1991: 154. Amm. XX.11.5 records that Constantius wept when he approached the walls he had constructed and surveyed only a heap of ashes. On the water supply, see below at n. 46. 8 On the fortifications, see CIL III.6730 with Lenski 2002: 378. On the former inhabitants of Nisibis see Zos. III.34.1; Malalas XIII.27; Chron. Pasch. p. 554. 9 For a useful map see Trombley and Watt 2000: map V. More on the walls of Amida at van Berchem and Strzygowski 1910; Gabriel 1940: 85-182; van Berchem 1954: 262-67; Gregory 1997: II.59-65; cf. Trombley and Watt 2000: 120. No attempt at dating is without difficulty. Gabriel 1940 would date the current walls entirely to the fourth century, but van Berchem 1954 entirely to the sixth. 10 Though mining under Amida’s northern and western walls was possible, cf. Ps. Josh. Styl. 71[295-6]. Dara and Petra-Justinianopolis were constructed on similar formations to restrict undermining, Procop. Bell. II.13.20; VIII.11-14.

220

NOEL LENSKI: TWO SIEGES OF AMIDA (AD 359 AND 502-503)

Fig. 1 Western wall with rounded towers (Photo Wikipedia GNU Free Use)

Fig. 2 Rock Escarpment and southern wall with square towers (photo Dick Osseman, with permission)

221

THE LATE ROMAN ARMY IN THE EAST east, only the northern and western sections were approachable from the level, a fact taken into account by the builders, who put more numerous and stronger rounded towers on the walls with a level approach, but fewer and weaker square towers on the walls above the escarpments. The terrain was also taken into account by Shapur II in 359 when he posted his fiercest warriors, the Segestani, and his elephants to the western quarter (Amm. XIX.2.3).

both sides (XIX.2.12-15). Thereafter a plague broke out in the city because of the high number of putrefying corpses from the war dead lying about in the summer heat13; this subsided only after 10 days (the 27th since Shapur’s arrival) when rain showers put an end to the unsanitary conditions (XIX.4.8). Meanwhile, the Persians at some point had moved into a second phase of siege activity: setting up sheds and mantlets behind which to work, they began constructing towers outfitted with ballistae in order to drive the defenders from the ramparts, though skirmishing by slingers and archers also continued (XIX.5.1; cf. 6.6; 7.2). At a point unspecified by Ammianus, a crack unit of seventy Persian sagitarii ex agmine regio were led under cover of darkness by a deserter from the city through a secret passageway hewn out of the escarpment between the Tigris and the fortress; these occupied a tower and set up a purple flag signaling the Persians to make an all out assault on the walls (XIX.5.4-8). Reacting quickly, the Romans simultaneously maneuvered ballistae into place to destroy the intruders and drove back the Persian infantry from their parapet. Though elated with their success, the Romans’ euphoria was soon dashed when they witnessed a throng of prisoners from the neighboring fortress of Ziatha being paraded as captives before them the following day (XIX.6.1-2). The desperation and anger this generated led the Roman command to allow two Gallic legions in the city, the Magnentiaci and Decentiaci, stationed in the east since their defeat at the close of the Magnentius revolt in 353, to make a clandestine night attack on the outposts of the Persians while they were sleeping; though these achieved considerable success in their initial attack, their actions awakened the Persians who soon countered and forced the Gauls, in an orderly retreat, back to the city walls where they were received by the Roman defenders (XIX.6.2-5; 7-11).

The Siege of 359 The siege of 359 had not been intended or planned in advance. Shapur II arrived at the city with a massive expeditionary army which he had hoped to lead deep into Roman territory following the advice of the Roman deserter Antoninus. Bringing with him a force of 100,000 men, he had already terrified several Roman legions into flight, and six of these together with a cavalry unit had taken refuge in Amida, joining its normal garrison of one legion and one cavalry turma. Together with refugee farmers from the neighboring countryside and throngs of visitors who had come to the city for the annual fair, these swelled Amida’s population to an unwieldy 20,000 by the time Shapur reached the walls.11 Once there, Shapur stopped, intending only to reconnoiter, but after he was shot at from the walls, he grew angry and began contemplating a siege; when on the next day his client king Grumbates again approached the walls and attracted a shot from a ballista that proved fatal to the king’s beloved son, Shapur’s anger flared and he resolved to take the city by siege. The siege lasted 73 days,12 and though it eventually succeeded, proved extremely costly, forcing Shapur to abandon his planned invasion deeper into Roman provincial territory (XIX.9.1; cf. XVIII.7.1011; XIX.1.6) and killing 30,000 of his men (XIX.9.9). The siege as such was set in motion only eleven days after Shapur’s arrival, following a weeklong celebration of funeral games for Grumbates’ son and a two day deliberation period. It began when Shapur ringed the city with a curtain of soldiers five-deep which remained silent and immobile for the entire first day, as if to strike terror into the hearts of the defenders by their sheer numbers and organization (XIX.2.2-5). On the second day of the siege (the 12th from Shapur’s arrival) Grumbates himself initiated the fighting by throwing a blood-stained spear toward the fortress in a ritualized drama that set serried ranks of Persians and their allies charging to the walls for two full days of effort at storming the city (XIX.2.6-11). Two days of respite seem to have followed, but on the fifth day of the siege (the 17th since Shapur’s arrival) another day was passed in open assault on the Roman walls by the Persian infantry with heavy casualties on

Although the Gauls had succeeded in killing many Persians, they also strengthened Shapur’s resolve to finish off the city (XIX.6.13-17.1). After a three day truce following the Gallic sortie, the Persians moved their siege towers into place and ordered ranks of mail-clad infantry and cavalry to march on the city to the measured cadences of trumpets. Because the iron-clad Persian siege towers were higher than Amida’s walls, the Romans suffered serious casualties on the first day they faced them; by the second they had devised a strategy of stationing scorpiones (onagri) against them with the effect that they successfully smashed the towers’ joints and caused them to collapse into piles of wood, easy targets for fire arrows (XIX.7.2-7). As with the Gallic sortie, however, this setback only strengthened Shapur’s resolve to take the city (XIX.7.8-8.1). In the midst of

11 Amm. XVIII.2.13; XIX.2.14: ad usque numerum millium viginti cunctis inclusis. Clark 1910: 160 emends the MS reading to ‘millium centum viginti,’ no doubt on the assumption that late antique legion approached an effective strength of 5,000. de Jonge 1982: 54 proves the case for the MS reading. 12 On the chronology, which leaves many gaps, see Matthews 1989: 58, whom I generally follow, with some alterations.

13 Amm. XIX.4.1. To be sure, Ammianus was aware of the plague described at Thuc. II.47-54, which he mentions at XIX.4.4. His familiarity with this literary plague, however, need not draw into question the veracity of his report, cf. de Jonge 1982: 65-66.

222

NOEL LENSKI: TWO SIEGES OF AMIDA (AD 359 AND 502-503) their various other stratagems and devices, the Persians had been constructing two siege mounds to overtop the walls; the defenders themselves had also been building counter-mounds (XIX.6.6; 8.1). On the day following his discomfiture with the towers, Shapur ordered his infantry to attack from the Persian mounds. As luck would have it, a Roman counter-mound gave way under its own weight, simultaneously pitching off its Roman defenders and bridging the gap between the Persian mound and the Roman wall (XIX.8.2-4). The Persians poured into Amida across the unoccupied Roman mound and met with disorganized resistance as the Roman defenders struggled against each other to get at their attackers. After a general slaughter, the commander of the troops in Amida, Aelianus, was gibbeted together with the military tribunes; any remaining survivors were taken prisoner and led off into captivity (XIX.8.4; 9.2). Ammianus escaped only by hiding until darkness and then fleeing through an unguarded postern gate (XIX.8.5).

hides on the ramp to prevent the Persian armored infantry from gaining a purchase. Thus the Persians were unable to effect a seizure of the walls and their mound eventually collapsed from under them after six hours of combat.17 Kavadh is said temporarily to have withdrawn in shame and to have been mocked by the defenders for his ineffectualness; Procopius even reports that prostitutes stood on the parapets and exposed themselves to him in scorn (Bell. I.7.16-19; cf. Ps. Zach. VII.3 [HamiltonBrooks 23]; Chron. 1234 148). Despite these setbacks, however, Kavadh resumed construction of his mound in angry determination to take the city (Ps.-Josh. 52-53[2789]). Again, the defenders harried this with an enormous scorpio, termed a “striker” by Pseudo Joshua the Stylite, which tore through the canvas screens the Persians used to protect the mound-builders and killed them.18 Ps. Joshua the Stylite (53[279]) indicates that the Persians became so demoralized that they nearly abandoned the siege, and Ps. Zachariah (VII.4 [Hamilton-Brooks 25]) that Shapur attempted to open negotiations to withdraw on the condition that the defenders pay him an indemnity. This the Amidenes rejected, posing instead their own counterdemands. Ps. Zacharaiah (VII.4 [Hamilton-Brooks 25]; cf. Procop. Bell. I.7.30-1) then indicates that Kavadh had a vision of Christ informing him that he would soon capture Amida because its inhabitants had sinned through their unrestrained insults. Heartened by this, Kavadh began preparations for a final attempt on the city.

The Siege of 502-503 The 502-503 siege was part of a larger campaign undertaken by the Persian shahanshah Kavadh who had broken the longest peace in Romano-Sasanian history in search of a way to generate plunder and force Anastasius to pay him annual tribute.14 Kavadh approached Amida from the north after having invaded Armenia Interior and captured its capital Theodosiopolis. He arrived on the 5th of October (Ps. Josh. 50[276]) and found the city unguarded by a Roman garrison.15 Even so, the abundant citizens of Amida, well prepared for a siege, put up soldierly resistance and thus detained Kavadh for three months and cost him 50,000 men (Ps. Josh. 53[279]). Amida’s defenders proved to be remarkably ingenious in their counter-tactics, using all the stratagems recommended in the ancient and Byzantine manuals. Against Kavadh’s battering rams, they are said by Ps. Zachariah to have lowered bundles of rushes on chains so as to dampen the impact of the blows, and by Procopius to have broken their shafts by dropping heavy timbers across their necks.16 When Kavadh constructed a siege mound, the Amidenes first added courses to their own walls to equal its height. When these were destroyed by rams, they undermined the Persian mound and propped it up on a wooden under-girding. The day Kavadh sent a force of 500 armored infantry to lower a wooden ramp between their siege mound and the city wall and storm the battlements, the Amidenes simultaneously lit the wooden under-girding ablaze and strewed oil-soaked

The Persian kanarang19 had noticed that a daring Amidene named Kutrigo regularly exited the city to harry the Persians and plunder their cattle. The kanarang gave this man considerable latitude to come and go and was thus able to learn where he reentered the city, through the “small watercourses” near the Tripyrgion on Amida’s western side. A guard of monks from the monastery of John Urtaye had been appointed to patrol this sector of the wall, but one dark and stormy night in early January, the kanarang noticed that the monks offered no resistance as he approached the walls. They had apparently fallen asleep; some sources report that they were overcome by excessive eating and drinking, a detail probably added to slander them in the aftermath of their costly lapse of vigilance.20 The kanarang quickly sent for reinforcements, entered the city, ascended a tower, and sent a message to Kavadh reporting his success. When some of the defenders heard the clatter, the Roman governor of Mesopotamia Cyrus came to investigate with a party of men carrying torches; their lights merely drew attention and bowshots from the Persian occupiers, thus

14 Ps. Josh. Styl. 48-50[274-6]; cf. 54[281], with Greatrex 1998: 77-81. Many of the sources are conveniently collected, translated, and commented upon at Greatrex and Lieu 2002: 63-69. On the siege see Greatrex 1998: 83-94 and the notes in Trombley and Watt 2000: 53-63. 15 At least according to Procop. Bell. I.7.4; Theophan. a.m. 5996, though see Trombley and Watt 2000: 54 n.260. 16 Ps. Zach. VII.3 (Hamilton-Brooks 23); Procop. Bell. I.7.12. Cf. Amm. XX.11.15, 19, where the Persians used similar tactics in defense of Bezabde. On the recommendations of the manual writers, see Aen. Tact. XXXII.4; Veget. DRM IV.23.1-5; Ps. Maur. Strat. X.3.2-3; Peri Strategias 13.

17 Ps. Zach. Styl. VII.3 (Hamilton-Brooks 23); Ps. Josh. Styl. 50[276-7], 53[279]; cf. Procop. Bell. I.7.14-15. 18 Ps. Josh. Styl. 53[279]. Marsden 1969: 81 holds that the largest shots a scorpio/onager could hurl weighed 35 to 40 kg. 19 On this title see Christensen 1944: 107 n.3. On its application here see Greatrex 1998: 90 n.50. 20 Ps. Zach. VII.4 (Hamilton-Brooks 25-26); Chron. 1234 148-49; cf. Ps. Josh. Styl. 53[280]; Procop. I.7.20-24; Theophan. a.m. 5996. Greatrex 1998: 91 calculates that the night of the assault, January 10, 503, would have been a Friday, not apparently a feast day.

223

THE LATE ROMAN ARMY IN THE EAST Table 1: Numbers of Besiegers and Defenders * City & Date

Number of Besiegers

Amida 359

100,000

Number of Defenders (soldiers / civilians) 7-10,000 / 10-13,000

Singara 360 Bezabde 360 Peroz Shapur 363

? ? 65,000

2,500-3,500 / ? 3,500-5,000 / ? 2,500-5,000 total

Amida 502-503 Amida 504-5

? 40,000

0 / 80,000? 2,500-3,000 / ?

Antioch 540 Sisauranon 541 Rusafa-Sergiopolis 542

? ? 6,000

6,000/? 800 / ? 200 / ?

Petra-Justinianopolis 549

8,000

1,500 / 0

Petra-Justinianopolis 550

6,000

2,300 / 0

Archaeopolis 550

c. 20,000

3,000 / 0

Phasis 556 Dara 573

60,000 63,000

? ?

Source(s) Amm. XVIII.9.3-4, XIX.2.14, 6.11 Amm. XX.6.8 Amm. XX.7.1 Zos. III.13.1, 18.4; Amm. XXIV.2.22; cf. Lenski 2002: 308 See above n. 23 Ps. Josh. Styl. 53-54 [280-81]. Procop. Bell. II.8.2 Procop. Bell. II.19.3 Procop. Bell. II.20.12, 14 Procop. Bell. II.29.10, 34 Procop. Bell. VIII.11.42 Procop. Bell. VIII.13.8, 14.44, cf. II.30.30-1 Agath. III.17.4 Chron. 1234 66**

*Unit sizes are calculated at roughly 1,000 men per infantry formation and 500 men per cavalry formation, see Coello 1996. Schmidt 2001: 96-98 argues for infantry units no larger than 750, and Tomlin 2000: 169-73 for infantry units of 500 and cavalry units of less than 200. ** This number seems credible, but the additional 120,000 farmers supposedly recruited as manpower seems impossibly large, for logistical reasons alone, as does the number 98,000 for the prisoners taken after Dara’s fall.

immobilizing this band of defenders. After the sun rose, Kavadh ordered his infantry to scale the walls on ladders while the kanarang and his men distracted the Amidenes from their tower. When some scalers met resistance and retreated, Kavadh threatened them with execution at the point of his own drawn sword.21 Thus goaded, the Persians began overtaking tower after tower along the rampart. Though one gargantuan man named Peter of ‘Amkhoro was able to block the Persians’ progress in one direction for a time, he gave up when he saw five or six towers taken in the other and fled for his life. The Persians gained the whole circuit and held it for 24 hours before descending and opening the gates to their countrymen after more than three months of siege, January 12, 503.22 Three days worth of general slaughter followed in which as many as 80,000 Amidenes were killed.23

Operations Now that we have examined these two sieges in some detail, it is time to step back and consider further the battle experience they generated. Our observations will be divided into three groups: operations, human limitations, and psychology.24 They will be made from both the perspective of the defenders and the besiegers, even if in both instances we have testimony only from the defenders. To fill out the picture from both sides, recourse will be had to other sources for late Roman sieges, with the necessary result that something of the composite picture described in the opening of this paper will emerge. Many of the operational details of these two sieges have already been outlined. Furthermore, these two instances followed patterns familiar from any number of ancient sieges that need not be explained in detail (see e.g. Marsden 1969; Kern 1999; Campbell and Hook 2005). It is nevertheless worth calling attention to certain striking

21

Ps. Zach. VII.4 (Hamilton-Brooks 27); Procop. Bell. I.7.27-9; cf. II.27.38. 22 Reckoned inclusively, Ps. Josh. Styl. records a siege of 99 days, from Oct. 5, 502 to Jan. 11, 503 (50[276], 53[280]). This accords well with the 97 days recorded at Elias Nis. a. 814 and Chron. Edess. 80 (a. 814). Chron. 724 a. 814, places the date of Amida’s fall on Dec. 24, 502, which accords better with the notice at Procop. Bell. I.7.29 that the war lasted 80 days. Ps. Zach. VII.3 (Hamilton-Brooks 22) estimates ‘more than three months’; cf. Theophan. a.m. 5996. Joshua’s dates are to be preferred. See also Greatrex 1998: 92 n.57. 23 This number, reported at Ps. Zach. VII.4 (Hamilton-Brooks 28); Ps. Josh. Styl. 53[280]; Elias Nis. a. 814, would have included both the residents of Amida and those from the countryside who had fled to its

walls for protection, cf. Trombley and Watt 2000: 93 n. 444. Even so, it seems extremely high. On the slaughter see Joh. Eph. Lives 58 (PO XIX.219-221). 24 One extremely important element in late Roman sieges which I leave aside is the role of religion, on which see Whitby 1998, which focuses heavily on sieges; cf. Lightfoot 1988: 122-24 on the divine apparitions during the third siege of Nisibis in 350.

224

NOEL LENSKI: TWO SIEGES OF AMIDA (AD 359 AND 502-503) features of operations during these two test cases as they relate to the soldier’s experience.

to the defenses. This reality, combined with the defenses afforded by their fortifications, led even fairly small defensive forces to hold their own against much larger armies of besiegers.30 Such calculations no doubt impacted commanders, soldiers, and citizens alike.

The first thing to note is that the number of soldiers besieging generally far exceeded the number of soldiers defending: in the 359 siege Shapur’s forces outnumbered the Roman soldiery ten to one, and the 502-503 siege there was no officially appointed garrison (see table 1).25

At the operational level, both the defenders and besiegers will also have been concerned to know the length of time necessary to complete a siege. Here the data is scanty31 and at times contradictory, but by and large indicates that successful sieges were either brought to a close within about one week or lasted something on the order of three months. In the first category we could number Julian’s siege of Peroz Shapur in 363 (Amm. XXIV.9.1-19). In the second, we must place the two Amida sieges under discussion, lasting respectively 73 and 99 days (see above nn. 12 and 22), the siege of Nisibis in 350 — which was not successful — lasting three to four months32, and that of Dara in 573, lasting rather longer at six months (Evag. V.10; Chron. 1234 66; Joh. Epiph. 5 [FHG IV.275]. Of itself Shapur’s termination of the Nisibis siege in 350 despite nearly achieving success is an indicator that a besieging army was only prepared to continue a siege for one third to one half of a year. This time limit was at least partly a result of the length of the campaign season in upper Mesopotamia and the sub-Caucasus. By the time an army reached a large fortress and placed it under siege, only a few months remained before the harsh winters of central Kurdistan and the sub-Caucasus will have put a halt to operations. By the onset of extreme cold temperatures, it was extremely difficult to carry on a siege, as Constantius II found at Bezabde in 360 (Amm. XX.11.24-5, 31-2. Cf. Amm. XIX.9.1). The capture of Amida in mid-winter 503 and of Dara in December 57333 required exceptional effort, something Ps. Zachariah attests when he reports that the Persians were hard put to continue their efforts at Amida in winter 502-503 because of the cold. For this reason, the classic distinction between sieges by assault and those by starvation, made for example by Vegetius,34 generally did not apply to sieges in the late Roman near east. To be sure, blockades and tampering with the water supply were regular features, but we shall see below that inadequate provisioning or water supply only rarely played a decisive role. When besiegers arrived, they arrived to fight and not simply to surround the enemy and drive him into submission with hunger and thirst. The logistical problem of provisioning a fortress for a siege period of

An average fortress or fortress city had a standing garrison of between 1,000 and 3,000 men, whereas expeditionary armies could number as many as 100,000 (Nicasie 1998: 202-7; Whitby 2000: 292-3). Indeed, we have already noted that in 359 Amida’s normal garrison of one legion, the quinta Parthica, and one cavalry unit (indigenarum turma), was swollen with the addition of six further legions and a cavalry squadron, making its defensive force exceptionally large.26 The odds were thus stacked in favor of the besiegers, though the besieged had other strengths to rely on, not least their walls, to even out the equation. For this reason, casualties were generally much higher on the side of the besiegers, as at Amida 359, where 30,000 Persian soldiers died in order to exterminate one third to one quarter as many Roman troops.27 Moreover, even when military garrisons were small, the local citizenry in these cities and forts was usually quite accustomed to participating in their own defense. Thus Amida was defended entirely by its own citizenry in 502-503 with considerable success. When Antioch was besieged in 540, the youth of the circus factions fought more bravely than the tiny imperial garrison sent by Justinian, which eventually fled.28 At Edessa in 544, rustics and citizens joined the garrison troops in fending off a Persian attack on the Great Gate, and, during Khusro’s final desperate assault, even women and children became involved in throwing projectiles and hot oil to hold off the Persians.29 Siege operations were thus Clausewitzian “total war” on a micro-scale, at least for those in a besieged city. The garrison, small though it normally was, was only one of the elements contributing 25 This was true whether we calculate the defenders based on the entire population under siege or on the smaller numbers represented by the military personnel only. Amida was of course a fortress city with a sizeable civilian population, but there are a number of instances on the eastern frontier where the number of military personnel was essentially coterminous with the total population of the fort, particularly in the subCaucasus region. 26 Amm. XVIII.9.3-4. At the end of the fourth century, Amida was garrisoned by the Equites scutarii Illyriciani and the Equites ducatores Illyriciani, ND Or. XXXVI.19, 21. 27 Veget. DRM IV.12.1 was well aware of this problem: violenta autem impugnatio quando castellis vel civitatibus praeparatur, mutuo utrimque periculo sed maiore oppugnantium exercentur luctuosa certamina. 28 Procop. Bell. II.8.11, 17-29. Ps. Maur. Strat. X.3.9 insists that citizens who remain within the fortification during a siege must participate in the fighting. 29 Procop. Bell. II.27.22-23, 33-36; cf. VII.38.16; Tac. Hist. V.13.3; Jos. BJ III.303, V.336-38, with Price 1992: 119-20. For the citizenry of a city bombarding soldiers from rooftops, see Amm. XXIV.2.13; Veget. DRM IV.25.2-6; cf. Barry 1996. More on the effectiveness of the local inhabitants of fortress cities on the eastern frontier as citizen militias at Isaac 1992: 252-55.

30 Cf. Procop. Bell. II.30.15-17 where Mihr Mihroes mocks the Roman besiegers of Petra-Justinianopolis for failing to take the city even when there were only 150 effective defenders left standing. 31 Ammianus’s chronology sometimes drops off before the end of a siege, as with the two Bezabde sieges of 360 reported at XX.7 and XX.11. 32 Chron. Pasch. p. 536 reports 100 days; Jul. Or. I.28d; II.62d four months; Theod. HE II.30.4 seventy days. 33 Theophylact III.11.2; Joh. Eph. HE VI.5. More on Dara at Whitby 1988: 258; Greatrex and Lieu 2002: 147-8. For the effects of weather see Shaw 1999: 139-40, 144. 34 IV.7.1-2, though the reading here is usually heavily emended. In his recent edition Reeve 2004: 127 omits the distinction.

225

THE LATE ROMAN ARMY IN THE EAST about three months will have factored into the tactical decision to keep garrisons small.

precisely because they could be effective at demoralizing and even crippling the besieging force and disabling its engines. A series of desperate sorties by the Persians from Bezabde in 360 destroyed the Romans’ siege engines and in turn their mounds, supported as they had been by timbers (Amm. XX.11.16-17, 22-3, cf. 13); emperor Julian was nearly killed in a sortie by the Persian defenders of Maiozamalcha in 363 (Amm. XXIV.4.4-5; cf. Zos. III.20.2-3); the sally by the Armenian defenders of Artogerassa in 368 essentially wiped out the opposing Persian siege force (Amm. XXVII.12.7-9, cf. Lenski forthcoming); two sorties by the defenders of Edessa in 503 essentially put a halt to Kavadh’s hopes for a siege there (Ps. Josh. Styl. 60-2[288-90]); Procopius speculates that a well-timed sortie by the defenders of Antioch in 540 could have saved the city by securing the rocky heights of Mt. Silpion (Bell. II.8.13-14); and he reports how a well timed sortie at Archaeopolis in 550 drove off the Persian army just as it was about to capture the city (Bell. VIII.14.29-44). Given their effectiveness, it is fair to speculate that sorties were probably a more common feature of siege warfare even than our sources usually indicate. Ammianus, who describes the one nighttime sortie by the Gallic units, indicates that this was only the most spectacular of their many sorties (XIX.5.2), and the siege narratives of Iotapata and Jerusalem in Josephus – even more detailed than that of Ammianus – make it clear that sorties could be an almost daily occurrence.37

The normal operational methods for conducting ancient sieges are of course well known: undermining, moundbuilding, battering, attacks from towers and artillery. Less well acknowledged is the amount of hand to hand fighting that went on outside a fort’s walls. Indeed, sorties by the besieged represented a common feature of all siege operations. They feature in both Amida sieges, with the night raid by the Gallic legions of Magnentiaci and Decentiaci in 359 and the raiding sorties of Kutrigo from Amida in 502-503. Ammianus’s note on the first is revealing for what it says about the distinction between western and eastern warfare, the latter of which was much more heavily dependent on siege: There were with us two Magnentian legions, recently brought from Gaul (as I have said) and composed of brave, active men, experienced in battle in the open field, but to the sort of warfare to which we were constrained, they were not merely unsuited, but actually a great hindrance; for whereas they were of no help with the artillery or in the construction of fortifications, they would sometimes make reckless sallies and after fighting with the greatest confidence, return with diminished numbers, accomplishing just as much as would the pouring of a single handful of water (as the saying is) upon a general conflagration.35

Equally dangerous was the mirror equivalent for besiegers, storming operations, which could also bring soldiers into hand-to-hand combat and certainly opened them directly to missile fire. At Amida in 503, Kavadh personally threatened execution for those who retreated down the Persian scaling ladders, a testament to the fear storming could arouse (see above n. 21). Here too the job was extremely dangerous and thus unattractive to all but the most war-hardened and courageous. One can only imagine, for example, what it would have been like for the 500 Persian storm troops who attempted to cross the ramps into Amida in 502 but came sliding instead onto the Amidene rampart in a helpless – and soon fatal – clatter of armor. So too, the 70 sagitarii ex agmine regio who secretly took the tower of Amida in 359 only to be exterminated by ballistae showed extraordinary courage in the face of almost certain death. These were risks most were unwilling to take. When the Romans attempted to retake Amida in 504, they successfully excavated a tunnel into the city, but once the first man out of the hole was killed, and the second, a dauntless Goth named Ald, met with fierce resistance, the remaining storm troopers retreated in fear and were eventually drowned after the Persians flooded the tunnel with water.38 Similarly, at the Roman siege of Petra-Justinianopolis in 549, the Armenian John had the courage to storm a breach in the

These Gallic units were indeed fish out of water, so much so that, despite the success of their nighttime sally in 359, they were unable to arouse any of the other units in Amida to join them and eventually had to retreat after suffering serious losses. Yet such operations were by no means as insignificant as Ammianus implies. For this reason, the Gauls’ valor won for their campidoctores statues erected in a public square of Edessa that Ammianus himself had seen and remembered (XIX.6.12). So too Ammianus recalls that the Comes Aelianus, commander of the refugee legions in Amida, had won his fame by conducting a nocturnal sortie at Singara in 348 while serving as a protector (XVIII.9.3 with Dodgeon and Lieu 1991: 193). Sorties were thus a valued weapon in the arsenal of tactical operations. The fact is, of couse, that sorties were also extremely high risk operations and rarely attracted volunteers except when they had more definite goals than those outlined by the Gallic legions in 359.36 Even so, they were desirable 35 Amm. XIX.5.2 (trans. Rolfe, with modifications). Further examples of sorties at Amm. XX.11.16-23; XXI.12.13; XXIV.4.2-4, 5.8; Jul. Or. II.65a; Procop. Bell. I.15.13; II.17.3-10, 26.25-8, 27.43-4; VI.27.1-22; VII.37.21; Agath. III.26.3, 7; III.27.6; Veget. DRM IV.18.1; Ps. Maur. Strat. X.1.1; Miracula S. Demetrii 147, and below. 36 Ps. Maur. Strat. X.3.10-11 strongly discourages sorties. Frontin. Strat. III.10 describes stratagems for baiting an enemy to sally forth in order to entrap him. See for example Procop. Bell. VII.28.11-14; Agath. III.22.3-8.

37

Jos. BJ III.152, 155, V.75, 115-18, 266, 277-90, 304-8, 358-59, 47389, VI.14-22, 157-60, 392-93; cf. Price 1992: 122-23. 38 Ps. Josh. Styl. 71[295-6] with PLRE II Ald. Cf. Procop. Bell. II.13.267, 26.26. On flooding out sappers, see Veget. DRM IV.5.2; Peri Strategias 13.

226

NOEL LENSKI: TWO SIEGES OF AMIDA (AD 359 AND 502-503) wall with 50 Romans, but when no others had the courage to follow, all were forced to retreat. The same John displayed similar courage at Petra two years later and was, unsurprisingly, killed in battle.39 When successful, such feats of derring-do could prove decisive to an engagement; even when they failed, they often served to set an example for others or at least to fire the courage of those who hung back. The most dangerous and difficult operational maneuvers were also the most useful. They were thus encouraged and rewarded within the community of soldiers and defenders. Such inducements to performance will be discussed further in the section on psychology.

The Persian besiegers of Bezabde in the first half of 360 were vulnerable to the clouds of Roman missiles in part because they advanced blind (caeci) behind wicker hurdles (Amm. XX.7.6). So too, the Persian defenders of Bezabde in late 360 were prevented from reconnoitering the Roman siege operations for fear of attracting fire (Amm. XX.11.20-22). To be sure, missiles were not simply an impediment to clear lines of sight, they were also quite deadly, or, more accurately, they were an impediment to visibility precisely because they could be deadly. Everett Wheeler (2001) has recently argued convincingly that missile fire in antiquity had the dual effect of first degrading an enemy’s performance by limiting visibility, mobility and action, and, secondarily, inflicting casualties – albeit not at a particularly high rate per shot fired. These bipartite functions are nowhere clearer than in an incident from the siege of Cremna in 278 during which the Isaurian captain Lydius was forced to reconnoiter the Roman operations through a little known window in the city’s defenses in order to escape notice and avoid missile fire; when Lydius’s top marksman deserted and revealed this to the Romans, he was asked to exploit Lydius’s patterned behavior in order to afford himself a clear shot of his former leader and thereby kill him with a ballista bolt (Zos. I.70.1-4). The incident encapsulates well the fact that a clear view of the enemy was gained only at tremendous risk to one’s personal safety. Occasionally the blind spots created by cilicia and other coverings could be used to the advantage of those who strung them up in defense: during the Roman siege of Bezabde in late 360, the Persian defenders hid behind the cover of the cilicia along their rampart and then popped out like puppets from behind a curtain to assault the Romans as they scaled the wall (Amm. XX.11.9). More often, however, these coverings and even the very fortifications themselves blocked sightlines with serious impact on the outcome of combat: during the siege of Antioch in 540, the defenders thought their upper wall had been breached after they heard the makeshift ramps they had strung between its towers collapse; fearing the worst, the defenders fled their posts and thus enabled the Persians to overtop their still unbreached walls (Procop. Bell. II.8.16). Not unlike the trench fighters of World War I, defenders and besiegers were thus forever hampered in their ability to see effectively in battle; though the weapons they faced were not nearly as accurate or longrange, their technologies for viewing the battle under cover – in a world before the periscope, telescope, and binoculars – were also much more limited.

Human Limitations The limitations of warfare’s human participants also seriously affected the practice of combat during a siege. Take for example the issue of visibility and lines of sight. Eastern warfare – both field engagement and siege – was particularly given to the use of missiles, especially bowshots. Ammianus reports how, in the first onslaught of 359, “a thick cloud of arrows in compact mass darkened the air”, phraeseology he and other authors often repeat.40 Such missile volleys seriously compromised lines of sight, above all by forcing soldiers to hem themselves round with armor and form up walls of shields; during a siege, defenders could also hide behind battlements and besiegers behind siege machinery or counterwalls in defense. Regardless, visibility was limited for besiegers and defenders alike. In an effort to protect against missiles, both besiegers and besieged also strung up curtains of iron, wool, linen, wicker, rawhide, or goathair (cilicia) around their battlements and siege machines in hopes of deflecting projectiles or dampening their impact and preventing incendiary devices from setting fires.41 These various processes of “taking cover” impeded action on the operational and tactical levels, limiting both the soldier’s ability to fight, work, and move, and his ability to reconnoiter targets and plan or coordinate actions. At times the very coverings they used to protect themselves became the cause of failure. Thus the water-soaked canvas curtains used by the Persians against bowshots and fire-darts in 502-503 provided the very screen which prevented their soldiers from seeing the crushing rocks hurled on them by the Amidenes’ “striker” (Ps. Joshua Styl. 53 [279] Cf. Amm. XX.9.11). 39

Procop. Bell. II.30.1-6, VIII.11.57-58, 64 with PLRE III Ioannes qui et Guzes 44. Compare Jos. BJ VI.54-67 on the small party willing to assault the makeshift wall the Jews had constructed behind the Antonia. 40 Amm. XIX.2.8: sagittarum creberrima nube auras spissa multitudine obumbrante; cf. XX.7.6: sagittarum enim nimbi crebrius volitantes; XXIV.2.11 utrubique saxorum manualium nimbis; XXIV.2.17: obumbrata caeli facie fragmentis montium et missilibus aegre repulsus; XXV.3.11: contra animosius Persae sagittarum volantium crebritate conspectum sui rapiebant oppositis; cf. Agath. III.25.1; Theophylact I.12.5; Miracula S. Demetrii 139. 41 E.g. Amm. XIX.7.5; XX.7.13, 11.9; XXIV.2.10, 4.15; Priscus fr. 6.2 (Blockley); Procop. Bell. II.26.29-30; Ps. Josh. Styl. 59[286]; Miracula S. Demetrii 152; Jos. BJ III.172-73, 241-42; Veget. DRM IV.6.2-3, 14.1, 15.1-6; Philo Byz. Polior. 95.34; Ps. Maur. Strat. X.3.2; Peri Strategias 13.

Darkness also played a significant role in ancient sieges – as in all ancient combat situations. This is well evidenced by the occupation of the tower at Amida in 359 by seventy Persian royal archers, who snuck in under cover of darkness (Amm. XIX.5.5). So too, the Persian kanarang effected his midnight penetration of Amida in January 503 on an overcast night with limited visibility (Ps. Zach. VII.4 [Hamilton-Brooks 26]; Ps. Josh. Styl. 227

THE LATE ROMAN ARMY IN THE EAST 53[280]). By the same token, the Roman governor Cyrus was all the more foolish to have used torches to investigate the kanarang’s penetration, for these served only to make a target of his band of defenders in the darkness. In 359, the Gallic legions so anxious to sortie against the Persian outposts were forced to wait for a stormy and moonless night before being granted permission to move (Amm. XIX.6.7: observata nocte squalida et interluni). Julian also used the cover of darkness to move his siege engines against Peroz Shapur during his 363 invasion.42 Even during daylight hours, smoke and dust played a role in clouding visibility, for ancient siege operations kicked up tremendous dust clouds and regularly employed smoke producing incendiary devices to destroy wooden breastworks, mobile towers, ramps, and siege machinery.43

(Procop. Bell. II.20.10-14) or twice from plans to besiege Dara (Procop. Aed. II.2.19-21). Food also seems to have held out at Amida in both 359 and 502-503, though just a year later, the Persian garrison left there by Kavadh nearly ran out of supplies and had to grant permission to the “comfort-women” whom they had retained in the city to eat the corpses of the dead.47 The fact is that food was generally well-supplied for all but the lengthiest sieges, for Romans and Persians alike were well aware that adequate provisioning was perhaps the single most important fallback of the defenders.48 This was also taken into account when a fort was garrisoned. The anonymous sixth-century treatise Peri Strategias (9) cautions specifically against stationing too many men inside a fort so as to avoid provisioning troubles. With this in mind, strategists took advantage of the operational principle that it took fewer men to defend a fort than to besiege one in order to keep their garrisons small and thus easier to feed. Where a fort could not be adequately provisioned, it was simply abandoned (Procop. Bell.VIII.13.15-19). The only variable in this equation was the number of civilians who would flee to neighboring fortifications for safe haven in anticipation of an invasion. Amida faced this problem in 359, yet even here supplies did not run short during its 73 day siege (See above n. 11; cf. Miracula S. Demetrii 197). Perhaps because of this heightened preparedness we do not generally hear of fighting-age males turning out large numbers of feebler civilians in late Roman sieges on the eastern frontier, even despite advice along these lines in Vegetius.49 As already discussed, the civilian populations of these frontier cities more than compensated for the drain they imposed on provisions by supplying fighting power, and, in the case of women, countless services that go unmentioned or are barely alluded to in the sources, from cooking, to caring for the wounded, to manufacturing missiles, to providing psychological and sexual comfort.

Limitations of sight were minor in their effects on combat performance compared to the gnawing grind of thirst, hunger, injury, and fatigue. Nevertheless, we have already mentioned the fact that the first two – logistical – problems, were not paramount in most eastern sieges of the late empire. Thirst, for example, played little role in either Amida siege, for, as Ammianus reports, that city not only overlooked the Tigris River but also had a spring at its center.44 Indeed, the technical writers Vegetius (DRM IV.10.1-5), Ps. Maurice (X.2.8-3.1, 3.15-16, 4.911, cf. X.1.2) and the anonymous author of the Peri Strategias (9-10) make it clear that fortresses should always be constructed with attention to a water source. In a number of instances – Dara,45 Petra-Justinianopolis (Procop. Bell. VIII.12.20-7), Archaeopolis (Procop. Bell. VIII.14.3) – we learn that this advice was heeded in a fort’s collocation and construction. In his narrative of the 359 siege, Ammianus (XIX.5.4) reports that not only did Amida have a secret, rock-hewn passage between its walls and the river, but he had observed a similar passageway in all forts of the region that bordered rivers. Even so, water supply problems could arise, as when the citizens of Beroea chose to place their livestock between their outer and inner walls during the siege of 540 and thus witnessed their single spring drunk dry.46 At certain forts water supply difficulties could also plague the besiegers, as when the Persians were forced to withdraw their siege of Rusafa-Sergiopolis in 542 for lack of water

Injury and illness also posed a serious threat to the battle readiness of soldiers on both sides. From 359, Ammianus poignantly describes how, in the face of the first all out assault on the walls of Amida, “those mangled overturned in their writhing those who stood next to them, or at any rate, so long as they remained alive kept calling for those

42 Amm. XXIV.2.11-12. Compare the role played by moonlight in permitting a late night sally against Julian’s forces at Amm. XXIV.5.8. Darkness also allowed for the investment of the fortress opposite Phathousa in 363, Zos. III.13.2-3; cf. Frontin. Strat. III.9.2. 43 Amm. XX.11.19: et quia conspectum abstulerant fumi nigerrimae nubes. Cf. Procop. Bell. II.27.7-17 where the Romans used incendiary missiles to mask the smoke rising from the mine they had dug under the Persian mound at Edessa as its under-girdings were burning. On the problem of dust clouds in Roman combat see Lee 1996: 201-2 44 Amm. XVIII.9.2. Gabriel 1940: 90-2 reports several springs. 45 Procop. Bell. VIII.7.7-9; cf. Theophylact III.11.2; Joh. Eph. HE VI.5. See also Croke and Crowe 1983: 149-50, 156-59, with a detailed analysis of the notices on Dara’s water supply in Procop. Aed. II. 46 Procop. Bell. II.7.12. At times a city’s or fort’s water supply was deliberately attacked or polluted, Procop. Bell. VI.3.2-7, 27.1-22; VIII.12.21-27; Chron. 1234 66; Joh. Eph. HE VI.5; cf. Frontin. Strat. III.7.

47 Ps. Josh. Styl. 76-77[299-300]; cf. Procop. Bell. I.9.1-3, but see I.8.7. Most, though not all, of the documented incidents of starvation in late antique sieges are from the west: Procop. Bell. I.21.8 (Martyropolis); II.19.19 (Sisauranon); VI.3.8-12 (Rome); VI.27.25-9 (Fisula); VI.27.2834 (Auximum); VII.39.5; VIII.23.2 (Ancona); VIII.25.24-25 (Croton); Theophylact VII.13.1-4 (Tomi); Men. Prot. fr. 25.2, 27.2-3 (Blockley) and Joh. Eph. HE VI.32 (Sirmium). 48 For explicit evidence of adequate provisioning, see Amm. XX.7.16, 11.24; XXIV.2.22; Procop. Bell. I.19.20; II.29.12, 30.30-31; VIII.7.7, 12.17-20; Agath. II.13.5; Men. Prot. fr. 23.9. On measures to be taken to ensure adequate provisioning see also Veget. DRM III.3.5-6; IV.7.4-10. More on logistics during siege warfare at Roth 1999: 314-19. 49 Veget. DRM IV.7.10. For examples of this phenomenon in earlier periods, see Xen. Hell. VI.2.15 ; Caesar BG VII.78.3-5; Frontin. Strat. III.4.5; Zos. I.69.2-4, and the detailed analysis of the famine during the siege of Jerusalem at Price 1992: 144-59, 271-80.

228

NOEL LENSKI: TWO SIEGES OF AMIDA (AD 359 AND 502-503) who had the skill to pull out the arrows implanted in their bodies” (XIX.2.9, trans. Rolfe, modified). Shortly after this passage he describes rudimentary MASH units helping the wounded, including triage by paramedics unwilling to treat the severely injured lest they prolong their suffering and the agonizing death by necrosis suffered by many who had already faced the painful extraction of arrows.50 Most gruesome of all is his description of the dismemberment of an architectus ripped apart by a scorpio that misfired at the siege of Maiozamalcha in 363 (XXIV.4.28). The operation of these machines must have resulted in regular injuries and death. The presence of so many comrades wounded in combat will have put an additional strain on the time, resources, and psyches of both sides, and particularly the besieged.51 Infectious disease was also a constant threat, particularly for the besieged. The 359 plague was, by Ammianus’s account, caused by the unsanitary presence of too many rotting corpses inside Amida’s walls (XIX.4.1, 8; cf. XIX.2.14). Even so, defenders had little choice but to keep the bodies in their walls and little time to bury them. Quite apart from the ritual consequences of exposing one’s dead to the elements—not so much a concern for Zoroastrians—jettisoning bodies from the walls would have given the besiegers a clear estimate of the depletion of a city’s defenders. Thus the 500 remaining Persian defenders of Petra-Justinianopolis in 549 were compelled to keep 1,000 decomposing cadavers within their walls so as not to reveal their casualties (Procop. Bell. II.30.16).

combat at the siege of Maiozamalcha in 363, both Persians and Romans became so fatigued by the midday Iraqi sun that they broke off the engagement for the remainder of the day (Amm. XXIV.4.17). Little wonder, then, that the monks garrisoning the Tripyrgion of Amida in early 503 slept through their guard duty (Ps. Zach. VII.4 [Hamilton-Brooks 26]; Ps. Josh. 53[280]). So too, the Gallic attackers on the Persian outposts in 359 took full advantage of the Persians’ need for sleep to wreak havoc on their camp. At the siege of Edessa in 544, the Persians nearly scaled the walls unnoticed while a guard unit caught some sleep in the early morning hours (Procop. Bell. II.27.18-21; cf. Jos. BJ III.318-19). Well aware of the problem of fatigue, the sixth-century tactical writer Ps. Maurice recommended that the besieging force never deploy its full complement every day but rather stagger the men into shifts, thus allowing them time to recover. By the same token, he argued, the besiegers should take advantage of their superior numbers by never relaxing the constant pressure of their assault, by day or by night.54 The daily grind element of siege warfare must have rendered it much more taxing, in many ways, than field battle. For both sides, vigilance was ever the watchword. The rest and relaxation both sides so desperately needed was in many ways their worst enemy. One final element of the soldier’s experience that would have directly affected combat operations was the element of labor involved in carrying out and defending against a siege. Most sieges that did not succeed in taking a city by storm in the first few days moved into a second phase during which preparation and construction were necessary before more fighting could resume. Ammianus describes just such a transition at Amida in 359:

Siege warfare was hard work and the fatigue it caused was often compounded by lack of sleep. In several passages Ammianus mentions his weariness following engagements and his relief when truces allowed him to catch up on rest.52 To judge by Ammianus’s report, these truces often followed particularly grueling days of fighting.53 All too often, however, sleep was interrupted by the exigencies of warfare. During the first day of open

Then the enemy, horrified and maddened by the unexpected mishap, set aside all delay and, since force was having little effect, now planned to decide the contest by siege works (operibus).55 His account is larded with the vocabulary of work and labor: constant toil and sleeplessness sapped the men’s strength (XIX.2.14 continuus cum insomnia labor); to defend against the Persian capture of Amida’s tower, work was divided among the defenders (XIX.5.6 dividitur opera); prior to the nighttime Gallic sortie, Amida’s walls were defended with toil and watchfulness (XIX.6.6 laboribus et vigiliis); so too, the Roman counter-mound was constructed through lengthy labor (XIX.8.2 diu laborata moles).56 To spur on the speed and efficiency of

50

Amm. XIX.2.15: medebatur ergo suis quisque vulneribus pro possibilitate vel curantium copia, cum quidam graviter saucii cruore exhausto spiritus reluctantes efflarent, alii confossi mucronibus, †prostrati curam, animis in ventum solutis proiciebantur extincti, aliquorum foratis undique membris mederi periti vetabant ne offensionibus cassis animae vexarentur adflictae, non nulli vellendis sagittis in ancipiti curatione graviora morte supplicia perferebant. 51 See for example the Persian defenders of Petra in 549, where only 500 of an original 1,500 men survived the siege and only 150 of these were fit to fight, Procop. Bell. II.30.15. 52 XIX.2.14: tandemque nox finem caedibus fecit et satias aerumnarum indutias partibus dederat longiores. ubi enim quiescendi nobis tempus est datum, exiguas quae supererant vires continuus cum insomnia labor absumpsit; XIX.6.13: indutiisque ob haec tridui datis adsensu communi nos quoque spatium ad respirandum accepimus; XIX.8.1: somno per breve otium capto; XIX.8.6: sed fatigato mihi lassitudine gravi; XXIV.4.20: diei finisset occasus, tum fatigationi consulitur; cf. XX.11.9, 13, 20; XXI.12.12. 53 Goldsworthy 1999: 203 describes similar lulls during the siege of Jerusalem and emphasizes the need to ‘restore the soldier’s aggressive spirit.’ More to the point, such rest periods were necessary for simple physical recovery. By contrast, it should be remembered that the average infantry engagement lasted no more than a few hours, see Sabin 2000: 4-5, 10-11; cf. Goldsworthy 1996: 201-6.

54 Ps. Maur. Strat. X.1.9-10. Veget. DRM IV.27.2-28.2 also describes sneak attacks at meal or rest time. See also Frontin. Strat. III.12 on stimulating vigilance. 55 Amm. XIX.7.1 (trans. Rolfe with modifications); cf. XIX.5.1. See also Lendon 2005: 242-45, 250-52 on the physical labor of soldiers. 56 Cf. Amm. XX.11.7: periculisque et laboribus indurati cuncta obsidioni congrua parabantur; XXI.12.7: indefesso labore; XXI.12.17: cum in reliquis opera consumeretur in cassum; XXIV.4.13: cumque apparatu omni excindendae urbis labore multiplici consumato; XXIV.4.20: obsidentium labor, obsessorumque industria vicissim

229

THE LATE ROMAN ARMY IN THE EAST labor, tasks were regularly divided up among units or regiments who competed against each other in completing their individual tasks or their portions of a larger task.57 Failure to do so offered the prospects of a deadly outcome for those who had not exerted themselves with sufficient effort or speed. When Khusro approached Callinicum in 542, its citizens had simply to abandon their city because they had failed to complete the task of rebuilding its decrepit walls in time.58 By contrast, the successful deployment of labor could rescue a city from the brink, as when the citizens of Nisibis in 350 rebuilt their collapsed wall to four cubits high under a fierce Persian barrage during the course of a single twenty four hour period.59

(forthcoming; cf. Rostovzeff et al. 1936: 188-205) has demonstrated that, when the Romans intersected the Sasanian mine under their tower with their own countermine, a furious battle immediately ensued in which approximately nineteen Romans were killed and their bodies eventually stacked at the entrance to their own countermine to block access so that the Persians could gain time to fire the wooden supports under the tower without encountering further skirmishers. At the moment the two groups of sappers broke through each other’s mines, they will have had to transform their activity instantly from sapping to soldiering, from excavation to execution. This same tug of war between labor and combat can be seen at Petra Justinianopolis where the Persian defenders who were creating a series of naphtha flame bombs became so absorbed in their work that they failed to recognize that one bomb had set their own tower alight (Procop. Bell. VIII.11.59-62). Labor and combat were thus uncomfortable bedfellows that had to be managed with skill and considerable caution.

Siege warfare was thus in some sense a paradoxical tug of war between creation and destruction: the creation of tunnels and towers to destroy a fort’s walls; the creation of countermines and extension of walls and towers to destroy the besiegers’ tunnels and towers; the creation of mounds and counter-mounds to overwhelm the enemy with missiles or storm their walls; the construction, maintenance and repair of artillery and the manufacture of missiles which were then destroyed in their use. Vegetius recommends that the inhabitants of a fortress collect supplies to manufacture weapons and build defenses in anticipation of siege, confirmation that any city under siege was at once a war zone and a manufactory.60 Even in the midst of combat, then, many soldiers spent much of their time creating the tools with which to wreak destruction. At times this meant not just creating, but inventing, as at the siege of PetraJustinianopolis in 550 when the Sabiri invented a new kind of portable battering ram light enough to be carried up the steep escarpment to the base of the fort’s walls.61 This creation-destruction paradox also made siege warfare particularly dangerous, for soldiers faced the gravest threats precisely in the effort to straddle these two opposing goals. This problem is illustrated brilliantly in a forthcoming article on the fate of the Persian sappers and Roman counter sappers at Dura Europos tower 19. Reanalyzing the archaeological remains, Simon James

Psychology The psychology of soldiers involved in siege operations will have been extremely complex. Sieges could be long, expensive, and perilously dangerous, to individuals and to armies. Morale was crucial to success, yet it had to be maintained over the longue durée by fostering staying power and maintaining intensity of effort in the face of constant, grueling resistance. Many of the psychological factors detracting from morale were beyond the control of the commanders. For this reason both sides needed to work hard to manipulate the psychological landscape at the command and structural levels using sophisticated strategies—vis à vis both their own men and the opponent. Siege operations almost universally began with attempts to convince the besieged to surrender through promises and threats. Thus it was in 359.62 Indeed, even before his arrival at Amida, at the beginning of his invasion of Mesopotamia, Shapur showed especial leniency toward the first two forts to surrender to him, Busan and Reman, in hopes of winning other surrenders more easily through his display of clemency (Amm. XVIII.10.2-4). If a major leader was present, as at Amida in 359 and 502-503, the counterpoint to these negotiations was that they offered the attackers a chance to display the level of power they were deploying against the defenders. Shapur’s desire to make his own presence and that of his client king Grumbates known to the Amidenes—even if it backfired—was intended, at least in part, to put them in fear of the consequences that might ensue from their

facinoribus speciosis inclaruit; XXIV.4.26: divisa itaque perpensis meritis et laboribus praeda. 57 Amm. XXIV.4.12: divisisque operibus officia quisque distributa capessit occissime; Zos. III.21.3-5; Jos. BJ V.357, 466-68, VI.150, 220, 502-3 with Goldsworthy 1999: 201-2. Ps. Maur. Strat. X.1.8 describes the need to estimate the number of men to assign each task during a siege. 58 Procop. Bell. II.21.30-34. So too Carrhae in 359, Amm. XVIII.7.3. Ammianus (XX.6.5, 7.9) describes how the inadequate repairs to damage at Singara rendered the wall more vulnerable to ramming; cf. XX.11.6. 59 Jul. Or. II.66b; cf. Zon. XIII.7.10; Theod. HE II.30.7-8; HR I.11; Efraem Syr. Carm. Nisib. XIII.17. Most of the sources on the 350 siege of Nisibis are conveniently collected and translated at Dodgeon and Lieu 1991: 193-207. 60 Veget. DRM IV.8.2-7. Such manufactories are confirmed by the archaeology at Dura Europos, cf. Leriche 1997: 88-92. Cf. Veget. DRM IV.19.1-2, which describes the construction of additional wall courses during a siege. 61 Procop. Bell. VIII.11.27-31; Cf. Procop. Bell. II.8.9, which describes the improvisation of platforms at Antioch in 540.

62

Amm. XIX.1.3-9; cf. XX.6.3, 7.3, 18, 11.7; XXI.12.4; XXIV.1.8, 2.13, 9, 4.11; Zos. III.14.3-15.2. Ps. Maur. Strat. X.1.6-7 recommends just such an initial approach: Cr¾ to‹j poliorkoumšnoij m¾ prote…nein ™n prooim…oij sklhr¦ kaˆ barša, †na m¾ tÍ duscere…v tîn proteinomšnwn tÕn k…ndunon koufÒteron logizÒmenoi, e„j ›nwsin kaˆ ¢pÒnoian œlqwsin.

230

NOEL LENSKI: TWO SIEGES OF AMIDA (AD 359 AND 502-503) refusal to acquiesce.63 If wheedling and threats failed, the besiegers usually began sieges much as Amida 359 began, by surrounding the city with armed men lined up with interlocking shields in pickets.64 The depth of these ranks would have given the defenders some idea about the size of the forces they were facing and no doubt affected their outlook on their chances of survival.65 Thus Ammianus writes of his reaction to the sight of Shapur’s 100,000 strong expeditionary force arrayed in five rings around the city:

The group psychology that united a soldier with his own comrades was also a regular factor in sieges. Indeed, recent work on ancient warfare has greatly emphasized the importance of group cohesion, especially at the unit level, in facilitating military success.69 Such cohesion was achieved in a variety of ways among both the attackers and defenders of a city or fort. Defenders were of course unified by the very fact of having lived together as common residents of the same fort or city for an extended period. For those who were, stricto sensu, citizens of a fortress city, their sense of civic identity and cohesion will have supplemented their unity as a fighting corps. Garrison troops will also have spent years together in the same locale, often together with their families, in whose defense they will have increased the strength of their resistance.70 In the course of battle, both sides often propagated and repeated simple acclamations to spur on loyalty and rouse esprit de corps. Thus in 359, Ammianus tells us that:

beholding such innumerable peoples, long got together to set fire to the Roman world and bent upon our destruction, we despaired of any hope of safety and henceforth strove to end our lives gloriously, which was now our sole desire.66 This ice-cold reaction to the threat of annihilation cannot have been uncommon given the number of sieges in which the defenders held out bravely against all odds (cf. Amm. XX.11.22). This same sort of pluck must have been what allowed the defenders of Nisibis to repel a Persian assault even after Shapur had caused their wall to collapse in 350 by diverting the River Mygdonius so that its waters came hurtling at the city (Efraem Syrus Carm. Nisib. II.18; Chron. Pasch. p. 537; Jul. Or. II.64c-66a). A related reaction, and indeed one that was also present in 359, was defiant mockery. It was the violent mockery and threats against Shapur and Grumbates that diverted the king from his original plans and toward the siege in the first place (Amm. XIX.1.3-2.1). Julian is said to have experienced the same reaction on more than one occasion from Persian cities that resisted him during his 363 invasion of Mesopotamia.67 At times the mockery was less violent and more crude, quite literally waving one’s private parts at the besiegers. Thus, local sources were quick to attribute the fall of Amida in 503 to divine retribution for the crude insults dished out to Kavadh by prostitutes during the siege. This sort of defiance, natural though it was, offered an affront to men like the shah or emperor whose honor demanded violent defense.68

The hills reechoed from the shouts rising from both sides, as our men praised the power of Constantius Caesar as lord of the world and the universe and the Persians called Shapur ‘saansaan’ and ‘pirosen’, which can be translated ‘king ruler over kings’ and ‘victor in wars’ (XIX.2.11 [trans. Rolfe]; cf. XXIV.4.23; Procop. Bell. II.29.1-2). The Antiochene youth who held out against Khusro’s Persians as they invaded the city in 540 rallied themselves by acclaiming “Emperor Justinian Triumphant!”71 Their identity as soldiers and citizens of an empire, as expressed through loyalty to the emperor or shah, thus had palpable, indeed audible effects in fostering group cohesion even as it provided a rhythm for group action. Group cohesion was balanced with inducements to competition based on honor, awards, and wealth. Undertaking and completing difficult operational maneuvers could guarantee public glory. Both Ammianus and Zosimus record, for example, the names of the first three sappers who penetrated the walls of Maiozamalcha in 363 – Exsuperius, Magnus, and Iovianus – for these had later been roundly praised and handsomely rewarded at a mass assembly with venerable military decorations.72

63

On the power exerted by a major leader during a siege see Amm. XIX.1.3, 7; XX.6.3, 7.3-4; Chron. Pasch. p. 538; Theod. HE II.30.8-10. 64 Amm. XIX.2.2: quinquiens ordine multiplicato scutorum cingitur civitas; XXI.12.4: ordine itaque scutorum gemino; XXIV.2.9: et armatorum triplici corona circumdatis muris; XXIV.4.10: muris duplicibus oppidum ordine circumdatum trino scutorum; cf. Jos. BJ III.146-48, 262. 65 This type of psychological effect is described at Veget. DRM IV.12.2: Illi enim qui muros invadere cupiunt terrifico apparatu expositis copiis in spem deditionis formidinem geminant; cf. Ps. Maur. Strat. X.1.3-4. 66 Amm. XIX.2.4 (trans. Rolfe); Cf. XIX.2.13: tot enim nos circumstantibus malis non obtinendae causa salutis, ut dixi, sed fortiter moriendi studio flagrabamus, et a diei principio ad usque lucem obscuram neutrubi proelio inclinato ferocius quam consultius pugnabatur. 67 Amm. XXIV.2.11, 5.4, 6; Zos. III.18.1. For violent mockery see also Amm. XX.7.3-4; XX.11.7. 68 Procop. Bell. I.7.17-19. See further examples at Amm. XIX.1.3-2.1; XX.7.2-4; XXI.12.7, 18-9; XXVI.8.2; Procop. Bell. II.27.13, 40.

69 MacMullen 1984: 228-34; Goldsworthy 1996: 252-57; Lee 1996: 207-10; cf. Holmes 1986: chs. 2 and 7. 70 On the tendency of garrison troops to settle in with family see Peri Strategias 9; cf. CJ I.27.2.8. 71 Procop. Bell. II.8.29: ‘IoustinianÕn basilša kall…nikon, ¤te nenikhkÒtej, ¢nškragon. Cfr. II.30.3: basilša ‘IoustinianÕn ¢nebÒwn kall…nikon. Ps. Maur. Strat. II.18 recommends religious acclamations like ‘Deus nobiscum’; cf. Veget. DRM III.5.4-5. 72 Amm. XXIV.4.23-24; Zos. III.22.4. Ammianus reports that coronae obsidionales were awarded, though this honor was traditionally reserved for soldiers who had raised a siege. By late antiquity, the honor had either taken on a new significance, or this instance reflects Julian’s taste for fustian, albeit tinged with the amnesia brought on by disuse. See Maxfield 1981: 67-9, and, more generally, Lendon 2005: 247-49. See also Josephus BJ III.233, which catalogs the names of the bravest defenders of Iotapata and BJ VII.13-16, where Titus personally

231

THE LATE ROMAN ARMY IN THE EAST Similar praise is reserved for the tribune Aliso who employed his Herculean strength and steely courage to break the chain blocking the harbor of Cyzicus singlehandedly during Procopius’s siege of that city in 365. This feat, carried out under heavy fire, won him his life after Procopius had been suppressed (Amm. XXVI.8.910). So too the giant Peter of ‘Amkhoro is known to us because of his exploits single handedly fending off Persians from the walls of Amida.73 At times even the commander was called to the pursuit of individual glory. Ammianus praises Julian for personally participating in a storming operation at Peroz Shapur, a feat which he compares with the bravado of Scipio Aemilianus.74 On top of his desire to win lasting fame and public recognition, a soldier might be tempted to feats of courage with monetary inducements. For the besiegers this will surely have included the knowledge that success would open the wealth of a city or fort to plundering and the enslavement of survivors who could be sold at a profit (e.g. Amm. XXIV.4.26-27; Procop. Bell. VI.27.31-34). Money was also offered as a reward for noteworthy performance even before a siege was decisively concluded. During their siege of Bezabde in late 360, a number of Roman soldiers removed their helmets on the battlefield in hopes of being recognized by the emperor Costantius II and therefore winning immediate rewards; instead such foolhardiness cost many their lives.75 Similar vain hopes prompted Dagisthaeus to write Justinian requesting rewards in anticipation of success with his siege of Petra-Justinianopolis in 549 (Procop. Bell. II.29.40). At times it was simply the more nebulously defined but very real phenomenon referred to by Keegan as the “will to combat” that spurred intensified performance, as at the siege of Maiozamalcha in 363 (Amm. XXIV.3.3, 4.11-14; cf. XIX.6.3-5, 7). Often this same euphoria coupled with the release of rage accumulated during the course of a long and grueling operation sparked indiscriminate slaughter if the besiegers succeeded in taking a city.76

engines were destroyed, commanders were killed, besiegers overtopped a tower or toppled one with their rams, the enemy produced a new machine, or forced his way through walls or a gate, resistance often melted away into resignation and even chaos. Thus the very sight of Julian’s helepolis induced the formerly defiant citizens of Peroz Shapur to surrender in 363, and the Roman defenders of Petra Justinianopolis lost heart and surrendered shortly after their brave commander John was shot in the neck and killed.77 Just as it often took just such a shocking event to break the resolve of one side to continue an orderly fight, the perception of imminent success often fired the winning side to deal the coup de grace to its enemies. After days, weeks, months of grit and resistance, the symbolic walls could come tumbling down following a major breakthrough that operated very much like the trope in the Greek phalanx (cf. Hanson 2000: 177-184). Watching one’s efforts and stratagems, so carefully planned and executed, succeed and the enemy crumple into chaos and despair could create a psychological turning point that fed simultaneously “the thrill of victory and the agony of defeat.” The psychology of siege warfare was profoundly communitarian and resulted in the creation of strong bonds of esprit de corps among both defenders and besiegers. Thus in the 359 siege of Amida, Ammianus can barely conceal his scorn for the ungainly and rude Gallic legionaries – classics outsiders in this eastern context – with whom he was forced to coexist (XVIII.9.3; XIX.5.2, 6.3-5); yet following their nighttime foray against the Persian outposts, great risks were run to ensure their safe return to the city by opening the gates in the face of a Persian onslaught. At the siege of Antioch in 540, the youth of the circus factions, who had formerly fought with each other in the streets, formed strong bonds that led them to fend off the Persian besiegers with considerable success.78 Much of the success in generating cohesion among the defenders, be they soldier or civilian, was related to a predominant lifeboat mentality (cf. Jos. BJ III.261). This was especially true because, regardless of the fact that the besiegers were likely to suffer higher total casualties, the besieged could expect extermination or at best enslavement as a consequence of defeat. Because survival and not just victory were at stake, social and personal divisions melted away in the heat of a mortal threat and were recast as solidarity.

Dejection also regularly set in among both sides, especially when massive efforts failed or the tide turned against one’s efforts: when siege mounds collapsed, siege

recognizes the bravest soldiers from the Jerusalem siege. Caesar’s (BG V.44) heroes Pullo and Vorenus, singled out for bravery in the relief of Q. Cicero’s camp during the Belgic revolt of 54, have been skyrocketed to contemporary fame in the TV blockbuster Rome. 73 Ps. Zach. VII.4 (Hamilton-Brooks 27); cf. Jos. BJ VI.81-90. Similar feats of brinkmanship are cataloged at Lendon 1997: 243-52 and 2005: passim. 74 Amm. XXIV.2.14-17, cf. XXIV.4.3-4. In the last hours before the fall of Amida, even Shapur became directly involved in the storming operations, Amm. XIX.7.8. Compare Titus’s direct involvement at Jerusalem, Jos. BJ V.85-97. 75 Amm. XX.11.12. Other examples of the power of monetary inducements at Amm. XXIV.3.3, 7; Procop. Bell. II.26.13; Jos. BJ V.349-51, VII.13-16. On the power of monetary inducements to stimulate performance cf. MacMullen 1984: 231; Goldsworthy 1996: 259-61; Lee 1996: 205-6. 76 Amm. XIX.8.4, 9.2; XXIV.4.25; Ps. Zach. VII.4 (Hamilton-Brooks 28); Vita Sym. Styl. Iun. 57; cf. Jos. BJ III.304-6, 329-31, VI.403-22. More on the treatment of captured cities at Kern 1999: 323-51.

77 On Peroz Shapur see Amm. XXIV.2.19-21; cf. Zos. III.18.2-3, with Matthews 1989: 293-4 on the helepolis. On Petra-Justinianopolis see Procop. Bell. II.17.16. Other examples of the collapse of morale following a major setback at Amm. XIX.5.8, 6.1, 7, 13, 7.1, 8, 8.1, XX.11.11-13, XXI.12.11-12; XXIV.2.19-20, 25; XXXI 15.12; Procop. Bell. II.5.11-13; VIII.33.22-3. Agath. II.14.3-6; III.26.1-6; Joh. Eph. HE VI.5. 78 Procop. Bell. II.8.17: toà RÁm oân d»mou nean…ai polloˆ, Ósoi t¦ prÒtera prÕj ge ¢ll»louj stasi£zein ™n to‹j ƒppodromšoij e„èqesan, ™peid¾ ¢pÕ toà peribÒlou katšbhsan, oÙdamÁ œfeugon, ¢ll’ aÙtoà œmenon; cf. Procop. Bell. II.8.28: ™ntaàqa dÁ aàto‹j tîn ‘Antiocšwn nenan…ai polloˆ ™j ce‹raj ™lqÒntej t¦ prîta katupšrteroi œdoxan tÍ xumbolÍ emai.

232

NOEL LENSKI: TWO SIEGES OF AMIDA (AD 359 AND 502-503) The pull toward cohesion will also have resulted from the profound need for unwavering unity in order to avoid any form of betrayal. Desertions, intelligence leaks, and even simple carelessness could spell the collapse of a city’s defense, and even of a siege operation. Amida was set in crisis in 359 when a deserter revealed the presence of its secret passageway to the Tigris and again in 502-503 when the bravado of the skirmisher Kutrigo allowed the Persians to figure a clandestine way into the city.79 From the other side, the Armenian defenders of Artogerassa in 367 were able to annihilate their Persian attackers when the Armenian naxarark charged by Shapur with conducting the siege betrayed the forces under their command to their compatriots.80 Mutatis mutandis, unity in the concealment of intelligence could offer the key to success. The Persian defenders of Amida in 504 were able to negotiate a favorable peace precisely because they succeeded in concealing the fact that they were on the brink of starvation (Procop. Bell. I.9.1-21). Of course, when the sacred bonds of secrecy and solidarity were felt to have been betrayed, blame and even reprisals followed. Thus the monks from the monastery of John of Urtaye in Amida were blamed for failing the city by falling asleep the fateful night of January 10, 503, and a priest of Bezabde who negotiated with the Persians in 360 was accused of revealing to them the city’s weak spots.81 At times like these, the same bonds that united a community could boomerang on one’s compatriots. Nevertheless, in its own way such collective resentment could further build community: the Amidenes’ collective anger with the deserter who led the Persian royal archers into their city helped them to rally to the city’s defense and drive the archers from the tower they had captured (Amm. XIX.5.8: augebat iras militum scelestum facinus perfugae). In the worst cases, negative communal sentiment could be leveled at scapegoats, as when the citizens of Aquileia excused their resistance to Julian’s siege forces by offering three of their citizens for execution as the instigators of their intransigence (Amm. XXI.12.19-20; cf. Evag. V.10). So too, on the side of the besiegers, Dagisthaeus was eventually imprisoned by Justinian after being blamed by the Lazi for his failure to capture Petra-Justinianopolis in 549 (Procop. Bell. VIII.9.3-4), and Nachoaragan was flayed by Khusro

when he was turned back from Phasis in 556 (Agath. IV.23.2-3). Just as the community exercised its collective power to rally strength against the enemy, it could redeploy that same power against those who were blamed for its failure to achieve military objectives. One final factor that would have had a particular effect on battlefield psychology was acoustics. To be sure war is noisy business, whether conducted in the field or around fortifications, but siege operations involved particular kinds of noises – as well as silences – that must be factored into any attempt to account for the interplay between a soldier’s experience and combat operations: the rhythmic pounding of battering rams, the crackling of a siege machine set aflame, the clank of hammer and shovel in a sap, the scream of ballistae, the clatter of projectiles striking armor, the rumbles and groans of a rolling siege tower, the shouts and cheers of stormers attacking a wall, the ululations of those civilians who lost loved ones, the crack of a catapult ball bashing against a stone wall, the rumbling crash of a collapsing wall or mound, all would have factored into the enervation and fatigue, the terror and mayhem experienced by the soldier. The conventions of tragic history of course influenced the manner in which historians like Ammianus chose to report the cacophony of a siege, yet it is as difficult to argue that Ammianus’s descriptions are irrelevant to a soldier’s experience as it is difficult to contend that he did not experience at first hand the noises he reports. His account of the Amida siege resounds with the clangor of battle: the blare of horns calling the Persian soldiers to their first day of combat,82 the whir of ballista bolts,83 the trumpeting of elephants “than whose clamor and immense bodies the human mind can conceive nothing more terrible,”84 the howls and pleas of those shot by missiles,85 even the lugubrious strains of the Persian army lamenting its war dead would have colored the soldier’s view of battle as much as they enliven the reader’s second-hand experience (Amm. XIX.6.13; cf. Matthews 1989: 63). At times noise was even used as a sort of weapon, as when the Edessenes succeeded in driving back a rampaging elephant by dangling from their walls a pig whose squeals frightened the beast away (Procop. Bell. VIII.14.35-38), or when the Persians besieging Phasis in 556 banged drums and raised shouts with the deliberate intention of terrifying the city’s defenders (Agath. III.25.7; cf. Jos. BJ III.265).

79

On 359 see Amm. XIX.5.5-6. On 502-503 see above at n. 19. For further examples see Amm. XX.6.5, 7.9; Procop. Bell. VII.37.9-10; VIII.14.23-28; Ps. Josh. Styl. 71[296]; Theophylact Sim. III.5.11-13; Evag. HE VI.14; Chron. 1234 78; Jos. BJ III.317. Captives too could reveal important intelligence, as at Procop. Bell. II.19.19-24; Agath. III.6.1-3; Theophylact II.15.10-12, though at times they were sent as double agents, on which Frontin. Strat. III.6. 80 Amm. XXVII.12.6-8. See also Procop. Bell. II.13.22-26, where a Persian skirmisher collecting arrows inside the inner walls of Dara revealed that the Persians were about to dig a sapping tunnel below the eastern walls, allowing the Roman defenders to dig a counter tunnel, and Procop. Bell. II.20.10-14, where the Christian Lakhmid Ambrus approached the walls of Rusafa-Sergiopolis by night in 542 to report that the Perisans had only two days supply of water, thus allowing the Sergiopolitans to force a Persian retreat, cf. Fowden 1999: 133-34. 81 On the monks of John of Urtaye see Ps. Zach. VII.4 (HamiltonBrooks 26); Theophanes a.m. 5996; Marcellinus Comes s.a. 502; cf. Trombley and Watt 2000: 61 n.293. On Bezabde see Amm. XX.7.9.

Paradoxically, quiet and even silence also played an important role in sieges. On the first day of their siege of Amida in 359, the Persians stood still “as through rooted in the same spot; no sound was heard, no neighing of 82

Amm. XIX.2.5, 6.10; cf. XX.11.19; XXIV.4.15, 22; XXIX.7.3; XXXI.15.13. Amm. XXIV.4.16: Tum aptatae ligneis sagittis ballistae, flexus stridore torquebantur; cf. XXIV.4.12. 84 Amm. XIX.7.6. On Ammianus’s frisson in the face of elephants see also XXV.1.14, 3.4; Matthews 1989: 64. 85 Amm. XIX.2.15, quoted above at n. 50; cf. XIX.1.8. Josephus BJ III.248 recalls with uncanny poignancy the sound of the bodies of men thudding to the ground from off the battlements after being struck by artillery. 83

233

THE LATE ROMAN ARMY IN THE EAST horses”. 86 Perhaps the intention was to model their discipline before the foe, perhaps to demonstrate the solemnity of the operation they were about to undertake, or even to forebode the eerie desolation of the wasteland they planned to create. At times silence was a tactical imperative, as at Amida in 503 when the Persians snuck past the sleeping monks, or when the Gallic Magnentiaci needed to sneak silently into the Persian camp to wreak havoc but were nonetheless betrayed by the groans of those they were murdering (Amm. XIX.6.7-8; cf. Jos. BJ III.324). As the Persians attackers first approached PetraJustinianopolis to besiege it in 541, the Roman commander John ordered his men to hide in total silence until the unsuspecting attackers approached closely enough to offer easy targets for a surprise assault (Procop. Bell. II.17.3-10). Sappers in particular had to work quietly even as they excavated earth and busted rock below ground (Amm. XXIV.4.22-3; Agath. I.10.5; cf. Procop. Bell. I.7.14). The anonymous author of the Byzantine treatise Peri Strategias (13) recommends that defenders literally keep an ear to the ground to detect the sounds of undermining. The effectiveness of this simple technique can be seen at the Persian siege of Edessa in 544 when the Persians detected the sounds of the Roman sappers who were attempting to undermine their siege mound and therefore began a countermine of their own (Procop. Bell. II.27.1-4). To counter this problem and drown out the sounds of their digging, the intensity of an assault was deliberately increased as sappers approached the enemy lines. In this sense, there was an enigmatic tension between quiet and noise inherent to sieges. This was true not only in the instance of undermining, but in others as well, as when the Persian royal archers who took the tower of Amida in 359 went from utter silence to ululation in the moment they wished to notify their comrades of their success (Amm. XIX.5.5; cf. Procop. Bell. VII.37.9-10).

sorties, storming, mining, clandestine raids – were both common and extremely important in deciding the outcome of a siege. At the level of human limitations, the evidence shows how important were the limitations placed on visibility by the various devices used to provide cover as well as by volleys of missiles and darkness; it also shows that starvation and thirst played only a limited role in these late Roman sieges in the east, in large part because of deliberate limitations in the size of garrisons and a studied concentration on provisioning them adequately; and the evidence also points to the limitations placed on combat performance by wounds and sickness and the need for hard, physical labor, much of it to create things meant for destruction. Finally, at the psychological level, this investigation has explored the gradual escalation in pressure on a fort and the fact that this often simply strengthened the resolve of the defenders; each side was forced to work together as a community, and, in the instance of the defenders, this esprit de corps was reinforced by the presumption that failure would bring total annihilation; competition simultaneously forced cohesion and encouraged the sorts of bravado that could decide a siege; yet at some point, often a quite specific and definable point, dejection set in on one side and weakened that side’s resistance even as it fired the other with a stronger will to combat. Bibliography Allen P. 1980 Zachariah Scholasticus and the Historia Ecclesiastica of Evagrius Scholasticus. In JThS 31: 471-88. Allen P. 1988 An Early Epitomator of Josephus: Eustathius of Epiphaneia. In BZ 81: 1-11. Barry W.D. 1996 Roof Tiles and Urban Violence in the Ancient World. In GRBS 37: 55-74. Campbell D.B. and Delf B. 2003 Greek and Roman Siege Machinery, 399 BC-AD 363, Wellingborough. Campbell D.B. and Hook A. 2005 Siege Warfare in the Roman World, 146 BC – AD 378, Wellingborough. Christensen A. 1944 L’Iran sous les Sassanides, 2d ed., Copenhagen. Clark C.U. (ed.) 1910 Ammiani Marcellini Rerum Gestarum Libri qui Supersunt, 2 vols., Berlin. Coello T. 1996 Unit Sizes in the Late Roman Army. BAR IS 645, Oxford. Croke B. and Crow J. 1983 Procopius and Dara. In JRS 73: 143-59. Crump G.A. 1975 Ammianus Marcellinus as a Military Historian, Wiesbaden. de Jonge P. 1982 Philological and Historical Commentary on Ammianus Marcellinus XIX, Groningen. den Hengst D. 1999 Preparing the Reader for War: Ammianus’ Digression on Siege Engines. In Drijvers J.W. and Hunt D. (ed.), The Late Roman World and its Historian: Interpreting Ammianus Marcellinus, London: 29-39.

Conclusion The city of Amida has thus provided us with something of the soldier’s eye view of combat like that so deftly reconstructed by Keegan. The sources for the sieges of 359 and 502-503 do not, unfortunately, give us the level of detail nor the multiple perspectives that Keegan and other modern historians have available, but they do provide a remarkably vivid window onto the distant world of combat in antiquity. When combined with material from other extant siege narratives, they can inform the discussion of ancient warfare on several levels. On the operational level, they show the almost universal disparity in the size of defending versus besieging forces and the fairly tight temporal limitations placed on siege operations by the short campaign season in the Roman Near East; they also show that acts of bravery in what moderns might call special operations – 86

Amm. XIX.2.5: a sole itaque orto usque diei ultimum acies immobiles stabant ut fixae nullo variato vestigio nec sonitu vel equorum audito hinnitu; cf. XIX.2.2.

234

NOEL LENSKI: TWO SIEGES OF AMIDA (AD 359 AND 502-503) Lee A. D. 1996 Morale and the Roman Experience of Battle. In Lloyd A.B. (ed.), Battle in Antiquity, London: 199-217. Lendon J. E. 1997 Empire of Honour: The Art of Government in the Roman World, Oxford. Lendon J. E. 1999 The Rhetoric of Combat: Greek Military Theory and Roman Culture in Julius Caesar’s Battle Descriptions. In ClAnt 18: 273-329. Lendon J. E. 2005 Soldiers and Ghosts: A History of Battle in Classical Antiquity, New Haven. Lenski N. 2002 Failure of Empire: Valens and the Roman State in the Fourth Century AD, Berkeley. Lenski N. forthcoming The Chronology of Valens’s Dealings with Persia and Armenia, 363-378 CE. In den Boeft J. - den Hengst D. - Drijvers J.W. and Teitler H. (ed.), Ammianus After Julian, Leiden. Leriche P. 1997 Tranchée sur la rue principale et fouille d’un sanctuaire et d’une maison de l’îlot M5 à Doura Europos. In Leriche P. and Gelin M. (ed.), DouraEuropos Études IV, 1991-1993, Beirut: 81-94. Lightfoot C.S. 1988 Facts and Fiction: The Third Siege of Nisibis (AD 350). In Historia 37: 105-125. Lightfoot C.S. 1989 Sapor before the Walls of Amida. In French D.H. and Lightfoot C.S. (ed.), The Eastern Frontier of the Roman Empire: Proceedings of a Colloquium held at Ankara in September 1988, Oxford: 285-294. Lloyd A.B. (ed.) 1996 Battle in Antiquity, London. MacMullen R. 1984 The Legion as Society. In Historia 33: 440-56 = Changes in the Roman Empire: Essays in the Ordinary, Princeton: 225-235 and 368-374. Marsden E.W. 1969 Greek and Roman Artillery: Historical Development, Oxford. Marshall S.L.A. 1947 Men against Fire: The Problem of Battle Command in Future War, New York. Matthews J.F. 1989 The Roman Empire of Ammianus Marcellinus, London. Maxfield V.A. 1981 The Military Decorations of the Roman Army, Berkeley. Mitchell S. 1995 Cremna in Pisidia: An Ancient City in Peace and in War, London. Nau F. 1915-1917 Résumé de monographies syriaques. In ROC 2nd ser. 10: 3-12. Naudé C.P.T. 1988 Battles and Sieges in Ammianus Marcellinus. In AClass 1: 92-105. Nicasie M.J. 1998 Twilight of Empire: The Roman Army from the Reign of Diocletian until the Battle of Adrianople, Amsterdam. Price J.J. 1992 Jerusalem under Siege: The Collapse of the Jewish State 66-70 C.E., Leiden. Reeve M.D. (ed.) 2004 Vegetius Epitoma Rei Militaris, Oxford. Roth J.P. 1999 The Logistics of the Roman Army at War (264 B.C. – A.D. 235), Leiden. Rosen K. 1970 Studien zur Darstellungskunst und Glaubwürdigkeit des Ammianus Marcellinus, Bonn. Rostovtzeff M.I. et al. 1936 Excavations at Dura Europos, Sixth Season, London.

Dodgeon M.H. and Lieu S.N.C. 1991 The Roman Eastern Frontier and the Persian Wars, AD 226-363: A Documentary History, London. Elton H. 1996 Warfare in Roman Europe, AD 350425, Oxford. Foss C. and Winfield D. 1986 Byzantine Fortifications: An Introduction, Pretoria. Fowden E.K. 1999 The Barbarian Plane: Saint Sergius between Rome and Iran, Berkeley. Gabriel A. 1940 Voyages archéologique dans la Turquie orientale, 2 vols., Paris. Goldsworthy A.K. 1996 The Roman Army at War, 100 BC-AD 200, Oxford. Goldsworthy A.K. 1997 The othismos, Myths and Heresies: The Nature of Hoplite Battle. In War in History 4: 1-26. Goldsworthy A.K. 1999 Community under Pressure: The Roman Army at the Siege of Jerusalem. In Goldsworthy A. and Haynes I. (ed.), The Roman Army as a Community. JRA Supplements 34, Portsmouth RI: 197-209. Gregory S. 1997 Roman Military Architecture on the Eastern Frontier, 3 vols., Amsterdam. Greatrex G. 1998 Rome and Persia at War, 502-532, Leeds. Greatrex G. and Lieu S.N.C. 2002 The Roman Eastern Frontier and the Persian Wars Part II, AD 363-630: A Narrative Sourcebook, London. Haldon J. 1999 Warfare, State and Society in the Byzantine World, 565-1204, London. Halsall G. 2003 Warfare and Society in the Barbarian West, 450-900, London. Hanson V.D. (ed.) 1991 Hoplites: The Classical Greek Battle Experience, London. Hanson V.D. 2000 The Western Way of War: Infantry Battle in Classical Greece, 2nd ed., Berkeley. Haury J. (ed.) 1905 Procopii Caesariensis opera omnia, vol. 1, Leipzig. Holmes R. 1986 Acts of War: The Behavior of Men in Battle, New York. Isaac B. 1992 The Limits of Empire: The Roman Army in the East, rev. ed., Oxford. James S. forthcoming. The Deposition of Military Equipment during the Final Siege at Dura-Europos, with Particular Regard to the Tower 19 Countermine. James S. 2004 The Excavations at Dura-Europos conducted by Yale University and the French Academy of Inscriptions and Letters 1928 to 1937: Final Report VII The Arms and Armour and other Military Equipment, London. Keegan J. 1976. The Face of Battle, New York. Kern P. B. 1999 Ancient Siege Warfare, Bloomington. Kugener M. A. 1900 La compilation historique de Pseudo-Zacharie le rhéteur. In ROC 5 : 201-214. Lander J. 1984 Roman Stone Fortifications: Variation and Change from the First Century A.D. to the Fourth. BAR IS 206, Oxford. Lawrence A. W. 1983 A Skeletal History of Byzantine Fortification. In ABSA 78: 171-227.

235

THE LATE ROMAN ARMY IN THE EAST Sabbah G. 1978 La méthode d’Ammien Marcellin: Recherches sur la construction du discours historique dans les Res Gestae, Paris. Sabin P.A.G. 2000 The Face of Roman Battle. In JRS 90: 1-17. Schmidt O. 2001[2004] Stärke, Struktur und Genese des comitatensischen Infanterienumerus. In BJ 201: 93111. Shaw B.D. 1999 War and Violence. In Bowersock G. Brown P. and Grabar O. (ed.), Late Antiquity: A Guide to the Postclassical World, Cambridge MA: 130-169. Southern P. and Dixon K.R. 1996 The Late Roman Army, New Haven. Tomlin R. S. O. 2000 The Legions in the Late Empire. In Brewer R.J. (ed.), Roman Fortresses and Their Legions, London: 159-181. Trombley F. 1999 Ammianus Marcellinus and FourthCentury Warfare. In Drijvers J.W. and Hunt D. (ed.), The Late Roman World and its Historian: Interpreting Ammianus Marcellinus, London: 17-28. Trombley F. and Watt J. (trans. and comm.) 2000 The Chronicle of Pseudo-Joshua the Stylite, Liverpool. van Berchem D. 1954 Recherches sur la chronologie des enceintes de Syrie et de Mésopotamie. In Syria 31: 254-267. van Berchem D. and Strzygowski J. 1910 Amida, Heidelberg. Wheeler E.L. 1998 Battles and Frontiers. In JRA 11: 644-651. Wheeler E.L. 2001 Firepower: Missile Weapons and the ‘Face of Battle.’ In E. Dąbrowa (ed.), Roman Military Studies. Electrum 5, Krakow: 169-184. Wheeler E.L. 2004a The Legion as Phalanx in the Late Empire (I). In LeBohec Y. and Wolff C. (ed.), L’Armée Romaine de Dioclétien à Valentinien Ier, Lyon: 309-358. Wheeler E.L. 2004b The Legion as Phalanx in the Late Empire, Part II. In Revue des études militaries anciennes 1: 147-175. Whitby M. 1988 The Emperor Maurice and his Historian: Theophylact Simocatta on Persian and Balkan Warfare, Oxford. Whitby M. 1998 Deus Nobiscum: Christianity, Warfare and Morale in Late Antiquity. In Austin M. - Harries J. and Smith C. (ed.), Modus operandi: Essays in Honour of Geoffrey Rickman, London: 191-208. Whitby M. 2000 The Army, c. 420-602. In Cameron A. Ward-Perkins B. and Whitby M. (ed.), The Cambridge Ancient History Volume XIV, Late Antiquity: Empire and Successors, A.D. 425-600, 2nd ed., Cambridge: 288-314. Yadin Y. 1966 Masada: Herod’s Fortress and the Zealot’s Last Stand, New York.

236

Naval Operations During Persian Expedition of Emperor Julian (363 AD)* Edward Dąbrowa Uniwersitet JagielloĔski, Kraków

later wars against the eastern neighbour, the Roman emperors likewise employed their fleet on a larger or smaller scale. The use of ships made it possible for Septimius Severus to capture Seleucia-on-the-Tigris and Ctesiphon.6 We also know that the fleet provided the transportation of supplies during the Persian venture of Gordian III.7

The Emperor Julian’s expedition against the Persians in 363 AD has long interested scholars. The many studies devoted to it have arisen out of research on his reign and analyses of the operation itself, as well as from interpretation of sources,1 foremost among which are unquestionably the works of Ammianus Marcellinus and Zosimos.2 There is now a vast body of literature which analyzes the military aspects of Julian’s Persian venture. What this article proposes to scrutinize is its accompanying naval involvement, for the fleet played an active part in the campaign, as is unambiguously testified by the sources. They describe the fleet’s operations in greater or lesser detail, but their real value is that they enable us to determine the tasks set for the fleet to perform at the planning and preparation stage, and the fleet’s subsequent practical implementation of its objectives. As a side benefit, the naval scenario of 363 may be helpful in reconstructing other similar operations conducted during some earlier Roman wars against its eastern neighbour, whether against the Parthians under the Early Empire, or later, when Rome was pitted against the Sasanian Empire. Many references to such prior engagements were naturally made in the context of Julian’s military action.3

The strategic importance of the Euphrates lay in the fact that much of its course constituted the boundary between Rome and the Iranian world. The river’s flow naturally determined the direction of an advancing Roman army. In the deserts and semi deserts of northern Mesopotamia, the Euphrates was the chief and only natural source of water to supply great masses of men and beasts, while it also provided a convenient route of transportation. The latter function, given the river’s strong current, was only available to the Romans. War against the Persians had become a priority in emperor Julian’s policy right from his accession to power.8 The intention was radically to alter relations with Persia so as to improve the prestige of Rome and its rulers in the eyes of subjects and neighbours alike.9 The emperor’s determination was such that he could not be made to abandon his plans either by negotiations proposed by ShƗpnjr II of Persia10 or by resentment toward the expedition voiced by parts of society and the political elites.11

At least from the early 1st century AD, the Euphrates or Tigris had been the theater of war in almost all such conflicts. During fighting with the Parthians over Armenia in Nero’s reign, the Syrian bank of the Euphrates was fortified by Cn. Domitius Corbulo against possible enemy attack.4 In the course of Trajan’s Parthian expedition (113-117), the Roman fleet operated on the Tigris, enabling the emperor and his army swiftly to advance to the walls of Seleucia-on-the-Tigris and Ctesiphon.5 We can almost take it for granted that in all

In planning his operation, Julian studied accounts of previous engagements in Rome’s wars against the Parthians and Persians.12 Such studies probably influenced his own strategy and tactics employed on enemy territory. The knowledge he thus gained proved useful in making decisions as events were unfolding. It made him direct his fleet to the so-called Trajan canal, which had been masked by the Persians, enabling the

* I would like to acknowledge the friendly assistance of Professor Anthony R. Birely with correction of the English of this paper. 1 Studies on sources on Julian and his Persian expedition dwell on controversies concerning the dependences between respective ancient authors and their mutual borrowings, see Thompson 1947: 28-34, 134137; Norman 1957: 129-133; Chalmers 1960: 152-160; Dillemann 1961: 110-135, 144-151; Cameron 1963: 232-236; Fontaine 1968: 2528; Blockley 1973: 54-55; Ridley 1973: 317 and note 2; Arce 1974: 340 ff.; Matthews 1989: 161-179; Paschoud 1979: xii-xix; Bleckmann 1992: 375-383; Den Boeft et al. 1998: xii ff. Still, such contentions have little impact on the subject of this discussion. 2 Of particular interest in recent years has been the work by Ammianus Marcellinus: Fontaine 1977; Fontaine 1996; Den Boeft et al. 1995; Den Boeft et al. 1998; Den Boeft et al. 2002. Zosimos’ account has gained an excellent commentary: Paschoud 1979. 3 Cf. Lib., Or. 18, 233; 18, 245; Amm. XXIV.6.1; Zos. III.24.2. 4 Cf. Dąbrowa 1986: 97 ff. 5 Dio Cass. LXVIII.26.1-3; 28.1-2; Suda, s.v. ȞĮ³Ȣ; cf. Amm. XXIV.6.1. We have indications suggesting that the Euphrates was also considered in planning the war, probably as a supply route for troops

operating in Babylonia. Unfortunately, our fragmentary sources leave us to speculations in this respect, cf. ILS 9471; Lepper 1948: 9-10, 133 ff. 6 Herodian III.9.9-10; cf. Amm. XXIV.6.1. Herodian’s account suggests that the capture of those cities by Septimius Severus resulted from an unexpected coincidence. 7 SHA Gord. 29, 2. 8 According to Zosimos (III.11.3) Julian began preparations for the campaign soon after arriving in Constantinople. They lasted continuously for the whole period of his stay in Antioch, which lasted for several months (Zos. III.11.5; cf. Amm. XXII.12.1). See also Andreotti 1930: 245 ff.; Blockley 1992: 24 ff.; Den Boeft et al. 1995: 214 ff.; Fontaine 1996: 319 f., note 939. 9 Cf. Lib., Or. 17, 19; Amm. XXII.12.1-2; Andreotti 1930: 241 ff.; Seager 1997: 253 ff., 263 ff. 10 Lib., Or. 12, 76-77; 17, 19; 18, 164. 11 Amm. XXII.12, 3-5; XXIII.5.4; Mathews 1989: 134 ff.; Blockley 1992: 26; Den Boeft et al. 1995: 217 ff. 12 Lib., Or. 18, 233, 245. Cf. Greg. Naz., Or. 5, 8.

237

THE LATE ROMAN ARMY IN THE EAST ships to reach Ctesiphon relatively uneventfully.13 Choice of the Euphrates as the main line of attack was allimportant,14 for it determined where the fleet was to be built and the necessary provisions stored. A witness of and participant in Julian’s Persian expedition, Magnus of Carrhae, suggests that the site for shipbuilding was chosen at and around Samosata, it being an area rich in timber.15

mentions that they were built of beams joined together, which suggests that they looked more like rafts than enclosed-hull vessels. This type of unit was called šahtnjr. Contrary to views widely accepted not long ago, the leather vessels were by no means a separate type of craft – known as kelek – but rather a variation of šahtnjr probably intended for transporting foodstuffs. The use of leather in their construction was meant to protect their precious cargo from moisture.22 They were designed to be simple – they would only be used in downstream river navigation – and sufficiently solid to assure safety to crews and cargoes. Once their destination was reached, their structural timbers would normally be sold.23 Depending on size, a šahtnjr vessel could carry loads between ten and 20 tons.24

Extant descriptions of Julian’s fleet permit a fairly accurate knowledge of its size and composition. According to Ammianus Marcellinus, its core consisted of 1000 transport craft (naves onerariae). In addition, there were 50 warships (naves bellatrices) and the same number of auxiliary vessels that could be used to build a floating bridge.16 Magnus of Carrhae says that the fleet numbered 1250 units in total,17 some of them built of wood (IJÀ ȝÁȞ įȚÀ ȟÌȜȦȞ), and others of leather (IJÀ dÁ įȚÀ ȕuȡıȫȞ).18 More chaotic in his account, Zosimos states that the fleet possessed 600 wooden ships (Ñț ȝÁȞ ȟȪȜȦȞ ʌİȠȚȘȑȞȦȞ), 400 vessels made of leather (ÐʌÄ įİȡȝȐIJȦȞ), 50 warships (ıIJȡĮIJȚȦIJȚțĮà Ȟ±İȢ), an unspecified number of floating platforms to serve as a pontoon bridge, plus many other units of various types.19 Zonaras gives the total number of 1100 units, of which as many as 700 were warships, while the remaining ones were transports.20 There is no reason to question the veracity of Magnus of Carrhae’s figure. It will be sensible to assume that it embraces all manner of vessels including smaller boats, probably the same ones that Zosimos had in mind in referring to a large number of other units of various types (įÁ țĮà ðȜȜĮ ʌȐȝʌȠȜȜĮ ʌȜȠ²Į).21 Some interesting details can be deduced concerning the construction of the ships. Ammianus Marcellinus

A serious problem to solve was how to man the fleet with a sufficient number of trained sailors. We do not know how this question was handled by Julian’s predecessors. Since the Romans did not regularly practice navigation on the Euphrates, finding a suitable number of men acquainted with sailing and with that particular river was next to impossible. What we know is that Julian’s shipboard crews numbered as many as 20,000 men.25 Had that body been composed of civilian sailors and members of other Roman flotillas, not only would their experience have been of little use on the Euphrates, but on reaching the fleet’s destination they would have become a useless encumbrance for a fighting army. It seems therefore that the crews of Julian’s vessels were most probably formed of suitably trained soldiers.26 This supposition is confirmed by the active part played by ship crews in operations in the march on Ctesiphon and their subsequent inclusion in the army when the fleet had been disbanded.27

13

Lib., Or. 18, 244-247; cf. Amm. XXIV.6.1-2; Zos. III.24.2. The canal attributed to Trajan by Roman authors (Dio Cass. LXVIII.28.1-2) was probably started during the reign of Seleucos I: Paschoud 1978: 353 ff., 357 ff.; Den Boeft et al. 2002: 172. 14 Julian’s alleged hesitation as to his choice of route is better explained by a need to keep his strategic plans secret (cf. Lib., Or. 18, 213) than by indecision. Trajan had chosen for his attack a path along the Tigris and built his fleet at Nisibis using timber from the surrounding forests, from where he transported the ships overland to the launch site: Dio Cass. LXVIII.26.1-2. Choice of a shipbuilding site was conditional on the abundance of timber. Such policy suggests that the Romans had a good knowledge of the area and of navigation. The latter determined the design of the vessels. By contrast, Alexander the Great in preparing his attack on Mesopotamia had had his fleet built in Cyprus and Phoenician cities before it was transported in parts to Thapsacus: Strabo XVI.1.11 (741). 15 Magnus, FGrH, no. 225, F 1 = Paschoud 1979: 242 (1) = Malalas 13, 21/Thurn 2000; Zos. III.12.1. 16 Amm. XXIII.3.9. 17 Magnus, FGrH, no. 225, F 4 = Paschoud 1979: 242 (4) = Malalas 13, 21/Thurn 2000. 18 Magnus, FGrH, no. 225, F 1 = Paschoud 1979: 242 (1) = Malalas 13, 21/Thurn 2000. 19 Zos. III.13.2-3. Zosimos’ account gives the impression of being composed of two stylistically mismatched fragments: one on the number of ships, the other on their design: Ridley 1972: 319. 20 Zon. XIII.8: ¶ıĮȞ įÁ IJȡȚȒȡİȚȢ ȝÁȞ ØʌIJĮțȩıȚĮȚ, ijȠȡIJȘȖȠÃ įe IJİIJȡĮțȩıȚĮȚ. 21 Zos. III.13.3. For a discussion of the size of the fleet, see Ridley 1972: 319; Fontaine 1977: 32 f., note 71; Paschoud 1979: 113 f., note 35; Den Boeft et al. 1998: 54 f.

22

For earlier discussions on identification of vessel types in Julian’s fleet, see Tardieu 1990: 95-102. Such craft, called kelek, which utilized in their design not just timber but also sheep skins inflated with air for added buoyancy, were only used in navigating the Tigris and its tributaries. Owing to the more difficult navigation on the Euphrates, they were never used there: Tardieu 1990: 81 f. 23 Dillemann 1961: 152; Matthews 1989: 147; Tardieu 1990:75 ff., 83 ff. The ships Alexander the Great had built for his Mesopotamian campaign were probably shell-type craft which the Greeks were commonly using at the time. It is not known whether their design stood the test on the Euphrates. It is a fact, however, that when Alexander ordered more ships built in Babylon, their design was similar to that of boats used by the locals: Strabo XVI.1.11 (741); cf. Amm. XXIV.3.11. 24 Tardieu 1990: 101. 25 Amm. XXIV.7.4: armatorum fere viginti milia in trahendis occuparentur eisdem navibus et regendis. 26 The basic skill involved in navigating Mesopotamian rivers with their swift current was the effective use of steering oars: Tardieu 1990: 101. 27 The conclusion that ship crews were recruited from among soldiers has a significance for a discussion of the size and composition of Julian’s army in his Persian venture. It is fairly generally assumed that it numbered 65,000 men (Zos. III.13.1; cf. Paschoud 1979: 110 f., 186). Yet this figure completely ignores the presence of ship crews, which must have accounted for almost a third of the entire army. Considering what has been said above, it seems justified to presume that the overall figure included the crews. This is confirmed by the fact that not a single historical source details such crews in a separate statistic (cf. Zos. III.13.1-2); their silence in this matter is perfectly understandable.

238

EDWARD DĄBROWA: NAVAL OPERATIONS DURING PERSIAN EXPEDITION OF EMPEROR JULIAN (363 AD) offer no clues as to any mishaps during the journey.37 The sole mention concerns an unusual predicament: the sinking of several transport craft due to a sudden rise in water level in the Euphrates caused probably by deliberate destruction of riverside engineering installations by the Persians.38 Sailing down the river, the fleet was protected by forces marching along on both banks of the Euphrates.39 Although no indication is available as to any division of the fleet into smaller sections, we may surmise that it included squadrons that were separate entities. Such a division contributed to a much more efficient cooperation between crews. This was of especial importance when naval support had to be given to ground troops.40

The large number of transport craft in the fleet indicates that its chief task was to carry sufficient food supplies to last the army a long time.28 Thanks to this, Julian’s army did not have to rely on uncertain local sources of provisions along its route and could advance swiftly. It was likewise efficacious to have the fleet transport heavy war equipment and materials for its construction and repair. Another advantage consisted in the inclusion in the fleet of pontoons to be used in building temporary bridges. It would be wrong to assume that in the ensuing warfare the fleet played a passive, secondary role. Transports carrying war machines on board were used in fighting as floating artillery batteries and offered efficient support to Roman ground troops against targets on land. As provisions were being consumed, emptied ships were used to transport assault troops.29 The largest among them could accommodate dozens of soldiers.30

To appraise the role of the naval force in Julian’s Persian expedition we must consider all the episodes and events in which it made its presence felt, if only to a modest degree. First among such episodes was an operation against the Persian stronghold on the Euphrates island of Anatha (now Ana).41 The emperor himself took the decision to assault it. The task was to be performed by a thousand lightly armed troops transported to the fortress by ships under cover of darkness. Then the ships surrounded the island in a tight embrace to prevent the defenders’ contact with mainland. Aware of the precariousness of an attack on a well fortified stronghold, the emperor ordered that floating artillery batteries be introduced into action and personally supervised their preparedness. The assault never actually came to pass, for the defenders negotiated a surrender.42 Moving further down the river, the Romans chose not to capture two other powerful Persian fortresses: Thilutha43 and Achaiachala.44 In the first case, the garrison, on seeing the enemy flotilla filing past down the Euphrates, decided against taking any action against the Romans.45

We are told that the fleet included 50 craft described in sources as warships.31 Those must have been intended for other purposes than the transport craft and were differently built. Our sources for Trajan’s expedition permit the conclusion that those ships were intended for special purposes.32 One such purpose could have been to offer the emperor and his entourage a convenient quarter and means of transportation, which, however, it seems that Julian never used for any extended period.33 What we know is that they covered his crossing of the Euphrates.34 It would be rational to expect such units to perform reconnaissance along the Euphrates ahead of the approaching flotilla, suppress smaller points of resistance on the banks and isles, and maintain liaison between commanders of squadrons, detachments, and corps.35 The latter task became vital as the flotilla was stretched over many miles along the river.36 The movement of such a multitude on the river must have constituted a problem in its own right as any false maneuver could lead to a disaster of grave proportions. Apart from one mention, accounts of the naval operation

A pontoon bridge was first deployed by the Romans on enemy territory as they were crossing the river near the city of Baraxmalcha.46 On another occasion, floating bridges proved useful in crossing canals in the vicinity of 37 It must have been skillful navigation indeed if the Roman army reached Dura Europos in two days from leaving Circesium ( Amm. XXIV.1.5) and after another four days attained the isle of Anatha (Ana) on the Euphrates (Amm. XXIV.1.6): Den Boeft et al. 2002: 12-13, 13 f. 38 Amm. XXIV.1.1; Den Boeft et al. 2002: 22 f. 39 Amm. XXIV.1.4: Classis autem licet per flumen ferebatur assiduis flexibus tortuosum, nec residere nec praecurrere sinebatur. Researchers agree that the army marched along the left bank of the Euphrates, for it was there that a road led which was used by the Romans to attack Mesopotamia, cf. Gawlikowski 1988: 89 f. The fragments quoted of Ammianus Marcellinus and Zosimos (III.14.1) suggest that both banks were covered by land forces. Such cover was probably needed only in some sections of the route. 40 Cf. Amm. XXIV.6.4. 41 Lib., Or. 18, 218; Amm. XXIV.1.6; Fontaine 1977: 137 f., note 300; Gawlikowski 1997: 151; Den Boeft et al. 2002: 14 f. 42 Lib., Or. 18, 218; Amm. XXIV.1.6-9; Zos. III.14.3-4. 43 Gawlikowski 1997: 152; Den Boeft et al. 2002: 30. 44 Den Boeft et al. 2002: 32. 45 Lib., Or. 18, 219; Amm. XXIV.2.1-2; Zos. III.15.1-2. 46 Amm. XXIV.2.3; Fontaine 1977: 141, note 313; Den Boeft et al. 2002: 33-34.

28

Cf. Jul., Ep. 98 (402a/Bidez); Lib., Or. 18, 214; 18, 215; Zos. III.13.3; Fest. 28; Greg. Naz., Or. 5, 9; 5, 11. 29 Amm. XXIV.1.6; 6.4. 30 Cf. Amm. XXIV.6.4. 31 The same number of warships, which were a flat-bottomed equivalent of the trireme in size, were commanded by Trajan: Suda, s.v. ȞĮ³Ȣ. The presence of warships in Julian’s fleet did not, as some maintain (cf. Andreotti 1930: 249), result from inclusion in it of craft from a regular Roman flotilla operating on the Euphrates as the Romans never had such a flotilla on that river: Dąbrowa 1997: 109; Saddington 2001: 583 f. Presumably, the ships had been built specifically for that mission. 32 Cf. Suda, s.v. ȞĮ³Ȣ. 33 Apart from Magnus of Carrhae (FGrH, no. 225, F 5 = Paschoud 1979: 242-243 (5) = Malalas 13, 21/Thurn 2000), no source mentions Julian making longer trips on board a ship. 34 Cf. Amm. XXIV.1.8. 35 No record contains the least mention of those ships in action. We are left to pure speculation on their role. 36 Taking as a reference the length of a marching military column that flanked it, the fleet extended over about 10 Roman miles along the Euphrates: Amm. XXIV.1.2-4; Den Boeft et al. 2002: 10.

239

THE LATE ROMAN ARMY IN THE EAST Macepracta.47 The bridges made a signal contribution to a skirmish that ensued on the far bank of a canal when a cavalry detachment swimming across it was unexpectedly attacked by the Persians. It was only through a timely intervention of some lightly armored Roman soldiers who had reached the enemy bank across a pontoon bridge that the cavalry got away with only minor losses.48 Another instance when floating bridges proved helpful was at Maiozamalcha.49 Their deployment enabled the Roman soldiers to take more favorable positions in laying siege to that city.50 As battle was waged on land outside the city walls, while Roman soldiers engaged in a parallel struggle in their ships on the water.51 In subsequent actions as well, the use of floating bridges proved indispensable to a Roman advance in unfavorable terrain crisscrossed with many canals.52 At Coche,53 they rendered important services in crossing an artificial limb of the Naarmalcha canal54 called the Trajan canal.

fought on ships in that battle were awarded naval decorations (corona navalis) by the emperor.57 The struggles near Ctesiphon were a turning point for the fleet as well as for the entire expedition. Despite his victory, Julian abstained from storming the city and decided to change his route. From that point on, it would lead inland, offering hopes for easier procurement of much needed provisions.58 Useless now, the fleet was burned so as not to fall to the enemy.59 Of the entire armada, only some smaller craft were taken along to be used as a floating bridge if necessary.60 Although Julian’s decision seems justified and his arguments convincing, it caused not only mixed responses from his entourage but also criticism from ancient authors. Those of the latter who sympathized with Julian justified his step with the same arguments that he voiced. According to Libanius, destroying the ships had been planned in advance.61 But Christian authors heaped accusations on Julian’s decision. They began by emphasizing the negative consequences of his step for the army’s commissariat, calling it near-suicidal behavior that resulted in great losses and troop demoralization in the face of the adversities that awaited them. In addition, they wasted no time in suggesting that the move was closely bound up with the emperor’s faith in pagan deities and fortune-telling.62 The undisguised enmity of Christian authors toward Julian makes them much less convincing.63

The benefits of having ships at hand were again demonstrated by an episode in Pirisabora. During the fight for the city, the Romans captured a large amount of food stockpiled there which, instead of being wasted, was loaded on the ships and helped replenish the Romans’ shrinking provisions.55 Preparing to strike at the Persian capital, Julian decided to include his fleet in the operation. All larger transport craft were to ferry an assault force across the Tigris. A large number of ships were prepared for this role. For efficient conduct of the operation, they were divided into squadrons, presumably each with its own deployment area. However, the idea of an assault was opposed by the emperor’s staff. Despite such objections, at the emperor’s express command, some craft attempted a landing, encountering fierce resistance from the Persians. Immediately Julian gave further orders and the rest of the vessels joined in the struggle, obtaining success and securing a bridgehead across the river.56 Soldiers who

Libanius’s contention that Julian’s action was planned and calculated is much more credible. Even at the planning stage, the emperor must have learned that the current in the Euphrates and Tigris in their upper and middle course was so strong as to make any upstream navigation practically impossible.64 A further circumstance that could only reaffirm his determination, and the hopelessness of any efforts to the contrary, was the unusually high water mark on the Tigris, caused by a late spring thaw in the Armenian mountains.65 A wait

47

Amm. XXIV.2. 6-7; Zos. III.17.3. Cf. Fontaine 1977: 142, note 315; Den Boeft et al. 2002: 41. 48 Amm. XXIV.2.8; Zos. III.16.1-17, 1. Zosimos has more detail to relate about this event: Fontaine 1977: 145 f., note 326; Den Boeft et al. 2002: 46 ff. 49 Zos. III.17.5; cf. Amm. XXIV.2.22. 50 Amm. XXIV.4.6: Den Boeft et al. 2002: 108. 51 Amm. XXIV.4.9; Fontaine 1977: 164, note 377; Den Boeft et al. 2002: 111-112. 52 Cf. Amm. XXIV.4.31. 53 Lib., Or. 18, 244-247; Amm. XXIV.6.1-2; Magnus, FGrH, no. 225, F 7 = Paschoud 1979: 243 (7) = Malalas 13, 21/Thurn 2000; Zos. III.24.2; Greg. Naz., Or. 5, 10; Sozomenos, HE VI.1.5. The fragment quoted from Ammianus Marcellinus contains mentions of Roman actions which are difficult to understand, see Ridley 1973: 321; Den Boeft et al. 2002: 170 ff. For topography, see Coche: Matthews 1998: 141 ff. 54 The course of the canal and its identification has become the subject of many studies: Dillemann 1961: 153 ff.; Fontaine 1977: 180 f.; Paschoud 1978: 245 ff.; Paschoud 1979: 246 ff.; Gawlikowski 1988: 8687; Matthews 1989: 149 ff.; Den Boeft et al. 2002: 171 f. 55 Matthews 1989: 155 ff. 56 Lib., Or. 17, 21; 18, 250-252; Amm. XXIV.6.4-6; Zos. III.25.1-4; Fest. 28; Sozomenos, HE VI.1.7. For differences between Ammianus Marcellinus’ and Zosimos’ accounts of those struggles, see Andreotti

1930: 267 f., note 145; Ridley 1972: 321; Austin 1979: 158 ff.; Den Boeft et al. 2002: 177-183. 57 Amm. XXIV.6.16. 58 Amm. XXIV.7.1-3; Zos. III.26.2. 59 Lib., Or. 18, 262-264; Amm. XXIV.7.4-5; Zos. III.26.2-3; Sozomenos, HE VI.1.9; Zonaras XIII.5-9. Ammianus Marcellinus’ story contains a marked contradiction resulting from simultaneous presentation of two different attitudes: Austin 1972: 301 ff.; Austin 1979: 97 ff. 60 According to Ammianus, 12 vessels were kept (Zonaras quotes the same number (XIII.13.7), Libanius has 15 (Or. 18, 263), but Zosimos speaks of 18 Roman-type and four Persian-type craft. Cf. Fontaine 1977: 192 f., note 460; Paschoud 1979: 185 f.; Den Boeft et al. 2002: 214, 217-218. 61 Lib., Or. 18, 262. 62 Greg. Naz., Or. 5, 12; Theodoret, HE III.25.1-4; Aug., de Civ. Dei IV.29; V.31. 63 Gregory of Nazianzus devoted two speeches to a criticism of Julian (4 and 5), cf. Moreschini 1975: 416-430. 64 Paschoud 1978: 350 f. and the next note. 65 Lib., Or. 18, 262-263. Typically the Tigris rises with spring flooding in late April and early May, cf. Den Boeft et al. 2002: 183, 213.

240

EDWARD DĄBROWA: NAVAL OPERATIONS DURING PERSIAN EXPEDITION OF EMPEROR JULIAN (363 AD) until the water levels dropped before upstream navigation could be contemplated, in the face of an ever more active enemy, would have been waiting to be surrounded and courting disaster as there was no hope of outside help. The decision to burn his boats was therefore fully justified under the circumstances as the fleet had served its purpose and its further maintenance at all costs would have been useless effort and would have severely restricted Roman mobility on enemy territory.

Bleckmann B. 1992 Die Reichskrise des III. Jahrhunderts in der spätantiken und byzantinischen Geschichtsschreibung. Untersuchungen zu den nachdionischen. Quellen der Chronik des Johannes Zonaras, München. Blockley R.C. 1973 Festus’ source of Julian’s Persian expedition. In CPh 68: 54-55. Blockley R.C. 1992 East Roman Foreign Policy. Formation and Conduct from Diocletian to Anastasius, Leeds. Cameron A.D.E. 1963 An alleged fragment of Eunapius. In CQ 13: 232-236. Chalmers W.R. 1960 Eunapius, Ammianus Marcellinus and Zosimos on Julian’s Persian expedition. In CQ 10: 152-160. Dąbrowa E. 1986 The Frontier in Syria in the First Century AD. In Ph. Freeman, D. Kennedy (ed.), The Defence of the Roman and Byzantine East. Proceedings of a Colloquium held at the University of Sheffield in April 1986, vol. 1, Oxford: 93-108. Dąbrowa E. 1997 The Rivers in the defensive system of Roman Syria (from Augustus to Septimius Severus). In W. Groenman-van Waateringe, B. L. van Beek, W. J. H. Willems, S. L. Wynia (ed.), Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Studies, Oxford: 109-111. Den Boeft et al., 1995 = Den Boeft, J., Drijvers, J. W., Den Hengst, D., Teitler, H.C., Philological and Historical Commentary on Ammianus Marcellinus XXII, Groningen. Den Boeft et al. 1998 = Den Boeft, J., Drijvers, J. W., Den Hengst, D., Teitler, H.C., Philological and Historical Commentary on Ammianus Marcellinus XXIII, Groningen. Den Boeft et al., 2002 = Den Boeft, J., Drijvers, J. W., Den Hengst, D., Teitler, H.C., Philological and Historical Commentary on Ammianus Marcellinus XXIV, Leiden-Boston-Köln. Dillemann L. 1961 Ammien Marcellin et les pays de l’Euphrate et du Tigre. In Syria 38: 87-158. Dodgeon M.H. - Lieu S.N.C. 1991 The Roman Eastern Frontier and the Persian Wars (AD 226-363). A Documentary History, London-New York. Fontaine J. 1968 Introduction. In E. Galletier, Ammien Marcellin, Histoire, t. I: livres XIV-XVI, Paris: 7-55. Fontaine J. 1977 Ammien Marcellin, Histoire, t. IV: livres XXIII-XXV. Commentaire, Paris. Fontaine J. 1996 Ammien Marcellin, Histoire, t. III: livres XX-XXII, Paris. Gawlikowski M. 1988 La route de l’Euphrate d’Isidore à Julien. In P.-L. Gatier, B. Helly, J.-P. Rey-Coquais (eds), Géographie historique au Proche-Orient (Syrie, Phénicie, Arabie, grecques, romaines, byzantines), Paris: 76-98. Gawlikowski M. 1997 L’empereur Julien sur les bords de l’Euphrate. In A. Sérandour (ed.), Des Sumériens aux Romains d’Orient. La perception géographique du monde. Espaces et territoires au Proche-Orient

The facts quoted above lead to the conclusion that the severe judgments on the emperor Julian’s Persian expedition passed by some researchers,66 do not equally apply to all its elements and are unjustified in relation to his naval strategy. It should be recognized with full conviction that this strategy significantly contributed to the course of and effects won in the first stage of the campaign. Efficient in its task of providing transportation, the fleet made it possible for the land force to advance rapidly. It should be remembered that a mere three months had elapsed from the launch of the campaign to the burning of the flotilla and the emperor’s death.67 During that time, the army had not only made the long, strenuous march,68 but had also fought a number of battles and clashes, surmounting many water obstacles big and small. Without an effective support from the fleet, which not only afforded continued supply of provisions but often also actively engaged in combat, the army could not have progressed at the rate it did. The sources for the naval operations leave no doubt that the fleet was well prepared to fulfill its mission and completed it satisfactorily. Bibliography Andreotti R. 1930 L’impresa di Giuliano in Oriente. In Historia 4: 236-273. Austin N.J.E. 1972 Julian at Ctesiphon. A fresh look at Ammianus’ account. In Athenaeum 50: 301-309. Austin N.J.E. 1979 Ammianus on Warfare. An Investigation into Ammianus Military Knowledge, Bruxelles. Arce J.J. 1974 On Festus’ sources for Julian’s Persian expedition. In Athenaeum 52: 340-343. 66

Critics try to demonstrate Julian’s lack of suitable preparedness for the expedition, his poor knowledge of the enemy territory, and indecision. They look for causes of the emperor’s defeat in his religious policies and emotional state. Among the most severe of them is G. Wirth 1978. Not all researchers share this point of view. Cf. Ridley 1973: 326. 67 This much time passed from the moment he left Antioch on Mar. 5, 363 (Jul., Ep. 98). Assuming as the start of the campaign the time he crossed the empire’s border in early April, (cf. Amm. XXIII.5.1; Zos. III.14.2), the period would be shorter still. The emperor was killed on June 26: Lib., Or. 18, 268-274; Amm. XXV.3.6 – 23; Zos. III.28.4-29; Magnus, FGrH, no. 225, F 12-13 = Paschoud 1979: 244 (12-13) = Malalas 13, 21/Thurn 2000. A detailed chronology of Julian’s expedition is given by: Dodgeon/Lieu 1991: 231-237; Den Boeft et al. 1998: xv-xx; Den Boeft et al. 2002: xiii-xxiii. 68 For the army’s route and distances between respective points, see Gawlikowski 1988: 81 ff., 92.

241

THE LATE ROMAN ARMY IN THE EAST ancien. Actes de la table ronde du 16 novembre 1996, Paris: 145-155. Lepper F.A. 1948 The Trajan’s Parthian War, Oxford. Matthews J. 1989Roman Empire of Ammianus, London. Moreschini C. 1975 L’opera e la personalità dell’imperatore Giuliano nelle due ‘Invectivae’ di Gregorio Nazianzeno. In Forma Futuri. Studi in onore del Cardinale Michele Pellegrino, Torino: 416430. Norman A.F. 1957 Magnus in Ammianus, Eunapius, and Zosimos: new evidence. In CQ 7: 129-133. Paschoud F. 1978 Le Naarmalcha: à propos du tracé d’un canal en Mésopotamie moyenne. In Syria 55: 345-359. Paschoud F. 1979 Zosime, Histoire Nouvelle, t.II , 1re partie (Livre III), Paris. Ridley R.T. 1973 Notes on Julian’s Persian expedition (363). In Historia 22: 317-330. Saddington D.B. 2001 The Roman naval presence in the East, the Classis Syriaca and the Roman approach to the Euphrates. In Archäologisches Korrespondenzblatt 31: 581-586. Seager R. 1997 Perceptions of Eastern Policy in Ammianus, Libanius and Julian (337-363). In CQ 47: 253-268. Tardieu M. 1990 Les paysages reliques. Routes et haltes syriennes d’Isidore à Simplicius, Louvain-Paris. Thompson E.A. 1947 The Historical Work of Ammianus Marcellinus, Cambridge. Thurn I. 2000 Ioannis Malalae, Chronographia, recensuit I. Thurn, Berlin-New York. Wirth G. 1978 Julianus Perserkrieg. Kriterien einer Katastrophe. In R. Klein (ed.), Julian Apostata, Darmstadt: 455-507.

242

‘Amr Ibn ‘AdƯ, Mavia, the Phylarchs and the Late Roman Army: Peace and War in the Near East Ariel S. Lewin Università deglu Studi della Basilicata, Potenza

the Arabian desert. The army was deployed both at the periphery of the provinces and in the interior.5

Introduction Several studies have rightly stressed that the late Roman army was deployed to an unparalleled degree in the most marginal areas of the Near East.1

Isaac, however, admits that in late antiquity the Arab tribes had become an important new element on the Near Eastern scene. In fact, since that time it is possible to detect the activity of some powerful confederations, acting in turn as allies of one or the other of the superpowers, the Romans or the Persians. It is in this changed context that the same scholar admits the existence of a new dynamic between forts and roads. Isaac states: “At this period there were many military systems in which the roads were constructed specifically to connect their bases - in other words, systems in which the roads were a secondary element, developed to serve the military structure”.6 Isaac would explain the presence of the army in the marginal areas, however, with the expansion of the settled area.7

Dissension has arisen, however, among the scholars about the meaning of this military presence. Some scholars have argued that the soldiers scattered in forts along the fringe of the steppe were part of a system of defence against Arab tribes; these focus attention on the seasonal migrations of the bedouins and the possibility that the army was deployed in order to monitor their movements and consequently to prevent any damage to the agricultural lands. They contend that Diocletian’s activity was a response to an increased threat from nomadic Arab tribes from the North Arabian desert.2 Moreover, M. Gichon in several studies has argued that in the Negev a military system, the limes Palaestinae, was organized in order to cope with occasional infiltrations by bedouins from Sinai.3

The idea that defensive aims were not primary aim of such a presence has been recently advanced by other scholars. At least as far as specific areas are concerned, we are invited to consider the socio-economic framework of the forts and to evaluate their presence in the context of settlement patterns and as a factor in the exploitation of the territory.8

A new wave of scholars has introduced fresh perspective: they encourage us to evaluate the presence of the army as a complex element in Roman provincial life. Ben Isaac in a convincing paper4 restates what he had previously written in earlier works, that is, military units deployed along desert roads provided road security. For readers of his previous works, this is a confirmation of what he had already argued, that is, the roads predated the forts. Military units were deployed in order to keep roads safe and to permit commerce and travel. Roads running on the fringe of agricultural land did not represent a barrier. There was no system organized to defend against an external enemy. The bedouins lived both outside of and within the provinces. The problems they caused were on the scale of the normal banditry. Consequently, it would be completely misleading to envisage the existence of a defensive line organized in order to block attacks from

According to yet another view the fact that Diocletian erected a series of military structures does not have anything to do with a presumed Arab threat. Such buildings were the result of “Diocletian’s determination to restore the Roman east”.9 On the other hand Fergus Millar has rightly stressed that at the time of Diocletian the Saracens had become “a strategic threat”.10 Finally, a complex and fruitful theory has been put forth by Bill Leadbetter: he stresses the broader military importance of the Tetrarchic apparatus and emphasizes 5

Isaac 1998b: 444-460. See already Isaac 1984 = 1998: 122-158; 1990. For similar views see also Mayerson 1986 = 1994: 271-283; 1989 = 1994: 313-321. See also Graf 1989; Kennedy 1992; Young 2001: 130133. 6 Isaac 1998b: 459. But see also his discussion of the problem ibidem: 450. 7 Isaac 1998b: 459. On the extension of cultivation in the steppe zone and the presence of military structures designed against Arab attacks see Millar 1993: 186. 8 Graf 1997: 271-282; Fiema 1995: 261-269; Findlater 2002: 137-152. 9 Eadie 1996: 79. 10 Millar 1993: 433-444. At 222 he acutely affirms that “How we should see the tripartite relations of the Roman forces, the settled population and the ‘Arabs’ or ‘Saracens’ is precisely the most complex question raised by the military history of the Near East”.

I am very grateful to several scholars who helped me in the preparation of the present paper, in particular to Noel Lenski and Maurice Sartre who have read it in its entirety. Noel Lenski deserves special thanks for having greatly improved my English. Eran Argov, Glen Bowersock, Leah Di Segni, Denis Feissel, Robert Hoyland, Hartmut Leppin and Jagoda Luzzatto supplied information on specific issues. Jason Moralee very kindly let me read and quote his forthcoming paper. In no way they are responsible for any remaining mistakes. 1 For an overview see Lewin 2002. 2 Parker 1986: 6-10; 136-143; 2006. 3 Gichon 1967; 1971; 1991; 1997; 2002. 4 Isaac 1998b.

243

THE LATE ROMAN ARMY IN THE EAST the great meaning of the activity of Diocletian and Galerius in supporting international commerces and rivitalizing the Empire’s infrastructures.11

John Malalas affirms that Diocletian greatly reinforced the defences of the empire in the Near East; a famous passage is generally seen by scholars as hinting at the construction of the strata diocletiana. It is also interesting to note that according to the same source the emperor built an ammunition factory at Damascus in order to supply the army against the incursions of the Saracens.15 Moreover, in another confused passage Malalas affirms that Constantius Chlorus sent Maximus Licinianus with a strong army to protect the territories of the East against the Persians and Saracen incursions. These last were causing trouble as far away as Egypt.16 It is probable that the emperor Malalas describes is Maximinus Daza whose involvment in wars in the Near East is known. Alternativelly, he could have had in mind Licinius who campaigned in 313-314 against the Persians in Cappadocia, Adiabene, Media and Armenia.17

I do not have any serious objections to the opinions of the scholars who have argued that during the first centuries of the empire Arab tribes did not represent a serious threat and that the Roman army was essentially employed as a police force which controlled the provincial territory and performed administrative tasks. Nonetheless I intend to argue that in late antiquity the presence of the army took on a new character. As a result, we would assume that although the late Roman army was deployed in marginal areas in order to protect economic activities, such as agriculture, mining and traffic along commercial routes, to cope with banditism, to settle conflicts between pastoralists and sedentarists, nonetheless it was also used as a defensive system aiming to intercept potential attacks by the Saracen tribes, especially those brought about by the ones allied with the Persians. Consequently, we must posit a flexibility and the multifunctionality of the late Roman army. Examples will be used to elucidate some features of relations on the Near Eastern frontiers between the Arab tribes and the Roman empire. Moreover, we will discover that the integration of these tribes was not a minor element in the development of the periphery of the empire.

Let us now look to another kind of literary sources which points to the existence of an important Arab confederation led by a powerful leader who was a precious ally of the Persians. The inscription of Paikuli lists the Arab king Amru, the Lakhmid, among the approximately 40 kings and dignitaries subordinated to the authority of Narses (292302).18

1. Two literary sources inform us about the existence of wars between Romans and Arabs at the time of Diocletian. A Latin panegyric extols the emperor for a victory he had achieved against the Saracens near the boundaries of Syria. The event is firmly dated to the summer of 290.12 This presumed victory was not officially recognized in the imperial titulature and consequently we might doubt that the episode was an important one. Other considerations, however, indicate a greater importance than has sometimes been acknowledged. At the beginning of April, Diocletian was still in Byzantium, in May his presence was attested in Syria, at Antioch, Emesa and Laodicea, but on July 1 he was in the Danube area, at Sirmium.13 Consequently, the emperor spent only a very few weeks in the Near East. We do not know if he came to fight against the Arabs or if he came to Syria only in order to settle a treaty after a war which had been fought by a Roman general in the area. Nonetheless it is impossible to imagine that he submitted himself to the stress of this journey in order to deal only with some local banditry. It seems wiser to argue that after a war against the Saracens he intended to ratify an important treaty with influential tribes; in so doing he intended to guarantee peace on the frontier and to furnish the empire with precious allies against the Persians.14

The presence of Amru, the Lakhmid, in the sphere of influence of the Persian empire is also proven by a Coptic Manichaean text which informs us that a teacher of this sect, Abiesou, after having been received with full honours by Zenobia, sent missionaries to a place called “the tower of Abiran”. There the Manichaeans became famous for the cures they provided to the local population. Amru, after having been informed of their activity became their patron and succeded in convincing the Sasanian king Narses to suspend the persecution of the Manichaeans in his kingdom.19 prestigious victory against the Saracens and that he only stipulated a treaty with them. However the panegyrist appears to stress the attainment of a military victory against the Saracens, which he put on the same level as the victories obtained against Germans, Sarmatians and Franks. See Pan. Lat. XI.5.4; 7.1. He may have amplified the extent of the military success, but he can not have invented it. On the other hand, it remains difficult to explain the absence of the title Arabicus in the official titulature of Diocletian and his colleagues. It is interesting to note that the title Gothicus was left out from the official titulature almost immediately after the imperial victorious campaign against the Goths. For a possible explanation of this see Brennan 1984. 15 Malal. XII.38-40 (307-308). The fabrica at Damascus is listed in ND Or. XI.23: Scutaria et armorum Damasci. Scholars believe that the military factories were organized by Diocletian. See James 1988. 16 Malal. XII.48 (313). 17 See Barnes 1981: 39; 41; 65. See also the discussion on the eastern wars fought in the times of Galerius, Maximinus and Licinius provided by Zuckerman 2002: 630-631. It is probable that the Saracen incursions described by Malalas were connected with these wars. 18 Humbach - Skjaervo: 1983 part. 3,1: 71; part. 3,2 120; Frye 1984: 375-378. 19 Giversen 1986: pl. 99-100 (non vidi); Tardieu 1992: 15-24; Pedersen 1997: 193-201; De Maria 2003: 165-166; 394-396. See also Shahîd 1984: 32 n. 6; Hartmann 2001: 308-315. For the identification of the

11

Leadbetter 2000; 2002. Pan. Lat. XI.4.2; 5.4; 7.1. On the date of the episode see Barnes 1982: 51. 13 On the movements of Diocletian see Barnes 1982: 51-52. 14 Leadbetter 2004: 247 believes that the absence of the title Arabicus in the Tetrarchic titulature proves that Diocletian did not obtain any 12

244

ARIEL S. LEWIN: ‘AMR IBN ‘ADƮ, MAVIA, THE PHYLARCHS AND THE LATE ROMAN ARMY The famous inscription of Imru’l-qais, dated to 328, found at Nemara, a site that had been occupied by the Roman army during the first centuries of the empire, reveals the burial of an important Arab chief at the fringes of the Roman world. Although inscribed in Nabataean letters the text is in an Arabic, a language which did not belong to the north Arabic Safaitic or Thamudenic sub groups. The linguistic group of Imru’lqais and his people was that of the tribes of central and south-central Arabia.20

same writer, ‘Amr ibn ‘AdƯ avenged the death of his uncle and caused the ruin of the queen of Palmyra.22 This piece of information appears not to be entirely correct, nonetheless it can hint at the fact that the Arab king somehow cooperated with Aurelian in order to avenge the killing of his uncle by Zenobia.23 It is worth noting that ‘Amr ibn ‘AdƯ appears already as a player in the near Eastern arena in the time of Aurelian. Moreover, Tabari affirms that he was a vassal of the Persians since the reign of Shapur I. We must conclude that at the time of Narses ‘Amr was an aged man who had already supported several other Persian sovereigns.24

In the inscription Imru’l-qais enumerated a long series of victories he had obtained against several tribes on the Arabian peninsula. The range of his action was remarkable because such tribes lived in areas quite distant one from the other. Scholars have offered different readings and translations of the text, and consequently some major issues remain open to debate. In particular, we must observe that recent studies have suggested a reading of the text capable of proving that the king proclaimed himself independent from the two empires. On the other hand, other scholars remain skeptycal about this reading; moreover, they argue that the king’s burial place was so close to the Roman provincial world that we can not help suspecting that at that time he was an ally of the Romans. According to the Arab sources, Imru’l-qais had once lived in Sasanian territory. If so we can tentatively reconstruct the dynamics of his movements: at first he was an ally of the Sasanians, but at a certain stage he came in conflict with them and launched a series of campaigns in the peninsula. Finally he was buried near Roman territory and logic would indicate that at that time he was an ally of the Romans.21

We cannot be certain in our identification of the Saracens who were defeated by Diocletian in 290 or at least coerced into a treaty of alliance with the Lakhmids led by ‘Amr ibn ‘AdƯ. However, the aforementioned conflict had as its theater the borders of Syria and this fact would seem to overlap the area where the Lakhmids had their range of activity. It is important to remark that this event occurred during the reign of the weak king Vahram II and that, as we have seen, ‘Amr is mentioned in the Paikuli inscription among a series of dignitaries and kings who cooperated with the successor of Vahram II, Narses. If ‘Amr was the chief of the Arabs subjugated by Diocletian he must have transferred his allegiance to the Persians again after little more than a year, as a consequence of the fact that a new and much stronger Sassanian king than Vahram II had appeared on the scene. ‘Amr ibn ‘AdƯ’s alliance with the Romans would have been quite ephemeral. An alternative hypothesis might be that the Saracens mentioned as defeated in the Panegyric belonged to another important Arab group, the Tanukh.25

One important piece of information provided by the inscription is that Imru’l-q-ais describes himself as the son of ‘Amr. Chronology itself and the Arab tradition of a chain of Lakhmid kings stretching from ‘Amr ibn ‘AdƯ to Imru’l-qais, preserved by Tabari, point to the identification of the father of Imru’l-qais with the king mentioned in the Paikuli inscription and in the Manichaean texts.

It is now possible to make some important observations. First of all, the ambitions of the Sassanians and the continuous wars of the third century weakened the Roman frontiers and gave the Arab tribes an opportunity to damage the provinces.26 Moreover, we have detected the existence of a considerable activity by important Arab tribes at the time of Aurelian, in the Tetrarchic age and again during the reign of Constantine. Ambitious chiefs like ‘Amr ibn ‘AdƯ and his son Imru’l-qais distinguished themselves in the conduction of wars and in stipulating alliances with the one or the other of the empires. We could argue that the two empires had learned how to employ Arab tribes in order to damage the interests of their rival and that, at

Moreover, Tabari describes ‘Amr ibn ‘AdƯ as the nephew of Jadhima, the king of Tanukh who having entered into conflict with Zenobia was killed by her. According to the Amru of the Coptic text with the Lakhmid king see also Shahîd 2000: 124 n. 122 who rejects an alternative opinion argued by de Blois 1995. 20 On the character, use and diffusion of ancient Arabic, attested for the first time at al-Faw, most probably at the end of the Ith century BC, see Macdonald 2000: 48-59. 21 For different views about the nature of his relations with the two empires see Shahîd 1984: 31-53; Bellamy 1985: 31-48; Kropp 1993: 6393; Robin: 1997: 265-269. Imrul’q-ais’ place of burial, so close to the provincial Roman world, does not seem to fit with the idea that he was an independent king. See MacAdam 1980; Bowersock 1983: 139-140. It is possible to suspect that eventual claims of independence from the two empires expressed in the inscription were other than the official position Imrul’q-ais detained at that time. The modern bibliography on the subject is discussed by Shahîd 2000; Gnoli 2005: 518-523. This last scholar is sceptycal about the idea that Imru’l-qais after having left HҝƯra and the Sasanian alliance became a Roman ally.

22 Against the sceptycism by Millar 1993: 433 on the reliability of Arab sources see Bowersock 1983: 133-134; Hartmann 2001: 336-344. See also Sartre 1982: 134-136. 23 See the important observations by Bowersock 1983: 134-137. 24 See also Caskel 1966: 370-371; Sartre 1982: 137 who notes that, according to al-Kalbi, ‘Amr was already active as the chief of HҝƯra at the time of Bahram II (273-276); Hartmann 2001: 346-348. 25 On the Tanukh see Sartre 1982: 134-139; Shahîd 1984 366-381. 26 See Sartre 1982: 132; Isaac 1998: 451; Parker 2006: 538-541.

245

THE LATE ROMAN ARMY IN THE EAST the same time, the Arab tribes were beginning to perceive the gains they could obtain from this highly contentious situation. The chiefs could have strived to negotiate profitable treaties in order to advance their position within their own society. Consequently the Arab raids would have become part of a greater struggle between Rome and Persia. All these are quite reasonable suppositions, but we must proceed to the times of Constantius II before we find a literary source capable of proving the existence of a series of Arab incursions and a system of diplomatic activity undertaken by the empires in order to shape alliances with the Arab tribes and to damage the interest of the enemy. The future emperor Julian in an oration addressed to this ruler, recalls that at the beginning of Constantius’ reign, at the time of a Persian incursion into Mesopotamia, the Arab tribes were plundering the Near Eastern provinces. Constantius thus found himself in the same situation Diocletian had experienced almost fifty years before: he was in the Balkans at that time but quickly moved into Syria. His appearance on the scene of the conflict quelled the conflict and several years after the event Julian memorialized this by praising the emperor for his effective diplomatic activity “Your embassies have turned the Arab bandits against your enemies…”27

represented an important step in the process of Christianization of the Arab tribes.30 The military aspect of the revolt was, however, secondary to the primarily religious focus of their interests. Even so, we must be grateful to these ecclesiastical historians for having provided information about an event that otherwise would have gone ignored, for it is not narrated by Ammianus or Zosimus. Sozomen is our best and most detailed source regarding the episode: he reports that after the death of a king of the Saracens the foedus between the Arab tribe and the Romans was broken. Mavia, who succeeded her husband as chief of the tribe, attacked the Roman empire and devastated several towns in Phoenice and Palaestina up to Egypt.31 Sozomen emphasizes that it was not a negligible event, even though conducted under the leadership of a woman. Sozomen’s narrative is noteworthy, for by describing a major battle that occurred between the imperial army and Mavia’s men he discloses an extraordinary chapter in late Roman military history. In response to the incursion of Mavia’s tribe the dux of Phoenice summoned the intervention of the magister equitum et peditum per orientem. When the magister approached the enemy, he ordered the dux to step aside from the battle because he intended to fight the Saracens only with his soldiers alone, that is, the comitatenses. But the facts revealed that he had underestimated the level of the danger, for the battle turned out quite unsuccesfully for him and he was nearly defeated in humiliating fashion. Ultimately this powerful general and his men were saved by the intervention of the spurned dux, whom in this second mention of him Sozomenos calls now ZÄm ÙcelËma t´m Ñm VoimÊj\ jaà PakaistÊm\ stqatiyt´m. The most obvious interpretation of this appellative is that the dux in addition to the soldiers he normally had at his disposition, the limitanei of the Phoenice, had also assembled some units from the neighbouring duchy of Palaestina.32 He had realized that it would be a serious mistake to remain outside of combat and thus by fighting the barbarians offered an escape route to the magister. The dux managed to protect the retreat of the army of the magister with a dense volley of arrows.

We can argue that the Romans exercised with the Arab chiefs the same art of clients management a scholar has detected in the case of the relations with western barbarians, where a fluid system of alliances and payments was continuously reshaped through diplomacy and wars.28 2. The second topic I intend to deal with is dated to the end of the reign of Valens. Ecclesiastical historians inform us about the uprising of a powerful Arab tribe led by a woman, Mavia. According to them, she caused serious problems for the Roman authorities.29 The story was important for these historians because at the end of the conflict a treaty was arranged and Mavia’s tribe requested a holy man, Moses as its bishop. He was an orthodox who refused to be consecrated by the Arian metropolitan of Alexandria but who finally received consecration by the orthodox bishops who were still in exile. This was an episode of great importance in the conflict between Orthodoxs and Arians and was thus worthy recounting by ecclesiastical historians because it

The battle constituted an epic event: Sozomen affirms that even up to his own times the inhabitants of the area 27

Jul. Or. I 19a; 21b. See Tantillo 1997: 254-255 with discussion of Constantius’ movements at 429-446. The intervention of Constantius in Syria is firmly dated: on October 11th 338 he was at Antioch, at October 28th at Emesa, near the place of the events, on December 27th he is again attested in Antioch. 28 Heather 2001. 29 Rufin. HE XI.6; Socr. HE IV.36; Soz. HE VI.38; Theod. HE IV.23; Bowersock 1977; Sartre 1982: 140-144; Shahîd 1984: 138-202; Lenski 2002: 204-209; Roberto 2003; Lewin 2004. See also Thelamon 1991: 128-147. For a minimalist view Mayerson 1980 = 1994: 164-172; Schmitt 2004.

30 However according to Schulz 1993: 100 the appointment of a bishop to the Saracens of Mavia was not a provision included in the treaty. 31 Soz. HE VI.38. Rufin. HE XI.6 says that: Mavia...Palaestini et Arabici limitis oppida atque urbes quatere vicinasque simul vastare provincias coepit. 32 So Shahîd 1984: 150 giving one example of limitanean troops crossing provincial borders in order to help the dux of a different province. On the strategy of gathering several units in order to fight an Arab enemy see Mayerson: 1986: 40 = 1994: 276. For fuller discussion of the passage see my appendix.

246

ARIEL S. LEWIN: ‘AMR IBN ‘ADƮ, MAVIA, THE PHYLARCHS AND THE LATE ROMAN ARMY where the battle had occurred still remembered the event. The Saracens for their part celebrated it in poems (£da²r).

Recently, however, the identification of the arrogant general with Iulius has been challenged by David Woods. This scholar prefers to have Victor as the general who almost suffered an humiliating defeat by Mavia and Iulius as the valiant man who saved the situation. In other words he argues that the arrogant magister of the east with authority over equites and pedites was the magister praesentalis while the strategos of Phoenice and Palaestina must be identified with a magister militum per orientem.37 This new interpretation must be rejected. First of all, it is impossible that in calling a person strategos of the soldiers of Phoenice or strategos of the soldiers of Phoenice and Palaestina Sozomen could refer to a magister militum per orientem. In fact the magister militum per orientem maintained authority over the comitatenses of the Near Eastern provinces and should not be confused with the dux, a provincial commander. It is much more logical to argue that Sozomen’s shortening of his written source slightly obscures the sequence of events and the manner in which the dux Phoenices intervened. We can imagine that he found in his source a more detailed and longer version that made explicit that the dux who had authority over the soldiers of Phoenice first asked for the intervention of the magister then started collecting under his command the limitanei of his province and some military units from the neighboring province of Palaestina as well.38

It is worth stressing the quality of our source: Sozomen was an accurate and ambitious historian. Moreover, as a native of Palaestina he was particularly informed about facts regarding his province and most probably relied on Gelasius of Caesarea, another Palestinian historian, whom he supplemented with other sources.33 His narration of Mavia’s revolt is fulsome and sufficiently detailed and we must assume that it is both convincing, and based on a sound knowledge of the event. Good evidence of Sozomen’s veracity is easily adduced. He describes the anonymous arrogant general as a stqatecÄr pÇsgr t±r ÐmÀ tÂm †Ey Úppij±r te jaà pef±r stqati°r. An inscription from Umm el Jimal dated to 371 reveals that the magister peditum et equitum per orientem Iulius had ordered the construction of a military structure there (burgus, dispositione Iuli viri clarissimi comitis magistri equitum et peditum).34 Sozomen thus provided the official position of the arrogant general in its precise, formal terms.35 Iulius maintained his command after the battle of Hadrianopolis. Consequently, it is best to assume that he was the arrogant commander who had disdainfully refused the help of the limitanei.36

More decisive still, it is impossible to assume that the arrogant magister can possibly be identified with Victor, who, as we will see, participated in the events only at a later stage as an imperial ambassador. While Sozomen describes the arrogant general as a magister peditum et equitum per orientem, at the time of his recorded intervention Victor held the position of magister equitum prasentalis.39

33 On Sozomen’s personal agenda and his interest in Palaestina see Argov 2006. In general see Urbainczyk 1997; Stevenson 2002/2003; Leppin 2003; Motta 2005. Schmitt 2004: 864; Van Nuffelen 2004 490 argue that the Arab poems mentioned by Sozomenos represent the source of this writer for the description of Mavia’s episode. This idea must be rejected. Robert Hoyland, in a personal communication, notes that “even accepting that Arabic poems in the fourth century would be similar to those in the sixth (no way of actually demonstrating this) and accepting that those from the sixth have been reliably transmitted (most granted these days) then I still have to say that such poetry would not make an easy source, since it is declamatory. It is addressed to an audience that is already familiar with the events and so only needs to be able to allude to specific details to elicit a feeling (usually pride in a shared victory). That means it could give a detail like a title (though it might get corrupted), but would not provide a sustained continuous narrative. Preislamic poetry tends to be portrayed by later writers (eighth - ninth century) as existing in a fully oral environment, so that would also pose the question of how it got to a Roman author).” For an introduction to poetry in Arabia before Islam see Hoyland 2001. On the contrary, we must assume as other scholars have stressed that Gelasius was the common source for the ecclesiastical historians account of Mavia’s story. See Bowersock 1977. Sartre 1982: 140 argues that Rufinus was the main source for Socrates and Sozomen and that they supplemented his narrative with other literary sources. For an eccentric idea see Van Nuffelen 2002 who thinks that the Historia Ecclesiastica of Gelasius of Cesarea was a forgery written in the second part of the Vth century. But see on it Lewin 2004: 234 n. 52. On the importance of Gelasius see Winckelmann 1966; Van Deun 2003. 34 ILS 773. 35 For some examples on the magister militum per orientem called in the sources stqatgcËr or stqatgkÇtgr either èy or Ðmatok±r see Cameron 1993 225-226, n. 127. 36 Such an identification is not registered in PLRE. On Iulius see Demandt 1970: 710-711. For Iulius as the general almost defeated by Mavia see Shahîd 1984: 151; Lenski 2002: 206; Roberto 2003: 73 n. 13; Lewin 2004: 235.

As I have already argued elsewhere, Sozomen’s narrative is of the utmost importance for military history. It reveals how the late Roman army functioned in the case of an attack initiated by Saracen tribes. The limitanei had the task of watching over the movements of the Arab tribes. When a major attack occurred, they were to ask for assistance from the comitatenses. Even so, the limitanei were themselves supposed to participate in battles;40 this 37 Woods 2004: 732-733 changing the interpretation he gave in Woods 1996: 68; 1998. 38 See my discussion at length in the appendix. 39 Feissel 2002: 207 n. 34 notes that Ammianus and Zosimus describe Victor as magister equitum and it is only Sozomen who states that he was a magister per orientem equitum et peditum. Obviously, the discrepancy must be settled by assuming that Sozomen is describing the position held by Iulius. Thelamon 1991: 135 identifies the arrogant general with Victor. It is worth stressing that Victor had been magister peditum during the reigns of Julian and Jovian and that Valens appointed him to the position of magister equitum. At the time of the battle against Mavia the magister peditum militum praesentalis was Arintheus. See Demandt 1970: 705-709; Roberto 2003: 67-71. Moreover, it is worth stressing that most probably at the time of the battle where the arrogant general was almost defeated by Mavia Victor was busy with his embassy to Persia. See Amm. XXXI.7.1. 40 Lewin 2004: 235-236.

247

THE LATE ROMAN ARMY IN THE EAST confirms that at that time they were still far from becoming a decaying, part time army of peasantssoldiers.41

marriage was surely dictated by diplomatic considerations. We can posit that the weddings were surely authorized if not imposed by the emperor. On his part, Victor with this marriage was increasing his already very strong position, becoming the key element in the political arena of the Near East.46

At least two points in the narrative provided by the ecclesiastical historians lead us to conclude that Mavia’s uprising was indeed very serious, was conducted by a powerful tribe, or a confederation of tribes, and was not a routine episode of normal banditry. First and obviously enough, is the fact that the dux asked for the intervention of the magister militum and the latter was almost annihilated along with his men. We must note here the intensity of the danger.42

We must now discuss another important issue: how long did the revolt last? Scholars who deny any serious importance to the event emphasize that Socrates claims it started only after Valens had left Antioch, that is, in the spring of 378. Only a few weeks later, however, a group of Mavia’s Saracens is mentioned by Ammianus as having been recently enlisted (recens illuc accersitus) and having played a key role in defending Constantinople from siege by the Goths and their barbarian allies. By this interpretation, the revolt would not have lasted more than a very few weeks and the Saracens would not have had any chance to affect seriously the tranquillity of the Roman eastern provinces.47

Second Socrates mentions another general as somehow connected with the event when he reports that at the end of the conflict the magister equitum praesentalis Victor married the daughter of Mavia.43 It is logical to assume that Victor was the person who led the Roman embassy which discussed the treaty terms with Mavia. A short time earlier, he had been sent as ambassador to the Persians for the second time in a few months. He was a very experienced man, on whom the emperor relied on for difficult peace negotiations. He was active as ambassador in the eastern arena at that time and surely he also received instructions from Valens to negotiate a peace with the Saracens.44 His participation in the peace negotiations does not imply that he was the arrogant magister who had fought with Mavia.

A different scenario has been argued, correctly in my opinion, by Umberto Roberto. He noted that the beginning of the revolt must be dated to a first departure of the emperor from Antioch after Easter of 377. This departure, mentioned by Socr. HE IV.36.1 must not be confused with the definitive one from Antioch to Constantinople in March 378. Accordingly, some imperial constitutions show that Valens was at Hierapolis from the spring to the autumn of that year. The confrontation between Moses and the Arian bishop can not have happened later than the last weeks of the same year 377. In fact, the orthodox bishops were recalled by Valens from their places of exile to their seats in December.48 If so, there is time for the revolt to have lasted for several months, from the spring of 377 to the very last weeks of the same year. This is a wide enough span of time allowing the revolt to have acquired a certain prominence without however having lasted too a long time.

A case which occurred in 453 offers a significant parallel example of the modes in which late Roman diplomacy operated. The magister militum per orientem Ardabur after having fought against the Saracens near Damascus initiated peace negotiations with them. In a second embassy, Priscus came with the strategos Maximinus to the spot. We can assume that they concluded the discussions and struck the treaty. After these negotiations Maximinus, again together with Priscus, went to Egypt in order to discuss the terms of the peace with the Blemmyes and the Nobadae.45 It is reasonable to assume that Maximinus was a very influential person, an expert ambassador and possibly a magister praesentalis who supervised the diplomatic strategy of the empire. I argue that the same obtains for Victor.

Mavia’s uprising occurred at a critical moment for the empire: the Gothic revolt was undermining the Roman 46 See the case mentioned in an inscription from Anasartha that must be dated to the fifth century where the lamprotatos phylarchos Silvanus gave his daughter in marriage to another phylarch after a permission by the imperial authority. See Feissel 2002: 210-220. If the imperial authority controlled such matrimonies there would have been an even greater necessity for Valens to authorize the marriage between Victor and Mavia’s daughter. 47 Amm. XXXI.16.5-6; see also Zos. IV.22.1-3; Eun. Hist. fr. 42; Socr. HE V.1.1. Schmitt 2004 stresses the short time lapse between Valens’ departure from Antioch in Mai 278 and the first attested presence of the Saracens in Constantinople before the battle of Adrianople. It must be observed that the Saracens valiantly defended the capital at two different moments, the first before the battle of Adrianople, and the second one in the aftermath of it. See Shahîd 1984 179-181; Lenski 2002 335-336 who rightly defends Shahîd’s interpretation of the sources against Woods 1996 who tries to reduce the defence of Constantinople by the Saracens to only one episode, after the battle of Adrianopolis. 48 Roberto 2003: 77-79. See also the important observations by Lenski 2002: 208-209 who through a different path arrives to the same conclusions offered by Roberto on the timing of the uprising.

The fact that a very high level official of the Roman empire such as Victor, second in importance only to the emperor himself, agreed to marry Mavia’s daughter confirms the importance of the tribe of which she was the leader. It is doubtful that he decided to marry the daughter of an insignificant Arab tribe. Furthermore, we can exclude the existence of a sincere love story between the old Roman general and the young Arab princess. The 41

Isaac 1988 = 1998 345-387; Schmitt in the present volume. See also Shahîd 1984: 142-152; Lenski 2002: 205-206; Lewin 2004: 234-236. 43 Socr. HE IV.36.12. 44 See the detailed discussion in Roberto 2003. 45 Prisc. fr. 26; 27. 42

248

ARIEL S. LEWIN: ‘AMR IBN ‘ADƮ, MAVIA, THE PHYLARCHS AND THE LATE ROMAN ARMY position along the Danube and Valens needed to leave the Near East and take a large part of the army with him.

bypassed Damascus, was approaching the Jordan valley through one of the main east-west routes. From north to south, the three main routes are the one leading to Paneas, the route to Nawa and Iulias and the axis leading through Adraha to Capitolias and Gadara.51

Mavia came to know of the imperial difficulties and decided that it was time to renegotiate the terms of the treaty between the Romans and her people. Moreover, she perceived that her becoming the leader of the revolt gave her the chance to redifine her own authority among the tribes. 49

We may exlude this last because Adraha lies well within the interior of Arabia and consequently we could hardly explain the absence of limitanei from this province among the forces collected by the dux Phoenices in order to fight Mavia. Also Nawa is in Arabia, but in this case Mavia would have approached this province for only a short way before penetrating Palaestina. We remain then with the first two possibilities: Mavia took the northern route to Panias or, alternatively, the central one leading to the basin of the river Yarmouk. If so, the confrontation between Mavia and Iulius occurred in the same area where in the seventh century the Arab invaders were victorious in the battle that sealed the destiny of the Near East.52

But let us now return to Sozomen’s description of the event. As we have observed, he is the only ecclesiastical historian to offer a relatively detailed account of the revolt. Consequently, it is worth reading the informations he provides with particular attention. At the beginning, while introducing the background of the uprising, he recalls in general terms how after the death of the king of the Saracens, his widow, Mavia, led the ethnos to devastate the towns of Phoenice, Palaestina up to Egypt. After this introduction, he narrates the events of the famous battle explaining that, as a result of it, the conflict became more serious. Finally, the situation compelled the Romans to call for peace. ‘Epibaqo³mtor dÁ to³ pokÈlou, Ðmacja²om ÑdËjei peqà eÓqÉmgr pqesbeÌsashai pqÄr LauÊam.

The treaty settled with the powerful tribe led by Mavia had very important results. As we have seen, a small group of Saracen allied forces was capable of saving Constantinople through their skill and courage from siege by the Goths and bands of Alans and Huns. But it is only Socrates who clarifies that this group of Saracens had been recently dispatched to Constantinople by Mavia.

This kind of reconstruction of the phases of the uprising allows us to surmise that the famous battle was the first episode of the military campaign. Mavia invaded the duchy of Phoenice and, as we have seen, the dux Phoenices supplemented his forces with some units of limitanei from Palaestina in order to fight her. If so, Sozomen’s description offers clues about the location of the famous battle: Mavia and her army apparently entered into Phoenice and intended to continue their incursion along an axis of penetration leading to Palaestina.50 Further details are most likely lost forever. Nonetheless we may at least establish in broader terms the area that was affected by the battle. We could reasonably suppose that the comitatenses arrived from the route AntiochTyre-Paneas while the army of Mavia, after having

Nevertheless a serious question awaits resolution: if Mavia’s uprising was not a minor event, why did Ammianus not narrate it in his res gestae? The reasons behind his silence are puzzling, nonetheless some explanations can be posited.53 Irfan Shahîd gives us an ideological explanation. The writer did not intend to admit that the treaty between Victor and Mavia played a role in securing Constantinople itself. A beneficial alliance between Christians, whom he disliked, and Arabs whom he clearly hated, was for Ammianus inconceivable.54 This explanation could be right.55

49 See Heather 2001 whose model for the relations between the Roman government and the barbarian chiefs on the european frontiers in late antiquity can be applied to our case. 50 See also Thelamon: 134-135. A different reconstruction of Mavia’s movements is advanced by Sartre: 1982: 141-142 who argues that Mavia held sway over tribes that lived near to the Sinai. If so, she would have started her uprise from the the border between Egypt and Palaestina. Sartre stresses that Rufin. HE XI.6 affirms that Mavia’s tribe lived close to the place where the holy man Moses used to heal people and that Theod. HE IV.23 says that Moses lived at the border of Palaestina and Egypt. I prefer to argue that Rufinus’ statement must not be taken literally and that Mavia’s confederation was capable of a large range of movement on the border of the eastern provinces. See also Gnoli 2005. According to Shahîd 1984: 140-141; 196-197 Mavia was the widow of a king of the Tanukhs/Lakhmids, but she originated from another tribe, the Kalb, which was encamped in the steppes between Palmyra and Tabuk. If so, the range of her tribe’s activity could fit with Rufinus’ description. According to Rufin. HE XI.6 the towns of Palaestina and Arabia were disturbed by the revolt and the neighbouring provinces were affected as well. Socrates HE IV.36.2 probably too emphatycally says that the entire Near East was devastated at that time.

51

On the three main east-west roads see Roll 2002. For the location of the battle of the year 636 at the river Jarmuk see Kaegi: 1992: 112-144. For a different view Woods in the present volume. 53 See Woods 2002: 128: “Strangely, Ammianus does not mention Mavia anywhere in his history, despite the fact that she fought a borderwar against the Romans c. AD 377 during which she inflicted at least one serious defeat upon them, so that it is tempting to believe that this omission is deliberate rather than a result of any lack of information.” 54 Shahîd 1984: 263-267. 55 For the very bad opinion Ammianus had on the Arabs see his description in Amm. XIV.4.1. They are pungently branded as a natio perniciosa at XIV.4.7. Quite probably his hatred towards them was exacerbated by the fact that his hero, Julian, had likely been killed by an Arab. See Bowersock 1978: 116-117. On the famous description provided by Ammianus of the Saracen warrior sucking blood from a Goth in the battle before the walls of Constantinople in 378 see Woods 2002. 52

249

THE LATE ROMAN ARMY IN THE EAST Moreover, problems related to the organization of material in his work could have compelled Ammianus to neglect the episode. Perhaps he was reflecting on the fact that he had already dealt with the Saracens in too many passages of the res gestae. In fact, Ammianus reveals that in his lost books he had already dedicated some pages to the Saracens, in his account of Marcus Aurelius and also elsewhere (in actibus principis Marci, et postea aliquotiens memini rettulisse, 14.4.2). In another passage, Ammianus recalls that he had already several times described the origin and the customs of the Saracens in other parts of his work (diversis in locis rettulimus plura, 31.16.5). In a lost passage Ammianus narrated that at the time of Costantius II the town of Thelsea - Dumeir on the fringes of the desert in Phoenice had been devastated due to the indolence of the dux Phoenices, Serenianus.56 It is probable that in this case as well Saracen tribes were behind the devastation.

orientales magnis.58

provinciae

discriminibus

ereptae

sunt

Iulius is exalted here as a true Roman general, acting with readiness and with the necessary severity. He manifested the traditional virtus because he was a Roman by birth. At that time, so Ammianus tells us, Romans still maintained leading positions in the army, as opposed to what was going on as he composed the res gestae when barbarians occupied the highest positions. The extermination of the Goths is the last episode described in the res gestae and consequently Iulius’ harsh decisiveness is offered as an example. Iulius was the champion of romanitas.59 With such considerations in mind, we can now understand why Ammianus chose not to describe the revolt of Mavia. He wanted to hide an episode that was controversial and probably embarrassing in the career of one of his favorite generals, because he had almost been defeated by a woman.

We might argue that the historian felt that he had the task of offering an agreeable and varied selection of materials, in order not to annoy his readers. Therefore, he decided to avoid the description of other events in which the Saracens were involved.

By contrast, Ammianus openly berated Serenianus, the commander whose ignavia led to the destruction of Thelsea. At XXVI.10.1 the writer tells that this man was killed during the usurping of Procopius. Cuius mors saluti plurimis fuit. Nam si victoriae superfuisset incultis moribus homo, et nocendi acerbitate conflagrans ... multas innocentium ediderat strages.

Another literary reason for neglecting Mavia’s uprising was the fact that he had plenty of material to fill the last books of his work. We do not have to think that he described all the major facts that occurred in the last years of Valens. He did not mention an important Isaurian uprising of 375 either; around the same time a Gothic civil war probably occurred, but Ammianus is silent about it.57 Mavia’s war, although not a minor event, lasted only some months and did not change dramatically the turn of history.

Ammianus astutely selected his material. He hated Serenianus and consequently inserted in his res gestae the episode about the destruction of Thelsea caused by his indolence. By contrast, he admired Iulius and decided to neglect the narration of the revolt led by Mavia who almost defeated him. To conclude: various considerations invite us to argue that Mavia’s revolt was not a minor episode and indeed the ecclesiastical historians describe it as a polemos. We must assume that the Saracens were a more dangerous force along the frontier of the Near East than often supposed by scholars.

Moreover, the reason for explaining Ammianus’ silence can be further investigated with relation to the strong personal ties the author maintained with the military world of his time. His likes and dislikes together with his own ideology, dictated his selection of themes.

3. A recently published epigraphical text, from Petra offers a precious piece of information on the problems which affected Palaestina in late antiquity. In the lines where the text is clear we read: “City walls of the city.... Fighting he saved the city, the neighbours, together with all the inhabitants of Palaestina Salutaris. He destroyed the enemies, who speak a barbarian language. Thanks to the diligence of Orion this solid building was erected (or enlarged)”.60

Iulius appears in the last sentences of the res gestae as the protagonist of a wild episode which happened after the battle of Adrianopolis in the wake of the Gothic victory and the hatred it aroused. In his position as magister militiae trans Taurum, that is magister peditum et equitum per orientem, after becoming informed about the Roman disaster in Thracia, he ordered the extermination of all the young Goths kept as hostages in the cities of the provinces of the Near East. Quite significantly, Ammianus praises Iulius for having given this order: His diebus efficacia Iulii... enituit salutaris et velox. The project was carried out sine strepitu vel mora and the

58 Amm. XXXI.16.8. See on it Elbern 1987; Zuckerman 1991; Speidel 1998; Barnes 1998: 184-186. 59 On Julius’ episode in Ammianus see Matthews 1989: 16; 471. On the date of the composition of the res gestae see Sabbah 1997. 60 Merkelbach & Stauber 2002: 445. For a first notice about the discovery of the inscription see Bikai 1999. Moralee forthcoming offers a good discussion on some of the problems raised by this text.

56 Amm. XIV.7.7: Serenianus ex duce cuius ignavia populatam in Phoenice Celsein ante rettulimus. For the identification of Thelsea with Dumeir see Gatier 1999. 57 Lenski 1995; 1999.

250

ARIEL S. LEWIN: ‘AMR IBN ‘ADƮ, MAVIA, THE PHYLARCHS AND THE LATE ROMAN ARMY In analyzing this text we must discuss at least three important points, the position held by Orion, the date of the events and finally the identification of the enemies described as barbarians.

The text is written with a certain stylistic pretension, in dactylic hexameter, a good testament to the ongoing vibrancy of classical culture. Nothing prevents us from dating it even as late as the VIth century.66

The scholars who have offered an initial publication of the text before the appearance of a definitive edition with more extensive commentary have argued that Orion was a dux of Palaestina whose sphere of activity was focused more specifically on the territory of the Palaestina Salutaris. Alternatively, one could hypothesize that he was a municipal police official, responsible of the public order on the territory of Petra.

Nor does its mention of the province of Palaestina Salutaris exclude such a dating. Around 390 Palaestina was divided into three administrative entities, one of them comprising southern Transjordan and the Negev. In a letter written between 389 and 392 Jerome mentioned the existence of this newly created provincial entity called Palaestina Salutaris. In CTh VII.4.30 issued in 409 the provinces of Palaestinae are listed as prima, secunda, and tertia and in several other sources of the fifth and sixth century the province is called Palaestina tertia. However, in the acts of the Council of Ephesos in 431 our province is refered to as Palaestina Salutaris.67 Moreover, we must take into consideration the anonymous passio Aeliani, a text concerning a Christian who was martyred at Arabian Philadelphia. The events go back to the time of Diocletian, but the text - which is transmitted in a Georgian translation of an Arab translation of the original Greek! - quite probably was written at the end of the sixth or the beginning of the seventh century. The passio Aeliani states that the governor who later condemned the martyr had arrived in Arabia from Salutaris. It is difficult to avoid the conclusion that the anonymous writer was describing the administrative situation of the area as it was in his own time.68 If so, we can infer that the name Salutaris continued to be used to designate the provincia simultaneously with the designation Palaestina tertia until the Arab conquest.

As far as the chronology of the events is concerned palaeographical evaluations indicate a date around the mid to late fifth century although the early sixth century can not be excluded.61 The inscription was found in a structure located in the so called “Piazza” just south of the Ridge Church. It is apparently in secondary use and consequently may have been taken from the so called “inner city wall” that is nearby. The ecclesiastical complex with the Ridge Church has been dated to the fifth or sixth century.62 The city walls of Petra are mentioned in only one literary text, the deeds of the monk Barsauma. When this holy man approached to Petra, in the year 423 according to the tradition, the citizens were asking relief from a long period of drought and after his prayers the rains were so intense that they caused damage to the city wall.63 If so, we could infer that the city walls were already in existence at the time of Barsauma’s visit, that is during the reign of Theodosius II. Nevertheless, considerable caution is needed in the evaluation of this information. First, archaeological research has revealed the existence at Petra of more than one system of defence: the so called outer wall, built most probably in Nabataean times, does not seem to have represented a coherent line of walls, but rather a series of isolated strongholds and curtaine walls while the so-called inner wall shows some late antique features, including re-used materials; therefore it should be dated to after the earthquake of 363.64 Moreover, the text concerning the deeds of Barsauma is considered a suspect source, written in the sixth century and devoid of historical information about the time of Theodosius II.65 Finally, we must observe that the phraseology in the epigraphical text does not allow us to clarify if a renovation of the city walls or a building ex novo is to be assumed.

It is time now to turn to the identification of the barbarians mentioned in the inscription. I intend to suggest that most probably Arab tribes living beyond the provincial territory are described here. We must not forget, however, that Arab people lived in the interior of the province as well.69 Some of them were seminomadic. Theoretically we can not exclude the possibility that the persons who were behind the composition of the text and who belonged to the Greek speaking elite wanted to remark the distance that separated themselves from people they deemed inferior. One scholar, however, has suggested that usually late antique literary sources called all the Arabs Saracens while preferring to designate transprovincial Arabs explicitly as barbarians. This negative stigma was usually

61 See L. Di Segni personal communication. G. Bowersock (personal communication) pointed out to me that that some letter forms of the Petra inscription show palaeographical similarities with those in other epigraphical texts, some dated to the beginning of the fifth century and other, to a later time. Consequently, the beginning of the fifth century represents a terminus post quem for our inscription, but a date in Anastasius’ or in Justinian’s reign can not be ruled out. 62 See Moralee forthcoming with the relevant bibliography. 63 Nau 1927. 64 Fiema 2002: 201-202. 65 Honigmann 1954.

66

Bowersock 1997 = 2000; 2003. On the complex problem of the division of the Palaestina see Mayerson 1987 = 1994: 284-293; 1988 = 1994: 294-300; Sipilä 2004. See also now Sipilä in the present volume. Contrast Pietri 1991: 319338 = 1997: 609-628, who posits that Palaestina Salutaris was a creation of Constantius II. 68 Garitte 1961. In the text a martyrium which was erected in honour of Aelianus at Philadelphia is also mentioned. Archaeologists have identified this with a structure which was built after 590. See Michel 2001: 281. 69 See e. g. Hoyland 2004 67

251

THE LATE ROMAN ARMY IN THE EAST avoided for the Christianized Arabs who lived inside the territories of the empire or for the Arabs who were bound to Rome by a treaty.70

both Stephanus and Summus were coping with similar disturbances occasioned by Arab groups. Should we deduce that the conflict in which Stephanus was involved was a very serious one, on the scale of a large attack brought on by extraprovincial Arab tribes? Or, alternatively, did Choricius simply exaggerate Stephanus’ deeds?

Worth examining in this context is or. III of Choricius of Gaza written around 536 in praise of the dux Palalestinae Aratius and the consularis of the same province, Stephanus. This text, which describes two different actions executed by the dux Aratius against some Arab groups, appears to employ precise terminology that distinguishes the different typologies of Arab groups who were causing problems for the military authorities in the ‘30s of the sixth century.

As far as we know civil provincial governors before Stephanus did not command a large group of soldiers. Usually they had only some guards and municipal police under their authority. If so, their effective force was strong enough to fight only a small group of bandits or rebels. When a serious menace arose they had to work in cooperation with the dux.73

In the first instance (or. III.27-32) the dux succeeded in reestablishing order, after a perioud of troubles, along a regional route infested by a band of Saracens who were terrorizing travellers. He employed less than twenty soldiers in order to accomplish his task leading us to deduce that this was a minor police action, emphatically exaggerated by the orator for encomiastic purpose: the Saracens who were threatening the security of the provincial road were merely a group of petty bandits living within the provincial realm.

In 536, a short time after Choricius had delivered his oration in honor of Aratius and Stephanos, Justinian in his Novel 103 upgraded the position of the governor of Palaestina I to the rank of proconsul and in order to strengthen his authority attributed to him the command of a comitatensian military unit.74 In this law the emperor asserts that Stephanus had already succeeded quite well in providing security; moreover, Justinian recalls that Stephanus had previously received another imperial disposition that attributed to him the command of the aforementioned soldiers. To conclude, Stephanus was given a special authority over some military units by the emperor and at a slightly later stage when Nov. 103 was issued this disposition was confirmed, presumably to be maintained in the future. However, we must admit that we do not know exactly when during his tenure Stephanus received for the first time the right of having authority over soldiers. It is not clear if it was connected with the uprising of the Samaritans or with another act of banditry.75 As far as the Samaritans are concerned,

By contrast, at 20-26 Choricius praises Aratius for an operation the general had accomplished in a territory which was external to the province, or at least on its periphery. He succeeded in capturing from the enemy a stronghold that had previously been used by the Roman state for storing the output from a gold mine. Scholars have argued that such gold mines were located in the Hidjaz.71 If this is true, the aforementioned operations were accomplished in an area located at the margin of the provincial world and the attack had been conducted by people living beyond it. Choricius describes the group who had captured the stronghold as barbarians. (or. III.20).

episodes. The modality with which he introduces the second episode he says that the success obtained by Aratius reminds him a deed by Stephanus - reveals that the connection between the two episodes was that in both cases the defeated enemy was a Saracen group. In the second case Choricius used a variatio : in order to not repeat that the enemies were Saracens he preferred to describe them as Aegyptians. On Summus’ activity see Chor. or. IV.21-23. 73 For some examples Di Segni 1996: 578-580. In my view none of these can prove the presence of a military unit under the command of the civil governor. See also Liebeschuetz 1972: 124-125 on the consularis Syriae who in the last decades of the fourth century did not have soldiers at his disposition. On Stephanus see Holum 1989. Alternatively, we should not exclude the possibility that in special cases the civil governor was allowed to utilize some, presumably few, units among the ones normally available to the dux. Tantillo forthcoming argues convincingly that some civil governors of the fourth century maintained some military authority in their province, cooperating in this sphere with the duces. We can not exclude the possibility that the same happened in the sixth century. It is possible to suspect that sometimes disputes over authority arose between the dux and the civil governor. The same fact that in a quite unusual vein Choricius dedicated an oration to two different officials, Aratius and Stephanus, seems to conceal tensions and an overlap of competencies between the civil governor and the dux. 74 See Jones 1964: 661. 75 However, Mayerson 1986: 40 = 1994: 276 has pointed out that according to Choricius or. III.34 Stephanus had gathered a sufficient force to punish the Egyptians. He notes that “the phrase gathering a

In his turn the consularis Palaestinae Stephanus was praised by Choricius for having suppressed banditry by some Egyptians who had previously threatened the neighbouring towns. Most probably, his intervention against these bandits had occurred in Palestinian areas bordering the Egypt. It is important to note that in another oration, referring to events that had happened in the very same period, Choricius reports that the dux Palaestinae Summus had entered Egypt in order to fight some people who were causing troubles there.72 It is fair to assume that 70

See Christides 1969, a short study with a limited number of examples. Shahîd 1995: 183. 72 Chor. or. III.33-34. According to Shahîd 1995: 184 the cities mentioned as having been threatened by the Egyptians at the time of Stephanus’ intervention were in Aegypt. The same scholar underlines that it was not unusual for a commander of Palaestina to campaign beyond the provincial border. Unfortunatelly he attributes this campaign to Aratius, but Choricus or. III.33 after having praised Aratius for having cleaned the highway form the banditry of a Saracen group introduces the new episode saying that the deeds done by the general remind himself of the actions done by our governor of the Palaestina. This proves that two different authorities wee involved in the two 71

252

ARIEL S. LEWIN: ‘AMR IBN ‘ADƮ, MAVIA, THE PHYLARCHS AND THE LATE ROMAN ARMY undertaken by the men that the orator was praising.78 On the other hand it is probable that the renovation of the city walls at Gaza was completed for fear of impending attacks by the Saracens living in Egypt. In fact, as we have seen, such an enemy had already slaughtered people of some towns of Egypt and Stephanus’ activity against them could reveal that the authorities were worried about the expansion of the threat.79

Choricius seems to affirm that Stephanus did not fight against them, but succeeded in convincing them to not rebel. This entailed persuasion and diplomacy, not a military force. Oration III written by Choricius in honour of the dux Aratius and the consularis Stephanus reflects the same cultural and social milieu in which the people who were behind the writing of the inscription on the city walls of Petra operated. Interestingly, among the noteworthy acts done by Stephanus, Choricius emphatically praises the construction of a solid city wall which made the city of Gaza impregnable to its enemies. (Chor. or. III.54; 56). The erection of powerful city walls must be seen as a peculiar aspect of late antiquity. Several such construction or restauration projects were the object of praise in the epigraphical texts that reflect the same stylistic adornment and emphasis seen in the text from Petra concerning Orion’s activity.76

Let us now return to the Petra inscription: it is worth stressing that in this case the enemies are described as barbarians and that Orion is introduced as the man whose intervention saved all of Palaestina III. Moreover, the text affirms that he had killed very many enemies. All these deeds, building the city walls, providing for the security of the entire province, and killing the enemies are interconnected one with another and must be considered elements reflecting a period of noteworthy troubles for the province. Orion was almost certainly a dux and not a municipal official; in fact, it is impossible that a city official had the authority to lead operations in the entire province, well beyond the city territory. Moreover, it was normal for a dux to have authority over the construction of city walls, although the case of Stephanus reveals that sometimes the civil governor became involved in the construction as well.

Reading Choricius we get the impression that Stephanus had completed the renovation in anticipation of an enemy that remained only hypothetical. In other words, the rhetorical element seems to predominate. Indeed there is no mention of a readily identifiable people that was actually threatening the security of the city.

As we have seen, Choricius’ consistency in describing different Arab groups seems remarkable and corresponds with the picture portrayed by other intellectuals who distinguished between provincial robbers and tribes living beyond the province. If we could prove that such consistency was universal we could justifiably assert that the barbarians mentioned in the Petra inscription were extraprovincial peoples who were attacking Palaestina III.80

Two other orations by Choricius, however, reveal that around the same time the bishop Marcianus cooperated with the civil governor Stephanus in order to launch the project for the renovation of the city walls of Gaza. In Or. II.16 Choricius affirms that the old city wall had collapsed in some parts while other sections were still standing but easily scaled. It was a dangerous situation, but nonetheless the enemies had still remained quiet. Finally, the project was carried out and the city walls received an additional fosse. In a slightly different description in Or. I.7 the same writer stressed that in the times before the completion of the works the city had continued to be exposed to the attacks of enemies.77

Unfortunately we cannot be completely certain about it: in some cases the sources describe Saracens living in the provinces as barbarians as well in order to convey the idea that they had customs incompatible with the civilized world or that they were given to pagan cults.81

The potential aggressors against Gaza’s city walls remain generic in Choricius’ orations, so as to appear more a rhetorical device than a pressing reality. It might be argued that the mention in the text of a possible threat was a device for stressing the importance of the works

There are several arguments to be made against the notion that the Petra inscription offers mere bombast and that the barbarians were only local petty bandits. As far as we know the Arab seminomadic people living in the provinces of Palaestina III and Arabia did not pose a 78 Glucker 1987: 55-57 who nonetheless points to the existence of some ruins of late antique fortifications in some minor sites along the coast, near Gaza. 79 As far as it is known, the menace never materialized. 80 The intervention of a military leader, a dux or a stratelates, in order to save cities from barbarian threat seems to be a topos. See Barsanuphius ep. 131. 81 Christides 1969: 321. Libanius ep. 1236 describes the governor of Arabia Ulpianus in 364 as busy in fighting the barbarians. He may have detained both civil and military command. See Sartre 1982: 104. If so, there is the chance that Libanius records here an important conflict between the Romans and an extraprovincial tribe. See Lenski 2002: 203-204.

sufficient force is the key to understanding the military strategy employed to meet the threat of large numbers of attacking Saracens”. In fact, he gives some important parallels to which we can add the aforementioned episode of the intervention of the anonymous dux against Mavia’s forces. On the other hand, we know that in certain circumstances the civil governor gathered police forces or guards from various towns of the province. It is possible to argue that in such cases he did not have to fight a very large group of enemies. See for example Marcus Diaconus Vita Porphyrii 27. 76 Lewin 1991. 77 Chor. Laud. Marc. I.7 (4.8-18); II.16 (5-13). For an inscription datable to the VIth century implying some works at the city walls of Gaza see Glucker 1987: 140-141.

253

THE LATE ROMAN ARMY IN THE EAST serious threat to public order. It is true that at times they emerged as local bandits who in tiny populated areas robbed travellers or isolated monastic communities,82 yet it seems difficult to envisage how such small scale banditry could have affected the entire province and could have necessitated the construction of fortification walls.83

Ardabur. As we have seen the conflict was finally settled by a quite influential ambassador. This must be seen as a sign of the importance of the event. This evidence points to a difficult situation for the Roman authorities, with a powerful group of Arab tribes that was threatening more than one province.86 In the year 473 the Arab sheikh Amorkesos was able to occupy the island of Jotabe and a neighbouring territory. The historian Malchus affirms that Amorkesos after having conquered Jotabe devastated several villages nearby in his efforts to win appointment as “phylarch of the Saracens under Roman rule on the borders of “Petraea” or, according to a different reading of the text, “in Petraea”.87 Malchus deliberately employs here an archaizing name to describe the province of Palaestina Salutaris. Amorkesos’ attacks seem then to have been succesful at damaging some areas of this province.

The more dangerous Egyptians are not known to have carried their raids in the Petra area. If we keep in mind that the literary sources disclose that in the Vth and in the VIth century transprovincial Arab tribes again and again attacked Palaestina III causing devastations and the intervention of Roman generals, we must infer that the erection of Petra’s city walls and the bloody conflict and Orion’s victory also belong in this bleak context of a serious menace. As far as the fifth century is concerned it is important to note the description provided by the historian Priscus concerning the apprehensions felt by the Roman authorities in 447, when faced with Attila’s demands. They were afraid of the Persians, of the Vandals, of the Isaurians, of the Ethiopian tribes, and of the Saracens “who were ravaging the eastern parts of their dominion”.84

In 497/8 the dux Palaestinae, Romanus, was able to defeat three different groups of Arabs. “Both by good planning and generalship, he captured in battle Ogaros... Before the battle Romanus had worsted and put to flight another tent-dweller, Gabalas, who had overrun Palestine before Romanus’ arrival. At the same time also the island of Yotabe was set free by Romanus after fierce battles”.88 We can discern here the existence of various tribes. Gabalas was the chief of the Ghassanids, but scholars have differed in their identifications of the group led by Ogaros. Sartre, following a well grounded tradition, argued that the tribal group lead by Ogaros - Hudyr was the Kindite; Ogaros’ father, Arethas, was the ruler over a wide territory in central and northern Arabia; subsequently he delegated his powers to his son. Robin and in turn Whittow argue that Ogaros belonged to the banu Tha’laba, a separate branch of the Ghassanids or, alternativelly, a completely unrelated tribe.89

A literary source reveals that in 452 the comes et dux Palaestinae Dorotheus fought against enemies in the Moab. The Moab included the area between the wadi Hasa and the wadi Mujib which belonged to the province of Arabia until some time between 451 and 535 when the provincial boundary was moved to the north and the Moab became part of Palaestina III.85 It is not impossible that as a consequence of an enemy attack Dorotheos was forced to move with his men slightly beyond the provincial borders in order to fight very mobile enemies. It is worth noting that a scholar has convincingly argued that the military activity of Dorotheos must be connected to other military operations of the same year in the Damascus area, led by the magister militum per orientem,

Romanus, the succesfull military commander involved in the fights with these groups was dux Palaestinae, and consequently we can posit that Palaestina III in particular was the target of the enemy incursions he crushed in 497/498.

82

Isaac 1984 = 1998: 171-203; Mayerson 1985 = 1994: 35-47. Note that, as we have seen, Choricius’ description of the construction of the city wall at Gaza has a different flavour, devoid of a drama. Its construction is not described as strictly necessary, nor is it collocated amidst bloody conflicts. If we accept the premise that the authorities erected the city walls in anticipation of an enemy attack by Arab tribes living in Egypt, such an attack did never materialize. Caesarea also received a new city wall, probably in the late fifth century. See Lehmann 1994: 121-131. An inscription which must be dated to the end of the fifth century or to the beginning of the sixth, attests to the building of a bourgos. See Lehmann - Holum 2000: 79-80 n. 57. As far as we know, these works cannot be connected with any known threat. 84 Prisc. fr. 10 (ed. Blockley). 85 Niceph. Call. XV.9. Feissel 1984: 545-558 has shown that in an inscription now stored in Jerusalem, but probably transferred there from elesewhere, is praised a Dorotheos who could be identified with the person mentioned by Nicephoros. He was honoured with a statue upon formal authorization from the emperor. His sector of authority was the limes Palaestinae. Consequently he was a dux Palaestinae. On the meaning of limes as a military district under the command of a dux see Isaac 1988 = 1998: 345-387. On the transfer of the Moab into Palaestina see Sartre 1982: 73-74. 83

Furthermore Saracen attacks continued to affect the Roman Near East in the years following. In 501/502 according to Theophanes there was an incursion by an Arab tribe led by Badicharimos in Phoenice, Syria and Palaestina. These swiftly penetrated the provinces capturing booty and this time Romanus was not able to intercept them before they retreated into the desert.

86

Prisc. fr. 26; Shahîd 1989: 55-58. Malch fr. 1 (ed. Blockley 1983: 404-406). On the episode see Letsios 1989; Rubin 1989: 388-389. 88 Theoph. AM 5990; Mango - Scott - Greatrex 1997: 217. 89 Sartre 1982: 155-160; Robin 1996; Whittow 1999: 212-213. For a different interpretation see Shahîd 1989: 127-129. 87

254

ARIEL S. LEWIN: ‘AMR IBN ‘ADƮ, MAVIA, THE PHYLARCHS AND THE LATE ROMAN ARMY Badicharimos was another son of the king Arethas and was therefore the brother of the aforementioned Ogaros.90 Theophanes affirms that in the following year “Anastasios made a treaty with Arethas..after which all Palestine, Arabia and Phoenice enjoyed much peace and calm.” Scholars have debated whether this Arab king can be identified with Arethas the Kindite with whom, according to Nonnosus, the Romans stipulated a treaty. If not, Theophanes would be referring to a chief of the banu Tha’laba.91

old walls rebuilt and in some cases new defences were constructed so that they were better protected from enemy incursions. Military structures capable of guaranteeing security to the provincials were built everywhere while the army was greatly reinforced with new recruits.93 It is tempting to associate the time of Anastasius with the deeds described in the Petra inscription, but this must be ventured only with considerable caution: during late antiquity city walls were repeatedly rebuilt94 and, as we have seen, attacks by the Arab tribes necessitating the intervention of the Roman military authorities were commonplace. It is safer to assume that the troubles recorded in the Petra inscription trace to an incident otherwise unreported in our scanty source record.

Nonetheless the situation deteriorated again very soon afterward because at the end of 503 or the beginning of 504 the Lakhmids, taking advantage of the absence of the dux Romanus who was engaged in military operations on the Persian front, attacked Palaestina and Arabia. Their chief, Al-Mundhir inflicted serious damages on the provinces and captured thousands of prisoners. At the same time some Arab bands who were roaming the territory also reached the desert of Judaea. Al-Mundhir repeated these hostile incursions again and again, practically undisturbed, for approximately fifty years and succeeded in devastating the Roman territories from the boundaries of Egypt up to Mesopotamia. This negative trend was caused by Roman negligence and by the absence of efficient Roman commanders, but, according to Procopius, the sack and capture of provincial populations by the Lakhmid king continued even after Justinian appointed the Ghassanid Arethas as phylarch of all the Saracens at the service of the Roman empire in 531.92

To conclude: it is difficult to avoid the impression that Orion was a dux or another high military authority and that he fought against a serious menace represented by a group of external Arab tribes who were trying to raid the province. Conclusion In late antiquity the level of the confrontation between Persia and Rome increased dangerously beyond that of the first centuries of the empire. The rise to power of the warlike Sasanian dynasty precipitated an escalation of the conflict between the two superpowers marked by a series of wars in the third and in fourth centuries, down to at least the time of Julian’s expedition. In this context of intensified tension, the Arab tribes saw the gains they could achieve in offering themselves as allies to one or the other of the two empires. Powerful Arab chiefs, such as ‘Amr ibn ‘AdƯ, supported the Persians. But the Romans also quickly learned how to employ the Arab tribes in the near eastern arena in order to protect themselves and eventually to damage the interests of the Sasanids.

The reign of Anastasius appears to have witnessed repeated conflicts between Arab tribes and Roman authorities. Palaestina was among the provinces affected by such attacks. A panegyric written by Procopius of Gaza in honour of Anastasius dedicates an entire chapter to the victories obtained by this emperor against the Arabs who had attacked the provincial territories. The orator reports that the barbarians living in the neighbouring zones described as stereotypical nomadic marauders, quick in their attacks and retreats - had plundered the towns of the Near East which were lacking military protection. The Arabs inflicted a series of attacks on the population, but thanks to imperial foresight and intervention these barbarians finally received what they had deserved and were forced to agree to live in peace. The cities had their

As we have seen, it is possible but not certain that, at least for a quite short time, ‘Amr ibn ‘AdƯ was forced to transfer his allegiance to the Romans. Imru’l-qais was buried at Nemara and presumably in the last years of his life had become an ally of the Romans. But at least in the time of Constantius the fact that the Romans used Arab tribes as allies against the Persian is solidly attested. The conflict between the two empires had the effect of increasing violence in the paeninsula arabica: the chiefs of tribes intensified hostility and the most powerful

90

Theoph. AM 5994; Mango - Scott - Greatrex 1997: 222. Evagr. III.36 mentions an incursion of the Arab Scenitae who were capable of devastating Mesopotamia, both the provinces of Phoenice and the Palaestinae. They were roundly defeated by the Roman commanders and forced to a treaty. Evagrius probably compressed the events of more than one year. 91 Theoph. AM 5995; Mango - Scott - Greatrex 1997: 223. Nonn. FHG IV 179; Cedr. I 628. See Robin 1996: 697-698, followed by Whittow 1999: 212-214. 92 Cyr. Scyth. V. Ioh. 211.15-17; V. Euthym. XLVI; Procop. I.17.41; PLRE II, Romanus 7; Shahîd 17-18; 27-28; Greatrex 1998: 107. See also Whittow 1999.

93

Procop. Gaz. Pan. Anast. 7 (ed. Chauvot 1986). For a commentary see Chauvot 1986: 129-135. Specifically on the rebuilding of the forts in Palaestina see John of Nikiu LXXXIX.93; Capizzi 1969: 225. 94 In particular the city walls of Bostra and Gerasa underwent works of renovation in 440/1 after an order of the magister militum Anatolius . See IGLS XIII 9118 (Bostra); Welles n. 273 (Gerasa). For inscriptions attesting to the building of city walls in late antiquity see Lewin 1991: 79-98; Di Segni 1999: 166-178.

255

THE LATE ROMAN ARMY IN THE EAST among them succeeded in promoting themselves as partners of one or the other of the empires.

Since Diocletian’s time the entire military system had been reinforced: ammunition factories and a chain of forts along the fringe of the desert, also in areas not previously occupied, were built. 99 Throughout late antiquity great attention was given to the maintenance of city walls. It is worth stressing that the military apparatus, although particularly useful for coping with the Saracen menace, was a multifunctional system, not specifically built for defence against external aggressors. It was used for policing the provincial territory, for supporting the logistics of the army and for administrative duties.

Literary sources stress that the Saracens were untrustworthy and ready to change their allegiance. We might suppose that the negative description offered by the literary sources and, in particular, by Ammianus about this people was dictated by literary topoi concerning nomadic people95; however a more objective analysis of the feelings of the author of the res gestae shows that he indeed hated the Arabs, a sentiment that becomes palpable in his derogatory characterization of the Arabs as a perniciosa natio. The influence of the topos can not be completely denied, but, on the other hand, it is equally possible to argue that this negative attitude towards the Arabs by Ammianus and other writers was dictated by the perception of actual events and customs. In particular, the untrustworthiness seemed a disgusting aspect of their behaviour.

Ammianus affirms that Arabia was opima varietate commerciorum, castrisque oppleta validis et castellis, quae ad repellendos gentium vicinarum excursus sollicitudo pervigil veterum per opportunos saltus erexit et cautos. Haec quoque civitates habet inter oppida quedam ingentes, Bostram et Gerasam atque Philadelphiam, murorum firmitate cautissimas.100 This passage appears to reveal the two most remarkable features of the Near Eastern provinces in Late Antiquity: the first was the prosperity of the territory itself which developed as a result of the importance of its commercial activities. The second was the presence of a system of military infrastructures and of military units deployed in the territory which the writer thought indispensable for protecting it from enemy incursions.101

The Arab threat had a certain impact on the life of the Near East. In certain cases their incursions struck several provinces at once; at the time of the Tetrarchy the Saracens terrorized Roman territory from the Euphrates into Egypt; so too, Mavia’s revolt affected several provinces; in the first years of the reign of Marcianus, while the dux Palaestinae Dorotheos was fighting in the Moab, the magister militum per orientem Ardabur fought against the Saracens near Damascus; a peace was later negotiated under the leadership of an expert ambassador, Priscus, sent by the central government and by a very important personality of the time, Maximinus who was probably a magister praesentalis, leaving us to infer once again the existence of trouble in the area.

According to some scholars, this does not imply that we must describe the Roman military apparatus of the Near East as part of a larger grand strategy or that a so-called defence in depth was at work.102 In the case of Mavia’s revolt, however, when a major enemy attack threatened the Roman provinces, the military apparatus changed its character. The units scattered along the territory informed one another about the imminent menace and prepared a response to it.

Some decades later Amorkesos succeeded in devastating some villages and in occupying for a time a stronghold which had economic importance for the Romans; and the attacks by the Arab tribes in the times of Anastasius and Justinian were also important; in particular, the Lakhmid king al-Mundhir was able to terrorize all the provinces of the Near East, and in 523 succeeded in capturing two provincial duces.96

Moreover, in certain special situations, the Roman army was led by its generals to attack the enemy into zones well beyond the imperial borders. In particular, a quite important event occurred in 528/9 when a well coordinated expedition against the Lakhmids was led by several Roman generals, at least two of whom were provincial duces, and also a phylarch allied with the Romans participated in order to strike the Lakhmids in areas very far from the Roman realm. The casus belli had been a Lakhmid king’s murder of a Roman phylarch who had fled the Roman empire after having quarrelled with a dux Palaestinae: “Enmity developed between the dux of Palestine, Diomedes, a silentiarius, and the phylarch

It is not clear to what a degree the escalation of the Arab threat in the late fifth and into the sixth was the result of a weakening of the late Roman army stationed in the provinces of the Near East. During the fifth century some forts on the periphery of the empire had been abandoned because the soldiers had been transferred to other areas.97 On the other hand it must be recalled that a law issued in 492 reveals that at that time not only some comitatensian units but units belonging to the central army as well had been put under the command of the provincial duces.98

99

See Lewin 2001; 2002; 2004. Amm. XIV.8.13. 101 For a different interpretation of Ammianus’ passage see Isaac 1990: 171; Kennedy 1992: 479. They doubt that the aforementioned castra et castella were structures built in order to repel the attacks brought by Arab tribes. But see Lewin forthcoming. 102 See Whittaker 1994: 59. Nonetheless, according to Ammianus the presence of military structures was dictated by defensive necessity. 100

95

See Shaw 1982/1983. Procop. Bell. I.17.43-44; Evagr. HE IV.12; Nonnosus FHG IV.179. Fisher 2004. See also Lewin 2007. 98 CJ XII.35.18; Jones 1964: 660. 96 97

256

ARIEL S. LEWIN: ‘AMR IBN ‘ADƮ, MAVIA, THE PHYLARCHS AND THE LATE ROMAN ARMY Arethas. Arethas took fright and went to the inner limes towards India. On learning this Alamoundaros, the Persian Saracen, attacked the Roman phylarch, captured him and killed him, for he had 30.000 men with him. On learning this the emperor Justinian wrote to the duces of Phoenice, Arabia and Mesopotamia and to the phylarchs of the provinces to go after him and pursue him and his army. There set out at once the phylarch Arethas, Gnouphas, Naaman, Dionysios, dux of Phoenices, John, dux of Euphratesia, and the chiliarch Sebastianus with their military force. Learning of this Alamoundaros the Saracen fled to Indian territory with the Saracen force he had. The Roman duces and phylarchs went in with an accompanying force and, not finding him anywhere there, they set off toward Persian territory. They captured his camp and took prisoner a number of men, women and children, as many dromedaries as they found and other animals of various kind. They burnt four Persian fortresses, capturing the Saracens and Persians in them and they returned victorious in Roman territory in the month of April of the 6th indiction”.103

Tetrarchic period, Priscus the Saracen the menace of mid fifth century and Procopius the attacks led by the Lakhmids in the sixth century. Nevertheless the bleak situation we have described represented only one side of the coin. In fact, we must envisage the existence of a fluid context where contacts between the Romans and Arab tribes fluctuated between open warfare and official recognition of allies. Over time, the Romans forged new kinds of alliances and in addition to the traditional tribute payments started to appoint chieftains to the office of provincial phylarch.108 The effects of this political dynamics deserve a fuller discussion elsewhere, but for the moment it is worth stressing at least some of the most important issues. First, the phylarchs became strictly interconnected with provincial life at various levels. Around the year 420 an Arab chief, Aspebetos, transferred his alliance from the Persians to the Romans. His son was healed by a holy man and since that time, if not previously, they both father and son embraced Christianity. Aspebetos, after having been appointed phylarchos of the province of Arabia moved with his tribe to the desert of Palaestina where he was consecrated bishop of the Parembolé with the name of Peter by Juvenal of Jerusalem. Significantly, the acts of the councils have left traces of his involvement in the doctrinal disputes of the following years. It appears that he played a prominent part in the council of Ephesos in 431.109

This text also reveals that this late in the sixth century provincial units led by their duces were fighting troops which were also employed for campaigning far away from their bases, in the heart of the land inhabitated by the Arabs.104 The confrontation between the Romans and the Sasanians also revolved around the control of commerce.105 The Arab tribes played an important role in this context because they could damage the interest of one of the superpowers by attacking caravans or taking possession of caravan stations and strongholds.106

During Justinian’s reign, in the years 529-530, the phylarch of Palaestina cooperated with the Roman authorities in the repression of the Samaritan revolt. At the end of the hostilities he took 20.000 captives and sold them to the Persians and Indians.110 On the other hand, one of the Petra papyri shows that some years before 544 the Ghassanid phylarch Abu Karib, who could have been the same phylarch involved in the repression of the Samaritans, had been invited to act as an arbitrator in a local dispute over houses and properties at Zadacathon.111 These two examples give us some idea of the range of activity such chiefs performed in the provinces.

Ultimately, the importance of the section of the Roman frontier running from the Euphrates to the Red Sea cannot be denied. In Late Antiquity it began to assume a quite significant role.107 As we have seen, Malalas, Priscus and Procopius each in his turn describe the Saracens as a threatening force against all the provinces located along this frontier. Significantly we have at least one explicit description of this entire frontier as threatened in each century. Malalas narrated the Saracen incursions in

Secondly, in order to comprehend the importance of this presence at the borders of the Near Eastern provinces we must look at the structures and the varicoloured mosaics that were built at sites located in the marginal areas of the provinces, such as Resafa in Syria and Nitl in Arabia. We will be able to appreciate the emergence of a fascinating

103

Malal. Chron. tr. E. and M. Jeffreys 1986: 252. See also Theoph. AM 6021 = AD 528/9. For the identification of the two Arethas involved in the episode see the different interpretations given by Sartre 1982: 166168; Shahîd 1995: 70-76; 149-153; Robin 1996; Whittow 1999: 212214. 104 In the VIth century, duces had at their disposition, in addition to limitanei , some units of comitatenses and foederati. See Jones 1964: 660. Consequently, Liebeschuetz 1977: 497 suggests that usually comitatenses and foederati participated in expeditions while the limitanei remained at their places in the province. 105 See the pioneering work by Andreotti 1969. See also e.g. Winter 1987; Rubin 1989. 106 For example, it is interesting to observe that in the fifth century the Sasanians tried to extend their political influence in the HқejƗz through the activity of the Lakhmids. See Bosworth 1983: 600-601. The attacks to the carovans had a well established tradition in the Arabian peninsula. See Maraqten 1996. 107 Lewin 1990.

108 Sartre 1982; Shahîd 1989; 1995; Grouchevoi 1995; Liebeschuetz 2006. 109 Sartre 1982: 149-152; Shahîd 1989: 40-49; Millar 2006: 105-107. 110 Chor. Or. III.10; Malal. Chr. XVIII.446-447. 111 For Abu Karib see P. Petra inv. 83; Fiema 2002: 134; see also Fiema in the present volume. For the problem of the identification of the phylarch involved in the repression of the Samaritans see the discussion by Sartre 1982: 168-170; 1993: 150-153 who argues that two different phylarchs participated in the suppression of the revolt. Shahîd 1995: 8289 prefers to have Arethas as the partner of the Roman authorities in the episode.

257

THE LATE ROMAN ARMY IN THE EAST world where proficuous relations between peoples and cultures were fostered. The vitality of the villages on the fringe with the desert, such as Umm al Rasas, Nitl and several others, illustrates the character of the Christian Ghassanid presence and its ability to create the conditions necessary for a new phase of development of the Near East.112 Monasteries and defensive structures were constructed there, and a similar situation obtained in Syria where scholars have detected the traces of the devices for improving agriculture utilized by the Ghassanids. Moreover, it is very important to note that these allies of the Romans settled treaties with other tribes who were external to the empire and that eventually settled in marginal lands of the provinces and thereby contributed to the spread of Christianity.113

order to make it conform to the second one. In other words, the copyist himself probably added the jaà PakaistÊmg not found in a. If by contrast the reading provided by V is correct we might posit that the anonymous dux had been appointed to an ad hoc command over both provinces in order to cope with a serious threat. Bibliography Ammianus Marcellinus 1874 Res Gestae. Ed. W. Gardthausen, Leipzig. Andreotti R. 1969 Su alcuni problemi del rapporto fra politica di sicurezza e controllo del commercio nell’impero romano. In RIDA 16: 215-257. Argov E.I. 2006 A Church Historian in Search of Identity: Aspects of Early Byzantine Palestine in Sozomen’s Historia Ecclesiastica. In Zeitschrift für Antikes und Christentum 9: 367-396. Barnes T.D. 1981 Constantine and Eusebius, Cambridge Mass. Barnes T.D. 1982 The New Empire of Diocletian and Constantine, Cambridge Mass.–London. Barnes T.D. 1998 Ammianus Marcellinus and the Representation of Historical Reality, Ithaca–London. Barsanuphe et Jean de Gaza 1998 Correspondance. Vol. I (CERF 427), Paris. Bellamy J.A. 1985 A New Reading of the Namarah Inscription. In Journal of the American Oriental Society 105: 31-48. Bikai P.M. 1999 Petra North Ridge Project. In American Journal of Archaeology 103: 510-511. Blockley R.C. 1983 The Fragmentary Classicising Historians of the Later Roman Empire. Eunapius, Olympiodorus, Priscus and Malchus II, Leeds. Blockley R.C. 1992 East Roman Foreign Policy, Leeds. Bosworth A. 1983 Iran and the Arabs before Islam. In E. Yarshater (ed.), The Cambridge History of Iran. Vol. 3, Cambridge: 593-612. Bowersock G.W. 1978 Julian the Apostate, Cambridge Mass. Bowersock G.W. 1980 Mavia, Queen of the Saracens. In W. Eck - H. Galsterer - H. Wolff (ed.), Studien zur Antiken Sozialgeschichte. Festschrift Friedrich Vittinghof, Köln: 477-495. Bowersock G.W. 1983 Roman Arabia, Cambridge Mass. Bowersock G.W. 1997 Two Greek Historians of PreIslamic Arabia. In G. Most (ed.), Aporemata I: Collecting Fragments / Fragmente Sammeln, Göttingen: 173-185 = Bowersock 2000: 123-134. Bowersock G.W. 2000 Selected Papers on Late Antiquity, Bari. Bowersock G.W. 2003 The Nabataeans in Historical Context. In G. Markoe (ed.), Petra Rediscovered, London-New York: 19-25. Brennan P. 1984 Diocletian and the Goths. In Phoenix 38: 142-146.

Appendix In the edition of the Patrologia graeca of Sozomen the anonymous Roman commander who intervened in the war against Mavia and saved the magister militum per orientem is first mentioned as a dux Phoenices (ÒcelÆm t´m Ñm VoimÊj\ stqatiyt´m) and few lines later as a dux Phoenices et Palaestinae (ÒcelËmor t´m Pakaistim´m jaà VoimÊj´m stqatiyt´m). However the edition by Bidez and Hansen prefers the reading tÄm ÙcelËma t´m Ñm VoimÊw\ jaà PakaistÊm\ stqatiyt´m for the first passage as well. If so, my interpretation of the passage would be wrong. However, as far as is known, a militarily unified command over these two duchies never existed.114 Bidez - Hansen’s preference for the reading of the codex V rather than tradition b could be justified because generally V is much better than b, but on the other hand, we must note that V seems to have been transcribed by a copyst who intervened regularly and seriously in the texts. Manuscript V is not particularly old; it is dated to the time of the Paleologans when it was fashionable to correct texts after eliminating possible mistakes. It is worth noting that Hansen XXVII, after having observed that V is filled with omissions, additions and intentional changes, affirms that “Trotz dieser offenkundigen Mängel (die zumeist eben mit Hilfe von b leicht zu eliminieren sind) hat V für uns den grossen Vorzug, von allen Fehlern der Rezension b frei zu sein...” V’s redactional zeal should rather arouse suspicion: the result is a manuscript characterized by heavy intervention on the part of the copyist. In this instance, it is probable that the copyist, noting that in the second passage the anonymous general was called ÙcelËmor t´m Pakaistim´m jaà VoimÊj´m stqatiyt´m corrected the first passage in 112 See Lewin 2007. The Ghassanid structure at Resafa is studied by Key Fowden 1999. On the Ghassanids at Nitl see Piccirillo 2001; Shahîd 2001; Hamarneh 2004. 113 See Liebeschuetz 2006. 114 A look at PLRE I shows that duces having special commands are attested in the near East only in the first part of the fourth century. See the dux Aegypti Thebaidos utrarumque Libyarum , Maximinus 10 (308310), or Val. Rometalces (324-337), and also the dux orientis Fl. Platanius Serenianus 3.

258

ARIEL S. LEWIN: ‘AMR IBN ‘ADƮ, MAVIA, THE PHYLARCHS AND THE LATE ROMAN ARMY Fischer G. 2004 A New Perspective on Rome’s Desert Frontier. In Bulletin of the American Schools 336: 4960. Frye R.N. 1984 The History of Ancient Iran, München. Garitte G. 1961 La passion de S. Elien de Philadelphie (‘Amman). In Analecta Bollandiana 79: 412-446. Gatier P.L. 1999 Romains et Saracènes: deux forteresses de l’Antiquité tardive dans des documents méconnus. In Topoi 9: 218. Gichon M. 1967 The Origin of the Limes Palaestinae and the Major Phases of its Development. In Vorträge des 6. Internationalen Limeskongressus, Köln: 175193. Gichon M. 1971 The Military Significance of Certain Aspects of the Limes Palaestinae. In Roman Frontier Studies 1967. Proceedings of the 7th International Congress Tel Aviv: 191-200. Gichon M. 1991 When and Why the Romans Commence the Defence of Southern Palestine? In V.A. Maxfield - M.J. Dobson (ed.), Roman Frontier Studies 1989, Exeter 1991: 318-325. Gichon M. 1997 The Strategic Conception and the Tactical Functioning of the Limes Palaestinae after the Diocletianic Reorganisation. In W. Groenmannvan Waateringe - B.L. Beek - W.J.H. Willems - S.L. Wynia (ed.), Roman Frontier Studies 1995, Oxford: 25-31. Gichon M. 2002 Years of Research on the Limes Palaestinae - the Findings and their Assessment in the Light of the Criticism Raised (C1st - C4th). In Ph. Freeman - J. Bennett - Z.T. Fiema - B. Hoffmann (ed.), Limes XVIII. Proceedings of the XVIIIth International Congress of Roman Frontier Studies, Oxford: 185-206. Giversen S. 1986 The Manichean Coptic Papyri in the Chester Beatty Library, vol. II: Homilies and Varia. Facsimile edition, Genève. Glucker C.A.M. 1987 The City of Gaza in the Roman and Byzantine Periods, Oxford. Gnoli T. 2005 Dalla “hypateia” ai “phylarchoi”: per una storia istituzionale del “limes arabicus” fino a Giustiniano. In Ravenna da capitale imperiale a capitale esarcale, Spoleto: 495-536. Graf D. 1978 The Saracens and the Defense of the Arabian Frontier. In Bulletin of the American Schools of Oriental Research 229: 1-26. Graf D. 1989 Rome and the Saracens: Reassessing the Nomadic Menace. In T. Fahd (ed.), Arabie préislamique et son environment historique et culturel, Leiden: 341-400. Graf D. 1997 The via militaris in Arabia. In Dumbarton Oaks Papers 51: 271-281. Graf D. 1997 Rome and the Arabian Frontier: From the Nabataeans to the Saracens, Aldershot. Greatrex G. 1998 Rome and Persia at War, 502-532, Leeds. Grouchevoy A.G. 1995 Trois “niveaux” de phylarques. Etude terminologique sur les relations de Rome et de Byzance avec les Arabes avant l’Islam. In Syria 72: 105-131.

Cameron A. 1993 Barbarians and Politics at the Court of Arcadius, Berkeley. Capizzi C. 1969 L’imperatore Anastasio, Roma. Caskel W. 1966 Gamharat an-nasab. Das genealogische Werk des Hisam ibn Muhammad al-Kalbi, Leiden. Chauvot A. 1986 Procope de Gaza, Priscien de Cesaree. Panegyriques de l’empereur Anastase Ier, Bonn. Choricius 1929 Opera. Ed. R. Foerster - E. Richtsteig, Leipzig. Christides V. 1969 Arabs as “Barbaroi” before the Rise of Islam. In Balkan Studies 10: 315-324. De Blois F. 1995 Who is King Amaro? In Arabian Archaeology and Epigraphy 6: 196-198. Demandt A. 1970 Magister militum. In RE Suppl. 12: 553-790. De Maria S. 2003 Testi copti. Gli atti. In G. Gnoli (ed.), Il manicheismo, Milano 165-166; 394-396. Di Segni L. 1996 Metropolis and Provincia in Byzantine Palestine. In A. Raban - K.G. Holum (ed.), Caesarea Maritima. A Retrospective after two Millennia, Leiden: 575-589. Di Segni L. 1997 Epigraphic Documentation on Building in the Provinces of Palaestina and Arabia, 4t - 7th c. In J.H. Humphrey (ed.), The Roman and Byzantine Near East; Vol. 2. Some Recent Archaeological Research, Portsmouth: 149-178. Donner F. 1981 The Early Islamic Conquests, Princeton. Eadie J. 1996 The Transformation of the Eastern Frontier 260-305. In R.W. Mathisen - H.S. Sivan (ed.), Shifting Frontiers in Late Antiquity, Aldershot: 72-82. Elbern 1987 Die Gotenmassaker in Kleinasien (378 n. Chr.). In Hermes 17: 99-106. Evagrius 1898 Ecclesiastical History. Ed. J. Bidez - L. Parmentier, London. Feissel D. 1984 Notes d’épigraphie chrétienne (VII). In Bulletin de Correspondance Hellénique 108: 545-579. Feissel D. 2002 Les martyria d’Anasartha. In Travaux & Mémoires. 14. Mélanges Gilbert Dagron: 201-220. Fiema Z.T. 1995 Military Architecture and the Defense “System” of Roman Byzantine Southern Jordan - A Critical Appraisal of Current Interpretations. In Studies in the History and Archaeology of Jordan V, Amman: 261-269. Fiema Z.T. 2002 Late-antique Petra and its Hinterland: Recent Reasearch and New Interpretations. In J.H. Humphrey (ed.), The Roman and Byzantine Near East. Vol. 3, Portsmouth R.I.: 191-252. Fiema Z.T. 2002 The Military Presence in the Countryside of Petra in the 6th Century. In Ph. Freeman - J. Bennett - Z.T. Fiema - B. Hoffmann (ed.), Limes XVIII. Proceedings of the XVIIIth International Congress of Roman Frontier Studies, Oxford: 131-136. Findlater G. 2002 Limes Arabicus, via militaris and Resource Control in Southern Jordan. In Ph. Freeman - J. Bennett - Z.T. Fiema - B. Hoffmann (ed.), Limes XVIII. Proceedings of the XVIIIth International Congress of Roman Studies, Oxford: 137-152. 259

THE LATE ROMAN ARMY IN THE EAST Leadbetter B. 2002 Galerius and the Eastern Frontier. In Ph. Freeman - J. Bennett - Z.T. Fiema - B. Hoffmann (ed.), Limes XVIII. Proceedings of the XVIIIth International Congress of Roman Studies, Oxford: 8589. Leadbetter B. 2004 Trade, Frontiers and the Limes Palaestinae. In Studies in the History and Archaeology of Jordan 7: 247-251. Lehmann C.M. 1994 The Combined Caesarea Expeditions: The Excavation of Caesarea’s Byzantine City Wall, 1989. In Annual of the American Schools of Oriental Research 52: 121-131. Lehmann C.M. - Holum K. 2000 The Joint Expedition to Caesarea Maritima. Excavations Reports. The Greek and Latin Inscriptions of Caesarea Maritima, Boston. Lenski N. 1995 The Gothic Civil War and the Date of the Gothic Conversion. In Greek Roman and Byzantine Studies 36: 51-87. Lenski N. 1999 Basil and the Isaurian Uprising of 375. In Phoenix 53: 308-329. Lenski N. 2002 Failure of Empire. Valens and the Roman State in the Fourth Century A.D., BerkeleyLos Angeles-London. Leppin H. 2003 The Church Historians: Socrates, Sozomenos, and Theodoretus. In G. Marasco (ed.), Roman Historiography in Late Antiquity. Fourth to Sixth Century A.D., Leiden: 219-254. Letsios D.G. 1989 The Case of Amorkesos and the Question of the Roman Foederati in Arabia in the Vth Century. In T. Fahd (ed.), L’Arabie préislamique et son environnement historique et culturelle, Leiden: 525-538. Lewin A. 1990 Dall’Eufrate al Mar Rosso: Diocleziano, l’esercito e i confini tardo-antichi. In Athenaeum 78: 141-165. Lewin A. 1991 Studi sulla città imperiale romana nell’oriente tardoantico, (Biblioteca di Athenauem 17), Como. Lewin A. 2001 Kastron Mefaa, the equites promoti indigenae and the Creation of a Late Roman Frontier. In Liber Annuus 51: 293-304. Lewin A. 2002 Diocletian: Politics and limites in the Near East. In Ph. Freeman - J. Bennett - Z.T. Fiema B. Hoffmann (ed.), Limes XVIII. Proceedings of the XVIIIth International Congress of Roman Studies, Oxford: 91-101. Lewin A. 2004 Limitanei and comitatenses in the Near East from Diocletian to Valens. In Y. Le Bohec - C. Wolff (ed.), L’armée romaine de Diocletien à Valentinien I er, Lyon: 227-236. Lewin A. 2007 The Impact of the Late Roman Army in Palaestina and Arabia, In L. De Blois – E. Lo Cascio (ed.), The Impact of the Roman Army, Amsterdam: 463-480. Lewin A. forthcoming La provincia di Arabia in Ammiano Marcellino. In Studi in onore di M. Mazza. Liebeschuetz J.H.W.G. 1972 Antioch. City and Imperial Administration in the Later Roman Empire, Oxford.

Hamarneh B. 2004 Mosaici pavimentali delle chiese rurali di Nitl della provincia Arabia. In Musiva & sectilia 1: 199-215. Hartmann U. 2001 Das palmyrenische Teilreich, Stuttgart. Heather P. 2001 The Late Roman Art of Client Management: Imperial Defence in the Fourth Century West. In W. Pohl - I. Wood - H. Reimitz (ed.), The Transformation of Frontiers: From Late Antiquity to the Carolingians, Leiden: 15-68. Holum K.G. 1989 Flavius Stephanus. proconsul of Byzantine Palestine. In Zeitschrift für Papyrologie und Epigraphik: 231-239. Honigmann E. 1954 Le couvent de Barsauma et le Patriarchat jacobite d’Antioche et de Syrie, Louvain. Hoyland R. 2001 Arabia and Arabs from the Bronze Age to the Coming of Islam, London. Hoyland R. 2004 Language and Identity: The Twin Histories of Arabic and Aramaic (and: Why did Aramaic Succeed where Greek Failed?). In Scripta classica israelica 23: 183-199. Humbach H. - Skjaervo P.O. 1983 The Sassanian Inscription of Paikuli. Part 3, 1, Wiesbaden. Isaac B. 1984 Bandits in Judaea and Arabia. In Harvard Studies in Classical Philology 88: 171-203. Isaac B. 1988 The Meaning of limes and limitanei in ancient Sources. In Journal of Roman Studies 78: 125-147. Isaac B. 1990 The Limits of Empire. The Roman Army in the East, Oxford. Isaac B. 1998 The Near East under Roman Rule. Selected Papers, Leiden-New York-Köln. Isaac B. 1998b The Eastern Frontier. In A. Cameron P. Garnsey (ed.), The Cambridge Ancient History. Vol. XIII, Cambridge: 437-460. James S. 1988 The Fabricae: State Arms Factories of the Later Roman Empire. In J. C. Coulston (ed.), Military Equipment and the Identity of Roman Soldiers, Oxford: 257-332. Jeffreys E. and M. 1986 The Chronicle of John Malalas, Melbourne. John of Nikiu 1916 Chronicle of John Bishop of Nikiu. Tr. R.H. Charles, London. Jones A.H.M. 1964 The Later Roman Empire. 284-602 AD, Oxford. Julianus 1932 Œuvres complètes. Tome I. Ed. J. Bidez, Paris. Kaegi W.E. 1992 Byzantium and the Early Islamic Conquests, Cambridge. Key Fowden E. The Barbarian Plain. Saint Sergius between Rome and Iran, Berkeley. Kennedy D. 1992 The Roman Frontier in Arabia (Jordan). In Journal of Roman Archaeology 5: 473489. Kropp M. 1993 Vassal - Neither of Rome nor of Persia. Mar’al-qays the Great King of the Arabs. In Seminar for Arabian Studies 23: 63-93. Leadbetter B. 2000 Galerius and the Revolt of the Thebaid in 293/294. In Antichton 34: 83-95.

260

ARIEL S. LEWIN: ‘AMR IBN ‘ADƮ, MAVIA, THE PHYLARCHS AND THE LATE ROMAN ARMY Liebeschuetz W. 1977 The Defences of Syria in the Sixth Century. In D. Haupt - H.G. Horn (ed.), Studien zur Militärgrenzen Roms, Köln-Bonn: 487-499. Liebeschuetz W. 2006 Nomads, Phylarchs and Settlement in Syria and Palestine. In A. Lewin - P. Pellegrini (ed.), Settlements and Demography in the Near East in Late Antiquity, (Supplemento di Mediterraneo antico 2), Roma: 131-145. MacAdam H.I. 1980 The Nemara Inscription: Some Historical Considerations. In Al-abhath 28: 3-16. Macdonald M. 2000 Reflections on the Linguistic Map of pre-Islamic Arabia. In Arabian archaeology and epigraphy 11: 28-79. Malalas 2000 Chronographia. Ed. I. Thurn, Berlin. Mango C. - Scott R. - Greatrex G. 1997 The Chronicle of Theophanes Confessor, Oxford. Maraqten M. 1996 Dangerous Trade Routes: on the Plundering of Caravans in Pre-Islamic Near East. In Aram 8: 213-236. Matthews J. 1989 The Roman Empire of Ammianus, London. Mayerson Ph. 1980 Mavia. Queen of the Saracens - A Cautionary Note. In Israel Exploration Journal 30: 123-131. Mayerson Ph. 1986 The Saracens and the Limes. In Bulletin of the American Schools of Oriental Research 262: 35-47. Mayerson Ph. 1987 Libanius and the Administration of Palestine. In Zeitschrift für Papyrologie und Epigraphik 69: 251-260. Mayerson Ph. 1988 Justinian’s Novel 103 and the Reorganization of Palestine. In Bulletin of the American Schools of Oriental Research 269: 65-71. Mayerson Ph. 1989 Saracens and Romans: Micro-Macro Relationships. In Bulletin of the American Schools of Oriental Research 274: 71-79. Mayerson Ph. 1994 Monks, Martyrs, Soldiers and Saracens, Jerusalem-New York. Merkelbach R. - Stauber J. Steinepigramme aus dem griechischen Osten 4. Die Südkuste Kleinasien, Syrien und Palaestina, München-Leipzig. Michel A. 2001 Les èglises d’epoque byzantine et ummayade de Jordanie (provinces d’Arabie et de Palestine), Ve-VIIe siècle. Typologie architecturale et amenagements liturgiques, Turnhout. Millar F. 1993 The Roman Near East 31 BC - AD 337, Cambridge Mass.-London. Millar F. 2006 A Greek Roman Empire. Power and Belief under Theodosius II, Berkeley. Moralee J. forthcoming Petra and the Saracens: New Evidence from a Recently Discovered Epigram. In Shifting Frontiers in Late Antiquity VI. Motta D. 2005 La rappresentazione dei barbari in Sozomeno. In Mediterraneo antico 8.2: 495-521. Nau F. 1927 Deux épisodes de l’histoire juive sous Théodose II (423 et 438) d’après la vie de Barsauma le Syrien. In Revue des études juives 83: 184-206. Nixon C.E.V. - Saylor Rodgers B. 1994 In Praise of Later Roman Emperors. The panegyrici latini.

Introduction, Translation and Hidtorical Commentary, Berkeley. Parker S.T. 1985 Romans and Saracens: A History of the Arabian Frontier, Winona Lake. Parker S.T. 2000 The Defense of Palestine and Transjordan from Diocletian to Heraclius. In L. Staeger - J.A. Greene - M.D. Coogan (ed.), The Archaeology of Jordan and Beyond. Essays in Honour of James A. Sauer, Winona Lake: 367-388. Parker S.T. 2006 The Roman Frontier in Central Jordan. Final Report on the Limes Arabicus Project 19801989, Washington D.C. Pedersen N.A. 1997 A Manichaean Historical Text. In Zeitschrift für Papyrologie und Epigraphik 119: 193201. Piccirillo M. 2001 The Church of Saint Sergius at Nitl. A Centre of the Christian Arabs in the Steppe at the Gates of Madaba. In Liber annuus 41: 267-284. Pietri Ch. 1991 Les provinces “Salutaires”: géographie administrative et politique de la conversion sous l’Empire chrétien (IVe s.). In Aevum inter utrumque. Mélanges G. Sanders, Le Haye: 319-338 = 1997 Christiana respublica. Elements d’une enquete sur le christianisme antique, Roma: 609-628. Procopius 1963-1964 Opera omnia. Ed. J. Haury, rev. G. Wirth, Leipzig. Roberto U. 2003 Il magister Victor e l’opposizione ortodossa all’imperatore Valente nella storiografia ecclesiastica e nell’agiografia. In Mediterraneo antico 6, 1: 61-93. Robin Chr. 1996 Le royaume hujride, dit “royaume de Kinda”, entre Himyar et Byzance. In Comptes rendùs de l’Academie des inscriptions et belles lettres: 665714. Robin Chr. 1997 In Y. Calvet - Chr. Robin - F. BriquelChatonnex - M. Pic (ed.), Arabie hereuse, Arabie déserte. Les antiquités arabiques du Musée du Louvre, Paris: 265-269. Rubin Z. 1989 Byzantium and Southern Arabia - The Policy of Anastasius. In D.H. French - C.S. Lightfoot (ed.), The Eastern Frontier of the Roman Empire, Oxford: 383-420. Rufinus Historia ecclesiastica. Patrologia latina XXI 467-540. Sabbah G. 1997 Ammien Marcellin, Libanios, Antioche et la date des derniers livres des “Res Gestae”. In Cassiodorus 3: 89-116. Sartre M. 1982 Trois études sur l’Arabie romaine et byzantine, Bruxelles. Sartre M. 1993 Deux phylarches. In Museon 106: 143153. Schulz R. 1993 Die Entwicklung des römischen Völkerrechts im Vierten und fünften Jahrhundert n. Chr., Stuttgart. Schmitt O. 2004 Noch einmal zu ‘Mavia die Königin der Sarazenen’. In Mediterraneo antico 7, 2: 859-877. Shahîd I. 1984 Byzantium and the Arabs in the Fourth Century, Washington D.C. Shahîd I. 1989 Byzantium and the Arabs in the Fifth Century, Washington D.C. 261

THE LATE ROMAN ARMY IN THE EAST (ed.), The Roman and Byzantine Near East. Vol 2. Some Recent Archaeological Research. Journal of Roman Archaeology Supplement Thirty-one, Portsmouth R.I.: 207-224. Winkelmann F. 1966 Untersuchungen zur Kirchengeschichte des Gelasios von Kaisareia, Berlin. Winter F. 1987 Handel und Wirtschaft in Sasanidisch Römischen Vertragen und Abkommen. In Münsterische Beiträge zur antiken Handelsgeschichte 6: 46-74. Woods D. 1996 The Saracen Defenders of Constantinople in 378. In Roman and Byzantine Studies 37: 259-279. Woods D. 1998 Maurus, Mavia and Ammianus. In Mnemosyne 51: 371-378. Woods D. 2002 Ammianus and the Blood-Sucking Saracen. In K. Sidwell (ed.), Pleiades Setting. Essays for Pat Cronin on his 65yh Birthday, Cork: 127-145. Woods D. 2004 Valens. Valiant but Flawed. Review of N. Lenski, Failure of Empire. Valens and the Roman State in the Fourth Century. In Journal of Roman Archaeology 17: 729-734. Young G.K. 2001 Rome’s Eastern Trade. International Commerce and Imperial Policy 31 BC - AD 305, London-New York. Zosimus 1971-1979 Histoire nouvelle. Ed. F. Paschoud, Paris. Zuckerman C. 1991 Cappadocian Fathers and the Goths. In Travaux et Mémoires 11: 473-486. Zuckerman C. 2002 Sur la liste de Vérone et la province de Grande Arménie, la division de l’Empire et la date de création des diocèses. In Mélanges Gilbert Dagron Travaux et Mémoires 14: 617-637.

Shahîd I. 1995 Byzantium and the Arabs in the Sixth Century. Vol. I, Washington D.C. Shahîd I. 2000 Byzantium and the Arabs in the Reign of Constantine. The Nemara Inscription. An Arabic Monumentum Ancyranum. In Byzantinische Forschungen 26: 73-124 Shahîd I. 2001 The Sixth Century Complex at Nitl, Jordan. The Ghassanid Dimension. In Liber annuus 41: 285-292. Shatzman I. 1983 The Beginning of the Roman Defensive System in Judaea. In American Journal of Ancient History 8: 130-160. Shaw B. 1982/1983 “Eaters of Flesh, Drinkers of Milk: the Ancient Mediterranean Ideology of the Pastoral Nomad. In Ancient Society 14/4: 5-31. Sipilä J. 2004 Roman Arabia and the Provincial Reorganisations of the Fourt Century. In Mediterraneo antico 7, 1: 317-348. Sokrates 1995 Kirchengeschichte. Ed. G.C. Hansen, Berlin. Sozomenos 1960 Kirchengeschichte. Ed. J. Bidez - G.C. Hansen, GCS 50, Berlin. Speidel M.P. 1998 The Slaughter of Gothic Hostages after Adrianople. In Hermes 126: 503-506. Stevenson W. 2002/2003 Sozomen, Barbarians, and Early Byzantine Historiography. In GRBS 43: 51-75. Tantillo I. 1997 La prima orazione di Giuliano a Costanzo. Introduzione, traduzione e commento, Roma. Tantillo I. forthcoming Praesides, duces, comites. Amministrazione civile e militare nell’Africa tardoromana, con particolare riferimento al caso della Tripolitania. In Studi in onore di M. Mazza. Tardieu M. 1992 L’arrivée des Manichéens à al-HҕƯra. In P. Canivet - J.P. Rey-Coquais (ed.), La Syrie de Byzance à l’Islam, Damas: 15-24. Thelamon F. 1991 Paiens et chrétiens au IVe siècle. L’apport de l’Histoire ecclésiastique de Rufin d’Aquilée, Paris. Theodoretus 1998 Kirchengeschichte. Ed. L. Parmentier G.C. Hansen, Berlin. Theophanes 1883 Chronicon. Ed. C. de Boor, Leipzig. Urbainczyk T. 1997 Observations on the Differences between the Church Histories of Socrates and Sozomen. In Historia 46: 355-373. Van Deun P. 2003 The Church Historians after Eusebius, in G. Marasco (Ed.), Greek and Roman Historiography in Late Antiquity. Fourth to Sixth Century A.D., Leiden: 151-176. Van Nuffelen P. 2002 Gélase de Césarée. Un compilateur de cinquiéme siècle. In Byzantinische Zeitschrift 95: 621-639. Van Nuffelen P. 2004 Un héritage de paix et pieté, Leuven. Welles C.B. 1938 The Inscriptions in C.H. Kraeling, Gerasa, City of the Decapolis, New Haven: 355-615. Whittaker C.R. 1994 Frontiers of the Roman Empire. A Social and Economic Study, Baltimore. Whittow M. 1999 Rome and the Jafnids: Writing the History of a 6th-c. Tribal Dinasty. In J.H. Humphrey 262

L'armée romaine et la défense de la Syrie du Sud. Questions de méthode et nouveau document Maurice Sartre Université François-Rabelais (Tours), Institut Universitaire de France

La Syrie du Sud ne représente pas, dans l'ensemble de la frontière orientale de l'Empire romain entre l'Euphrate et la mer Rouge, un enjeu d'une particulière importance. Si j'ai choisi néanmoins de limiter mon enquête à un secteur relativement peu étendu de cette longue frontière, ce n'est pas tant pour l'originalité de sa situation que pour les possibilités de réflexion méthodologique qu'il offre. En effet, en ce qui concerne les aspects géographiques ou stratégiques du Hauran, il y a peu à dire. Je ne soulignerai que deux points. D'une part, entre Euphrate et mer Rouge, le Hauran se situe en l'un des rares secteurs où le pays des sédentaires et la steppe sont en contact en zone de montagne, notamment dans le Jebel Druze, même si l'on trouve des situations assez semblables sur les hauts plateaux de Moab et d'Edom. D'autre part, la région est le point d'aboutissement de l'une des voies de communication entre la Méditerranée et le golfe Persique, le wadi Sirhan, dont l'importance commerciale est plus supposée que réellement documentée, mais dont le rôle est maintenant bien mis en valeur grâce à la découverte de l'inscription mentionnant une route de rocade de Bostra à Dumata.1 Cela ne constitue pas, on le voit, une situation exceptionnelle, quelle que soit la richesse – très réelle – du Hauran durant l'Antiquité tardive. Peut-être faut-il ajouter que la période IVe-VIe siècle marque probablement dans l'histoire de l'occupation du sol de la région celle de la plus grande extension: des villages sont alors installés sur les marges de la steppe, comme à Diyathé, Deir en-Nasrani, Umm al-Quttein, ou même Umm al-Jimal, dont beaucoup furent ensuite désertés jusqu'à une époque toute récente (Diyathé reste presque seul désert). On ne peut sans doute pas ignorer cette poussée agricole vers l'Est et le Sud, c'est-à-dire en direction des nomades, dans une étude de la protection militaire de la région. Cela dit, en dépit de cette particularité, il va de soi qu'une étude du système de défense ne peut se concevoir pour le seul secteur hauranais, et doit donc être mise en relation avec les installations militaires relevées plus au Nord et plus au Sud.

Si la région est intéressante d'un point de vue méthodologique, c'est que l'on peut y croiser des informations nombreuses et d'origines diverses. Aux données de la Notitia Dignitatum pour l'Arabie, on peut en effet confronter non seulement l'exploration archéologique – beaucoup plus développée sur ce point du côté jordanien que du côté syrien – mais aussi l'enquête épigraphique, extrêmement riche et précieuse des deux côtés de la frontière. Or la confrontation de ces trois séries révèle quelques surprises qui incitent à la prudence dans l'interprétation. * * * Mais on prendra pour point de départ une inscription inédite de Teima (à paraître dans IGLS XVI 708), dans le nord du Jebel Druze. On peut en donner le texte tel que j'ai pu le restituer, avec malheureusement encore quelques lacunes, notamment à la fin de la ligne 3, la fin de la ligne 6 et la ligne 7 presque entière : [Pro salute] D[ominorum] n[ostrorum] Dioclet[iani et Maximi]ani Augg[ustorum] et Constanti et Maximian[i Nob. Caes. Re]stituta est castra noua coh[orte] prima [------nus, dd[ominis] nn[ostris] Diocletiano Augusto et [Consta] ntio Nob. Caes. II cons[ulibus] instan[tibus Aurelio Anti]ocho v[iro] p[erfectissimo] pr[aeside] et Aur[elio] Rufino PPROR.I

----------TP--------------------«Pour le salut de nos Seigneurs Dioclétien et Maximien Augustes et Constance et Maximien Très Nobles Césars ont été remis en état les nouveaux casernements (castra noua) pour la cohorte Ière [- - - ]nus, sous le consulat de nos Seigneurs Dioclétien Auguste et Constance le Très Noble César pour la 2e fois, en présence d'Aurelius Antiochus, vir perfectissimus, gouverneur, et d'Aurelius Rufinus - - - - - - -» La fin est difficile et, malheureusement, aucune photographie n'était possible. Aurelius Rufinus porte un titre que l'on aurait tendance à lire comme préfet du prétoire mais, d'une part, le titre de préfet du prétoire d'Orient n'est pas attesté avant 338, d'autre part, son nom devrait devancer celui du gouverneur provincial. Or, la date est absolument assurée par la mention du 2e consulat de Constance Chlore, qui géra ce consulat avec Dioclétien, dont c'était le 6e. On observe que cette indication n'est pas fournie par le texte. Aurelius Rufinus doit exercer une fonction qui m'échappe pour l'instant. Le gouverneur Fl. Antiochus qui préside à la construction ou reconstruction de castra noua dans ce village était déjà connu à Gerasa, peut-être à Umm al-Jimal, mais ne

1 Cette inscription a fait l'objet de multiples commentaires depuis qu'elle a été signalée pour la première foi par Gregory - Kennedy 1985: 269270 et 416-417; cf. en particulier Speidel 1987 ; Bauzou 1996. Christol et Lenoir 2001 ont suggéré de façon convaincante que ce texte appartient à l'époque d'Aurélien plutôt qu'à celle de la tétrarchie. Je crois, comme d'autres, que rien n'indique que l'ensemble de la route était contrôlée et patrouillée par l'armée romaine, jusqu'à Dumata, même si cela pouvait se produire de façon occasionnelle (cf. l'épitaphe de Flavius Dionysius à Dumata : Bowersock 1983 : pl. 14); mais le texte de l'inscription qui n'indique qu'une seule étape après Basie (Azraq), Amata, puis, d'une seule traite, 208 milles jusqu'à Dumata, me semble prouver que les patrouilles n'allaient pas, habituellement, au-delà de cette étape à 22 milles de Basie (déjà Zuckerman 1994: 83-84).

263

THE LATE ROMAN ARMY IN THE EAST pouvait être daté avec précision entre 264 et le début du e IV siècle. L'inscription de Teima fixe son gouvernement en 296 et peut-être un peu avant et/ou un peu après.2

inconnue. Pour nous ici, l'intérêt de ce texte est de nous signaler un camp de cohorte totalement ignoré des listes, même si la région abrite des installations militaires assez denses comme on le verra plus bas.

Le terme de castra est plus vague qu'il n'y paraît.3 Il est attesté pour désigner le fortin de Qasr Bshir (castra praetorii Mobeni) construit vers la même époque dans le pays de Moab, un peu au nord du camp légionnaire de Lejjun, sous le gouvernement du praeses Aurelius Asclepiades (attesté entre 293 et 305).4 A Palmyre, des castra furent également édifiés entre 293 et 303.5 On ne peut, à partir de là, préjuger de l'ampleur de la construction; sans doute est-ce davantage qu'une simple tour puisqu'il faut y loger une cohorte, mais il n'est pas utile d'imaginer un camp étendu. Aucune trace n'est actuellement repérable sur le terrain de manière directe, mais l'enquête archéologique reste à conduire.

On notera enfin que cette date haute confirme une affirmation de W. Seston qui notait que l'on ne connaissait aucun fortin construit par Dioclétien avant 297, sauf en Arabie.10 En tout cas, cette construction ne peut avoir de lien direct avec la reprise de la guerre avec les Perses en 297, puisque celle-ci fut déclenchée de manière imprévue et imprévisible par Narsès. * * * Cette découverte nous fait toucher du doigt, me semble-til, la fragilité des reconstructions qui ne prennent pas en compte la totalité de la documentation, et, pour ce qui est de la partie syrienne du Hauran où l'exploration archéologique paraît moins développée que du côté jordanien, il convient de ne pas surinterpréter les vides de la documentation. On peut commencer par l'examen de la documentation non-archéologique.

Les nouveaux castra ont été inaugurés en présence du gouverneur d'Arabie. Ils sont destinés à une cohors prima dont le nom a malheureusement disparu, sauf peut-être l'extrême fin, --nus, qui paraît incongrue grammaticalement. La Notitia n'en donne qu'une, la Prima miliaria Thracum à Adtitha, qui ne peut donc convenir ici. De plus, la construction des castra noua précède de plus d'un siècle la Notitia, et il y a pu y avoir des changements. Au total, six cohortes I sont attestées en Arabie : cohors I Hispanorum, cohors I Augusta Canathenorum equitata, cohors I Thebaeorum, cohors I Augusta Thracum equitata, cohors I Thracum, cohors I miliaria Thracum. La cohors I Augusta Thracum equitata est bien attestée par des inscriptions de Mothana et d'Umm al-Quttein et ce peut être la même qui n'est appelée que I Thracum à Qasr al-Hallabat 6. Moins probablement, ce pourrait être la même qui est devenue miliaria dans la Notitia.7 On peut sans doute exclure les deux cohortes venues d'Egypte en Judée à la veille de la conquête de l'Arabie et qui participe à l'annexion, la I Hispanorum et la I Thebaeorum8 car elles ne paraissent pas avoir séjourné longtemps dans la nouvelle province. Reste la I Canathenorum, dont le nom n'est pas très sûr dans une inscription de Mothana.9 Je ne vois pas le moyen de trancher, et les restes de lettres ne permettent pas d'arriver à une conclusion ferme. On ne peut exclure la présence plus ou moins éphémère d'une cohorte

Si l'on part de la liste des postes de garnisons conservée dans la ND Or. 37, on obtient une liste relativement courte et dont beaucoup de sites sont mal localisés. On peut en retenir les postes suivants, pour ce qui relève de la région sous examen : • le camp de la légion III Cyrenaica à Bostra, capitale de la province (21). • les Equites scutarii Illyriciani, à Motha (14) • les Equites promoti Illyriciani à Tricômia (15) • les Equites promoti indigenae à Speluncis (18) • les Equites sagittarii indigenae à Gadda (20) • les Equites sagittarii indigenae à Dia-Fenis (23) • l'ala sexta Hispanorum à Gomoha (26) • l'ala prima Valentiana à Thainatha (29) • l'ala secunda felix Valentiniana près d’Adittha (30) • la cohors prima miliaria Thracum à Adtitha (31) Sur cet ensemble, peu de sites sont identifiés avec une parfaite assurance: Bostra à Busra, Motha (généralement appelée Mothana) à Imtân (ou Khodr Imtân, la colline qui domine le village actuel d'Imtân). Pour les autres, la situation reste douteuse, même si je crois que l'on peut écarter des identifications anciennes et admettre quelques propositions récentes.

2 Cf. à Gerasa (Welles 1938: n° 162) et Umm al-Jimâl (PAES III A 234), v.p. praes. provinc. Arabiae; sur la date, Sartre 1982: 98, n° 54. Cf. encore PLRE Antiochus 12. 3 Cf. Gregory 1995-1997 I: 8-9. 4 CIL III 14149; cf. Gregory 1995-1997 II: 338-344. 5 CIL III 133, cf. 6661; cf. Gregory 1995-1997 II: 189-194. 6 Speidel 1978: 710-711 (= 1984: 252-253). L'un de ses cavaliers est enterré à Kurnub dans le Negev. 7 La Notitia la signale à Asabaia, que Avi-Yonah 1976: 32, localise à Qasr Supai‘a, à la suite de Abel 1938 II: 189 : Quseir Dubai‘a, « vaste site nabatéen ruiné, dominé par une tour, à 7 km d'un fortin romain appelé Kh. ez-Zôna, au sommet du plateau à droite du w. et-Temed, ensemble situé à l'ouest du Qala‘at Daba‘a, station du hagg ». On est donc bien loin de Teima si Abel et Avi-Yonah ont raison. Mais ont-ils raison? 8 Speidel 1978: 709 (= 1984: 251). 9 Speidel 1978: 709 (= 1984: 251); inscription CIL III 14379.

On a proposé de placer Tricômia à Salkhad11 ou à Qasr al-Ba‘idj,12 mais la première solution doit être retenue de 10

Seston 1946: 164. Dunand 1934: 108, n° 222; IGLS XVI 1116: Triakômè. La forme Tricômia se trouve à la fois dans la Notitia et dans les listes de Georges de Chypre, n° 1074. 12 Abel 1938 II: 188, suggère Qasr al-Bâ‘ij pour autant que la construction de la forteresse qu'on y connaît grâce à l'épigraphie soit antérieure à 412, date donnée par l'inscription dédicatoire ; Butler 1919: 11

264

MAURICE SARTRE: L'ARMÉE ROMAINE ET LA DÉFENSE DE LA SYRIE DU SUD préférence ; elle peut au moins s'appuyer sur un document provenant de la ville, bien qu'il ne soit pas d'usage de mentionner le toponyme dans les documents du lieu même; en effet, des pistoi de Triakômè dédient à Salkhad un édifice en 292-293.13

de soldats. De plus, Deir al-Kahf abrite aussi un vaste fortin (cf. ci-dessous) et nombre de traces d'une importante occupation militaire. Dia-Fenis, selon Abel, renverrait à la fois à Dion (Tell Ashari) et à Phaena (Mismiyyeh), qui se partageraient donc la garnison d'une unité de cavalerie indigène (p. 187); cette solution très étrange me semble devoir être écartée car Dion paraît abandonnée ou peu importante au Ve siècle;23 de plus, ce serait l'exemple unique d'une unité localisée en deux points éloignés. Gomoha serait Djumhâ, non loin au nord de Der‘a (Adraha) selon Abel,24 mais je ne connais aucun toponyme de ce nom dans la région et Dussaud l'ignore aussi; Avi-Yonah, dans son Gazetteer, sans citer Abel, et sans donner aucune justification le place (sur la carte p. 105) à mi-chemin entre Adraha et Tricomias.

On peut présumer qu’Adittha et Adtitha sont le même nom, l'une des unités étant stationnée dans le village même, l'autre à proximité (apud). Pour Abel et bien des commentateurs, ce serait al-Hadid, Hatita de la Tabula Peutingeriana, à 9 milles de Gadda,14 tandis que Gadda même serait Khirbet as-Samra, à 11 milles au Sud de Thainata.15 Tout repose en fait sur l'identification de Thainata, qu'Abel confondait avec Thantia de la Tabula Peutingeriana et plaçait à Umm al-Jimâl.16 En réalité, cette localisation a été écartée à juste titre par D. Kennedy17 car Umm al-Jimâl ne se situe pas sur la route Bostra-Philadelphia alors que Thantia en est clairement une étape. Thantia est en réalité Tughrat alJubb où l'on a reconnu des vestiges antiques. En conséquence Adittha/Adtitha sera Khirbet as-Samra et Gadda al-Hadîd18 ou, un peu plus au nord, Khirbet Khaw.19 Quant à l'identité entre Thainata et Thantia, elle est possible mais non certaine.20 On peut, au moins provisoirement, les distinguer.

Lorsque l'on place ces postes de garnison sur une carte, on observe que la IIIe légion Cyrénaïque se trouve assez loin à l'arrière, comme on peut s'y attendre. En revanche, si nos localisations sont exactes, on ne vérifie que partiellement les affirmations des auteurs au sujet des promoti indigenae, qui se situeraient toujours en première ligne, ou les Illyriciani, en seconde ligne. Pour les promoti indigenae, Speluncis, que ce soit Deir al-Kahf ou Umm al-Quttein, se trouve bien en première ligne, à la limite de la steppe et du désert. Mais les Illyriciani de Mothana et de Tricômia ne peuvent passer pour être sur des positions de l'arrière, même si ce n'est pas la toute première ligne. Le vrai problème est en réalité ailleurs : où est la première ligne, ou, en d'autres termes, où se situent le danger ? La définition d'une ligne de front suppose résolue l'identification de l'adversaire, ce qui est loin d'être le cas ici. Certes, on peut imaginer que l'ennemi est dans le désert, qu'il soit un ennemi proche (les bédouins), ou un ennemi lointain (les Perses). Mais cela reste à prouver.

Le reste paraît largement spéculatif. Speluncis a été localisé à Deir al-Kahf,21 mais l'identification repose en partie sur la similitude de sens (« les grottes », « le couvent de la grotte »), alors que ce phénomène naturel n'est pas rare dans la région. Henry Macadam a proposé l'identification avec Umm al-Quttein,22 mais sans cacher que le risque demeure d'user d'un argument circulaire, à savoir que le site militaire de Speluncis doit être à Umm al-Quttein parce qu'on y a trouvé un fortin et la mention 80-83, estimait que le fortin avait été construit d'un seul coup et qu'il n'y avait aucune trace d'occupation antérieure. Mais il note au passage que l'épigraphie (IGLS XVI 1116) localise Triacômè à Salkhad. Avi-Yonah 1976, recopiant à la va-vite la notice d'Abel, place cette Trikômias (sic) à « Qasr el Bâghiq near Salkhad », et le renseignement est repris sans aucun contrôle par Dodgeon et Lieu 1991: 346, avec renvoi au seul Gazetteer. Non seulement on crée ainsi un toponyme inconnu (Qasr alBâghiq, qui s'ajoute, il est vrai, aux formes Ba’ik de Butler et Baagué de l'Aviation Française au Levant, 1927, citée par Bauzou 1998: 112, n. 28), mais on associe deux sites qui sont pourtant éloignés de 35 km. environ à vol d'oiseau. 13 MnhvsÙqh B(a)ªdºaÙrou Kanhsoªuº. Ino" Auousou ke; Qeod(wv)ro" Oujatraniko;" ke; Maseco" Anamou ke; Ourero" Qe(m)ou ke; Akrabo" Sabivnou ke; Oujavlh" Sabivnou pistoi; Triakwvmh" e[(k)tisan. Diogevnh", kwvmh(") Omrwn, AÙmbro" oijkodovmo" oijkodovmhsen, (e[tou") rpq v. «Qu'on se souvienne de Badaros fils de Kanèsos. Inos fils d'Auousos, Théodoros, le vétéran, Masechos fils d'Anamos, Oureros fils de Thaimos, Akrabos fils de Sabinus et Valens fils de Sabinus, Pistoi de Triakômè, ont érigé (ce monument). Diogénès du village des Omroi; Ambros, constructeur, a construit, l'an 187». 14 Abel 1938 II: 188. 15 Abel 1938 II: 188. 16 Abel 1938 II: 188. 17 Kennedy 1982: 152-154. 18 Bauzou 1998: 252. 19 Kennedy 2000: 96. 20 Bauzou 1998: 252. 21 Abel 1938 II: 187, suivant Brünnow - von Domaszewski 1909 III: 256, et Butler 1919: 148. 22 Macadam 1989: 295-309.

* * * Si l'on se tourne maintenant vers l'archéologie, on ne parvient que partiellement à faire coïncider les données des textes avec les découvertes. L'exploration a été beaucoup plus largement conduite du côté jordanien grâce aux travaux de Butler et des savants américains au début du XXe siècle, puis de David Kennedy dans les années 1970-1980, l’ensemble répertorié et synthétisé par Shelagh Gregory.25 En revanche, du côté syrien, il reste beaucoup à faire malgré quelques travaux conduits par

23

Les philosophes qui trouvent près de là une entrée du Styx (Dam. Isid. fr. 6;2, avec les commentaires de Tardieu 1990: 49-51) estiment la ville déserte, mais, comme l'observe Michel Tardieu, cela signifie souvent qu'elle est entièrement chrétienne. Cependant, à l'appui d'un déclin de l'antique cité, on relève qu'elle est la seule cité attestée (notamment par un monnayage de l'époque des Sévères) qui ne soit pas devenue un évêché. Cela ne l'empêche évidemment pas d'avoir abrité un poste militaire, d'autant que le site est très favorable : un haut tell dominant la plaine de Batanée à l'Est, la vallée du Yarmouk à l'Ouest. 24 Abel 1938 II: 188. 25 Gregory 1995-1997.

265

THE LATE ROMAN ARMY IN THE EAST François Villeneuve (à Diyathé) et Maurice Lenoir (Bostra, Saané) dans les années 1990. • •

•

•

•

•

Le camp de Bostra, reconnu depuis longtemps, a été exploré récemment par M. Lenoir, qui cherchait notamment à en fixer les limites avec exactitude.26 La question qui se pose est surtout celle de son importance à partir du IVe siècle et au début du Ve siècle. Pour les autres découvertes, on peut en faire la liste sans grande difficulté, même si les questions de datation restent plusieurs fois en suspens.

•

à Qasr al-Bâ‘ij, Butler avait repéré un fortin romain construit en une seule fois;27 une inscription (PAES III A 21) en fixe la construction exactement en 411-412, au temps du dux Flavius Pelagius Antipater.28 Th. Bauzou29 n'y voyait plus, en 1984, que des débris informes. L'hypothèse d'une occupation tardive du site paraît confirmée par les relevés de céramique de surface de S. Th. Parker 1986: 24 (qui présuppose néanmoins la présence d'un fortin dès le IIe siècle, ce que rien ne confirme); l'absence du site sur la Tabula Peutingeriana milite de même dans le sens d'une occupation tardive, sans précédent aux IIe-IIIe siècles.30 à Tughrat al-Jubb (Thantia), présence possible d'un castellum avec quatre tours d'angle sur les photos aériennes; néanmoins Robinson Lees avait observé une enceinte rectangulaire avant sa destruction autour de 1900.31 S'agit-il de la même construction ? Et cela correspond-il au fortin occupé à par l'ala prima Valentiana? Probablement, si l'on admet l'identité de Thantia avec Thainata. Khirbet as-Samra (Adittha), une forteresse existe au e IV siècle, mais la présence d'une inscription inachevée en latin en l'honneur des empereurs Valentinien, Valens et Gratien, c'est-à-dire de l'époque où est créée l'ala secunda felix Valentiniana, milite en faveur d'une installation militaire de cette aile à cette époque précise. Le bâtiment doit être plus ancien, sans doute dès le IIe ou le IIIe siècle.32 al-Hadîd (Gadda?): R. E. Brunnow et A. von Domaszewski avaient relevé une enceinte rectangulaire de 107 x 158 m, avec quatre tours aux angles et une large porte sur le long mur sud.33 En dehors de la céramique de l'âge du Bronze, Parker n'y a trouvé que des tessons romains tardifs, byzantins et ayyoubide-mamelouk.34 Cela n'infirme donc pas la

•

présence d'une unité de cavalerie au IVe - début Ve siècle, même s'il n'y en a aucune trace concrète. Khirbet Khaw (Gadda?): un fortin grossièrement carré (100 x 100 x 94 x 88), de près d'un hectare. D'un point de vue typologique, le fort paraît appartenir à l'époque tétrarchique (comme Da'ajaniya et Umm alJimâl), mais la céramique ramassée sur place est soit plus ancienne (Ier-IIIe siècle), soit très tardive (fin VIedébut époque islamique).35 Umm al-Quttein (Speluncis): les nombreuses tours que Butler avait observées sont des tours de maisons36, mais l'examen des photos aériennes révèle l'emplacement d'un fortin au nord du village actuel, encadré par deux grands réservoirs, datés d'après la forme et les inscriptions des IIe-IIIe siècles.37 Umm al-Jimâl (Suratha d'après H. I. Macadam).38 Butler identifiait un praetorium, qu'il considérait comme équivalent au pyrgos de l'inscription PAES III A 233, datée de 371, mais le rapprochement est peutêtre illusoire. En revanche, une inscription mentionne un castellum (en grec kavstello")(PAES III A 237) que l'on reconnaît dans les casernes de la partie sudouest du site actuel39 et que Butler datait des environs de 412 par comparaison avec le fortin daté de Qasr alBa‘ij. La publication sommaire du monument par S. Thomas Parker40 ne contredit pas cette datation, au contraire, au vu de la typologie (absence de tour d'angle en saillie). C'est un vaste rectangle de 33,75 x 55 m., avec une chapelle en verrue sur le long côté est et une haute tour dans l'angle sud-est.41 Une faible garnison pourrait l'avoir occupé au Ve siècle et il aurait été abandonné un peu après la conquête islamique. Plus important, un castellum de bien plus grande ampleur (env. 95 x 112) se dresse le long de la muraille est de la ville, de type tétrarchique si l'on se fie au plan. Mais l'ensemble n'a pas encore fait l'objet d'une publication et a été seulement signalé en passant par S. Thomas Parker et D. Kennedy.42

• Kom ar-Ruff: forteresse antique (pré-romaine?), impossible à visiter car dans le no man's land entre Syrie et Jordanie.43 • Kom al-Manara: une tour à la jonction de deux routes.44 • Deir al-Kahf: forteresse à peu près carrée, d'un peu plus de 60 m de côté, avec entrée sur le côté est et quatre tours de trois étages aux angles45. Sur le côté ouest de la cour, chapelle. Le fortin aurait été d'abord

26

35

Lenoir 1998 et 2002. Butler 1919: 80-83. 28 Cf. Gregory 1995-1997 II: 261-265. 29 Bauzou 1998: 113. 30 Kennedy 2000: 86-87. 31 Lees 1895: 16; sur l'enceinte vue sur la photo aérienne, Bauzou 1998: 116, n. 49. Cf. aussi Kennedy 1982: 152-154; Kennedy 2000: 96-97. Le site est absent dans l'inventaire de Gregory. 32 Kennedy - Riley 1990: 198-200; Gregory 1995-1997 II: 284-288; Kennedy 2000: 97-98. 33 Brünnow - von Domaszewski 1905 II: 222-223; Kennedy 2000: 99102. 34 Parker 1986: 32; cf. Gregory 1995-1997 II: 298-299.

Kennedy 2000: 96. Butler 1919: 141. 37 Kennedy 2000: 76-77; cf. Gregory 1995-1997 II: 274-275. 38 Macadam 1986: 16-17; Suratha apparaît chez Ptol. V.16.4. Cf. Gregory 1995-1997 II: 266-299. 39 De Vries 1986: 227-241. 40 Parker 1998: 131-142; cf. description sommaire par Kennedy 2000: 85. 41 Butler 1919: 166-171. 42 Cf. Parker 1988; Kennedy 2000: 83. 43 PAES II A : 110. 44 PAES II A : 100. 45 Cf. Gregory 1995-1997 II: 276-281.

27

36

266

MAURICE SARTRE: L'ARMÉE ROMAINE ET LA DÉFENSE DE LA SYRIE DU SUD

•

• •

Mismiyyeh51 n'est qu'un temple, sans rapport particulier avec la garnison. A Mothana, on ne relève pas de traces d'édifices militaires, mais Butler signale à l'est d’Imtân, en un lieu nommé Umm Quseir (Est) une tour et un ensemble qui semble constituer une caserne et des étables.52 De plus, au sud d'Imtân le site de Gharabeh abriterait également une « petite forteresse ou une station militaire » selon Butler.53 On peut se demander si le nom de Mothana ne recouvre pas un ensemble de postes répartis autour du village même, ce qui expliquerait la dispersion des inscriptions militaires entre Imtân et Inak, notamment.

construit en 306 (PAES III A 328), puis réparé ou agrandi sous Valentinien, Valens et Gratien (PAES III A 329) – on le nomme alors castellum –. Une troisième inscription, non datée (PAES III A 330), rappelle la construction d'un réservoir et d'un aqueduc au temps du préfet Agrippa, et elle peut remonter au e 46 II siècle. Je ne suis pas sûr qu'il faille rapporter à ce site l'inscription d'Inat (située à 10 km au Nord) mentionnant la construction d'une tour (puvrgo")(PAES III A 224; Dussaud-Macler Voyage: 47) que le paysans prétendent avoir été apportée de Deir el-Kahf. Deir al-Qinn: fortin de 73 x 56 environ, repéré par photographie aérienne d'Aurel Stein, peut-être à partir d'une tour plus ancienne; mais les découvertes de surface militent en faveur d'une occupation précoce, peut-être comme tour de surveillance d'époque nabatéenne, abandonnée au IVe siècle quoique Gregory et Kennedy le juge caractéristique des castella du limes.47 Qasr Burqu: une tour ancienne (IIIe-IVe), et un établissement plus important tardif, peut-être seulement d'époque islamique.48 Qasr al-Hallabat:49 une inscription d’époque sévérienne (212-214) fait connaître la construction d’un castellum nouum par les soins de quatre cohortes (VI Hispanorum, I Thracum, V Afrorum Seuerianae et un quatrième non identifiée avec certitude, peut-être une III Thracum), au temps du gouvernement de Furnius Iulianus (PAES III A 17). Le fortin, bien conservé quoique reconstruit, est encore occupé à la fin du Ve et au début du VIe siècle, puisqu’on y a gravé in extenso le célèbre édit d’Anastase pour l’Arabie. En 529, le dux Flavius Anastasios faisait rénover le castrum (PAES III A 18).

Par ailleurs, le site de Diyatheh abrite une imposante forteresse rectangulaire (51,7 x 71,7 m.), non datée, avec quatre tours aux angles, deux au milieu des deux longs côtés et une porte sud flanquée de deux autres tours. Comme il n'y a aucune construction en dur à l'intérieur de cette enceinte, peut-être n'est-ce qu'un fortin occupé temporairement, avec logement sous la tente pour les patrouilles de passages.54 En revanche, il n'a rien été repéré pour l'instant à Malikiyyeh dont on verra qu'il a sans doute abrité aussi un détachement militaire. Il faut aller plus à l'Est encore pour trouver un dernier site où la présence militaire romaine est attestée: Nemara.55 Le fortin, mal conservé et peu exploré, n’est pas daté, mais des graffiti sur le linteau de la porte nord et sur les rochers environnants ont été laissés par des soldats de la légion III Cyrenaica, au IIe ou IIIe siècle. * * * La carte des sites militaires que l'on peut établir à partir de la Notitia et des découvertes archéologiques doit encore être confrontée à l'enseignement des inscriptions. Dans quelques cas, on peut relever une concordance étroite, par exemple à Mothana (Motha de la Notitia) où une importante série d'inscriptions confirme la présence d'une garnison aux IVe et Ve siècles (cf. plus bas) malgré l'absence de vestiges archéologiques. C'est moins net à Mismiyyeh car les textes paraissent pour l'essentiel d'une période plus ancienne (IIe-IIIe siècles) et l'identification entre Phaina avec le Fenis de la Notitia reste sujette à caution. De même, il y a de réelles correspondances pour quelques sites du Hauran jordanien, comme on va le voir.

Plus au Nord, en territoire syrien, l'exploration est moins systématique. Néanmoins, M. Lenoir a mis en évidence la présence d'un camp à Sa‘aneh, sur le rebord oriental du Jebel Druze.50 Ce camp, qui couvre 0,75 ha, a sans doute été créé à l'époque sévérienne et abandonné avant le milieu du Ve siècle. En revanche, dans trois sites mentionnés probablement par la Notitia, Salkhad (Triakômia), Mismiyyeh (Phaina-Fenis), Imtân (MothaMothana), les édifices militaires restent entièrement à découvrir, bien que les vestiges antiques ne manquent pas. Mais il n'y a pas eu d'étude systématique. On rappellera seulement que le prétendu « prétoire » de

Mais pour le reste, la situation paraît bien plus compliquée et nécessite quelques remarques préalables. En premier lieu, la plupart des inscriptions relatives à des soldats ne sont pas datées. C'est le cas de la totalité des épitaphes de soldats de Bostra, par exemple, même si l'on

46 Butler 1919: 145-148 ; Gregory - Kennedy 1985: 253-259, avec photographies; Kennedy 2000: 67-73. 47 Gregory-Kennedy 1985: 251-253; Kennedy 2000: 73-74; Gregory 1995-1997 II: 282-283. 48 Kennedy 2000: 74-75; Gregory 1995-1997 II: 239-243. 49 Gregory 1995-1997 II: 289-297. 50 Lenoir 2003: 145-157; le camp avait été repéré par photographie aérienne par Poidebard 1934: pl. XLVIII-XLIX, mais sans vérification au sol, et cette découverte avait été quelque peu oubliée par la suite, malgré de rapides mentions de Lander 1984: 145; Kennedy-Riley 1990: 189 et fig. 135-136; Gregory 1995-1997 II: 235-236; Gregory 19951997 III: fig. E18.1.

51

Weigand 1938; Hill 1975: 349, fig. 4; Ma'oz 1990: 41-46. Butler 1919: 143; un autre site de même nom se situe au sud-ouest d'Imtân. 53 Butler 1919: 144. 54 Butler 1919: 340-342; cf. Villeneuve 1986: 697-715; Kennedy-Riley 1990: 195-196; Gregory 1995-1997 II: 230-234; Kennedy 2000: 206207. 55 Gregory 1995-1997 II : 237-238. 52

267

THE LATE ROMAN ARMY IN THE EAST est en droit de supposer que les épitaphes en latin datent d'une période assez haute, probablement les IIe-IIIe siècles. Mais on trouve nombre d'inscriptions relatives à des soldats, en grec, que nous ne savons pas dater et qui peuvent descendre sans peine au IVe siècle. Mais il serait tout aussi anormal de placer au IVe siècle toutes les inscriptions militaires privées en grec, sauf datation explicite plus ancienne. Dans ces conditions, on peut difficilement faire usage de nombre d'inscriptions relatives à des soldats.

certaine importance. Adraha, Canatha, Dionysias, Philippopolis et Shaqqa (Maximianopolis) possèdent le statut de poleis. Ezraa (Zorava) peut leur être assimilée, comme métrokômia,57 et Sanamein (Airès) jouit sans doute du même statut en même temps qu'elle abrite les services d'un procurateur chargé d'administrer le saltus de Batanée,58 ce qui justifie aussi la présence de soldats. Restent tous les autres villages, Hit59, Malikiyyeh, Inak, Mseikeh, Majadel, Ghariyyeh Gharbiyyeh, Sleim, Kafr, Hebran, où cette présence ne peut être interprétée à coup sûr. Le plus simple est d'admettre que leur présence revêt un caractère strictement privé, comme le confirment pratiquement tous les textes. Mais cette dispersion traduit l'ancrage profond des soldats en activité dans la vie villageoise, ce qui correspond bien à l'image que l'on se fait des limitanei, dispersés sur l'ensemble du territoire pour bénéficier aux mieux des ressources qu'il offre. Or, plusieurs sont des soldats de la légion : un cavalier de la IIIe Cyrenaica à Hit (XVI 608), un beneficiarius à Rameh (XVI 808-809), un cavalier légionnaire à Mellah (XVI 1004) et un centurion sûrement chrétien (XVI 984), un soldat légionnaire à Orman (XVI 1068, en 250), un optio princeps de la IIIe Cyrenaica à Uyun (XVI 1171), un primipile en 295 à Qreyyeh (XVI 1189), un immunis carcerarius et un centurion ordinarius à Anz (XVI 1294 et 1299), des soldats de la IIIe Gallica (XV 137) et de la IIIe Cyrenaica (XV 143, 145) à Mseikeh sur le Leja, des centurions légionnaires encore à Ariqah (XV 338, en 169170), à Rimet al-Lohf (XV 395), à Breikeh (XV 405), des légionnaires à Majadel (XV 464), à Jubeib (XIII 9814), à Rimet Hazem (XVI 56, aux IIe-IIIe s.), un optio legionis à Kholkholeh (XV 513), et l'on pourrait facilement allonger la liste.

En second lieu, la répartition des soldats et officiers sur le territoire du Hauran montre qu'il s'en trouve à peu près partout, en tout cas dans bien plus de villages que ceux où sont attestés de façon certaine une garnison ou un fortin. Lorsqu'il ne s'agit que d'une inscription isolée, de caractère privé – consécration à un dieu, participation à la vie villageoise, épitaphe – on en déduira que le soldat avait noué des relations particulières avec ce village, sans en déduire qu'il y était nécessairement en garnison. Mais l'activité privée des soldats et officiers dans les villages pose évidemment la question de leur insertion dans la société hauranaise. S'agit-il de recrutements locaux ? Ontils fait souche au point d'être à la fois soldats et propriétaires fonciers dans les villages ? On a quelques exemples de soldats évergètes ou magistrats. Je n'y insisterai pas car la question a été récemment traitée.56 Dans quelques cas cependant, la mention de plusieurs soldats pose la question de la présence d'un détachement, permanent ou temporaire. L'imprécision de la chronologie rend les conclusions difficiles à tirer, mais on doit au moins souligner le fait. Ainsi, plusieurs soldats ou officiers sont connus dans des villages comme Shaqqa, Hit, Malikiyyeh, Mushennef, Bousan, Salkhad, Inak, Nemara, Mseikeh, Ezraa, Philippopolis, Majadel, Ghariyyeh Gharbiyyeh, Sleim, Atil, Canatha, Sia, Dionysias, Kafr, Hebran, Adraha, Sanamein. A Salkhad, les découvertes renforcent peut-être l'identification de Trikômia-Triakômè avec cette petite ville, et cela peut donc traduire la présence d'une garnison des equites promoti Illyriciani; mais les inscriptions mentionnent des soldats d'une cohorte de Canathéniens sous la 1ère Tétrarchie (IGLS XVI 1099) et deux centurions d'une unité non identifiée en 345-346 (IGLS XVI 1126). A Nemara, les inscriptions ne laissent guère de doute sur la présence d'un poste militaire avancé, car on est déjà dans le désert, sans aucune possibilité d'établissement sédentaire.

Quelles que soient les difficultés soulevées par ces inscriptions pour ce qui relève de la défense du Hauran entre l'époque de la Tétrarchie et le VIe siècle, le bilan très positif de leur apport mérite néanmoins d'être souligné. En premier lieu, les inscriptions relatives à des constructions militaires permettent de densifier le réseau de garnisons tel que l'on pouvait l'établir à partir des seules listes mentionnées plus haut. Certaines inscriptions sont trop vagues pour être utilisées, mais elles doivent être néanmoins signalées. Dans deux agglomérations, il est fait mention de la construction d'un mur (tei'co"), probablement défensif, sans que cela puisse être mis en rapport avec la présence de l'armée romaine. C'est le cas à Nimreh (XVI 736-737) où il s'agit d'un mur de clôture du village, mais il peut avoir un rôle défensif. C'est le même mot qui est employé à Adraha pour désigner le rempart de la cité. Bien que la reconstruction ou remise en état des remparts d'Adraha se fasse sous la surveillance

Dans quelques rares cas, au contraire, on peut expliquer la présence de soldats par l'importance d'un sanctuaire local sans que cela implique une garnison: ainsi à Mushennef bien connu pour son sanctuaire d'Allat, à Si‘a (Seeia) qui abrite un vaste complexe sacré (dont un mithraeum), peut-être à ‘Atil (Athela) où se trouvent deux temples dédiés aux empereurs. Mais ailleurs? Plusieurs des sites mentionnés plus haut sont des bourgs d'une 56

57

Cf. Sartre 1999. Sartre 1999. 59 On relève en particulier un préfet d'aile de l'armée hérodienne sous Agrippa Ier ou Agrippa II (XVI 615), un préfet (e[parco")non daté, un soldat – sans doute du IIe siècle – revenu de Germanie, un autre envoyé en Mésie. On relève encore un cavalier de la IIIe Cyrenaica. Le village paraît davantage un lieu de recrutement que de stationnement. 58

Cf. Sartre-Fauriat 2005.

268

MAURICE SARTRE: L'ARMÉE ROMAINE ET LA DÉFENSE DE LA SYRIE DU SUD étroite du gouverneur d'Arabie, rien ne signale la présence d'une garnison romaine permanente au IIIe siècle (les travaux se déroulent de 259 à 274).60

s'étend sur le Djebel Druze depuis Bteineh et Juneineh au Nord jusqu'à Orman et Dhibin au Sud: •

On est de même embarrassé pour faire usage des inscriptions mentionnant la construction d'un puvrgo". J'en ai relevé 14 mentions dans le Hauran; toutes celles qui sont datées appartiennent à la période IVe-VIe siècles, sauf une à Adraha sous Gallien. La difficulté vient de ce que l'on ignore si cela recouvre à chaque fois un ouvrage de type militaire.61 On sait en effet que, dans le Hauran tardif, des maisons sont flanquées d'une tour (Orman, Mellah as-Sarrar). Parfois, cette construction semble placée sous la responsabilité de notables villageois: à Orman, deux tours sont ainsi construites en 372-373 (IGLS XVI 1065-1066). A Athela, cela désigne à coup sûr un pigeonnier (XVI 127), à Canatha la dédicace est faite pour le dieu ancestral (XVI 158), ce qui n'invite pas à y reconnaître un ouvrage militaire, tout comme à Mjaidel où elle est dressée par un particulier (XV 127). Mais à Nahiteh, la tour est dédiée en latin (burgus) par le dux Flavius Bonus en 392-393. A Rameh, c'est l'œuvre - non datée - d'un beneficiarius du consulaire (XVI 808-809), à Inak, en 358-359, elle relève de la responsabilité du dux Fl. Silvinianus (XVI 1367), mais peut-être l'inscription at-elle été apportée de Deir al-Kahf si l'on en croit les paysans. A Meshquq, une tour est construite par un vétéran, ancien soldat ordinarius en Mésopotamie, en 351, mais ce peut être une tour de maison.62 Le terme apparaît aussi à Bostra à propos des travaux de restauration des remparts vers 539-541 sous la responsabilité de l'archevêque Jean (XIII 9128-9134, et XIII 9660, provenant de Jmarrin, mais peut-être bien transportée de Bostra). Enfin, dans la partie jordanienne du Hauran, le terme apparaît aussi à Umm al-Jimal dans une inscription de 371 (PAES III A 233) qui a sûrement un aspect militaire (le texte est en latin) mais sans que l'on sache à quelle structure militaire le rapporter.63

• • • • •

Bteineh (XVI 650-651): un frouvrion construit par les villageois, à une date non précisée. Juneineh (XVI 671): un frouvrion construit par le village, date inconnue. Irajeh (XVI 701): un frouvrion construit en 351-352 par le dux Fl. Silvinianus Rdaymeh Sharqiyyeh (XVI 688-689): un frouvrion construit en 349-350 par le comes Fl. Archélaos et par le dux Fl. Silvinianus. à Dhibîn (XVI 1211): un édifice dont le nom n'est pas précisé mais qui est sûrement militaire est construit vers 365-370. à Sleim enfin (XVI 75), un peu plus tôt en 345-346, un frouvrion est construit par le dux Flavius Victor « pour la sécurité des voyageurs ».65

Dans le même ordre d'idée, il ne peut y avoir de confusion sur la nature de la construction d'un castellum (en grec kavstello") à Qasr al-Ba‘ij en juin 412,66 d'un autre non daté (mais peut-être contemporain du précédent) à Umm al-Jimal (PAES III A 237) et d'un troisième à Deir al-Kahf en 306 (PAES III A 328), réparé ou agrandi sous Valentinien, Valens et Gratien (PAES III A 329). Si on s'en tient aux inscriptions datées, on constate donc une forte densité de construction militaires éparpillées du nord au sud du Jebel Druze entre la fin du IIIe siècle (Teima, 296) ou les premières années du IVe siècle (Deir al Kahf en 306) et le tout début du Ve siècle (Qasr alBâ‘ij, en 411-412). Si l'on cartographie les données épigraphiques relatives à des postes militaires romains, on aperçoit une nette densification par rapport aux seuls renseignements donnés par la Notitia et les relevés archéologiques. Cela permet de rééquilibrer le dispositif en territoire syrien, alors que l'archéologie avait privilégié la partie jordanienne du Hauran. On obtient un réseau presque continu sur la bordure orientale de la région, avec des concentrations dans le nord du Jebel Druze et de Salkhad à Azraq.

Le terme frouvrion a l'avantage de ne pas prêter à confusion, a priori, dans la mesure où le terme désigne un fortin ; mais les usages que l’on relève oblige à considérer que c’est un terme vague, désignant aussi bien un réel fortin qu’un camp militaire tout entier, par exemple à Palmyre ou à Mefaa.64 Or, à s'en tenir aux seules constructions de frouvria aux IVe-VIe siècles, on constate que la carte diffère de ce que nous apprennent les textes littéraires. Une ligne continue de frouvria

Cette ligne de fortins à l'Est du pays des sédentaires doitelle être mise en rapport avec des postes situés plus à l'Est? La question se pose pour deux raisons. D'une part, des postes militaires romains bien identifiés se trouvent à Nemara, c'est-à-dire dans le harra, à Jathum, à Qasr Burqu, pour le plus à l'Est. S'agit-il de postes avancés par rapport à la ligne de fortins définie plus haut? A Nemara,

60

IGLS XIV 8-22. Sur ce terme, cf. GREGORY 1995-1997 I: 9, qui estime que burgus et puvrgo" sont deux termes différents bien qu'ils découlent de la même racine. 62 IGLS XVI 1311: jAgaqhï' Tuvchï. Euj(tu)cw'" ejkodomhvqh oJ puvrgo". Basso" (o)aujtr(a)no;" ejx ojrdenarivwï s(t)ratiovmeno" ejm Mesopotamivaï. Soriano" oijdodovªmo"º. jAnelovqªhº muvria - - cili(av)de" - - -. [Htou" sm(e v). 63 Kennedy 2000: 84. 64 Cf. les exemples relevés par Gatier 1999: 215-218 ; noter que Mefaa, que Palladios nomme frouvrion s’appelle officiellement Kastron Mefaa ! 61

65 jEk prostavgmatoª" tou' kuºrivou mou Fl(aouivou) Oujivktoªro" tou' diºashm(otavtou) dou'ko" oijkodomhvqh to; frouvrion uJpe;r ajsfaliva" tw'n diodeuovntwn pronoiva/ jAlexavnrou (eJkatontavrcou) wjrd(inarivou) leg(ew'no") g v Kur(hnaikh'"), e[ti sm v. 66 PAES III A 21 = IGLJ V/1 92: jEpi; Fl(aouivou) Pelag(ivou) jAntipavtrou douko;" ejktivsqh kai; ejcrhmavtisen oJ kavstello", e[ti t" v, Daavssou hk v.

269

THE LATE ROMAN ARMY IN THE EAST les documents ne sont pas datés, mais ils semblent appartenir plutôt aux IIe-IIIe siècles, sans que l'on puisse en être certain. Surtout, en 328, le célèbre chef Imrulqays, « roi de tous les Arabes », se fait inhumer à Nemara ce qui semble exclure la présence d'une garnison romaine. A Qasr Burqu, il semble bien que seule la tour soit ancienne, les installations périphériques n'étant peut-être pas antérieures à la haute époque islamique.67 Dans ces conditions, la construction d'une ligne de fortins suivrait plutôt l'abandon d'une ligne située plus en avant. D'autre part, la route de praetentura ou praetensio qu'un milliaire d'al-Azraq permet de suivre de Bostra à Dumat al-Jandal apparaît clairement comme une voie de pénétration au cœur du désert. Après une première datation sous la tétrarchie, M. Christol a fait valoir qu'une date aurélienne convenait mieux. On peut se poser la même question que pour Nemara: fonctionne-t-elle en même temps que les fortins repérés? Le texte même de l'inscription suggère que l'armée romaine ne va pas au-delà de la première étape, à 22 milles d'Azraq/Basie.

dans le secteur nord-est du Jebel Druze. On observera que Teima est aussi le lieu de trouvaille de l'inscription des castra noua. C'est peut-être aussi le cas de l'ethnarque de Kafr, mais je crois que le titre est moins sûrement militaire que stratège ou même phylarque. De même, les syndics doivent être les porte-parole d'une communauté plutôt que des chefs militaires. Le phylarque Flavius Kathelgouos à Dhakir est sans doute un chef de tribu dont le rôle militaire reste à démontrer, quoique son nom de Flavius le désigne comme membre de l'administration impériale.69 Le phylarque Asaraèlos de Harran paraît plus indépendant. Quant à Abu Karib, à Samma’, c'est un homme extérieur à la région comme je l'ai montré ailleurs.70 Quoi qu'il en soit, je crois que l'on doit retenir au moins les quatre mentions de Malikiyyeh, Teima, Tharba et Rameh, sans doute celle de Dhakir. Ces officiers indigènes seraient donc l'équivalent du praepositus gentiliôn de Mothana (XVI 1368, daté 298, à Inak). On doit sans doute compléter notre carte des implantations militaires en y faisant figurer ces postes tenus par des troupes indigènes recrutées localement.

Il faut encore faire une place aux unités indigènes qui paraissent avoir stationné dans la région. Une série d'inscriptions mentionne en effet des phylarques ou ethnarques des nomades en plusieurs points du Hauran. On peut en établir la liste : • • • • • • • • • • •

* * *

Malikiyyeh: un ethnarque, stratège des nomades (XVI 698-699); dans le même village, un protector Caesaris (XVI 697). Teima: stratège des Paremboles et des nomades (XVI 713) Tharba: oiJ ajpo; e[qnou" nomavdwn (XVI 778) Rameh: stratège des Awidh (XVI 806) et dans le même village un beneficiarius (808-809). Mushennef: syndic des nomades (XVI 818) ; syndic sans précision (834-835) Bosana: syndics mais plutôt villageois (XVI 874-878) Dhakir: phylarque (XV 504-505), en 455-456. Jizeh: ethnarque (XIII, 9718) Samma: phylarque Abu-Karib, VIe siècle, (XIII 9843) Kafr: ethnarque (XVI 401) et dans le même village un protector (405) daté de 583-584. Harran: phylarque en 568-569, dans une bilingue gréco-arabe (XV 261).

Les conclusions que l'on peut tirer d'une telle enquête ne peuvent être que provisoires et de portée locale. On voit que pour un secteur limité, la combinaison de l'ensemble des sources donne une image assez différente de ce que l'on peut tirer d'une source unique comme la Notitia. Certes, celle-ci donne une image à un moment donné, mais je ne crois pas qu'elle soit aussi complète que l'on pourrait l'imaginer. Même si l'on admet que les troupes régulières sont toutes répertoriées, on voit que le dispositif d'ensemble et plus complexe et que nombre de fortins isolés doivent servir soit à des détachements de l'unité principale, soit à des garnisons temporaires, voire saisonnières (Diyatheh?). Une deuxième conclusion s'impose: la ligne principale de fortifications se trouve bien au contact des zones de sédentaires et de nomades, parfois un peu en retrait de manière à bénéficier de facilités de la route romaine Bostra-Philadelphie, mais il existe d'autres voies plus à l'est, sur la bordure orientale du Jebel Druze, de Salkhad à Mothana et Umm al-Quttein. La seule unité placée en retrait reste bien la légion de Bostra, selon le schéma repéré depuis longtemps un peu partout dans l'Empire.71 Ce déploiement de forces fixes dans cette zone correspond sans doute, comme l'a plusieurs fois souligné David Kennedy,72 à une extension des villages sédentaires en direction de l'Est; il fallait donc protéger

Une interprétation classique y voyait des structures indigènes, des chefferies locales au mieux reconnues par Rome, mais sans rapport direct avec l'administration romaine. Michael Macdonald a soutenu avec beaucoup de vraisemblance que l'ethnos des Thamoudéniens qui érige un sanctuaire des empereurs à Ruwwafa au temps de Marc Aurèle et de Lucius Verus n'était rien d'autre qu'un numerus intégré à l'armée romaine.68 Il en va probablement de même dans quelques autres cas, notamment pour les quatre premiers nommés, tous situés

69

Sartre 1993: 137-145. Sartre 1993: 137-145. Lewin 2004: 227-236. 72 Cf. Kennedy 2000: 48.

67

70

Gregory 1995-1997 II: 239-243; je me demande si le nom actuel ne dérive pas directement du latin burgus. 68 Macdonald 1995: 93-101.

71

270

MAURICE SARTRE: L'ARMÉE ROMAINE ET LA DÉFENSE DE LA SYRIE DU SUD ces paysans des marges. Or, il existe une réelle insécurité, bien que rien ne prouve qu'elle soit permanente. Dans l'inscription de Khirbet Umm al-Menara de 334, le protector Vincentius, qui commande le poste de Basie (Azraq) a fait construire un réservoir pour la garnison car des agrarienses ont été pris dans des embuscades et tués par des Saracènes.73

les documents sur une même carte donne peut-être une image excessivement militarisée de la région. Reste posée la question que l'on évoquait au début de cette communication: qui sont les ennemis? Les documents que l'on vient d'évoquer en Batanée excluen qu'il s'agisse exclusivement de gens arrivant du désert, qu'il s'agisse de bédouins ou des Perses.

Pourtant, la menace ne vient pas seulement d'éventuelles attaques des nomades. Dans la grande inscription dressée à Sanamein par la Batanée personnifiée, les dédicants – à coup sûr les habitants du saltus – remercient Julien César (ou le préfet du prétoire anonyme) d'avoir rétabli la paix et la sérénité dans la région,74 preuve de troubles récents d'une réelle importance. Comme l'inscription date de la période 355-360, les événements doivent être de peu antérieurs, entre 352/3 et 358/9. Il est tentant de mettre cette dédicace en rapport avec les troubles dus aux Saracènes dont parle Ammien Marcellin en 35375 et que Pierre-Louis Gatier a justement mis en relation avec la destruction de Thelséè, au nord-est de Damas, cette même année, dont le dux de Phénicie Serenianus est tenu pour responsable.76

Malgré ces limites, il paraît de plus en plus assuré qu'après les travaux de mise en défense effectués sous les Sévères, on assiste à des nouvelles vagues de construction d'abord sous la tétrarchie, puis vers le milieu du IVe siècle, puis à l'extrême fin et au début du siècle suivant. A chaque fois, on ne peut faire état d'une menace globale qui justifierait ces travaux. Peut-être faut-il chercher des causes locales (comme pour les constructions de Flavius Bonus dans les années 390). En tout cas, au moins sous la Tétrarchie et au début du Ve siècle, la mise en place de fortins de même type montre un plan concerté. Ne s'agitil pas en même temps de quadriller le pays pour éviter tout mouvement de révolte et de répartir la charge que représente l'entretien de l'armée sur l'ensemble de la région? Voire d'une mise en coupe réglée des campagnes?

Mais l’insécurité doit être constante, ou en tout cas antérieure à ce mouvement passager, car on observe des constructions assez nombreuses entre 345 et 352: une tour pour la sécurité des voyageurs à Sleim,77 les fortins de Rdaymeh Sharqiyyeh et de ‘Irajeh sous la direction de Fl. Salvinianus. Mais cela ne dut pas suffire puisque Flavius Bonus agit encore vers 392-393 à Jassem et à Nahiteh, c'est-à-dire à l'ouest de la région et au centre de la Nuqrah: à Nahiteh il construit un burgus,78 à Jassem il rétablit la paix et la sécurité.79 On est bien loin des frontières et l'armée se trouve mise à contribution dans tout le pays. Cela conforte l'idée qu'avait exprimée F. Jacques d'une région sous contrôle militaire dès le Haut Empire, avec omniprésence de l'armée; cette situation se poursuit au IVe siècle et sans doute dans une partie du Ve siècle. Mais on ne peut négliger que nombre d'inscriptions ne sont pas datées et que le cumul de tous

On aimerait enfin savoir quand change la situation repérée au IVe et début du Ve siècle. On a vu que le camp de Sa‘aneh était abandonné avant le milieu du Ve siècle. Aucune inscription ne mentionne des constructions après le milieu du Ve siècle et plus aucun soldat n'apparaît de façon explicite au-delà de la fin du IVe siècle. Or les inscriptions ne disparaissent pas pour autant. Il faut donc conclure à une mutation des habitudes et peut-être à un changement profond dans l'organisation des troupes. Peut-être n'est-ce qu'à ce moment-là que les troupes fournies par les alliés nomades jouent un rôle majeur. Il est trop tôt pour répondre, mais l'enquête continue. Abréviations IGLJ V/1 Bader N. Inscriptions grecques et latines de la Jordanie. T. V. Fasc. 1: Le Nord-Est de la Jordanie. Beyrouth. Sous presse. IGLS XIII Sartre M. Inscriptions grecques et latines de la Syrie. T. XIII. Fasc. 1: Bostra, Paris. 1982 (n° 90019472); Fasc. 2: Bostra (supplément) et la plaine de la Nuqrah, Beyrouth. Sous presse (n° 9473-9952). IGLS XIV Sartre-Fauriat A. - Sartre M. Inscriptions grecques et latines de la Syrie. T. XIV: La Batanée. A paraître. IGLS XV Sartre-Fauriat A. - Sartre M. Inscriptions grecques et latines de la Syrie. T. XV: Le plateau du Trachôn. A paraître. IGLS XVI Sartre M. Inscriptions grecques et latines de la Syrie. T. XVI: Le Jebel al-Arab. A paraître. PAES III A Littmann E. Publications of the Princeton University Archaelogical Expedition to Syria in 19041905 and 1909. Division III : The greek and latin inscriptions. Section A : Southern Hauran. Leyde. 1914.

73

AE 1948: 136. TÙo;n lªaºmÙprovtÙatÙoªn ---------------------------º to;n tou' iJerou' pretwrivou e[parcon tou' despotou' hJmw'n Kwnstantivou tou' Aujgouvstou kai; Nikhvtou kai; tou' ejpifanestavtou Kaivsaro" jIoulianou' di v ouJ' ijrhvnh kai; ajmerimniva u{phrxen hJ Bataneva to;n eujergeth;n kai; swthvra ajnevsthsen, ejpi; Fl(abivou) Maxivmou tou' diashmotavtou doukhnarivou tou' savltou; «Le clarissime - - - - - - - - - , préfet du sacré prétoire de notre maître Constance Auguste et Vainqueur, et du très illustre César Iulianus, par lequel paix et sérénité existent, la Batanée a dressé son évergète et sauveur, sous Flavius Maximus, l'éminent ducénaire du saltus». 75 Amm. Marc. XIV.4.3. 76 Gatier 1999: 214-215. 77 Sartre 2007. 78 IGLS XIII 9857. 79 IGLS XIV 450 : l'inscription n'est pas datée, mais celle de Nahiteh atteste la présence de ce dux en 392-393: JO kuvriov" mou Bovno" oJ lampr(ovtato") prwvtou Í tavg(mato") kovmh" kai; dou'ªxº a[rxa" hJm(w')n ejn eijrhvnhï Í kai; tou;" diodeuvonta" kai; to; e[qno" dia; Í panto;" eijrhneuvesqai hjsfalivsato. «Mon seigneur Bonos, le clarissime comte de premier rang et duc, nous ayant gouverné dans la paix, a rendu sûr et a pacifié (le pays) pour les voyageurs et pour le peuple, en tout». 74

271

THE LATE ROMAN ARMY IN THE EAST PLRE Jones A.H.M. – Martindale J. R. – Morris J. Prosopography of the Later Roman Empire I. Cambridge. 1971.

Lenoir M. 2002 Le camp de la légion IIIa Cyrenaica à Bostra. Recherches récentes. In Freeman Ph. Bennett J. - Fiema Z.T. – Hoffmann B. (ed.) Limes XVIII. Proceedings of the XVIIIth International Congress of Roman Frontier Studies held in Amman, Jordan (September 2000). Oxford: 175-184. Lenoir M. 2003 Sa‘neh ou le désert des tartares: un camp oublié du limes arabicus. In Syria 80: 145-157 Lewin A. 2004 Limitanei and comitatenses in the Near East from Diocletian to Valens. In Le Bohec Y. – Wolff C. (ed.) L'armée romaine de Dioclétien à Valentinien Ier. Actes du congrès de Lyon, 12-14 septembre 2002. Lyon: 227-236. Macadam H.I. 1986 Studies in the History of the Roman Province of Arabia. Oxford. Macadam H.I. 1989 Epigraphy and the Notitia Dignitatum. In French D.H. – Lightfoot C. (ed.) The Eastern Frontier of the Roman Empire, Proceeding of a Colloqium held at Ankara in September 1988. Oxford: 295-309. Macdonald M.C.A. 1995 Les Saracènes, l'inscription de Rawwafa et l'armée romaine. In Lozachmeur H. (ed.) Présence arabe dans le Croissant Fertile avant l'Hégire. Paris: 93-101. Ma'oz Z. U. 1990 The « Praetorium » at Mismiye, again. In DOP 44: 41-46. Parker S. Th. 1986 Romans and Saracens. Philadelphie. Parker S. Th. 1998 The later castellum (« barracks »). In de Vries B. Umm al-Jimal. A Frontier Town and its Landscape in Northern Jordan. Portsmouth (RI): 131142. Poidebard A. 1934 La trace de Rome dans le désert de Syrie. Paris. Sartre M. 1982 Trois Etudes sur l’Arabie romaine et byzantine. Bruxelles. Sartre M. 1993 Deux phylarques dans l’Arabie byzantine. In Le Muséon 106 : 137-145. Sartre M. 1999 Les métrokomiai de Syrie du Sud. In Syria 76 (=Mélanges offerts à Ernest Will): 197-222. Sartre M. 2007 Un nouveau duc d’Arabie, Flavius Victor. In Mélanges de l’Université Saint-Joseph (Beyrouth) (= Mélanges offerts à Jean-Paul ReyCoquais) 60. Sartre-Fauriat A. 2005 Les soldats : une élite sociale en Syrie à l’époque romaine? Antiquitas 28: 117-132 Seston W. 1946 Dioclétien et la tétrarchie. Paris. Speidel M. 1978 The Roman Army in Arabia. In ANRW II.8. Berlin-New York: 687-750. Speidel M. 1984 Roman Army Studies. I. Amsterdam. Speidel M. 1987 The Roman Road to Dumata (Jawf in Arabia Saudia) and the Frontier Strategy of praetensione coligare. In Historia 36: 213-221. Tardieu M. 1990 Les paysages reliques. Paris. Villeneuve F. 1986 Ad-Diyatheh : village et castellum romains et byzantins à l'Est du Jebel Druze (Syrie). In Freeman Ph. – Kennedy D. (ed.) The Defence of the Roman and Byzantine East. Oxford: 697-715. Weigand E. 1938 Das sogenannte Praetorium von Phaena-Mismije. In Würzburger Festgabe Heinrich Bulle. Stuttgart: 71-92.

Bibliographie Abel F.-M. 1938 Géographie de la Palestine. II. Paris. Avi-Yonah M. 1976 Gazetteer of Roman Palestine. Jérusalem. Bauzou Th. 1996 La praetensio de Bostra à Dumata (elJowf). In Syria 73: 23-35. Bauzou Th. 1998 La via nova en Arabie. Le secteur nord, de Bostra à Philadelphie. In Humbert J.B. Desreumaux A. (ed.) Khirbet es-Samra. I. Turnhout : 101-255. Bowersock G. W. 1983 Roman Arabia. Cambridge (Mass.). Brünnow R. E. - von Domaszewski A. 1905 Die Provinvia Arabia. T. I-II. Strasbourg. Brünnow R. E. - von Domaszewski A. 1909 Die Provinvia Arabia. T. III. Strasbourg. Butler H.C. 1919 Publications of the Princeton University Archaelogical Expedition to Syria in 19041905 and 1909. Division II: Ancient Architecture in Syria. Section A: Southern Syria. Leyde. Christol M.- Lenoir M. 2001 Qasr el-Azraq et la reconquête de l'Orient par Aurélien. In Syria 78: 163178 de Vries B. 1986 Umm al-Jimal in the first three centuries A.D. In Freeman Ph. - Kennedy D. (ed.) The Defence of the Roman and Byzantine East. Oxford: 227-241. Dodgeon M.H. - Lieu S.N.C. 1991 The Roman Eastern Frontier and the Persian Wars AD 226-363. Londres. Dunand M. 1934 Le Musée de Soueida. Paris. Gatier P.-L. 1999 Romains et Saracènes : deux forteresses de l’Antiquité tardive dans des documents méconnus. In Topoi 9: 209-218. Gregory Sh. 1995-1997 Roman Military Architecture on the Eastern Frontier. I 1995; II 1996; III 1997. Amsterdam. Gregory Sh. – Kennedy D. 198 Sir Aurel Stein's Limes Report. Oxford. Hill St. 1975 The Praetorium at Musmiye. In DOP 29: 347-348. Kennedy D.L. 1982 Archaeological explorations on the Roman frontier in north-east Jordan: the roman and byzantine military installations and road network on the ground and from the air. Oxford. Kennedy D. L. 2000 The Roman Army in Jordan. Amman. Kennedy D. L. – Riley D. 1990 Rome Desert Frontier from the Air. Londres. Lander J. 1984 Roman Stone Fortification. Variation and change from the First Century A.D. to the Fourth. Oxford. Lees R. 1895 Across Southern Bashan. In GJ 5: 1-27. Lenoir M. 1998 Bosra (Syrie) : le camp de la légion IIIe Cyrénaïque. In MEFRA 110: 523-528.

272

MAURICE SARTRE: L'ARMÉE ROMAINE ET LA DÉFENSE DE LA SYRIE DU SUD Welles C.B. 1938 The inscriptions. In Kraeling C.H. Gerasa, City of the Decapolis. New Haven. Zuckerman C. 1994 Aur. Valerianus (293/305) et Fl. Severinus (333), commandants en Arabie, et la forterese d’Azraq. In AntTard 2: 83-88.

273

ࡣͱ͝ࢲͧ͟͢͡, ͇ࢼ͒ͧ͜͟͢͡ and ̈́ͤ͒ͤ͘͡ ͔͘ࢹ͢ ͟͝uࢲ͕ͩ͝ in Roman Arabia (1st – 3rd Century). Central Power, Local Administration, and Nomadic Environment Thomas Brüggemann Institut für klassische Altertumswissenschaften, Martin Luther Universität, Halle

Roman rule in the Southern part of Syria remained rather provisional until the end of the 1st century A.D. With the death of Agrippa II in 92/93 and the annexation of his princedom the Romans began to strengthen their position in the region..1 Decisive for the further development of this region has been also the acquisition of the Nabataean kingdom and the following establishment of the province of Arabia in 106.2 Due to these processes the southern border of Syria had become a purely interior one.3 That meant that the northern part of the new province Arabia now was protected against raids by a desert-belt hardly passable, at least for the great armies of central states.4 So the strategic situation of the exercitus Arabicus had become, in this respect quite similar to that of the exercitus Iudaicus, and different to the Syrian frontier at the Euphrates: It was not a military unit in an area of permanent hostile threat any longer, at least from outside the empire’s territory. Consequently the presence of Roman permanent military installations in Arabia could be called rather sparse. The Roman occupying power seems to have concentrated the distribution of its troops in the border region with Syria to the urban centres in the north and northeast of the new province, at least to the end of the 2nd century AD.5 In fact there are till now no hints for a permanent Roman military presence in these steppe regions until the reign of Marcus Aurelius.6 This rural area between the two provinces, i.e. in the northern Arabia and the southern Syria, was inhabited by indigenous nomadic tribes,7 who were obviously not that kind of menace that the Romans, took really serious in the 1st century AD.8

Some Greek inscriptions known from this territory called Djabal HҏaurƗn, which seem to be erected by the tribes living there, may rather prove that at least some of them formally recognized Roman authority.9 That does surely not mean that the discussed area could be called quiet and peaceful. Permanent nomadic raids especially from the mountainous regions of the Trachonitis determined the Roman agenda between Antilebanon, Hermon, and Jabal Drnjz.10 But as these circumstances did not seem to question Roman rule generally, the central power confined itself to leaving the nomads-problem with the local allies11 — indeed without any substantial success.12 At first during the Antoninian times the Romans began to change there strategy towards these nomads significantly.13 So, the situation in the Trachonitis could be cleared ultimately by Marcus Aurelius and Commodus, who raised the number of Roman troops in the region considerablely.14 The mentioned Greek inscriptions appear with the beginning of serious Roman efforts to calm down the Trachonitis since the end of the 1st century, which seem to imply that the central power obviously tried at first to support the effectiveness of its own proceedings by incorporating members of the local tribes.15 Within this that Isaac 1990: 237 could convincingly demonstrate “that the provinces of Arabia and Judaea were not, before the fourth century, faced with very intense pressure of nomads […]”; the contrary opinion of Parker 1986 passim has to be strongly rejected. 9 See for the earlier discussion concerning these texts Sartre 1982b: 123128; MacDonald 1993: 368-376; Isaac 1990: 237-240. 10 In this respect has to be emphasized that not only assaults upon merchants and travellers are well documented, see for the Arabia Felix for example Strab. XVI.2.20, but also raids and invasions onto the territory of the empire up to the Damascus area were common, see Jos. Ant. XIV.9.2 and bell. Iud. I.10.5. 11 Already Augustus had charged Herod to eliminate that problem, as known from Jos. Ant. XV.10.1 and bell. Iud. I.20.4; see for Herod’s (unsuccessful) campaigns also Isaac 1984: 175-178 and Id. 1990: 62-66; Bowersock 1983: 50f. 12 Even the shifting of the legio III Cyrenaica to the border region of southern Syria by Trajan did not seem to have lowered the threat, see Gebhardt 2002: 91f; Eadie 1985: 410f; Isaac 1984: 172-181, Id. 1990: 65; Sartre 1982b: 122-128. 13 The building of the fortified military road through the Trachonitis between Damascus and the garrison at Bostra has to be called a fundamental turn in Roman politics: Dunand 1933; MacAdam 1986: 6871; Bauzou 1990: 294. 14 Especially the extension and completion of the garrison at Bostra in connection to the military road has to be mentioned in this respect; that there also were stationed new contingents results from an inscription, which names a centurion of the legio IV Scythica from 161/162; see Bauzou 1989: 208; MacAdam 1986: 23 and 68; Speidel 1998: 186 No. 33. 15 In the recent debate some scholars suggest convincingly that the

1

See for the death of Agrippa II. Jos. Vit. 32-410; the incidents are documented by Gebhardt 2002: 83-86; also Frankfort 1962: 659-672. 2 The sources for these events are rather sparse, see Cass. Dio LXVIII.14.5 and Amm. Marc. XIV.8.13; also Gebhardt 2002: 87ff, Bowersock 1970: 39ff and Sartre 1982b: 77-80. 3 The exact route of the border is documented by Sartre 1982b: 42-54 with maps 1-3; see also Freeman 1996: 104 with further literature. 4 Except the well documented Roman legion at Bostra it still is a controversial debate how the balance between legionarii and auxiliarii in the province Arabia could have been organized; see Tac. IV.5.4 and Gebhardt 2002: 88. 5 See for the debate concerning the garrison of the legio III Cyrenaica Bowersock 1983: 81 with note 18; Isaac 1990: 120; Freeman 1996: 98101; Pollard 2000: 42-52 and 242ff. 6 See Gebhardt 2002: 90ff and passim with further literature. 7 Strab. XVI.1.28 describes the first Arabic nomads as allies of the Romans in this region at the beginning of the 1st century AD as follows: ࢔ ͚͟͡͝ ͕ࣽ ࡝ͣͤࢷ ͤः͢ ͂͒͡ͱͥ͒ࢸͩ͝ ࡍͧ͡ः͢ ࢊ ̷࢕ͦ͡ࢲͤ͘͢ ͛͒ࢷ ࡪ ͖͒͠͡ࢸ͒. ͅࢱ ͕ࣻ ࡝ͤ͝ࢹ͢ ࡡ͚ͧͥͣ͟ द͟u͒ऑ͚͟ ͛͒ࢷ ͤभ͝ ࡕ͡ࢲ͓ͩ͝ ͟ࡺ ͦࢼ͚͒ͧ͜͟͡ uࢴ͚ͧ͡ ̴͓͒ͥͩ͜͝ࢸ͒͢, ͟ࡺ uࢳ͝ uࣴ͜͜͟͝ ࡝͖͛ࢸ͚͟͢͝, ͟ࡺ ͕ࢳ ͤ͟ऑ͢ द͟u͒ࢸ͚͟͢ ͣ͟͠͡ࢴ͖ͧͤ͟͢͝, ͟ࢀ͖ͣ͠͡ ͛͒ࢷ ͚ͣ͘͜͠ࢺͧͩ͟͡ࢸ ͖ࡹ͚ͣ͝, ࡰͤͤ͟͝ uࢳ͝ ̈́͛͘͝ऑ͚ͤ͒ ͟ࡺ ͟͝uࢲ͕͖͢ ͟ࡺ ͤम ͤ͒͟͠uम ͣ͘͜͠ࢸ͟͝, uࣴ͜͜͟͝ ͕ࣻ ͟ࡺ ࡑͩ͠ͱ͖͝ ͛͒ࢷ ͠͡ࢹ͢ ͤऄ ͖࢕͕͒ࢸu͚͟͝ ࡕ͓͒͡ࢸࣲ. 8 This seems to be true at least until the second half of the 3rd century so

275

THE LATE ROMAN ARMY IN THE EAST scenario fits perfectly the discovery that the central power, not seeing the necessity to set up permanent military facilities until the first half of the 2nd century, seems to have delegated police duties to these nomadic populations by recruiting detachments from them for temporary purposes.16

statements about Arabic or Saracen nomads, their ways of life or their modus vivendi with the Roman central power.21 That a large number of the indigenous tribes living in the Djabal HҏaurƗn in ancient times were nomads follows from many findings in the region,22 but by no means from the few Greek inscriptions discussed in the contribution.23

The epigraphic evidence for those phenomena, i.e. nomads ‘allied’ to Rome, can be dated between the 1st and early 3rd century AD, and it derives exactly from the above mentioned area between the two Roman provinces of Syria and Arabia, about 100 km southeast of Damascus.17 In a few Greek inscriptions the titles ͔͘ࢹ͢ ͟͝uࢲ͕ͩ͝ seem to be ࡝ͱ͝ࢲͧ͟͢͡, ͦࢼ͒ͧ͜͟͢͡ and ͣͤ͒ͤ͘͡ in a similar manner connected with the rather unspecific term ‘nomad’.18 Furthermore all of our inscriptions have in common, however, that they seem at least to be influenced from ‘outside’, i.e. by the Roman central power and not by the mentioned indigenous people. The paper tries to link the unfortunately mainly fragmentary epigraphic evidence with the Roman administrative action in the Near East in the early empire.19 The texts seem to reflect a ‘period of experimentation’ with offices, titles, and methods used by a central administration facing specific regional forms of life in the first two centuries AD. The way of experimenting seems to make sense in dealing with segmentary societies and their manners. That meant absolutely not distinct Roman politics or even an intended sedentarization of these populations.20 The inscriptions do not allow any

Although the three mentioned offices ࡝ͱ͝ࢲͧ͟͢͡, ͦࢼ͒ͧ͜͟͢͡ and ͔ͣͤ͒ͤ͘͡ࢺ͢ could occur in Greek and afterwards in Roman times under very different circumstances generally,24 their explicit connection with ‘nomads’ in our case should be called remarkable. One should perhaps assume that these offices did not have a direct equivalent in the indigenous (nomadic) part of the population, i.e. these titles would not be immediate Greek translations of native titles, but idioms, which may have come close to the Roman perception of tribal ‘functionaries’. Especially the offices of ࡝ͱ͝ࢲͧ͟͢͡ and ͦࢼ͒ͧ͜͟͢͡ and were related to regional social phenomena since the Seleucids in the Near East.25 Only the ͦࢼ͒ͧ͜͟͢͡ appears in segmentary or nomadic contexts before our inscriptions occur.26 The latter title includes at least implicitly an aspect referring to nomads already during societies, i.e. not only in the Near East, but also in the North African provinces; here as well the Romans created new offices trying to treat the nomads suitably, see Weiß 2006 forthcoming; for the diversification of mechanisms established by the Roman administration concerning nomadic environments compare also Brüggemann 2004: 156-187. 21 See in general for example Ball 2000; Graf 1989: 332-356; in respect to our time and region also Id. 2002: 153-160; Parker 1986; MacAdam 1986. 22 Nonetheless some scholars recently claimed that the indigenous inhabitants of the Jabal HҏaurƗn could be sedentaries (for the debate about the so-called ‘Safaïtes’ see MacDonald 1993: 306-310 and Id. 1992: 303-307, who is able to show clearly that this term could by no means used as an ethnicon for these tribes), because a few of the inscriptions from the indigenous tribes in the region were found in or around settled areas, as for example Milik 1980: 46 who argued that “à partir du 1er siècle avancé de notre ère, les tribus ‘awidhéniennes (d’abord) et les dafaïtes (ensuite) étaient quasi complement sédentarisées. Des fractions de cette population, en principe villageoise, nomadisaient encore, tout au moins par des sorties saisonnières”; in the same direction as well Sartre 1991 esp. 333ff, but MacDonald 1993: 311-323 has proved convincingly the mainly pure nomadic habits of these populations. 23 For the nomadic tradition in the area see for example Akkermans 2003 esp. 327-397; MacDonald 1993: 311ff; Sartre 1982a: 77ff and Id. 1983: 139-160. 24 For the phylarch’s office in the area see MacAdam 1989: 289-320; Id. 1994: 49-93; Mayerson 1991: 291-295; for the specific Jewish use of the title ࡝ͱ͝ࢲͧ͟͢͡ in Palestine compare Schalit 2001: 223ff and appendix XIV, esp. 782ff; for the sources and the epigraphic evidence see E. Schürer 1973/95 here esp. Vol. I, 330-335. 25 Ibid. passim. 26 In the area of the Jabal HҏaurƗn evidence for those two offices in indigenous environments is known from some inscriptions in which the dedicatees are called simply ࡝ͱ͝ࢲͧ͟͢͡ or ͦࢼ͒ͧ͜͟͢͡ with no other qualification, i.e. esp. not to nomads: ̳͘͜͟͢͝͝ ̈́͒u͖ͱͥ͟ [͠]͒͝ࢲ͖͡[ͤ]͖ ࡝ͱ͝ࢲͧ͒͡ ͧ͒ऑ͖͡ (JƯza), publ. in: MacDonald 1991 here 115, n. 89; ͇͜. ̼͒ͱ͖͔͜͟(͢) ͥࡺࢺ͢ ͦͥ͜ࢲͧͥ͟͡ ࢊ͚͛͡ࢺ͢ [sic!] (Dhakir), publ. in: Brünnow 1899 here 87, no. 55; PUAES III A 675 ͇ࢸ͚͜͜͟͢͠ ́ͣ[- - ࡝ͱ] | ͝ࢲͧ͘͢͡ ࡺ͖͡[ͤ͒͟uࢸ͒͢? … ͟ࡹ͕͛͟ࢺ] | u͖ͣ͘͝ ͤࢹ [- -] | u͕͖͘͝ࢹ͢ ͣͥ[u͓͒͜͜͟uࢴͥ͟͝? -] (Kafr), here the editors note that the restoration of ࡝ͱ] | ͝ࢲͧ͘͢͡ “must be regarded as very uncertain, especially in connection with the following word, which, whatever be the proper restoration, must denote some sacred or semi-sacred function” has to be emphasized.

region around the northern hills of the Djabal Drnjz, where the inscriptions (from the end of the 1st to the beginning of the 3rd century AD) under discussion here mainly were found, could have been a centre for temporary military campaigns in the area since Herodian times; esp. the village of Caesarea Eitha seems to be noteworthy in this respect, as MacAdam 1986: 61-67 and Gebhardt 2002: 249 with notes 1-7 assume. 16 This shows for example Wadd 2203: The phrase ͟ࡺ ࡍ͠ࢹ ࡟ͱͥ͟͢͝ ͟͝uࢲ͕ͩ͝ seems to hint to specific units recruited from among the nomads and used for particular purposes; Sartre (1982b) 124 assumes something similar, suggesting “l’expression ͟ࡺ ࡍ͠ࢹ ࡟ͱͥ͟͢͝ ͟͝uࢲ͕ͩ͝ désigne sans doute des soldats recrutés parmi les nomads”; see also the discussion below. 17 See for the region under discussion generally Sartre 2000: 635-663; Rey-Coquais 1978: 44-73; the epigraphic examples are mainly found in the Jabal HҏaurƗn, see MacDonald 1993: 303-413; Sartre 1982a: 77-91; Id. 1982b: 123ff. 18 Until now there are only twelve to fifteen Greek inscriptions with similar offices known from Roman times form 1st to 3rd century. Although M. Sartre in his paper ‘L’armée romaine et la défense de la Syrie du sud au IVe-VIe siècles’, given by him at our conference, hinted to more and new epigraphic data for the 4th to 6th century, this contribution has to stay with the already known and published material: The author’s efforts to learn from Maurice Sartre of which kind his (unpublished) material would be, to avoid two contradictory papers in the conference’s proceedings, remained unfortunately unsuccessful. 19 Several other epigraphical texts from the Roman Near East referring to these three offices mention only the titles ࡝ͱ͝ࢲͧ͟͢͡ and ͦࢼ͒ͧ͜͟͢͡ without reference to nomads and should therefore not to be discussed in this paper; only five Greek inscriptions from Roman times are mentioning ‘nomads’ at all, OGIS 321 = IGR III 1136 = Wadd 2112, OGIS 616 = Wadd 2196, 2203, PUAES III A 752 and AAES 383 and finally a sixth, OGIS 617 = Wadd 2236, mentions a ͔ͣͤ͒ͤ͘͡ࢺ͢ of the ࡕ͚͕ͥ͟ः͚͟͝, which may or may not imply nomads. 20 That a sedentarization of nomadic populations has never been a goal of Roman politics, has become clear for every area of the empire meanwhile where the central power came into contact with segmentary

276

THOMAS BRÜGGEMANN: ࡣͱ͝ࢲͧ͟͢͡, ͇ࢼ͒ͧ͜͟͢͡

AND

͔̈́ͤ͒ͤ͘͡ࢹ͢ ͟͝uࢲ͕ͩ͝ IN ROMAN ARABIA (1ST – 3RD CENTURY)

Seleucid times. Finally the ͔ͣͤ͒ͤ͘͡ࢺ͢, however, was a well known and rather conventional office in the military services in antiquity not only in the Palestine region.27

OGIS 616 offers an ࡝ͱ͝ࢲͧ͟͢͡ and ͔ͣͤ͒ͤ͘͡ࢹ͢ ͟͝uࢲ͕ͩ͝, who, as may be concluded by one of his names, seems to have been Romanized.31

The following remarks are divided into two main sections: In the first one the implications caused by each of the three titles will be dealt with directly by some noteworthy examples and in the second part the historical context of the inscriptions, i.e. the special significance of at least two of these offices in the Near East will be described briefly.

ࡖ͕͚͒͟͡͝ठ ͤ͟ठ ͛͒ࢷ ̈́͒͟ࢸ͕ͥ͟ | ̾͒͜ࢴͧͥ͟, ࡝ͱ͝ࢲͧͥ͟͡, ͣͤ͒͡ | ͔ͤ͘͟ठ ͟͝uࢲ͕ͩ͝, ͤࢹ | u͘͝u(͖)ऑ͟͝ ࡝ͤभ͝ ͓͜ऴ. | ࡙͕͕͟͢ ࡍ͕͖ͦ͜ࢹ͢ ࡝ͤभ͝ ͛͘ऴ.32 The first publisher of the inscription, William H. Waddington, calls the Ethnarch33 in this case a nomadic ‘chieftain’, but the frequency of his name ‘Hadrianos’ does not allow more precise statements.34 The 32 year old Hadrianos was honoured by his 25 year old brother Addos. The fathers name and that of the brother show clearly, that this family came from the tribal environment of the HҏaurƗn. That the deceased Hadrianos was a member of the Roman military service becomes evident not only by his special military rank as ͔ͣͤ͒ͤ͘͡ࢹ͢ ͟͝uࢲ͕ͩ͝, but also by giving his age. Doing the latter was normal in soldiers’ graves, while this habit did not become common in civilian funerals before Christian times.35 That even the brother mentioned his age might show that he as well has been a (still active) soldier: He may have thought that Roman soldiers generally had to mention their age in inscriptions — the result was a kind of hyper-conformism. If Waddington’s assumption should be true that this Hadrianos Souaidos was the leader of a (nomadic) tribe or clan, this would at least contradict the way a tribal shaykh is commonly identified, i.e. by the group he leads, not by the way of life of this group, in our case expressed

A sadly fragmentary inscription seems to represent a typical Roman equestrian career. OGIS 321, from the second half of the 1st century AD,28 shows an ࡡ͒ͧ͟͢͠͡ ͖ͣ͠ࢸ͘͢͡, i.e. a praefectus cohortis, who has been a ͖͔ͣͤ͒ͤ͡ࢹ͢ ͟͝uࢲ͕ͩ͝ as well. ࡣ͠ࢷ ͓͚͒ͣ͜ࢴͩ[͢ u͖͔ࢲͥ͜͟ ̾ࢲ͛ͥ͟͡ ࢁͥ͟] | ͜ࢸͥ͟ ࡕ͔͡ࢸ͒͠͠ | [ࡡͤͥ͟͢ . . , - - ࢊ ͕͖ऑ͒͝] | ͈͒͡ःͤ͟͢ ࡡ͒͠[ͧ͟͢͡ ---9] | ͖ͣ͠ࢸ͘͢͡ ̳࢕ [͔͟ࢼͣͤ͘͢ (?)͛͒ࢷ ͣͤ͒ͤ͘͡] | ࢹ͢ ̿͟uࢲ͕ͩ͝ [---11] | ͘͢ ͛͒ࢷ ͈͒͜ [͚͕͛͘͝ः͢ -----]29 The rank of an ࡡ͒ͧ͟͢͠͡ ͖ͣ͠ࢸ͘͢͡ leads to the assumption that ͈͒͡ः͢ was Romanized quite well and that he may have had a career in the Roman military service.30 These details distinguish him considerably from all other individuals in the following texts of the ‘group’ dealt with here, who seem to be integrated into the ‘Roman context’ rather ad hoc, spontaneously. The duties of a ͔ͣͤ͒ͤ͘͡ࢹ͢ ͟͝uࢲ͕ͩ͝ are not referred to in our fragment. As it can be seen clearly, the inscription has been an official document, most probable erected by ͈͒͡ः͢ himself or his tribe to honour king Agrippa II. If the protagonist of the first inscription seems to stand at least with one foot (i.e. as being an ࡡ͒ͧ͟͢͠͡) in the traditional Roman cursus yet, this could not be said about the following text. The funeral inscription of a ‘Hadrianos, also named Soaidos son of (?) Malechos’ in

31 The fact that Hadrianos mentions his indigenous name, Soaidos, in the formal text also could document the in the Near East since Seleucid times quite common habit to adopt traditional names of a central power with regard to official duties; for this hint the writer has to thank A. Mehl. 32 OGIS 616 = Wadd 2196 from Batanaea in the Trachonitis (Malka or El-Mahkiye), second half of the 2nd century AD; the assumption of Sartre 1982c, who 125 points out that the two titles mentioned may well have been held by different men, in which case the title ethnarch need have had nothing to do with nomads, seems convincing; he attributes this interpretation to Waddington 1870 who in his commentary to 2196 seems to believe that not only the names refer to one man, but also the both titles given in the text. 33 The title ࡝ͱ͝ࢲͧ͟͢͡ is not very common and denotes a somewhat higher rank than that of tetrarch, if indeed it was a title generated by the Romans, see Schürer 1973/95: I 333f; Sartre 1982b: 125; Isaac 1990: 238 and the discussion in this paper below note 69. 34 If in this case the adjective ͟͝uࢲ͕ͩ͝ could be also connected with ࡝ͱ͝ࢱͧ͟͢͡ so that an ࡝ͱ͝ࢱͧ͟͢͡ ͟͝uࢲ͕ͩ͝ would be implied in the text as well, as Sartre 1982b: 123 suggests (in contradiction to his own assumption in (1982c), see note 32), and if with the latter office was meant “un chef de tribu ou de clan, reconnu par Rome comme tel”, then this would surely be a curious way to identify such a man in a public document probably erected in behalf of the central power – one would expect the name of the tribe he should be the chief of and not only the term ‘chief of nomads’; rather the ͦࢼ͒ͧ͜͟͢͡ ࡕ͚͕ͥ͘͟͝भ͝ of OGIS 617 = Wadd 2236 is the way one might expect the chief of a tribe to be described, but also in this case it is not at all certain that ࡕ͚͕ͥ͘͟͝भ͝ refers to the both nouns, i.e. ͦࢼ͒ͧ͜͟͢͡ as well as to ͔ͣͤ͒ͤ͘͡ࢺ͢ the latter inscription will be discussed below more detailed, see esp. note 46. 35 See in this context the remarks and included examples from Schmidt 2004: 69ff and also Schumacher 1996: 246-258; for the epigraphic habit in Christian times still see in general Marucchi 1911: 37-74.

27 In connection with nomads there are some epigraphic examples at the end of the 1st and beginning of the 2nd century known from the Palmyrene region, which seem to reflect the relationship between Romans and nomads better then the texts here under discussion: The situation emphasized in these inscriptions would be expected as being the ‘normal’ way of the central power in dealing with the nomadic ‘menace’ – but the here mentioned ͖͔ͣͤ͒ͤ͟͡ࢸ ͛͒ͤࢱ ͤभ͝ ͟͝uࢲ͕ͩ͝ should be not subject of this paper, see therefore Gebhardt 2002: 287ff esp. with note 2. 28 Jos. Ant. XIX.354 – XX.213; Id., Bell. II.220; Id., Vit. 32-410 passim; Marcus Julius Agrippa II., 28-93 AD, son of Herod the Great; Schalit 2001: 313-330; Schürer 1973/95: I 471-483; Frankfort 1962: 659-672. 29 OGIS 321 = IGR III 1136 = Wadd 2112 from the Djabal al-Drnjz, Batanaea in the Trachonitis (Eitha or el-HҏƯt), probably from the end of the 1st century AD, found on 2 fragments laying in a distance from about 50 m so that the inscription was surely not in situ; Sartre 1982c: 123-126; Id. 1982b: 123-125; Gebhardt 2002: 249 with note 2. 30 M. Sartre 1982c here 123 rightly concludes that all the text provides is “le contexte militaire (͖ͣ͠ࢸ͘͡) et la mention des nomads”; Wadd 2203 refers to ࡟ͱͥ͟͢͝ ͟͝uࢲ͕ͩ͝ but not to a leader; AAES 383 refers to a ͣࢼ͕͚͛͟͢͝ ͟͝uࢲ͕ͩ͝ with no clue to his duties, though see already Harper 1928: 139.

277

THE LATE ROMAN ARMY IN THE EAST by the word ͟͝uࢲ͕͖͢.36 A name of a tribe or a shaykh should be given in the text.37 Moreover also the syntactic structure of OGIS 616 shows that there is no direct link between the ‘nomads’ and the title ࡝ͱ͝ࢲͧ͟͢͡, because the noun in the genitive plural refers to the ͔ͣͤ͒ͤ͘͡ࢺ͢; to refer it to the other title is only an assumption. So the ͖͔ͣͤ͒ͤ͡ࢹ͢ molÇdym seems to refer to a Roman office on behalf of nomads, while the ࡝ͱ͝ࢲͧ͟͢͡ may be or not a Roman office in connection with nomads. Therefore the question of the title, which was given by the Roman central power to a nomadic chieftain, has still to be answered.

would have been the fact important if the person in the ͖͔ͣͤ͒ͤ͡ࢹ͢ charge was or was not a nomad. Thus the title does not make sense if taken as an (self-) identification of a tribal chief, but makes excellent sense as a Roman military title. One would therefore suggest, that ͔ͣͤ͒ͤ͘͡ࢺ͢ ͖͒͠͡u͓͟͜भ͝ ͟͝uࢲ͕ͩ͝ does not mean ‘chief of nomads’ camps’, but ‘captain of a troop of nomads’.42 Especially the low age, what implies his office not being a very high rank, of the in our text mentioned ͔ͣͤ͒ͤ͘͡ࢺ͢ seems to show that he was the indigenous commander of a smaller unit beneath the level of a ͖͒͠͡u͓͟͜ࢵ – one even could assume that he could have been the chief of his tribe as well.43

A further funeral inscription published in the Princeton Expeditions mentions a ͔ͣͤ͒ͤ͘͡ࢺ͢ ͖͒͠͡u͓͟͜भ͝ ͟͝uࢲ͕ͩ͝.

Thus, although the possibility cannot be excluded that some of these ͔ͣͤ͒ͤ͘͡ࢺ͢ may have been leaders of their tribes, at least the ͔ͣͤ͒ͤ͘͟͡ࢷ ͖͒͠͡u͓͟͜भ͝ ͟͝uࢲ͕ͩ͝ should have been professional soldiers, and one would therefore assume that they are more likely to have been chosen from outside the immediate indigenous environment, i.e. they would have been probably Romanized tribesmen, at least partially firm in Roman military services already. The mentioning of unspecific ͒͠͡ࢴu͓͚͒͟͜ ͟͝uࢲ͕ͩ͝ without the name of a tribe leads to a ‘summarizing’ office, which could have bundled detachments of several tribes, as just mentioned. Such ͒͠͡ࢴu͓͚͒͟͜ therefore would have consisted of some smaller units, which probably were not commanded by Roman or Romanized officers, but rather by members of the tribes they were recruited from. Only leaders with ‘tribal’ reputation would have had the necessary authority in their own tribal environment.44

[࢐ ͕͖ऑ͒͝] | ͣͤ͒ͤ͘͡ | [͔]ࢹ͢ ͖͒͠͡ | [u]͓͟͜भ͝ | [͝]͟uࢲ͕ͩ | [͝], ࡝ͤभ͝ | ͛͗ऴ. [࡙]͖ͩ͡ ࡑ | [͜]͚͖͟͠ ͧ | ͒ऑ͖͡.38 ͖͒͠͡u͓͟͜ࢵ means ‘camp’ as well as ‘troop, army, host’ or ‘a section of an army’.39 The last meaning seems to make the best sense in this case. If here was meant an indigenous chief remains uncertain again, because, as already seen above, one would expect at least his tribe’s name and not only the general term ͟͝uࢲ͕͖͢. Again, for him as for the rest of his tribesmen, what identified them would not have been the fact that they were nomads, but their kin-group.40 On the other hand to any central power the mostly distinctive topic about these tribesmen would have been the fact that they were nomads, the names of their tribes or sections being more or less irrelevant.41 For the Romans, who gave these titles to the natives, just

Actually there are known some other Greek inscriptions from the Djabal HҏaurƗn, which show that the nomads indeed used the names of their tribes, but just in official texts.45 This becomes clear in the following inscription,

36

See also above note 34. The partly very long family trees of the so called ‘Safaïtic’ inscriptions prove this for the HҏaurƗn clearly, see MacDonald (1993) 323-352 and id. (1992) 303-307; this cannot be discussed here in detail. 38 PUAES III A 752 (Malka), dated roughly to the end of the 2nd or the beginning of the 3rd century AD; MacDonald 1993: 374f; Sartre’s assumption 1982b: 124 that this ͔ͣͤ͒ͤ͘͡ࢺ͢ was the chief of encampments of nomads and Malka in the Djabal HҏaurƗn “a dû être sinon le centre unique des estivages nomads, du moins le site où s’instaillait traditionnellement le chef des tribus passant l’été dans Nord du Djabal al-DrƗz, chef désigné ici par le titre de stratège” has to be rejected, because it is not only nearly impossible to say what this man’s duties could have been, but also to clarify, if he was a tribal chief at all. 39 See slightly different MacDonald 1993: 374; also Liddell/Scott 1996 s.v. ͖͒͠͡u͓͟͜ࢵ. 40 It has to be strongly emphasised again in this respect that for the shaykh of a tribe or indeed for any tribesman identity is embodied in the group of which he is a member; to refer to the ‘chief of a tribe of nomads’ in this way would be simply, as MacDonald 1993: 371 puts it “So-and-so son of So-and-so, a bedouin chief”; this is although highly probable true for the nomads of the HҏaurƗn, who wrote the Safaïtic texts; those inscriptions are full of self-identification by genealogy and social groups; a word for ‘nomad’ has not been found so far, see ibid. passim, different Jamme 1970: esp. 510. 41 However, even in this case it could not be totally rejected that the deceased Roman officer was a tribal chieftain as well and that his Greek title would have been mentioned in the lost part of the text. But even then his title would have referred to the duties on behalf of the government, not to the status within the tribe. That the 27 year old donator of the inscription has been a Romanized tribesman derives from his (Roman) military rank clearly. 37

42 In the same sense also Sartre 1982b: 124f; MacDonald 1993: 375ff and in the editio princes, PUAES III A 752, it was already translated as “general of armies of nomads”. 43 Actually there is no evidence before the 4th century that leaders of bigger confederations of Arabic nomads were ‘involved’ personally in Roman military services; the ‘first’ prove in this respect is an inscription from 328, which refers to a ‘lord of the Arabs’, Sartre 1982b: 134 with note 50; found in Namara, north-east of the Jabal Drnjz, it is written in Nabataean characters: Imru’l-quais provided, however, soldiers to the Roman central power but did not command them personally; he seems to have been a loyal ally of the Romans, see Isaac 1990: 239f and Shahid 1984: 46 for the recent discussion. 44 MacDonald emphasises a different view 1993: 376 assuming that “while we cannot exclude the possibility that commanders of these units may have been leaders of the ’ls from which they were recruited, military discipline and effectiveness would surely have required them to be officered by professional soldiers”; as we have already seen above in OGIS 321 = Wadd 2112, and MacDonald 1993 passim generally agrees with that, such commanders of indigenous units could be drawn from Romanized members of ‘nomadic’ tribes – they do not need to be Roman citizens; in this respect convincingly Sartre 1982b: 124. 45 There is only one Safaïtic inscription known, where the author identifies himself, i.e. by his profession as a horseman, rather than by his tribe. This exception from the rule symbolizes the change in selfperception from the purely tribal to the professional sphere, Ms 64; this example was found in the HҏaurƗn region of Ruwayšid; it says following the translation of MacDonald 1993, “By ‘qrb son of ’bgr, a horseman in

278

THOMAS BRÜGGEMANN: ࡣͱ͝ࢲͧ͟͢͡, ͇ࢼ͒ͧ͜͟͢͡

AND

͔̈́ͤ͒ͤ͘͡ࢹ͢ ͟͝uࢲ͕ͩ͝ IN ROMAN ARABIA (1ST – 3RD CENTURY)

other title – as mentioned above in a different context – is just an unreliable possibility. If the here mentioned ͖͔ͣͤ͒ͤ͡ࢹ͢ ࡕ͚͕ͥ͘͟͝भ͝ has been the commander of an ethnic unit from the same tribe, could this unit then have been one of several lower detachments of a ͖͒͠͡u͓͟͜ࢵ ͟͝uࢲ͕ͩ͝? Unfortunately this has to remain a hypothesis. The ͔ͣͤ͒ͤ͘͡ࢺ͢ in the context of our inscriptions is surely not a translation of a native title but a specific ‘new’ office within the Roman administrative hierarchy, which made the respective officeholder a ‘functionary’, who was defined by the Roman administration qua ‘duty’ and not by ‘personality’. Therefore it is quite logical that the epigraphic evidence does not show, whether these men were indigenous individuals or not. In this respect an indigenous name can serve only as a hint to the origin (decent) of the holder of such a title and to nothing else.49 The Greek inscriptions, so far, looked at the nomads from a perspective from which they seem just simply lumped together under the collective notion ‘nomads’ as an undistinguished mass of people, rather than as members of social groups with distinctive names, but such groups are also well known in the HҏaurƗn.50 Our next inscription seems at least to insinuate the existence of concrete tribes in the Djabal HҏaurƗn:

which offers perhaps the opportunity to identify the ethnic environment by giving an indigenous tribal name in context with the title ͔ͣͤ͒ͤ͘͡ࢺ͢. This example has to be called remarkable, because in the Safaïtic graffiti, i.e. texts for the ‘internal use’ of these tribes, the connection of personal names and profession is not common at all. That means that the representation of a person by name and profession has to be seen as particular way expressing the loyalty with the Roman central power. ࢏͕͚͒͝ࢲͱफ ̈́͒ͥ͟ࢲ͕ͥ͟ | ͔ͣͤ͒ͤ͘͡ࢶ͚ͣ͒ͤ͝ ࡕ͚ͥ͟ | ͕͘͝भ͝ ͛(͒ࢷ) ͦ(ͥ)͜(͒ͧ͡ࢶͣ)͒(͝) | ͚ͤ ̺͟u͒͜ࢴͧ͘ ͔ͥ͝ࢵ ͛(͒ࢷ) ̈́ | ͕͒ͥ͟ࢺ͢ ͒ͤ͠ࢵ͡ ͒[࢕ͤ]͟ठ ࡍ͝ࢴͣͤͣ͒͘͝.46 The tribe of the ࡕ͚͕ͥ͟ः͚͟͝ is not mentioned outside this inscription, although it sometimes has been identified with a similar spelled, in the Djabal HҏaurƗn epigraphically much better documented and obviously larger tribe. This ͦͥ͜ࢶ ࡕ͕ͥ͘͝भ͝ is mentioned, for example, in ̳࢕͡. ࡕ[ͥ͟ऑ]ͤ͟͢ ࡕͤ[͚ͤ͛͟ठ u͘]ͤ͛ͩ͟͡uࢸ͒͢ ̴͖ͧ͟͡ࢱͱ ͓̈́͒ࢲͩ͝ ࡺ͖͡ࢺͤ[͟] | u͖ࢻ͢ ͟ࡹ͕͛͟ࢺu͖ͣ͘͝ ͦͥ͜ः[͢ ̳]࢕͕͘͝[भ͝].47

[ͅࢹ͝ ͕͖ऑ͒͝ ͖ͣ͠͡]͓(͖ͥͤࢵ͝) ͖͓̈́(͒ͣͤ͟ठ) ࡍ͚ͤͣͤ͝͡ࢲ(͔ͤ͘͟͝) ͟ࡺ ࡍ͠ࢹ ࡡͱͥ͟͢͝ | ͟͝uࢲ͕ͩ͝, ࡎ͔͖͝ࢸ͒͢ ͧࢲ͚͡͝.51

The latter inscription may only show that there is a problem in identifying the ‘same’ name, given that a great many Semitic names are distinguished one from the other only by differences in vocalization of the same consonantal frame, or by the doubling of consonants. Therefore it is obviously doubtful, if not impossible, to tell, even in a Greek transcription, if the ࡕ͚͕ͥ͟ः͚͟͝ and the ࡕ͕ͥ͘͟͝ࢸ were the same tribes.48 Moreover OGIS 617 again shows that there is no direct link between the ࡕ͚͕ͥ͟ः͚͟͝ and the title ͦࢼ͒ͧ͜͟͢͡, because the noun in the genitive plural refers just to the title ͔ͣͤ͒ͤ͘͡ࢺ͢. Thus only the tribe and the office of the ͔ͣͤ͒ͤ͘͡ࢺ͢ are syntactically connected and the former’s extension to the

The text seems to corroborate that ͟ࡺ ࡍ͠ࢹ ࡟ͱͥ͟͢͝ ͟͝uࢲ͕ͩ͝ might express an individual tribe of the HawrƗn. The expression is part of an inscription honouring a senatoric governor, ࡍ͚ͤͣͤ͝͡ࢲ͔ͤ͘͟͢, and was found in the village of Tarba in the northeast of the Djabal HҏaurƗn. Those setting up such a text would surely have had referred to themselves in a way that would ensure that they were identified by the Roman central power. But a ‘nation (or ‘tribe’) of nomads’ is a very imprecise self-description and therefore would be absolutely senseless from the indigenous perspective.52 It has a sense, however, if one cautiously suggests that the term could be a technical one, within the Roman provincial administration in the HҏaurƗn, referring perhaps

the unit of the ’l ‘mrt”; it seems to be comparable to IGR III 1257 = Wadd 2271 ͖̾ͣ͒uࢲ͟͢͡ ࡺ͖͠͠ࢻ͢ ̼ͥ͡(͚͒͛͘͝ः͢) ͔ࢴ͟͝[͢] ͓̿͒ࢱ͢; here has to be agreed with MacAdam 1986: 111, 145 and MacDonald 1993: 374 and Id. 1991: 106 with note 38, who both reject the traditional interpretation of the last term meaning ‘the Nabatean race’ and suggest that the erector is just identifying his clan, see also note 22. 46 OGIS 617 = Wadd 2236 from RƗma, between Mušannaf and Tarba; 3rd century AD; the first correction of the text was made by ClermontGanneau 1903: II 147f; for the different interpretations of this text see above notes 30 and 34 as well. 47 Wadd 2396b of unkown date; it should be emphasized that SabÇym is the name of a village in the Trachonitis, see already the commentary of Waddington. 48 Sartre, Tribus, 80 has distinguished rightly the ࡕ͚͕ͥ͟ः͚͟͝ from the ࡕ͕ͥ͘͟͝ࢸ, when he notices that the personal names ࡕ͕ͥः and ࡕ͕ͥࢺ͢ are well known in the Greek evidence of the region, but due to the fact that ‘Awd, ‘AwdƗ and ‘Awÿ are transcribed in the same way, one has to be cautious to sump up these two terms to one tribe; by no means this observation could be explained in the manner of MacAdam’s 1986: 121 hypothesis that “at some point the ࡕ͕ͥ͘͟͝ࢸ split, since two members live in villages some distance apart. This undoubtly happened to other tribes [as well] and was a normal part of the sedentarisation process” since there is absolutely no convincing evidence for such a process in the HҏaurƗn at all, as MacDonald 1993: 346-352 is able to point out.

49

This due to the fact that there is the problem of identifying the ‘same’ name in Greek, Nabataen, or Safaïtic inscriptions, see MacDonald 1993: 353ff; thus a name in a Greek inscription which seems to be a possible parallel to a Safaïtic never should lead to too obvious assignations; see above note 48. 50 Forms of social organization that were called ‘tribal’ in antiquity were not restricted to nomads so that not all groups called ‘tribes’ were necessarily organized in an identical way, as Sartre 1982a Tribus, 86 points out, when he suggests for the HҏaurƗn that “lors de la création des nouvelles cités dans le HҏaurƗn , certaines tribus bien représentées dans la population de telle ou telle cité ont dû donner leur nom à une tribu civile” and 87 “les ͦͥ͒͜ࢸ peuvent être des tribus indigènes, c’est-à-dire le type de groupement traditionnel des populations hauranaises”; in this respect see also MacDonald 1993: 310 with note 47. 51 Wadd 2203 from Tharba in the northeast of the Djabal HҏaurƗn, 1st half of the 2nd century AD; the text honouring the governor of the province, was written on a little column, found not in situ; see although notes 16 and 30 above. 52 See above remarks and notes 34 and esp. 40.

279

THE LATE ROMAN ARMY IN THE EAST to specific units recruited from among the nomads and used for particular purposes, which unfortunately are not known. So the ࡍ͚ͤͣͤ͝͡ࢲ͔ͤ͘͟͢, the propraetor, might have supervised the commanders of several indigenous auxiliary units inside his province, in this view comparable with the ͔ͣͤ͒ͤ͘͡ࢺ͢ ͖͒͠͡u͓͟͜भ͝ ͟͝uࢲ͕ͩ͝. Especially in this respect, i.e. the semi-official (diplomatic) methods of the Roman administration in dealing with nomadic populations in the region, a bilingual inscription, in Greek and Nabataean, found almost 200 km south-east of Aela, at Ruwwafa in the Hejaz seems to be particularly important as well.53 The inscription gives evidence for the activities of the governor of Arabia well beyond the provincial boundaries. It shows, as the texts under discussion in this contribution, that the Roman central power had allies which recognized its suzerainty without themselves forming part of the provincial administrative structure, a relationship often thought to belong the Byzantine period.54

͇ͥ͒ͧ͜͟͡ࢸ, insofar as they were Arab tribal chieftains, were not appointed by government officials, but acquired the title by right of succession. In dealing with indigenous tribes the term meant a tribal chieftain, a Sayyid or Shaykh. It is doubtful, whether the Roman government, on its own authority, could place tribes under the leadership of a ͦࢼ͒ͧ͜͟͢͡ without the willingness of the tribal leaders themselves.58 That this is true can be shown with a late example: Justinian (530/531), in an attempt to counteract the military successes of Mundhir, a Saracen ally of the Persians, decided in Procopius’ words, “to put in command of as many tribes as possible Arethas, the son of Gabalas, who ruled over the Saracens of Arabia, and bestowed upon him the dignity of king”.59 The statement of Procopius that Arethas was placed in command of “as many tribes as possible” is an indication that Arethas had to negotiate his position with other tribal leaders who, for a variety of reasons, especially the hope for material benefits, may or may not have agreed to be under him. The idea that Arethas could exercise effective control over all Saracen tribes federated with Rome, is difficult to accept if one considers the geographical distribution of Arab tribes over the vast eastern frontier.60 From Justinian’s point of view, his primary objective was to amass an army of Saracens big enough to neutralize the activities of Mundhir in the area of the Persian conflict. It is in this connection that the political administration considered the chieftain the ࡍͧ͡ࢺ͢ of his ͦͥ͜ࢶ (tribe(s)).61 But in a more general sense, a ͦࢼ͒ͧ͜͟͢͡ was just the head of a group of families, or of a body of people united by ties of descent from a common ancestor, or he was an official in charge of a tribal unit of operation including a political dimension. During late antiquity and Byzantine times the term had no precise specific meaning that would limit its use to a particular organized group of people.62 The office of the ࡝ͱ͝ࢲͧ͟͢͡ has been a traditional one in the Near East as well, but proved to be too wide and too great for a nomadic phenomenon, i.e. the connotations

The ͦࢼ͒ͧ͜͟͢͡ was the only of the three offices in our inscriptions, which im-plicitly included if not a direct ‘nomadic’ at least a ‘tribal’ feature. The ͦࢼ͒ͧ͜͟͢͡ has been the leader of a ͦͥ͜ࢶ in Athens, the tribunus, the emir, the Hebrew,55 and last but not least also the Arabic chieftain.56 It should be emphasized, that the term, at least during the period of our inscriptions, was not yet part of the official terminology of the Roman government.57 53 AE 1977, 834; the inscription records the building of a temple by the people, or the confederation, of the Thamud in honour of Marcus Aurelius and Lucius Verus, under the auspices of the governor Claudius Modestus, legate of Arabia 167-169; the Nabataen text mentions his predecessor Antistius Adventus as having made peace, presumably in an intertribal dispute, Sartre 1982b: 84; for the detailed debate of the inscription see Graf 1988: 171-211 with further literature; Gebhardt 2002: 100 with note 5; Stoll 2001 passim. 54 It may further provide evidence of the existence of a nomad confederation such as were formed from time to time in the pre-Islamic period, Graf 1988 passim. 55 In classical and Hellenistic times this office in many Greek ͠ࢺ͖͚͜͢ was the title of the leaders of ͦͥ͒͜ࢸ; only in Athens this office had already in the 6th and 5th century B.C.E. military implications; in indigenous contexts the ͦࢼ͒ͧ͜͟͢͡ title was used as the Greek term for Hebrew tribal chiefs, see for example Dt XX.31.28. 56 As we have seen above, note 7, Strab. XVI.1.28 uses the term in his description of the first alliance between Arabic nomads and the Romans in the region. 57 The use of the term in Late Antiquity and Byzantine is not subject of this paper and, besides, it is still not quite clear: Sartre 1982b: 150ff is aware of the ambiguity in the term since it was used to denote a tribal chief not only of the Arab nomads, the Saracens, but of other tribal organizations, when he points out that “il est des cas où ce titre, reconnu par Rome, acquiert un valeur officiel et le phylarchat devient un titre romain”; a survey of the appearance of the word ͦࢼ͒ͧ͜͟͢͡ is presented by F. Gschnitzer, s. v. ͇ࢼ͒ͧ͜͟͢͡, RE Suppl. 11, 1067-1090, but he does not enter into any discussion of the term as an official administrative title; this does, however, I. Shahid 1984: 31 when he says, that the transliteration of the “Greek word into Latin phylarchus reserved the use of the term to the Arab chief and this contributed to the emergence of the phylarchate as a distinctive military office”; it has to agreed with MacDonald 1993: 371 that “the titles ethnarch and phylarch in these [indigenous] contexts are not ‘translations’ of native titles but are specific offices within the Roman administrative hierarchy”; brief remarks about the controversial debate also in MacAdam 1994: 49-93; Mayerson 1991: 291-295.

58

Ibid. passim; see in this respect also MacDonald 1993: 376 who convincingly points out that “a tribal shaykh can only lead by persuasion, other than in exceptional circumstances; the chief is no more than primus inter pares and he does not ‘rule’ but can only persuade or inspire”; Lancaster 1981: 87-91; MacDonald 1991: 104f. 59 Proc. Bell. I.17.47; see Shahid 1984: 20 and 46; the view of Mayerson 1991: 292, that the ͦࢼ͒ͧ͜͟͢͡ “was not part of the official terminology of the Roman government, and that the word means, in dealing with Saracens, a tribal chieftain, a sayyid or a shaykh or a malik (king)” has to be rejected since one cannot exclude, however, the possibility that the commanders of native units may have been leaders of the (their) clans from which they were recruited at the same time, see above note 44. 60 Compare in this respect Schmitt 2004: 859-877, who convincingly points out that there is no evidence for any important confederation of the Saracens allied to the Romans in the 4th century; furthermore MacAdam 1994: 49ff and Mayerson 1991: 291ff. 61 For such ‘agreements’ between Romans and indigenous Arab tribes see for example Theoph. Chron. 336 (de Boor); Nik. Opusc. Hist. 23 (de Boor). 62 But as already mentioned above, the late antique and Byzantine implications and the development of the office are not subject of this paper; the writer is planning soon to deal with those topics, i.e. the shifting disposition of the ͦࢼ͒ͧ͜͟͢͡, in a different setting; for the moment see MacAdam 1994: 49ff; Mayerson 1991: 292-295.

280

THOMAS BRÜGGEMANN: ࡣͱ͝ࢲͧ͟͢͡, ͇ࢼ͒ͧ͜͟͢͡

AND

͔̈́ͤ͒ͤ͘͡ࢹ͢ ͟͝uࢲ͕ͩ͝ IN ROMAN ARABIA (1ST – 3RD CENTURY)

title,69 but an instrument of the Roman rule, which Caesar did not want Hyrcanus II to be just seven years before. Herod became one of several client-kings established in the eastern part of the empire by Romans. Although the title ‘king’ represented a formal recognition compared to that of an ࡝ͱ͝ࢲͧ͟͢͡, he nevertheless was no longer an outstanding Jewish ruler, as Hyrcanus II, but just a ‘king’ among others. After Herod’s death in Josephus’ description of the appointment of Herod’s son Archelaus it becomes apparent again, that the Roman central power in the first century B.C.E. did not yet found a clear concept in dealing with Jewish rulers. Than Augustus “appointed Archelaus not king indeed, but ethnarch of half of the territory that had been subject to Herod, and promised to reward him with the title of king if he really proved able to act in that capacity”.70 Compared with origin and use of the title ࡝ͱ͝ࢲͧ͟͢͡, especially in the region’s Jewish context, the segmentary structures of the Djabal īaurƗn and its tribal societies would have made necessary several tiny ࡝ͱ͝ࢲͧ͟͡i. The term proved to be unfit for the planned purposes quickly and therefore could have been given up from the middle of the 2nd century onwards.71

linked to this title did not fit with the position of a local tribal chieftain. The ࡝ͱ͝ࢲͧ͟͢͡ was well known in Judaea in Roman times, at least since the title was given to John Hyrcanus II. and his sons by official decree of Caesar in 47 B.C.E. in addition to the office of High Priest.63 Already some twenty years before in 63 B.C.E. “Pompey restored the high priesthood to Hyrcanus [II.] and permitted him to have the leadership of the nation”,64 he nevertheless “forbade him to wear a diadem”.65 Even if the title of the ࡝ͱ͝ࢲͧ͟͢͡ within the Roman sphere was introduced not later than the time of Caesar, its meaning, which excluded the title or the rights of a king, remained the same as the ‘leadership of the nation’ Pompey permitted Hyrcanus before.66 This office and the High Priesthood were both traditional Jewish institutions of power. So both offices and titles were not created by Caesar following the common idea of transferring Roman institutions to conquered regions so that they might serve the central power as instruments of its rule, but he restored ancient institutions, which may have been traditionally given to the Jews by other big powers before. By conceding Hyrcanus the rank of an ࡝ͱ͝ࢲͧ͟͢͡ of the Jews, i.e. a ‘prince of the God-Folk’, as the corresponding Hebrew title should be read,67 Caesar accepted the High Priest in a political position again, in which the latter’s ancestors had once been before Judaea had fallen under Roman rule.68 In this respect the position of Hyrcanus was based on a native Jewish fundament. By allowing the political head of the Jews to be ࡝ͱ͝ࢲͧ͟͢͡ again the people’s autonomy under foreign rule became accepted as well. After Herod had been appointed king in 40 B.C.E. by Antonius, he was as a rex socius et amicus populi Romani no more a ruler with a traditional Jewish

However, with regard to the military situation the Romans from the end of the 1st century no longer confined to uphold existing structures, but began to implement ‘new’ ones. They intended to use the military potential, which was given by recruiting indigenous men. Therefore the ͔ͣͤ͒ͤ͘͡ࢹ͢ ͟͝uࢲ͕ͩ͝ is the most often mentioned office in the inscriptions cited here. His duties and his position in between the Roman provincial hierarchy are, however, hardly to clarify. Nevertheless the question arises, if the supposed ‘higher’ ͖͔ͣͤ͒ͤ͟͡ࢸ ͖͒͠͡u͓͟͜भ͝ ͟͝uࢲ͕ͩ͝ could have been in the first place no regular military commanders but men appointed by

63

Jos. Ant. XIV, 194 …͕͚ࢱ ͤ͒ࢼͤ͒͢ ͤࢱ͢ ͒ࡹͤࢸ͒͢ ࢝͛͒͡͝ࢹ͝ ࡕ͖͜͞ࢲ͕ͥ͟͝͡ ͛͒ࢷ ͤࢱ ͤࢴ͛ͤ͒ ͒࢕ͤ͟ठ ࡝ͱ͝ࢲͧ͒͢͡ ࢁ͕ͥ͒͟ࢸͩ͝ ͖ࡿ͚͒͝ ࡍ͚͖ͧͩͣ͡͡ࢼ͘͝͝ ͖ͤ ࢁ͕ͥ͒͟ࢸͩ͝ ͕͚ࢱ ͒ͤ͠͝ࢹ͢ ࡡ͖͚ͧ͝ ͛͒ͤࢱ ͤࢱ ͠ࢲ͚ͤ͒͡ ࡡͱ͘ …; Schalit, Herodes, 223-256; furthermore Schäfer 1983 esp. 97 and 116; Schürer 1973/95: I 330-335. 64 Pompey made Hyrkanos High Priest again, Jos. Ant. XIV 73, and appointed him to be leader of the nation but took away from him the dignity being King, Schalit 2001: 14f; see although Jos. Ant. XX 244 …ͤम ͕’ ࢝͛͒͡͝म ͠ࢲ͚͜͝ ͤࢵ͝ ࡍ͚͖ͧͩͣ͡͡ࢼ͘͝͝ ࡍ͕͟͠ࢻ͢ ͤࢵ͝ uࢳ͝ ͤ͟ठ ࡡͱͥ͟͢͝ ͣͤ͒ͣ͟͠͡ࢸ͒͝ ࡝͠ࢴ͖͖ͤͨ͡͝… 65 Jos. Ant. XX 244 …͕͚ࢲ͕͖u͒ ͕ࢳ ͖ͦ͟͡ऑ͝ ࡝͛ࢾ͖ͥͣ͜͝; Pompey, as Schalit 2001: 15 puts it “begnügte sich im Jahre 63 v. Chr. […] mit der Einrichtung einer Verwaltung, die eine Zwischenstellung zwischen Selbst- und unmittelbarer Provinzialverwaltung einnahm und dem Lande einen Schein von Freiheit ließ, nicht von politischer Freiheit, aber von Selbstherrschaft in gewissen Grenzen. Diese Ordnung war als Vorstufe zu einer späteren Provinzialherrschaft gedacht”; in this respect Id. 1937: 4-6. 66 See Schalit 2001: 224f and above note 63. 67 The commonly accepted translation of ࡝ͱ͝ࢲͧ͟͢͡ as ‘prince of the God-Folk’, ʬʠ ʭʲ ʸʹ, wich was made by Schürer 1973/95: I 249f, is rejected by Schalit 2001: 781-787; in our discussion the exact translation is less important then the fact that a traditional institution related to the Jews was revived by Caesar. 68 Jos. Ant. XIV.192ff; it has to be emphasised, however, that “indessen hier das Rechtsprinzip nie auf jene Spitze getrieben worden ist, wo es die nationale Existenzberechtigung negiert; vielmehr ist die tunlichste Schonung der Stammeseigentümlichkeiten einer der Grundsätze, in welchen die Römer ihren Frieden mit den unterworfenen Völkerschaften suchten. Bezeichnend ist hierfür die weitgehende Rücksichtnahme, welche den nationalen Ansprüchen der Juden entgegengebracht wurde”, so already rightly Mitteis 1871: 90.

69 The title ࡝ͱ͝ࢲͧ͟͢͡ evidently had become a rank somewhat higher than that of ͖ͤͤ͡ࢲͧ͟͢͡; in this respect the Hasmonean princes bore the former before assuming the royal title, see I Mac. XIV.47 and XV.1f; Herod the Great and his brother Phasael had already received it from Antonius, see Jos. Ant. XIV 326, see above note 33. 70 Jos. Ant. XVII.317 …ࡕͧ͡ࢴ͒͜͟͝ ͓͚͒ͣ͜ࢴ͒ uࢳ͝ ͟࢕͛ ࡍͦ͒͟͠ࢸ͖͚ͤ͒͝, ͤ͟ठ ͕’ ࡪuࢸ͖ͣ͟͢ ͤः͢ ͧࢾ͒͢͡ ࡮͖͠͡ ࡲ͡ࢾ͕ँ ࢖͖ͤ͠ࢴ͖͚͜ ࡝ͱ͝ࢲͧ͘͡͝ ͛͒ͱࢸ͚ͣͤ͒ͤ͒, ͚ͤuࢶ͖͚ͣ͝ ࡍ͚͞ࢾu͚͒ͤ ͓͚͖͒ͣ͜ࢸ͒͢ ࢖͚ͣͧ͟͠͝ࢼu͖͟͢͝, ͖ࡽ͖͠͡ ͤࢵ͝ ͖ࡹ͢ ͒࢕ͤࢵ͝ ࡍ͖ͤ͡ࢵ͝ ͣͦ͟͠͡ࢴ͚ͤ͟͟͡; the writer has to thank Israel Shatzman, Jerusalem, for the hint regarding the ongoing use of title ͓͚͖͒ͣ͜ࢼ͢ u͖͔ࢲ͜͟͢ by Herod’s sons, as the example of Marcus Iulius Agrippa in OGIS 321 = Wadd 2112 shows, a long time after Iudaea had become dependent on Rome, although being a Roman ally Agrippa II. had a special rank as ͚ͦ͛͒͜͟ऑͣ͒͡ and ͚ͦͩ͜͟͡u͒ऑ͟͢, OGIS 419-424, which permitted him the use of the other title as well. 71 In the Near East the title ࡝ͱ͝ࢲͧ͟͢͡ seems normally to mean simply the ruler of a particular ethnic group under Seleucid and later Roman authority; the nature of the ethnic group, the type of leadership and the relationship with the imperial power seem to have varied considerably, and there is nothing in the title to give it a particular association with nomads, so that in this respect the assumption of MacDonald 1991: 114f has to be rejected; see therefore rightly Schalit 2001: 225ff about the title in Jewish context; after the successful political integration of Judaea into the administrative structures of the Roman Arabia, i.e. supposedly beginning with the annexation of the princedom of Agrippa II., the title seems to have kept only his religious function, see Noethlichs 2001: 207.

281

THE LATE ROMAN ARMY IN THE EAST the Roman administration to act as ‘connecting officers’ between the government and the tribes, among whose principle duties may have been to maintain good relations between the nomads and the Roman authorities and, possibly, the provision of auxiliary troops from the tribes.72 It has to be emphasized again that such men needed to have the acceptance of almost all members of a tribal section or confederation, because for the Roman authority they were responsible for the behaviour of these groups and moreover, they were the (often only direct) ‘channel of communication’ between central power and native populations.73

Near East, the ࡝ͱ͝ࢲͧ͟͢͡ and the ͦࢼ͒ͧ͜͟͢͡. With the ‘new’ structure of the ͔ͣͤ͒ͤ͘͡ࢹ͢ ͟͝uࢲ͕ͩ͝ the central power tried to satisfy its own military needs during the period of pacifying the Trachonitis from the end of the 1st century onwards. This is the only total new office, only for this office the texts prove the connection to the ‘nomads’ even syntactically doubtless. From the beginning of the 3rd century finally only the ͦࢼ͒ͧ͜͟͢͡ remained, and shifted from a more civil to a military office.76 Consequently the Romans gave up the office and title ͔ͣͤ͒ͤ͘͡ࢹ͢ ͟͝uࢲ͕ͩ͝ after it served its needs in appeasing the Trachonitian trouble spot at the beginning of the 3rd century. The total disappearance of the two other titles in this period is an indication not only for the integration of the tribes of the HҝaurƗn, but also of the whole area into the Roman Empire.

The combined view on the three offices ࡝ͱ͝ࢲͧ͟͢͡, ͦࢼ͒ͧ͜͟͢͡ and ͔ͣͤ͒ͤ͘͡ࢹ͢ ͟͝uࢲ͕ͩ͝ shows that the Roman central power tried to integrate those nomadic tribes, whose migration routes were totally on the territory of the empire, generally into the provincial administration, but did not force the indigenous tribes to give up their traditional forms of life.74 The area in the north-east corner of the Djabal īaurƗn must by no means therefore be called a ‘reservation’, in which native populations should be settled and sedentarized. Even in Roman times this territory remains the traditional environment of nomadic (Safaïtic) tribes.75 Thus Roman politics did not seem to push the assimilation of these populations. The implementation of the ࡝ͱ͝ࢲͧ͟͢͡, ͦࢼ͒ͧ͜͟͢͡ and ͔ͣͤ͒ͤ͘͡ࢹ͢ ͟͝uࢲ͕ͩ͝ as well served by no means the settlement respectively the sedentarization of nomadizing tribes, but it was the way, how Romans reacted on local peculiarities without disturbing the indigenous ethnicities in their traditional way of life at all.

Bibliography Akkermans P.M.M.G. and Schwartz G.M. 2003 The Archaeology of Syria. From Complex HunterGatherers to Early Urban Societies (ca. 16000-300 B.C.), Cambridge. Ball W. 2000 Rome in the East. The Transformation of an Empire, London. Bauzou Th. 1990 Les voies romaines entre Damas et Amman. In Gatier P.-L. et al. (ed.), Géographie historique au Proche-Orient (Syrie, Phénicie, Arabie, grecques, romaines, byzantines). Actes de la Table Ronde de Valbonne, 16-18 septembre 1985, Paris: 292-300. Bauzou Th. 1989 Les routes romaines de Syrie. In Denzer J.-M. et al. (ed.), Archéologie et histoire de la Syrie II: La Syrie de l’époque achéménide à l’avènement de l’Islam, Saarbrücken: 205-221. Bowersock G.W. 1970 The Annexation and Initial Garrison of Arabia. In ZPE 5: 37-47. Bowersock G.W. 1983 Roman Arabia, Cambridge. Brüggemann Th. 2004 Nundinae als Bindeglied zwischen römischer Administration und indigenen Gesellschaften im antiken Nordafrika. In Streck B. (ed.), Segmentation und Komplementarität – Organisatorische, ökonomische und kulturelle Aspekte der Interaktion von Nomaden und Sesshaften, Halle: 156-187. Brünnow R. 1899 Reisebericht 1898. In Mitteilungen und Nachrichten des Deutschen Palästina-Vereins 5: 81-91. Clermont-Ganneau C. 1903 Recueil d’archéologie orientale. Vol 5, Paris. Dunand M. 1933 La voie romaine du Ledja. In Mémoires présentes par divers savants à l’Académie des Inscriptions et Belles- Lettres 13 de l’Institut de France, Paris: 521-556. Eadie J.W. 1985 Artifacts and Annexation: Trajan’s Grand Strategy and Arabia. In Id. et al. (ed.), The Craft of the Ancient Historians. Essays in Honour of

The poor Greek epigraphic documentation from the north-east Djabal īaurƗn , which gives only evidence for the offices ࡝ͱ͝ࢲͧ͟͢͡, ͦࢼ͒ͧ͜͟͢͡ and ͔ͣͤ͒ͤ͘͡ࢹ͢ ͟͝uࢲ͕ͩ͝, shows for the period from the 2nd to the 3rd century AD the Roman striving to find a modus vivendi with the segmentary societies of the region. The Romans tried this at first by using two titles well known in the 72

This practise could be observed in recent times when European powers, holding more or less nominal control over large areas of desert in the Middle East and North Africa, realized the need for ‘liaison’ with, in the hope of control over, the nomadic tribes in and on the edge of their spheres of influence; such duties were virtually always performed by men who were not nomads but who established good relations with them: This was the task, which Glubb 1983: 63-105 describes clearly he had under the British Mandates in both Iraq and Transjordan; from the point-of-view of the authorities he was “in charge of the Bedouin”. 73 This was true for Roman North Africa although and is well documented for this region of the empire, as esp. the responsibility of the tribal chief concerning the relationship between central power and nomadic Baquates in the Mauretania Tingitana can show, see Kuhoff 1993: 55-71 and Weiß 2004: 1-20; Id., 2006: forthcoming; Brüggemann 2004: 156-187. 74 Ibid. passim; comparable in this respect is again the situation of Roman North Africa during imperial times, when the provincial administration never tried to sedentarize the indigenous nomadic tribes; the Romans even seem to have ‘protected’ the migration routes of these pastoralist against the expanding agricultural landscapes in the 3rd and 4th, as Gutsfeld 1989: 166-176 could point out convincingly. 75 See MacDonald 1993: 346-352.

76

282

See Mayerson 1991: 292-295 and above note 39.

THOMAS BRÜGGEMANN: ࡣͱ͝ࢲͧ͟͢͡, ͇ࢼ͒ͧ͜͟͢͡

AND

͔̈́ͤ͒ͤ͘͡ࢹ͢ ͟͝uࢲ͕ͩ͝ IN ROMAN ARABIA (1ST – 3RD CENTURY)

culturel. Leiden: 289-320. MacAdam H.I. 1994 Settlements and Settlement Patterns in the Northern and Central Transjordania, ca. 550ca. 750. In King G.R.D. et al. (ed.), The Byzantine and Early Islamic Near East, Vol. II: Land Use and Settlement Patterns, Princeton: 49-93. MacDonald M.C.A. 1991 Was the Nabatean Kingdom a ‘Bedouin State’? In ZDPV 107: 102-119. MacDonald M.C.A. 1992 The distribution of Safaitic Inscriptions in Northern Jordan. In Zaghloul M. et al. (ed.), Studies in the History and Archaeology of Jordan IV, Amman: 303-307. MacDonald M.C.A. 1993 Nomads and the HawrƗn in Late Hellenistic and Roman Periods. A Reassessment of the Epigraphic Evidence. In Syria 70: 303-413. Marucchi O. 1911 Christian Epigraphy. An Elementary Treatise with a Collection of Ancient Christian Inscriptions mainly of Roman Origin, Cambridge (reprinted Chicago 1974). Mayerson Ph. 1991 The Use of the Term Phylarchos in the Roman-Byzantine East. In ZPE 88: 291-295. Milik J.T. 1980 La tribu des Bani ‘Amrat en Jordanie de l’époque grecque et romaine. In ADAJ 24: 41-54. Mitteis L. 1871 Reichsrecht und Volksrecht in den östlichen Provinzen des römischen Kaiserreiches, Leipzig (reprint Leipzig 1935). Noethlichs K.L. 2001 Die Juden im christlichen Imperium Romanum (4.-6. Jahrhundert), Berlin. Parker S.S.T. 1986 Romans and Saracens. A History of the Arabian Frontier, Winona Lake. Pollard N. 2000 Soldiers, Cities, and Civilians in Roman Syria, Ann Arbor. Rey-Coquais J.-P. 1978 Syrie Romaine de Pompée à Dioclétien. In JRS 68: 44-73. Sartre M. 1982a Tribus et clans dans le HawrƗn antique. In Syria 59: 77-91. Sartre M. 1982b Trois études sur l’Arabie romaine et Byzantine, (Latomus 178), Bruxelles. Sartre M. 1982c Inscriptions Grecques et Latines de la Syrie XIII, Fasc. 1: Bostra, Nos. 9001 à 9472, Paris. Sartre M. 1983 Les Arabes nomades d’Arabie du nordouest d’Alexandre au haut-empire. In Christe Y. et al. (ed.), Le désert: Image et réalité. Actes du Colloque de Cartigny 1983, Louvain: 139-160. Sartre M. 1991 L’Orient romain. Provinces et sociétés provincials en Méditerranée orientale d’Auguste aux Sévères (31 avant J.-C.-235 aprés J.-C.), Paris. Sartre M. 2000 Syria and Arabia. In The Cambridge Ancient History XI: 635-663. Schäfer P. 1983 Die Geschichte des Judentums in der Antike. Die Juden Palästinas von Alexander dem Großen bis zur arabischen Eroberung, Munich. Schalit A. 1937 Römische Palästinaverwaltung, Jerusalem. Schalit A. 2001 König Herodes. Der Mann und sein Werk, Berlin2. Schmidt M.G. 2004 Einführung in die Lateinische Epigraphik, Darmstadt. Schmitt O. 2004 Mavia - Königin der Sarazenen. In

Chester G. Starr, Lanham: 407-423. Frankfort Th. 1962 Le royaume d’Agrippa II et son annexion par Domitien. In Renard M. (ed.), Hommages à Albert Grenier, Bruxelles: 659-672. Freeman Ph. 1996 The Annexation of Arabia and Imperial Grand Strategy. In Kennedy D.L. (ed.), The Roman Army in the East, (JRA Suppl. 18), Ann Arbor: 91-118. Gebhardt A. 2002 Imperiale Politik und provinziale Entwicklung. Untersuchungen zum Verhältnis von Kaiser, Heer und Städten im Syrien der vorseverischen Zeit, Berlin. Glubb J.B. 1983 The Changing Scenes of Life. An Autobiography, London. Graf D. 1988 Qura ǥArabiyya and Provincia Arabia. In Gatier P.-L. et al. (ed.), Géographie historique au Proche Orient (Syrie, Phénicie, Arabie, grecques, romaines, byzantines). Actes de Table Ronde de Valbonne, 16-18 Septembre 1985, Paris. Graf D. 1989 Rome and the Saracens. Reassassing the Nomadic Menace. In Fahd T. (ed.), Le Arabie préislamique et son environnement historique et culturel, Leiden: 332-356. Graf D. 2002 Nomads and the Arabian Frontier. The Epigraphic Perspective. In Freeman Ph. et al. (ed.), Limes XVIII. Proceedings of the XVIIIth International Congress of Roman Frontier Studies held in Amman, Jordan (September 2000), Oxford, Vol. 1: 153-160. Gschnitzer F. s.v. ͇ࢼ͒ͧ͜͟͢͡. In RE Suppl. 11: 10671090. Gutsfeld A. 1989 Römische Herrschaft und einheimischer Widerstand in Nordafrika. Militärische Auseinandersetzungen Roms mit den Nomaden, Stuttgart. Harper G.M. 1928 Village Administration in the Roman Province of Syria. In YCS 1: 103-168. Isaac B. 1984 Bandits in Judaea and Arabia. In HSPh 88: 171-203. Isaac B. 1998 The Near East under Roman Rule. Selected Papers, Leiden-New York-Cologne. Isaac B. 1990 The Limits of Empire. The Roman Army in the East. Revised Edition, Oxford. Jamme A. 1970 Safaïtic mlk, ‘Lord’ of the Tribe. In Orientalia 39: 504-511. Kuhoff W. 1993 Die Beziehungen des römischen Reiches zum Volksstamm der Baquaten in Mauretanien. In Arctos 27: 55-71. Lancaster W. 1981 The Rwala Bedouin Today, Cambridge. Liddell H.G. and Scott R. 19969 A Greek-English Lexicon, Oxford. MacAdam H.I. 1986 Studies in the History of the Roman Province Arabia. The Northern Sector, Oxford. MacAdam H.I. 1989 Strabo, Pliny the Elder and Ptolemy of Alexandria. Three Views of Ancient Arabia and its Peoples. In T. Fahd (ed.), Le Arabie préislamique et son environnement historique et 283

THE LATE ROMAN ARMY IN THE EAST Mediterraneo Antico 7.2: 859-877. Schürer E. 1973/95 The History of the Jewish People in the Age of Jesus Christ (175 B.C.E. – 135 AD). A new engl. version, rev. and ed. by Geza Vermes and Fergus Millar, 3 Vol. Edinburgh2. Schumacher L. 1996 Römische Inschriften, Stuttgart2. Shahid I. 1984 Rome and the Arabs: A Prolegomenon to the Study of Byzantium and the Arabs, Dumbarton Oaks. Speidel M.A. 1998 Legio III Scythica, its Movements and Men. In Kennedy D.L. (ed.), The Twin Towns of Zeugma on the Euphrates. Rescue Work and Historical Studies, (JRA Suppl. 27), Porthsmouth: 163-204. Stoll O. 2001 Zwischen Integration und Abgrenzung. Die Religion des Römischen Heeres im Nahen Osten. Studien zum Verhältnis von Armee und Zivilbevölkerung im römischen Syrien und den Nachbargebieten, St. Katharinen. Waddington W.H. 1870 Inscriptions grecques et latines de la Syrie, recueilles et expliquées, Paris. Weiss A. 2004 Rom und die Baquaten oder: Die Grenzen der Integration. In Holzwarth W. et al. (ed.), Nomaden und Sesshafte – Fragen, Methoden, Ergebnisse, Halle: 1-20. Weiss A. 2006 Das Amt des praefectus gentis in den nordafrikanischen Provinzen, (forthcoming).

284

Moines, militaires et défense de la frontière orientale au VIe s. Geoffrey Greatrex Département d’études anciennes et de sciences des religions, Université d’Ottawa

Que les moines et les évêques aient joué un rôle important dans la défense de l’empire romain dans l’Antiquité tardive ne fait guère de doute. Pense-t-on à l’évêque Jacques de Nisibe, qui inspira les défenseurs de la ville durant le IVe s. ou bien à la résistance acharnée que livrèrent les moines du Tur Abdin contre les envahisseurs perses au VIIe s., et ce, même après le retrait des forces régulières, la notion du christianisme comme pilier de l'état romain paraît hors de doute.1 Plusieurs études ont mis en relief cet aspect important de la défense de l’empire, notamment l’article récent de Michael Whitby.2 Pourtant, l’épanouissement du christianisme à l’intérieur de l’empire ne s’accomplit pas sans conséquences moins favorables à la protection des frontières impériales, et c’est sur ces aspects, souvent sous-estimés, que notre contribution se penchera.

population locale de l'Orient. L’attitude méfiante à l'égard des moines dont témoigne Marcellin n’aurait fait qu’élargir ce clivage. Les relations entre soldat et civil, toujours tendues, ne s’améliorèrent point durant la période suivante, car les persécutions ponctuelles des opposants au concile de Chalcédoine finirent par user leur loyauté et miner leur confiance dans l'armée, souvent employée comme le bras fort des persécuteurs.4 Dans ce qui suit, nous tracerons l’évolution de ces persécutions ainsi que l’implication croissante des soldats dans les campagnes des évêques pour rallier la population au concile. Il est important de signaler maintenant que le chevauchement croissant des sphères religieuses et militaires menaça doublement la stabilité de la frontière: d’une part, les opérations des soldats, comme on l’a vu, affaiblirent la loyauté des victimes des persécutions, et d’autre part, ces activités, menées soit par un évêque, soit par un général, détournèrent les unités concernées de leur tâche principale: assurer la protection des frontières.

Pour illustrer cet aspect, d’une certaine manière, négatif, commençons par évoquer les rumeurs qui circulèrent après la prise d'Amid par les Perses en janvier 503. Plusieurs sources, dont le chroniqueur Marcellin comes, attribuent la perte de la ville à la trahison des moines chargés de la défense de la tour qui fut occupée par les Perses durant la nuit, ce qui conduisit inévitablement à la chute de la ville. Sans doute le fait que l’archimandrite ait été d’origine perse fut la cause de ces rumeurs, qui ont peu de chance d’ailleurs d'être justifiées.3 Il s’agit, il faut bien l’admettre, d’un cas relativement rare. Durant toutes les guerres qui secouèrent la frontière orientale tout au long du VIe s. la population locale, qu’elle fût pour ou contre le concile de Chalcédoine, fit preuve d’une loyauté remarquable. Il faut cependant retenir ce manque de confiance de la part de certaines sources puisqu'il reflète le fossé qui se creusa durant tout le reste du VIe s. entre le haut commandement romain (et ses chroniqueurs) et la

Nous entamerons notre analyse en traitant de la période 512-522, une décennie charnière pour l’Orient romain, marquée surtout par le revirement total de la politique impériale à l’égard du concile de Chalcédoine que produisit l’avènement de Justin I en 518. En novembre 512, le grand adversaire de Chalcédoine, Sévère, succéda au trône patriarcal d’Antioche, son prédecesseur Flavien ayant été déposé par un concile à Laodicée (en Isaurie) quelque temps auparavant. La correspondance volumineuse du patriarche nous permet de mesurer l’ampleur de ses activités et les difficultés auxquelles il se heurta. Elle témoigne de contacts nombreux avec l’élite civile et militaire de l’empire mais révèle également les limites de son pouvoir. Tandis qu’il communiqua avec le duc de Mésopotamie Timostrate concernant les problèmes financiers causés par le nombre excessif de personnes qui se faisaient consacrer prêtres,5 une affaire plutôt banale, il éprouva plus de mal à s’imposer auprès du comte Eutychien en Syrie II. Celui-ci n’hésita pas à communier avec des évêques qui avaient été déposés par Sévère. Il ressort clairement de la lettre du patriarche qu’il ne put rien faire, sauf proférer quelques menaces.6

1 Voir Dodgeon et Lieu 1991: 166-72 sur Jacques avec Blockley 1992: 16. Sur la résistance acharnée opposée aux Perses au début du VIIe s. par les moines du Tur Abdin voir plus bas n.61 et cf. le rôle remarquable de l'évêque Eunomius dans la défense de la ville de Théodosiopolis en 421 (en bénissant une catapulte), un épisode relaté par Théodoret de Cyr, HE, V.37.6-10, traduit dans Greatrex et Lieu 2002: 41. Je tiens à remercier de leurs commentaires les participants au colloque ainsi que Jan van Ginkel, Marie-Luce Constant et Catherine Collobert. 2 Whitby 1998: 191-208, surtout 196-9 sur la frontière orientale. Gaddis 2005 représente une contribution importante à l’étude des persécutions religieuses dans l’antiquité tardive, mais il se concentre surtout sur la période allant jusqu’au concile de Chalcédoine, cf. Barnes 2006: 721. 3 Marc. com. a.502, à lire avec Greatrex 1998: 91 et n.52. Debié 2003: 601-22 et Haarer 2006: 55. On rappellera aussi les accusations portées contre l’évêque monophysite Philoxène de Mabboug durant cette même guerre: voir Philoxène, Lettre aux moines de Senoun: 94-5/789. L’hostilité dont il fut l’object de la part des généraux et des gouverneurs romains ressort clairement, soit à cause de ses convictions monophysites, soit en raison de ses origines perses. Cf. Greatrex et Lieu 2002: 52.

4

Sur les relations entre la population civile et les militaires au Ve s. voir Greatrex et Lieu 2002: 43 (où l’on cite Nov.Th. IV.1), cf. le pseudoJosué le Stylite §§92-4, 100 avec Segal, 1970: 160-3. Quelques commandants pourtant nouèrent des relations plus étroites avec les indigènes, par exemple le maître des milices Anatolius au Ve s., qui offrit un naos pour abriter la dépouille de saint Thomas. Voir Segal 1970: 175 avec Di Segni 1995: 313-14 pour d'autres cas; aussi Pollard 2000: 99-100 (sur une période antérieure), 149-51 (une perspective plutôt positive sur l'Antiquité tardive). 5 Sévère, Sixth Book of the Letters, ep. I.8. 6 Sévère, Sixth Book of the Letters, ep.I.24. La même impuissance est perceptible dans ses lettres au maître des offices, Celer, ep. I.21, 24. Cf.

285

THE LATE ROMAN ARMY IN THE EAST

tranquillité des provinces frontalières. Pourtant les limites du pouvoir de l’empereur transparaissent clairement, tout comme celui de ses commandants. Un autre épisode de la même période, relaté seulement par Évagre, met en évidence le même phénomène. Après que deux évêques récalcitrants de Syrie II eurent réussi à excommuniquer Sévère par une ruse, Anastase riposta en ordonnant au duc de Phénicie Libanensis, Asiaticus, d’expulser les auteurs de l’excommunication de leurs sièges épiscopaux. Se rendant sur les lieux, Asiaticus constata que les deux évêques jouissaient de l’appui de la population entière: comme il ne tarda pas à communiquer à l’empereur, il ne saurait donc les évincer sans déclencher un bain de sang. Devant une telle opposition, Anastase se ravisa et déclara qu’il ne voulait plus que ses ordres soient exécutés s’ils avaient pour effet d’aboutir à une effusion de sang.9 L’épisode survint vers 516, la période même où l’empereur se vit obligé de faire marche arrière en Palestine, comme l’on a noté précédemment. On remarque, une fois de plus, l’ingérence des commandants militaires dans les affaires ecclésiastiques, bien que cela soit conforme aux ordres de l’empereur. Il y a lieu d’ailleurs de s'interroger sur l’implication du duc d’une autre province, puisqu’on s’attendrait à ce que le duc de Syrie II soit sommé par Anastase de prendre les mesures nécessaires. De toute évidence celui-ci se serait dérobé à ses ordres ou aurait tardé le plus possible à les mettre en application. Telle fut l’approche du haut commandant Patricius l’année précédente au moment où l’armée de l’insurgé Vitalien menaçait Constantinople: il refusa de prendre part à la défense de la ville en faisant remarquer qu’il était connu comme un proche du rebelle et qu’en cas de victoire de celui-ci, il serait soupçonné de lui être venu en aide.10 La mention du duc d’une province sinon frontalière, du moins très importante dans la défense de la frontière, nous force à reconnaître les répercussions possibles des redéploiements de forces nécessités par la politique ecclésiastique d’Anastase.

Sans le moindre succès, toutefois, car les actes d’une réunion tenue à Apamée après la fuite de Sévère en 518 et celle de son principal appui dans la région, l’évêque Pierre d’Apamée, rapportent que nul autre qu’Eutychien présida la séance et reçut les dénonciations de Sévère et de Pierre par les évêques.7 Quoi qu’il en soit, aucun élément cité jusqu'ici ne pouvait mettre en péril la stabilité des frontières. Mais il ne s’agit pas d’un cas isolé. Les manœuvres qui entourèrent la déposition du patriarche Elias de Jérusalem et son remplacement par Jean illustrent également l’imbrication des pouvoirs civils, militaires et ecclésiastiques. Sans entrer dans le détail du récit, qui fait l’objet de plusieurs études, on peut en esquisser les traits saillants. L’empereur Anastase, excédé par le refus d’Elias de communiquer avec Sévère, le fit déposer par l’intermédiaire d’Olympus, duc de Palestine, en août 516. Olympus arriva à Jérusalem à la tête d'une armée, rappelons-le, bien qu’il se soit contenté d’expulser le patriarche par des ruses plutôt que par la force. Ayant exilé Elias, Olympus installa Jean comme successeur, car celui-ci avait assuré à l’empereur qu’il entrerait en communion avec Sévère. Pourtant, sous la pression du vénérable saint Sabas et des moines des monastères avoisinants, Jean rechigna et crut préférable d’accepter le concile de Chalcédoine. Informé de cette volte-face, Anastase remplaça Olympus par un certain Anastase et lui ordonna d’obliger le nouveau patriarche à tenir sa promesse. Anastase alla à Jérusalem, emprisonna le patriarche, et lui ordonna d’entrer en communion avec Sévère. Jean réussit à se dérober, pourtant, en suivant les conseils du gouverneur de Palestine I, Zacharie: il assura à Anastase qu’il tiendrait cette fois sa promesse, mais lorsqu’il monta à l’ambon le dimanche suivant pour annoncer son adhésion au parti de Sévère, il anathématisa plutôt les adversaires de Chalcédoine et se rallia à Sabas et à ses associés qui l’entourèrent. On perçoit clairement l’implication de l’administration civile et militaire dans les affaires ecclésiastiques, allant jusqu’au déploiement d'une force armée, mais il nous reste à achever le tableau. En effet, le neveu aîné de l’empereur, Hypatius, un général expérimenté, assista à tout ce spectacle. Lorsque le duc Anastase s’enfuit, on peut comprendre que les regards de la foule rassemblée se soient braqués sur lui: comment réagirait-il à cet échec porté à la politique de son oncle? Sa réponse fut définitive: il ne communiait pas avec Sévère. Il renforça cette affirmation en distribuant des dons généreux aux monastères de la région; son oncle, quant à lui, ne réussit jamais à évincer Jean de son siège patriarcal.8 Cela n’était, cependant, pas suffisant pour troubler la

Il nous reste à discuter d’un autre événement marquant du patriarcat de Sévère dont les retentissements parvinrent jusqu’à Rome. Il s’agit du massacre qui survint en 516 vingt kilomètres au sud d’Apamée et qui a été l’objet d’une étude poussée de Paul Peeters.11 Un groupe de moines chalcédoniens de Syrie II était en train de se rendre au monastère de saint Syméon le stylite lorsqu'ils furent interceptés et attaqués dans une gorge près de Larisse; 350 hommes y trouvèrent la mort. Comme le suggère Peeters, il est fort probable que l’embuscade eût été tendue avec l’assentiment de l’évêque Pierre, l’allié de Sévère. Que des forces militaires régulières y aient participé semble vraisemblable: le récit des événements livré par les évêques syriens parle d’un véritable combat -

Peeters 1951: 261 et Frend 1972: 228. Sur l’administration de Sévère voir Allen et Hayward 2004: 11-24. 7 Voir Peeters 1951: 261, ACO III, 90-3, avec PLRE II, Eutychianus 4. 8 Le récit est donné par Cyrille de Scythopolis, Vie de Sabas, 149-52, tr. Festugière 1962: 76-81. Voir l’analyse de Peeters 1951: 237-9 et de Greatrex 1996: 120-3, cf. Maraval 1998a: 130-1, Binns 1994: 6-7 et Haarer 2006: 158.

9 Évagre, HE, III.34 avec les commentaires de Whitby 2000, cf. Honigmann 1951: 59 et Allen 1981: 154-5. Pour la date, voir Maraval 1998a: 130 et Peeters 1951: 259-60. 10 Voir Greatrex 1996: 125 et n.15, cf. Haarer 2006: 176. Nous n'avons aucun renseignement sur l’identité des ducs de Syrie à cette période. Sur le duché de Syrie, voir Greatrex 2007. 11 Peeters 1951: 252-8, cf. Honigmann 1951: 60-2.

286

GEOFFREY GREATREX : MOINES, MILITAIRES ET DÉFENSE DE LA FRONTIÈRE ORIENTALE AU VIE S. makhê.12 Quoique les historiens de l’église ne fassent pas la moindre allusion à cet affrontement, il n'y a aucune raison de douter qu'il a eu lieu.13

parvinrent même à capturer deux commandants romains, le duc Timostrate, dont il a déjà été question, et le duc Jean; les Romains se virent obligés de les racheter tous deux pour une somme élevée. Leur capture, qui eut lieu vraisembablement vers le début du règne de Justin, témoigne de l’état lamentable des défenses romaines.17 Rares sont les sources - anciennes ou modernes - qui mettent en parallèle ces deux séries d’événements. La tendance, dès l’Antiquité, a privilégié une distinction très nette entre politique militaire et politique ecclésiastique, exception faite de certains cas particuliers, tels les soulèvements des Samaritains.18 Pourtant, comme nous le verrons, le haut commandement de la frontière orientale avait déjà fort à faire avec la mise en opération de la politique ecclésiastique musclée de Justin, d’où son impuissance face aux raids lakhmides.

Il serait peut-être possible de rétorquer que les dernières années du règne d’Anastase furent exceptionelles. Nul doute que sa politique ecclésiastique, qui sembla se durcir progressivement, ne lui rendit pas la vie facile. Le soulèvement de Vitalien dans les Balkans l’obligea peutêtre à agir avec plus de clémence qu'il ne l’aurait fait. Il faut, toutefois, souligner le danger que représentaient les querelles christologiques, qui risquaient toujours de déborder et de provoquer la révolte populaire, ce qui eut lieu d’ailleurs, à maintes reprises, en Égypte.14 Même les ducs durent parfois s’avouer incapables de s’opposer à la volonté populaire, comme nous l’avons vu à Jérusalem et en Syrie. Néanmoins, malgré la grogne que suscitait la politique d’Anastase en Palestine et en Syrie II, les sources ne font état d’aucune tentative de la part des Perses d’en tirer profit. Ce n'était en somme qu'un heureux hasard, comme le démontre les événements qui survinrent durant les règnes de Justin I et Justinien.

Les sources se font plus nombreuses sur la région frontalière durant la période suivante. On doit toutefois admettre que les informations ne peuvent pas toujours être prises au sérieux: il s’agit principalement de sources anti-chalcédoniennes, surtout d’auteurs syriaques tels que Jean d’Éphèse et Michel le syrien. Comme l’a souligné à juste titre Jan van Ginkel, ce dernier en particulier a tendance à exagérer l’ampleur des persécutions subies par les adversaires de Chalcédoine, sans doute dans le but de conforter la communauté monophysite de son époque et pour souligner son unité dès le VIe s.19 Tout en tenant compte de ces déformations qu’introduit parfois Michel et du parti pris de Jean d’Éphèse, nous pouvons dresser le portrait d’une région sinon bouleversée, du moins durement touchée par l’imposition de la nouvelle politique religieuse de Justin.

L’avènement de Justin I marqua un revirement total dans la politique ecclésiastique de l’empire, comme c’est bien connu.15 L’appui vigoureux du nouvel empereur au concile de Chalcédoine entraîna très vite des changements de personnel dans les évêchés de l’Orient, notamment dans les provinces qui s’étaient rangées du côté de Sévère: de nombreux évêques, dont le patriarche lui-même, furent obligés de quitter leur siège, souvent précipitamment. La plupart prirent le chemin de l’exil vers l’Égypte.16 Le roi perse Cavad, quant à lui, profita du changement de régime pour renouveler sa pression diplomatique sur l’empire. Il insista pour que Justin lui verse un paiement régulier, renouvelant ainsi une demande faite à Anastase sans jamais le persuader. Puisque Justin se montrait peu enclin à accéder à ses exigences, Cavad donna l’ordre à ses alliés lakhmides de lancer des raids en territoire romain, ce qu’ils ne tardèrent pas à entreprendre. Lors d’une de leurs razzias, ils

L’empereur agit dans un premier temps avec énergie. Les expulsions se multiplièrent et les agents du gouvernement n’hésitèrent pas à recourir à la violence dans certains cas. Vraisembablement cette politique était liée à l’influence de Vitalien, ennemi acharné de Sévère: on retiendra que le comes Orientis Irénée reçut l’ordre d’arrêter Sévère et le garda sous haute surveillance (avant sa fuite). Certains affirment que Vitalien méditait de lui arracher la langue. Vitalien, rappelons-le, avait un lien avec Flavien, le prédecesseur de Sévère au patriarcat d’Antioche.20 Or, même durant cette première phase, les persécutions n’englobèrent pas l’Égypte, ce qui paraît démontrer un

12

ACO III (1940), 107.31, ainsi que les détails dans Günther 1895-8, ep.139.5, cf. Peeters 1951: 252-3 et Fraisse-Coué 1998: 184. Les actes préparés par les évêques de Syrie II furent lus au synode du patriarche Ménas de Constantinople en 537, ce qui assura leur survie; voir Peeters 1951: 264. Voir aussi ACO III, 96.19-27, des reproches adressés à Pierre, où l’on parle d’un véritable siège du monastère de Dorothée et de la séduction de la femme d’un soldat (appelé Pséphas) par Pierre. Les accusations portent à croire à des liens plutôt étroits entre l’évêque d’Apamée et les soldats stationnés dans sa région. On retiendra d'ailleurs une inscription syriaque qui témoigne de la présence de Pierre près d'Apamée vers 515-516: voir Harrak 1995.

17

Voir Greatrex et Lieu 2002: 79, sur les sources avec Greatrex 1998: 131 et n.31 (sur la chronologie). 18 On pense à la distinction entre l’historiographie traditionelle, de nature profane, telle que composée par Procope ou Priscus et l’historiographie ecclésiastique, à l’exemple d’Évagre ou de Socrate. L’œuvre de Stein 1949 suit cette même structure: les chapitres IV et VII traitent de la politique ecclésiastique d’Anastase et de Justinien, tandis que les chapitres V et VI concernent l’histoire politique. 19 van Ginkel 1998: 351-8. Cf. Whittow 1996: 43-4, qui va pourtant trop loin en réduisant (ex.) l’importance des persécutions sous Maurice. 20 Évagre, HE, IV.4, sur la hargne de Vitalien; pseudo-Zacharie de Mytilène VIII.2, pour son lien avec Flavien avec les remarques de Peeters 1948: 166; ibid. 141-2 sur cet épisode et le rôle de Vitalien. Evans 1996: 108, a tort en affirmant que Vitalien fut désigné comes Orientis lui-même.

13

Voir Peeters 1951: 258-9, pour une explication du manque d’attention portée à cet événement plutôt choquant. Voir, par exemple, Frend 1972: 154-6, Gregory 1979: 175-92 et Blaudeau 2006: 138-56 (et ailleurs) sur la situation en Égypte au Ve s. Une émeute contre la politique ecclésiastique d’Anastase faillit lui coûter son trône en 512: voir Frend 1972: 220, Maraval 1998a: 131-2, Haarer 2006: 156-7. 15 Voir (par exemple) Vasiliev 1951: ch.4, Frend 1972: 234-54, Rosen 2001: 766-71, Maraval 1998a: 137-44. 16 Voir les statistiques dans Frend 1972: 248, Honigmann 1951: 146-8. 14

287

THE LATE ROMAN ARMY IN THE EAST

région à l’adhésion à Chalcédoine de façon systématique, et de nouveau Paul fut chassé de son siège, cette fois par le maître des milices Pharesmanès.25 Son successeur, Asclépius, mena une lutte acharnée contre les adversaires du concile, surtout contre les moines qui étaient nombreux aux alentours de la ville. Pour ce faire, il dut se servir des troupes sous le commandement de Pharesmanès. Les soldats balayèrent la région en chassant les moines de leurs monastères malgré les rigueurs d’un hiver particulièrement rude.26 Citons un autre cas où l’on demanda à Hypatius d’intervenir. En août 520, ayant été nommé maître des milices de l’Orient, il fut mandé par l’empereur pour mettre un terme aux activités d’une orthodoxie douteuse de l’évêque Serge de Cyr. Il régla la situation en déposant l’évêque.27

certain réalisme dans l’attitude de Justin. Les provinces au-delà de l’Euphrate semblent, elles aussi, avoir été épargnées, du moins jusqu’en 520.21 On pourrait supposer qu’une telle décision de concentrer la pression sur les régions plus éloignées de la frontière fut liée à des considérations pratiques relatives à la défense de l’empire, mais il n'en est rien. Bientôt les soldats et leurs commandants furent sommés d’intervenir dans la région et de tout mettre en œuvre pour assurer son adhésion au concile. Il convient ici de citer des exemples concrets. Dès la fin du règne d’Anastase on voit Hypatius, alors maître des milices, chargé d’enquêter sur un cas ecclésiastique impliquant le patriarche Sévère. Celui-ci dut comparaître devant Hypatius à Égée en Cilicie en août 517. A l’occasion de leur rencontre, le patriarche présenta un hymne au général; tout porte à croire qu’il voulait obscurcir ainsi la véritable raison de son déplacement en Cilicie.22 L’issue du cas nous échappe, mais ce qui mérite d’être retenu, c'est le fait qu’un commandant militaire se pose en arbitre dans une affaire qui n’a aucun lien avec son domaine de compétence. On a évoqué, plus haut, la façon dont le comes Orientis Irénée fut chargé de mettre Sévère sous garde en 518. En novembre 519, la chronique d’Édesse nous signale la présence de Patricius, l’ancien maître des milices, dans la ville. Sa mission était de convaincre l’évêque Paul d’accepter le concile, mais il se trouva aux prises avec une opposition farouche. Confronté à l’attitude intransigeante de Paul, Patricius le saisit en l’arrachant au baptistère où il s’était réfugié et l’exila à Séleucie.23 Patricius n’exerça plus de commandement officiel, il est vrai, mais il est certain qu’il fut appuyé par un contingent de soldats. Selon la version préservée par le pseudo-Denys, il fut accompagné d’une unité imposante de Goths qui n'hésitèrent pas à s’attaquer à la foule qui s’était mobilisée pour venir au secours de l’évêque.24 Justin eut beau revenir sur la décision de Patricius en lui reprochant sa fougue et en restituant Paul à son évêché, celui-ci refusa toujours de se plier à sa volonté. Il fallut intervenir de nouveau en 522, au moment où l’empereur s’appliquait à gagner toute la

Nous nous sommes concentrés délibéremment jusqu’ici sur une période restreinte. Il sera question ultérieurement de l’évolution des relations entre l’armée et la région frontalière, mais il serait utile d’esquisser d’abord quelques conclusions. En premier lieu on remarque le chevauchement croissant des domaines militaires et ecclésiastiques, une tendance qui se maintiendra et se renforcera tout au long du siècle, aboutissant enfin au VIIe s. à la nomination de Cyrus comme préfet et patriarche de l’Égypte. L’entrée en fonctions en 527 du patriarche Ephrem d’Antioche, ancien comes orientis, va dans le même sens.28 Cette tendance est compréhensible dans la mesure où il incombait à l’empereur d’assurer l’orthodoxie de son empire. S’il se dérobait à son devoir sacré il risquait non seulement le salut de l’empire, car il ne jouirait plus du soutien divin, mais aussi la colère de la population (surtout de celle de Constantinople).29 Le déploiement de l’armée afin d’assurer l’orthodoxie, déjà 25

Ps.-Denys 25-6, tr. Witakowski, 26 avec Vasiliev 1950: 236-7. Cf. ps.-Zach., HE, VIII.4, qui nomme le gouverneur Liberius, lui aussi appelé un Goth, comme le responsable. PLRE II, Libelarius, identifie ce Liberius avec le Libélarius nommé par Procope. 26 Ps.-Denys 27, tr. Witakowski, 27-8. Voir aussi Jean d’Éphèse, Lives, PO 18: 607-8. Peeters 1948: 180 place l’expulsion des moines en décembre 522, mais la Chronique d’Édesse, §89 (90), indique plutôt le mois d’octobre. Cf. Vasiliev 1950: 236-7. 27 ACO IV.1, 184, 186-7, et surtout 200.3-5, où on relate que l’évêque trahere autem ad se milites etiam, qui testimonia in gestis Antiochiae confectis praestitisse inveniuntur de numero tertio Stabilisianorum, nec non etiam Andronicum et Georgium). Andronique, presbyter et defensor et George, diacre, auraient promu activement la commémoration de Thédoret de Cyr. Cf. PLRE II, Hypatius 6, Maraval 1998a: 139, Honigmann 1951: 78 et Peeters 1948: 160-3. 28 PLRE II, Ephraemius. Frend 1972: 249, suivi par Evans 1996: 109, en font à tort un ancien maître des milices. Les Monophysites adoptèrent la même attitude: les deux évêques qui succédèrent à Thomas, celui qui organisa la construction de Dara (voir n.64 plus bas) furent des anciens gouverneurs, Nonnus et Moro bar Constant; voir ps.-Zach., HE, VIII.5. Sur Cyrus, voir n.70 plus bas. 29 Voir van Ginkel 1999: 61, qui cite Chadwick 1979: 10; cf. Meyendorff 1993: 228-9. On pourrait citer, par exemple, la déclaration du patriarche Nestorius à Théodose II, qu’il lui assurerait une victoire contre les Perses s’il lui permettait d’imposer l’orthodoxie à son empire, Socr. HE, VII.29.5. Jean d’Éphèse, HE, II.25, III.2, associe les revers soufferts par les Romains aux persécutions dont les Monophysites étaient victimes. Sur les manifestations qui eurent lieu à Constantinople pour insister sur une orthodoxie plus sévère, voir n.47 plus bas.

21 Voir ps.-Zach., HE, VIII.5 avec Maraval 1998a: 138-9, Jansma 1965: 488-92, Vasiliev 1950: 211, Honigmann 1951: 110-12. Ce dernier note le cas de l’évêque Sotérichus de Césarée en Cappadoce qui ne fut pas destitué, sans doute en partie à cause de sa grande popularité. L’empereur Justin lui-même témoigne du respect accordé à cet évêque, prope cunctus Oriens sine dubio veneratur, ep.193 dans Guenther 18958: 650-1. 22 Cf. Peeters 1951: 265-7, Greatrex 1996: 122, Allen et Hayward 2004: 55. L’hymne se trouve dans Brooks, éd., The Hymns of Severus, 661, cf. Honigmann 1951: 84. Il y prononça également ses homélies CX-CXI, tr. Brière 1943, sur lesquelles voir le commentaires de Peeters 1944, 27. 23 Chronicon Edessenicum §88, cf. Chronicon pseudo-dionysianum vulgo dictum, vol.2: 24-5 (tr. Witakowski, 25), avec Vasiliev 1950: 233, Jansma 1965: 194-5, Segal 1970: 130, et PLRE II, Patricius 14. Peeters 1951: 179-86 discute de cet épisode en détail. Il émet une théorie possible, selon laquelle Patricius aurait outrepassé son mandat en évinçant Paul en raison de son interprétation personelle plus rigide du concile. Justin aurait renversé la décision à la demande des représentants du pape Hormisdas présents à Constantinople. 24 Ps.-Denys 24-5, tr. Witakowski, 25.

288

GEOFFREY GREATREX : MOINES, MILITAIRES ET DÉFENSE DE LA FRONTIÈRE ORIENTALE AU VIE S.

lois de Justin et Justinien émises en 527 témoignent de son attitude intransigeante dans ce dossier.33 Le récit de Jean d’Éphèse concernant les monastères d’Amid nous permet de suivre les péripéties des communautés touchées par les persécutions durant cette période. Ayant été expulsés de la ville en 521, les moines se dirigèrent vers un autre monastère, celui de Mar Mama, où ils passèrent cinq ans. Malgré les dangers, les moines décidèrent alors de rentrer dans la région d’Amid et s’installèrent à la limite de son territoire à l’ouest, à Abarné. Après y avoir passé un certain temps - le texte de Jean indique deux ans et demi, mais Brooks fait remarquer à juste titre qu’il s’agit plutôt de quatre ans et demi - les moines reçurent l’ordre de rentrer à Amid où ils purent constater l’ampleur de la destruction qu’avaient subie leurs monastères durant leur absence.34

connu au Ve s., se fit plus fréquemment, sans doute du fait du durcissement des attitudes doctrinales des deux parties. Cette mobilisation des effectifs militaires ne put avoir que des conséquences néfastes pour la défense des provinces: alors que les raids lakhmides perturbaient les régions frontalières, se soldant par la capture de deux généraux, le commandant en chef Hypatius fut obligé de s’absenter du front pour régler des questions de discipline ecclésiastique dans une ville fort éloignée de la zone touchée.30 Dans certains cas, il a même fallu affecter des troupes à certains évêques qui changèrent de camp pour se ranger du côté de l’empereur, afin de les protéger de la colère de leurs ouailles: ce fut le cas, par exemple, de Sotérichos de Césarée, ainsi que de plusieurs patriarches d’Égypte.31 Pis encore, l’intervention des commandants et de leurs unités se fit souvent au détriment des citadins et de la population locale en général. Même si l’on tient compte de l’exagération possible de nos sources, il est évident que les troupes romaines - souvent qualifiées de Goths - n’étaient pas appréciées des autochtones; la description de leur comportement que livre le pseudoJosué le stylite, une source contemporaine et fiable, est éloquente à cet égard.32 Pour conclure, il y a lieu de souligner la différence entre l’attitude de l’empereur Anastase et celle de ses successeurs: alors que le premier renonça plusieurs fois à imposer sa volonté par la force sans doute en sachant qu’il ne pouvait être assuré de son succès, vu l’opposition si acharnée des partisans de Chalcédoine - ses successeurs n’y sont pas allés de main morte. L’ingérence des soldats et des généraux dans la vie quotidienne des villes de l’Orient se renforça, ce qui ne manqua pas d’influer sur l’attitude de la région vis-àvis de l’empire.

Quelle que soit la date du retour des moines en 528 ou en 530, le contexte est le même: un état de guerre entre les deux grandes puissances, plus ouverte, toutefois, en 530 qu’en 528. À notre connaissance, peu de chercheurs ont perçu un rapport entre la reprise de la guerre et la suspension des persécutions des adversaires de Chalcédoine. Pourtant le changement de politique paraît brusque et même inattendu. Selon le pseudo-Denys de Tel-Mahré l’ordre fut transmis à Bélisaire, alors duc de Mésopotamie ou plus vraisembablement maître des milices de l’Orient, de permettre le retour des moines.35 L’intervention d’un commandant ne doit pas surprendre: nous avons déjà pu apprécier l’implication des commandants et des soldats dans le harcèlement des communautés monophysites. Il est donc logique que l’empereur revienne sur sa politique musclée, plutôt dangereuse pour la stabilité de la zone frontalière, au moment où la guerre venait de reprendre ou bien était imminente. Les conséquences dangereuses de soulèvements à l’intérieur de l’empire, capables d’être exploités par les Perses, furent clairement illustrées par la révolte des Samaritains, qui éclata en 529 et qui fut brutalement réprimée par le duc de Palestine avec l’aide des alliés jafnides; le chroniqueur Théophane nous indique que les insurgés étaient entrés en contact avec Cavad en lui suggérant de diriger son invasion vers le

Il nous reste à éclairer la façon dont les interventions de plus en plus nombreuses des forces romaines dans les persécutions religieuses finirent par miner la loyauté des provinces concernées et quel en fut l’impact sur la défense de l’empire. Il n'est point nécessaire de relater l’histoire de l’église du règne de Justinien, une histoire fort complexe et marquée de revirements remarquables. Il suffira de mettre en exergue certains événements, surtout les persécutions des opposants du concile, échelonnées tout au long du VIe s. Notre discussion doit commencer toutefois en faisant état d’une amélioration vers 530 du sort de ces adversaires. Au début de son règne en 527 Justinien avait prôné la fermeté contre les hérétiques: des

33

Voir Vasiliev 1950: 246, Maraval 1998a: 144, Gray 2005: 229. Jean d’Éphèse, Lives, PO 18: no.24, 608, 618-20. Cf. ps.-Denys, 29, tr. Witakowski, 30, où l’on indique que Bélisaire permit le retour des moines après une période de six ans et demi. Ps.-Zach., HE, VIII.5, place la période d’exil des moines entre 525 et 531. La date du retour des moines ne saurait être établie avec certitude, tant les sources varient entre elles. Vasiliev 1950: 238-9, le place en 530, Devréesse 1945: 72-3, préfère 531, mais en s’appuyant sur la Chronique de 846, 223/169-70) qui donne plutôt une date de sél. 838, l’équivalent de 526/7. Vööbus 1988: 212 penche aussi pour 531. Il est certain que plusieurs furent rappelés à leur siège pour qu’ils puissent prendre part aux négociations à Constantinople en 532, mais le fait qu’une proportion élevée de ces évêques occupât des sièges très rappochés de la frontière n'est pas une coïncidence, nous le croyons. Voir aussi Vööbus 1988: 210-14, sur l’adoucissement de la politique impériale durant cette période. 35 29, tr. Witakowksi, 30. 34

30 Il est vrai qu’en 517 et en 520 les cas présentés à Hypatius eurent lieu au mois d’août, une période de l’année où les opérations militaires étaient rares. Patricius expulsa Paul d’Édesse en novembre - après la fin de la saison des campagnes militaires normalement. Toutefois, il a fallu rassembler les forces nécessaires pour ces opérations et rien ne porte à croire qu’il n'y a pas eu d’autres cas d’interventions à d’autres moments. 31 Mich. Syr. IX.20, tr. Chabot, 189, cf. Honigmann 1951: 112. Sur la situation en Égypte, voir (p. ex.) ps.-Zach., HE, IV.1-2 et Gregory 1979: 184. 32 Ps.-Josué, §§93-4. Sur les méthodes employées lors des persécutions, voir van Ginkel 1999: 65. On misait surtout sur l’éloignement du clergé monophysite de leurs ouailles, d’où les nombreuses expulsions. Les effets se firent sentir dès les années 530, comme en témoigne Jean d’Éphèse, Lives 24, PO 18: 516, 522 cf. Vööbus 1988: 217.

289

THE LATE ROMAN ARMY IN THE EAST sud.36 Certes, l’attitude des opposants du concile ne fut guère aussi hostile au pouvoir impérial que celle des Samaritains, ce qui est attesté par une lettre envoyée à la capitale impériale par l’entremise du duc Théodote vers la même période; bien que les représentants du camp monophysite insistent sur leurs différends avec l’empereur en ce qui concerne le concile, ils n'hésitent pas à témoigner de leur loyauté et de leur respect envers le pouvoir impérial.37 Néanmoins, le danger représenté par une communauté opprimée ne pouvait échapper à l’empereur. Une fois la guerre terminée en 532, les persécutions ne reprirent pas pour autant. C’est que la conjoncture politique à Constantinople favorisait plutôt ceux qui s’opposaient au concile, une situation sans doute à mettre en relation avec les négociations qui avaient été entamées en 532 entre les représentants des deux parties.38 Les débats se poursuivirent et s’approfondirent au fur et à mesure que les adversaires du concile se rassemblaient à Constantinople, forts de l’appui vigoureux de l’impératrice Théodora. Le moment sembla propice aux efforts des opposants du concile: en 535 le nouveau patriarche de Constantinople, Anthime, se rangea de leur côté et entra en communion avec le patriarche Sévère, présent dans la capitale pour les négociations, et Théodose, récemment consacré patriarche d’Alexandrie.39 Sur ces entrefaites intervint un brusque revirement, provoqué par l’arrivée du pape Agapet à Constantinople: Anthime fut remplacé par Ménas et Sévère dut fuir les lieux. L’empereur renchérit en condamnant celui-ci explicitement: en août 536 il émit une loi qui cibla directement Sévère et ses alliés et ordonna la destruction des œuvres du patriarche déposé. Pourtant, cinq ans s’écoulèrent avant que Justinien ne décrète que les «Acéphales», ceux qui suivaient les doctrines de Sévère et de Dioscore, étaient des hérétiques de la même eau que les Nestoriens et touchés par les mêmes restrictions et pénalités que ces derniers, ainsi que les Juifs et les Samaritains.40

défrayent la chronique, le patriarche Ephrem d’Antioche et l’évêque Abraham bar Kaili d’Amid. Sans entrer dans les détails, il suffit de noter que tous les deux furent accompagnés d’unités de soldats romains, souvent qualifiés de Goths. Tandis qu’Abraham se prévalut du soutien du duc Thomas, lui aussi désigné comme un Goth, Ephrem semble avoir mené ses «descentes» luimême. Vu son ancienne expérience de comes Orientis, une telle démarche n’a rien d’implausible.41 De toute cette triste histoire, relevons quelques éléments importants. Premièrement, soulignons l’alliance entre les évêques et l’armée. Peu importe si les sources affirment avec insistance que la plupart, sinon la totalité, des troupes impliquées étaient des «barbares», leur traitement brutal non seulement des moines et des religieuses, mais aussi de toute la population des provinces orientales, n’a pu avoir que des conséquences fâcheuses pour les relations entre soldats et civils. Le fossé qui séparait le soldat, fut-il «Goth» ou «Romain», et le civil continuait à se creuser. En deuxième lieu il faut rappeler que ces persécutions survinrent au moment de la «paix éternelle». Tous croyaient alors à la durée de cette paix, conclue en 532. Grâce à elle Ephrem put compter sur la collaboration du marzban de Nisibe dans l’arrestation de l’évêque monophysite Jean de Tella en 537. L’évêque lui-même témoigne des bonnes relations entre les deux pouvoirs et la façon dont il put traverser la frontière sans encombres.42 Il y a ainsi toute raison de croire à un lien entre le déferlement de cette vague de persécutions et l’existence d’une paix durable - du moins aux yeux des contemporains - sur la frontière. Les troupes qui n’avaient pas été affectées aux campagnes en Occident se voyaient donc employées comme le bras fort des partisans du concile. Il n’est guère surprenant que l’armée romaine fît si piètre figure lors de l’invasion de Chosroès en 540.43 On n’entend plus parler de persécutions en Orient durant les années 540. La première source après cette date à en parler, c’est le pseudo-Zacharie, qui relate l’intervention d’un duc en 553 dans un chapitre digne d’attention.44 Lors du cinquième concile oecuménique qui se déroulait à Constantinople,45 un concile convoqué par Justinien dans le but d’adopter une formule qui rallierait les Monophysites à Chalcédoine, certaines personnes firent savoir à l’assemblée des évêques qu’il existait de nombreux adversaires du concile dans les provinces

Les conséquences ne tardèrent pas à se faire sentir à travers tout l’Orient. Les sources monophysites s’appesantissent sur les souffrances de leur communauté durant cette période en insistant sur la brutalité de ceux qui menèrent les persécutions. Deux personnages surtout

41

Voir le récit du pseudo-Denys, 32-3, 37-43, tr. Witakowski, 32-3, 3741 et ceux de ps.-Zach., HE, X.4 et de Michel le Syrien, IX.16, tr. Chabot 180-1, IX.19, tr. Chabot 185-8, IX.26, tr. Chabot 206, avec Frend 1972: 284 et Vööbus 1988: 214-17. 42 Vit. Joh. Tel., 71-3, traduit dans Greatrex et Lieu 2002: 98-9. 43 Mich. Syr. X.21, tr. Chabot 193, rapporte l’envoi de 6 000 troupes en Égypte en 536 pour permettre l’instauration de Paul le Tabennésiote, cf. Stein 1949: 384-5, 389-91, Frend 1972: 274-5. 44 Ps.-Zach., HE, XII.6. Jean d’Éphèse, Lives, PO 18: 622, semble indiquer que les moines ne purent réintégrer leurs monastères qu’après 23 ans de persécutions, vers 559, mais Brooks, loc. cit., citant le passage de Ps.-Zach. dont il est question, affirme qu’ils durent être rentrés dès 553 au plus tard. Vraisemblablement dès les années 540 certains réussirent à retourner à leurs monastères. Cf. van Ginkel 1999: 63. 45 Sur lequel voir (p. ex.) Maraval 1998b: 419-23, Gray 1979: 70-1, van Ginkel 1999: 64.

36

Greatrex 1998: 191-2 avec Théophane, Chronographia, 179. Sur l’écrasement de la révolte voir (p. ex.) Evans 1996: 116-17, Sartre 1982: 168-70. 37 Ps.-Zach., HE, IX.15 (p.115), cf. Mich. Syr. IX.22, 198, tr. Chabot (p.282 du texte). Pour la date voir PLRE III, Theodotus 2 (où l’on omet le passage du Ps.-Zach., pourtant), qui la place en 535; il est plus probable pourtant que la lettre fut envoyée plus tôt, en 532. 38 Sur lesquels voir (p. ex.) Frend 1972: 269-72, Allen 1994: 228-9, Meyendorff 1993: 242-3, Maraval 1998b: 403-5, Gray 2005: 229-30. 39 Frend 1972: 270-1, Maraval 1998b: 404-5. 40 Justinien, Novelle 42 (août 536), cf. Frend 1972: 273, Evans 1996: 110-11 et Noethlichs 2001: 691-4. Comme le note van Ginkel 1999: 63, cette phase fut décisive; Justinien se décida même à intervenir en Égypte, jusque-là indemne. Nov.109 (541), préf., pour la mention des Acéphales, cf. Vasiliev 1950: 248, Meyendorff 1993: 248-9.

290

GEOFFREY GREATREX : MOINES, MILITAIRES ET DÉFENSE DE LA FRONTIÈRE ORIENTALE AU VIE S.

revenait normalement à l’empereur de déclencher une vague de persécutions, comme il en fut en 536: les «descentes» d’Abraham et d’Ephrem survinrent après l’échec des négociations à Constantinople et la diffusion de lois ciblant les adversaires du concile.

frontalières. Selon l’information fournie aux évêques, ces schismatiques tenaient leurs propres conciles et gagnaient des adhérents parmi toute la population. Inquiet du fait de ces renseignements, Justinien convia le duc Audono à prendre en main la situation en agissant de concert avec le presbyter Basiliscus. Leur enquête se concentra sur Amid - du moins le récit du pseudo-Zacharie ne parle que de cette région - et les moines monophysites furent menacés d’expulsion. Toutefois, on parvint à un compromis puisque les moines se décidèrent plutôt à s’installer sur le mont Izla. L’année suivante l’ambassadeur Pierre le patrice, de passage dans la région, eut vent de la démarche du duc. Il le réprimanda et retira ses menaces. Tout porte à croire que les moines rentrèrent alors à Amid.46

Il ne serait pas inapproprié de voir dans cette tendance un coup fatal porté à la défense de l’Orient. Les insultes lancées contre Tibère à cause de sa tolérance envers les soldats goths démontrent à quel point la hargne religieuse avait éclipsé des considérations politiques ou militaires aux yeux de la majorité de la population byzantine. Justinien lui-même durant les années 530 et 540 en subit les lourdes conséquences en Afrique du Nord, lorsqu’il céda aux exigences des catholiques et décréta des mesures draconiennes contre les ariens: la province fut désormais aux prises avec des soulèvements dangereux qui faillirent effacer tous les acquis de la reconquête.48 Les dangers présentés par cette intolérance croissante n’échappèrent ni à Justinien, ni à ses successeurs. A certains moments les empereurs furent en mesure de refuser les exigences des militants, prétextant les besoins militaires de l’empire, comme le firent Justin II, Tibère et Maurice. Michel le Syrien rapporte la réponse suivante de Tibère lorsqu’il fut interpellé par le patriarche Eutychius pour lui demander de renouveler les persécutions des adversaires de Chalcédoine:

Cet épisode, dont le pseudo-Zacharie est le seul témoin, est révélateur d’un nouveau phénomène dans la politique ecclésiastique de l’empire, au moins en ce qui a trait aux provinces orientales. Ce n’était plus l’empereur qui menait la charge, mais plutôt des individus, et notamment des hommes d’église. Fait intéressant, le pseudo-Zacharie connaît les noms de ceux qui interpellèrent les évêques orientaux lors du concile, mais se garde de les dévoiler. Il suggère que leur démarche fut provoquée par la jalousie (si l’on accepte la lecture du texte proposée par Hamilton). Or, il est probable qu’elle fut motivée autant par zèle doctrinal que par jalousie, puisqu’on constate un durcissement progressif des attitudes christologiques au cours du siècle. Ceux qui furent les plus marqués par cette sévérité croissante étaient, comme on aurait pu s’y attendre, les hommes d’église et notamment ceux qui avaient gravi les marches de la hiérarchie ecclésiastique. Dorénavant les sources indiquent clairement que ce furent plutôt les patriarches ou les évêques qui prônèrent la rigueur contre les hérétiques que les empereurs euxmêmes; ces derniers pour leur part s’empressèrent de refréner leurs activités, quitte à risquer de provoquer la colère de la population, tout aussi intolérante que ses prêtres. En témoigne, par exemple, l’affaire des soldats goths recrutés par Tibère pour la guerre contre les Perses: le peuple de Constantinople s’insurgea contre la décision de Tibère de leur permettre l’usage d’une église dans la capitale. Déjà en 552 Procope relate un épisode pertinent au sujet d’une armée expédiée par Justinien pour venir en aide aux Lombards. Plusieurs commandants furent envoyés, mais un seul parvint aux alliés; les autres durent s’arrêter à Ulpiana en Illyricum pour mettre un terme à des luttes internes qui opposèrent certains chrétiens à d’autres.47 Jusqu’aux années 530 ou 540, par contre, il

Nous avons assez des guerres contre les Barbares; nous ne pouvons exciter une autre guerre contre nous, ni faire en sorte que les chrétiens soient aux prises les uns avec les autres. Allez et restez tranquilles. Tâchez de les persuader en les exhortant par la parole. Sinon, cesse de persécuter.49 L’exaspération de l’empereur devant l’insistance des militants est perceptible. La transformation totale de la situation depuis le début du VIe s. ressort clairement également. Au Ve et au début du VIe s. les persécutions et les mesures fortes émanèrent normalement de l’empereur, la plupart du temps pour rallier la population à l’orthodoxie du jour (soit l’Hénotikon, soit le monophysisme durant le règne d’Anastase, soit l’adhésion au concile). Il fallut ainsi plusieurs interventions sanglantes pour imposer les décisions de 465-6. Sur l’armée envoyée aux Lombards voir Proc. Guerres VIII.25.13, cf. Stein 1949: 534. Le durcissement des attitudes doctrinales de la population de la capitale peut être relié à l’émergence d’une nouvelle identité romaine, fortement imbue de son orthodoxie. Les acclamations de la foule à l’avènement de Justin I attestent de la ferveur du peuple à ce sujet, voir Vasiliev 1950: 136-44, Maraval 1998a: 133-4, cf. Greatrex 2000: 277-8. 48 Justinien, Nov.37 (535), cf. Greatrex 2001: 80 et Maraval 1998b: 398. 49 Mich. Syr. X.8, tr. Chabot, 310 (texte, 347). Jean d’Éphèse, HE, III.12 attribue un pareil discours au même empereur en réponse au patriarche Jean de Constantinople, cf. Jean d’Éphèse, HE, III.21 et Mich. Syr. X.18, 349, où il est question de Tibère et le patriarche Eutychius). Cf. Jean d’Éphèse, HE, V.21 pour une réponse semblable de Maurice. Pour une analyse moderne, voir Maraval 1998c: 468, Frend 1972: 332 et Shahîd 1995: 898-900.

46 Ps.-Zach., HE, XII.6. Jean d’Éphèse, Lives, PO 18: 622, note une troisième expulsion, en 561, lors de laquelle les moines se réfugièrent dans le désert, dans la région de l’`Arab, sur le mont Izla et aux alentours de Dara. Cette vague de persécutions pourrait être liée à la conclusion d’un autre traité de paix en 562: les négociations auraient abouti déjà en 561. Voir Greatrex et Lieu 2002: 131-4. 47 Sur les soldats goths voir Jean d’Éphèse, HE, V.15, cf. Greatrex 2001: 80-1. L’échec des négociations entamées au début du règne de Justin II atteste de l’impossibilité d’un compromis et de l’attitude intransigeante des moines monophysites; voir Frend 1972: 318-19, Maraval 1998c:

291

THE LATE ROMAN ARMY IN THE EAST

Ce durcissement des attitudes eut raison finalement des tentatives de négociation, notamment lors du dernier véritable effort au début du règne de Justin II.54 Il ne restait aux partisans de Chalcédoine que d’imposer par la force leurs doctrines aux populations frontalières qui résistaient toujours. Durant la guerre entre 572 et 591, des considérations politiques finirent par primer sur les revendications des patriarches, mais dès la conclusion de la paix en 591, ils revinrent à la charge.55 Deux individus se chargèrent de faire accepter à la région le concile de Chalcédoine, le patriarche Grégoire d’Antioche, qui avait déjà joué un rôle de premier plan dans la dernière phase de la guerre perse, et l’évêque Domitianus de Melitène, un proche conseiller de l’empereur Maurice. Ce premier entreprit une tournée de l’Orient vers 591, tandis que Domitianus y mena une persécution sanglante vers la fin du siècle. Michel le syrien affirme que ce fut Domitianus qui demanda à l’empereur la permission d’intervenir dans la région, piqué de jalousie à l’égard les Monophysites. Selon Michel, ceux-ci auraient fait des progrès dans l’évêché de Domitianus durant son mandat: bien qu’il puisse s’agir d’un cliché des historiens monophysites, on y perçoit la même source de mécontentement qu’en 553 la jalousie des Chalcédoniens - de même que l’inspiration locale (plutôt que centrale).56 Michel n’hésite pas à s’attarder sur les souffrances de la communauté monophysite lors des persécutions poursuivies par Domitianus, qui fut appuyé par des soldats et employa tous les moyens possibles pour obliger ses adversaires à adhérer au concile. Tout en tenant compte de l’intérêt de Michel à exagérer ces faits, il n’y a aucune raison de remettre en question la nature plus brutale, voire sanglante, de cette dernière vague de persécutions avant la conquête perse. Plusieurs trouvèrent la mort en refusant d’accepter le concile. Qui plus est, comme l’affirme Michel, «ils raillèrent l’empereur et Domitianus».57 Cette réaction tranche nettement avec l’attitude de cette même communauté un demi-siècle auparavant. Théodose, le patriarche monophysite d’Alexandrie destitué, qui résida à Constantinople jusqu’à sa mort en 566 - avec l’aval de

Chalcédoine en Palestine et en Égypte, bien que dans ce dernier cas le bilan de ces efforts fût plutôt maigre.50 Par contre, il revenait désormais aux évêques et à la population de mener la charge et d’insister sur des mesures fortes contre ceux qui s’opposaient à l’orthodoxie officielle. Un représentant tout à fait typique de cette nouvelle tendance est incarné par «saint» Serge, dont la vie est racontée par Jean d’Éphèse. Pendant que les habitants d’Amid, monophysites et chalcédoniens ensemble, assistaient à l’homélie du prêtre à l’église un dimanche matin, celui-ci pénétra dans l’église, monta les escaliers menant à l’ambon, saisit le prédicateur par le cou et le jeta par terre. Ce faisant, il le somma de ne pas jeter des perles aux pourceaux et se mit à lancer des injures contre le concile de Chalcédoine.51 Le gouvernement ne pouvait réagir qu’en faisant tout son possible pour calmer le jeu et pour trouver une solution aux querelles doctrinales.52 Certes, même au Ve s. certaines villes furent troublées par des luttes entre les représentants de doctrines divergentes - par exemple, Édesse, où les actes du latrocinium de 449 nous renseignent sur le bras de fer qui opposa les adhérents d’Ibas à ceux de Rabbula - mais rares étaient ceux qui voulaient troubler la situation et chercher à étendre les persécutions. L’empereur Anastase, comme l’on a vu, a même préféré reculer, et ce à plusieurs reprises, plutôt que de recourir aux armes pour imposer ses interprétations doctrinales à la population.53

50 Voir Evans 1996: 74-5, Chitty 1966: 86-91. Sur l’Égypte, voir plus haut n.14. 51 Jean d’Éphèse, Lives, PO 17: 102-3, cité par Evans 1996: 110. Cf. en revanche les commentaires de Whitby 2000: xliv et n.30. Serge fut saisi par les Chalcédoniens après la messe et incarcéré dans un monastère. Cf. Jean d’Éphèse, HE, IV.31, qui rapporte des cas de moines monophysites arrêtés par les autorités civiles pour avoir assassiné des adversaires de leur parti. 52 Aucune de leurs initiatives ne porta fruit, cependant, malgré des discussions parfois prometteuses. On a déjà noté l’échec des négociations qui prirent fin à Callinicum en 568. Les divisions entre les Monophysites eux-mêmes réduisaient davantage les chances de succès, ce qui explique pourquoi les autorités s’ingérèrent même dans ces querelles, comme le rapporte Jean d’Éphèse, HE, IV.35, 40. 53 Sur la situation à Édesse, voir Frend 1972: 37, 42, Segal 1970: 93-4 et 130 et Drijvers 1996: 242-8. La mission des commandants de l’armée déployés à l’occasion, comme à Édesse, fut plutôt de rétablir «la tranquillité», voir (ex.) ACO I.4, ep.240, 169, une lettre du maître des milices Dionysius au comes Titus où il est question de la tranquillitas urbana, cf. ep.265, 196.17-18 (où le patriarche Jean d’Antioche caractérise l’évêque Mélitius comme un fauteur de troubles). Cf. ibid., ep.195, 139-40, où Paul d’Émèse insiste dans une missive au maître des milices Anatolius sur son orthodoxie et sur son désir de tranquillité; ep.139, 91.20-1). Voir n.21 plus haut pour un cas où Justin refusa de déposer un évêque pour ne pas troubler la paix des provinces. Cf. aussi van Ginkel 1999: 65-6. Signalons que par cette interprétation nous prenons nos distances par rapport à celle avancée par van Ginkel, loc. cit., qui préfère mettre en relief le rôle de l’empereur et les moyens que pouvaient adopter ses ministres pour ralentir, voire réduire à néant, la force de ses initiatives. Il croit toujours à la primauté de l’intervention impériale tandis que selon nous, l’initiative locale primait sur les décisions de la cour. On pourrait citer de nombreux cas où les ministres de l’empereur ont pris l’initiative sans l’approbation de celui-ci, ex. en sabotant les négociations avec les Perses en 526, voir Greatrex 1998: 135-7, ou bien dans le cas de l’ambassadeur Jean à la cour perse en 567 (relaté par Ménandre, frg.9.1-2).

54

Voir n.52 plus haut. Malgré le coup porté aux négociations à Callinicum en 568, des discussions se poursuivirent jusqu’en 571, au moment où Justin trancha l’affaire en émettant un édit qui condamna encore une fois les adversaires du concile. Quelques prêtres monophysites acceptèrent de communier avec les chalcédoniens à Constantinople, mais plusieurs autres, dont Jean d’Éphèse, subirent un resserrement des persécutions. Voir Maraval 1998c: 466-7, Frend 1972: 321-2, Allen 1994: 230-1. 55 Les persécutions qui suivirent l’échec des négociations en 568 se maintinrent jusqu’en 578, surtout grâce au zèle des patriarches de Constantinople, mais seulement dans une zone limitée. Voir Jean d’Éphèse, HE, III.21 avec Frend 1972: 322-3, Maraval 1998c: 466-7, Shahîd 1995: 898-908. 56 Mich. Syr. X.23, tr. Chabot 372 (386 du texte), cf. Chron. 1234, ch.82, 217-18/171. Sur les deux interventions, voir Frend 1972: 333-4 et Greatrex et Lieu 2002: 176. 57 Mich. Syr. X.23, 373, tr. Chabot (386 du texte). Étant donné que Michel ne rapporte aucune critique de ce genre de la part des victimes jusque-là, il n’y a aucune raison de douter du ton du récit qu’il nous livre ici: malgré un indéniable préjugé, il s’agit d’une attitude conséquente dont les traces sont visibles dans son œuvre entière. La montée d’un cran de la tension sur la frontière, dont il témoigne ici, est donc digne de foi.

292

GEOFFREY GREATREX : MOINES, MILITAIRES ET DÉFENSE DE LA FRONTIÈRE ORIENTALE AU VIE S.

pouvaient pas participer à la défense de la frontière, mais elles contribuèrent également à effriter la solidarité qui avait existé entre les habitants des provinces et leurs protecteurs. Il en va de même pour leurs commandants qui eurent à trancher des disputes ecclésiastiques partout en Orient. Certains, il est vrai, se montrèrent plus conciliants que d’autres, mais les sources donnent l’impression que leur tendance était plus souvent de passer outre à la légalité afin d’obtempérer aux ordres de l’empereur et d’assurer l’orthodoxie doctrinale; il arriva donc parfois à l’empereur ou à son délégué d’être obligé de revenir sur les actions, jugées excessives, des commandants.62 Mais à ce moment, le mal était déjà fait. On a aussi remarqué le rôle croissant des évêques dans la gestion des provinces, une tendance concrétisée, par exemple, par une novelle de Justin II émise en 569.63 En Orient, nous avons observé l’implication des évêques dans les persécutions des adversaires de Chalcédoine où on les retrouve souvent à la tête d’unités de soldats, mentionnés par les sources monophysites comme des «barbares» ou des «Goths». Rappelons, toutefois, qu’il y a un aspect positif dans cette activité episcopale: ils furent chargés parfois de la construction d’œuvres défensives, comme notamment à Dara sous l’empereur Anastase, mais aussi à Birtha et ailleurs, des tâches qu’ils surent mener à terme avec efficacité et empressement.64 Le patriarche Grégoire aida à désamorcer la révolte des soldats qui menaça toute la frontière durant le règne de Maurice; son influence fut telle qu’il réussit même à triompher de ses adversaires à Antioche dont le comes Orientis lui-même.65 Reste à signaler les liens souvent étroits entre dirigeants ecclésiastiques, civils et militaires, qui confondaient, voire effaçaient, les distinctions entre les différentes sphères d’administration. Le patriarche Flavien, rappelons-le, fut le «père spirituel» du maître des milices Vitalien, celui qui leva l’étendard de la révolte contre Anastase au nom du concile de Chalcédoine et qui voulut, selon Évagre, couper la langue à Sévère.66 Le patriarche Ephrem, nous l’avons noté, avait occupé le poste de comes Orientis; durant son patriarcat son frère fut gouverneur d’Osrhoène ou de Mésopotamie.67 L’évêque d’Édesse qui remplaça Paul durant le règne de Justin I, Asclépius, était le frère du préfet du prétoire Démosthène, et il ne manqua pas de se prévaloir de ses contacts avec la cour impériale.68 Plus curieux est le cas de Damien, patriarche monophysite d’Alexandrie sous le règne de Tibère, dont le frère fut gouverneur d’Osrhoène: on pourrait donc supposer que celui-ci put atténuer les

l’empereur Justinien d’ailleurs - insista toujours dans ses discours sur l’importance de la loyauté à l’empereur et à l’empire.58 On pourrait supposer que l’attitude de Théodose ne fut pas forcément partagée par la population frontalière monophysite, plus touchée par les persécutions. Mais la lettre des moines de l’Orient à ceux qui débattaient de la foi à Constantinople en 532, citée plus haut, témoigne de la loyauté constante à l’empereur, du moins avant les persécutions qui suivirent.59 De là à supposer que les habitants des provinces orientales accueillirent les armées perses (et arabes) à bras ouverts serait exagéré. Toutefois force est de reconnaître l’écart presque infranchissable qui sépara alors la population locale et les soldats chargés de la défendre. Lorsque commencèrent les assauts perses, les soldats romains ne tardèrent pas à se replier vers l’ouest, du moins après la prise de Dara; selon Michel, encore une fois, les «Romains» furent la cible particulière des Perses, qui, toujours selon Michel, ne firent aucun mal aux habitants des provinces. Il revint enfin aux moines de se mobiliser pour assurer la défense des places fortes restantes, notamment la forteresse de Mardin. «Ils étaient tous prêtres et ils envoyèrent demander à Basilius, évêque de Kephar Touta, s’il leur était permis de tuer les Perses.»60 De toute évidence Basilius leur répondit par l’affirmative, car le siège de la ville dura plusieurs années. En dernier recours la défense des provinces orientales a dû être assurée par des moines monophysites; il ne manquèrent pas à leur devoir. Grâce à leur ténacité il fallut environ six ans aux Perses pour traverser la Mésopotamie et l’Osrhoène et atteindre l’Euphrate. En dépit de toutes les persécutions, une loyauté à l’empire romain subsista toujours, ne fût-ce que par opposition au paganisme des Perses.61 Le moment est venu de tirer les conclusions de notre discussion. On a déjà souligné les conséquences négatives du redéploiement des troupes pour assurer l’adhésion des provinces au concile de Chalcédoine. Non seulement les unités impliquées dans ces rafles ne 58 Voir Frend, Rise, 288, 292, bien que ses références soient douteuses. Cf. Blaudeau 1996: 123-8, sur l’appui du patriarche à l’empire (au Ve s.). Il est vrai, par contre, que Jean Rufus, Plérophories, PO 8: no.4, 1516, paraît associer des soldats romains et des démons (en parlant du patriarche Juvénal). 59 Ps.-Zach., HE, IX.15 (voir plus haut, n.37). Cf. Mich. Syr. IX.19, tr. Chabot 188, sur la loyauté des Monophysites en 536. 60 Mich. Syr. X.25, 378, tr. Chabot (texte, 390-1). Au début du VIe s. soldats et civils avaient fait front commun pour s’opposer aux envahisseurs, notamment à Constantia et à Édesse en 503, voir le pseudo-Josué §58-60; cf. le cas de Resafa/Sergiopolis en 542, voir Greatrex et Lieu 2002: 110. On n’entend plus parler d’une telle collaboration vers la fin du VIe s., mais il est possible que ce changement reflète plutôt le tarissement de sources aussi détaillées que le pseudo-Josué. Voir (p.ex.) Hoyland 2004: 193 pour la suggestion que les persécutions des Monophysites aplanirent le chemin des conquérants musulmans. 61 Il faut signaler néanmoins que Chosroès II, sensible à l’importance de l’église jacobite, amorça une politique favorable aux évêques monophysites qui purent ainsi regagner leurs sièges après la conquête perse. Une telle tournure n’était pas prévisible au début de l’invasion, pourtant. Voir Greatrex 2003: 78-88.

62

Voir les cas cités plus haut. Nov.149 (569), cf. Liebeschuetz 2000: 220 et Flusin 1998: 490-5 (sur le rôle croissant des évêques). 64 Voir Greatrex et Lieu 2002: 240, Segal 1970: 129, Trombley et Watt 2000: 106 n.498 et Trombley 1997: 162 avec Di Segni 1995: 332, Palmer 1990: 117 et Codex Justinianus I.4.26 (530). 65 Évagre, HE, VI.7, sur l’opposition du comes à Grégoire, avec Maraval 1998c: 475, Allen 1981: 250 et Lee 2007a. 66 Ps.-Zach., HE, VIII.2, cf. Peeters 1951: 234. Delmaire 1984 ne s’adresse point à cette question. 67 Ps.-Denys, 40, tr. Witakowski, 38, cf. Jean d’Éphèse, Lives, PO 17 (1923): 294. 68 Ps.-Zach., HE, VIII.4. 63

293

THE LATE ROMAN ARMY IN THE EAST

provinces plus occidentales, là où l’appui au concile était unanime.74

mesures prises contre les Monophysites durant son mandat, bien que rien ne nous affirme qu’il partageait les opinions christologiques de son frère.69 Nous avons évoqué plus haut le point culminant de cette tendance, la cumulation de tous les pouvoirs - ecclésiastiques, civils et militaires - en la personne de Cyrus, le dernier gouverneur de l’Égypte avant la prise de la province par les Arabes.70

Il convient de relever un autre fait saillant qui concerne les lois introduites par Justin et Justinien. Jusqu’en 536, l’année décisive où l’empereur rompit les négociations avec Sévère et ses partisans, aucune loi ne cibla les Monophysites. De nombreux groupes se trouvèrent en proie aux mesures rigoureuses prises par le gouvernement contre les non-chrétiens et les hérétiques, dont les Juifs, les Samaritains, les Manichéens, les Ophites et les Montanistes, mais on n’y retrouve aucune mention des Monophysites. Évidemment ceux-ci durent subir des persécutions importantes durant les années 520, mais il est significatif que ce ne fut qu’en 541 qu’on les désigna spécifiquement dans la législation.75 La position des opposants du concile se dégrada par la suite, mais il est évident que les mesures ne furent pas toujours appliquées de façon rigoureuse. On retiendra également que les partisans et les adversaires du concile partageaient le même souci pour l’orthodoxie de l’empire et la suppression des hérésies et du paganisme: Michel le Syrien rapporte en les approuvant les confiscations décrétées par Justinien qui frappèrent les hérétiques et qui persuadèrent nombre d’entre eux d’intégrer l’église orthodoxe.76 En effet, l’opposition acharnée des deux communautés à tout autre groupe religieux fut un instrument utile que les empereurs ne se gardèrent pas d’exploiter: déclencher une persécution de païens ou d’une autre minorité pouvait assouplir les relations tendues entre les deux groupes en les réunissant contre un ennemi commun. Tel fut sans doute l’objectif de Tibère lorsqu’il réagit vigoureusement aux activités des païens aux alentours d’Héliopolis en 577: il somma le commandant Théophile de venger les chrétiens qui avaient été tués par les païens.77

Il n’a pas été question jusqu’ici de la loyauté des soldats eux-mêmes. Jamais les sources ne font état d’un refus de ceux-ci à s’en prendre à la population civile. Ils prêtèrent main forte aux généraux et aux évêques sans broncher, motivés en partie par le butin qu’ils pouvaient gagner lors de leurs interventions.71 Or, il faut préciser que tel fut le cas à partir du règne de Justin: l’empereur Anastase, on l’a vu, eut beaucoup de mal à imposer ses décisions et dut même y renoncer parfois. En effet, les sources ne relèvent aucune trace de soldats monophysites. Lorsque Justin I émit des lois sévères en 519 ou 520 pour que tous les militaires acceptent le concile, aucune source ne rapporte la moindre opposition parmi les forces romaines. On doit souligner pourtant que ni les fédérés ni les alliés ne furent touchés par ces mesures, si contraignantes fussent-elles. Une loi de 527 fait mention spécifiquement de l’exception faite aux Goths qui reçurent ainsi le droit d’être enrôlés parmi les fédérés sans être obligés d’adhérer au concile.72 Les alliés jafnides des Romains se trouvèrent eux aussi à l’abri de ces lois et maintinrent jusqu’à la dissolution de la dynastie leur soutien aux Monophysites; à maintes reprises leurs chefs s’évertuèrent à appuyer la recherche d’une solution aux disputes doctrinales et protégèrent les communautés monophysites de la zone frontalière.73 Il est difficile de déterminer les causes de cette tendance chalcédonienne des soldats, perceptible dès le début du VIe s., mais elle pourrait être liée aux régions d’origine des soldats: sans doute une grande partie de l’armée fut recrutée dans des

Notre principale conclusion peut se résumer par la formule paradoxale suivante: au VIe s. les quelques

69 Mich. Syr. X.17, 344 tr. Chabot. Les opinions très différentes des neveux d’Anastase à l’égard de Chalcédoine suffisent à démontrer que les familles n’étaient pas forcément unies dans leurs attitudes doctrinales: voir Greatrex 1996: 129-32. 70 Voir PLRE III, Cyrus 17 avec Kaegi 2003: 216. Cf. l’implication du préfet Pergamius dans les affaires de l’église dès le Ve s., ps.-Zach., HE, V.6-7 avec PLRE II, Pergamius 2. On rappellera également le cas de Photius, un commandant muni de pleins pouvoirs par Justin II et entouré d’une garde de moines et de soldats; fort de ses pouvoirs, il alla jusqu’à extorquer des fonds aux évêques. Voir Jean d’Éphèse, HE, I.32 avec PLRE III, Photius 2. Durant sa carrière il fut général, consul honoraire, et finalement abbé, avant de recevoir sa mission générale de Justin II. Cette imbrication des compétences explique par ailleurs l’érosion de la distinction entre l’histoire ecclésiastique et classicisante: on remarque dans l’œuvre de Théophylacte des discours d’évêques (ex. Histoires IV.15.17-16.28), tandis qu’Évagre et Jean d’Éphèse n’hésitent pas à relater des événements profanes. 71 Pourtant le Ps.-Denys, 39, rapporte que les soldats préférèrent s’attaquer aux moines d’Amid, installés à Tella de-Tuthe, indirectement, par le biais de pression exercée sur la population locale. 72 Sur les lois, voir Vasiliev 1950: 233, 242-6. Selon Vasiliev, le texte de la loi de 519/20 est préservé par C.J. I.4.20 qui fut attribué à tort à Justinien seul. Jacques d’Édesse et Michel le Syrien y font allusion aussi. Celle de 527 est C.J. 1.5.12, voir section 17 sur les Goths avec Greatrex 2001: 79, Evans 1996: 67. 73 Cf. Frend 1972: 284-5, Shahîd 1995: 876-90.

74 Il n’est pas nécessaire de supposer qu’il s’agissait toujours de «Goths», malgré l’insistance des sources monophysites à ce sujet. Or, le fait qu’elles désignent souvent les soldats comme «barbares» signale qu’ils n’étaient pas originaires de la région. Le rapport mentionné plus haut (n.27) semble indiquer que certains soldats étaient attirés par des doctrines nestoriennes. Michel le Syrien X.3, tr. Chabot 294, est une rare mention de (quelques) soldats monophysites. Comme l’on a vu au début, le haut commandement semble avoir eu peu de sympathie pour les opposants du concile. On pourrait ajouter aux cas notés ci-dessus celui de l’a secretis Astérius, qui, sous Anastase, déclara que la nomination de Sévère au siège d’Antioche représentait la descente de l’hiver sur l’empire: voir Sévère, Collection of Letters, PO 12: 321, cf. Frend 1972: 220. Sur l'attitude des soldats aux disputes doctrinales voir dernièrement Lee 2007b: 198-204. 75 Voir n.40 pour les références. 76 Mich. Syr. IX.21, 191-2 tr. Chabot, cf. IX.30, 248 tr. Chabot, sur les hérésies qui prospérèrent durant le règne de Justinien. 77 Jean d’Éphèse, HE, III.27-8, cf. Mich. Syr. X.12, 318 tr. Chabot. Jean, op. cit. III.20 rapporte la plainte des Monophysites, qu’on préférait les persécuter plutôt que s’occuper des autres hérétiques et païens de l’empire. La réaction brutale de Tibère doit être perçue à la lumière de ces plaintes; il n’est pas sans rapport que Théophile vînt tout juste de mater des révoltes parmi les Samaritains et les Juifs (ibid. III.27). Cf. Allen 2000: 830-1, sur la pression que dut subir Tibère.

294

GEOFFREY GREATREX : MOINES, MILITAIRES ET DÉFENSE DE LA FRONTIÈRE ORIENTALE AU VIE S.

périodes de paix entre les Romains et les Perses furent plus nuisibles à la défense des frontières que les décennies de guerre. Durant les guerres les empereurs n’avaient ni les moyens ni le désir de s’adonner à la persécution de leurs propres citoyens: il leur incombait plutôt de protéger leur territoire. Même les conflits dans le camp monophysite se calmèrent durant la guerre entre 572 et 591, comme le note Michel le Syrien.78 Par contre, ce fut durant les courtes périodes de paix que sévirent les partisans du concile, notamment durant la période de la «paix éternelle», du moins après l’échec des négociations à Constantinople, et durant la dernière décennie du siècle. Nos sources sont moins explicites pour le bref interval qui sépara les deux grandes guerres du siècle, entre 562 et 572, mais il semble que Justin II se rallia lui aussi à une attitude sévère lorsque les négociations entre les deux parties échouèrent en 568.79 Le phénomène qui sous-tend ces persécutions ponctuelles est sans doute le durcissement des attitudes de la population entière de l’empire. Ce n’étaient plus les empereurs qui insistaient sur l’orthodoxie de leurs sujets mais leurs ministres, fussent-ils évêques ou généraux. Les contacts et la collaboration entre les commandants et les évêques se resserrèrent. L’armée romaine, de plus en plus souvent affectée à des missions policières à l’intérieur des provinces, ressembla davantage à une armée étrangère, composée largement de «barbares» (aux yeux des Monophysites). A force d’être la cible de tant de séries de persécutions, les moines et certains Monophysites finirent par conspuer les troupes impériales. Cet effritement de la loyauté mérite d’être souligné, car nous discutons d’une région qui avait longtemps fait preuve d’un dévouement remarquable envers l’empire.80 Le christianisme, peut-on conclure, loin d’apporter un soutien décisif à la défense de l’Orient, a fini par la miner à cause de l’intolérance qu’il engendrait parmi ses adhérents.

Bibliographie ACO. Acta Conciliorum Oecumenicorum, éd. E. Schwartz, J. Straub, vols.3-4.1. Berlin 1940-71. Allen, P. 1981. Evagrius Scholasticus, the Church historian, Louvain. Allen, P. 1994. «Monophysiten», Theologische Realenzyklopädie 23: 219-33. Allen, P. 2000. «The Definition and Enforcement of Orthodoxy» in A. Cameron, B. Ward-Perkins et M. Whitby, éds., The Cambridge Ancient History, vol.14, Cambridge: 811-34. Allen, P., et Hayward, C. 2004. Severus of Antioch. London. Barnes, T.D. 2006. Compte rendu de Gaddis 2005, JTS 57: 720-5. Binns, J. 1994. Ascetics and Ambassadors of Christ. The Monasteries of Palestine, 314-631. Oxford. Blockley, R.C. 1992. East Roman Foreign Policy. Leeds. Blaudeau, P. 1996. «Timothée Aelure et la direction ecclésiale de l'empire post-chalcédonienne», REB 54: 107-34. Blaudeau, P. 2006. Alexandrie et Constantinople (451491). De l'histoire à la géo-ecclésiologie. Rome. Chadwick, H. 1979. The Role of the Christian Bishop in Ancient Society. Berkeley. Chitty, D. 1966. The Desert a City. Londres. Chronicon ad A.D. 846 pertinens, CSCO Scr. Syr. 3-4, éd. et tr. J.-B. Chabot. Louvain 1904. Chronicon anonymum ad a.c. 1234 pertinens, CSCO Scr. Syr. 55-6, éd. et tr. J.-B. Chabot. Louvain 1937. Chronicon Edessenicum, CSCO Scr. Syr. 1-2, éd. et tr. I. Guidi. Paris 1903. Chronicon pseudo-dionysianum vulgo dictum, vol.2, ed. J.-B. Chabot, Paris, 1933, tr. R. Hespel, Louvain, 1989, tr. aussi par W. Witakowski, Pseudo-Dionysius of Tel-Mahre, Chronicle, known also as the Chronicle of Zuqnin. Part III, Liverpool, 1996, Codex Justinianus, éd. P. Krueger, onzième édition, Berlin, 1954. Cyrille de Scythopolis, Vie de Sabas in E. Schwartz, éd., Kyrillos von Skythopolis, Texte u. Untersuchungen XLIX.2, Leipzig, 1939. Debié, M. 2003. «Du grec en syriaque: la transmission du récit de la prise d'Amid (502) dans l'historiographie byzantine», BZ 96: 601-22. Delmaire, R. 1984. «Les dignitaires laïcs au concile de Chalcédoine: notes sur la hiérarchie et les préséances au milieu du Ve s.», Byz 54: 141-75. Devréesse, R. 1945. Le patriarcat d'Antioche depuis la paix de l'église jusqu'à la conquête arabe. Paris. Dodgeon, M. et Lieu, S.N.C. 1991. The Roman Eastern Frontier and the Persian Wars A.D. 226-363. Londres. Drijvers, H.J.W. 1996. «The Man of God of Edessa, Bishop Rabbula, and the Urban Poor: Church and Society in the Fifth Century», JECS 4: 235-48.

78 X.12, 318, tr. Chabot. Jean d’Éphèse, HE, IV.42 relate que le chef jafnide al-Harith demanda à Tibère de mettre fin aux persécutions des monophysites, ce qu’il consentit à faire, cf. Shahîd 1995: 900-8. Voir aussi n.49 plus haut sur le lien explicite entre la cessation des persécutions et la guerre contre les Perses; il est vrai pourtant que les persécutions persistèrent durant les premières années de la guerre de 572, cf. n.55. Théodore Lector 466 fait état de ce même lien dans le contexte de la guerre de 502-5: selon lui, Anastase fut empêché de poursuivre ses persécutions des Chalcédoniens à cause des hostilités. 79. Voir Frend 1972: 319 sur l’échec des négociations à Callinicum en 568. Jean d’Éphèse, HE, II.25, cf. III.2, perçoit les revers cuisants essuyés par les Romains au début de la guerre de 572 comme le châtiment des persécutions déclenchées par Justin et le patriarche Eutychius. Ibid. II.28 sur les interventions musclées d’un certain diacre Théodoulos, chargé d’assurer l’orthodoxie et pourvu d’unités de soldats. 80 Noter, par exemple, les acclamations loyales rapportées dans les actes du concile de 449, notées par Frend 1972: 67, ainsi que la résistance vigoureuse des populations d’Édesse et d’Amid durant la guerre de 502506, pour ne citer que quelques exemples. Cf. plus haut n.1.

295

THE LATE ROMAN ARMY IN THE EAST

Harrak, A. 1995. ‘Notes on Syriac Inscriptions, I. The Inscription of Ma'ar-zaytâ (Syria)’, Orientalia 64: 110-19. Honigmann, E. 1951. Évêques et évêchés monophysites d'Asie antérieure au VIe siècle. Louvain. Hoyland, R. 2004. « Language and Identity: the Twin Histories of Arabic and Aramaic (and Why did Aramaic succeed where Greek failed? », SCI 23: 10399. Jansma, T. 1965. « Credo (Encore le) de Jacques de Saroug », L’Orient syrien 10: 75-88, 193-236, 331-70, 475-510. Jean d'Éphèse, Historiae Ecclesiasticae Pars Tertia, CSCO Scr. Syr. 54-5, éd. et tr. E.W. Brooks. Louvain 1952. Jean d'Éphèse, Lives of the Eastern Saints, éd. et tr. E.W. Brooks, PO 17 (1923), 18 (1924) et 19 (1925). Jean Rufus, Plérophories, éd. et tr. F. Nau, PO 8 (1911). The Chronicle of Joshua the Stylite, éd. et tr. W. Wright, Cambridge, 1882, tr. aussi par Watt, J. et Trombley, F.R., Liverpool, 2000. Justinien, Novellae in Corpus Juris Civilis vol.3, édd. R. Schoell et W. Kroll, sixième édition. Berlin 1954. Kaegi, W. 2003. The Emperor Heraclius. Cambridge. Lee, A.D. 2007a. « Episcopal power and perils in the late-sixth century: the case of Gregory of Antioch » in Salway et Drinkwater, éds: 99-106. Lee, A.D. 2007b. War in Late Antiquity. A Social History. Oxford. Liebeschuetz, W. 2000. «Administration and Politics in the Cities of the fifth to the mid-seventh century» in A. Cameron, B. Ward-Perkins et M. Whitby, éds., The Cambridge Ancient History, vol.14: 207-37. Cambridge. Maraval, P. 1998a. «La réception de Chalcédoine dans l'empire d'orient» in Pietri, éd.: 107-45. Maraval, P. 1998b. «La politique religieuse de Justinien» in Pietri, éd.: 389-426. Maraval, P. 1998c. «L'échec en Orient: le développement des Églises dissidentes dans l'empire» in Pietri, éd.: 457-81. Marcellin comes, Chronique, ed. T. Mommsen, tr. B. Croke, The Chronicle of Marcellinus. Sydney 1995. Ménandre. The History of Menander the Guardsman, éd. et tr. Blockley, R.C. Liverpool 1985. Meyendorff, J. 1993. Unité de l'Empire et divisions des Chrétiens, tr. F. Lhoest. Paris. Michel le Syrien, éd. et tr. J.-B. Chabot, Chronique de Michel le Syrien, patriarche jacobite d'Antioche (1166-1199), 4 vols. Paris 1899-1924. Noethlichs, K.L. 2001. «Justinian I», RAC 19: 668-773. Palmer, A. 1990. Monk and Mason on the Tigris Frontier — The Early History of Tur `Abdin. Cambridge. Peeters, P. 1944. Compte rendu de M. Brière, Les Homélies cathédrales de Sévère d'Antioche, AnBoll 62: 257-60. Peeters, P. 1948. «Jacques de Saroug, appartient-il à la

Évagre, Ecclesiastical History, éd. J. Bidez and L. Parmentier, Londres, 1898. Traduction par M. Whitby, The Ecclesiastical History of Evagrius Scholasticus, Liverpool, 2000. Evans, J.A.S. 1996. The Age of Justinian. The Circumstances of Imperial Power. Londres. Festugière, A.J. 1962. Les moines d'Orient: Les moines de Palestine, vol.3.2. Paris. Flusin, B. 1998. «Évêques et patriarches. Les structures de l'Église impériale» in Pietri, éd.: 485-544. Fraisse-Coué, C. 1998. «L'incompréhension croissante entre l'Orient et l'Occident (451-518)» in Pietri, éd.: 147-96. Frend, W.H.C. 1972. The Rise of the Monophysite Movement. Cambridge. Gaddis, M. 2005. There is No Crime for Those Who Have Christ. Religious Violence in the Christian Roman Empire. Berkeley. van Ginkel, J. 1998. «Making History: Michael the Syrian and His Sixth-Century Sources» in R. Lavenant, éd., Symposium Syriacum VII. Rome: 3518. van Ginkel, J. 1999. «Persuasion and Persecution: Establishing Church Unity in the Sixth Century» in H.L.J. Vanstiphout, éd., All those Nations... Cultural Encounters within and with the Near East. Groningen. Gray, P.T.R. 1979. The Defense of Chalcedon in the East (451-553). Leyde. Gray, P.T.R. 2005. « The Legacy of Chalcedon » in M. Maas, éd., The Cambridge Companion to the Age of Justinian. Cambridge: 215-38. Greatrex, G. 1996. «Flavius Hypatius, quem vidit validum Parthus sensitque timendum. An investigation of his career», Byz 66: 120-42. Greatrex, G. 1998. Rome and Persia at War. Leeds. Greatrex, G. 2000. «Roman identity in the sixth century» in G. Greatrex et S. Mitchell, éds., Ethnicity and Culture in Late Antiquity. Londres: 267-92. Greatrex, G. 2001. «Justin I and the Arians» in M.F. Wiles et E.J. Yarnold, éds., Studia Patristica 34: 7181. Leyde. Greatrex, G. et Lieu, S.N.C. 2002. The Roman Eastern Frontier and the Persian Wars, A.D. 363-630. Londres. Greatrex, G. 2003. «Khusro II and the Christians of his empire», Journal of the Canadian Society for Syriac Studies 3: 78-88. Greatrex, G. 2007. «Dukes of the eastern frontier» in Salway et Drinkwater, éds: 87-98. Gregory, T.E. 1979. Vox Populi. Popular Opinion and Violence in the Religious Controversies of the Fifth Century A.D. Columbus, OH. Günther, O., éd., Collectio Avellana: epistulae imperatorum, pontificum, aliorum, A.D. 367-553, CSEL 35. Vienna 1895-8. Haarer, F.K. 2006. Anastasius I. Politics and Empire in the Late Roman World. Cambridge. 296

GEOFFREY GREATREX : MOINES, MILITAIRES ET DÉFENSE DE LA FRONTIÈRE ORIENTALE AU VIE S.

Trombley et Watt 2000. Voir The Chronicle of Joshua the Stylite. Vasiliev, A.A. 1950. Justin I. Washington, D.C. Vit. Joh. Tel., éd. E.W. Brooks, Vitae Virorum apud Monophysitas celeberrimorum, CSCO Scr. Syr.7-8. Paris 1907. Vööbus, A. 1988. History of Asceticism, vol.3. Louvain. Whitby, M. 1998. «Deus Nobiscum»: Christianity, Warfare and Morale in Late Antiquity» in Modus Operandi. Essays in Honour of Geoffrey Rickman, M. Austin, J. Harries et C. Smith, éds. Londres: 191-208. Whitby, M. 2000. Voir Évagre plus haut. Whittow, M. 1996. The Making of Orthodox Byzantium, 600-1025, Londres. Pseudo-Zacharie de Mytilène, éd. et tr. E. W. Brooks, Historia Ecclesiastica Zachariae Rhetori vulgo adscripta, CSCO Scr. Syr. 38-9, 41-2. Paris 1924.

secte monophysite?», AnBoll 66: 134-98. Peeters, P. 1951. «Hypatius et Vitalien: autour de la succession de l'empereur Anastase» in idem, Recherches d'histoire et de philologie orientales. Bruxelles. L'article a paru en premier lieu dans l'Annuaire de l'institut de Philologie et d'Histoire orientales et slaves 10, 1950 (Mélanges H. Grégoire), 5-51. Philoxène, Lettre aux moines de Senoun, éd. et tr. A. de Halleux, Louvain, 1963. Pietri, L. 1998. Histoire du Christianisme des origines à nos jours, vol.3. Les églises d'orient et d'occident. Paris. PLRE II-III. Prosopography of the Later Roman Empire, ed. Martindale, J., et al., Cambridge 1981-92. Pollard, N. 2000. Soldiers, cities and civilians in Roman Syria. Ann Arbor. Procope, Guerres, éd. Haury, J., rév. Wirth, G., 2 volumes. Leipzig, 1963. Rosen, K. 2001. «Justinus I», RAC 19: 763-78. Salway, B. et Drinkwater, J. éds. 2007. Wolf Liebeschuetz Reflected. Londres. Sartre, M. 1982. Trois études sur l'Arabie romaine et byzantine. Bruxelles. Segal, J.B. 1970. Edessa. The Blessed City. Oxford. Di Segni, L. 1995. «The involvement of local, municipal and provincial authorities in urban building in late antique Palestine and Syria», in J.H. Humphrey, éd., The Roman and Byzantine Near East, vol.1: 312-32. Portsmouth, RI. Sévère, Hymnes, éd. et tr. Brooks, E.W., The Hymns of Severus, PO 7 (1911). Sévère, Homélies cathédrales (CIV-CXII), éd. et tr. M. Brière, PO 25 (1943). Sévère, The Sixth Book of the Letters of Severus, Patriarch of Antioch, éd. et tr. E.W. Brooks. Londres 1903. Sévère, Collection of Letters, éd. et tr. E.W. Brooks, PO 12 (1919). Shahîd, I. 1995. Byzantium and the Arabs in the sixth century, vol.1.2. Washington D.C. Socrate, Kirchengeschichte, ed. G.C. Hansen, GCS. Berlin 1995. Stein, E. 1949. Histoire du Bas-Empire, vol.2. Paris. Théodore Lector, Kirchengeschichte, éd. G.C. Hansen, GCS. Berlin 1995. Théodoret de Cyr, Historia Ecclesiastica, éd. L. Parmentier, rév. G. Hansen, GCS. Berlin 1998. Théophane, Chronographia, éd. de Boor, C. Leipzig 1883; tr. C. Mango et R. Scott, The Chronicle of Theophanes Confessor. Oxford 1997. Théophylacte Simocatta, Histoires, éd. de Boor, C. rev. Wirth, P. Stuttgart 1962; tr. M. et M. Whitby. Oxford 1986. Trombley, F.R. 1997. «War and Society in Rural Syria c.502-613 A.D.: Observations on the Epigraphy», BMGS 21: 154-209. 297

Archaeological and Ancient Literary Evidence for a Battle near Dara Gap, Turkey, AD 530: Topography, Texts and Trenches Christopher Lillington-Martin York College

This paper explores how images attained from satellite1 and field photography2 can be studied alongside historical literary sources to improve our understanding of the context for, and details of, an historical event. The event is a battle fought in June/July AD 530 between the Romans and the Sassanian Persians near Dara in the south east of modern Turkey. After outlining the importance of the region near this battle site, three translations of sixth century historical sources will be analysed alongside the satellite and field photography. The sources are Zachariah and Malalas, who offer secondary accounts of the battle, and Prokopios, General Belisarios’ legal adviser, who provides us with an eyewitness account.

unspecified, year, “extraordinarily heavy rains fell…the river…rose in high flood… spread over…whole city…market place… streets… houses… sweeping onward a great mass of furniture (etc)… plunging into this pit it disappeared underground. Not many days later it emerged near Theodosiopolis… 40 miles from Dara…recognised by objects… carried off from houses of that city”.5 In 1895, Dara was described as “one of the best watered villages in Mesopotamia”.6 In October 2005 a villager informed me that the river had not flowed since 1990 but that cereal crops grow on the plain to the south without artificial irrigation. Darker areas on the satellite images (figures 4 and 13) suggest water availability near Dara in 2004. Perhaps not surprisingly, Dara’s fortified existence provoked major Persian invasions in 530, 540, 544 & 573. In 540 and 544 Dara was attacked (unsuccessfully) by Khosrau I who eventually captured it in 573. In 586 Maurikios’ Roman army defeated the Persians at Solachon7 (near Dara) and recaptured the fortress. In 604 the Persians under Khosrau II captured and destroyed Dara but it was rebuilt in 628 by the Romans. From 639942 the Arabs captured and held Dara and it was sacked by Romans in 942 and re-sacked in 958 by John Tzimisces. In 530 the Romans and Persians had been at war intermittently since 5028 and although both sides purported to be willing to seek peace they each had tens of thousands of troops in the region in the early summer of that year. The Persians seized the initiative and invaded Roman territory from Nisibis with the apparent intention of capturing the fortress of Dara. With 25,000 men the Roman commanders had too many to supply during a siege, if they had garrisoned Dara, but enough to meet the Persians in the field at a well prepared defensive position of their choice.

Figure 1: NASA World Wind screenshot showing location of battle site. The site of the battle near Dara is worth investigating further because it is one of very few in the ancient world to have a relatively secure location: south of the fortress city of Dara (modern Oguz) in the Mardin region of SE Turkey close to the Syrian border. Dara had been heavily fortified by Anastasias in AD c.506 and its site was chosen because, apart from being close to the Roman– Persian border of the time and being protected by arid hills to north, east and west, it had an excellent supply of two fundamental natural resources. Firstly, there was an abundant supply of stone3 for building its fortifications out of direct line of sight of Nisibis. Additionally, Dara had another important element for a fortress’s defence: water. Prokopios wrote “close by there is a spring placed by nature among precipices, forming a large river”,4 which winds down from the north and in one,

The battle near Dara is of interest to historians and archaeologists alike because it was recorded by an eyewitness, Prokopios. The testing of written battle accounts through archaeological investigation has achieved some degree of success with battles from later periods (e.g. Pollard 2005)9 but there is scope for it to be applied to the field of ancient warfare. Prokopios’ literary technique and perception of battles can be linked to Averil Cameron’s aim “to lay down the foundation for a

5

Prokopios Buildings II.ii.10-16 Parry 1895 7 Haldon 2001 8 Greatrex 1998 9 Pollard 2005

1

6

World Wind 1.3.2 2 Author’s photography from field visits in 2002, 2005 and 2006. 3 Nicholson 1985 4 Prokopios Wars VIII.vii.7

299

THE LATE ROMAN ARMY IN THE EAST

Figure 2: from Ambar north to Dara with dry pebble riverbed winding through centre ground (October, 2005). This is the traditional site of the battle.

Figure 3: from Ambar church roof eastwards with low hills in background and probable battlefield between, (May 2006). This is the proposed site of the battle. “THE THIRD CHAPTER OF THE NINTH BOOK, CONCERNING THE BATTLE BEFORE DARA The Persians were proud and puffed up and boastful; and, indeed, the Mihran and the marzbans assembled an army and came against Dara and encamped at Ammodis, being fully confident in the expectation of taking the city, because the Roman army had been diminished by their sword. And their cavalry and infantry approached and came up on the south side of the city, intending to encompass it all round for the purpose of blockading it; but a Roman force met them by the help of our Lord, who chastises but does not utterly deliver over unto death. For a certain Sunica, a general, who was a Hun, and, having taken refuge with the Romans, had been baptized, and Simuth (?),*a Roman tribune, and their armour-bearers with twenty men each drove the whole Persian army away from the city several times, passing boldly and vigorously from one part of the field to another, and cutting men down right and leftwith the lance. And they were practised in the use of the sword; and their cry was loud and terrific, and made them appear terrible to the Persians, so that

different and more realistic way of interpreting Prokopios”10 described as “a talented reporter who saw history in terms of personality, and whose forte was not analysing but describing events”.11 In his battle account Prokopios describes a Roman “deep trench” which could probably be located during an archaeological survey. The trench design dictated the tactics used by both armies and according to Prokopios the Roman victory was gained via advanced control of cavalry along it. Naturally, this is more difficult than controlling infantry and it is conceivable that the trench also served as a physical point of reference for the various unit commanders so that overall army cohesion could be maintained by the Roman army commanders (Belisarios and Hermogenes). Therefore, if the trench were to be located, we may, to some extent, be able to learn about late Roman temporary fortification design and better appreciate some of the generals’ planning ideas. Our first ancient historical literary source is Zachariah of Mitylene / Zacharias Rhetor, “The Syriac Chronicle”:12 10

Tougher 1997 Tougher 1997 12 Pearse, 2002 11

300

CHRISTOPHER LILLINGTON-MARTIN: ARCHAEOLOGICAL AND ANCIENT LITERARY EVIDENCE FOR A BATTLE NEAR DARA GAP

Figure 4: NASA World Wind screenshot showing regional topography. The Persian army was based at Nisibis (modern Nusaybin, Turkey) and the Persian pre-battle camp was sited at Ammodios (modern Amudah, Syria). The Roman fortress-city of Dara (the modern village of Oguz) and the hill at Ambar - both within Turkey – are also shown. The white line indicates a line of site from Ambar to Nisibis whilst the black one demonstrates that Dara city could not be seen from Nisibis. they fell before them : and two of their leaders were killed, besides no small number of horsemen; while of the faige, who are the Persian infantry, many were cut down and hurled back by the Helurians,** under Butzes, to the east of the city.

the Romans and carried some of their dead back to their base at Nisibis.

And the Persians, when they saw how great the number of the dead was, acted craftily, and sent to Nisibis, asking them to bring as many baggageanimals as possible and come at once to Dara, and take as much spoil as they could. And, when large numbers came, they loaded them with the bodies of their slain, and then they returned in shame. However, the rest of the Persian force invaded Roman ‘Arab, and burned it with fire”.13

“In that year (AD 530) during the consulship of Lampadios and Orestes, Hermogenes and the magister militum Rufinus were sent as ambassadors of the Romans into Persian territory. When they had reached Dara, which had been renamed Anastasioupolis, they sent a message to Koades, the emperor of the Persians, who put off receiving them. While they were staying at Dara together with the magister militum Belisarios, the other exarches and the army, they encamped outside the city and awaited the response from the Persian emperor. Meram, the chief Persian exarch, and the son of the Persian emperor, with other Persian exarches were based at Nisibis. When they discovered that the Romans were encamped outside Dara, the Persians divided their forces into three commands and attacked with 70,000 men.

The second source is Malalas’ account which emphasises the political context:

*

Probably the “Simmas” of Prokopios ....The Herulians, whom the Greeks frequently called Helouroi, are meant. **

So Zachariah states that the Persians approached Dara from the south (confirming the logical approach dictated by topography cf. satellite imagery above), were met by 13

cf. Greatrex and Lieu 2002 for an alternative translation.

301

THE LATE ROMAN ARMY IN THE EAST Realizing this, the Roman exarchs came out against the Persians and attacked them. In the battle that followed the Persian and Roman armies fought at close quarters; the Romans cut the Persians completely to pieces and also captured a Persian standard. Meram fled with a few men, including the emperor’s son, and escaped to Nisibis. In this conflict a Persian exarch named Sagos was slain; the dux Sounikas, a Roman exarch, had challenged him to single combat. Here with the dead strewn over the ground, could be seen a victory over Persian folly”. Book 18, 50. Bo. 452.13-453.1414

perhaps even partially rhetorical because an army of that size would require several square kilometres in which to manoeuvre. Indeed Malalas mentions a camp outside the city (see above). However, if we consider the topography, we can see Dara’s funnel shaped plain has a strategic point (from Ambar in the west to the hills in the east) which would be a likely site to construct a defensive trench as it would have allowed space for the Roman army to have manoeuvred north of it whilst defending a narrow gap at the trench. This interpretation brings the trench and battle south 2.5 km away from Dara’s walls. This is significant not only to identify a more precise site for the battle but also because the site is less likely to have been affected by subsequent siege actions which would have taken place within a few hundred metres of the walls in AD 540, 544, 573, 604, 628, 639-942 and 958. In addition the Persians would have been funnelled into an ever narrowing space as they advanced upslope from Ammodios towards the Roman lines at the trench near Ambar.

So Malalas states the Roman army was encamped outside Dara, 70,000 Persians were divided into three commands and they were totally defeated by the Romans. The Persians are described as losing a standard, escaping back to their base at Nisibis, leaving many dead on the field of battle and all ascribed to “folly”. Perhaps the descriptions of Persian pride and folly can be explained when we consider Prokopios’ eye-witness account. Firstly, he describes Roman battle preparations:

NASA World Wind 1.3.2 topography analysis also allows for the capture of terrain elevation data. The Persian march westwards from Nisibis to Ammodios was practically level but the approach towards Dara was up a noticeable incline.

“Not far from the gate15 (ʌȪȜȘȢ) which lies opposite the city of Nisibis, about a stone’s throw away…they (Belisarios and Hermogenes) dug a deep trench with many passages across it...not dug in a straight line. In the middle there was a short portion straight, and at either end of this there were dug two cross trenches at right angles to the first...they continued two straight trenches...to a very great distance...” 16 However, published plans17 and photography of Dara (figure 5) show there is only the circuit wall with towers and no evidence for a “gate…opposite… Nisibis” in the literal sense. Indeed, there are hills between Dara and Nisibis which are not in sight of one another (figures 4 and 6). Prokopios also uses the measurement of “a stone’s throw” during his account of the siege of Rome where he describes Hadrian’s Mausoleum as having four equal sides of “about a stone’s throw in length”18. Fortunately, as the structure survives, this gives us a possible distance of approximately 84 metres for this expression of measurement.

If we deploy an alternative translation of Prokopios’ reference to the word ʌȪȜȘȢ as a gap19 rather than a gate this interpretation is further supported, and the ancient source is better appreciated because, as we look at the topography, we see that there is a gap in the terrain to the east of Ambar which does indeed lie opposite and within sight of Nisibis (figure 4). Therefore, Prokopios has provided us with an accurate description of the battlefield and Roman defensive positions within the landscape. This adds weight to his reliability for other topographic descriptions related to military actions elsewhere in his writings and we should perhaps reconsider the accuracy of other ancient sources in this regard. A possible approximate line of the Roman trench is indicated by a white in the gap between Ambar in the west and the low hills in the east. Note the dry river-bed (arrowed) which is not mentioned by Prokopios.

No precise distances are given from the ʌȪȜȘȢ to the trench and from the central trench section to the trench ends. However, at least Prokopios has provided an approximate position and shape of the trench. The idea of such a significant military fieldwork being only a stone’s throw from a city wall gate with an army of 25,000 men placed between it and the city could lead us to dismiss Prokopios’ description of distances as impractical and

Prokopios explains that the battle centred on the “deep trench” with the 25,000-strong Roman army between it and Dara. The Loeb translation uses the English word “trench” five times whereas Prokopios uses two different terms: IJȐijȡȠȞ for “trench” and țİȡĮȚĮȚ for “arm of the

14

Malalas Previous scholars have interpreted this as meaning the battle was fought very close to the walls of Dara. 16 Prokopios, Wars I.xiii.13-15 17 Ahunbay, Croke & Crow, Greatrex, Sinclair, Whitby and Zannini. 18 Prokopios, Wars V.xxii.13 15

19

I am grateful to Dr Noel Lenski (Colorado) for pointing out this alternative translation.

302

CHRISTOPHER LILLINGTON-MARTIN: ARCHAEOLOGICAL AND ANCIENT LITERARY EVIDENCE FOR A BATTLE NEAR DARA GAP

Figure 5: Dara’s wall facing in the direction of Nisibis.

Figure 6: Dara’s walls from hills to south east.

Figure 7: NASA World Wind screenshot showing the topography of plain to the south of Dara with its narrowest point indicated by -------------. Note the dry riverbed (pale line) running north-south from Dara skirting Ambar and then heading south west.

303

THE LATE ROMAN ARMY IN THE EAST Terrain Elevation

From

To

Distance Km Stades

550m

Dara (Oguz), Turkey

6? km 27 km

28 98

Nisibis

8 km 24 km

20

460 m

Frontier, c 530 Nisibis (Nusaybin) Ammodios (Amudah, Syria) Ammodios

470 m

Ammodios

Ambar Battlefield (on way to Dara)

4 km 4-5 km

510 m 540 m

Battlefield Ambar hill

Dara Dara

3-4 km 3 km

Figure 8: World Wind 1.3.2 terrain elevation data showing the relative elevation of the sites concerned.

Figure 9: this high resolution digital topography data (by Dr Mark Matney NASA) shows how the approach to Dara (marked with by a small oval) from the south funnels into a narrow valley. I have added a large oval to suggest an area for the trench and battle.

304

CHRISTOPHER LILLINGTON-MARTIN: ARCHAEOLOGICAL AND ANCIENT LITERARY EVIDENCE FOR A BATTLE NEAR DARA GAP

Figure 10: view north towards Dara with Ambar on the hill west of the gap (October 2005).

Figure 11: view south from Dara with Ambar on the hill to the west of the gap (October 2005).

Figure 12: NASA 1.3.2 World Wind screenshot view north towards Dara.

Figure 13: NASA 1.3.2 World Wind screenshot view south over Dara towards Syrian plain.

trench”. On the Roman left (eastern) flank the trench is described20 as extending from “..the left straight trench...as far as the hill..” (assuming Prokopios was describing things as if observing from Dara to the north). The Greek ȜȩijȠȞ implies a low hill. Prokopios relates Roman deployment and tactical manoeuvres to the trench and hill in the landscape.

On the Roman left flank Bouzes commanded a “large force of horsemen” and Pharas supported him with 300 Herulian cavalrymen. Between them and the centre “..outside the trench, at the angle formed by the cross trench and the straight section..” were positioned Sunicas and Aigan with 600 Hun cavalry. “On the other (western)...extremity of the straight trench....at the angle on the right, 600 (Hunnic) horsemen took their stand commanded by Simmas and Ascan,..that… they might move out from there…” . The Roman right (west) wing

20

Prokopios, Wars, I.xiii.19, 20, 21& 22

305

THE LATE ROMAN ARMY IN THE EAST

Figure 14: area of low hills to the east of the plain (south of Dara and east of Ambar).

Figure 15: Roman deployment at the battle south of Dara, AD 530. 25,000 Romans, they would have had to have been deployed in formations at least 10 ranks deep to fit in the space available. Therefore the Persians would have had to have been deployed approximately 20 ranks deep or in two or more lines of fewer ranks. Prokopios’ trench location and design therefore warrants detailed consideration. It is perhaps worth re-evaluating the Loeb (1979: 106) diagram of the trench which, if inverted and lengthened, might help make the Roman tactics described by Prokopios easier to appreciate. Roman deployment may have actually looked something like figure 15 below.

was commanded by John, son of Nicetus, with subordinate commanders: Cyril, Marcellus, Germanus and Dorotheus. So “..all along the trench stood the detachments of cavalry and the infantry”. “Behind these in the middle stood the forces of Belisarios and Hermogenes” as a reserve. Therefore understanding his description of the “deep trench” design would go a long way to understanding the battle tactics better. This would be something to seek remains of on the Dara plain if it were surveyed archaeologically. In order to protect the Roman army’s front, the trench would have had to have been practically as wide as the plain and with height and depth at least matching those of a marching camp. The Dara plain is approximately 2 km wide east of Ambar. If there were

It is also worth noting that from the Persian viewpoint it would have only been possible to easily perceive the two “arms” of the trench defended by Roman cavalry with 1,200 Huns manoeuvring in between perhaps concealing 306

CHRISTOPHER LILLINGTON-MARTIN: ARCHAEOLOGICAL AND ANCIENT LITERARY EVIDENCE FOR A BATTLE NEAR DARA GAP

high ground (“..with the help of God..”) before gaining a victory “..by fastening the letters written by each of us on top of our banners”23. This can be compared with the previous Persian attack on the Ephthalitae Huns where Prokopios relates their leader “...hung from the top of the royal banner the salt over which Perozes had once sworn...”24 again before securing a great victory. Incidentally, Prokopios records that the Persians were commanded by another Perozes when they approached Dara. Perhaps, as Prokopios is the only one of our sources to offer “Perozes” as the Persian commander’s name, he was applying irony by giving the Persian army commander a name meaning “victory”.

the central trench. The Roman infantry and Belisarios’ Roman cavalry reserve may have been visible to the Persians beyond the Huns because the slope rises northwards. This description by Prokopios may be linked to another trench21 apparently dug by Ephthalitae Huns to thoroughly defeat the Persians which is worth considering briefly as he reports that the Persians under King Perozes (a name meaning “victory”) had invaded Hunnic territory (AD c.482) where the Huns also “...made a deep trench...” concealed by means of reeds and earth. Then “...in the centre he (the Hunnic leader) left a small portion of land intact,..as a way for ten horses…” Prokopios describes the battle as starting after a small force of Huns was sent across the “portion of land” to provoke the Persian cavalry army which launched into a headlong charge across the plain and fell into the trench. Although not impossible this story begs certain questions about the dimensions of the trench, how reeds could have concealed it and where the surplus soil would have been deposited. Rather than a deep trench, perhaps it is more likely to have been a series of “lilia” style holes, possibly with stakes inside, which could have been more easily prepared and covered with reeds and the surplus earth scattered on top. These would have disabled a cavalry charge and allowed the Ephthalitae Huns to finish off the Persian cavalry.

To return to the battle south of Dara, the “Persian army...of 40,000...all stood close together.(.in a ) ... phalanx...deep”. On seeing the good order of the Romans the “Persians seemed...at a loss what to do”. On the first day of battle a Persian probe from their right wing against the Roman left resulted in a skirmish with seven Persian bodies taken by the Romans. Overnight 10,000 more Persians arrived. The next morning Prokopios has Pharas, the Herul, make a tactical proposal to Belisarios and Hermogenes25 stating “..if we conceal ourselves on this slope (țȐIJĮȞIJİȢ)...and then, ...we climb up by this hill...” (this is probably same hill referred to before26 as the same word (ȜȩijȠȞ) is used) “..and suddenly come upon their rear, shooting from behind them, we shall…do them the greatest harm”. This could be one of Prokopios’ rhetorical devices perhaps to show Belisarios as an approachable commander taking advice from one of his (non-Roman) subordinates as it would be reasonable to expect a commander-in-chief, who had ordered the construction of field defences at a strategic point, to have noticed the ambush potential of low hills to the east. Prokopios goes on to describe the ensuing battle as beginning with “vast clouds” of arrows being exchanged with the Persian advantage in numbers and rotating fresh troops compensated by the Romans having a steady wind blowing in their favour. Considering tens of thousands of these soldiers were equipped with the bow, several hundred thousand arrows will have been shot. Then the Persian right wing commander, Pityaxes, led the Cadiseni and others to force the Roman east wing back. At this moment the Heruls under Pharas descended the hill on the east flank and attacked the Persians from behind with the Cadiseni suffering badly. Next Sunicas and Aigan charged from the centre into the now exposed Persian internal left flank resulting in 3,000 Persian casualties and the remainder escaping to return to their original position on the right wing of the Persian centre. The Romans returned to their positions along the eastern arm of the trench which they will have been ordered to remain at. With so many casualties on the ground to their right,

Figure 16: “Lilia” at Rough Castle, Scotland22 It may also be the case that Prokopios places this story to foreshadow the events at Dara but instead of having the Persians attacking the Huns employed by the Romans in the centre of the line and falling into “... a deep trench...” as he has described them do against the Ephthalitae Huns, he has them avoid the trench by attacking the Roman flanks. Apart from the mention of trenches, Prokopios presents another potentially rhetorical parallel between these two battles, for his careful readers. This is where he writes that Belisarios and Hermogenes seized the moral

23

Prokopios, Wars, I.xiv.10 Prokopios, Wars, I.iv.9 Prokopios, Wars, I.xiv.33 26 Prokopios, Wars, I.xiii.19 24

21 22

25

Prokopios, Wars, I.iv.7-8 http://www.ourpasthistory.com/roman_scotland/Imgp0696.jpg

307

THE LATE ROMAN ARMY IN THE EAST

DARA

AMBAR

NISIBIS

Figure 17: World Wind screenshot showing the topography with proposed Roman and Persian deployment. Perozes committed his whole reserve (the Immortals) and his left wing, commanded by the one-eyed Baresmanas, to an attack on the Roman right. Belisarios and Hermogenes were well-placed (being on slightly higher ground) to observe this and ordered Sunicas and Aigan to join Simmas and Ascan at the trench angle west of centre and placed many of Belisarios’ men behind them. As Baresmanas’ command and the Immortals forced John’s Roman left wing command to make a hasty retreat, the Romans in the trench angle (led by the Huns) attacked the Persians, via the crossing points, against their unshielded internal right flank and cut their force into two leaving the standard bearer of Baresmanas in the group behind. Sunicas killed him which provoked the Persians ahead to turn around but they were now exposed on two sides so John’s men returned to join in the melee. The Immortals attacked Sunicas who killed Baresmanas which caused the Persians to panic and rout with 5,000 casualties.

angles; cavalry commands under John, son of Nicetus, (west) and Bouzes (east) with 300 mounted Heruls in ambush behind a low hill further east. The Persian army has a reserve of “Immortals” to the south, Perozes “at the middle of the front”, cavalry on each flank commanded by the one-eyed Baresmanas (west) and Pityaxes (east). The exact position of the infantry is not specified but it may have been surrounded by the cavalry. The Persian camp is to the south at Ammodios and the Persian base at Nisibis is to the east. According to Prokopios the Persians were therefore outmanoeuvred by Roman cavalry tactics including an ambush from behind the low hill and internal flank attacks from the centre (made possible by the trench design having crossing points) executed by wellpositioned cavalry units and Belisarios’ reserve. The entire plan depended on the topography and trench design to use tactics which exploited Persian folly and pride.

The NASA World Wind screenshot below (figure 17) shows the topography with proposed overall Roman and Persian27 deployment superimposed upon it. The Roman deep trench stretches across the plain at its narrowest point from Ambar towards the hills in the east. The Roman army is positioned between the fortress city of Dara and the trench in seven commands: a reserve under Belisarios to the north; infantry between it and the trench; two units of 600 mounted Huns in the central trench 27

All three sources agree that the Persians attacked from the south, that Nisibis was their base and that the Persians suffered very heavy casualties. Zachariah claims Persian dead were taken to Nisibis but does not mention a trench. Malalas states that there were many Persian dead left on the battlefield and indicates that the Roman army had previously encamped outside Dara. It was standard practice for Roman (and Persian) armies to build camp defences so Malalas’ testimony would increase the probability of at least Roman camp defence trenches

Prokopios, Wars, I.xiii.19-22 & I.xiv.29-33

308

CHRISTOPHER LILLINGTON-MARTIN: ARCHAEOLOGICAL AND ANCIENT LITERARY EVIDENCE FOR A BATTLE NEAR DARA GAP

Figure 18: eastern end of battlefield with the field boundary dip in mid ground.

Figure 19: Dr Tony Pollard, Centre for Battlefield Archaeology, Glasgow University beside the dip crossing the field diagonally for at least 100 metres background). In the middle ground there is a field boundary with a distinctive dip in it.

having been dug on the plain south of Dara. It may be that such trenches were originally to protect the camp and then made use of during the battle or, perhaps more likely, they were designed with a dual purpose. Malalas agrees with Prokopios about the Romans capturing a Persian standard but records an army of 70,000 Persians. Prokopios’ 50,000 is possibly more reliable as he was an eye-witness.

Whether or not this dip is a natural or archaeological feature with any relation to the trench of AD 530 remains to be confirmed as part of a disciplined archaeological survey. Artefacts and debris related to the battle (arrowheads, blades, armour fittings, horse furniture etc.) possibly remain to be located there to contribute to our knowledge of late antique military equipment. Data collected during such a survey could be used to create a model of the battle and there is the potential to inform us of Belisarios’ tactics.

Combined, this evidence improves our understanding of the battle near Dara, late antique cavalry tactics and Prokopios’ literary technique within conflict descriptions on which there is scope for more work to be done. As the area is now used for arable agriculture within field systems, to the south of Dara and east of Ambar, it would be reasonable to expect part of the field defences to survive in the landscape as an archaeological feature. The likelihood of re-locating the trench is reasonable and the potential for good levels of archaeological preservation are high. For example, the photograph below was taken in October 2005 from the eastern edge of the plain looking south west towards Ambar (on the hill in the

The ruin of a Christian church on the hill at Ambar overlooking the probable battlefield has been dated to the early sixth century28 and it is conceivable that this building may have been associated with the battle. Perhaps it had a partially military function to observe

28

309

Mango 1982

THE LATE ROMAN ARMY IN THE EAST

Figure 21: part of a sixth century Christian church, Ambar, with battlefield to east.

Figure 20: view from the south-west end of the dip looking towards the low hills to the east of the plain (perhaps where the Heruls laid their ambush).

granted me funds to visit SE Turkey (including Dara) in May 2002 as well as additional funding to support a further Dara battlefield specific visit in October 2005 when the villagers of Dara-Oguz and Ambar were most helpful. In addition, I should also like to thank the organisers and participants of the congress who discussed my ideas with me and the many others who have responded to my enquiries enabling me to further my research.

Persian movements from Nisibis or there is the possibility that it was built as a monument to the Roman victory. In conclusion, the reinterpretation of the topographic and literary evidence presented suggests the battle was fought approximately 3 km south of Dara along a trench running “...a stone’s throw away” from a geographical gap, in the vicinity of Ambar, eastwards towards the hills rather than “...a stone’s throw away” from the walls of Dara. Field photography and satellite imagery show the contemporary landscape is farmland conducive to an archaeological survey. An archaeological survey could aim to locate and map the trench and seek battle debris to create a battle model. The survey data could be used for further detailed reconciliation of archaeological and literary evidence. If successful, further investigations seeking debris finds could better inform us about battlefield tactics and late antique Roman and Persian equipment as it would be reasonable to expect that some debris would be found there. It is hoped that the method used to explore how images attained from satellite and field photography can be studied alongside historical literary sources to improve our understanding of the context for, and details of, historical events more accurately, could be applied to other sites.

Bibliography Ahunbay, M. 1991 “Dara - Anastasiopolis 1990 Yili Çalismalari,” Kazi Sonuçlari Toplantisi 13/I Croke, B. & Crow J. 1983, ‘Procopius on Dara,’ Journal of Roman Studies 73, 143-59. Greatrex, G. 1998 Rome and Persia at War, 502-532. Leeds: Cairns Greatrex, G. and Lieu, S.N.C. 2002 The Roman Eastern Frontier and the Persian Wars, A.D. 363-630, London: Routledge Haldon, J. F. 2001, The Byzantine Wars : Battles and Campaign of the Byzantine Era, Stroud: Tempus http://www.ourpasthistory.com/roman_scotland/Imgp 0696.jpg Malalas, ed. Dindorf, 1986 translated by E. Jeffreys, M. Jeffreys, and R. Scott, The Chronicle of John Malalas (Melbourne) Matney, M. http://www.ghg.net/matney/Dara/Mundell Mango, Marlia 1982 “Deux Eglises de Mesopotamie du Nord: Ambar et Mar Abraham de Kashkar,” CahArch 30, 47–70 Nicholson, O. 1985. ‘Two notes on Dara’, AJA 89: 66371 Parry, Oswald H. 2004 Six Months in a Syrian Monastery, 1895 & Gorgias Press Pearse, R. 2002 http://www.ccel.org/p/pearse/ morefathers/zachariah09.htm

Acknowledgements I would like to acknowledge the support I have from my supervisors, Dr Janet Huskinson (The Open University) and Dr Tony Pollard (University of Glasgow); assistance with Prokopian Greek from Dr Marina Bazzani (University of Oxford) and with satellite imagery from Dr Mark Matney (NASA Johnson Space Center). I would also like to thank Dr Geoffrey Greatrex (University of Ottawa) for frequent support and who made my visit to Dara possible in 2002. Dr Ken Dark (University of Reading) has been another encouraging supporter. The British Institute of Archaeology at Ankara (BIAA). 310

CHRISTOPHER LILLINGTON-MARTIN: ARCHAEOLOGICAL AND ANCIENT LITERARY EVIDENCE FOR A BATTLE NEAR DARA GAP

Pollard, T. 2005. Archeological Survey at Culloden Battlefield. GUARD Monograph report No 1981 Prokopios. Edited by Dewing H. B. 1914-79. 7 vols. Loeb Classical Library. Cambridge, MA: Harvard University Press and London, Hutchinson, Greek text and English translation. Sinclair, T. A. 1987-1990 Eastern Turkey: An Architectural and Archaeological Survey, Volumes IIV, Pindar Press Tougher, S. 1997 http://www.dur.ac.uk/Classics/histos/1997/tougher.ht ml as retrieved on 23 Sep 2005 Whitby, M. 1986 “Procopius’ Description of Dara”,’ in ed. P. Freeman & D. Kennedy, The Defence of the Roman and Byzantine East (vol.2) 737-83. “World Wind” 1.3.2 lets you zoom from satellite altitude to any place on Earth. Leveraging Landsat satellite imagery and Shuttle Radar Topography Mission data, World Wind lets you experience Earth terrain in visually rich 3D. Free download available from http://worldwind.arc.nasa.gov/ Zachariah of Mitylene, Syriac Chronicle (1899). Book 9: http://www.ccel.org/p/pearse/morefathers/zachariah09 .htm (text transcribed Roger Pearse, Ipswich, 2002) Zanini, E. 1994, 1998 ‘Introduzione all’Archeologia Bizantina’. Roma.

311

The Byzantine Military in the Petra Papyri - a Summary Zbigniew T. Fiema The Academy of Finland

The excavations of a large Byzantine church complex in the center of Petra in southern Jordan by the Petra Church Project, between 1992 and 1997, have provided a plethora of information concerning the ecclesiastical architecture and the material culture of Byzantine Petra.1 However, it was the discovery of the large corpus of papyrus documents in Room I of the church complex in 1993 which has unexpectedly opened a new avenue of research by providing much needed documents related to the socio-economic history of the city in the 6th century. The conservation of the carbonized papyrus scrolls was carried out by the Finnish team under the direction of Jaakko Frösén. The publication of the Petra Papyri is being carried out by teams from the University of Helsinki and the University of Michigan, headed by Jaakko Frösén and Ludwig Koenen, respectively.2

emerged, at the end of the 4th century, as a new province of Palaestina Salutaris/Tertia, with Petra as its main administrative center. This province included southern Jordan, southern Palestine (Negev) and the Sinai. The security of this province as well as the other two Palaestinae was accorded by the army commanded by dux Palaestinae. Under his disposition were four units of Equites Illyriciani and one of Equites Thamudeni Illyriciani, two units of Equites promoti indigenae, five of Equites sagittarii indigenae, and six alae, all of them cavalry. Palestine was also garrisoned by eleven cohorts of infantry and one, possibly two, legions. The available information is, unfortunately, either incomplete or ambiguous concerning the location of the military units in the area of Petra. Whether the city itself was garrisoned or not during the Byzantine period is unknown. An inscription related to the conversion of the Urn Tomb to a church, dated to AD 446, mentions a numerus6 stationed there either permanently or temporarily. The unit was tentatively reconstructed as numerus Tertiodalmatarum.7 In the Petra’s hinterland, according to the Notitia Dignitatum (Or. 34; 1876: 7274), major garrisons were stationed in Admatha (modern al-Ⱦammam) to the east of Petra, in Arieldela ('Ain Gharandal) to the southwest, in Robatha (Khirbat Ruwath)8 to the north, and in Zodocatha/Kastron Zadakathon (ɋadaqa) and Hauana=Hauarra (Ⱦumeima), both to the south of Petra. Some military units were presumably stationed in Bi'r Madhkur, Wadi Sabra, and in the environs of Udhrudz, if not in Udhrudz itself.9 Other military installations - fortlets or watchtowers - around Petra, Udhrudz and Ⱦumeima, which cannot otherwise be identified with ancient names, include QaǺr Umm alRattam, QaǺr Wadi at-Tayyiba, al-Mutrab, Ail, and Fardhakh, among others.10 On the basis of the survey material, it appears that all these sites were occupied, and presumably garrisoned, at least during the 4th century and most probably also later. Unfortunately, only a few of these structures have been accorded more extensive exploration in terms of excavations or soundings. In most of the other cases, only surface sherding and a general description of the ruins or surface wall lines was effected.

The papyri are private papers of Theodoros son of Obodianos, a deacon and then archdeacon of the Church of the Virgin Mary in Petra, presumably the church where the papyri were found. The earliest dated scroll bears the date of AD 537, while the latest one is from AD 592.3 Basically, all of them are documents related to Theodoros and his extended family, which pertain to the sale, acquisition, transfer, and registration of landed property for tax purposes as well as sworn and unsworn statements concerning possession, disposition, inheritance, marriage agreements, and various disputes. The texts have also provided some information on the military presence in Petra’s hinterland in the 6th century. Although this information is not extensive, it is certainly worth reviewing in light of a general dearth of sources concerning this subject. The preliminary observations have already been published some time ago,4 but the subject definitely needs updating and reassessment, as presented in this article. Military Presence in southern Jordan (4th-7th centuries AD) Extensive but still poorly understood administrative changes which affected the area of Roman Palestine and Arabia during the 4th century are still the subject of considerable debate.5 However, there is common agreement that the entire area south of the Wadi al-Ⱦasa,

6

Brünnow and Domaszewski 1909: 345. Sartre 1993: 82. 8 For recent explorations at the site, see Walmsley and Barnes 2002. 9 For site identifications, dating of occupation and further information, see Fiema 1991, Appendix 3, and 1995: 263-266; Tables 1, and 2. The question of the military garrison at Udruh is unresolved. For the review of opinions and the alternative interpretation, see Isaac 1995: 140-141, 149, note 112. 10 Graf 1992, 1995, 1997a: 279-280, and 1997b, for a review of these and other sites. Doubts about the military purpose and nature of occupation at al-Mutrab are expressed by Findlater, based on his recent exploration of the area (2002: 141-142). 7

1

The project was organized by the American Center of Oriental Research, and directed by Pierre M. Bikai, ACOR Director, with the author in charge of the fieldwork. 2 See P. Petra I, 2002, for the full bibliography of publications. 3 P. Petra I 1; Frösén 2004: 141. 4 Fiema 2002a and 2002b. 5 E.g., Brünnow and Domaszewski 1909: 273-80; Mayerson 1984, 1987; Tsafrir 1986, Fiema 1991: 128-33, and Sipilä 2004.

313

THE LATE ROMAN ARMY IN THE EAST As such, not only the occupation span but also the identifaction of some of these structures with military purposes (forts/fortlets) vs. civilian use (caravanserais, fortified farmsteads) may not always be secure.11

Lakhmid clients of Persia, and the participation of foederati as auxiliaries in Byzantine campaigns. After the elevation of Harith b. Jabala, the Ghassanid, to the status of the patrician and the king of all Arab foederati in the Near East, his brother Abu Karib (Abochorabos) declared his vassalage to Justinian in AD 529, and was granted the phylarchate of Palestine (Procop. Bel. Per. I.19.8-13), including southern Jordan and Petra. However, the phylarchates were neither created nor invested with the power to fully replace the existing administrative and military systems in the frontier provinces but rather to implement them. Novellae 102 and 103, issued by Justinian in 536 and referring to the provinces of Arabia and Palaestina Tertia, specify that all military forces in Palestine, including the comitatenses, limitanei and Arab foederati (in this order) were to remain under the command of dux Palaestinae.

Even this incomplete and sometimes ambiguous picture of the military dispositions in the Petra area indicates that military arrangements there were not of an offensive character. No further expansion to the east or southeast was envisaged, nor is one evidenced. Hence, if these arrangements were indeed defensive, the locations of troops and fortifications would reflect two primary considerations: the nature and direction of the threat, and the location of the economically/politically sensitive areas to be defended. Generally, the defences in southern Jordan do not exhibit concern for a large-scale, wellorganized invasion, and the Petra area is no exception. Such a threat did not materialize in the region until the early 7th century Persian invasions. As such, the distribution of the military infrastructure in southern Jordan, as exemplified by the Petra area, was conspicuously uneven and regionalized. The Via Nova Traiana was still a major communication line connecting the major settlements of the countryside, such as Aila, Ⱦumeima, ɋadaqa, Petra and Tuwaneh. Therefore, the distribution of forts along the road guaranteed security of communication. The overall strategy concentrated on internal security, including main population centers or settlement clusters, associated communication lines, probable imperial estates and the resources of the countryside.

The 6th century may also mark some sweeping changes in the military arrangements in southern Jordan. The 1980s surveys claimed evidence for a widespread abandonment of fortifications in the East during the later Byzantine period. South of the Wadi al-Ⱦasa, the major military sites, such as Jurf ed-Darawish, Da'janiya, elMutrab, al-Ⱦammam, Qirana and Quweira, were supposedly abandoned in the late 5th or early 6th century. According to these surveys, occupation probably continued throughout the 6th century only at the forts of Khalde, Kithara, and the garrisoned sites of Udhrudz and Ⱦumeima. Some scholars have regarded this evidence as a sign of military decline and a neglect of security by the central government,13 further reinforced by Procopius’ remark (Anecdota XXIV.12-14, 1969: 282-285) on the neglect of limitanei under Justinian. Concurrently, the military and administrative control over southern Jordan, including the Petra area, was thought to have been largely relinquished to the Ghassanid phylarchs.

For example, the area of Udhrudz, located east of the Via Nova, experienced a notable expansion of agricultural land and settlements during the Byzantine period. Not surprisingly, that region possessed its own micro-defence in the form of military installations proceeding southeast from the cluster of forts near Udhrudz to Ma’an.12 These marked the greatest extent of agricultural and settlement expansion in southern Jordan during the Byzantine period. It is doubtful if all these military measures were simultaneously undertaken in the manner of a planned buildup. Apparently, forts were constructed, rebuilt, regarrisoned and abandoned throughout the Byzantine period, presumably with regard to a constantly changing political and economic environment, and neither a fully linear, nor a defense-in-depth system, were ever established.

However, neither the widespread abandonment nor the total dependence on the phylarchates can be fully sustained in light of the new evidence below. At first, while some military sites indeed feature archaeologically detected discontinuity in the 6th century (probably Ail, Khirbet al-Qirana, B'ir Madhkur, QaǺsr Umm al-Rattam, QaǺr Wadi at-Ʉayyiba, and 'Ain Gharandal), no wholesale abandonment occurred either in the Petra area or in southern Jordan. In fact, some major sites in the Petra area were also garrisoned by regular army units in the later 6th century.

The security would have been provided not only through the presence of regular army units but also through the military alliances between the Byzantine state and the local Arab tribes. The foedus with the Tanukhids in the later 4th century, the Salihids in the 5th century and the Ghassanids in the 6th included the defence of the eastern provinces against nonaligned nomads, the preservation of peace and order in the countryside, the containment of the 11 12

The Evidence from the Petra Papyri Although the papyri do not imply military presence in Petra, two other, well-known localities in the area Zodocatha/Kastron Zadakathon and Ammatha - were certainly garrisoned. The mentioning of these localities in the papyri is not accidental; most of the paternal property 13 See Parker 1986: 149, 152-153, 158, for the survey evidence, and 1986: 152, for his historical interpretation of this evidence.

Fiema 1995, for discussion. Graf 1997: 279.

314

ZBIGNIEW T. FIEMA: THE BYZANTINE MILITARY IN THE PETRA PAPYRI - A SUMMARY of Theodoros was located in these towns.14 It will be instructive to first briefly summarize the history of these towns as known before the Petra papyri were discovered:

central Jordan, is specified in the Notitia Dignitatum as Mefa, and garrisoned by Equites promoti indigenae.25 It would be reasonable to assume that, at a certain point of time, the military moved into the walled town. The mosaics (6th and 8th centuries) mention Kastron Mefaa.26

Kastron Zadakathon can be firmly identified with ɋadaqa some 25 km southeast of Petra.15 The town was mentioned by Ptolemy as ,GSGGNG (Georg. V1.7.5, ed. Nobbe), in the Tabula Peutingeriana as Zadagatta (8.5, ed. Weber), and in the Beersheva Edict (document 2) as ,GJGPGNG.16 The Notitia Dignitatum (Or. 34.24) lists the town as Zodocatha, garrisoned by the Equites promoti indigenae, a cavalry unit. Greek inscriptions from Sinai (Wadi Haggag) mention P­YZWU[ ,GJGPGNG.17 The site is currently occupied by extensive ruins, including a large trapezoidal fortification (ca.120x80 m) with corner and interval towers and a complex internal plan, a necropolis, remains of a temple(?), a water reservoir, and a watchtower. Recent soundings in the fort yielded Byzantine and Early Islamic pottery.18 Kastron Ammatha, also mentioned in the Petra texts, is most probably Admatha, garrisoned by the Ala Antana dromedariorum (ND Or. 34.33). The Beersheva Edict (document 2) mentions ’Allaha.19 This toponym should probably be linked with the site of al-Ⱦammam, some 30 km southeast of Petra, which features ruins of a castellum, a large reservoir connected with an aqueduct, and extensive ruins of a settlement.20 Recently, the military character of the occupation at al-Ⱦammam was questioned, although later Byzantine and early Islamic sherds were found at the site,21 thus eliminating the previous opinion of its early abandonment.

At any rate, a similar situation applies and seems to be supported by the archaeological evidence coming from Sadaqa. The large fortification with corner towers might originally have been a purely military installation. However, the internal arrangement of structures and walls, as can be discerned from the ground-level remains, appears somewhat chaotic and tight. It may be that, at some point of time, the civilian population moved inside the fortified enclosure for protection. This seems the most reasonable interpretation, especially as the Petra texts clearly indicate the presence of a regular army unit in Sadaqa throughout the 6th century and a civilian population there is indicated by the existence of churches, chapels, and private dwellings. Persons having military rank occur in various documents related to the private affairs of the family of Theodoros in Zadakathon and Ammatha.27 The military ranks and titles mentioned include: prior (pqÊyq), and ordinarios (ÔqdimÇqior), both being grades of non-commisioned officers,28 as well as soldier (stqatiÍtgr)29 and recruit (ñlbahlor).30 Notably, priores are also attested in Kastron Nessana.31 An honorific title devotus/devotissimus (jahysiylÈmor), often of military connotation and, for example, denoting some soldiers in Nessana,32 is also present in the Petra texts. Notably, priores from Kastron Zadakathon, Flavius Barakhos and Flavius Thomas, the latter married to Kyra, possibly a granddaughter of Theodoros, both carry the high status name of Flavius.33 The documents mentioning these persons well indicate close ties and social relations between the local population and the military stationed in the countryside.34

The Petra Papyri are the only Byzantine-period documents where both localities are specified with the addition of kastron to the original toponyms. This should not be surprising. By the early 6th century, the term kastron designated a fortified settlement of a smaller size, but later on, the terms polis and kastron could have been synonyms in the popular language.22 The appearance of kastron in the toponyms of Zodocatha and Ammatha may also indicate the fusion of the town and the castrum into a single unit, either by civilians gradually moving into the protection of the fort, or by the civilian settlement expanding to include the military one. Either situation is well known from Palestine and from other parts of the Byzantine Empire. For example, Nessana is referred to both as kome and kastron23 and the military (walled) compound could have been a part of the town.24 Kastron Mayfa'ah (Umm er-Rasas), a fortified settlement in

25

Lewin 2001: 293. E. g., Inscriptions 2 and 9a, Piccirillo 1994: 244-246, 252. Lehtinen 2001: 788; Koenen 2003: 214-215. 28 Grosse 1918: 126; Jones 1964: 662, 674; but ordinarii could also be civil officials. For the list of military titles from Nessana, see Rubin 1997: 66-7. 29 Also in P. Nessana, e.g., 15, AD 512, and 16, AD 512 (Kraemer 1958: 41-4; 45-51), etc. For the use of this term, see Müller 1912: 101107. 30 Also in P. Nessana 24, AD 569 (Kraemer 1958: 77-80) with the understanding of the term as applying to low-ranking soldiers or recruits. 31 P. Nessana 26, AD 570, and 35, 6th century, Kraemer 1958: 81-83, 108-110. 32 E.g., P. Nessana 15, AD 512, Kraemer 1958: 41-44. But this title also occurs in relation to elite imperial officer corps, palace guards and secret police (scholae, protectores, domestici, silentiarii, agentes in rebus) and various officials of the imperial chancellery (Koch 1903: 7881). 33 According to the reconstructed reading offered by T. Gagos, Flavius Thomas might have been a prior domesticorum (Koenen 2003: 215: notes 31-32), probably an honorary title rather than indicating that he belonged to the elite imperial guards. 34 Koenen 2003: 215. 26 27

14

Gagos 2001: 501. Fiema 1991: 300; Graf 1995: 250. 16 Alt 1921: 8-10. 17 Negev 1977, nos. 72 and 104. 18 Graf 1995: 250. 19 Alt 1921: 8-10. 20 Brünnow and Domaszewski 1905: 3-5; Parker 1986: 101-102. 21 Findlater 2002: 141. 22 See Haldon 1999: 11-12, 15-17; and Saradi-Mendelovici 1988: 365366, 369-370, for discussion 23 P. Nessana 16, Kraemer 1958: 46-51. 24 Isaac 1995: 147, note 94. 15

315

THE LATE ROMAN ARMY IN THE EAST praepositus.44 At any rate, the addition of Kastron Ammatha after his title would indicate the location of the former post of Dusarios, where he could previously have been the commander of a military unit.45

It needs to be stressed that regular army personnel is attested in Kastron Zadakathon, even towards the end of the 6th century, as prior Flavius Barakhos is mentioned in P. Petra inv 44a, dated to AD 593.35 But whether the unit in ɋadaqa was still the same equites promoti indigenae as in the 4th century or not, cannot be verified. Current opinion favors the recruitment of these type of units from among the local populations rather than from original legionary cavalry.36 The nature of the unit is less certain but was more likely of a limitanei-type, not unlike a numerus known from Nessana.37 The Nessana soldiers were local recruits who had served in the garrison of their hometown where they could own houses and landed property,38 and a similar situation must have existed in Kastron Zadakathon. This assertion is further supported by the evidence discussed below, which indicates the involvement of the military as arbitrators in local disputes

The local tax-system mentioned in the texts also included extra payments in kind. For example, four tax receipts P. Petra I 7, AD 568; 8, AD 570?/571; 9, AD 575/576 and 10, AD 578 - mention “...extra taxes in farm products, in oÓmËjqeom and in other products accordingly...” Oinokreon is generally understood as a formula of rationing of the actual foodstuffs which specifically included one unit of meat and two units of wine, rather than just meat marinated in wine,46 and is known from Egypt, seemingly as forming a part of annona militaris payments.47 Although annona was also disbursed to the Arab allies, in this case this may better pertain to annona mostly collected for the regular army units. Notably, annona militaris appears to have still survived in Byzantine Syria, Mesopotamia and Egypt at the time of the Muslim conquest, and was still a form of government disbursement rather than the produce of the soldiers-farmers left for their upkeep.48

Apparently, Ammatha was occupied in the 6th century as well, probably as a civilian-military kastron not unlike that at ɋadaqa, and the texts also indicate potential military presence there. One Flavius Dusarios, the greatuncle of Theodoros, is styled in one of the documents as Ð[p]Ä pqaivÈjtym JÇstqou ’Allah´m - ex-prefect of Kastron Ammatha.39 The use of ÐpÄ, which began in the 3rd century, indicated that the ex-holder of a particular office was automatically entitled to a particular rank or status with all its privileges and exemptions formally attached to that.40 Generally, the ÐpÄ ÑpÇqwym is a designation of a former prefect,41 as well as an honorific title specifically denoting the social class.42 The title of Fl. Dusarios indicates his current social rank or status in relation to his previous appointment. The use of pqaivÈjtym rather than ÑpÇqwym was perhaps meant to emphasize the former military rather than civilian appointment. By the 6th century, the original military prefect of the Early Empire continued, although on limited scale, to denote the commander of an auxiliary alae or cohortes as well as the numeri, legiones and vexillationes legionis,43 all being primarily the limitanei forces. Furthermore, praefectus, in the Late Roman meaning of the term, is often used in the context of a commander, who could also be titled tribunus or

There is no indication of Arab foederati, whether the Ghassanid or any other allied groups in the Petra Papyri. However, it is significant that Abu Karib ibn Jabala (in the text as Abu Kherebos), the Ghassanid phylarch of Palaestina Tertia appears in P. Petra inv. 83, most possibly dated to AD 544. The document is a settlement of a long-lasting dispute between Theodoros son of Obodianos and Stephanos son of Leontios, over some houses and properties in Zadacathon. Notably, among the arbitrators was Flavius Thomas, the prior of the kastron. As for Abu Karib, he is mentioned as an arbitrator who was asked to give his decision in this dispute some years prior to AD 544, apparently during his active service for the Byzantine state as a phylarch.49 As it was a purely private dispute, it seems that Abu Karib became involved in it not ex officio but rather as a person of high authority and respect, well known in the local community. Thus, his involvement as a mediator reflects perhaps less his military rank and official political position but rather indicates his competence in settling local, civil affairs and the respect accorded to the military by the local population. Undoubtedly, his understanding of the local situation, customs and traditions would have been a great asset in dealing with micro-scale conflicts and potential disturbances in the area.

35

Koenen 2003: 214. Lewin 2001, 2004, for full discussion. Isaac 1995: 145-147, and Rubin 1997: 64-66, argue against the identification of the Nessana unit as Theodosiani. The reference to that unit in P. Nessana 15 (Kraemer 1958: 41-44) is unclear. 38 Isaac 1995: 145-147. 39 Gagos and Frösén 1998: 475; Koenen 2003: 214. 40 Millar 1983: 90, 94, note 78, Roueché 1989: 23-24. 41 Also granted to consistoriani upon their retirement (Delmaire 1995: 14-15). 42 Guilland 1982: 30-31. These honorifics were also applied to toll collectors, notaries, supervisors of state workshops, or chartularioi, i.e., the class of low-level, provincial administrators (Kazhdan 1991: 13334). The difference between the high and low-level officials holding this title could be further discerned by a specific rank added, e.g., apo eparchon for a person being ÑmdonËtator; late 6th-early 7th century? from Aphrodisias (Roueché 1989, inscription no. 91). See the exdomesticus (ÐpÄ - dol[estijËm]) in P. Nessana 26, AD 570 (Kraemer 1958: 82-84). 43 Southern and Dixon 1996: 60-1; Jones 1964: 640; Haldon 1995: 6. 36 37

44

Grosse 1918: 152. Fl. Abinnaeus, praefectus alae at Dionysias, was often addressed as praepositus, or praefectus castrorum Dionisiados (Southern and Dixon (1996: 61). 46 Gascou 1989: 292; Fournet 2003: 401. 47 Syrcou 1996: 86-91; P. Berol. 21891, Aphrodito, 6th century; Koenen 2003: 215, note 36, 216; Bagnall 1990: 90-92. 48 Kaegi 1985: 592-595. 49 Kaimio 2001: 720-721, 724. 45

316

ZBIGNIEW T. FIEMA: THE BYZANTINE MILITARY IN THE PETRA PAPYRI - A SUMMARY in 629 resulted in a sound Byzantine victory near Mu'ta. Arab sources imply that tribal segments of the Bali, Judham, Lakhm, Ghassan and al-Qayn fought at Mu'ta on the Byzantine side.51 However, during the major invasion of AD 634, southern Palestine was deprived of effective, formal defence. Instead, the population would have followed the pattern of self-protection, probably already extant in the later 6th century, when the local communities had increasingly to depend upon private arrangements with local phylarchic groups or friendly nomads. Even before the main invasion, three larger towns in the area - Aila, Udhrudz and Jarba - peacefully capitulated to the Muslim forces in AD 630,52 and the conquest of southern Jordan passed without any specific references in either Byzantine or Arab sources to any military engagements.

Conclusions Although the pattern of the military presence in the Petra area in the 6th century is more ambiguous than in the 4th century, yet it indicates its continuity there. The possible selective abandonment or withdrawal of troops was probably related to a changed emphasis in the political and economic vitality of the area for the Empire. However, the major towns in the area were still accorded protection by the military, which apparently included the regular army units, and annona militaris was still duly collected. The soldiers in Kastron Zadakathon appear to have been town dwellers, fully integrated with the civilian population and actively participating in the social and economic affairs of the town. As such, their involvement in purely civilian affairs is not surprising.

Acknowledgements

Furthermore, it is then reasonable to assume that local security concerns would still have been shared by the local limitanei and foederati, probably even after the official dissolution of the Ghassanid phylarchy in AD 581, the fact the significance of which has been much exaggerated by modern scholars. The Byzantine phylarchical system was transformed rather than eliminated, returning to the earlier form of independent alliances with individual local groups; e.g., the foedus with Salih was revived.50 Similar, independent arrangements probably prevailed with the tribes of the northern Hejaz and southern Transjordan, such as the 'Amila, Judham and Bali. These tribes remained actively on the Byzantine side even after the interlude of the Persian invasions.

The research on the subject of this article has been carried out within the framework of the Research Center ‘Ancient and Medieval Greek Documents, Archives and Libraries’ at the University of Helsinki, under the direction of Prof. Jaakko Frösén, which is part of the ‘Centres of Excellence in Research program of the Academy of Finland. All errors of omission and interpretation are mine. Bibliography Alt A. 1921 Die Griechischen Inschriften der Palaestina Tertia westlich der Araba. Wissenschaftliche Veröffentlichungen des DeutschTurkischen Denkmalschutz-Kommandos. T. Wiegand, ed., Berlin. Bagnall R.S. 1990 Five Problematic Fourth-Century Pieces. In Bulletin of the American Society of Papyrologists 17: 79-93. al-Baladhuri, Abu-l’Abbas Ahmad ibn-Jabir 1966 The Origins of the Islamic State. Translation of Kitab futuh al-Buldan by Hitti P., Beirut. Brünnow R. - von Domaszewski A. 1904-9 Die Provincia Arabia I-III, Strassburg. Delmaire R. 1995 Les institutions du Bas-Empire romain, de Constantin à Justinien. I. Les institutions civiles palatines, Paris. Donner F.M. 1981 The Early Islamic Conquests, Princeton. Fiema Z.T. 1991 Economics, Administration and Demography of Late Roman and Byzantine Southern Transjordan, Ph.D. Dissertation, University of Utah. Salt Lake City. Fiema Z.T. 1995 Military Architecture and the Defence “System” of Roman-Byzantine Southern Jordan - A Critical Appraisal of Current Interpretations. In

Nevertheless, such arrangements in the early 7th century would never again resemble the scale of the 6th century alliances. After all, the successors of Justinian could not envisage the scale of warfare which would be conducted by the Persians in the early 7th century, as previous conflicts had been restricted to the regions of Armenia or northern Syria. The shock of the military defeat and subsequent occupation must have seriously affected provinces like Palestine, which had not experienced any large-scale, non-Roman, invasion from the East since the Palmyrene episode in the later 3rd century. It is not surprising then that the Persian invasion and occupation (AD 614-629) effectively disrupted and eliminated any military arrangements which previously existed in Palestine in the 6th century. A brief respite (AD 629633/634), following the Byzantine victory and the Persian withdrawal, apparently did not suffice to fully reconstruct the military strength and re-garrison outlying parts of the empire, such as the Petra area. Some sort of a regional defense, most probably of local foederati, was apparently reconstituted in southern Moab just north of the Wadi el- Ⱦasa since the first Muslim raid

51 50

52

Shahîd 1989: 303, 476.

317

Donner 1981: 103-107; 115-116. Ibn Ishaq in Ibn Hisham 1955: 525-526; al-Baladhuri 1966: 92-95.

THE LATE ROMAN ARMY IN THE EAST Studies in the History and Archaeology of Jordan V, Amman: 261-269. Fiema Z.T. 2002a The Military Presence in the Countryside of Petra in the 6th Century. In Freeman Ph. –Bennett J. – Fiema Z. T. and Hoffmann B. (ed.), Limes XVIII. Proceedings of the XVIIIth International Congress of Roman Frontier Studies held in Amman, Jordan (September 2000). (BAR International Series 1084, Vol. I and II), Oxford: 131-136. Fiema Z.T. 2002b Petra and Its Hinterland during the Byzantine Period: New Research and Interpretations. In Humphrey J. (ed.), Roman and Byzantine Near East: Some New Discoveries III. (JRA Supplementary Series no. 49), Portsmouth, RI: 191-252 Findlater G. 2002 Limes Arabicus, via militaris and Resource Control in southern Jordan. In Freeman Ph. – Bennett J. – Fiema Z. T. and Hoffmann B. (ed.), Limes XVIII. Proceedings of the XVIIIth International Congress of Roman Frontier Studies held in Amman, Jordan (September 2000). (BAR International Series 1084, Vol. I and II), Oxford: 137-152. Fournet J.-L. 2003 Review of The Petra Papyri I. Frösén J. - Arjava A. - Lehtinen M. (ed.), Amman, 2002. In AntTard 11: 398-404. Frösén J. 2004 Archaeological Information from the Petra Papyri. In al-Kraysheh F. (ed.), Studies in the History and Archaeology of Jordan V, Amman: 141144. Gagos, T. 2001 Negotiating Money and Space in Sixth Century Petra. In Andorlini I. - Bastianini G. Manfredi M. - Menci G. (ed.), Atti del XXII Congresso Internazionale di Papirologia, Firenze, 23-29 agosto 1998, Firenze: 495-509. Gagos T. - Frösén J. 1998 Petra Papyri. In Annual of the Department of Antiquities of Jordan 42: 473-481. Gascou J. 1989 La table budgétaire d’Antaeopolis (P. Freer 08.45 c-d). In Lefort J. - Morrisson C. (ed.), Hommes et richesses dans l’Empire byzantin. Vol 1: IVe-VIIe siècle, Paris: 279-313. Graf D.F. 1992 Nabataean Settlements and Roman Occupation in Arabia Petraea. In Studies in the History and Archaeology of Jordan IV, Amman: 253260. Graf D.F. 1995 The Via Nova Traiana in Arabia Petraea. In Humphrey J. (ed.), The Roman and Byzantine Near East: Some Recent Archaeological Research. (JRA Supplementary Series, no. 14). Portsmouth, RI: 241-65. Graf D.F. 1997a The Via Militaris in Arabia. In Dumbarton Oaks Papers, 51: 271-282. Graf D.F. 1997b The Via Militaris and the Limes Arabicus. In Groenman-van Waateringe W. - van Beek B. L. - Willems W. J. H. - Wynia S. L. (ed.), Roman Frontier Studies 1995, Oxford: 123-133. Grosse R. 1918 Die Rangordnung der römischen Armee des 4.-6. Jahrhunderts. In Klio 15: 122-161. Guilland R. 1982 Etudes sur l’histoire administrative de L’Empire Byzantin. III L’apoeparque - ÐpÄ ñpaqwym. In Byzantinoslavica 43: 30-44.

Haldon J.F. 1995 Administrative Continuities and Structural Transformations in East Roman Military Organisation c. 580-640. In Haldon J., State, Army and Society in Byzantium. London [reprint from Vallett F. - Kazanski M. ed., L’armée romaine et les barbares du 4e au 7e siècle. Paris 1993]: 1-20. Ibn Hisham 1955 Sirat Rasul Allah. Mustafa al-Saqa et al. (ed.), Cairo. Isaac B. 1995 The Army in the Late Roman East: the Persian Wars and the Defence of the Byzantine Provinces. In Cameron Averil (ed.), The Roman and Early Islamic Near East III, Princeton: 125-155. John Malalas 1986 The Chronicle of John Malalas. Jeffreys E. - Jeffreys M. - Scott R., transls., Melbourne: Australian Association for Byzantine Studies. Jones A.H.M. 1964 The Later Roman Empire, 284-602. A Social, Economic and Administrative Survey, Norman. Kaegi W.E. 1985 The Annona Militaris in the Early Seventh Century. In Byzantina 13/1: 591-596. Kaimio M. 2001 P. Petra inv. 83: A Settlement of Dispute. In Andorlini I. - Bastianini G. - Manfredi M. - Menci G. (ed.), Atti del XXII Congresso Internazionale di Papirologia, Firenze, 23-29 agosto 1998, Firenze: 719-724. Kazhdan A. 1991 Apo Eparchon. In Oxford Dictionary of Byzantium: 133-134. Koch P. 1903 Die Byzantinischen Beamtentitel von 400 bis 700, Jena. Koenen L. 2003 The Decipherment and Edition of the Petra Papyri: Preliminary Observations. In Schiffman Lawrence H. (ed.), Semitic Papyrology in Context. A Climate of Creativity. Papers from a New York University Conference marking the Retirement of Baruch A. Levine, Leiden-Boston: 201-226. Lehtinen M. 2001 Preliminary Remarks on the Prosopography of the Petra Papyri. In Andorlini I. Bastianini G. - Manfredi M. - Menci G. (ed.), Atti del XXII Congresso Internazionale di Papirologia, Firenze, 23-29 agosto 1998, Firenze: 787-794. Lewin A. 2001 Kastron Mefaa, the Equites Promoti Indigenae and the Creation of a Late Roman Frontier. In Liber Annuus 51: 293-304. Lewin A. 2004 Limitanei and comitatenses in the Near East from Diocletian to Valens. In Le Bohec Y. Wolff C. (ed.), L’Armée romaine de Diocletian a Valentinien Ier. Actes du Congrés de Lyon (12-14 septembre 2002), Paris: 227-236 Mayerson Ph. 1984 “Palaestina” vs. “Arabia” in the Byzantine Sources. In ZPE 56: 223-230. Mayerson Ph. 1987 Libanius and the Administration of Palestine. In ZPE 69: 251-260. Millar F. 1983 Empire and City, Augustus to Julian: Obligations, Excuses and Status. In JRS 73: 76-96. Müller A. 1912 Das Heer Justinians (nach Procop und Agathias). In Philologus 71: 101-138. Negev A. 1977 The Inscriptions of Wadi Haggag, Sinai (Qedem 6), Jerusalem.

318

ZBIGNIEW T. FIEMA: THE BYZANTINE MILITARY IN THE PETRA PAPYRI - A SUMMARY Notitia Dignitatum 1876 Notitia Dignitatum. Seeck O. (ed.), Berlin. P. Petra I 2002 The Petra Papyri I. Frösén J. –Arjava A. – Lehtinen M. (ed.), ACOR Research Publications No. 4, Amman. P. Nessana 1958 Excavations at Nessana 3. Nonliterary Papyri. Kraemer C.J.Jr. (ed.), Princeton. Parker S. Thomas 1986 Romans and Saracens: A History of the Arabian Frontier, Philadelphia. Piccirillo M. 1994 Le iscrizioni di Kastron Mefaa. In Piccirillo M. - Alliata E., Umm al-Rasas Mayfa’ah I. Gli scavi del complesso di Santo Stefano, Jerusalem: 241-270. Procopius of Caesarea 1971 The Persian War. In Procopius, with an English Translation. Vol. I, History of the Wars, Books I and II. Dewing M.B. (transl.), Cambridge. Ptolemy 1845 Claudii Ptolemaei Geographia. Tomus II. Nobbe C.F.A. (ed.), Leipzig. Roueché Ch. 1989 Aphrodisias in Late Antiquity, London. Rubin R. 1997 Priests, Soldiers, and Administrators: Society and Institutions in the Byzantine Negev. In Mediterranean Historical Review 12: 56-74. Saradi-Mendelovici H. 1988 The Demise of the Ancient City and the Emergence of the Medieval City in the Eastern Roman Empire. In Echos du Monde Classique/Classical Views 32, n.s. 7: 365-401. Sartre M. 1993 Pétra et la Nabatène méridionale du Wadi al-Hasa au golfe de 'Aqaba. Inscriptions de la Jordanie. Tome IV. (Inscriptions Grecques et Latines de la Syrie. Tome XXI.), Paris. Shahîd I. 1989 Byzantium and the Arabs in the Fifth Century, Washington, D.C. Sipilä J. 2004 Roman Arabia and the Provincial Reorganizations of the Fourth Century. In Mediterraneo Antico 7/1: 317-348. Southern P. - Dixon K. 1996 The Late Roman Army, London. Syrcou A. 1996 Six Byzantine Documents. In AFP 42: 84-99. Tabula Peutingeriana 1976 Tabula Peutingeriana. Codex Vindobonensis 324. Weber E. (ed.), Graz. Tsafrir Y. 1986 The Transfer of the Negev, Sinai and Southern Transjordan from Arabia to Palaestina. In Israel Exploration Journal 36/1-2: 77-86. Walmsley A. - Barnes H. 2002 A Brief Ground Survey of Ruwath (Robatha), South Jordan, 28 October - 16 November 2000. In Annual of the Department of Antiquities of Jordan 42: 485-492.

319

The Operational Methods of the Late Roman Army in the Persian War of 572-591 Frank R. Trombley Cardiff University (Wales)

The Strategikon of Maurice, published in the first quarter of the seventh century, gives the clear picture of a welldefined system of tactical training and method that was being implemented in the last decade of the sixth century, in the wars against the Avars and Slavs in Illyricum and Thrace.1 The wars of the later sixth century provided the proving ground for the fully developed system reflected in this tactical manual, as comparison with the military histories of that century reveals. Harder to grasp are the operational methods of the late sixth-century armies, which are often only implicitly hinted at in the narratives of Procopius, Agathias and Theophylact Simocatta.2 There are also questions about the overall political aims of the regime in Constantinople, that is, the economic advantages and power configurations it was seeking to achieve or protect.3

upon this at a conference near Dara with the Roman negotiator Zachariah in 581-582:6 [The] Roman Empire, which is facing a large number of enemies, is fighting in many parts of the world and has its forces divided in hostilities with almost every barbarian nation… You are at war with many peoples, we with you alone; thus the necessity for the treaty… [So] we shall surely conquer since we have a dispute with no one other than the Romans and since we are committed to only one war. I hope to show that, by this time, the Late Roman army was practising a type of operational method that foresaw and applied particular solutions to particular problems of managing armies and handling enemy forces, and that this had a conceptual framework that found its culmination in the Strategikon of Maurice. There were therefore criteria against which the success or failure of a particular operation could be measured by the time the Strategikon was published.

In what follows, I intend to look at the Oecumenical History of Theophylact Simocatta and its treatment of the Persian War of 572-591, as it gives a broadly coherent account based on official and semi-official versions.4 As such this is an artificial construction, since ongoing operations against the Avars in Thrace and Illyricum exercised a powerful influence on the situation in Oriens and Armenia, particularly in terms of manpower deficits and the commitment of revenues.5 The regime in Constantinople was well aware of its operational overreach, and the Sasanid ambassador Andigan played

Operations What was an ‘operation’ in the context of the Persian War of 572-591? In general, it involved the movement or positioning of a large force to further the overall aims of the war: weakening the Sasanids’ military strength by the destruction of their armies, capture of defensive installations, disruption of their civil population and the tax revenue it contributed through taking captives or outright killing, and the destruction of agricultural capital in the territories of the Sasanid cities.7 The long-term aim was to break the will of the Sasanid monarch, or that of his nobles, to wage war.8 The Roman armies came tantalisingly close to these aims at different times, particularly in 578 not long before Khusrau I’s death, but disagreements about the final settlement intervened to prevent definitive peace negotiations taking place before he died. The Syrian church historian John of Ephesus has

1

Maurikios, StratƝgikon. On broader questions, see Grosse 1920. Apart from general histories, there have been few studies specifically addressing military personnel, operations and intelligence. Among them are Lee 1993, 109-128, Maspero 1912a, Maspero 1912b and Cameron 1995. Cf. Intelligence, military, in Margiotta 1996: 506-520, and Trombley 2005 (provisional analysis). 2 This can also be said of the narratives of Menander Protector and John of Epiphaneia, whose histories survive only as fragments. On the importance of John’s history as a source for Theophylact Simocatta, see Olajos 1988: 14-66, 96-106. 3 See broadly Stein 1919: 1-25, as well as Goubert 1951 and Christensen 1944. The regime in Constantinople had an interest in canalising the silk trade to itself own hands. Tate 1999. This led to diplomatic manoeuvres with the Kindite Arabs and Axumite empire, causing tensions with the Sasanid monarchy, but the issues that brought about the Persian War of 572-591 were, among other things, Justin II’s support for the Armenian insurrection of 571/2 and political problems in Constantinople caused by the annual payments to the Persian king. Turtledove 1983. 4 Theophylact Simocatta, Historiae: 1887 and 1986. Whitby 1988 (essential). See the source collection in Greatrex and Lieu 2002: 141175. 5 On the Avars in 6th c. Thrace and Illyricum, see Pohl 1988: 58-93. On their impact on Illyricum and Thrace, see Velkov 1977: 52-59, etc. Whitby 1988: 140-156. Curta 2001: 90-107.

6 Menander Protector, Fr. 26 (Blockley 1985: 230f. and note 303). Blockley puts these negotiations in the winter of 580-81. The date is immaterial to the thrust of the present argument. 7 For the implicit dependence of Sasanid revenues on agricultural production, see Rubin 1995: 292. The rationale for this kind of warfare is succinctly set forth, albeit for a different period, in Smail 1956: 25. On basic concepts of grand strategy, see Liddell Hart 1967: 333-372. 8 For example, in 586 the Persians attempted to end the war through the agency of the archbishop of Nisibis. The Romans had however implemented new tactical systems and operational procedures, and were perhaps loath to conclude hostilities without testing their newly acquired capabilities, culminating in the battle of the Solachon plain.

321

THE LATE ROMAN ARMY IN THE EAST it that Khusrau I exclaimed at the time of these negotiations: ‘We learn that the Caesar of the Romans is a young and warlike man; and I, as I see it, have grown old, and can no longer bear the fatigue of wars. Let us therefore make peace with the Romans, for we cannot overcome them’.9 In that year Maurice, newly appointed as magister militum of Oriens, took large numbers of Christian captives who were eventually transported to Cyprus.10

large forces at his disposal. If the deployments listed in the Notitia Dignitatum are still a relevant guide — there being only two sixth-century references to known formations to corroborate this — the magister militum would have disposed of perhaps ten cavalry arithmoi directly under his command.14 He could in theory have drawn upon the combined forces of the duces of eight provinces in Oriens: Phoenice Libanensis with ca. twelve arithmoi, Syria I-II and Euphratensis with ca. ten, Palaestina I-III with ca. twelve, Osrhoene with ca. nine, Mesopotamia with ca. ten, and Arabia with ca. eight, all of them cavalry, coming to a total of sixty-one arithmoi and giving a figure of ca. 20,000 men.15 This assumes that the manpower of the individual cavalry squadrons averaged well under 500, as low as the figure of 310 men given in the Strategikon of Maurice.16 It is doubtful that any of these units would have been at full strength,17 nor is it known which provincial cavalry forces were drawn upon and incorporated into the field army under Marcian’s command in 573, or if the practice existed of leaving fractions of the provincial cavalry moira behind to cover the eastern approaches of the individual provinces. A concentrated Roman field army was a formidable and intimidating force. An undated fragment of Menander Protector emphasises this point, although the size of the Roman formation in question is not completely certain:18

The term ‘operation’ is potentially synonymous with ‘campaign’. The annual spring to late autumn season might see more than one Roman operation, or a combination of competing Sasanid and Roman operations in different sectors of the long frontier, which ran from Lazica and the Armenian mountains round the eastern spurs of Tnjr ‘AbdƯn and through Beth ‘Arabaye to the Syrian desert. Operations also involved the organisation of supplies in the zone of operations, defensively covering the Roman cities and their civilian population, and the provision of boats to supply river operations and to construct bridges. Roman operations might broadly be characterised as offensive and defensive. Since the regime in Constantinople under emperor Justin II provoked the war, the burden of taking the offensive fell to the empire. The principal zone of operations was upper Mesopotamia, in the Sasanid provinces of Arzanene and Arbayistan. The most important objective was Nisibis, a populous city with an extensive territory.11 Two attempts were made against Nisibis in 573, both of which failed, but the reasons for this are far from clear. The magister militum per Orientem normally concentrated the organic troops of his command, and the moirai or droungoi of the provinces of his circumscription in the vicinity Dara, which may have been his headquarters between 540573.12

When the enemy’s army had gathered, it was afraid of the size of the Roman force (YZWGZO­). It therefore waited for the Romans to disperse and offer an 14

Ten cavalry squadrons with the designation equites or cuneus equitum are mentioned at Notitia Dignitatum Or. VII. Insignia viri illustris magistri militum per Orientem, 23-34. 15 In arriving at these figures, I have included only the cavalry squadrons designated by the term equites or cuneus equitum in the Notitia Dignitatum. This assumes that cavalry formations designated by the term ala had disappeared from the order of battle in Oriens by the late 6th c. There is a strong likelihood that formations of limitanei designated by the term legio, like the IV Parthica at Beroea-Aleppo ca. 586, still existed in some form. Although firm evidence is lacking, I assume that infantry units designated by the term cohors had all but ceased to exist by this time, on the analogy of the cavalry alae. New evidence could correct this supposition. The relevant sections of the Notitia Dignitatum are: Or. XXXII. Dux Foenicis. 17-29, Or. XXXIII. Dux Syriae 15-27, Or. XXXIV Dux Palaestinae 18-29, Or. XXXV Dux Osrhoenae 15-23, Or. XXXVI Dux Mesopotamiae 19-28, and Or. XXXVII Dux Arabiae 11-23. On the size of these formations, see note 16 below. It is unknown how many of these formations were transferred to the magister militum per Armeniam in ca. 528. See below, note 42. 16 This figure is the estimate of Grosse 1920: 274. See Maurice, Strategikon III 2 for the figure of 310. The Strategikon indicates that the arithmoi were deliberately made asymmetrical in size to prevent the enemy from counting the heads of a single Roman squadron and thereby calculating the size of the army by multiplying this figure by the number of arithmoi visible on the battlefield. Ibid. On the other hand, if the 6,000 cavalry of Phoenice Libanensis who rode to the relief of Antioch consisted of twelve arithmoi, as indicated in the Notitia Dignitatum in the early 5th c., the average size of each formation would have been 500 men, conforming exactly to the theoretical establishment of the arithmos. Procopius, De Bellis II 8. 1. Cf. Hoffmann 1969-70. See also A. Kazhdan, Notitia Dignitatum, in Kazhdan 1991: 1496. 17 Cavalry squadrons may often have been under strength. See Grosse 1920: 274, for a cautious assessment. 18 Menander Protector, Fr. 10 (Blockley 1985: 214f.).

The first large-scale operation of the war was the attempted capture of Nisibis by siege in 573. Marcian magister militum per Orientem conducted the two advances against Nisibis in that year.13 He had potentially 9 John of Ephesus, HE VI 21 (Payne Smith 1860: 421f.). There is a series of translations from Book VI of the Third Part of John of Ephesus’ ecclesiastical history by Jan van Ginkel in Lieu and Greatex 2002: 141, 144-147, 154f., 156f., 164f. The quotations from Payne Smith’s translation have been checked against van Ginkel’s and compared to the Syriac text where necessary. See Chosroes I Anoushirvan, PLRE IIIA: 303-306. 10 See the observations of Oleson 1976: 161-164. Oleson’s remarks are well taken, but it should be noted that the seals he discusses could also have ended up in Cyprus as a consequence of refugee traffic during the Sasanid conquest of the littoral of Syria, Palestine and Egypt between 613-19. 11 M. Mango, Nisibis, in Kazhdan 1991: 1488. 12 M. Mango, Dara, in Kazhdan 1991: 588. 13 Stein, Studien, 43-46. Marcianus 7, PLRE IIIB: 821-23. One of Marcian’s predecessors was a certain Marthanius, who was magister militum per Orientem in 559/60. His administrative achievements, including the refurbishment of fortifications, are known only from the epigraphy and are difficult to gauge. Dagron and Feissel 1987: 141-146 (no. 89).

322

FRANK R. TROMBLEY: THE OPERATIONAL METHODS OF THE LATE ROMAN ARMY IN THE PERSIAN WAR OF 572-591 opportunity to attack them…, since some of the Persians thought that, while [the Romans] would be difficult opponents united, they would be easier to attack when scattered over enemy territory.

militum per Orientem ca. 573. It gives a list of the camels assigned to eight decarchies (troop bodies of ten men) and the men responsible for them.23 It is not exactly clear whether the men named in each decarchy represent its full complement, or only those who had been assigned camels. The end of the document is missing, so it is difficult to say even how many decarchies were in Nessana at this time. If all the ordinary soldiers of the decarchies had camels, the average strength was 4.5 men per decarchy, excluding decarchs, suggesting that this formation was operating at an establishment well under fifty percent. In view of the fact that some decarchs’ names are given without camels, however, it would seem that some of the officers rode horses instead. The editor assumes that this force consisted of supply troops but, if this were the case, it is difficult to understand why two decarchies (or possibly individual men) are designated as going to Egypt, and others possibly to Caesarea, the provincial capital of Palaestina I, as the editor supposes, because the provincial supply dumps in Palaestina III would have been located much closer to hand. Many of the men, including some of the commanders of the individual decarchies, have Arabic personal names and patronymics. They may therefore have been largely recruited in Palaestina III.

The Roman army was invariably successful in engagements with the Persian army when it succeeded in concentrating most of the provincial moirai, as occurred at the battle of the Solachon plain in 586. In the Persian War of 502-507, all these ducal formations, in addition to those of Armenia I and II, were eventually brought into the conflict. When all the cavalry arithmoi of a particular province joined the field army, they were combined into a single moira or, in the language of the seventh century, a droungos. This is suggested in an important report of John of Epiphaneia about operations in 572, the first year of the war:19 After Marcian crossed the Euphrates and reached Osrhoene, he sent 3,000 armoured soldiers (ÛpkÊtai) against Arzanene, in what is called a moira, setting as their commanders Theodore and Sergius who drew their birth from Rabdion, and Iuventius, commander of the squadrons (tagmata) in Chalcis.

It is difficult to generalise from this document. It comes from a fortress far from the zone of operations in Mesopotamia and Osrhoene. It may well be that cavalry arithmoi throughout Oriens were under strength when operations began in 573. On the other hand, the magister militum per Orientem may have received the pick of the new recruits in the winter of 572/3, before the abortive expedition against Nisibis. It is possible but unknown if the troops of the dux Palaestinae were committed to action in Mesopotamia in 573 or at any time thereafter.24 Flavius Paulus dux Palaestinae is mentioned in a building inscription of 580 at Ziza (present-day Zizia, near Madaba in Jordan), but the actual work was delegated to a locally based officer (ðqwym) and a clarissimus.25 The whereabouts of the dux Paulus at this time are therefore unknown, but it is likely that the cavalry squadrons of Palaestina had returned to winter quarters in the towns and forts that normally accommodated them, for the inscription dates from the autumn (9 November in the fourteenth indiction); the dux may have begun refurbishing the fortifications at kastra like Ziza once the troops were in their billets.26

From the early sixth century onward, the moira of a province was normally designated by the city in which its headquarters lay.20 In this instance, our source it refers to the forces at the disposal of Iuventius, dux of Syria and Euphratesia, whose headquarters were at Chalcis.21 The 3,000 men sent into Arzanene were the strength of some ten cavalry arithmoi at the strength of ca. 300 men each, plus any men the other commanders — attached to the moira because of their local knowledge — brought with them. This is roughly consistent with the figure of twelve cavalry squadrons of comitatenses given in the Notitia Dignitatum. The dux Iuventius’ forces undoubtedly returned to their billets in the winter of 572/3 and rejoined the field army for the expedition against Nisibis in the spring 573. There is very little information about the individual provincial cavalry squadrons (arithmoi) of Oriens— which were rated as stratiǀtai or comitatenses—or about their commanders.22 An exception is found in a papyrus from Nessana listing military camels ([JQR VL @]jal[Èkom](sic)) and dating perhaps ca. 560580 — and therefore from first phase of the Persian War of 572-591. This document sheds an indirect light of the strength of the arithmoi at the disposal of the magister

23

Kraemer 1958 (P.Nessana III): no. 37. This was done in the Persian War of 502-7, but only later in the conflict, after the troops of the northern provinces of Oriens had suffered years of attrition. Ps.-Joshua, Chronicle §§ 72, 92 (Trombley and Watt 2000: notes 427, 516). 25 Gatier 1986 (IGLS XXI ): no. 155. 26 It is not known if a formation descended from the Equites Dalmatici Illyriciani mentioned in the Notitia Dignitatum was still based in the fort at Ziza in the late 6th c. Notitia Dignitatum Dux Arabiae Or. XXXVII 16. On the dux as supervisor of constructing fortifications in his province, Trombley 2004: 83-87.

19

24

John of Epiphaneia, Fr. I 3 (Dindorf 1870: 378, lines 21-28). On the importance of John’s history as a source for Theophylact Simocatta, see Olajos 1988: 14-66. 20 Ps.-Joshua the Stylite, Chronicle § 51 (Trombley and Watt 2000: 56 and notes, etc.). 21 Cf. Iuventius, PLRE IIIA: 760, where the doubts about Iuventius’ command as dux Euphratensis et Syriae seem unjustified. For the site, see Mouterde and Poidebard 1945: 7-9, Plan I, Plates II-IV. Cf. Whitcomb 1999. 22 Grosse 1920: 276-280.

323

THE LATE ROMAN ARMY IN THE EAST The Nessana papyri of ca. 560-580 may reflect a policy of starving distant provincial commands of the better recruits and forcing them to recruit local camel riders, who as Arabs shouted their patronymics and did not bother to take the fashionable gentilic praenomen ‘Flavius’, like ‘the most loyal Flavius […ca. l6 spaces] soldier and native of kastron Nessana’ mentioned in a papyrus dated 23 December 590, the final year of the war.27 On the other hand, it may not have seemed necessary to record onomastic details of this type in internal army lists and inventories. In either case, the low troop establishments of the decarchies at Nessana reveal very little about the situation further north in the zone of operations.

have left the provinces in the interior open to staggering losses in captives and destroyed agricultural capital. Writing in the 580s, John of Ephesus sums up the point of view of the provincials in apocalyptic terms:31 It has seemed to us not improper or alien to our purpose to attach to these ecclesiastical narratives a short account also of the wars and battles, desolation and bloodshed which have happened in our days, for the information of those who come after us, should the world contine to exist for so long… On his march [to Apamea, the Persian marzban] Adarmahan stormed numerous castles, which fell in his way, and he razed and burned them together with several strong and well-fortified towns, and at length arrived at Apamea… Having thus completed their work of destruction, they took the captives and loot of Apamea and the other towns, and returned to the king [at Dara]. The captives were counted in the king’s presence, and their number [was] 292,000. They were divided among the troops and taken to Persian territory.

It is possible that the provincial cavalry squadrons of Arabia, another out of the way province, were eventually committed to operations Mesopotamia or were involved in local fighting, for an inscription at al-Kufr mentions a building project of some kind that the protector Maximus supervised in 583, adding that ‘he fought in the war(s)’ (ÑpokÈl[g]sem) for thirteen years.28 This suggests that Maximus may have been involved in low intensity fighting for two or three years before the Persian war began. As a protector he would normally have assisted the magister militum in administrative tasks and what might be called ‘staff work’.29 It is impossible to say, however, whether Maximus spent the entire thirteen years in Oriens or initially performed his duties elsewhere, as for example in Thrace and Illyricum, being assigned to one of the magistri militum of Oriens later, sometime between 572-583.

Michael of Antioch mentions that the territory of Antioch and various towns of the Chalcidice and Euphratesia were ‘plundered and laid waste’.32 The number of captives taken after the fall of Dara is said to have been another 98,000.33 Whether accurate or inflated, the latter figure would have included soldiers, townspeople and refugees from the villages in the city’s territory. The events of 573 marked the last attempt to take Nisibis by siege. Little is known about the Sasanid defensive arrangements that made this operation so difficult, apart from the fact that in 573, and on later occasions when Roman armies reached the environs of Nisibis, relieving armies were promptly dispatched. This happened during the first siege in 573. The Sasanid relief force was defeated at Sargathon, a fortress some eight Roman miles west of Nisibis, but this may have convinced the Roman commander Marcian not to become too closely involved in a direct assault on the city.34 Instead he attempted to take a fortress thirty miles southeast of Nisibis called Thebothon.35 The siege failed, but it is indicative of an altered operational framework. The Romans were now concentrating on trying to isolate Nisibis from its further territory and from contact with the rest of Sasanid Arbayistan. This approach was particularly evident later on, when Maurice commanded the army of Oriens in the last years of Tiberius Constantine’s reign.

To return to the operational narrative, the presence of the provincial moira or droungos of the dux Syriae et Euphratensis in 572, and quite possibly that of the dux Phoenicae Libanensis, in the vicinity of Dara and Nisibis, and away from their respective provincial billets, may well explain the ease with which the Sasanid general Adarmahan rode to the environs of Antioch and from there to Apamea in November 573, destroying installations in the countryside, rounding up captives and seizing the latter city by ruse. John of Ephesus claims that 292,000 captives were taken, a figure that may be somewhat exaggerated, but his assertion suggests a serious loss of population in the territories of Antioch and particularly Apamea.30 From the standpoint of operations, the inference to be drawn from this catastrophe is that the magister militum per Orientem had to exercise good judgement in determining which provincial moirai to incorporate into the army of operations, and which to leave in their local billets facing the enemy. Errors in deployment, like the one seemingly made in 573, could 27

31 John of Ephesus, HE VI 1 and VI 6 (adapted from Payne Smith 1860: 366, 385-87). 32 Greatrex and Lieu 2002: 147. 33 Syriac chronicle of 1234, quoted in Greatrex and Lieu 2002: 148 (M. Greatrex). 34 Simocatta, Hist. III 10. 4. 35 Simocatta, Hist. III 10. 5 (Whitby-Whitby 1986: 87, note 44).

Kraemer 1958 (P.Nessana III): no. 29. Littmann, Magie and Stuart 1921 (PAES IIIA): no. 672. 29 This argument is provisionally set forth in Trombley 1999: 17f., 2327, and Trombley 2005: 396-398. 30 I have dealt with this from the standpoint of the epigraphy in Trombley 1997: 174-178. On Apamea, J.-P. Rey-Coquais, Apamea, in Stillwell 1976: 66f. 28

324

FRANK R. TROMBLEY: THE OPERATIONAL METHODS OF THE LATE ROMAN ARMY IN THE PERSIAN WAR OF 572-591 507.42 The praesental forces that participated in later wars were deployed in small packets, like the formation commanded by Theodoric, who may have been comes foederatorum (ca. 577/8-581), which raided Media on the bank of the Tigris in 581.43 An influential provincial like Evagrius of Epiphaneia could say that guards formations from Constantinople were simply not up to the job of operating in Mesopotamia. Of the campaigns of 581 and 582, where Evagrius’ chronology is conflated, the ecclesiastical historian observes:44

A concerted attempt to seize Nisibis never materialised again because of the loss of Dara in late 573.36 From that date, until Khusrau II handed it back to the Roman authorities in 591, the Persian garrison in Dara would have threatened the rear of any Roman attempt to renew siege operations against Nisibis. In consequence, the Roman armies confined themselves to entering Sasanid Arbayistan further south, from the direction of Circesium on the Euphrates, only for the purpose of raiding the territories of Nisibis and the towns of Beth ‘Aramaye. The Sasanids had shown themselves capable of undertaking more than one operation in a single season: Adarmahan’s raid against northern Syria coincided with Khusrau I’s personal intervention with an army to relieve the second siege of Nisibis in 573—which Justin II had demanded despite Marcian’s objections.37 The army’s second retreat to Dara in a single season left it depressed and possibly incapable of further offensive action. In the disorganised retreat Marcian’s standard had been lost and one infantry formation had been left behind and nearly wiped out by the pursuing Persians.38 Moreover, the duces of Syria and Euphratesia and Phoenice Libanensis may have demanded the release of their cavalry troops to cover their respective provinces, as happened in 541, fearing imperial censure for leaving the provincials in the lurch.39 If this happened, Marcian would have had only the troops of Osrhoene and Mesopotamia, along with the organic formations of his command.40 This would have left him with something in the range of twenty-nine under-strength arithmoi, or perhaps some 9,000 cavalrymen with which to contest the siege of Dara — a large force, but too small to confront a besieging army consisting of many tens of thousands of men. Khusrau I is said (whatever we may choose to make of these figures) to have had 23,000 cavalry, 40,000 infantry, and 120,000 farmers to help him with the excavations and demolitions that the siege of Dara required.41 Marcian may thus have been unable to achieve the concentration of forces necessary to operate effectively in Khusrau’s rear. At the same time, a Sasanid advance, or one conducted by their allies the Lakhmid Arabs against the verdant territories of Damascus and Emesa, could not be risked. This need to protect these cities and their territories was an operational fact of life.

… Tamkhusrau fell in the encounter, not through the courage of the Roman army but solely through its general’s devoutness and trust in God… Theodoric too, who was leader of the Scythian nations, fled with those around him, even though he had not come within range. It is conceivable that Theodoric commanded a newly raised barbarian force consisting of one or more squadrons of comitatenses (stqatiÍtai) designated as the Skythai Tiberianoi.45 If so, the troops’ poor performance may have been a consequence of not yet being fully conversant with the demands of operating with the Roman army in Mesopotamia.46 A Greek funerary inscription from a military cemetery outside the land walls of Constantinople at Mevlana Kapi mentions a certain Armatos who was a soldier in the numerus (or arithmos) of the Brachiati.47 It could date anytime after the later fifth century, so the question remains open as to whether the Armatos’ was interred before or after the apparent break-up of the praesental armies and their redeployment to Armenia ca. 528 and in the expeditions against Africa, Italy and Sicily in the 530s. Another 42

Praesental troops (Ðqihloà stqatiyt´m) were permanently redeployed to Armenia, along with some formations from Oriens, when Justinian established a new command, the magister militum per Armeniam, ca. 528. It is not known how many cavalry squadrons were transferred. John Malalas, Chronographia XVIII 10. Adonts‘ 1970:103125, 415-423. 43 Simocatta, Hist. III 17. 3-4. On the arguments for Theodoric as comes foederatorum, see Theodericus 2, PLRE IIIB: 1237. There is some doubt as to whether ‘praesental’ would be the best way to characterise the ‘Scythian’ troops organised by Tiberius Constantine in the mid570s. The term is applicable only if the force in question was normally stationed in Constantinople and used as an operational reserve or reinforcement. On the dearth of praesental troops in the Persian War of 540-545, see Trombley 2005: 393f., 404f. 44 Evagrius, HE V 20. (Adapted from Whitby 2000: 282f.). 45 On the origin of this designation, which went back to the time of Justinian (or possibly even Constantine the Great), see Maspero 1912a: 47 and note, 50f. 46 In a passage of doubtful date, but broadly ca. 580, Menander Protector observes: ‘…the indiscipline of the other [Roman] generals and their failure to follow [Maurice’s] orders brought the Romans into the gravest danger’. Menander Protector, Fr. 23.11 (Blockley 1985: 214f.). The praesental armies performed badly at the battle of Opadna in the Persian War of 502-7. Ps.-Joshua, Chronicle § 57 (Trombley and Watt 2000: 69f. and notes). Their lack of readiness for operations may well have convinced Justinian to keep the number of formations in the vicinity of Constantinople to the absolute minimum necessary to maintain security, and to redeploy the rest to provincial commands, reasoning that this would keep them in a better state of discipline and make them earn their pay more usefully. 47 Kalkan and ùahin 1995: no. 3.

The general absence of praesental armies in the war of 572-591 made it impossible to concentrate large forces in the way it had been done during the Persian War of 502-

36

For the site, see Whitby 1986. Simocatta, Hist. III 10. 5. 38 John of Ephesus, HE VI 2 (Payne Smith 1860: 369). 39 Procopius, Wars II 16.17-18. Trombley 2005: 398. 40 It is likely that the arithmoi of Armenia I and II — which Areobindus, magister militum per Orientem, had used to such good effect in the Persian War of 502-7 — now fell under the command of the magister militum per Armeniam. See Ps.-Joshua, Chronicle §§ 51-52 (TrombleyWatt 2000: 55-59 and notes 404, 436). 41 Greatrex and Lieu: 147f. (M. Greatrex) (Chronicon anonymum ad annum Christi 1234 pertinens LXVI: 203-5 (textus), 160f. (versio). On the approximate strength of these provincial moirai, see above, note 15. 37

325

THE LATE ROMAN ARMY IN THE EAST poorly spelled inscription in the same cemetery appears at first sight to mention the burial of the wife of a certain Theodore the cubitor (HeËdoqor joubÊtoqor). It could refer to a decempedator or officer in charge of surveying ground and presumably laying out camps and field fortifications. On the other hand, the grammatical and orthographic anomalies of the text might perhaps be resolved in this way: ‘Here lies Lapetia the wife of Theodore the excubitor, etc.’ (+ ÑmhÇde jatÇjite KapetÊa cum± (to³) HeodËqo(u) (Ñ)sjoubitËqor...).48 The excubitors and their commanders repeatedly turn up in the narratives of the Persian War of 572-591.

construct the siege engines that Marcian was eventually forced to leave in the hands of the enemy after he abandoned the siege.53 The name of only one ostensible infantry formation is given in the course of Theophylact Simocatta’s narrative, the legio IV Parthica, a formation of limitanei whose base had been moved since the early fifth century from Circesium to Beroea-Aleppo.54 This does not clarify the picture perfectly, for the legion is mentioned as playing the role of a mobile formation (and not ‘static’) in the battle of the Solachon plain in 586, which was a cavalry battle to judge from Theophylact Simocatta’s description.55 The circulation of coins at some sites where limitanei are known to have been based in Syria, as for example at Cyrrus in Euphratesia, may actually have increased in the decades after 550. Although the archaeological samples are limited, they might be taken to suggest that the limitanei were actually being reinforced in the decades before the outbreak of war in 572.56 There is an analogous rise in the number of papyri at Nessana involving soldiers during the same period.57 It is possible that, by the late sixth century, all formations in the frontier provinces of Oriens were horsed, and that their names were kept as points of reference for reasons of paying the roga and perhaps tradition.58 We know that the cavalry of the centre division or tourma in the Roman army at the battle of the Solachon plain was ordered to fight dismounted in the face of the shock tactics of the Sasanid centre division.59 The front ranks of well-trained cavalry formations clearly had infantry skills, and of course men fighting dismounted had the advantage of being able to fight using both their hands without having to balance themselves on their horses. The accumulation of dead and disabled men

Sometime after the loss of Dara in late 573, however, a new concentration point further back from the frontier was created. It consisted of a fortified position south of Mt. Aisouma northwest of Tella-Constantina, and was called Monocarton. The site was undoubtedly developed after Tiberius Constantine became Caesar on 7 December 574, for the place was renamed Tiberiopolis in his honour.49 Menander Protector explains its use as a staging area for operations in connection with peace negotiations that have been dated to 580/1 or 581/2:50 Meanwhile, the Persian general Tamkhusrau was encamped with the Persian army by the Mygdonius river in an area near Nisibis that was suitable for cavalry (Ñr tÀ ÚppÇsila). Maurice had established his army behind a defensive work at a place called Monocarton in the neighbourhood of the city of Constantina. The whole area around Monocarton has water and fodder for horses and is suitable for an army to encamp. Thus they protected themselves with a ditch. The place appears not to have been a permanent base but a staging area whose field fortifications had to be repaired annually. It remained in use until at least June 586, when the army of Oriens concentrated there under the command of Philippicus to receive the attack of Kardarigan’s army from the vicinity of Dara. The subsequent engagement was known as the battle of the Solachon plain.51

53

John of Ephesus, HE VI 2. On infantry formations in the 4th-5th c., see Schmitt 2001. It is possible that the infantry deployed at the battle of Tell Beshmai in November 503 were the legio I Parthica Nisibena based at Tella-Constantina, a formation of limitanei. Ps-Joshua, Chronicle § 51 (Trombley - Watt 2000: note 271). 55 Casey 1996: 218f. 56 Casey 1996: 218f. 57 Kraemer 1958 (P. Nessana III): nos. 15, 16, 19-22, 24, 26. 58 I consider Maspero’s suggestion decisive at Maspero 1912a: 48f. For other examples of 6th c. use of the names of formations found in the Notitia Dignitatum, see Grosse 1920: 273. Discoveries since Maspero and Grosse bear this out. For example, the Nessana papyri contain a letter dated 30 May 512 from a regular soldier (stqatiÍtgr) named Flavius Aws in the squadron of the Most Loyal Theodosiakoi (ÐqihlËr t´m jahosiylÈmym Heodosiaj´m) at Rhinicorura, Egypt. P. Nessana III, no. 15. It may have been replaced or renamed from the ala veterana Gallorum of the Notitia Dignitatum sometime between 425450. Notitia Dignitatum, Or. XXVIII.28. Cf. Maspero 1912a: 135. For the date of formations bearing the name Theodosiaci, see Hoffmann 1969-70: 242. Formations with this designation were still in existence in the West (viz. at Ravenna and Rome) ca. 600. Ibid. For their continued existence in the 10th c., see Haldon 1993: 13f. 59 It became standard practice for cavalry to fight dismounted on rough ground. Maurice, Strategikon XI 1, lines 64-67. The Sasanid tactical manual found in the ƖyƯn-nƗmeh recognises the centre and right divisions as the offensive formations of a field army drawn up for battle. On this, see Inostrancev 1926: 13. In the battle of the Solachon plain, the Sasanid left wing, which had been pushed aside by the Roman attack, added its weight to the attack on the Roman centre division. 54

One feature of Roman offensive operations that the sources leave in the dark is the employment of noncavalry formations.52 There was no point in undertaking siege operations with an army composed only of cavalry. Infantry were involved in the two marches against Nisibis in 573, along with engineering personnel required to 48

Kalkan - ùahin 1995: no. 4. For chronology, see Tiberius Constantinus 1, PLRE IIIB: 1323-1326. 50 Menander Protector, Fr. 26.1 (adapted from Blockley 1985: 230-33). 51 Goubert 1951: 94-98. 52 In describing operations in Thrace and Illyricum, Simocatta frequently mentions the cooperation of infantry and cavalry formations. This could be a consequence of the fact that the wars against the Avars were more recent, and are dealt with in greater detail than operations in Oriens, which ended in 591. It could also be that Simocatta regarded it as sub ratione to mention that all or most of the regular forces involved in offensive operations in Oriens were cavalry. 49

326

FRANK R. TROMBLEY: THE OPERATIONAL METHODS OF THE LATE ROMAN ARMY IN THE PERSIAN WAR OF 572-591 in the killing zone would also have worked to the advantage of dismounted men as well.60

the Romans broke down some of the Euphrates bridges to prevent the Sasanid army from crossing into Euphratesia.64 Bridges were undoubtedly an operational factor in the Persian War of 502-507, not least because the Euphrates used to flood from November to the end of March, necessitating floating bridges of boats in many localities.65 The Persians later developed portable bridging equipment that was available for use at the time of their campaign in Euphratesia in 542.66 The Romans had certainly developed the technique of building floating bridges (cevuqylÇtym jatasjeuÀr ... tÀr kecolÈmar pkytÇr) by the last decade of the sixth century, but undoubtedly much earlier.67 There were several important bridges on the Khabur river, one of them linking Thannouris on the east bank with the rest of Osrhoene.68 A good example of the perceived importance of bridges in operations is seen in a major Roman attempt to cross into Beth ‘Aramaye in June 582 that aimed, as Michael Whitby has argued, to reach the vicinity of Ctesiphon:69

Finally, naval troops must have been used in one operation to manoeuvre the supply ships during Maurice’s march down the Euphrates in 581. One would suppose that the boat men constructed their vessels somewhere near the point where the middle Euphrates becomes navigable, using timber that was floated down from the Armenian mountains, then collected soldier’s bread (bucellatum) and other comestibles from the cities of Osrhoene and Euphratesia lying along the great bend of the Euphrates.61 The combined professional skills of these personnel proved insufficient in themselves to gain decisive, long-term advantages for the Roman Empire. Infrastructure Infrastructure in the broadest sense consists of the material and administrative environment necessary for the execution of operations. In the sixth century it involved the construction and maintenance of fortifications, grain silos, barracks, roads and bridges. It also involved the procurement and allocation of personnel and building materials to carry out this work. Finally, it entailed the procurement of the transport wagons, horses and pack animals necessary to ensure the mobility and tactical effectiveness of the cavalry squadrons that provided the main striking force in offensive and defensive operations.

A subsequent time Maurice and al-Mundhir b. HarƯth king of the Arabs concentrated their forces and marched into Persian territory by the route through the desert, and penetrated into the enemy’s dominions for a distance of many score of leagues, as far as Beth Arabaye. But on arriving at the great bridge there, upon which they had relied for crossing over [to Ctesiphon?] and subduing the wealthy cities on the opposite side, they found it cut away, for when the Persians learned their intentions they destroyed it.

Bridges were an important but possibly underrated factor in the scheme of operations practised in Roman Mesopotamia, Osrhoene, and Syria-Euphratesia, as also in Sasanid Arbayistan and Asurestan, the regions known as Beth Arabaye and Beth Aramaye in the Syriac sources.62 There were a number of permanent masonry bridges on military roads leading from Antioch across the Orontes (Jisr al-HadƯd, Derkous and Jisr al-Shughnjr).63 It will be recalled that, during Shapur II’s campaign in 359, the location of bridges had a direct impact on the movements of the Persian and Roman armies, and that

John of Ephesus was well connected, and apparently got this piece of information from someone who participated in the operation; it reflects the opinions of ordinary civil officials and military officers.70 The importance of the campaign is apparent through the fact that supply ships sailed down the Euphrates to meet the army at Callinicum—being subsequently burned when the operation was called off. The official explanation for the failure to cross the river was that the Sasanid general Adarmahan had — using the ‘indirect approach’ — ridden past the Arab-Roman army, raided Osrhoene and attacked the fortified town of Callinicum, which lay on

60 It would be useful to know if their horses were kept near the front line, or whether they were gathered in the rear. 61 The steppes of Syria-Euphratesia and Osrhoene-Mesoptamia seem to have lost their forestation in earlier antiquity. Mouterde and Poidebard 1945: 15. In or after 545 a Sasanid commander is said to have hauled large numbers of wooden beams into Lazica ostensibly for the construction of ships (Ñr mg´m poÊgsim), but their point of origin is not indicated. Procopius, De Bellis II 29. 1. The Persians also cut down trees in the territory of Edessa during the siege of 544, but these may well have been olive trees and therefore not suitable for ship construction. Procopius, De Bellis II 26. 23. It seems doubtful that timber suitable for this purpose existed in the steppe zones of Osrhoene, where there is little good evidence of forestation. In general see Geyer 2001: 89-218. In contrast, we learn of tall cypress trees at Daphne on the Orontes below Antioch. Procopius, De Bellis II 14. 5. This was perhaps characteristic of rivers lying near massifs. On supplying comestibles to the soldiery in Osrhoene, see Novel 163 of Tiberius Constantine, as in Appendix II below. 62 Poidebard 1939 gives a map of the road network south of Tnjr ‘AbdƯn. Cf. A. Kazhdan, Bridges, in Kazhdan 1991: 324f. 63 Mouterde and Poidebard 1945: large folding map.

64

Trombley 1999: 23f. For references, see of Ps.-Joshua, Chronicle (Trombley and Watt 2000: 157). W. Kaegi, Euphrates, in Kazhdan 1991: 748. 66 Procopius, De Bellis II 21.21-22. 67 Maurice, Strategikon XI 4, line 75. 68 Poidebard 1939: Planche CXVII. For other Khabur bridges, see map at Planche CXI. 69 John of Ephesus, HE VI 16 (adapted from Payne Smith 1860: 413). Simocatta fails to mention this incident at Hist. III 16. 5-11 (Whitby and Whitby 1986: 100 and note 82). Cf. Greatrex and Lieu 2002: 164 (J. van Ginkel). Because Anatha was on the east bank of the Euphrates, any bridge that the Roman army encountered must necessarily have been on the Tigris in the vicinity of Ctesiphon-Seleucia. It is peculiar that Simocatta nowhere states that the expedition aimed at reaching Ctesiphon. 70 Maurice’s co-operation with al-Mundhir took place in 581. The Syriac historian got his account of battle of Melitene in 576 from the interpreters who accompanied one of the armies. HE VI 9 (Payne Smith 2000: 397). 65

327

THE LATE ROMAN ARMY IN THE EAST the east bank of the Euphrates at its confluence with the Khabur river, and whose loss would have threatened the Roman army’s eventual retirement.71 The bridge was thus only one factor in the cancellation of the expedition. John of Ephesus’ statement must therefore reflect a soldier’s eye view of the situation, one that was inevitably confused by the impression of unpleasant events and seldom had an overall grasp of the operational situation.72 John of Ephesus has told only one side of the story, for the Syriac Chronicle of 1234 indicates an apparently unsuccessful attempt to seize Anatha by coup de main, a town lying between Peroz-Shapur and Circesium on the east bank of the Euphrates previously during the same expedition:73

lay in the vicinity of Seleucia Pieria (present-day Magharajik) in the Orontes basin southwest of Antioch in Syria I. The work was completed in 524:78

They laid siege to a certain fort which was called ‘Anat (Anatha). They fought greatly against it; they armed themselves and got on small boats in the waters of the Euphrates, and made war from the boats. But the Persians fought them from the wall with stones and ballistae, and many were drowned and perished.

It appears that masonry piles supported the structure. The foundations of another Roman bridge have been observed on the Jaghjagh river at Soufeiye south of Nisibis,79 and the building inscription of another survives at Tafas northeast of Damascus in Phoenice Libanensis.80 Wooden bridges and bridges of boats were also constructed during a rehabilitation of the Sasanid infrastructure that took place during the reign of Khusrau I (531-579), but there is little archaeological evidence to fill out the picture:81

(Cross) In the time of the most magnificent and glorious Ephremius, Count of the Sacred Largess and (Count) of the East, there were constructed the bridges of the two Black Torrents (aÚ cÈvuqai t´m dÌo LekÇmtym), and (the bridge) at the fifth milestone. The sixty-four arches (or ‘piers’) (lÈtqai) were renewed and the corresponding construction work (was done) with cut-waters (joilotËloi) (sic), in the month of Dios in the third indiction in year 573 (of the era of Antioch).

Sasanid territory was a land of bridges, and they existed in large numbers on the Tigris and Euphrates in Asuristan near the Sasanid royal palaces, near al-Hira (the Lakhmid Arabs’ principal settlement), on the Greater and Lesser Zab rivers and in the Zagros mountains.74 Some of the bridges in the vicinity of Ctesiphon were built of stone,75 as are the surviving archaeological examples, including the Roman bridge at Dara constructed of massive voussoir blocks and two others reported on the Khabur river.76 The investment required to build these installations is in itself proof of their importance. Knowledge of the engineering techniques needed to construct strong bridges was widely disseminated. Menander Protector observes:77

He likewise ordered the rebuilding of every wooden bridge or bridge of boats (jisr) that had been destroyed and of every masonry bridge (qantara) that had been smashed… Along the highways he built castles and towers. These bridges were undoubtedly still in use during the Persian War of 572-591. Similarly, the Roman bridges left inside the Sasanid frontier after the Persions’ acquisition of Nisibis in 363, like the one on the Jaghjagh river, were undoubtedly kept in repair thereafter. Their continuing maintenance might be inferred from a report of Menander Protector dating from shortly after the death of Khusrau I (February or March 579). It remarks the existence of Sasanid towers and small forts, but unfortunately fails to mention any bridges:82

For bridges there is a triangular structure with a sharp edge to the front like a triangle. The engineers (lgwamopoioÊ) called it a cutwater (julotËlor), and it acts like the ram of a warship. Such devices could have been made either of masonry or wood. The Romans constructed masonry bridges, some of them quite extensive, across the rivers and seasonal torrents of the Syrian provinces. These served economic as well military functions. The building inscription of one

[Speaking of the Persians’ transit arrangements for the Roman peace envoys Zachariah and Theodore:] After him came another Persian who slowed down their journey… He halted at stations (stahloÊ) that were very close together, having covered only a few parasangs… This was done with purpose and hostile intent… during which time the Persians would be… collecting supplies to be stored in the cities of Nisibis and Dara and laying up food stocks in the smaller

71 The resounding victory at Callinicum claimed by Simocatta is doubtful. Whitby 1988: 273. 72 John Keegan has emphasised the ‘personal angle of vision’ vis-à-vis tactical situations in Keegan 1976: 128-133. The operational situation was likely to be even more difficult to grasp for the average soldier or officer. 73 Greatrex and Lieu 2002: 165. 74 al-TabarƯ, Ta’rƯkh: 837, 885, 888 (Bosworth 1999: 52, 130, 140). On the literary evidence for al-Hira, see Rothstein 1899: 18-40. 75 al-TabarƯ, Ta’rƯkh, 1028 (Bosworth 1999: 357). 76 Dillemann 1962: 130 and Planche XI. 77 Menander Protector, Fr. [38] (Blockley 1985: 246f.). Menander’s observation concerns the bridging of a tributary channel other than the Danube by the Khagan of the Avars.

78

IGLS 1142. Ed. pr. Chapot 1901: no. 6. Poidebard 1939: 142f., Planches CXVIII, CXIX, large folding map A-B/h. 80 SEG 7, no. 256 (343-348 AD). On bridges see also Dagron and Feissel 1987: 96f. (no. No. 51). For two bridges of Justinian, one over the Sangarius in Galatia, the other at Adana in Cilicia Pedias, see Feissel 2000: nos. 29 and 43. 81 al-TabarƯ, Ta’rƯkh, 897f. (Bosworth 1999: 157). 82 Menander Protector, Fr. 211 (Blockley 1985: 210f.). 79

328

FRANK R. TROMBLEY: THE OPERATIONAL METHODS OF THE LATE ROMAN ARMY IN THE PERSIAN WAR OF 572-591 Antioch highway to Beroea and Chalcis, are likely to have guaranteed movement in all seasons.89 The ongoing maintenance of roads in provinces far behind the frontier as operational routes for moving troops and supplies is known from an inscription of 536. It records the widening of an artificial defile in the vicinity of Anazarbus in Cilicia Pedias.90 Wagon haulage was seldom practicable, however, in raids and deep penetration operations, because of the slow pace and limited endurance of oxen.91 The ecclesiastical historian Evagrius of Epiphaneia was aware of the problems of moving supplies thanks to his connections with the officials of the praetorian prefecture of Oriens. In his version of the speech Gregory patriarch of Alexandria made to the dissident soldiery at Litarba in 588, the bishop says to them:92

forts (vqoÌqia) both on the far side and on this side of the Tigris. Many bridges are mentioned in Theophanes Confessor’s account of Heraclius’ campaign of 627-628 in Sasanid territory. Many others bridges at unspecified locations are reported to have been restored during the brief reign of Buran daughter of Khusrau II (630-631).83 It is likely that the Romans and Persians maintained a system of fire signals to alert the towns in the interior of the enemy’s approach and accelerate the dispatch of relief forces. Examples of these survive at many locations within visual range of key administrative centres. These installations consisted of circular masonry platforms some six to ten metres in diameter, atop which large fires could be ignited.84 There were nevertheless many occasions when the attacking party achieved operational surprise in these incursions. The only reported instance of fire signals during the Persian War of 572-591 is an ad hoc case concerned with a Roman attack on Akbas, a Persian fortress in Arzanene:85

[For the army] to remain all together is an impossibility: for from where will there be the conveyance of seasonal crops, or those things which the sea grants to the land in commercial exchange…? The Romans are known to have organised a flotilla of supply ships on the Euphrates especially for the expedition of 582. This method had certain inherent disadvantages, for the construction of large numbers of river craft was likely to have attracted the attention of enemy spies. If the expedition ran into trouble, as happened in 582, the ships would have had to be burned.

When the [Roman] general attacked those in the fort, the besieged lit signal-fires (vqujtyqÊai) on the parapet of the fortification, indicating the magnitude of the danger to [the Sasanid general] Kardarigan’s men who were striving to reduce Aphumon; for the Persian general had established such an agreement with the men in Akbas.

Theophylact Simocatta’s semi-official account indicates the use of mules to carry Philippicus’ baggage into Arzanene during the exploitation phase after the battle of the Solachon plain in 586, but it is not immediately clear whether the mules pulled wagons or were employed merely as pack animals.93 Both sides used camels in large numbers along with other pack animals, particularly in cavalry operations.94 Camels also played an important tactical role in the battle of Solachon:95

Fire signals were thus an obvious ploy during a defensive operation, particularly as the Roman commander adopted the ‘indirect approach’ to relieving the siege of Aphumon by riding deep into Arzanene and attacking Sasanid installations.86 The transport of supplies is known only from occasional references in the sources.87 The road system and wellconstructed bridges would have enabled the Roman army to use wagon transport over considerable distances from its weapons fabricae in eastern Asia Minor and supply sources in Euphratesia, Mesopotamia, Osrhoene, and Syria I and II.88 Long sections of road, including the

Then [the Persians] set out from the river Bouron [near Dara] and moved towards the Romans [on the Solachon plain]. Next they loaded up herds of camels with a heavy burden of water-skins, so that the Romans should not conquer them easily because of their shortage of the watery element.

83

al-TabarƯ, Ta’rƯkh, 1064 (Bosworth 1999: 404). Poidebard 1939: Planches XXXIX, LIX, XCIV, XCV. 85 Simocatta, Hist. I 12. 3 (adapted from Whitby and Whitby 1986: 37). 86 The operational principle of creating a diversion by sending a striking force into enemy territory is recognised at Maurice, Strategikon X 2, lines 23-29. 87 In the campaign against Euphratesia in 542, the Sasanid army had exhausted ‘the supplies they first had when they invaded the land of the Romans’, and in consequence Khusrau I decided to risk battle by crossing the Euphrates into Osrhoene to replenish his supplies in the agricultural zone along the river. Procopius, De Bellis II 21. 16. 88 A. Kazhdan, Roads, Kazhdan 1991: 1798. Belisarius’ force that invaded Sasanid Beth ‘Arabaye in 541 used wagons to evacuate its sick and wounded. It is not stated whether they were the army’s supply wagons or vehicles requisitioned from the locals. Procopius, De Bellis II 19. 45. Vehicles of this type (ÜpofÌcia) were normally drawn by oxen. For the road system of the entire limes south of Tnjr ‘AbdƯn, see Poidebard 1939: large folding map. 84

89 Poidebard 1939: Planches CLVI, CLVII, CLVIII.2, large folding map. 90 Dagron and Feissel 1987: 167f. (no. 105). 91 See the detailed discussion of using ox-drawn haulage in brush country during the Zulu War of 1879 in Morris 1994: 310-313. 92 Evagrius, HE VI 12, (Whitby 2000: xiv-xv, 304). Goubert 1951: 112f. As to Gregory’s mention of the importation of fish, see Kraemer 1958 (P.Nessana III): nos. 47 (ante 605?) and 85 (late 7th c.), which mention the shipment of fish from Gaza to the inland kastron of Nessana, which lay deep in the Negev desert. 93 Simocatta, Hist. II 9. 12. 94 Simocatta, Hist. II 8. 1. The capabilities of camels for military purposes half a century later are examined in Hill 1975. 95 Simocatta, Hist. II 1. 4 (Whitby and Whitby 1986: 45).

329

THE LATE ROMAN ARMY IN THE EAST The Ghassanid Arabs possessed great herds of camels.96 The Praetorian Prefect of Oriens will have purchased and requisitioned large numbers of these useful beasts to support military operations.97

the villages,103 while the Sasanids could contemplate the prospect of an economically revitalised region that would be liable to future predatory raids. In April 575 Tiberius Constantine issued a law granting tax remissions to various provinces, including Osrhoene and Roman Mesopotamia.104 There is no corresponding edict for other affected provinces, including Euphratesia and Syria I and II, but we know from a papyrus of Petra that a twenty-five percent tax remission was granted to Arabia.105 This is a clear indication of the damage done to the provincial infrastructure, in terms of the depopulation of the countryside, destruction of agricultural capital and lost productivity through abandonment of the land. Operations shifted northward to Roman Mesopotamia, Sasanid Arzanene and Roman Armenia for the three years between 575-578. Defensive manoeuvre warfare involved blocking the passes of north-eastern Anatolia from Sasanid penetration in 576. It proved relatively easy for the Roman force led by Justinian son of Germanus, magister militum per Orientem, to stop this and administer a check near Melitene in Armenia II. Khusrau I found it difficult to use his possibly numerically superior cavalry to full advantage in the eastern Anatolian passes.106

Camels were also requisitioned from the towns of the eastern provinces and used to equip military formations. They were employed by merchants and for agriculture in the territories of the towns.98 For example, a papyrus of Nessana (6th c.) mentions a requisition (dgkgcatÊym) of camels (mostly dromedaries but also ordinary ‘camels’) sent by a regional administrator, nineteen of which were assigned to junior officers (pqÊoqer) and officials on the staff of the dux Palaestinae (doujijoÊ). The lists also mention five scouts and two koursores, who were either front-line soldiers trained in the formation tactics of mobile warfare or perhaps simply couriers.99 It is not clear whether the military personnel were supposed to take the camels from stock-raising families in the environs of the kastron of Nessana or from Bedouin tribes living more widely dispersed in its territory. The administrative tasks of civil officials based in Nessana extended quite far, to judge from the fact that two of the requisitioned dromedaries were destined for the town of Charachmoubǀn (present-day al-Karak), lying on the road between Petra and Amman-Philadelphia.100 It seems likely that the troops billeted Nessana patrolled this route, and had some responsibility, along with troops stationed in other small fortresses, for watching the eastern boundaries of Palaestina III, the province in which all these towns lay. It is unknown whether an entire arithmos of cavalry (also designated as a bandon or numerus in the Strategikon of Maurice), numbering anywhere from 200 to 500 men (on the assumption that most units were under-strength) was billeted in Nessana.101

Thereafter Roman defensive operations were confined mainly to covering the territories of key cities from Sasanid raiding. For example, in June 578 the Sasanid generals Sarnachorgan and Tamkhusrau crossed the frontiers several days before the expiry of the truce of 575 and raided the rural territories of Constantina, Resaina and Amida in Mesopotamia, involving, it was said, some 12,000 Persian clibanarii (PÈqsai huqeovËqoi) and horse-archers, and perhaps as many as 8,000 allied Arabs and Sabir Huns, who fought using their own ethnic weapons.107 The raid lasted a mere fifteen days.108 The Roman fortress at Thannouris on the Khabur river seems to have been evacuated, once the size of the Persian army became apparent.109 Sasanid generals like Tamkhusrau and Kardarigan and their Arab allies were usually hoping to catch local garrisons and agricultural personnel by surprise through swift

Defensive Operations Another aspect of the war of 572-591 was defensive operations. The regime in Constantinople had no desire to repeat the catastrophic depopulation of Syria caused by Adarmahan’s raid in late 573. Under an agreement negotiated in 575, Syria I and II, Osrhoene, Mesopotamia and Euphratesia were put under truce for three years.102 The scheme was reciprocally advantageous: the Romans could rebuild the provincial infrastructure and repopulate

103 If the imperial authorities sought to do this, they seem to have been unsuccessful in repopulating Syria II, particularly the territory of Apamea. Trombley 1997: 177f. 104 Novella CLXIII, in Corpus Iuris Civilis III (Schoell 1972: 749-751). 105 It is therefore likely that analogous laws were issued for the other affected provinces, particularly Syria I and II, but that they established no new legal principle, being a consequence of uniquely catastrophic conditions, and were therefore not included in the codification. To judge from Theophylact Simocatta’s account, Arabia was less affected yet received a twenty-five percent tax remission. See Appendix II below. 106 John of Ephesus, HE VI 8 (Payne Smith 1860: 392). See below for further discussion of the campaign of 576. 107 For the date, see Szádeczky-Kardoss 1976: 110f. On the fortifications of the cities, see Gregory 1966: 59-65 (Amida), 89-93. 108 Menander Protector, Fr. 23.1 (Blockley 1985: 198f. and notes). The Sasanid cavalry levy is often reported to have numbered in the range of 12-15,000, e.g. the Sasanid invasion of Euphratesia in 531. Procopius, De Bellis I 17. 1-2. Cf. R. P. Harper, Amida, in Stillwell 1976: 49. 109 Menander Protector, Fr. 23.5.

96

John of Ephesus, HE VI 3 (Payne Smith 1860: 370). On camels’ capabilities for transport and supply work, see Bulliet 1975: 176-215. Cf. A. Kazhdan, Camels, in Kazhdan 1991: 368. 98 Cf. the example of a merchant who grazed his camels in the territory of Cyrrhus in Euphratesia. Gheyn 1900: 315. Camels are mentioned in the fauna of the Syrian provinces, even in largely sedentary villages. Trombley 1994: 183, 187, etc. 99 Kraemer 1958 (P.Nessana III): no. 35. Cf. L. MacCoull, Nessana papyri, in Kazhdan 1991: 1458f. 100 Kraemer 1958: 110, note 12. 101 Egypt is one of the very few places where developments in military organisation can be traced throughout the 6th, including the size of formations. See the remarks of Maspero 1912a: 104-113. 102 The negotiations are covered in detail by Menander Protector and to a lesser extent by John of Ephesus. Goubert 1951: 71f. 97

330

FRANK R. TROMBLEY: THE OPERATIONAL METHODS OF THE LATE ROMAN ARMY IN THE PERSIAN WAR OF 572-591 approaches.110 This led to dangerous situations in the interior, because the Roman field army and provincial moirai were frequently concentrated in forward positions and unable to move quickly enough to block these incursions. Theophylact Simocatta offers no information about major Roman formations resisting the raid of 578, so the task of defence presumably lay in the hands of the limitanei descended from the lists given in the Notitia Dignitatum (if these formations still existed), local militias, and any arithmoi that were covering their bases in the interior of their provinces.111 The provincials’ feelings of consternation at ‘rumours’ of war, quite possibly after Tamkhusrau’s raid of 578, are reflected in a striking passage from the Syriac chronicle of 1234:112

To the best-chosen (or ‘most beloved’) emperor Justinus: his army of Oriens constructed (this monument) to his fame: a warning to the barbarian nation, coming from the statue of the triumphal chariot. (Iustino imperatori dilectissimo exercitus eius Orientis gentem barbaram e curru statuario cautionem ad gloriam construxit). The threat was well grounded. Roman offensive operations in Persia after this created much destruction in the territories of the Sasanid cities, but seldom resulted in the destruction of Sasanid armies. In 585 the Persians staged an eight-day raid that reached Monocarton-Tiberiopolis — an important concentration point for the Roman armies throughout the war — and Martyropolis.117 Almost nothing is known of Roman defensive manoeuvres in this and other situations.118 There is only one piece of unambiguous evidence about this. In 588, during the mutiny against Priscus taking command of the army of Oriens, Germanus, dux Phoenicae Libanensis and the elected spokesman for the mutineers, led a scratch force of 1,000 men to save Constantina from Sasanid assault.119 This is a good example of how the arrival of a small mobile force could quickly alter the outcome of a siege. The mutiny demonstrated that the absence of an active Roman army in the zone of operations was an invitation to catastrophe.

At this time there was anxiety in the cities beyond the Euphrates, and they were steadfast, by night and day, on Sundays and feast-days, in building the walls of cities, and all were in distress and fear. In the summer and autumn of 578, Maurice indirectly shielded the towns of Osrhoene and Mesopotamia by drawing Sasanid forces away to the south-east, conducting an offensive operation with the cavalry squadrons of Oriens against Sasanid Arbayistan (Beth ‘Arabaye) by operating south and south-east of Nisibis and taking Singara.113 It is possible that Tamkhusrau’s fifteen-day raid in June-August 578 was cut short by the news of Roman troop concentrations to the south, for the Sasanid army retired eastward, took possession of Thannouris on the Khabur river, ‘fearing that the Roman attack would take place immediately’, and finally pulled back into Persian territory, probably to oppose Maurice’s incursion.114 It is a good example of the ‘indirect approach’ to defensive operations being adopted well before Philippicus began his study of the ancient tactical manuals in the 580s. The subsequent Roman retirement came near the beginning of the winter of 578/9.115 The operations of 578 provided the occasion for a gesture of defiance, the erection of a monumental chariot in Sasanid territory. Its accompanying inscription, no longer in existence, was seen by a traveller in 1673 at a bridge near Nisibis:116

Manoeuvre Warfare A new phase in Roman operational methods came with the arrival of two men who served as magister militum per Orientem, Maurice (577-582) and, after the latter became emperor, Philippicus (584-587/8, iter 588-589), along with a number of competent subordinate officers, some of them natives of the empire, but others of barbarian origin, who had been recruited in the mid 570s.120 Among the latter was the German Eilifreda, who as dux Phoenicae Libanensis rose to the rank of division commander (lÈqaqwgr or toÌqlaqwor) between ca. 586-588.121 Their contribution to operational methods lay in what Basil Liddell-Hart called the ‘indirect approach’, or what the classical Chinese writer Sun Tzu referred to

110

A good example of this is seen in the Lakhmid Arabs’ advance against Carrhae-HarrƗn in 503, during the Persian War of 502-507. Ps.Joshua, Chronicle § 52 (Trombley and Watt 2000: 57f. and notes). 111 On limitanei, see the remarks of Whitby 1995: 68-74. The rustics who took refuge in fortified towns seem to have made vigilant sentries, to judge from a report about the siege of Edessa in 544. Procopius, De Bellis II 27. 20. 112 Translated at Greatrex and Lieu 2002: 162 (M. Greatrex) = Chronicon anonymum as annum Christi 1234 pertinens LXXIII: 209 (textus), 164 (versio). 113 On the fortifications of Singara, see Gregory 1997: 104-108. 114 Menander Protector, Fr. 23.5 (Blockley 1985: 200f.). 115 Simocatta, Hist. III 16. 1-2. 116 CIL III, no. 212. We had previously supposed that the inscription belonged to the reign of Justin I, as in Trombley-Watt, ps.-Joshua, Chronicle, xlvii, but Maurice’s operations in 578 supply an appropriate occasion and context for the document. A Latin document of this date would have been unusual in any part of greater Syria, but I consider this to be a consequence of the large number of western ‘barbarian’ mercenaries in the army at this time, the so-called Tiberiani and officers

like the German Eilifredas dux of Phoenice Libanensis who could interpret the monument and inscription to their subordinates. For the likely Latinity of the army at this time, see Appendix III below. This discussion has profited from the paper of Pierre-Louis Gatier. 117 Simocatta, Hist. I 14. 6-7. 118 Menander Protector, Fr. 23.1-6. 119 Simocatta, Hist. III 3.9. A predecessor of Germanus Zenodorus, or perhaps commander of one of the provincial arithmoi, may have been Zenodorus, whose funerary inscription at al-Qunaitra near Damascus seems to indicate that he was dux of Phoenice Libanensis. He was killed in battle in 569 at the age of forty-five, fighting alongside the empire’s Ghassanid allies, perhaps against a raid conducted by non-allied Arabs. SEG 7, nos. 249, 250. 120 Fl. Mauricius Tiberius 4, PLRE IIIB: 855-860. Philippicus 3, PLRE IIIB: 1022-1026. 121 Eiliphredas, PLRE IIIA: 435.

331

THE LATE ROMAN ARMY IN THE EAST as ‘striking into the void’.122 Instead of marching to block Sasanid incursions, the magister militum per Orientem would strike deep into Sasanid territory with cavalry, threatening the captivity of village populations in the vast stretches of countryside running from the environs of Nisibis all the way to the Tigris river, and even beyond. There is an example of this in the campaign of 580:123

Theophylact Simocatta observes that in 585 Philippicus began studying the tactical manuals with a view to imitating the Scipionic strategy of the ‘indirect approach’ that had led to the failure of Hannibal’s enterprise in Italy during the Second Punic War, but the contemporary use of this method actually went back to the first year of Maurice’s command as magister militum of Oriens in 578, well before Philippicus took charge.126 Some of the modern concepts seen in the resultant improvement in Roman operational technique include standing on the operational defensive till attacked, followed by active defence and rapid exploitation, tactical envelopment in particular engagements, attacking the economic basis of the enemy’s war effort by depopulating his agricultural lands, and applying the principle of concentration consciously before engaging in a general battle.127

When Hormisdas’ arrogance was made plain to the emperor Tiberius and summer had again arrived, Maurice collected his forces and reached Persia, after sending Romanus, Theodoric and furthermore Martin to the far side of the Tigris to lay waste to the interior of Media. And so they invaded with the mass of the army, and pillaged the fertile and most fruitful areas of the Persians; after spending the whole summer in the slaughter of Persians, they ravaged Media and wrought extensive destruction.

The full application of this system can be seen in the campaign of 586, culminating in the battle of the Solachon plain. The Roman army under Philippicus magister militum per Orientem moved out of winter quarters to Mambrathon — apparently its assembly point — and from there advanced to a site known as ‘Bibas’. He undoubtedly concentrated the moirai or droungoi of the duces of several provinces there, those of SyriaEuphratesia, Phoenice Libanensis, Osrhoene and Mesopotamia, in addition to the organic troops of his command, all or most of whom were cavalry billeted in the vicinity of Antioch, having at his disposal some fifty arithmoi, most of them probably near full strength.128 If so, Philippicus’ army may have numbered in the range of 20,000 men. It also included formations of limitanei, like the legio IV Parthica from Beroea-Aleppo in Syria I.129

These Roman forces seem to have moved without siege personnel or artillery, having as their aim to overwhelm smaller fortresses by coup de main and destroy any Sasanid formations they encountered on the march. The Roman forces exploited the military highways running across the centre of Tnjr ‘AbdƯn, using the massif to mask their movements and as a secure transit route between the provinces of Osrhoene and Mesopotamia. It was possible to stage a raid into Arzanene, re-cross the Tigris and Tnjr ‘AbdƯn, and then suddenly appear in Sasanid Arbayistan, as happened in 584. Philippicus initially concentrated the forces of Oriens at Monocarton-Tiberiopolis, moved in the direction of Carcharoman, a site on ‘high ground’ (possibly Tnjr ‘AbdƯn), and then marched the forces of Oriens against territory of Nisibis where he gained large plunder:124

The location of ‘Bibas’ has been hotly disputed, and its correct identification affects our understanding of the initial movements of the Philippicus and the Sasanid commander the Kardarigan. Dillemann has identified ‘Bibas’ as Bebasa on the Bouron river, lying some 50 km. due south of Dara. If this were correct, Philippicus would have moved his army south-eastward, threatening to advance along the old Roman military road in the direction of Singara, but also leaving him in an exposed position in case the Sasanid army assembling near Dara and Nisibis moved westward in the direction of Osrhoene.130 In accordance with this reconstruction, Philippicus’ advance would have been a feint to draw the Sasanid main body southward along the Bouron river and

Then … the general [Philippicus] concentrated his forces, transferred his camp to Mount Izala [Tnjr ‘AbdƯn], because the locale offers security to those who surmount it, and handed over the booty to a very observant guard. The centre of the massif was a natural fortress and was used as a convenient concentration point for captives taken from the enemy, where they could be guarded against Sasanid attempts recover them until otherwise disposed of. Tnjr ‘AbdƯn did require garrisoning, for the enemy did penetrate Tnjr ‘AbdƯn on occasion, as in 581 when the Qartmin monastery was raided.125

126

See below. Applications of these principles discussed below include operational defensive followed by offensive operations (battle of the Solachon plain, 586), attacking the enemy’s economic potential (Maurice’s raid against the territory of Nisibis, 578), and concentration (advance against the Sasanid rebel BahrƗm in territory of Nanisenes, 591). 128 The figures for the number of available arithmoi here are estimates, based on the force structure outlined in the Notitia Dignitatum. See above, note 15. 129 On limitanei, see the remarks of Whitby 1995: 68-74. 130 Simocatta puts Bibas on the Arzamon river, but according to Dilleman this is a mistake for Bebasa on the Bouron. Dilleman 1962: 294. 127

122 Sun Tzu 6. 6: ‘Go into emptiness, strike voids, bypass what [the enemy] defends, hit him where he does not expect you’. Sun Tzu 1963: 96. Liddell Hart 1967: 59-74. Liddell Hart might have drawn interesting conclusions, had a translation of Theophylact Simocatta been available to him. 123 Simocatta, Hist. III 17. 3-4 (Whitby and Whitby 1986: 99). 124 Simocatta, Hist. I 13. 7 (Whitby and Whitby 1986: 39). 125 I follow Whitby’s date vis-à-vis Simocatta, Hist. III 17. 8. The Syriac chronicle of 819 places the event in 579/80, the chronicle of 1234 in 580/1. Greatrex and Lieu 2002: 165.

332

FRANK R. TROMBLEY: THE OPERATIONAL METHODS OF THE LATE ROMAN ARMY IN THE PERSIAN WAR OF 572-591 away from Dara, to tire it out in the mid-summer aridity of Beth ‘Arabaye, and perhaps to create an opportunity to retake Dara.131 After successfully drawing the enemy southward, Philippicus would then have pulled back the army to a site near the spurs of Tnjr ‘AbdƯn where, as Theophylact Simocatta indicates, he fortified a camp on high ground above the steppe; it probably lay to the east of the Arzamon river in the vicinity of Apadna.132 Whitby offers a much simpler interpretation, grounded on identifying ‘Bibas’ as Tell Beú, the southernmost site in the quadrilateral it forms with Monocarton-Tiberiopolis, Apadna and Solachon. If this identification is correct, the withdrawal from ‘Bibas’ to the camp near Apadna would have covered only a few kilometres, a more plausible sequence of events in view of the arid and hot summer than the one proposed by Dillemann. Whitby’s explanation also has the advantage being more consistent with Theophylact Simocatta’s account, which is here based on official or semi-official sources of information.

useful study to see how a Roman tactical success increasingly became the basis of operational exploitation between 576-590. This had become a regular feature of Roman operational doctrine by the time the Strategikon was drafted ca. 600. There were a great many features of the ecology and human-built infrastructure of Roman Armenia, Mesopotamia and Syria that favoured the operational movements of the field armies and slowed them down (‘friction’).134 Factors in the organisation and command of the armies had both negative and positive effects (‘genius’ or the lack thereof).135 Commanders with aggressive yet genial personalities like Philippicus succeeded in overriding the inertia of troops in their winter quarters and motivating them to ride to the army’s concentration points in a timely way. If the year’s expedition involved raiding Sasanid territory and ended in operational failure, the commander’s character and personality enabled the army to maintain its cohesion while retiring under enemy pressure, as happened during the dangerous retreat from Arzanene in 586.

Probably on the strength of intelligence provided by ubiquitous scouts, Philippicus calculated that the Sasanid army was a large one, and eager for battle. The Apadna position was designed to observe the movement of the Sasanid army on the steppe and threaten its northern flank if it made for the Arzamon river to water its horses and replenish the water skins carried by the Persian soldiery. The torrent known as Karin Su or Nahr al-Awaydi, lying roughly midway between the Arzamon and Bouron, had apparently gone dry by this time of the summer. As it turned out, all Philippicus had to do to confront the enemy was to move south-eastward down a shallow declivity in the direction of the Persian advance. It offered an easy approach to the eventual site of the battle at Solachon (present-day Salah), lying some 8 km. to the southeast. If retreat became necessary, the Roman army could follow the same route back to its source of water on the high ground near Apadna, continue to cover the steppe and the upper crossings of the river, and block the southern approaches to Tnjr ‘AbdƯn.133 The selection of this ground seems to have been carefully considered in light of the possibilities for manoeuvre and obtaining water, and in light of the aggressiveness of the Kardarigan. Philippicus waited in this position for three days, and then occupied the position at Solachon, all the while being informed of the direction being taken by Persian army. In its initial stages the operation was seen as a defensive one, but the outcome of the battle allowed a rapid exploitation of the victory into Sasanid Arbayistan. An analysis of course of the battle lies in the realm of tactics rather than operations, but it makes a

Among the operational tools and methods that enabled imperial commanders to overcome friction during the offensive operations of 578-591 was the network of pre363 Roman military highways that now lay in Sasanid territory, extending from Nisibis eastward to Bezabde on the Tigris, and another which crossed the river, running in a south-easterly direction towards the lower course of the Greater Zab.136 Another Roman highway ran south, crossing Jabal Sinjar and reaching Singara; east of Singara it then divided into two routes, one of them reaching the Tigris near its confluence with the Greater Zab at Nimrnjd, the other at Qal‘at Sharjat.137 The armies of Oriens and Armenia that operated with Khusrau II to suppress the rebellion of BahrƗm in the summer of 591 must have made use of these latter routes. Fortresses and Manoeuvre The small towns and desert fortresses of the Euphrates steppe had to be kept in good repair as part of the process of operational manoeuvre. They provided a framework of installations that could be used by individual cavalry squadrons as points d’appui for manoeuvre behind the frontier. They had the multiple functions of serving as refuges for the rural population, protective sites for Roman cavalry squadrons who found themselves outnumbered by advancing Persian forces or raiding Lakhmid Arabs, and refuges for defeated Roman forces.138 The relative density of these installations varied

131 It is unknown if the Sasanid main body took the bait and marched southward. 132 The three march stages taken by the Kardarigan’s army suggest that the Persians marched south as far as Bebasa, in the vicinity of Tell Cahker Bazar, then proceeded diagonally toward the northwest in the direction of Apadna, a distance of as many as 50 km., finding very little water along the route. Dilleman 1962: 293 (Fig. XXXVIII), 295. See below. 133 This is evident from Dilleman 1962: 55, Fig. VI.

134

Clausewitz 1976: 119-121. See also Heuser 2002: 88f. Clausewitz 1976: 100-112. Heuser 2002: 72-74. 136 Dillemann 1962: Fig. XXIX (p. 213), Feuille X (facing p. 135). Lightfoot 1983. 137 Dillemann 1962 Fig. XVIII (p. 149). Cf. Poidebard 1939: Planche CXL. 138 The larger fortresses were of course the focal points of the Roman civil administration. On functions, see Gregory 1997: 70-99. 135

333

THE LATE ROMAN ARMY IN THE EAST throughout the region, their occupation being largely dependent on the presence of water supplies during the arid summers. Particularly large concentrations of these kastra existed in the Jabal Hass district of Syria I, and along the Euphrates and Khabur basins.139 Fortresses elsewhere were dependent on springs for their water supplies, or made use of barrages to accumulate standing water, which might then be fed to the fortress by a system of underground or surface channels.140 A good example of this is seen in the patrician John son of Domentiolus’ supervision of the refurbishment of the water supply (viz. cisterns) and conduits (Üdqowe²a) at Dara in 567.141 There was a particularly large set of installations at Khan al-Manqnjra near Jabal RawƗd ca. 100 km. northeast of Damascus, on the route to Palmyra, and thereafter to Sura and Circesium on the Euphrates.142 Numerous rural cisterns and artificial wells (vqÈata) are known to have existed in the parts around Dara and Mardin, some of them regular installations constructed with large limestone blocks.143 There is not much in the way of epigraphic and literary evidence to corroborate the functions of small fortresses directly, but what survives is quite revealing.

These fortresses seem to have been located on the lower slopes of Tnjr ‘AbdƯn facing Persia,146 like Matzaron, where ‘the men under Theodore and Andrew restored Matzaron (this was also a fort) which had become antiquated in the course of time’.147 The walls of Phathakon and Alaleisos had been damaged in the course of Persian attacks against them, perhaps dating from the Persian operations during siege of Dara in 573.148 Philippicus’ troops were demoralised and ‘beaten’, after losing many of their comrades during a retreat that initially was took place in chaos over rough terrain.149 It is doubtful that they served as permanent garrisons in Phathakon and Alaleisos: their task was more probably a temporary one, to rebuild the fortresses’ walls and cover the approaches to Tnjr ‘AbdƯn in case the Persians raided Roman Mespotamia. Philippicus felt he could safely put his broken and de-horsed cavalry — probably provincial arithmoi of the dux Mespotamiae — to work on practical tasks, counting on them to stay in their positions behind partly demolished walls lying on high ground until they recovered from the depressing effects of the retreat. They will have had an important role to play in future operations, once rested and brought up to strength.

The raids directed into Arzanene after the battle of the Solachon plain in 586 were not uniformly successful. During the retirement phase Philippicus’ troops came under pressure from Persian troops defending the province, when a sudden panic affected the army after the siege of Chlomoron was abandoned.144 Even the existence of the Roman-controlled kastron of Aphumon on the Redwan river could not stop the rout, and the panic did not fully evaporate until after the army had crossed the Tigris:145

A contrasting example of fortified towns and kastra suggesting their importance and principal functions is seen in an inscription from the citadel gate of AnasarthaTheodoropolis in the Jabal Hass district in Syria I plausibly dated to 594/5, a few years after the conclusion of the war: 150 (Cross) Jesus. Emmanuel. (Cross) [---] God of all. [A city] which scorned the inroad of barbarians by the gifts of empire now sets up on its gates [the names of] its benefactors: the saviour Christ, the victorious sovereigns, the all-praiseworthy [phylarch of the Ghassanids?], the Praetorian Prefects, [---] its most holy bishop, [---] the most glorious engineer, in the month of Gorpiaios of the year 6 [of the Seleucid era] in the thirteenth indiction. (Cross)

On the following day [Philippicus] reached Amida, after losing much of his force, since the Persians were harrying the rear of the Roman force. The general, whose hopes had been dashed, restored the forts below Mount Izala [Tnjr ‘AbdƯn], and furthermore repaired the old forts which had been neglected and had broken down with the passage of time and war; one of these forts was called Phathakon, the other Alaleisos, and he stationed garrisons in them, organising the control of Mount Izala from there.

The inscription suggests that an enemy army besieged the town on at least one occasion between ca. 574-591, perhaps during Adarmahan’s advance against Antioch in late 573, and perhaps again thereafter at the hands of the Lakhmid Arabs of al-Hira who were allies of the Sasanid regime. Roman commanders were at least in theory interested in capturing or building small forts in enemy territory, sometimes for the purpose of isolating large fortified

139 Poidebard 1939: passim. Mouterde and Poidebard 1945: passim. Monchambert 1984. Fuller 1988-89: 279-290. 140 Calvet & Geyer 1992. 141 Menander Protector, Fr. 9 (Blockley 1985: 96-99). Ioannes 81, PLRE IIIA: 672-674. 142 Poidebard 1939: Planches XXIII-XXV. Gregory 1997: 213-216. 143 Simocatta, Hist. II 5. 6 (Whitby and Whitby 1986: 49). It would be useful to know more about the archaeology of wells in Osrhoene and Roman Mesopotmia. Cf. Dillemann 1962: 62f, 172, etc. Much of our knowledge of this comes from sources of the Islamic period. 144 Simocatta, Hist. II 9. 1-16. 145 Simocatta, Hist. II 9. 16-17 (with minor adaptations from Whitby and Whitby 1986: 56).

146

I deduce this from Dillemann 1962: 230f. Simocatta, Hist. II 18. 7 (Whitby and Whitby 1986: 69 and note 47). For a different interpretation, see Whitby 1988: 283, who puts the repair of the forts in a different locality and has fewer doubts about the effectiveness of these troops. 148 See previous note. 149 Simocatta, Hist. II 9. 5-6. 150 The editors of IGLS 288 favour a date of 594/5. See also Prentice 1908 (AAES III): no. 318. For photos of the site, see Mouterde and Poidebard 1945: Plates XXXI-XXXII. 147

334

FRANK R. TROMBLEY: THE OPERATIONAL METHODS OF THE LATE ROMAN ARMY IN THE PERSIAN WAR OF 572-591 towns. This has already been seen in the case of Nisibis. It is difficult to say whether this was an established operational principle or one dictated by opportunism and circumstance. During the investment of the Sasanid town of Chlomoron in Arzanene in 586 in the exploitation phase after the successful battle of the Solachon plain, Philippicus entertained a proposal from two Christian Armenian grandees:151

Defensive operations in Cappadocia and Armenia I-II in 576 Khusrau I’s campaign in eastern Asia Minor was an extension of operations into Armenia after the truce negotiated in 575, which banned further operations in Syria and Euphratesia for three years.152 These provinces had been depopulated, and in consequence Tiberius Constantine issued a law ca. April 575 granting a twentyfive percent remission on taxes in specie to the provincials of the empire, with certain additional remissions to the people of Osrhoene, Mesopotamia and the Black Sea dependencies.153 These measures extended as far south as Roman Arabia, where a papyrus of Petra indicates a twenty-five percent reduction in the annual indiction.154 Khusrau took the decision to lead the army himself.155 He advanced westward past Dvin-Karin, the administrative capital of the Sasanid province of Armin (Persarmenia), reaching the headwaters of the Araxes river.156 After crossing the Araxes-Euphrates portage he besieged Erzerum-Theodosiopolis (present-day Erzurum), but failed to take the city.157 After this, Khusrau followed the network of Roman highways leading in the direction of Caesarea, provincial capital of Cappadocia I and location of an important fabrica for Roman cavalry equipment.158 Caesarea was also a station on the main route across Asia Minor ending at Samosata on the Euphrates along which large contingents of new troops, including the Langobardi and other western barbarians being recruited by Caesar Tiberius Constantine, were likely to have begun reaching Oriens that year.159 These troops would have been designated the Tiberianoi along with their ethnic name (e.g. Skythai Tiberianoi), a system of unit nomenclature that dated from the time of Justinian.160

‘General, if you wish the Romans to get control of Arzanene, we will show you some places that are particularly impregnable because they are supported by their strong position. The course of good counsel enjoins that forts be built’. … Their words were not inconsistent with the general’s intentions, for the director of the Roman camp was eager to acquire places of this very type, but had meantime out of uncertainty restrained the enterprise by delaying the investigation. So… he dispatched his second-incommand Heraclius, and also sent the guides with him to the strong point. The reconnaissance was thus entrusted to a senior officer and about twenty men who could be counted on to estimate the suitability of the reported places for erecting fortifications, maintaining troops and assisting the blockade of Chlomoron, including the availability of stone and timber, water sources and its position relative to the Roman camp. The sources contain two extended narratives about Roman military operations, one of them in the ecclesiastical history of John of Ephesus, the other in Theophylact Simocatta’s Oecumenical History. It is necessary to trace the course of these expeditions in some detail, to get some understanding of how the Roman military planners managed their assets in manpower and equipment, supplied the army’s needs once it was in movement and how they overcame the technological and environmental effects of ‘friction’ to achieve a successful outcome for the strategic aims of the regime in Constantinople. It must be remembered that planning operations lay primarily in the hands of the magister militum per Orientem, in those of his senior subordinate, the lieutenant-general, who was often commander the offensive wing of the field army, the right meros or tourma, and to some extent in those of his junior colleague, the magister militum per Armeniam. The latter relationship was fraught with rivalries because, as it seems, the commander of the Armenian troops was constantly seeking to raise the reputation of his command organisation vis-à-vis that of Oriens, and because his command may well have consisted of large numbers of native troops who had an inferior ‘tone’ compared to those of Oriens.

The aims of Khusrau’s expedition are difficult to determine when considered in light of the practical difficulties of operating along the Araxes-Euphrates route. Whether or not Khusrau’s advance had a preemptive aim, he may well have been motivated by a pragmatic desire to reclaim the prestige his regime had lost in Armenia after the rebellion of 571.161 If Khusrau 152

Whitby 1988: 259-261. For a translation of the law, see Appendix II below. Frösén, Arjava and Lehtinen 2002 (P.Petra I): no. 10. 155 See Whitby 1994. 156 N. Garsoian, Duin, in Kazhdan 1991: 665f. 157 N. Garsoian, Theodosioupolis, in Kazhdan 1991: 2054. 158 Notitia Dignitatum Or. XI 26. R. P. Harper, Caesarea, in Stillwell 1976: 182. 159 Ps.-Joshua, Chronicle § 57 (Trombley and Watt 2000: note 332). On the fortifications of Samosata: Gregory 1997: 124-127. 160 Five cavalry squadrons were recruited from the Germanic Vandals and sent to Oriens in the 530s. They were known as the Ioustinianoi Bandiloi. De Bellis II 14. 17-18. A numerus known as the felices Persoarminii is reported in Italy. Grosse 1920: 278. 161 Whitby 1988: 264. Khusrau I’s predilection for pre-emptive attacks and operational manoeuvre is evident in his decision to break the threeyear truce of 575 several days before its expiry, sometime between June-August 578. Menander Protector, Fr. 23.1 (Blockley 1985: 198f. and notes 200, 271). He may also have challenged traditional Sasanid taboos about the start of the campaigning season; for the Persians it 153 154

151

Simocatta, Hist. II 7. 9-11 (adapted from Whitby and Whitby 1986: 52 and notes).

335

THE LATE ROMAN ARMY IN THE EAST really believed that the Roman armies were in no condition to resist invasion, he may initially have planned the operation of 576 as an opportunistic, limited liability raid. The early date of the advance corroborates this, for the Persian campaigning season did not normally begin till after mid-summer.162 Khusrau’s decision to advance into Roman Armenia after his failure to capture ErzerumTheodosiopolis meant, however, that should he suffer a reverse, his communications would be threatened in any retreat. This is precisely how the operation turned out, suggesting once again that the Persian king was relying on surprise to catch the Romans with their defences down. These events took place in the late winter and early spring of 576.

version of events and in John of Ephesus’ statements, which reflect the views of middle-ranking imperial bureaucrats. There seems to have been endemic ill feeling between the magistri militum of the two commands, possibly because of the need to operate defensively and protect the infrastructure of towns and fortresses along both sides of the boundary dividing their respective jurisdictions. In summarising operations up to the battle of Melitene, John of Ephesus observes:167 And had it not been for the envy and divisions which existed among the Roman generals, and prevented their acting in unison, they might utterly have destroyed both [Khusrau] and his army: for all that was necessary for this was that they should arrange for a common plan of operations, and with their several divisions surround him. As it was, the Persians advanced against Justinian the patrician, son of Germanus, and Justinian was afraid and fled the encounter; and his fellow generals did nor join him or come to his support. And when the king and his army saw this, they took courage and were emboldened to attack and set fire to the city of Melitene.

Roman observers thought the Sasanid army quite large, and it seems possible that Khusrau could have drawn sufficient supplies along the route of the expedition.163 He undoubtedly waited till there was sufficient grazing for cavalry horses, pack animals and other beasts before beginning the advance. He is furthermore likely to have endeavoured to get the jump on the Roman defences before Justinian, magister militum per Orientem, could draw upon the arithmoi of Osrhoene, Mesopotamia, Syria, Euphratesia and Phoenice Libanensis then in winter quarters and concentrate them into a field army. It is therefore quite possible to suggest that Khusrau began his concentration and then march into Armenia even before the spring thaw.164

It is also possible that Justinian found it difficult to persuade the dux Euphratensis et Syriae and dux Phoenicae Libanensis to respond to the invasion with all their forces. The duces, fearing imperial censure if the truce of 575 proved to be a ruse and the provinces under their jurisdiction were raided once again, may have insisted on keeping back some or all of their forces.

Justinian would have begun his march westward to cut off Khusrau’s advance once the latter was known to be moving in the direction of Caesarea.165 It is unknown how much of the army of Oriens Justinian had him, but his presence near Amida suggests that he could at least have drawn on the provincial moirai of Osrhoene, Mesopotamia and Syria-Euphratesia, plus the organic troops of his command, who had their winter quarters in Antioch.166 The provincial cavalry of Armenia I and II would undoubtedly have remained near their winter quarters to protect the towns and fortresses of western Armenia pending the arrival of the army of Oriens. The magister militum per Armeniam plays a peculiarly silent role both in Theophylact Simocatta’s semi-official

Whatever the difficulties of co-operation between the two commands, the defensive operations were an unqualified success. The road network of eastern Anatolia passes through numerous defiles.168 No matter which way the Sasanid army probed along these routes, Roman detachments, perhaps at times the entire force of a provincial moira or droungos, blocked the routes, evidently skirmishing with the Sasanid cavalry when it became necessary and taking advantage of the natural defensive features of the terrain like rivers fed by melting snow and using hilltop refuges as positions from which to attack the rear of passing Sasanid formations. The instructions given for defensive operations against enemy incursions in the Strategikon of Maurice (ca. 600) are consistent with this and are perhaps based in part on the lessons of the campaign of 576, involving manoeuvre and concentration:169

normally began in August or September, although spring attacks were always a possibility. Inostracev 1926: 32, note 1. Cf. Khusrau I’s surprise attack on Euphratesia and Syria I and II in March 540, which stuck to the west bank of the Euphrates, thus avoiding the more usual invasion route via Nisibis into Mesopotamia and Osrhoene. Procopius, De Bellis II 5. 1. Of the Nisibis route Procopius observes: ‘[Until 531] [the Persians had] always staged their invasions from here (ÑmhÈmde)’. De Bellis I 17. 25. See also Procopius, De Bellis I 23. 15. 162 Inostrancev 1926: 32, note 1 (without references). 163 John of Ephesus, HE VI (Payne Smith 1860: 392), who puts the size of the army at 120,000. It was certainly possible to move large armies along the Araxes, to judge from the Roman campaign there in 542/3. Trombley 2005: 404-411. 164 Whitby 1988: 264. 165 For sites and maps of the theatre of operations, see Ramsay 1890: map at pp. 266-267. Hild and Restle 1981: large folding maps. 166 The praetorium of the magister militum per Orientem was located inside the fortifications of Antioch. Malalas, Chronographia XVII 14.

If an enemy force attacks our territory, if it is superior in strength or even equal to ours, particularly at the beginning of the invasion, one must be sure not to engage in a pitched battle. One should instead carefully set ambushes by day or night, block his route of march (tÀr ÛdoipoqeÊar ÑlvqÇsseim), 167

John of Ephesus, HE VI 8 (adapted from Payne Smith 1860: 395). These are fully described in Anderson 1897. 169 Maurice, Strategikon X 2, lines 1-8, 15-18. 168

336

FRANK R. TROMBLEY: THE OPERATIONAL METHODS OF THE LATE ROMAN ARMY IN THE PERSIAN WAR OF 572-591 seize strong points beforehand [and] destroy supplies ahead of him. If it seems right, one should attack as the enemy is turning about and leaving our territory… One must above all keep the army intact and unbeaten. By keeping to this principle, it will not be easy for the enemy to besiege the fortifications or disperse his forces to plunder the countryside, for he will be under observation and will suspect that [our] army will concentrate (sumestÇmai) [against him].

(Helenopontus) and other towns, including the population of their territories, migrated westward, often following the valleys of the Lycus and Iris rivers, in the direction of Amasea, the metropolis of Helenopontus and best defended town of the region.173 These folk were fed from the grain reserves of Eutychius’ monastery there.174 It can be said without doubt that one of Justinian’s most important operational aims was the protection of the urban and rural population in the interior of Asia Minor from Sasanid captive hunting, to prevent them suffering the same disaster that befell Mesopotamia and Syria I and II in 573.

The insubstantial nature of the battle descriptions in the sources for the campaign of 576 is suggestive of small engagements. It is also possible that the mountainous terrain made it difficult for the Roman commander to supply concentrations of troops larger than a provincial moira. It is clear that the separate Roman forces, operating under Justinian’s and possibly his Armenian colleague’s instructions, engaged in blocking tactics in the defiles. John of Ephesus somewhat repetitive language is suggestive of this:170

The Sasanid army’s thrust in the direction of Amasea was a gesture of despair in the midst of the Roman forces building up on the road network along its flanks:175 And when [Khusrau] saw that [the Roman armies] had surrounded him, and were pressing upon him from all sides, he was compelled to flee in haste to the mountains, leaving behind him his camp, and all the equipage… [There follows an inventory of the contents of the camp.]

When [the Roman forces] met him, and drew up in order for battle, [Khusrau] was alarmed, and would not give [battle], but marched toward another city. And [the Roman forces] also hastened there, and threw themselves in his way, and repulsed him from there also. And as they had made trial of his army, their own courage grew, and they despised him. And as he saw that matters were not advancing according to his wish, he marched towards the mountains on the northern frontier [of Mesopotamia?] towards Cappadocia, with the intention of attacking Caesarea. But when the Roman armies saw this, they also marched there, and arrived before him, and posted themselves in his way, and met him in the mountains of Cappadocia, and stopped his further progress, and would not let him pass over.

Khusrau therefore retired in the direction of Melitene, the provincial capital of Armenia II, which the Sasanids burned down, presumably after its population had been evacuated ahead of the enemy advance.176 An engagement was fought there according to both Theophylact Simocatta and John of Ephesus.177 If the Sasanid force was as close to disintegration as John of Ephesus suggests, it is likely that Justinian would have employed stand-off tactics involving the use of archery, and have resorted to shock action only as a last resort. His operational aim was to escort Khusrau’s army out of Roman territory in such a way as to minimise damage to the provincial infrastructure. If this could be done by avoiding a general engagement, so much the better for Roman interests. John of Ephesus drew the basic details of the engagement from an account given by the official interpreters, who would have been soldiers (possibly protectores) accompanying the Roman army from the staff of the magister officiorum.178 The engagement and

It is evident that the Roman forces’ hard marching to seal off the Sasanids’ routes of advance improved their morale and toughness and, as John of Ephesus indicates, raised doubts in Khusrau’s mind about the soundness of the operational plan. The historian also indicates that it was the Zoroastrian priests in the imperial camp who advised the king to shift his direction of advance toward the northwest, enabling him to capture Sebastea,171 and adds: ‘[Khusrau] could take neither booty nor captives, because the whole land had fled before him’.172

173 Komana, Hild and Restle 1981: 208f. R. P. Harper, Nikopolis, in Stillwell 1976: 626. D. R. Wilson, Amaseia, Neocaesarea, and Zela, in Stillwell 1976: 47, 620, 999. 174 Eustratius, Vita S. Eutychii: 2344B-D. For the institutional context, see Trombley 1985: 53f. For the extent of the evacuation, see maps listed in note 165 above. 175 John of Ephesus, HE VI 8 (adapted from Payne Smith 1869: 393). 176 Melitene, Hild and Restle 1981: 233-237. Gregory 1997: 49-53. 177 The difficulty with Simocatta’s account of the battle is that it reads more like a ‘battle piece’ than a report about an actual battle, his rendition of Justinian’s pre-battle speech being longer than his account of the actual engagement. The main characteristic of the ‘battle piece’ is the oversimplification of human behaviour through the use of ‘collective images’. Keegan 1976: 36-46, esp. 39f. 178 See Menander Protector, Fragment 6.1 (Blockley 1985: 76f.), where I infer that the interpreters were immediate subordinates of Peter Patricius the magister officiorum.

This observation in strikingly confirmed in another wellinformed and near contemporary account of the campaign of 576, that of the biographer of Eutychius patriarch of Constantinople, who had founded a monastery on the estates he owned in the Helenopontus. Large numbers of residents from Nikopolis (Armenia I), Neocaesarea (Pontus Polemoniacus), Zela and Komana 170

John of Ephesus, HE VI 8 (adapted from Payne Smith 1860: 392). Sebasteia, Hild and Restle 1981: 274-276. 172 John of Ephesus, HE VI 8 (Payne Smith 1860: 393). 171

337

THE LATE ROMAN ARMY IN THE EAST subsequent exploitation of the Roman victory took this course:179

during its retirement into Arzanene so as to permit conducting raids against what Theophylact Simocatta calls ‘the interior of Babylonia’ (viz. Beth ‘Arabaye, Beth ‘Aramaye and Asuristan) and to winter there, re-entering Roman territory in the spring of 577.183

1.

The opposing armies drew up opposite each other on the plain east of Melitene ‘at some distance from it’. 2. The armies approached each other closely, but stood in their ranks ‘from morning until the sixth hour’. 3. In the early afternoon, the three Roman interpreters rode out defiantly between the armies, spurring their horses and wheeling, all this a high speed. Their aim was evidently to ascertain Khusrau’s intentions. 4. After another delay, Khusrau is reported to have sent a message indicating: ‘There can be no battle today, for the time has passed’. 5. During the night the Sasanid army retreated to the Euphrates, which was six miles to its rear. 6. Upon realising what was happening, Romans attacked the retiring Persians — it is not clear whether on the same night or the following morning — and destroyed about half the enemy’s troops, all of whom appear to have been cavalry. 7. Khusrau retired into Sasanid Arzanene, possibly crossing the Tigris bridges near Amida. 8. The Sasanid king ordered that all Roman villages along the retirement route be burned down. 9. Khusrau led the remnants of his forces into the Karch mountains. 10. Justinian led the Roman forces into Beth ‘Arabaye and Khuzistan, raiding to within a hundred Roman miles of Ctesiphon.

Operational manoeuvre in the expedition to lake Urmiah in June-July 591 The longest operational march of the Persian War took place in the spring and summer of 591. This expedition had a clear political purpose, to overthrow the rebel BahrƗm and restore Khusrau II AbarwƯz to the Persian throne.184 The political circumstances were complex; some of Maurice’s advisers opposed the intervention, recommending instead that it better served the Roman interests for the Sasanid state to sink into civil war. This operation brought together perhaps the largest number of Roman troops since the campaign against Dvin in Sasanid Armenia in sixth indiction (1 September 542 to 31 August 543), which had involved all the troops of the magistri militum per Orientem and per Armeniam, and possibly some praesental troops.185 The same two commands participated in the campaign of 591. It was also an allied operation, involving the Persian forces that Khusrau II had been able to pull together from the Nisibis military district during his stay inside the frontiers of the empire. Narses, magister militum per Orientem, was commander of the senior Roman formation and held overall command, with John Mystacon magister militum per Armeniam and his troops at his disposal.186 The armies of Oriens and Armenia may hypothetically have numbered as many as 20-30,000 in all.187 In practice, however, the cavalry squadrons must have been under strength because of casualties, sickness and lack of combat training and experience among the new recruits. A fair number of troops may have been left behind to police the provinces that their duces were charged with protecting. It is unknown how many arithmoi were disposed by the magister militum of Armenia. A cautious

John of Ephesus attributes Khusrau’s caution on the day of battle to the demoralisation of the Sasanid troops. It is more probable, however, that Khusrau was following conventional operational doctrine in awaiting the Roman attack. The Sasanid tactical manual quoted in the ƖyinnƗme observes:180 And let the captain take care that the rays of the sun and wind should be from the rear of his army. And let him not enter into action with the army unless in case of extreme necessity and being in such a position when the battle cannot be avoided. And if that happens, let the captain try protract the battle to the end of the day.

183

Simocatta, Hist. III 15. 1-2. Bahram 2 (ChobƯn), and Chosroes II Parwez, PLRE IIIA: 166f., 306308. Riedlberger 1998. (non vidi). Frendo 1989 (non vidi). On the routes taken during the campaign, see Minorsky 1944: 244-248. 185 On this, Trombley 2005: 406. It has been suggested that the campaign in fact took place in the late autumn or early winter of sixth indiction (542/3). Kislinger and Stathakopoulos 1999. In my view the assigned date might still remain in 543, but this position cannot be sustained without further analysis of the chronology of the events surrounding the impact of the bubonic plague on the morale and battleworthiness of the Persian army in 543, and of the administrative complications following on Belisarius’ dismissal as magister militum of Oriens and his replacement by Martinus. Unfortunately Kislinger and Stathakopoulos did not include these issues in their analysis. 186 Simocatta, Hist. V 5. 3. Mystacon served as magister militum per Armeniam in 589-91. Johannes qui et Mystacon, PLRE IIIA: 679-81. Narses served as magister militum per Orientem 591-603. Narses 10, PLRE IIIB: 933-935. 187 The magister militum of Oriens could have drawn on as many as ca. 60 arithmoi in this phase of the war, which at 300 men per squadron would have amounted to ca. 18,000 men. See note 15 above. 184

Another text recommends that the engagement should be put off as long as possible; but if it could not be avoided, the fight should take place towards the end of day.181 It may well be that Khusrau conducted the battle of Melitene ‘by the book,’ even as regards aligning his battle formation with the rising sun to its rear, for the army was in enemy territory and the level of risk was therefore much greater.182 Justinian’s force had sufficiently destroyed the cohesion of Khusrau’s army 179

John of Ephesus, HE VI 9-10 (Payne Smith 1960: 396-99). Inostrancev 1926: 14. 181 Inostrancev 1926: 29. 182 This principle is acknowledged at Maurice, Strategikon X 2, lines 814. 180

338

FRANK R. TROMBLEY: THE OPERATIONAL METHODS OF THE LATE ROMAN ARMY IN THE PERSIAN WAR OF 572-591 estimate of their numbers would be consistent with a Sasanid tradition embedded in a later Arabic source:188

dÇmeisla), dispatched it to Maurice and received abundant financial wealth; this he divided up among the attendant Persian forces.

An unheard of present was offered by AbarwƯz to the king of Rnjm on account of the struggle against BahrƗm ChnjbƯn… [In response, Maurice] assisted him, sending twenty thousand horsemen, armed from head to foot, two million dinars to be paid to the troops, etc.

It was a sensible policy for the Roman commanders to let Persian troops, mercenaries and ethnic allies do most of the hard fighting, as will be seen. Although the Roman army was a large one, its main function may have been as a security and support force, to help Khusrau keep his heterogeneous and recently recruited troops in a concentrated body during the march phase of what was to prove a deep penetration of the Zagros mountains, in case any pitched battles arose and particularly if BahrƗm succeeded in achieving surprise.

There is no easy way of estimating the number of Sasanid troops participating, some of whom belonged to the cavalry levy of Nisibis and some of whom may have been ethnic allies. The Persian cavalry levy in previous operations of the war — probably drawn from the military districts adjoining Roman Mesopotamia — was around 12-15,000 men, and this may not be far off the number of troops Khusrau II raised in 591. Eighteen years earlier, in 573, Khusrau I is said to have brought 23,000 cavalry to the siege of Dara. A reduction in the numbers of the Sasanid cavalry levy — if the figures are accurate — may be accounted to wartime attrition (casualties and disease) and the fact that a sizeable proportion of the available horsemen were partisans of BahrƗm.189 Theophylact Simocatta’s estimate of 60,000 men for the size of the combined Roman-Sasanid army must therefore be something of an exaggeration, but perhaps not an unreasonable one.190 Comentiolus, the previous commander of the troops of Oriens, was replaced shortly before the expedition after Khusrau accused him of delays, and in consequence of the king not having sufficient time to recruit allies.191 There may be something to the complaint, for Maurice extended a sizeable loan to the Sasanid king for this operation:192

The assembly of the two Roman armies undoubtedly took time, from the standpoint of marching the cavalry squadrons from their towns and fortresses to the concentration point, and of collecting supplies and transporting them to supply dumps on the roads leading to ‘AmmudƯn on the frontier opposite Nisibis. A senior Persian officer named Sarames was also active in this task in the territories now controlled by Khusrau in Sasanid Arbayistan.193 He seems to have deposited supplies along the old Roman military highway running eastward from Nisibis to Singara and on the west bank Tigris near its confluence with the Greater Zab river.194 The expedition did not begin its advance into Persian territory until the wheat harvest began, most probably in late June and July.195 The arithmoi of Oriens under Narses probably assembled at their usual point of concentration at MonocartonTiberiopolis south of Mt. Aisouma in Osrhoene. Theophylact Simocatta gives the stages of their subsequent advance. Narses’ force, accompanied by Dometianus metropolitan of Melitene, who was acting as Maurice’s political agent with plenipotentiary powers, took the Mardin-Dara-‘Ammudin highway to the frontier, repossessing Dara en route from the Persian garrison occupying it.196 At one point, Dometianus threatened to cancel the expedition and order the Roman troops to retire to Tella-Constantina, because of Khusrau’s presumptuous behaviour in occupying the episcopal church and residence in Dara with his retinue. The completion of the operation was therefore probably not a foregone conclusion in the minds of Maurice and his advisers, being contingent on Khusrau’s good conduct.

[Khusrau, waiting at Constantina] asked [emperor Maurice] to put a stop to delay among the allied forces and to assist with all speed, and that he be provided by the emperor with a financial endowment, guaranteeing to return the loan to the Romans once he had recovered his own kingdom. The Roman emperor bestowed on him the massive sum of money in addition to the military alliance. Then Khusrau wrote a receipt for the loan (weiqocqavÉsar tÄ 188

Abnj ‘UthmƗn ‘Amr ibn Bhar al-Jahiz al-BasrƯ, KitƗb al-mahƗsin wa’l-addƗd, in G. van Vloten (ed.), Le livre des beautés et des antithèses (Leiden 1898), 369-371, cited by Inostrancev 1926: 45f. 189 The figure comes from the Chronicle of 1234. Greatrex and Lieu 2002: 147. Theophanes of Byzantium’s figure of 40,000 cavalry at the siege of Dara is surely nothing more than a wild estimate. Theophanes of Byzantium, Fr. 4 (Dindorf 1870: 448, lines 29-41). 190 For a more optimistic view of this figure, see Howard-Johnston 1995: 167. 191 Simocatta, Hist. V 2. 7. It is reported that John Mystacon, magister militum. per Armeniam, jeered at Khusrau II’s tactical impulsiveness during the summer campaign of 591. Simocatta, Hist. V 15 .4. Comentiolus 1, PLRE IIIA: 321-325. This is undoubtedly a reflection of the Roman army’s increasingly effective operational and tactical methods that began with the training programme imposed by Maurice after his appointment as magister militum per Orientem in 578. 192 Simocatta, Hist. V 2. 5-6 (adapted from Whitby-Whitby 1986: 134). On the links between the two monarchs, see Goubert 1949.

Khusrau’s retinue of guardsmen was undoubtedly augmented with regular Persian troops from Nisibis after its populace acclaimed him king. At other stages of the expedition, recruiting officers were sent out in advance to announce that the king was coming and had a large pay chest. Khusrau’s personal bodyguard of about 1,000 men 193

Simocatta, Hist. V 5. 4. Sarames 2, PLRE IIIB: 1113. Simocatta, Hist. V 5. 4. 195 ‘Early summer’. Simocatta, Hist. V 4. 3. 196 The city was handed over with a grand gesture. Simocatta, Hist. V 3. 10-11. 194

339

THE LATE ROMAN ARMY IN THE EAST 541.204 The rapid advance to the Tigris also suggests that the army consisted mainly of cavalry. Here Narses and Khusrau waited for the arithmoi of John Mystacon, magister militum per Armeniam. The route that the latter took is unknown. Some think he took a route through Azerbaijan, passing round the northern edge of the Zagros mountains. One wonders in light of the Narses’ halting on the Tigris whether, after their initial concentration, the troops of Armenia could have followed a route near the east bank of the Tigris round Tnjr ‘AbdƯn to join the main army, accompanied by such ship-borne supply transport as was needed and by ethnic allies from Armenia and the Caucasus. Areobindus seems to have taken this route in a march round Tnjr ‘AbdƯn from Arzanene to the territory of Nisibis in July 504;205 the existence of a Roman military highway running along the east bank of the Tigris has been surmised.206 Theophylact Simocatta explicitly states that Narses actually delayed on the west bank of the Tigris while waiting for John Mysticon and the troops of Armenia to come up.207 This suggests that Mystacon was taking a route near that river and not through Atropatene.

was made up of Roman personnel, possibly hired on the same principle as bucellarii, because he is said to have distrusted his own troops.197 Theophylact’s narrative gives little information about the arrival of the ethnically Persian contingents; we are told that Khusrau dispatched 2,000 troops under the command of the MƗhbǀdh — undoubtedly cavalry — to advance by way of Singara on the royal cities of Seleukeia and Ctesiphon and seize the royal palaces.198 There the MƗhbǀdh was probably hoping to find personnel for Khusrau’s army, coined silver and bullion, weapons and agricultural supplies.199 This is apparent from Theophylact’s report on the difficulties that confronted the MƗhbǀdh when he reached to vicinity of Seleukeia and Ctesiphon:200 On approaching the royal cities he sent the overseer of the royal treasury a message that he should prepare a great supply of food and set aside a suitable grant of money, since the allied forces were converging on him in large numbers… And so the man invested with the care of the treasury showed the MƗhbǀdh’s letter to the comptroller of the royal treasures, who had been left by BahrƗm at the capital... [In] the middle of the night [the comptroller] collected his soldiers, left Seleukeia and went to Ctesiphon… After occupying Ctesiphon, the MƗhbǀdh took charge of the royal treasures…

The first engagement, a battle of reconnaissance detachments, took place immediately, well before the arithmoi of Armenia had begun to come up. Khusrau is said to have sent 1,000 men from his personal bodyguard — undoubtedly his Roman-recruited bucellarii — across the Tigris to find the enemy main body and determine its strength. The river was fordable at this time, and would therefore have failed to provide a sufficient barrier if the enemy was close at hand and present in large numbers.208 Khusrau’s troops destroyed a Persian reconnaissance force led by a certain Bryzakios in a night attack on their camp.209 Khusrau evidently gained information from captives about BahrƗm’s plans and persuaded Narses to advance at once, for with the destruction of Bryzacius’ force the enemy was potentially unaware of the Roman army’s presence.210 In achieving this success Khusrau was broadly following a set of operational and tactical principles set forth in the Sasanid tactical manual whose fragments are embedded in the epic ƖyƯn-nƗmeh, but here of course the executors were the Roman-Sasanid troops of his special guard:211

After Khusrau’s acclamation as king by the officials and people of Nisibis, Narses began to move the arithmoi of Oriens forward to Dara.201 The advance of the army of Oriens into Persian territory began in late June or early July after a complimentary speech by Dometianus which concluded with the advice that Christ would be the soldiers’ principal commander (ÐqwistqatgcËr). This was the precursor to a type of religious persuasion that became common in the wars of the seventh century.202 Narses’ troops first camped in the territory of Nisibis where the Mygdon river provided an abundant supply of water, then advanced quickly to the Tigris, passing along the old Roman military highway through the territory of Singara, covering a distance of some 150 km. and reaching the river in three days.203 The motive for this swift movement may have been concern about the effects of the extraordinarily hot summer weather typical of Mesopotamia on the Roman troops. It was undoubtedly remembered that this nearly ruined the campaign of

And let the vanguards pass over even places and halt on heights, and not pass any locality without having explored the same minutely. And let the ambuscades 204 Procopius, Wars II 19. 31-32. Desert warfare, in Margiotta 1996: 281-287. 205 Ps.-Joshua, Chronicle § 75 and note 440. 206 Anderson 1897: Plate I (map). Simocatta, Hist. (Whitby and Whitby: 138, note 20). 207 Simocatta, Hist. V 5. 4. 208 Simocatta, Hist. V 5. 6. 209 Byzacius, PLRE IIIA: 251. 210 Bryzacius’ captors mutilated him by cutting off his ears. It is uncertain whether it was he, or other men under his command, who made the disclosures. 211 Inostrancev 1926: 14-16. I have adapted Bogdanov’s translation of Inostrancev’s original Russian text in the interest of clarity.

197

Simocatta, Hist. V 5. 2. 198 Simocatta, Hist. V 6. 7. Mebodes 3, PLRE IIIB: 870. 199 The contents of the Sasanid palaces are known mainly from Theophanes’ account of Heraclius’ expedition which reached Dastagerd in December 627. Theophanes Confessor, Chronographia AM 6118: 322, lines 1-16. 200 Simocatta, Hist. V 6. 1-10 (adapted from Whitby-Whitby 1986: 139f.). 201 Simocatta, Hist. V 3. 3. 202 Simocatta, Hist. V 4. 15. E.g. Kitzinger 1954: 111f. Kaegi 2003: 113f. 203 Simocatta, Hist. V 5. 4.

340

FRANK R. TROMBLEY: THE OPERATIONAL METHODS OF THE LATE ROMAN ARMY IN THE PERSIAN WAR OF 572-591 Zab.216 The commander of the right meros or tourma, the attacking wing of the late 6th c. Roman army, was at times designated hypostratƝgos or ‘lieutenant-general’ (ÜpostqatgcËr). As such, he was supposed to be an experienced commander with accurate powers of observation and inference, on whom the magister militum (stqatgkÇtgr) could rely for accurate intelligence;217 further, command of the army might temporarily devolve upon the lieutenant-general if the commanding general were incapacitated or recalled 218 The important tasks given to Comentiolus during this expedition show that Maurice replaced him as magister militum per Orientem at the beginning of the campaign solely to ensure that political co-operation with Khusrau did not break down, and not because of any lack of capability in Comentiolus as a field commander. The Lesser Zab was spanned by several bridges, and the latter’s men seem to have captured one intact.219 It is not exactly clear why capturing a bridge was so important, inasmuch as the water level in the main channel should have been low at this point in the summer season, and in view of the fact that the army had already crossed the Tigris and Greater Zab without the assistance of such installations. Perhaps the bridges were located along regularly used roads where the river banks were less steep, allowing large numbers of troops to cross quickly and thereby minimising the risk of a sudden attack being directed against them while they were still forming on the far bank.220 The seizure of river crossings was an operational task, designed to ensure the rapid and safe movement of a large army and to minimise ‘friction’.221 Another force

be laid behind cover and in unperceived places. And let iron traps [caltrops?] be placed in localities where a night attack is feared. And let the captain guard against news being spread about him, because in the spread of the same [lies] the destruction and ruin of the army… And let those attacking by night avail themselves for the attack [at a time when] the wind is blowing or the gushing of some river nearby is audible, because that is the most convenient [way] for the noise of their [approach] not to be heard, and [let them] avail themselves for the attack of [the hours of] midnight or the hour of greatest darkness… And it must be known that night-attacks are necessary in order to frighten and terrify the enemy; and let them abstain from gathering chattels of the enemy, from catching their mounts and from taking spoils. In this instance the Romano-Sasanid commander found it necessary to take captives for purposes of interrogation, in order to get clear intelligence about the movements of BahrƗm’s army. It is otherwise uncertain to what extent Khusrau’s troops observed these practices, as the terrain may not have necessitated using all of them, but the nocturnal advance against the camp and destruction of the opposing reconnaissance force were fully consistent with the principles outlined in the ƖyƯn-nƗmeh. The Romano-Sasanid army then crossed the Tigris and constructed a fortified camp (tÇvqz peqistoiwÊfomtai) in a place called Dinabandon.212 Khusrau was eager to exploit his minor success in destroying Bryzacius’ reconnaissance force, so it was agreed on the following day that the army should push on and cross the Greater Zab, although John Mystacon and the troops of Armenia had not yet come up.213 There was a clear element of risk in advancing before the armies of the two magistri militum had concentrated, but an opportunity to finish the campaign through rapid exploitation was quite possibly uppermost in the mind of Narses, who had overall command of the Roman forces.214

216

This would have amounted to three or four cavalry arithmoi. Comentiolus is here designated taclatÇqwgr to³ denio³ jÈqator, which Whitby and Whitby translate as ‘brigadier’. Simocatta, Hist. V 8. 1. 217 Heraclius the elder conducted a reconnaissance during the siege of Chlomoron in 586, to observe strong points that might be turned into small forts in order to restrict the movement of Sasanid forces and facilitate the occupation of Arzanene. Heraclius had commanded the centre meros in the preceding battle of the Solachon plain. It had stood on the defensive, enduring a furious Sasanid assault and undoubtedly sustaining many casualties. In consequence, the left meros which had not been so heavily engaged conducted the initial pursuit after the battle. Simocatta, Hist. II 3. 2. During the exploitation phase after the battle, the army entered Arzanene and Heraclius was designated as hypostratƝgos, which Whitby and Whitby translate as ‘second-incommand,’ an accurate rendering. Simocatta, Hist. II 7. 11. 218 The elder Heraclius had performed similar tasks for Philippicus as hypostratƝgos. Simocatta, Hist. II 9. 17—10 . 7 and II 18. 2. 219 This is implied in Simocatta’s observation: ‘But BahrƗm tried to secure control of the next bridge’. Hist. V 8. 2. 220 This particular point requires field reconnaissance. Both permanent bridges constructed of masonry and bridges of boats are known to have existed on the rivers of Mesopotamia. They figure prominently in the emperor Heraclius’ operations there in the 620s. 221 Clausewitz classifies this as an example of die kleine Kriege, or ‘wars of detachments,’ which are the component parts of an operation. On river crossings: ‘A major river that cuts across the line of attack is a great inconvenience to the attacker… [He] is usually limited to a single bridge, so unless he stays close to the river his actions will be severely hampered… [A] river is a substantial factor, for it always weakens and dislocates the offensive’. Clausewitz 1976: 532f. On friction: ‘Countless minor incidents — the kind you can never really foresee — combine to lower the general level of performance [of an army], so that one always falls short of the intended goal… Friction is the only concept that more or less corresponds to the factors that distinguish real war from war on

The final phase of the operation against BahrƗm involved penetrating the Zagros mountains and reached the environs of Lake Urmiah. It took another four days’ march to reach a locality called Alexandriana, and another day after this to reach a district called Chnaithas.215 At this point Narses dispatched Comentiolus, commander of the right division (meros) of the army of Oriens, with ca. 1,000 men — undoubtedly cavalry — to reconnoitre the crossings of the Lesser

212

Simocatta, Hist. V 5. 8. This was presumably done the day after the recce force returned with Bryzacius and other captives. 213 Simocatta, Hist. V 6. 1. 214 It is likely that Narses and John Mystacon were in communication with each other through dispatch riders, an essential feature of campaigning in territory cut up by rivers and mountain ranges. If this was the case, Simocatta does not inform us about it. 215 Alexandriana may well have been Arbela, where Alexander the Great decisively defeated Darius III. Simocatta, Hist. V 7. 10-11.

341

THE LATE ROMAN ARMY IN THE EAST sent out under the command of Rufinus son of Timostratus succeeded in capturing some other crossings in a series of nocturnal forays.222 The rationale behind this lay in the fact that Narses had interrogated enemy scouts who had been taken prisoner and knew that they were observing his movements.223

the magister militum per Armeniam greater freedom to operate in these provinces and presumably to draw supplies. John Mystacon’s reconnaissance troops performed good service, reporting the advance of BahrƗm’s cavalry against them. The immediate outcome was an operational standoff because a river lay between the two forces. It may have been on the Tatavi which pours into lake Urmiah from the south.229 It is difficult to establish the respective positions of the two Roman forces. Did John Mystacon take the forces of Armenia round the eastern side of lake Urmiah and catch BahrƗm in the rear? Or did he make his approach from the northwest? Theophylact Simocatta’s statements about compass directions might be taken to support the former view.230 On the other hand, the relative ease with which the forces of Oriens and Armenia communicated and eventually linked up might be taken as evidence that John Mystacon was moving round the western side of lake Urmiah.231 BahrƗm would then have faced Narses and Khusrau towards the southeast and John Mystacon’s towards the north-west on the west side of the lake. As Theophylact remarks:232

The operational plans of the BahrƗm are central to the present analysis only as far as they bear on the Roman conduct of the expedition. The Sasanid general was well aware of the fact that the magister militum per Armeniam was soon expected in the zone of operations, but seems to have been completely ignorant of the route that the latter was taking, nor is Simocatta entirely clear on this point. In consequence, BahrƗm was hoping to destroy the combined force of Narses’ and Khusrau’s troops before they could unite with those of John Mystacon, for as Theophylact Simocatta observes: ‘[BahrƗm] fostered his hopes with the encouragement that the Roman contingent in Armenia would not be able to join up with the Roman eastern forces on account of the difficulty of the terrain’.224 This description could apply to any number of routes through Sasanid Mesopotamia and Atropatene. BahrƗm was in the end forced to divide his forces:225

So Narses and Khusrau… reached the vicinity of a certain village called Siraganon by its inhabitants. On the fifth day [i.e. the day after crossing the Lesser Zab and the standoff at the Tatavi river], Narses discovered that the Roman reinforcements from Armenia were nearby. For this reason he dispatched couriers to John and advised the general that he must not engage the Persians till the two forces had concentrated (sumaptolÈlai) and made their strength invincible.

But he, now that he had failed in his objective [of blocking the Lesser Zab crossings and keeping the Roman armies of Oriens and Armenia apart], separated his forces towards the north and east in order to cut off [Narses’ and Khusrau’s] concentration (sÌllinir) with the Roman force expected from Armenia… [BahrƗm] had arrived at a certain lake on this route, when it happened that the contingent from Armenia appeared in close proximity.

Whatever the geographical realities, the Roman-Sasanid forces were advancing cautiously and avoiding risks by stopping at river lines, quite possibly out of fear of BahrƗm’s cavalry, and observing the principle of concentration. John Mystacon’s Armenian troops would have had no reason to engage BahrƗm’s detachment on the Tatavi if he was informed about the proximity of Narses and Khusrau, but Theophylact’s narrative is not

The body of water in question must have been lake Urmiah. BahrƗm’s task was complicated by difficulties in tracking the movements of Khusrau and Narses, for his reconnaissance riders were taken prisoner and interrogated on one occasion, but possibly more often.226 BahrƗm had nevertheless formed a rough estimate of the size of the two armies closing in on him. He occupied a ‘central position’ between the two forces, and decided to throw his weight against the smaller Roman force from Armenia under John Mystacon’s command.227 Theophylact Simocatta indicates that Bindoes, Khusrau’s loyal supporter, was operating with John’s army, having overall command of Persian troops drawn from the Sasanid Armenia and Azerbayjan.228 Their presence gave

their military retainers) and then made contact with John Mystacon. Bindoes seems to have remained with the latter thereafter, along with Bestam, Khusrau’s agent in Sasanid Armenia (Armin). The men to whom Bindoes distributed the money provided by Maurice seem to have been military aristocrats of the two provinces who used it to pay their troops. Simocatta, Hist. IV 15.1 – 6. The combined forces of John Mystacon and Bindoes formed the northern wing of the Roman-Sasanid expeditionary force. Bindoes, PLRE IIIA: 231-232. Bistam, PLRE IIIA: 232. 229 Simocatta, Hist. V 8. 8 (Whitby-Whitby 1986: 143 and note 40). 230 See previous note. 231 This goes against Minorsky, who has John Mystacon’s Armenian troops operating to the northeast of BahrƗm’s main body, but elsewhere has John’s right (viz. the right moira or droungos, or perhaps a reconnaissance detachment) trying to link up with Narses and Khusrau on the Greater Zab, and advancing on the GƗdir river, which flows into lake Urmiah from the west. Minorsky 1944: 245. 232 Simocatta, Hist. V 8. 9-10 (adapted from Whitby and Whitby 1986: 143).

paper… Action in war is like movement in a resistant element’. Clausewitz 1976: 119f. 222 Rufinus 7, PLRE IIIB: 1099. 223 Simocatta, Hist. V 8. 4-5. The prisoners were subjected to torture, being interrogated ‘under the lash’. 224 Simocatta, Hist. V 8. 3. 225 Simocatta, Hist. V 8. 6 (adapted from Whitby-Whitby 1986: 143). 226 Simocatta, Hist. V 8. 4 (Whitby and Whitby 1986: 142). 227 On ‘central position’ in Belisarius campaign of 541 in Sasanid Arbayistan, Trombley 2005: 399f. 228 After defecting, Bindoes made his way to Sasanid Adrabignon (Azerbayjan), raised allies (who must have consisted of aristocrats and

342

FRANK R. TROMBLEY: THE OPERATIONAL METHODS OF THE LATE ROMAN ARMY IN THE PERSIAN WAR OF 572-591 entirely clear about how closely the two Roman armies communicated before the standoff on the Tatavi. In any case, the two magistri militum effected the junction of their forces somewhere south of lake Urmiah. Theophylact provides another estimate of the size of the respective armies, which may be a little exaggerated:233

(60,000 to 40,000 men). To judge from Theophylact’s report, BahrƗm’s force was at a clear disadvantage compared to the Roman troops because of the latters’ many years of hard marching and fighting.236 Two pitched battles were fought in rapid succession as a consequence of these operational manoeuvres, both of them in the mountainous country south of lake Urmiah. In the first of them, BahrƗm ordered his army to abandon the battlefield without any decisive result, but the second engagement went substantially in favour of the Romans. The course of these battles was more a matter of tactics than operations, but some of the incidents were nevertheless a direct continuation of the manoeuvre phase of the campaign. The Roman forces practised a more disciplined method of tactical manoeuvre and alignment in drawing up their formation than their Sasanid allies, as Theophylact claims, forming up the night before the first, indecisive battle ‘in disciplined silence and without noise’.237 In contrast Khusrau’s troops were champing at the bit and eager for action. The historian puts it in these terms:238

It is said that Khusrau brought with him more than sixty thousand soldiers, including the allies [viz. the troops of Narses and John Mystacon], whereas the battle was knit by BahrƗm with forty thousand men. The most important phase of a campaign is the transition from operational reconnaissance, manoeuvre and supply haulage to the actual engagement. The Byzantine and Sasanid tactical manuals both stress the risks involved in engaging in a pitched battle. The Ɩyin-nƗme observes (my italics):234 Let the captain take care, that the rays of the sun and wind should be from the rear of his army, and let him not enter into action with the army unless in case of extreme necessity and being in such a position when battle cannot be avoided. And if that happens, let the captain try to protract the battle to the end of the day.

Narses was angry with the [Sasanid] generals Bindoes and Sarames, who were unable to calm their barbarian forces into untroubled quiet. Then the Romans shared their discipline with the allied barbarian forces, and persuaded them to arrange their formation in order and refrain from foolish clamour.

Among the questions stemming directly from the operational phase of the campaign was the relative experience of one’s own troops and their numbers vis-àvis those of the enemy. The Sasanid tactical manual sets forth a high standard of caution (my italics):235

There were doubtless many other behavioural issues and ethnic differences that provoked arguments between Roman and Sasanid commanders and troops, but most of these have vanished from the historical record.

When the majority of the warriors of the army are tried, intelligent and valiant [men], it is best for the army if the enemy should attack first; whereas, when the majority are inexperienced and the battle cannot be avoided, then it is better for the army to be first in attacking the warriors of the enemy. It is not advisable for the army to fight with the enemy unless it surpasses the enemy three or four times; but when it is the enemy that attacks, then [the army] is allowed to fight, if it surpasses the enemy in number by one and a half approximately. But when the enemy invades [our] country, [the army] is allowed to fight even when it is in lesser numbers.

In the first, indecisive battle the Romano-Sasanid army was drawn up in three divisions, as prescribed in the Strategikon of Maurice and Ɩyin-nƗme.239 Khusrau’s generals Sarames and the MƗhbǀdh commanded the right division, which in Roman and Sasanid tactical doctrine was normally the attacking wing.240 Narses and Khusrau, with the troops of Oriens and the latter’s special guard, commanded in the centre, and John Mystacon and the troops of Armenia made up the left division, which normally stood on the defensive during the first phase of

The last of these conditions does not strictly apply to operations around lake Urmiah, because, although this took place in Sasanid territory, the Roman armies were acting as allies of the indigenous forces commanded by Khusrau II and Bindoes. If the ratio expressed in Theophylact Simocatta’s figures is taken at face value, it means that BahrƗm was outnumbered by three to two

236

Simocatta, Hist. V 9. 9. Simocatta emphasises the effective discipline of the Roman army throughout the campaign of 591. From this, it appears that standards of discipline mentioned in the Strategikon of Maurice had been brought into force not later than the late 580s. Among the principles of tactical warfare seen here that are also enunciated in the Strategikon are: not abandoning the ranks while awaiting an enemy assault (I 8. 16); the maintenance of complete silence as the enemy advances in order to terrify the enemy, calm the younger soldiers, and permit commands to be heard more easily (II 17-18); and keeping the battle line even and cohesive (II 18). It was recognised that the horses would become more excitable if the troops failed to maintain complete silence (II 17). 238 Simocatta, Hist. V 9. 6-7 (adapted from Whitby and Whitby 1986: 144). 239 Maurice, Strategikon II; XI 1. Inostrancev 1926: 13. 240 Maurice, Strategikon III 14. 237

233 Simocatta, Hist. V 9. 4 (adapted from Whitby and Whitby 1986: 144). 234 Adapted from Inostrancev 1926: 14. The Sasanid army made a sunrise attack against Belisarius’ force at Dara in July 530, coming up from the south where their camp was at ‘AmmudƯn. Procopius, De Bellis I 13. 15-19. 235 Adapted from Inostrancev 1926: 14, 30.

343

THE LATE ROMAN ARMY IN THE EAST a battle.241 These tactical arrangements were a direct consequence of the grand strategy and operational aims of the war. Khusrau’s troops would make a hopefully decisive initial attack, for the moment allowing the Roman troops to keep their status as a corps d’observation. Khusrau would after all be the main beneficiary of the destruction or weakening of BahrƗm’s army, so Sasanid troops would be expected to do most of the hard fighting. Roman interests were also in play, however, for Maurice was keen to end the conflict in the interest of reducing public expenditure in Oriens and Armenia, and of transferring senior officers and experienced formations to Illyricum and Thrace to oppose the Avars’ predatory raids, thereby exerting greater control over the Slavic Landnahme in the Balkans. Roman troops would therefore keep in formation and not advance against BahrƗm’s opposing divisions unless the initial Sasanid attack hesitated or failed.

wooded locality supplied with water, and to let the enemy have at his disposal the plains and lowlands’.245 Khusrau evidently allowed the right division to advance, thinking that BahrƗm’s troops were less than battle-worthy, but the terrain gave the latter a sufficient edge to stand their ground. Brevis verbis Khusrau should have known better. Nightfall ended the engagement. Marches to new campsites and the selection of ground are both operational and tactical questions, as can be seen in BahrƗm’s manoeuvres after retiring to his camp after the battle. Theophylact Simocatta observes (my italics):246 [BahrƗm] broke camp at day break and retired towards difficult country, which was steep and unsuitable for cavalry. After this, he came to the plain which extends nearby, where the city of Cancazon is situated. BahrƗm moved camp and from there led his forces elsewhere, planning thereby to shake off the aggressive thrusts of the Romans.

This was largely how the first battle turned out. Consistent with Sasanid tactical doctrine, BahrƗm prepared for a night action, probably because of his inferior numbers and the consequent need to achieve surprise.242 The ruse failed. Although there were some difficulties in getting the Romano-Sasanid right division to conform its deployment, as already mentioned, BahrƗm’s troops put in a poor showing against them:243

This confirms the supposition that BahrƗm, with the smaller number of troops, was methodically preserving his army by avoiding risks and relying on mountainous terrain to frustrate the movement and logistics of Narses and Khusrau, who would have experienced greater problems in maintaining adequate food supplies. BahrƗm seems to have aimed at controlling the defiles and exhausting the resources of the plains where water, rations and fodder were more abundant. Narses evidently intended destroy the enemy before his own supplies ran low and BahrƗm escaped. This led the Romans to make forced marches covering some ninety kilometres to the plain of Cancazon, and from there to the valley of the Blarathos river (present-day Saruk), where the opposing armies again made contact.247 Theophylact does not mention how many days elapsed in this phase of the operation.

[They were] dismayed by the strength, number and good order of the Romans, and withdrew to the mountain slopes… Five hundred of BahrƗm’s company laid down their arms and went over to the Romans… [Khusrau] urged the Romans to advance to the hillsides… [but the Romans] did not disperse at all from their ordered cohesion and … deferred any advance to the mountain heights. The allied success was owed in part to Narses’ success in concentrating the cavalry squadrons of Oriens and Armenia on the battlefield, and the timely execution of the supply and routing plans made before the expedition began, an aspect of operations. The Romano-Sasanid right wing, commanded by the MƗbǀdes and Sarames, then advanced pell-mell onto the slope and was defeated. Theophylact’s narrative reflects an officially inspired contempt for Khusrau and the Sasanid method of apparently using cavalry shock action to dislodge the enemy on broken ground, something which both Roman and Persian tacticians considered to be bad practice.244 The Ɩyin-nƗme specifically counsels the commander to make a ‘disposition of the troops in some covered,

The final battle took place there, and BahrƗm suffered a decisive defeat. It is significant that Narses and the cavalry squadrons of Oriens in the centre division made the decisive attack through shock action against BahrƗm’s centre after the latter had taken personal command of his left division and routed the allied right, which probably still consisted of Persian troops commanded by the MƗhbǀdh and Sarmanes.248 The operational success of advancing two Roman armies deep into Persian territory was thus converted into a tactical victory that tipped the scales and achieved a primary Roman war aim, the end of the nineteen-year conflict with the overthrow of BahrƗm and installation of Khusrau II on the Sasanid throne.

241 It is unknown if Bindoes with the pro-Sasanid troops of Armin and Azerbayjan was present with John. He may have been operating further afield at this juncture. 242 Inostrancev 1926: 15f., 32. 243 Simocatta, Hist. V 9. 9-10 (adapted from Whitby and Whitby 1986: 144f.). 244 Theophylact, Hist. V 9. 10-12. Maurice, Strategikon XI 1 (Dennis and Gamillscheg 1981: 356, lines 27-32). Inostrancev 1926: 48 (‘light weapons’).

245

Adapted from Inostrancev 1926: 16. Simocatta, Hist. V 10. 1-2 (adapted from Whitby and Whitby 1986: 145). 247 Simocatta, Hist. V 10. 2-3 (adapted Whitby and Whitby 1986: 145, notes 46 and 47). 248 Theophylact’s silence suggests that the previous tactical arrangements were still in force. 246

344

FRANK R. TROMBLEY: THE OPERATIONAL METHODS OF THE LATE ROMAN ARMY IN THE PERSIAN WAR OF 572-591 The actual combat is more a question of tactics, but Theophylact Simocatta’s narrative mentions some points bearing on operational issues. For one thing, the Romans appear to have issued a Latin password (perhaps ‘Ave Maria’) to their own men and their Sasanid allies, to prevent confusion during the chaos of battle.249 Khusrau’s Persian troops were undoubtedly equipped and manoeuvred like those of the enemy, and the Romans might otherwise have mistakenly destroyed them by missile fire or shock action, or both, when during the battle Narses re-deployed some of his cavalry squadrons to the right wing to stop its rout.250 The problems of coalition warfare were thus present even in combat. After the Sasanid line collapsed, the hard-fighting troops of Oriens destroyed their opposite numbers and succeeded in isolating some 6,000 of BahrƗm’s men on a hillock, killing some and taking many prisoners.251 At this point it was discovered that among the latter were Turks fighting as ethnic allies of the rebel.252

kindness and preservation (sytgqÊa) he had received from the Romans. Then, after depositing such words with him, the Romans returned homeward. But Khusrau, in fear that an assassination might perhaps be organised against him by his countrymen, asked the emperor Maurice to send a thousand of the army for the protection of his person. This implicitly reflects the operational and tactical effectiveness of the armies of Oriens and Armenia in the Persian war of 572-591. The reasoned bravado of the generals is also consistent with the public discourse of the regime in Constantinople dating back to the year of Justin II’s consulship.256 Aftermath and Prospectus The conclusion of hostilities allowed the regime in Constantinople to transfer numerous military forces to Illyricum and Thrace, because the defences of the Balkan peninsula required strengthening in the face of the Avars — a Turkic tribal confederation that demanded annual tribute to defray the costs of paying off its warrior retinues, failing which it threatened to attack the Late Roman cities and their territories.257 There was also the ongoing migration of the Sklavinoi — Slavic tribes, their Germanic allies and ethnic tribal confederations of steppe Iranians like the Sagudatai — who were beginning to establish settlements in Illyricum and Thrace. The political and demographic history of these events is little known, but it seems quite likely that the native East Roman population living north of the Jireþek line was becoming increasingly displaced, explaining such phenomena as the migration of the group later known as the Vlachs.258

After this a final operational task was put in front of the troops of Oriens. On the day after the battle, after the plundering of the dead, a force of 10,000 men was assembled under the command of Marinus, dux of Syria, and Bestam, who seems to have commanded troops from Sasanid Armenia.253 It may be that the cavalry squadrons of Oriens were still sufficiently fresh for this mission after the one-sided battle with BahrƗm’s centre division. On the other hand, it is possible that Marinus was acting as Narses’ lieutenant-general and took some troops of Armenia on the pursuit as well; the latter, apparently stationed on the left or defensive wing of the allied army, seem not to have been engaged on the day of the final battle on the Blarathos river, and may therefore have been available for further duty. It is conceivable that Marinus assembled his force from the most battle-worthy cavalry squadrons, irrespective of their origins, including his own provincial moira or droungos. This combined force pursued BahrƗm to the breaking point and returned to Khusrau’s camp — which had been moved to Canzacon because of the decaying corpses on the battlefield — over a stretch of ten days, thus completing the last Roman operation of the war.254 Khusrau then dismissed the generals of the two Roman armies in a manner that created great offence. The generals, undoubtedly in a delegation led by Narses, are said to have expressed their reservations publicly in front of the Sasanid king:255

The need to protect the provincial infrastructure of Illyricum and Thrace necessitated large-scale operations on an annual basis. Many of the generals who had fought in the Persian War of 572-591 — even troublesome and discredited characters like Priscus — found further employment in the Balkans, as did many of the arithmoi that had served under them.259 A new system of operations, using somewhat lighter cavalry, armed with equipment selectively borrowed from the Avars, but otherwise retaining Roman tactical methods, unit organisation and weapons, was an essential feature of adapting operations to the Balkans. It involved abandoning what might broadly be called the clibanarius tactics of the Syrian and Mesopotamian steppes. It is quite probable that the new, lighter system of equipment and manoeuvre still prevailed when the army of the magister militum per Thraciam was transferred back to Oriens after December 603 to confront the threatening

And so the Roman generals expounded a spoken moral to the barbarian that he should not forget the 249

Simocatta, Hist. V 10. 4-5. Simocatta, Hist. V 10.9. 251 Simocatta, Hist. V 10. 11-12. 252 Simocatta, Hist. V 10. 13-15. 253 See note 228 above. 254 Simocatta, Hist. V 11. 7. 255 Simocatta, Hist. V 11. 8-9 (adapted from Whitby-Whitby 1986: 147f.). Theophylact emphasises Khusrau II’s lack of gratitude to contradict the latter’s claim that he was avenging Phocas’s execution of Maurice by renewing hostilities in 603. The point is that Khusrau had always shown ingratitude to his benefactor Maurice, even in 591. 250

256

See Appendix III below. Maurice, Strategikon XI 2. Pohl 1988: 128-162. Whitby 1988: 140f. 258 Pohl 1988: 94-127, 232-236. On the Jireþek line, see Rosetti 1968: 206 (with references), and map entitled ‘Teritoriul uniunii linguistice balcanice’. The discussion is noted at Jireþek 1911: 38f. 259 Priscus 6, PLRE IIIB: 1052-1057. Whitby 1988: 156-165. 257

345

THE LATE ROMAN ARMY IN THE EAST situation in northern Syria and Mesopotamia, because of Khusrau II AbarwƯz’s avowed intention to overthrow the usurper Phocas and appoint as emperor Theodosios, the ostensible son of Maurice, all of which culminated in the great Persian War of 603-629.260

concentrated their forces mostly in the south to guard against Heraclius’ movement against Antioch, and so a serious crisis developed for them in Persarmenia. As Sebeos relates:266 But there soon arrived an order of the Persian king brought in all haste by swift couriers, enjoining his troops to march against the Romans and to destroy their army to the last man. The Persians therefore took the road, penetrated the province of Ararat by forced marches, and came to encamp on the banks of the Araxes. Their plan was to engage in combat on the morrow, but at the same time Philippicus broke camp at night, marched toward the canton of Nig and, after passing behind Mount Aragae, re-entered Roman territory across the cantons of Shirak and Vanand, not far from the city of Karin [Erzerum-Theodosiopolis].

The new ‘Avar model’ army was an important factor in the early operations of the new war in Oriens; they were sufficiently effective to ensure a stalemate. No important fortresses surrendered to the Sasanids during this phase of Phocas’ reign (December 603 to October 610); there is evidence to the contrary that fortifications were being repaired and maintained at sites in the Jabal Hass district, including Anasartha (present-day al-Khanasir) and Mu‘allaq.261 The real disasters came later, after Heraclius’ overthrow of Phocas in October 610, while the new regime was attempting to consolidate its position in Constantinople and Asia Minor.262 Veteran generals of the Persian and Avar wars who had served under emperor Maurice were given important operational tasks in the early years of Heraclius’ regime, among them Philippicus, Priscus and perhaps Heraclius the Elder.263

Sebeos adds that the Sasanid force pursuing Philippicus suffered serious losses in men and animals during its march. The third phase of the war (April 622-628) saw deep penetration operations in Sasanid Armenia and Mesopotamia. The operations of 622-628 are reported in the correspondence of Heraclius sent from the zone of operations, much of which appears to be the basis of the operational narratives in the Chronicon Paschale and the the Chronographia of Theophanes the Confessor.267 These materials have frequently been studied in detail, but not always from the perspective of the planning and execution of operations.

There are few contemporary narratives covering the first two phases of the Persian War of 603-628, the initial defensive operations under Phocas (December 603 – October 610) and then the defensive-offensive of Heraclius’ conduct of the war (between October 610 April 622).264 There is only one report about a significant offensive expedition, so it is difficult to develop a clear picture of the operational realities and tactical performance of Roman armies of Oriens. The event in question is mentioned in the Armenian historical work of pseudo-Sebeos. According to his account, the Roman army in 613 executed a double advance against the Sasanid forces on the far side of the Taurus and AntiTaurus barrier. Heraclius, taking overall command, advanced with the army of Oriens into the vicinity of Antioch and was defeated. Meanwhile, Philippicus with the army of Armenia penetrated Persarmenia after starting out from Caesarea in Cappadocia I, which must also have been Heraclius’ base of operations for this campaign.265 Philippicus struck eastward after reaching Erzerum-Theodosiopolis, crossing the portage of the headwaters of the Euphrates into the Ararat province of Persarmenia and down the Araxes river. He eventually encamped on the plain surrounding Valarshapat, the administrative capital of the Ararat province, which lies immediately to the west of Dvin. The Persians had

Concluding Observations The present discussion has raised many issues that require further clarification, for it was possible only to enumerate exemplary cases of Late Roman operational methods in it. The most pressing question is the need plausibly to establish an approximate order of battle for the armies of Oriens and Armenia, for the deployments listed in the Notitia Dignitatum certainly underwent various changes between their publication ca. 425 and the Persian War of 502-507 and thereafter.268 One suspects that the force levels implied in the document remained roughly the same, but cavalry squadrons (equites) and infantry formations (legiones) were sometimes shifted to new bases. There is also the question of how many cavalry squadrons were subtracted in order to fill out the 266

Sebeos, History § 114 (Thomson and Howard-Johnston 1999: 67, 205f.; cf. Macler 1904: 66f.). Thomson and Howard-Johnston set this event in 615, but Sebeos’ internal chronology seems to put it in 613. Thus Kaegi 2003: 75. 267 Chronicon Paschale (Dindorf 1832: 727-735; Whitby and Whitby 1989: 182-188). Cf. Oikonomides 1971. The level of detail given in Theophanes’ notices is suggestive of written communiqués dispatched from the zone of operations, which the chronographer has summarised. This is particularly the case with his account of operations between October 627 to February-March 628. Theophanes, Chronographia AM 6118 (De Boor 1883: I, 317-327). 268 This point is repeatedly made in sections of the commentary on Ps.Joshua, Chronicle (Trombley and Watt 2000: notes 244, 260, 267, 268, 271, 333, 391, 425, 519). This discussion clearly needs to be taken forward further.

260

Whitby 1988: 305-308. Phocas 7 and Theodosius 13, PLRE IIIB: 1030-1032, 1293f. Theophanes, Chronographia AM 6096 (De Boor 1883: I 292, lines 5-6, 11-14). 261 Trombley 1997: 182-198. 262 Kaegi 1973. 263 Heraclius 3 and Heraclius 4, PLRE IIIA: 584-586, 586f. 264 Between 603-October 610 pseudo-Sebeos’ Armenian history mainly reports operations conducted by the magister militum per Armeniam in eastern Armenia, which saw an overall poor Roman performance. Sebeos, History (Thomson and Howard-Johnston 1999: 57-65, 197202). 265 Sebeos, History § 114.

346

FRANK R. TROMBLEY: THE OPERATIONAL METHODS OF THE LATE ROMAN ARMY IN THE PERSIAN WAR OF 572-591 body armour including a plumed helmet,270 a breast-plate that could be worn over the mail, a bow with 30-40 arrows in a quiver, spare bow strings, shield and sword, but his equipment now also included an Avar lance and Avar-style round neck-pieces (with linen on the inside and wool outside).271 The men were supposed to equip their horses with Avar-style protective armour covering the animals’ heads, necks and breasts,272 and to provide themselves with Avar-style cloaks and encamp in Avarstyle round tents.273

order of battle of the magister militum per Armeniam when this command was newly established ca. 528. Force levels have operational implications, as demonstrated by Marcian’s concentration of too many Roman troops for the two sieges of Nisibis in 573, leaving the provinces of northern Syria open to Adarmahan’s raid in November of the same year. Another desideratum would be a thorough survey of defensive installations to determine levels of occupation in the run-up to the Persian War of 572-591, all of which with the road network played an integrated role in the manoeuvre of cavalry squadrons when provincial troops engaged in defensive operations, and provided habitations and winter quarters for the provincial troops commanded by the duces.

It is clear from this that the army had begun to divest itself of the catafractarius clibanarius, the heavily armoured cavalry lancer that had become common in the fourth century.274 Their name comes, ironically, from the neck-piece that the Persian clibanarii wore, the grivpan,275 for which Byzantine troops were now being made to wear the Avar equivalent instead. Heavy cavalry of this type was potentially useful in open terrain against infantry or barbarian steppe cultures whose armies fought in kin-based formations, but not against the armies of state formations like the Avar khaqanate, as the Strategikon notes elsewhere. Many clibanarius formations are mentioned in the Notitia Dignitatum. Such a formation of comitatenses at Arsinoe in Egypt, known as the Leones Clibanarii, may still have made use of the older heavy equipment as late as 531,276 supposing that shock tactics were still imagined as an effective tool against the barbarian Blemmyes, Nobadae, Mazikes and other peoples of desert, against whom there were few military operations between ca. 452-616/7.277 Clibanarius tactics were sometimes effective even against professional soldiers during pursuits and counter-pursuits.

It appears that the construction of new military narratives about the Persian wars of the sixth and early seventh centuries are of doubtful value without detailed operational analysis and an organic approach to a wide range of behavioural and environmental factors. This includes the study of how the infrastructure was maintained, how great the Roman supply capacity was, and how essential materièl was conveyed to armies in the zone of operations. The narrative of Theophylact Simocatta provides brief but nevertheless important glimpses into these processes. Future discussion might also provide analysis of how differences in ethnic attitudes affected the manufacture and use of particular weapons and armour, and variations in modes of small group organisation and co-operation that were severely tested in operational marches and tactical engagements, particularly between Roman and Arab commanders. The consequences of long periods of professional inactivity are apparent in the non-literary papyri of Nessana.269

270 The plume was presumably, along with the lance pennon, a coloured recognition symbol for the individual banda. 271 Strategikon I 2, lines 19-20. Bivar was unable to establish a clear connection with the Avar-style sword, which seems to be the origin of the European sabre. Bivar 1972: 288. 272 Maurice, Strategikon I 2, lines 35-39. 273 Maurice, Strategikon I 2, lines 46-49, 59-61. 274 On its origins, see: Eadie 1967. 275 Bivar 1972:. 277f. and note 28. 276 Cf. Jones 1964: 661. Maspero 1912a: 105f. 277 Maspero 1912a: 130. As noted above in my remarks on the battle of the Solachon plain, it is doubtful — in fact a fallacy — that shock tactics were ever effective against trained Sasanid soldiers. They may however have been effective against the tribal societies of the Egyptian steppes, just as they seem to have been against the Arabs, as in Procopius, De Bellis I 18. 35-37. Bivar quite possibly overrates overall effect of shock tactics, because, as he admits elsewhere, dropping enemy horses by steady fire with Roman heavy arrows was more likely to break up cavalry charges. Bivar 1972: 279, 281, 286. On the failure of shock tactics at the battle of Callinicum in 531, see Procopius, De Bellis I 18. 45-50. I am inclined to doubt Maspero’s view that the army formations deployed in Egypt were not intended for use in military operations (‘… l’armée d’Egypte n’était pas pour la guerre’). Maspero 1912a: 5. Even if the anticipated threats to this region were ‘low intensity’ in comparison with the situations likely to arise in Syria and Roman Mesopotamia, the mere existence of barbarian peoples with warriors who used mobile tactics posed a credible danger to the Nile settlements, as Maspero clearly implies elsewhere in his seminal study. This fully materialised in the Muslim advance led by ‘Amr b. al-‘As, which reached Rhinocoura (present-day al-Arish) at the frontier of the diocese of Egypt on 12 December 639. Ibid.: 25-34. Cf. the chronology at Butler 1978: 544f.

The sixth century was a culminating epoch, a new era of confrontation between Romans and allogenous peoples in increasing numbers. The dialectic of this ‘challenge and response’ environment required a systematic war effort far greater than the ad hoc arrangements of earlier antiquity. The adaptive qualities of Late Roman bureaucratic and organisational skills as seen in the Strategikon of Maurice and the narratives of Theophylact Simocatta were an important factor in the preservation of Mediterranean material culture during the political turmoil and military collapse of the seventh century. Appendix I. The ‘Avar model’ army in the Strategikon of Maurice The re-arming of the Roman regular cavalryman (jabakkÇqior stqatiÍtgr) is described in the Strategikon of Maurice, a military manual that integrated the lessons of the late sixth-century wars into Roman tactical and operational doctrine. The soldier was now trained to defeat the enemy by missile fire and resort to shock action only if that failed. His equipment included a suit of chain mail (fÇba) extending down to his knees, 269

Kraemer 1958 (P.Nessana III): nos. 15, 16, 19-22, 24, 26.

347

THE LATE ROMAN ARMY IN THE EAST It succeeded, for example, against a Persian sortie from Nisibis during operations Sasanid Arbayistan (Beth ‘Arabaye) in 541:278

misunderstandings in their dealings with tax officials. Elsewhere Menander Protector sometimes explains Latin terminology to keep the record clear of confusion, as for example: ‘[quaestor] which, I think, is named from the Latin for ‘investigate’),’ and ‘called a secretis in Latin’.283 The translation of the novel of 575 as given is a provisional attempt to progress the study of the impact of war on provincial society.

As the Goths came upon [the Persians] first with their long and densely arrayed spears, the Persians did not await their attack but were driven to retreat. The Romans with the Goths followed them and killed one hundred and fifty men.

Prooemium. Justice and love of humanity are great and good [principles] in the human affairs. The former principle allots an equal share to each person and is not desirous of other things, while the latter tends toward mercy, since it is necessary to liberate those who are caught in difficulties. These things make it possible to manage the empire, to preserve the state and to govern the human way of life well. Following from this it is also most desirable for us who have received imperial power from God to be conspicuous through the good practice of these [principles], in order that by granting something beneficial to our subjects we might in return receive virtue and glory from above. We are moreover aware that wide-ranging disasters coming one after the other have brought such great impoverishment to landed properties (aÚ jtÉseir) that it has not been possible for landowners to bring in their rents (aÚpqosËdoi) or sufficiently (aÕtÇqjyr) to make payment of their taxes. This has happened while military (stqatiytijÉ) and other expenditure have necessarily increased, become very urgent and put pressure on us to make many generous contributions of funds for many wars. We are therefore sufficiently anxious as to how the shortfall (wqeÊa) might be made up and the impoverishment of [our] subjects healed. Since our intentions in relation to this have for the most part been divided, it is nevertheless preferable to pay regard to love of humanity and give a measure of support (boÉheia) to our taxpayers (ÜpotekaÊ) in a manner pleasing to God and make up the unpaid taxes (tÄ ke²pom) needed for public expenditure from the imperial treasury.

On the other hand, it must be recognised that names like clibanarii may have been maintained for reasons of organisation designation and cadre tradition, and may not tell us anything about the way the troops in question were actually armed. Appendix II. Novel 163 of Tiberius Constantine (ca. April 575)279 Tiberius Constantine responded to the Persian armies’ pillage and depopulation of Mesopotmia, Osrhoene, and the Syrian provinces by publishing a law that temporarily reduced the capitation tax by twenty-five percent for a four-year period. The practice was not a new one; on 25 December 503 Anastasius issued the first in a series of seven tax remissions for the province of Mesopotamia, which had suffered catastrophically at the hands of Sasanid troops and their Arab and Hunnic allies during the year just ending. Payments were subsequently written off through 510/11. Partial tax remissions were also granted to Mabbug-Hierapolis in Euphratesia because of troops staging through it in 504/5 and the territory of Edessa in Osrhoene in 505/6, probably in view of the large troop concentrations there and damage to its territory during a poorly executed Persian blockade in September 503.280 The loss in 573 of the rural populations of two provincial capitals, Antioch and Apamea, will have significantly reduced the urban tax base.281 The remission of the tax in gold would particularly have benefited women who had been displaced from their communities, because in the Syrian provinces they were liable for the full capitation tax, unlike other regions of the empire where women were normally rated as only half a caput.282 In Tiberius Constantine’s novel of 575, the equivalents for particular Greek terms are given in Latin to avoid disputes about the meaning of language in the text of the law. This seems to be a clear recognition of the fact that noticeable numbers of Latin-speaking taxpayers resided in Oriens whose unfamiliarity with Greek sometimes led to

I. While therefore dedicating our cares to our master Christ and completing the liturgy in these present days of the holy festival of his saving Passion and holy Resurrection, we offer, commensurate with his will, the present gift (dyqeÇ) on behalf of the state, through which we concede (concedimus) one entire instalment (integrum unum canonum) of public taxes divided over a four-year period ([t±r] eÓrquadriennium diaiqoulÈmgr), that is a quarter part of this [amount] in the coming ninth indiction, and another such instalment in the tenth indiction after this, and similarly another quarter part in the eleventh indiction, thus distributing

278 Procopius, De Bellis II 18. 24-25. The historian does not explicitly state that cavalry were involved in this action. 279 Novella CLXIII (R. Schoell 1972: 749-751). 280 Ps.-Joshua, Chronicle § 59-63, 66, 78, 99 (Trombley and Watt 2000: notes 401, 448, 449, 544). Mabbug-Hierapolis had for centuries been a concentration point for Roman armies marching into Mesopotamia. Poidebard-Mouterde 1945: 20. 281 The taxation system of the Syrian provinces seems to have been governed by a law of 290. See Jones 1964: 62-64, 453, 768, 820, esp. 63 and 1078, note 49. 282 Jones 1964: 63.

283

Menander Protector, Fr. 18.3 (Blockley, 158f.) and Fr. 23.9 (Blockley, 208f.).

348

FRANK R. TROMBLEY: THE OPERATIONAL METHODS OF THE LATE ROMAN ARMY IN THE PERSIAN WAR OF 572-591 equally in this manner the allowed customary (paqewËlemai … sumÉheiai) [payments?] .

those commodities can be done all that easily, what with the unavoidable expense involved in disposing and regulating public affairs, so to speak. We trust that this gift will be of great assistance to the state because God makes auspicious provision through such policies. If anyone dares to transgress [the laws] which we have enacted for the assistance of our subjects, they will imperil their property and personal safety.

II. We remit for them the remaining [taxes] of the time passing until the end of the presently elapsing fifth (quinta ÑpimÈlgsir) indiction, and we enjoin that no collection of taxes whatsoever should be made by our honoured pious [officials], whether they are administering (Ûq§) them in the high office of the praetorian prefecture of Oriens, or in the prefecture of the Illyrians, or in the quaestura (ÐqwÉ) of the military companies of the islands, Scythia and Moesia, or in [the management of] our sacred largess, or in any other office. They should forbid city councillors, tax-collectors (ÑjkÉptoqer), keepers of records (scriniarii), commissaries (taneÍtai), officials of the militia (palatinii), exactores (ðmutai), treasury and also treasury officials (ÜpodÈjtai) and everyone who solicits or collects taxes, not to mention those who perform frauds against farmers and taxpayers or the owners of village estates (oÚ t´m wyqÊym despËtai) [to commit any injustice (Ðdije²m)], whether [the taxpayers] have previously received different written replies (ÐmtivymÉseir) about the tax arrears remitted to them, or whether they have received sureties (ÑccÌai) and written agreements (ÛlokocÊai) against fraud., for we reject the validity of every act of circumvention which has been or will be made against our [present act of] munificence. If anyone has anticipated the collection of our gift as regards the arrears of the fifth indiction, and appears not to have paid this, he should by all means present it for public accounting. In order to benefit the taxpayers in the present humanitarian gesture (vikamhqypÊa), we have shared [the burden] with them, but we have not made a remission (sucwÍqgsir) of the aforesaid temporary tax assessment (ñcwaiqor jÇmym) for public tax collectors who wish to satisfy their own greed, since the annual tax in kind (ÑlbokÉ=annona) has not been reduced at all. Every measure of grain and the other commodities [subject to the tax in kind] must be contributed and delivered in accordance with customary practice and law, since the quarter part of tax payments has been annualy remitted for a four year period as assessed at the public revenue office (Ù dglËsia tqÇpefa) for those who pay tax in the other [assessments], which are contributed by them in gold or which are paid from the public fisc. The same practice shall be observed in Osrhoene and Mesopotamia as regards the commodities that are contributed for stores and for military expenditure, and similarly as regards the so-called sea-borne commodities for Lazica, Bosporus and Chersonnesus. The assessment of these [taxes] shall be made by the public [fisc] in accordance with the prescribed procedure for the revenue offices of the [praetorian] prefects. These [taxes] shall be paid in such a way that the [taxpayers] will enjoy the benefit of our love of humanity, for it is far from certain that the delivery of

Epilogus. The present policies shall be made known to everyone in this blessed city and in the prefectures under its jurisdiction by ourselves and through the present manifestly divine law, with your glory making the formal announcements. . There is documentary evidence in the Petra papyri for the extension of tax remissions in the prefecture of Oriens outside the provinces enumerated in this law. It is in the last of a series of tax receipts for the years 568, 570/1, 575/6, and 578. The last in the series dates from JulyAugust 578, by which time the reduction of payments was fully in force:284 (Cross) The most honourable Patrophilus [son of Bassus] has given [on account of the taxes of the ninth,] tenth and eleventh indictions … one fourth, one eighth [iugera] from Petra, and from [Augustopolis one half], one third, one tenth, one hundred and seventy[-fifth …, all exacted and] extra taxes in farm products, [in oinokreon and in other products reduced by] one fourth share (tetqaloiqÊa) granted by our [most divine and pious] lords (Justin and Tiberius Constantine) etc. This document is of particular interest because the one quarter remission is applied to agricultural products, which are specifically excluded for Mesopotamia, Osrhoene and the other provinces named in Novel 163. It may well be the case that this law was only one of a series to be published in the aftermath of the disasters of 573, with more generous conditions granted to Syria I and II in view of the destruction of Apamea and the depopulation of its territory and that of Antioch.285 These hypothetical novels would have used protocols and terminology quite similar to Novel 163, but with variations in the amounts and duration of the exemptions. These documents, like the laws published by Anastasius in 503-510, may have been left out of future codifications of the Novellae on the ground of their repetitive language and phrasing. The law of ca. April 575 may therefore owe its inclusion in manuscripts of the Novellae to the need to have an exemplar in case such situations arose again. 284 Slightly modified from M. Vesterinen’s edition and translation in Frösén, Arjava and Lehtinen 2002 (P.Petra I): no. 10. 285 Trombley 1997: passim.

349

THE LATE ROMAN ARMY IN THE EAST the tax remission granted in Novel 163 (ca. April 575), and that of the soldiers because, for example, he paid their wages promptly before the campaign of 579.288 The esteem in which the Caesar was held is strikingly confirmed in his release of Persian captives as a gesture of good-will towards concluding peace with Khusrau I shortly before his death (February or March 579). Menander Protector — undoubtedly reflecting public opinion in the capital — characterises this act as ‘pleasing to God’ (he© jewaqislÈmom).289

Appendix III. CIL III 212, its political context and the use of Latin in the late sixth-century Roman army It was suggested above that Maurice’s operations in the summer of 578 against fortresses in the south-eastern territory of Nisibis gave occasion for a gesture of defiance against the regime in Ctesiphon, to wit, the erection of a monumental chariot in Sasanid territory, whose accompanying inscription — no longer in existence — was seen by a certain Jesuit named Rigord near a bridge not far from Nisibis (Nisibi prope pontem non longe ab urbe). His transcription was published in 1673. A traveller named Olivier, the second volume of whose work appeared in 1804 (Voyage dans l’empire Ottoman) seems to have found only traces of the text:286

It is not feasible to investigate the literary ‘archaeology’ of Rigord’s claims, but only to indicate the plausibility of such a monument in light of the public statements of the regime in Constantinople between 565-572. There are two pieces of evidence bearing on this.290 The first is a report in the ecclesiastical history of John of Ephesus, who (as I have remarked before) enjoyed important inside contacts with Roman mid-echelon officials. He mentions that Zemarchus, Justin’s envoy, encountered a Sasanid diplomat during his mission to the Turks, who had observed that the Romans were ‘slaves’ of the Persians for paying the annual tribute of 30,000 solidi in accordance with the terms of the treaty of 561/2.291 Zemarchus is said to have replied:292

To the best-chosen (or ‘most beloved’) emperor Justinus: his army of Oriens constructed (this monument) to his fame: a warning to the barbarian nation, coming from the statue of the triumphal chariot. (Iustino imperatori dilectissimo exercitus eius Orientis gentem barbaram e curru statuario cautionem ad gloriam construxit). The terminus ante quem for the raising of the monument was 5 October 578, the date of Justin II’s death, but since the news of this along with Tiberius Constantine’s accession to the imperial title will have taken several weeks, if not longer, to reach Maurice’s army in the zone of operations, it may be that the work was completed as a parting gesture as Maurice’s army retired from Beth ‘Arabaye. Theophylact Simocatta puts their withdrawal ‘just as season of winter was peeking in’ (Ñpeà weil´mor íqa paqÈjuptem), perhaps in December 578 or even as early as November if the severe weather began then.287

They speak falsely, for many Roman emperors have invaded their lands and have devastated them, and taken their people captive. When the Roman emperor Trajan invaded them, he overthrew and vanquished them to the extent that until these days they tremble and shake before the statue of himself that he set up in their land, nor will any of them venture even to this day to pass before it on horseback. The statue was erected at Seleucia on the Red Sea and showed emperor Trajan with the spolia of war piled at his feet, in the same manner as his relief image at Tropaeum Traiani in Scythia Minor (present-day Adamklissi).293 The khaqan of the Turks is then said to have confronted the Sasanid envoys with Zemarchus’ claims. John of Ephesus reports that Khusrau I thereafter commanded the statue of Trajan to be destroyed. The factuality of the incident — which Menander Protector does not mention in his account of the same embassy — is less important than the fact that the regime in Constantinople was engaged in a discourse of power similar to this in its public statements, including court inspired poetry.294 An anonymous verse epigram honouring Justin II runs:295

A Latin document of this date would have been unusual in any part of greater Syria, but I consider this to be a consequence of the large number of western ‘barbarian’ mercenaries in the army at this time, the so-called Tiberiani and officers like the German Eilifredas dux of Phoenice Libanensis who could interpret the monument and inscription to their subordinates. One can hardly expect perfect Latin in an inscription of this nature, for it was only a second language — but nevertheless the language of command — for the greater part of the officers and soldiers in the army. The use of dilectissimus to describe the emperor is unusual, but was certainly consistent with the official views of the regime in Constantinople, and may in fact be a species of special pleading on behalf of Justin II whose insanity was an embarrassment to the regime. More particularly, Caesar Tiberius Constantine — the public face of the Justin’s regime — earned the love of the provincials because of

288

Menander Protector, Fr. 23.9. For the law, see Appendix III above. Menander Protector, Fr. 23.8 (Blockley 1985: 204f.). 290 I am here indebted to Olajos 1981. 291 The political reaction to these payments varied with time and location. The payments made in connection with Justinian’s treaty of 532 with Kawad were seen in a positive way on the victory stele set up at Hierapolis in Euphratesia. Roussel 1939. 292 John of Ephesus, HE VI 23 (Payne Smith 1860: 426f.). 293 Olajos 1981: 379f. 294 Menander Protector, Fr. 10.3. 295 Greek Anthology XVI 72. For consular dates, see Grumel 1958: 354. 289

286 CIL III, no. 212. One had previously supposed that the inscription belonged to the reign of Justin I, as in Trombley-Watt, ps.-Joshua, Chronicle, xlvii, but Maurice’s operations in 578 supply a more appropriate occasion and context for the document. For the degree of Latinity in the army at this time, see below. 287 Simocatta, Hist. III 16. 2.

350

FRANK R. TROMBLEY: THE OPERATIONAL METHODS OF THE LATE ROMAN ARMY IN THE PERSIAN WAR OF 572-591 At Susa the bold Persian will set up another statue loaded with spoils to the emperor (ðman) for his victories, and another army of long-haired Avars across the Danube will crop the locks from their squalid heads. But the imperial city set this [statue] here for the righteousness of his rule thanks to his receipt of the consul’s wreath. May you stand firm, Byzantine Rome, after rewarding the divine might of Justin.

Italy and Africa, as well as from Illyricum and Thrace (where Latin was predominantly spoken north of the Jireþek line) to Oriens and Armenia.298 Data of this kind can be gleaned from the Liber Pontificalis, and other western sources that mention military commanders, including a Greek inscription at Ravenna. It provides a good example of the mobility of senior military officers between commands in the eastern and western Mediterranean. Isaac served in the west from before 2 October 625 until the winter of 642/3 (St. Vitale basilica, Ravenna, winter 642/3):299

The original statue at Susa would have been that of Trajan, the ‘other’ one of Justin II after the victory anticipated in the operations of 573. All this is consistent with the vainglorious character of other public pronouncements made by the regime. Grandiloquent threats and displays of defiance were characteristic of Justin II’s ‘new’ foreign policy of eliminating subsidies to the barbarians, as in the Sasanid envoy Sebokhth’s embassy to Constantinople in 572. The rhetoric was uncompromising:296

Here lies a man who well served our pacific masters (Heraclius and his co-emperors) as a general three times for six years each, keeping Rome and the Occident secure, Isaac, the ally of the emperors (and) the great ornament of all Armenia., for this man was an Armenian from a distinguished family. When this man died gloriously, his prudent wife Susanna, a parent of holiness, groaned painfully at being deprived of her husband, after her husband gained glory for his labours in East and West, for commanded an army in the Occident and Orient.

If the Persian king were to lift one finger, [Justin] would raise his arm and invade Persia; and he was confident that if he made war he would destroy Khusrau and himself give a king to the Persians. With these outlandish threats he dismissed Sebokhth.

Another inscription refers to him as ‘the great commander of the Italian armies who made a display of great tasks’.300 Isaac was the younger contemporary of the men who gained their initial experience in operational methods as junior and mid-level officers in the late sixthcentury wars against Sasanid Persia and the Avars, and whose professional consensus became the basis of the Strategikon of Maurice. It would be of interest to know the dates of Isaac’s earlier commands in Oriens, the sizes of the formations that accompanied him in his transfers to new theatres of operations, and the ethnic make-up of his troops. The inscription was composed in Greek partly because it was the koinƝ of Isaac’s family, partly because it was the lingua franca of the multi-ethnic army and civil population in Rome, Ravenna and elsewhere.301 The senior officers of Isaac’s acquaintance would have had subordinate officers, bucellarii and regular soldiers in their retinues, as well as notarii and other literate personnel, who were selected on the basis of their familiarity with Latin as well as Greek. The Strategikon makes it clear that Latin was the language of command and disciplinary procedure, and thus a unifying factor, even ca. 600:302

These circumstances suggest that executors of the regime’s policies like Maurice would not have hesitated — indeed may have been actively encouraged — to make displays of defiance such as that seen in the Nisibis bridge inscription. There are other questions. For example, why would Latin have been used in a public document of this type? Tiberius Constantine’s Novel 163 adduces Latin terms to the Greek text for the sake of precision. This suggests that administrative Latin was still widely understood in the officia of the provincial governors not only in Thrace and Illyricum, but also in Osrhoene and Mesopotamia. The barbarians recruited and inserted into the formations of Tiberianoi in 574 and afterwards evidently came from the western barbarian territories where their contacts with the sub-Roman administration would have given these troops an implicit understanding of the importance of Latin among the provincials. Furthermore, the language of command in the army remained Latin, as the Strategikon of Maurice makes clear. The Tiberianoi and other westerners would have had to understand that much Latin as part of their professional duties. Apart from the ‘Torna! Torna!’ incident that Theophylact Simocatta reports in 586/7, we have little to go on as far as the daily use of vulgar Latin in the army is concerned.297 One must also contend with the fact that senior officers were likely to be transferred back and forth from the commands in

298 See the vulgar Latin inscription for the wife of Prasinacius the maister (sic) militum found at Sveti Vlas near Burgas. His name is of Greek origin. Beševliev 1964: no. 169. 299 CIG IV 9869 = Guillou 1996: no. 109. Isaacius 8, PLRE IIIA: 719721. 300 CIG IV 9870 = Guillou 1996: no. 108 (church of St. Mauro, Comaccho, Italy, inter 625-43). 301 This can be deduced from the epigraphic data. Guillou 1996: nos. 46, 47, 60. Cf. the Latin subscription in Greek letters in a Ravenna papyrus testament. Tjader 1955: 222 (no. 6) (February and April 575). See also Guillou 1969: 78-88. 302 Maurice, Strategikon I 8, lines 2-3 II 18, lines 14-17, III 5, passim.

296

Menander Protector, Fr. 16.1 (adapted from Blockley 1985: 152f.). Simocatta, Hist. II 15. 9. Năsturel 1966. Dvoichenko-Markov 1984: 511. Cf. Theophanes, Chronographia AM 6079 (Mango-Scott, 380f. and note 18 (bibliography). 297

351

THE LATE ROMAN ARMY IN THE EAST After the organisation of the cavalry squadrons, the following list of punishments should be read out in Latin and Greek.

particular language groups. One thinks for example of the Christian federate (vedeq°tor) Ɯstotzas and his wife Dody, who were interred in the Mevlana Kapi cemetery outside the land walls of Constantinople; their burial inscription is in Greek (6th c. or later).307 There are also the cases of a certain Emalac, a spatharius in the comitatus patrici Bilisarii, perhaps a bucellarius, who died in Rome, and the dux Zerezindo who died at Hispalis in 578.308

[Regarding prayers:] Everyone together, led by the priests, general and other officers, should all together recite Kyrie EleƝson for a while. Then, to achieve a successful outcome, each division should shout Nobiscum Deus three times as it exits from the camp. It is likely that the words ‘Ave Maria’ became the password before the battle fought on the Blarathos river in the Zagros mountains in the summer of 591.303

The plausibility of a military expedition like the one led by Maurice against the territory of Nisibis in 578 erecting a vulgar Latin inscription in a corporate spirit of selfcongratulation is therefore not implausible. It may well be that the unifying force in the polyglot armies recruited by Tiberius Constantine was initially the Latin language, till the barbarian soldiers became better acquainted with Greek, the lingua franca of Arabs, Armenians and Syrians.

There is an interesting example of biculturalism in an inscription found outside the land walls of Constantinople that is worth quoting in full. A terminus ante quem of the Muslim Arabs’ capture of Tripoli inter 708-711 has been suggested, but palaeographically the inscription may belong much earlier:304

There is also evidence of Latin in civil society, and in the civil administration of the empire. The law school at Berytus in Phoenice Maritima was preparing students for administrative jobs until the great earthquake of 551. It is likely that the most recent classes remained in the civil service for several decades thereafter, leaving a certain residue of Latinity behind them in the officia of the provincial governors. The use of Latin terminology in Tiberius Constantine’s Novel of 575 is a possible reflection of this. There is also the striking example of a Greek-Latin glossary of Vergil’s Aeneid dating from the sixth century found among the otherwise almost exclusively Greek papyri at Nessana, a provincial kastron and military outpost in the Negev desert in Palaestina III that enjoyed contacts with the Mediterranean coast through the seaport of Gaza.309 Decisive evidence of a multilingual society is found in the funerary epigraphy of Syracuse in Sicily, to cite the example of an important western provincial capital where both Greek and Latin were used in the inscriptions of persons whose professions are mentioned. The relevant inscriptions are unfortunately not always dated.310 It seems best therefore to avoid simplistic and all-encompassing assumptions about the linguistic competence of army personnel without further inquiry into the evidence, which — if not rich — suggests a certain amount of cultural diversity.

[Here lies] Theodore of blessed memory who after serving as a soldier sailed from Loutr[-] in the province of Mauretania, Zargenos Zarkiano fleeing name (?) in the part of Tripoli. He died on the twenty-fifth day of October. The obscurity of the onomastics, sentence structure and toponymy should not distract us from the fact that a soldier from one of the Mauretanian provinces (Caesariensis, Sitifensis or Tingitana) where the predominant language of the provincials was Latin ended up in Constantinople and was commemorated with a Greek funerary inscription. There is in fact another Latin inscription from the reign of Tiberius Constantine mentioning the construction of a castrum in Numidia, the province immediately east of Mauretania Sitifensis.305 Theodore’s age at the time of his death is unknown, so he may well have continued to serve as a soldier after arriving in Constantinople, quite probably as a Latin speaker. It is worth asking how many other Greek- and Latin-speaking officers and men made similar reverse cultural journeys in the late sixth century and after. One thinks of the elder Heraclius who became the magister militum and praetorian prefect of Africa, and a certain Sergius, who was quite possibly an easterner and evidently held the roughly equivalent posts of prefect and dux (ÐpÄ ÑpÇqjym jaà do³n) in Italy sometime in the seventh century. He died at the age of fifty-three and was buried in Rhegium.306

Appendix IV. The Persian army’s use of the ApameaBeroea highway in 540 and 573.

There is also the problem of soldiers with ethnically diverse ‘barbarian’ names that cannot always be traced to

The first discussion I am aware of that dealt with the chronology of the plague versus the impact of war in the Massif Calcaire in Syria I and II is found in the footnotes

303

307

Simocatta, Hist. V 10. 4 (Whitby-Whitby 1986: 145 and note 48). This is my provisional view based on the clear script of the inscription. I take it that squadrons of troops from Mauretania were transferred eastward, as far as Cyrenaica, to oppose Arab raids against the praetorian prefecture of Africa. Kalkan and ùahin 1995: no. 9. 305 ILCV I, no. 28. Durlait 1981: no. 29 (30). 306 SEG 13, no. 469.

Kalkan-ùahin 1995: no. 1. ILCV 241, 447. Casson and Heltlich 1950 (P.Nessana II) nos. 1-2. Some of the Nessana non-literary papyri are of Umayyad date and are bilingual in Arabic and Greek. 310 Agnello 1953: nos. 18, 21, 24, 32, 37, 71, 74, 75, 76, 79, 80, 82, 84, 93, 95, 97.

304

308 309

352

FRANK R. TROMBLEY: THE OPERATIONAL METHODS OF THE LATE ROMAN ARMY IN THE PERSIAN WAR OF 572-591 of an article published in 1997.311 It was argued that the plague reached these provinces at least two years after Khusrau I’s devastating march on Antioch and Apamea in the spring and early summer of 540, and that Khusrau’s raid rather than the plague accounts for a decade’s lapse in building activity, as reflected in the Greek epigraphy of the region. This article was mainly interested in establishing the impact of the sixth-century Persian Wars on the provinces of Syria I and II. It was necessary, however, to deal with the chronology of the plague provisionally because there was likely to be controversy about the causes of the seeming economic stagnation in this region after 539.312 The main theses were brevis verbis:313

/ Litharba … ostwärts von Babisqa auf diese Route traf. This statement gives a very misleading idea of the main theses of the 1997 article, where it is made clear right at the start that the Sasanid withdrawal from Syria in 540 took the Roman military highway from Apamea to Beroea, and not through Babisqa:317 In the later stages of the campaign [of 540] the Persians turned up the [Orontes] river, reached Apamea and sacked it, thence retiring north-eastward on the Roman road to Chalkis and Beroia. The 1997 article indicates that it was not lost on Roman military architects that the territory of Apamea was potentially open to attack along the military highway from Beroea. Fortresses were constructed covering this route at ‘IjƗz, al-HabbƗt and Androna between 546/7 and 558/9.318 The notion that anyone has argued that Khusrau I’s army retired through Babisqa in Jabal Barisha in 540 therefore exists only in the minds of the authors of the 1999 article.

The central problem is the historical probability that military activity, in the form of the systematic deportation of agricultural and urban populations…destroyed a unique local culture in the invasions of 540, 574, and 611-613. Purely military explanations can never be entirely satisfactory, but the state of the evidence, including the chronology of the inscriptions, will remain suggestive if not determinative unless a suitable paradigm can account for the demographic implications of the epigraphy, such as the recurrence of the bubonic plague from 542 onward.

Bibliography Sources

The 1997 article reports:314

al-TabarƯ, Ta’rƯkh al-rusul wa’l-mulnjk. In C.E. Bosworth (trans.) The History of al-TabarƯ V. The SƗsƗnids, the Byzantines, the Lakmids, and Yemen, Albany. Chronicon anonymum as annum Christi 1234. (Ed. I. B. Chabot: Louvain, 1937). Chronicon Paschale. (Ed. W. Dindorf: Bonn, 1832; trans. L.M. and M. Whitby: Liverpool, 1989). Eustratius, Vita S. Eutychii (BHG 657) (Ed. J. P. Migne, Patrologia Graeco-Latina 86, 2273-2390). Evagrius, HE. (Trans. L.M. Whitby, The Ecclesiastical History of Evagrius Scholasticus: Liverpool, 2000). The Greek Anthology (Ed. and trans. W. R. Paton I-V: Cambridge, Mass. – London, 1916-1918). John Malalas, Chronographia (Ed. J. Thurn: Berlin, 2000). John of Ephesus, HE. (Ed. E. W. Brooks, Historiae Ecclesiasticae Pars Tertia, Corpus Scriptorum Christianorum Orientalium 105, Scriptores Syri 5455: Louvain, 1935-1936; trans. R. Payne Smith The Third Part of the Ecclesiastical History of John of Ephesus: Oxford, 1860). John of Epiphaneia, Fragmenta (Ed. L. Dindorf Historici Graeci Minores I: Leipzig, 1870: 375-382). Ps.-Joshua the Stylite, Chronicle (Trans. F. R. Trombley and J. W. Watt The Chronicle of Pseudo-Joshua the Stylite: Liverpool, 2000). Maurikios, StratƝgikon (Ed. G. T. Dennis and trans. E. Gamillscheg Das Strategikon des Maurikios: Vienna,

An adjoining structure was added to the East Church at Babisqa in Djebel Barisha in 540. The latter was built in the Antiochene month of Hyperberetaios (October = Tishri I), postdating Khusrau I’s march through the Limestone Massif in June and withdrawal later that summer. A more recent article by D. Stathakopoulos and E. Kislinger establishes a comparative chronology of the plague and the Persian War of 540-544.315 I would call the reader’s attention to a criticism of this analysis:316 F. R. Trombley … will … in einer Bauinschrift aus Babisqa (auf der Route Beroia / Aleppo-Antiocheia … vom Oktober 540 ein Indiz sehen, ‘postdating Khusrau I’s march … and withdrawal later that summer’. Chosroes hat jedoch auf seinem Rückmarsch von Apamaea über Chalkis diesen Streckenabschnitt gar nicht benutzen müssen, da er vom Süden kommend beim Knotenpunkt von Atharib

311

Trombley 1997. Oral communication, Michael Crawford, 1997. Trombley 1997: 201f., 206, and notes 8, 85, 97, 170, 187. 313 Trombley 1997: 166. A subsequent analysis revealed the existence of a Syriac inscription at Babisqa dating from 547. Littmann 1904 (AAES IV): nos. 14-15, with corrections listed in Littmann 1934): 64f. Cf. Trombley 2004: 341-344. 314 Trombley 1997: 166. 315 Kislinger and Stathakopoulos 1999: 76-98. 316 Kislinger and Stathakopoulos 1999: 80, note 23. 312

317 318

353

Trombley 1997: 156. IGLS 1598, 1609, 1682. Trombley 1997: 177-180.

THE LATE ROMAN ARMY IN THE EAST Curta F. 2001 The Making of the Slavs. History and Archaeology of the Lower Danube Region, c. 500700, Cambridge. Dagron G. and Feissel D. 1987 Inscriptions de Cilicie, Paris. Dillemann L. 1962 Haute Mésopotamie orientale et pays adjacents, Paris. Durliat J. 1981 Les dédicaces d’ouvrages de défense dans l’Afrique Byzantine, Rome. Dvoichenko-Markov D. 1984 The Vlachs: the Latin speaking population of eastern Europe. In Byzantion 54: 508-526. Eadie J.W. 1967 The development of Roman mailed cavalry. In Journal of Roman Studies 57: 161-173. Feissel D. 2000 Les édifices de Justinien au témoignage de Procope et de l’épigraphie. In Antiquité Tardive 8: 81-104. Frendo J.D. 1989 Theophylact Simocatta on the revolt of Bahram Chobin and the early career of Khusrau II. In Bulletin of the Asia Institute 3: 77-88. Frösén J., Arjava A. and Lehtinen M. (ed.) 2002 The Petra Papyri I, Amman. Fuller M.S. 1988-89 Tell Tuneinir on the Khabur: preliminary report on three seasons. In Annales Archéologiques Arabes de Syrie 38-39: 279-290. Gatier P.-L. 1986 IGLS XXI. Inscriptions de la Jordanie. Tome 2. Région centrale (Amman-Hisban-MadabaMain-Dhiban), Paris. Geyer B. (ed.) 2001 Conquête de la steppe et appropriation des terres sur les marges arides du Croissant fertile, Lyon-Paris. Gheyn J. van den (ed. and trans.) 1900 Acta Graeca S. Dometii Martyris. In Analecta Bollandiana 19: 285318. Goubert P. 1949 Les rapports de Khosrau II, roi des rois sassanide, avec l’empéreur Maurice. In Byzantion 19: 78-98. Goubert P. 1951 Byzance avant l’Islam I. Byzance et l’Orient sous les successeurs de Justinien, Paris. Greatrex G. and Lieu S.N.C. (ed.) 2002 The Eastern Roman Frontier and the Persian Wars, London. Gregory S. 1997 Roman Military Architecture on the Eastern Frontier, II, Amsterdam. Grosse R. 1920 Römische Militärgeschichte von Gallienus bis zum Beginn der byzantinischen Themenverfassung, Berlin. Grumel V. 1958 Traité d’études Byzantines I. La chronologie, Paris. Guillou A. 1969 Régionalisme et independence dans l’empire byzantin et de la Pentapole d’Italie, Rome. Guillou A. 1996 Recueil des inscriptions grecques médiévales d’Italie, Rome. Haldon J. 1995 Administrative continuities and structural transformations in East Roman military organisation ca. 580-640. In J.F. Haldon, State, Army and Society in Byzantium, Aldershot [= F. Vallett and M. Kazanski (ed.), L’armée romaine et les barbares du 4e au 7e siècle: Paris 1993, 45-51]. Heuser B. 2002 Reading Clausewitz, London.

1981; trans. G. T. Dennis Maurice’s Strategion. Handbook of Byzantine Military Strategy: Philadelphia, 1984). Menander Protector, Fragmenta (Ed. and trans. R. C. Blockley The History of Menander the Guardsman: Liverpool, 1985). Notitia Dignitatum (Ed. O. Seeck: 1876; repr. Frankfurt am Main, 1962). Novellae (Ed. R. Schoell, Corpus Iuris Civilis III: Berlin, 1972). Sebeos, History (Trans. R. W. Thomson and J. HowardJohnston The Armenian History Attributed to Sebeos: Liverpool, 1999; trans. F. Macler Histoire d’Héraclius: Paris, 1904). Theophanes Confessor, Chronographia (Ed. C. de Boor I-II: Leipzig, 1883; trans. C. Mango and R. Scott: Oxford, 1997). Theophanes of Byzantium, Fragmenta (Ed. L. Dindorf Historici Graeci Minores I: Leipzig, 1870, 446-449). Theophylact Simocatta, Historiae (Ed. C. de Boor: Leipzig, 1887; trans. L.M. Whitby and M. Whitby The History of Theophylact Simocatta: Oxford, 1986). Bibliography Adonts‘ N. 1970 Armenia in the Period of Justinian. The Political Conditions Based on the NaȤarar System, Lisbon. Agnello S.L. 1953 Silloge di iscrizione paleocristiane della Sicilia, Rome. Anderson J.G.C. 1897 The road system of eastern Asia Minor with the evidence of Byzantine campaigns. In Journal of Hellenic Studies 17: 22-44 and Plate I. Beševliev V. 1964 Spätgriechische und spätlateinische Inschriften aus Bulgarien, Berlin. Bivar A.D.H. 1972 Cavalry equipment and tactics on the Euphrates frontier. In Dumbarton Oaks Papers 26: 271-291. Bulliet R. 1975 The Camel and the Wheel, New York. Butler A.J. 1978 The Arab Conquest of Egypt, ed. P.M. Fraser, Oxford. Calvet Y. and Geyer B. 1992 Barrages de Syrie, LyonParis. Cameron A. (ed.) 1995 The Byzantine and Early Islamic Middle East III. States, Resources and Armies, Princeton. Casey P.J. 1996 Justinian, the limitanei, and ArabByzantine relations in the 6th c. In Journal of Roman Archaeology 9: 214-222. Casson L. and Heltlich E.L. (ed.) 1950 Excavations at Nessana II: Literary Papyri, Princeton. Chapot V. 1901 Antiquités de la Syrie du Nord. In Bulletin de Correspondance Hellénique 26: 161-208, 289-290. Christensen A. 1944 L’Iran sous les Sassanides, Copenhagen. Clausewitz C. von 1976 On War, trans. M. Howard and P. Paret, Princeton.

354

FRANK R. TROMBLEY: THE OPERATIONAL METHODS OF THE LATE ROMAN ARMY IN THE PERSIAN WAR OF 572-591 Hild F. and Restle M. 1981 Kappadokien (Kappdokia, Charsianon, Sebasteia und Lykandos). Tabula Imperii Byzantini 2, Vienna. Hill D.R. 1975 The role of the camel and horse in the early Arab conquests. In V.J. Parry and M.E. Yapp (ed.), War, Technology and Society in the Middle East, Oxford: 32-43. Hoffmann D. 1969-70 Das spätrömische Bewegungsheer und die Notitia Dignitatum, 1-2, Düsseldorf. Howard-Johnston J. 1995 The two great powers in late antiquity: a comparison. In Cameron 1995: 157-226. Inostrancev C.A. 1926 The Sasanid military theory. L. Bogdanov (trans.). In Journal of the K. R. Cama Oriental Institute 7: 7-52. Jireþek J. 1911 Geschichte der Serben, I, Gotha. Jones A.H.M. 1964 The Later Roman Empire 284-602: A Social, Economic, and Administrative Survey, Oxford. Kaegi W. 2003 Heraclius Emperor of Byzantium, Cambridge. Kaegi W. 1973 New evidence on the early reign of Heraclius. In Byzantinische Zeitschrift 66: 308-330. Kalkan H. and ùahin S. 1995 Epigraphische Mitteilungen aus Istanbul II: Kreuzförmige Grabstellen an Konstantinupolis. In Epigraphica Anatolica 24: 137148. Keegan J. 1976 The Face of Battle, New York. Kislinger E. and D. Stathakopoulos D. 1999 Pest und Perserkriege bei Prokop. Chronologische Überlegungen zum Geschehen 540-545. In Byzantion 69: 76-98. Kitzinger E. 1954 The cult of images before iconoclasm. In Dumbarton Oaks Papers 8: 83-150. Kraemer C. (ed. and trans.) 1958 Excavations at Nessana III. Non-Literary Papyri, Princeton. Lee A.D. 1993 Information and Frontiers. Roman Foreign Relations in Late Antiquity, Cambridge. Liddell Hart B. 1967 Strategy: The Indirect Approach, London. Lightfoot C.S. 1983 The site of Roman Bezabde. In S. Mitchell (ed.), Armies and Frontiers in Roman and Byzantine Anatolia, BAR IS 156, Oxford: 189-204. Littmann E. 1904 The Publications of an American Expedition to Syria in 1899-1900. Part IV: Semitic Inscriptions, New York. Littmann E. 1934 Syria. Publications of the Princeton University Archaeological Expeditions to Syria in 1904-5 and 1909. Division IV: Semitic Inscriptions, Section B: Syriac Inscriptions, Leiden. Littmann E., Magie D. and Stuart D.R. 1921 Syria. Publications of the Princeton University Archaeological Expeditions to Syria in 1904-5 and 1909, Division III: Greek and Latin Inscriptions, Section A: Southern Syria, Leiden. Margiotta F.D. (ed.) 1996 Brassey’s Encyclopedia of Land Forces and Warfare, Washington, D.C.London. Maspero J. 1912a L’organisation militaire de l’Égypte byzantine, Paris.

Maspero J. 1912b Voideq°toi et Stqati´tai dans l’armée byzantine au VIe siècle. In Byzantinische Zeitschrift 21: 97-109. Minorsky V. 1944 Roman and Byzantine campaigns in Atropatene. In Bulletin of the School of Oriental and African Studies, University of London 11: 243-265. Monchambert J.-Y. 1984 Le futur lac du Moyen Khabour: Rapport sur la prospection archéologique menée en 1983. In Syria 61: 181-218. Morris D.R. 1994 Washing of the Spears, London. Mouterde R. and Poidebard A. 1945 Le limes de Chalcis. Organisation de la steppe en Haute Syrie romaine, III, Paris. Năsturel P.ù. 1966 Quelques mots de plus à propos du tËqma, vqÇtqe de Théophylacte et de Théophane ‘ tËqma, tËqma, vqÇtqe..’ . In Byzantinobulgarica 2: 217-222. Oikonomides N. 1971 Correspondence between Heraclius and KavƗdh-ŠiroƝ in the Paschal Chronicle (628). In Byzantion 41: 269-281. Olajos T. 1981 Le monument du triomphe de Trajan en Parthe. In Acta Antiqua Academiae Scientiarum Hungaricae 29: 379-383. Olajos T. 1988 Les sources de Théophylacte Simocatta historien, Leiden. Oleson J.P. 1976 An unpublished Sassanian seal with a comment on the deportation of Armenians to Cyprus in A.D. 578. In Levant 8: 161-164. Kazhdan A. (ed.) 1991 The Oxford Dictionary of Byzantium, Oxford. Pohl W. 1988 Die Awaren. Ein Steppenvolk in Mitteleuropa 567-822 n.Chr., München. Poidebard A. 1939 La trace de Rome dans le désert de Syrie. Le limes de Trajan à la conquête arabe, Paris. Prentice W.K. 1908 The Publications of an American Archaeological Expedition to Syria in 1899, Part III: Greek and Latin Inscriptions, New York. Ramsay W.M. 1890 The Historical Geography of Asia Minor, London. Riedlberger P. 1998 Die Restauration von Chosroes II. In E. Dąbrowa (ed.), Electrum II. Ancient Iran and the Mediterranean World, Cracow: 161-175. Rosetti A. 1968 Istoria limbii române de la origini pină în secolul al XVII-lea, Bucharest. Rothstein G. 1899 Die Dynastie der Lahmiden in alHira, Berlin. Roussel P. 1939 Un monument d’Hiérapolis-Bambykè relative à la paix ‘perpétuelle’ de 532 ap. J.-C. In Mélanges syriens offerts à Monsieur René Dussaud, 1, Paris: 367-372. Rubin Z. 1995 The reforms of Khusro AnnjshirwƗn. In Cameron 1995: 227-297. Schmitt O. 2001 Stärke, Struktur und Genese des comitatensischen Infanterienumerus. In Bonner Jahrbücher 201: 93-110. Smail R.C. 1956 Crusading Warfare, 1097-1193, Cambridge.

355

THE LATE ROMAN ARMY IN THE EAST Whitby M. 1988 The Emperor Maurice and his Historian. Theophylact Simocatta on Persian and Balkan Warfare, Oxford. Whitby M. 1994 The Persian king at war. In E. Dąbrowa (ed.) The Roman and Byzantine Army in the East, Cracow: 227-263. Whitby M. 1995 Recruitment in Roman armies from Justinian to Heraclius (ca. 565-615). In Cameron 1995: 61-124. Whitcomb D. 1999 Notes on QinnsarƯn (sic) and Aleppo in the early Islamic period. In Annales Archéologiques Arabes de Syrie 43: 203-209.

Stein E. 1919 Studien zur Geschichte des byzantinischen Reiches, vornehmlich under den Kaisern Justinus II. und Tiberius Constantinus, Stuttgart. Stillwell R. (ed.) 1976 The Princeton Encyclopedia of Classical Sites, Princeton. Sun Tzu 1963 The Art of War. S. B. Griffith (trans.), Oxford. Szádeczky-Kardoss S. 1976 Bemerkungen zur Geschichte (Chronologie und Geschichte) der sassanidisch-byzantinischen Kriege (in den siebziger Jahren des sechsten Jahrhunderts). In Acta Antiqua Academiae Scientiarum Hungaricae 24: 109-114. Tate G. 1999 La route de la soie en Syrie au VIème siècle. In Annales Archéologiques Arabes de Syrie 43: 195-201. Tjader J.-O. 1955 Die nichtliterarischen lateinischen Papyri Italiens aus der Zeit 445-700, I, Lund. Trombley F.R. 1985 Monastic foundations in Anatolia and their role in the social and economic life of the countryside. In Greek Orthodox Theological Review 30: 45-59. (Reprinted in N.M. Vaporis (ed.), Byzantine Saints and Monasteries, Brookline, Massachusetts: 45-59). Trombley F.R. 1994 Hellenic Religion and Christianization c. 370-529, II, Leiden. Trombley F.R. 1997 War and society in rural Syria c. 502-613: observations on the epigraphy. In Byzantine and Modern Greek Studies 21: 154-209. Trombley F.R. 1999 Ammianus Marcellinus and fourthcentury warfare: a protector’s approach to historical narrative. In J.W. Drijvers and D. Hunt (ed.), The Late Roman World and its Historian: Interpreting Ammianus Marcellinus, London: 17-28. Trombley F.R. 2004a Demographic and cultural transition in the territorium of Antioch, 6th-10 c. In Topoi Supplement 5: Antioche. Histoires, images et traces de la ville antique, Lyon-Paris: 341-362. Trombley F.R. 2004b Epigraphic data on village culture and social institutions: an interregional comparison (Syria, Phoenice Libanensis and Arabia). In W. Bowen, L. Lavan and C. Machado (ed.), Recent Research on the Late Antique Countryside, Leiden: 73-101. Trombley F.R. 2005 The Late Roman practice of war on the Syrian frontier (A.D. 502-641): leadership, infrastructure and operations. In B. Meissner, O. Schmitt and M. Sommer (ed.), Krieg - Gesellschaft Institutionen Beiträge zu einer vergleichenden Kriegsgeschichte, Berlin: 387-416. Turtledove H. 1983 Justin II’s observance of Justinian’s treaty of 562. In Byzantinische Zeitschrift 76: 292301. Velkov V. 1977 Cities in Thrace and Dacia in Late Antiquity (Studies and Materials), Amsterdam. Whitby M. 1986 Procopius’ description of Dara (Buildings II, 1-3). In P. Freeman and D. Kennedy (ed.), The Defence of the Roman and Byzantine East. BAR IS 297 II, Oxford: 737-783.

356

Una guarnigione bizantina a Roma nell’età di Gregorio Magno: i Theodosiaci* Pietrina Pellegrini Università della Basilicata, Potenza

1. Nell’ottobre del 584 papa Pelagio II scriveva all’allora apocrisario della Sede pontificia a Costantinopoli, il venerabilis diaconus Gregorio, affinché informasse l’imperatore Maurizio di quante calamitates ac tribolationes tutta l’Italia stesse soffrendo a causa della perfidia dei Longobardi di fronte alla quale sembrava impossibile porre rimedio. Roma, in particolare, era sprovvista di ogni presidio e l’esarca, che riusciva a mala pena a difendere le zone intorno a Ravenna, certo non era nelle condizioni di potersi occupare dell’Urbe. Se almeno l’imperatore avesse potuto inviare un magister militum e un dux!1 Ma era necessario agire in fretta prima che l’esercito di quella spregevole gente riuscisse ad occupare i territori che ancora appartenevano all’impero.2

Spoleto, non lasciava spazio a esitazioni e Gregorio non si dava pace nel vedere che l’esarca Romano non solo trascurava di arginare l’irruenza dei nemici, ma gli impediva di intavolare trattative per una tregua. La strategia dell’esarca di spostare le linee di difesa lungo la via Flaminia per proteggere il corridoio che collegava Roma a Ravenna non convinceva il papa, allarmato nel vedere la sua città rimanere priva di difese. Scriveva allora Gregorio al vescovo di Ravenna Giovanni di intervenire presso Romano perché lo autorizzasse a stipulare un trattato con Ariulfo. Parte delle milizie era stata allontanata da Roma e a difesa dell’Urbe era rimasta la sola unità dei Theodosiaci che, non avendo ancora ricevuto la paga militare, a stento si occupava della custodia delle mura. E la città, abbandonata da tutti, se non aveva la pace, come avrebbe potuto resistere?5

Quello stesso Gregorio che aveva rappresentato all’imperatore le istanze e le preoccupazioni di Pelagio II nel 590 era salito al soglio pontificio. Non era ancora cambiata la politica di Costantinopoli nei confronti dell’Italia e il papa, subito dopo la sua elezione, con una lettera allo scolasticus Paolo manifestava di nuovo l’urgenza di organizzare la difesa di Roma. Al bisogno di fronteggiare la pressione delle spade dei nemici, si aggiungeva la necessità di reprimere la rivolta dei soldati di guardia all’Urbe.3

2. La presenza a Roma di una guarnigione denominata Theodosiaci costituisce una delle pochissime testimonianze di truppe bizantine attive in Italia alla fine del VI secolo. Si tratta di un numerus, cioè di una piccola unità militare al comando di un magister militum.6 Oltre alla lettera gregoriana, due papiri provenienti da Ravenna e una iscrizione funeraria rinvenuta a Roma attestano l’esistenza di un numerus di Theodosiaci. I papiri 16 e 17 pubblicati da J.-O. Tjäder, databili intorno al 600, sono il primo un atto di donazione alla chiesa di Ravenna di un fondo situato nei pressi di Rimini da Iohannes, spatarius quondam Georgii magistri militum et nunc primicerius numeri felicum Theodosiacus (sic!);7 il secondo è sempre una donazione di terreni che la gloriosissima femina Flavia Santippe offre ai mansionarii della basilica di S.

Finalmente nel 592 Maurizio aveva potuto esaudire le richieste che da tempo giungevano da Roma e l’invio del magister militum Casto, comandante delle guarnigioni imperiali,4 aveva ridato speranza alle aspettative del papa. La crescente minaccia dei Longobardi di Ariulfo, duca di *

Ringrazio Salvatore Cosentino, Sylvain Janniard, Ariel S. Lewin e Mario Mazza per gli utili suggerimenti ricevuti durante la stesura di questo lavoro dei cui risultati sono la sola responsabile. 1 Per Durliat 1979: 316-318, i termini magister militum e dux sono equivalenti. L’autore mette in evidenza come dei diciassette magistri militum conosciuti grazie alla corrispondenza di Gregorio Magno ben sei sono ricordati come duces dell’Italia bizantina. Ferluga 1991: 384, scrive, invece, che sia i magistri militum sia i duces ebbero inizialmente competenze militari; ma l’obbligo per i duces di restare a lungo nella regione loro assegnata finì per investirli anche di incarichi amministrativi. La conseguenza fu che i duces furono allo stesso tempo magistri militum ed è per questo motivo che la corrispondenza di Gregorio Magno attesta una sovrapposizione dei due titoli. Per un’utile rassegna bibliografica sulle diverse posizioni riguardo a questo problema cfr. Cosentino 1996-2000: I, 61, n. 272. Più di recente Zuckerman 2002: 172, ha sostenuto che dopo la metà del VI secolo il titolo di magister militum venne conferito ai duces come promozione onorifica. 2 Greg. M. epist. app., (Ewald – Hartmann II, 440, 1). 3 Greg. M. epist., I 3 (Norberg 4, 14): Veniente autem uiro magnifico domno Maurentio chartulario, ei quaeso in Romanae urbis necessitate concurrite; quia hostilibus gladiis foris sine cessatione confodimur, sed seditione militum interno periculo grauius urguemur. 4 Sul magister militum Casto cfr. Greg. M. epist., III 51 (Norberg 197, 26); V 30 (Norberg 297, 23) e V 36 (Norberg 307, 88).

5 Greg. M. epist., II 38 (Norberg 123, 27): Si quando tamen est aliquando locus optinendi, agat apud eum, fraternitas uestra, ut pacem cum Ariulfo, si ad aliquid parum possumus, faciamus, quia miles de Romana urbe tultus est, sicut ipse nouit. Theodosiaci uero qui hic remanserunt, rogam non accipientes, uix ad murorum quidem custodiam se accomodant, et destituta ab omnibus ciuitas, si pacem non habet, quomodo subsistet? Sulle convulse vicende che intercorsero fra l’Italia e l’impero d’Oriente negli anni finali del VI secolo cfr. le sintesi di Diehl 1888; Hartmann 1889 e Goubert 1965. In particolare sulla situazione di Roma restano ancora fondamentali i lavori di Bertolini 1941: 225-229 e 240-261; Bertolini 1952; Llewellyn 1971; Bavant 1979; Jenal 1988. Più di recente Noble 1995 prende in esame la situazione di Roma nell’età successiva al pontificato di Gregorio Magno (604-701), ma il suo lavoro offre una utile panoramica anche sul periodo precedente. Vd. inoltre Ravegnani 2004. Sulla politica di Gregorio Magno nei confronti dei Longobardi cfr. Richards 1984: 267286; Markus 2001: 113-124. 6 Per le occorrenze del termine numerus in Gregorio Magno cfr. epist., IV 37 (Norberg 259, 53): in numeris militasse; VIII 10 (Norberg 527, 17): ex militaribus numeris; dial., IV 36, 7 (Vogüé I, 120, 46): in numero optio fuit. 7 Pap. 16 (Tjäder I, 318-327 e commento 449-453) = pap. 90 (Marini 139-141 e commento 295-299, in partic. 297, n. 9).

357

THE LATE ROMAN ARMY IN THE EAST

Cripta di S. Cecilia in Trastevere, Roma CIL VI 32970 stata accettata nella sostanza da Tjäder,12 nuovo editore dei documenti ravennati il quale, nel suo commento al papiro 16, replicava a Marini facendo notare che al nome di uno dei due imperatori Teodosio erano da riferirsi i Felices Theodosiani iuniores ricordati nella ND Or. IX 41 e attestati a Firenze in un’iscrizione del 547 con il nome di numerus primus Theodosianorum.13 Aggiungeva infine che sempre la ND Or. VIII 27 ricordava una truppa di Equites Theodosiaci iuniores, sotto il comando del magister militum per Thracias, che avrebbe potuto avere relazione con l’unità d’Italia.

Maria Maggiore a Roma. Fra i firmatari dell’atto compaiono, tra gli altri, Theo/datus, adorator numeri Theodosiac(i).8 L’iscrizione funeraria di Roma del 22 febbraio 597, invece, conservata nell’ipogeo della basilica di S. Cecilia in Trastevere, ricorda Pasca[sia…] / coniux Vitaliani primicerii et autenta(e) numeri [felicum?…] / Theodosiac(orum).9 Intorno all’origine e alla provenienza di questa unità si è acceso un complesso dibattito che non è riuscito a far convergere le opinioni degli studiosi. Già G. Marini agli inizi del 1800, nella sua edizione dei papiri diplomatici, aveva raccolto ed esaminato le testimonianze sui Theodosiaci presenti in Italia avanzando la proposta che si trattasse di truppe istituite da uno dei due Augusti di nome Teodosio.10 Non condividevano questa interpretazione i primi editori del Registrum epistolarum di Gregorio Magno, P. Ewald e L.M. Hartmann, i quali, riprendendo le osservazioni di Alteserra e Marini alla lettera gregoriana,11 ritenevano che l’unità dei Theodosiaci avesse preso nome da Teodosio, figlio maggiore dell’imperatore Maurizio. Questa lettura era

Intanto nel 1935 H. Colt, durante una campagna di scavo a Nessana nel Sud della Palestina, riportava alla luce nei magazzini di alcune chiese cristiane una serie di papiri in lingua greca di straordinario valore. Pubblicati nel 1958 da C. Kraemer, fra questi documenti si distingueva un corpus di 17 testi che dovevano appartenere all’archivio di una unità militare definita nel papiro 15 Colt: ajriqmou' tw'n kaqosiwm(evnwn) Qeodosiakw'n: il numerus Theodosiacus.14 Si trattava, secondo l’interpretazione di 12

Tjäder 1955: 451, n. 12. Precedentemente Rasi 1937: 45, n. 2, avanzava l’ipotesi che i Theodosiaci ricordati da Gregorio Magno potessero anche essere stati reclutati in Italia. 13 CIL XI 1693 (= ILCV Diehl 486 = ILS Dessau 2806): b. [m.] / hic requiescit / in pace Macrobis / primicerius primi Tho/dosianorum numeri, qui / vixet annis pl. M. quinqu/aginta tantum; deposit. / est sb Idus Maias sex. P. c. Bas/ili v.c. ind. decima. Sulla storia dei rinvenimenti epigrafici nell’area di S. Felicita a Firenze cfr. Gunnella 1994. Diversamente da Gunnella, Ciampoltrini 1989: 249, n. 14, dubita della effettiva pertinenza al sepolcreto della nostra iscrizione dal momento che mancano dati sicuri di provenienza. 14 Pap. 15 in Kraemer 1958: 43, 3: Fl(avouio") [Stevfano" ºAbraamivou str]atiwvth["] ajriqmou' tw'n kaqosiwm(evnwn) Qeodosiakw'n. Su kaqosiwmevno" come devotissimus cfr. Hornickel 1930: 18. La devotio era l'atteggiamento psicologico che si richiedeva ai soldati e ai bassi

8

Pap. 17 (Tjäder: 327-334 e commento 453-455) = pap. 91 (Marini 141-142 e commento 299-302). Il testo della donazione, copiato dall’originale papiraceo, fu trascritto su una iscrizione del IX sec. Sul significato del termine adorator cfr. Marini 1805: 302, n. 16; Tjäder 1955: 455, n. 12; Jones 1964: III, 203, n. 116; Ravegnani 1988: 120. 9 CIL VI 32970 (= ILCV Diehl 489 = ICUR2 I 131): loCVM QVE EXILARATO PASCA / sia CONIVNX VITALIANI CONPARAVIT A PRIMICERII ET AUTENTA NUMERI felic. / THEODOSIAC ‚ SOL ‚ IIII ‚ ET TRIM ‚INO REQUIESCT DEP ‚ VIII KAL ‚ MAR ‚ IND / XV. Attualmente è visibile solo una delle due lastre che dovevano comporre l’iscrizione. Per il significato di autenta, aujqevnth", comandante subalterno con funzione di capo di un numerus, cfr. Babut 1913: 260, n. 3; Bellen1961: 243. 10 Marini 1805: 296. 11 Greg. M. epist., II 45 (Ewald-Hartmann I, 145, n. 2).

358

PIETRINA PELLEGRINI: UNA GUARNIGIONE BIZANTINA A ROMA NELL’ETÀ DI GREGORIO MAGNO: I THEODOSIACI Kraemer, di una guarnigione attivata tra il 421 e il 442 da Teodosio II dopo i suoi successi in due campagne militari contro i Persiani (421-422 e 441-442). I Theodosiaci erano truppe di limitanei reclutati fra i residenti, che agendo come unità mobile contro i beduini, si occupavano della difesa dei campi, delle città e delle carovane. L’accampamento, stanziato a Nessana come posto di guardia per i popoli del deserto del Negev, verso il 590 era stato abbandonato in conseguenza del cambiamento della politica estera di Maurizio che, per rafforzare le difese contro i Persiani, nel 582 aveva tagliato il sussidio militare. Poteva esserci un collegamento fra il numerus Theodosiacus presente a Nessana fino al 590 e l’unità che nel 592 difendeva la città di Roma? Con questo interrogativo L. Casson, che aveva curato la pubblicazione del secondo volume dei ritrovamenti di Nessana, terminava sull’American Journal of Philology del 1957 la recensione al libro di Tjäder sui documenti ravennati.15 Casson, anticipando alcuni risultati del lavoro di Kraemer che da lì ad un anno sarebbero stati pubblicati, si chiedeva insomma se il numerus smantellato a Nessana potesse essere stato riattivato in Italia. Oppure, dal momento che a partire dal 590 il numerus Theodosiacus non esisteva più, se fosse stata creata a Roma una nuova unità che aveva preso il nome dal figlio dell’imperatore Maurizio.

ha rivisto e corretto in molti punti l’interpretazione dei papiri Colt escludendo ogni connessione fra l’unità di Nessana, di cui è conservato l’archivio militare, e il numerus Theodosiacus.18 Nel P. Colt 15 del 30 maggio del 512, l’unico a riportare il nome di questa unità, si legge infatti che due fratelli nativi di Nessana si arruolarono nell’esercito e servirono nell’unità dei Theodosiaci stanziata a Rhinocorura, città che, sulle coste del Mediterraneo al confine tra le province di Palaestina Tertia e d’Egitto, costituiva un emporio per il commercio con l’Arabia e proteggeva le rotte fra Siria ed Egitto. Dopo aver completato il loro servizio militare, in età avanzata, i due fratelli erano tornati a Nessana portando con sé i loro documenti personali. Questo unico cenno al numerus Theodosiacus nei papiri Colt riguarda dunque l’unità militare di Rhinocorura e non quella di Nessana. Anche gli argomenti portati avanti da Kraemer, continua Negev, per attribuire a Teodosio II l’istituzione del numerus non sono convincenti. Le due guerre contro i Persiani che avrebbero risolto l’imperatore a creare una nuova unità perché fosse stanziata nel deserto, in realtà furono campagne brevi e insignificanti dal punto di vista militare. Queste puntuali indicazioni ci consentono di escludere la possibilità di ricostruire le vicende del numerus Theodosiacus attraverso l’archivio militare di Nessana e quindi ci obbligano a distinguere le storie delle due unità: se infatti le truppe di Nessana vennero smantellate nell’ultimo decennio del VI secolo – come risulta dalla documentazione dell’archivio –, i Theodosiaci di Rhinocorura non necessariamente subirono la stessa sorte.19 Inoltre, ciò che per noi è più importante, se davvero i Theodosiaci di Rhinocorura furono truppe di limitanei, essi non poterono essere trasferiti, al pari di un esercito mobile, nelle zone più indifese dell’impero. Ne consegue con sufficiente ragionevolezza la possibilità di respingere ogni collegamento fra il numerus Theodosiacus di Rhinocorura e la guarnigione di Roma.

La questione restava per allora irrisolta e si arricchiva di nuove, ulteriori interpretazioni. A.H.M. Jones riteneva che i Theodosiaci di Roma potessero identificarsi con una delle unità presenti nella Notitia Dignitatum: i Felices Theodosiani o i Primi Theodosiani, ma più probabilmente gli Equites Theodosiaci. Le truppe di Nessana in Palestina, invece, erano senza dubbio i Ballistarii Theodosiaci dell’esercito di campo orientale citati in ND Or. VII 57.16 Nel suo lavoro sull’esarcato della Pentapoli d’Italia del 1969, A. Guillou, a sua volta, ipotizzava che i Theodosiaci presenti in Italia fossero soldati di origine gota che traevano il loro nome dalla città di Teodosia a est della penisola di Crimea, l’antico Chersoneso Taurico. Soldati di questo numero, scriveva Guillou, si trovavano anche a Nessana in Palestina almeno fino alla fine del VI secolo e nel 592 erano stati incaricati della difesa di Roma.17

3. Nel tentativo di risalire alla provenienza della truppa ricordata nel Registrum gregoriano, cerchiamo ora di mettere ordine alle sporadiche notizie che ci sono giunte e all’intreccio di interpretazioni che si sono susseguite. Anzitutto è necessario esaminare in maniera più accurata i documenti a nostra disposizione per verificare se i Theodosiaci, il numerus primus Theodosianorum e il numerus felicum Theodosiacus attestati in Italia possano

Nuovi elementi però hanno contribuito più di recente a chiarire alcuni aspetti importanti del problema. A. Negev livelli della gerarchia amministrativa verso l'imperatore. Sul significato di devotus in Gregorio Magno cfr. Conti 1971: 128-134. 15 Casson L., AJPh 78, 1957: 103-108, in partic. 108. G.F. Gilliam, nella sua recensione al volume di Kraemer apparsa in CPh 54, 1959, 180-183, in partic. 182, escludeva invece ogni collegamento fra l’unità dei papiri di Nessana e le truppe d’Italia. 16 Jones 1964: III, 202, n. 111. L’autore avanza questa ipotesi riportando erroneamente quanto è scritto nel pap. 16 Tjäder. In esso, come abbiamo visto in precedenza, viene ricordata l’unità dei Felices Theodosiaci e non dei Felices Theodosiani. Condivide comunque l’interpretazione di Jones Ravegnani 1988: 30-31; Ravegnani 2004: 39-43; Ravegnani 2004a: 39-43. 17 Guillou 1969: 153 ss. L’autore in un suo successivo lavoro, Guillou 1980: 263 ss., tace sul collegamento fra l’unità di Roma e quella di Nessana.

18 Negev 1988: II, 1-2. Sulla situazione delle difese dell’impero nel Vicino Oriente tardoantico cfr. Isaac 1990: 207-213; un’ampia sintesi sulla trasformazione delle città e del territorio del Negev è proposta da Lewin 2002. 19 Isaac 1995: 148 ritiene che non ci siano ragioni per sostenere che l’unità di Nessana venne smantellata nel 590; al contrario il P. Colt 15 dimostra che la truppa in quella data è ancora attiva. Su questo vd. anche Rubin 1997: 64-66. Palme 2004: 165 ha studiato la presenza di Theodosiaci Isauri ad Alessandria escluendo ogni identificazione di questa unità con i Theodosiaci di Nessana.

359

THE LATE ROMAN ARMY IN THE EAST identificarsi con la stessa unità.20 Delle tre testimonianze provenienti da Roma, la prima in ordine di tempo è la lettera di Gregorio Magno nella quale la truppa a difesa dell’Urbe è definita semplicemente Theodosiaci. Si potrebbe spiegare l’assenza dell’aggettivo felix con il fatto che Gregorio, rivolgendosi ad un vescovo, non sentisse la necessità di indicare con precisione la nomenclatura ufficiale dell’unità. Tuttavia il termine felix manca di certo nel papiro 17 Tjäder della basilica di S. Maria Maggiore, sempre in Roma; inoltre l’iscrizione CIL VI 32970 presenta una lacuna che è stata integrata con felicum solo per analogia con l’unità di Ravenna. Messa in dubbio, dunque, la presenza dell’aggettivo felix nella titolatura del numerus di stanza a Roma alla fine del VI secolo,21 bisogna chiedersi quale rapporto esista fra queste truppe, il numerus felicum Theodosiacus di Ravenna e il numerus primus Theodosianorum di Firenze.

È difficile valutare se la variazione del suffisso –iacus e – ianus presente nella Notitia Dignitatum costituisca una differenza scrupolosamente rispettata nella denominazione delle unità militari. In un discusso articolo sulle truppe legate al nome di Costantino, Ralf Scharf ha ritenuto che i termini Constantiniaci e Constantiniani non siano intercambiabili, ma che la distinzione esista e venga mantenuta perché funzionale alla necessità di riferire con esattezza il nome di ogni unità.31 Questa posizione è stata contestata da Constantin Zuckerman che non crede nella perfetta rigidità militare immaginata da Scharf.32 Senza entrare nel merito della discussione, ricordo solo che, per quanto più da vicino ci interessa, proprio riguardo ai Theodosiaci Scharf fa notare come nel papiro 16 Tjäder l’aggettivo Qeodwsiakov" sia reso con Theodosiacus e non con Theodosianus. Il papiro cioè indica che la traduzione dal greco al latino del nome delle truppe abbia privilegiato il suffisso –iacus. Bisogna osservare inoltre che a distinguere le truppe di Roma e di Ravenna da quelle di Firenze non è solo il suffisso –iacus e –ianus; ma anche l’aggettivo primus presente nell’unità di Firenze. Tutti i documenti che abbiamo preso in esame hanno carattere ufficiale e difficilmente, a mio parere, si poteva registrare con approssimazione il nome della truppa. Si può perciò con ragionevolezza concludere che il numerus primus Theodosianorum di Firenze sia un’unità diversa rispetto ai Theodosiaci di Roma e al numerus felicum Theodosiacus di Ravenna tanto più se si considera la possibilità di riconoscere nella guarnigione fiorentina gli Equites primi Theodosiani o i Primi Theodosiani della Notitia Dignitatum:33 non è escluso che queste truppe giunsero in Italia in occasione della guerra greco-gotica.34

Un’indagine sulle unità militari dell’impero tardoantico non può prescindere dalla lettura della Notitia Dignitatum. Il testo ricorda diverse truppe legate al nome di Teodosio, tutte attive in Oriente, e che potrebbero avere relazione con le nostre guarnigioni: sono numeri di cavalieri – Equites Theodosiaci seniores,22 Equites Theodosiaci iuniores e Equites primi Theodosiani –;23 e di fanti – Primi Theodosiani,24 Felices Theodosiani,25 Ballistarii Theodosiaci,26 Felices Theodosiani iuniores e Balistarii Theodosiani iuniores.27 Come è evidente, la Notitia non menziona esplicitamente un numerus di Felices Theodosiaci. Si può ragionare allora su due possibilità: che i termini Theodosiani e Theodosiaci si equivalgano – e allora i Felices Theodosiaci potrebbero identificarsi con i Felices Theodosiani della Notitia Dignitatum –;28 oppure che i Theodosiaci di Roma o il numerus felicum Theodosiacus di Ravenna, siano nuove truppe istituite successivamente alla redazione della Notitia Dignitatum da Teodosio II29 o da Maurizio in onore del figlio, Teodosio anch’egli, nominato correggente insieme al padre il 26 marzo del 590.30

una nuova unità. I Primi Theodosiani ricordati a Firenze nel 547 dimostrano che ancora nel VI secolo fosse possibile trovare in Italia truppe legate al nome di Teodosio I o Teodosio II. Tuttavia Hoffmann ricorda che l’imperatore Maurizio nel 589 stabilì come corregente suo figlio Teodosio e anche da questo imperatore le truppe potevano aver preso il nome. 31 Scharf 1997: 206, n. 47. 32 Zuckerman 1998a. 33 Gli Equites Primi Theodosiani in ND Or. VIII 32 (Seeck 25), sono una uexillatio comitatensis sotto il magister militum per Thracias; i Primi Theodosiani in ND Or. V 64 (Seeck 14) sono un auxilium palatinum al comando del magister militum praesentalis. 34 Sull’insediamento a Firenze di una guarnigione bizantina e sulla battaglia del Mugello nel 541 che vide l’esercito bizantino sconfitto cercare scampo dentro le mura di Firenze riferisce Proc. BG III 5 (Haury 318, 1). Grosse 1920: 122, n. 8, proponeva di identificare l’unità di Firenze con l’auxilium palatinum dei Primi Theodosiani menzionati nella Notitia Dignitatum. Di diverso parere Hoffmann 1969-1970: I, 242, secondo il quale i Primi Theodosiani di Firenze sono un numerus dell’esercito bizantino in Italia, probabilmente la uexillatio comitatensis degli Equites primi Theodosiani al comando del magister militum per Thracias citati da ND Or. VIII 32. A conforto dell’interpretazione di Hoffmann si può ricordare il ruolo determinante che la Tracia, anche durante la guerra greco-gotica, rivestì come base per il reclutamento di truppe da spostare in Italia. Secondo Proc. BG, III 10, 1 (Haury 336, 20), Belisario nel 544, mentre si apprestava a giungere in Italia, riuscì a raccogliere in Tracia 4000 volontari. Su questo episodio e le difficoltà di coscrizione durante il VI secolo cfr. Fotiou 1988. Sull’esercito bizantino durante la guerra greco-gotica cfr. Müller 1912; Hannestadt 1960; Teall 1965, e, più di recente, Ravegnani 2002. Amory 1997, pone l’accento sulla presenza nell’esercito di molti gruppi etnici diversi, fra i quali sottolinea la folta rappresentanza proprio di Traci e di Illirici.

20 Di questo parere gli editori di CIL VI 32970; Diehl 489 e Tjäder 1955: 451 n. 12. 21 Sono di questo avviso Hoffmann 1969-1970: II, 97, n. 349 e Scharf 1997: 206 n. 47. Avevano già distinto l’unità di Ravenna da quella di Roma Diehl 1888: 197, n. 14 e Brown 1984: 91, n. 19. 22 ND Or. VI 33 (Seeck 17), al comando del magister militum praesentalis II. 23 Rispettivamente ND Or. VIII 27 (Seeck 25) e VIII 32 (Seeck 25) al comando del magister militum per Thracias. 24 ND Or. V 23 (Seeck 12) al comando del magister militum praesentalis I. 25 ND Or. VI 21=62 (Seeck 16 e 18) al comando del magister militum praesentalis II. 26 ND Or. VII 21=57 (Seeck 20 e 22) al comando del magister militum per Orientem. 27 Rispettivamente ND Or. IX 41 (Seeck 29) e IX 47 (Seeck 29) al comando del magister militum per Illyricum. 28 È questa una delle ipotesi avanzata da Jones 1964: III, 202 n. 111, che però, come abbiamo scritto supra alla n. 16, non è condivisibile. 29 Zuckerman 1998: 137-147, in partic. 144 ss., propone come data di redazione del documento il periodo tra la fine del 398-inizi 399 e il 401. 30 Hoffmann 1969-1970: I, 242, resta aperto a molte interpretazioni. Egli non esclude che Teodosio II, durante il suo lungo regno, abbia istituito

360

PIETRINA PELLEGRINI: UNA GUARNIGIONE BIZANTINA A ROMA NELL’ETÀ DI GREGORIO MAGNO: I THEODOSIACI Il numerus felicum Theodosiacus ravennate a sua volta solleva alcuni problemi riguardo al nome. Il termine Theodosiacus, nelle diverse occorrenze attestate in pap. 16 Tjäder, non è mai declinato. Ciò non può essere interpretato come un errore grammaticale perché il redattore del documento della donazione dimostra di conoscere il latino e rispettarne le desinenze. D’altro canto i soldati, che ovviamente sapevano bene la denominazione della propria unità di appartenenza, potevano verificare e correggere un’eventuale svista. Ne consegue che la truppa di Ravenna abbia il nome, non altrove con certezza attestato, di numerus felicum Theodosiacus. Rimane aperta l’ipotesi sostenuta ancora di recente da Cosentino,35 che questa unità possa identificarsi con i Theodosiaci ricordati da Gregorio Magno. La truppa potrebbe essere stata istituita dall’imperatore Maurizio in onore del primogenito Teodosio, nominato Augusto nel 590 dal patriarca di Costantinopoli, e battezzato da Gregorio Magno secondo la notizia riportata da Gregorio di Tours36. Del resto ancora nel VI secolo l’imperatore Tiberio creò una unità cui diede il suo nome.37

efficace. Menandro Protettore ricorda due ambascerie provenienti da Roma per perorare l’ennesima richiesta di aiuto. Nel 577 una delegazione del senato romano guidata dal patrizio Pamphronio aveva inviato una somma consistente di denaro a Costantinopoli per pagare l’allestimento e l’invio di un esercito. Tiberio, però, non avendo soldati da mandare in Italia, aveva consigliato a Pamphronio di impiegare quella somma per guadagnare i capi longobardi alla causa romana e convincerli a prendere servizio in Oriente fra le truppe imperiali.39 L’invito, rifiutato dai Longobardi, era stato esteso e accolto dai principi franchi che per un breve periodo riuscirono a contenere a Nord l’avanzata nemica.40 Qualche anno dopo, nel 579, alcuni senatori e sacerdoti provenienti da Roma si recarono di nuovo dall’imperatore per implorare l’invio di un contingente militare a difesa della città. La guerra contro i Persiani, scrive Menandro Protettore, travagliava l’Armenia e l’Oriente e le truppe bizantine erano appena sufficienti a contenere l’assalto dei nemici. Questa volta Tiberio riuscì a reclutare una piccola unità che risultò però inadeguata; fece invece grandi sforzi per convincere i capi longobardi a combattere a fianco di Costantinopoli promettendo loro lauti guadagni.41 Ancora nel 579, dunque, l’impero non era nelle condizioni di arruolare truppe da inviare a Roma; anzi cercava in Italia eserciti di mercenari disposti a prendere servizio in Oriente.

4. Tuttavia questa ricostruzione non esclude la possibilità di percorrere un’altra strada per cercare di risalire all’origine dei Theodosiaci di Roma. È necessario però estendere la ricerca oltre i confini occidentali dell’impero per comprendere le ripercussioni della politica estera di Bisanzio sulle vicende d’Italia alla fine del VI secolo.

Con l’avvento di Maurizio crebbe l’attenzione per la sicurezza dell’Italia. Nonostante fosse ancora prioritaria la questione persiana, l’imperatore non aveva rinunciato all’Occidente e con la costituzione dell’esarcato di Ravenna aveva tentato di assicurare all’Italia quella organizzazione militare e politica che le consentisse di difendere i propri confini. Ma anche le forze a disposizione dell’esarca risultarono ben presto insufficienti a fermare i Longobardi.

Il giorno prima di morire, l’imperatore Tiberio sceglieva come augusto uno dei suoi più abili e fidati generali, il comes excubitorum Maurizio. Insieme al potere, Maurizio ereditava una difficile situazione di politica estera che rendeva fragili i confini d’Oriente e metteva a repentaglio le conquiste di Giustiniano in Occidente. Alla pressione dei Persiani e degli Avari, si era aggiunta nel 568 l’invasione dei Longobardi in Italia.38 Già Tiberio, consapevole della difficoltà di contenere i nemici su tanti fronti di guerra, aveva scelto di combattere i Persiani e di ricorrere alla diplomazia e all’aiuto degli alleati barbari per assicurarsi il controllo dell’Occidente. Ma la sua politica era risultata poco

L’ultimo decennio del VI secolo rappresentò un momento di svolta per la difficile situazione orientale. In seguito ad un conflitto dinastico scoppiato fra Cosroe di Persia e Varanes, satrapo della Media Atropatene, nel 591 i persiani deposero le armi.42 Restava ancora aperto il problema della sicurezza nei Balcani.43 A questo

35

39

Ioh. Ephes., h.e. VI 13 (Brooks 234, 15-19), ricorda che 60.000 soldati longobardi, in occasione della campagna contro i Persiani del 585, avrebbero promesso di militare nelle armate bizantine. Sui rapporti diplomatici fra Bisanzio e Longobardi cfr. Bognetti 1967: 453 ss. 40 Men. Protec. fr., 22 (Blockley 196). Sulla figura di Pamphronio cfr. Bertolini 1968: 230-233 e Goffart 1957. 41 Men. Protec. fr., 24 (Blockley 216). Di certo il denaro offerto da Maurizio riuscì a guadagnare alla causa bizantina i capi longobardi Autarit e Nordulfo. Essi, riferisce Gregorio Magno in epist., II 38 (Norberg 123, 10), dopo aver militato nelle truppe dell’impero, erano tornati a combattere accanto ai loro fratelli di sangue chiedendo però un pari trattamento economico. 42 Seb. hist., cc. 9-10, (Gugerotti 59 ss). Su SebƝos cfr. più di recente Thomson - Howard-Johnston: 1999. Su questa fase della guerra contro i Persiani cfr. Goubert 1965: I, 122-170; sui problemi di cronologia cfr. Higgins 1939: 42-54. Vd. inoltre Christensen 1936: 436-490. 43 Per una dettagliata esposizione della politica estera di Maurizio nei Balcani dopo il 592 cfr. Goubert 1961-1964.

Cosentino 1996-2000: I, 67, n. 307, ritenendo che l’iscrizione CIL VI 32970 contenga l’aggettivo felicum – ma ciò in realtà, come abbiamo cercato di spiegare, non è sicuro –, suppone, sulla base delle parole riportate da Gregorio Magno in epist., II 38 (Norberg 123, 30): Theodosiaci uero qui hic remanserunt, che una parte dei Theodosiaci di Roma fosse stata trasferita a Ravenna. 36 Greg. Tour. HF X.1 (Krusch – Levison 407, 11): Unde factum est, ut epistolam ad imperatorem Mauricium dirigeret, cuius filium ex lavacro sancto susciperat. 37 Thphn. chron., A.M. 6074 (de Boor 251, 24): oJ de; basileu;" Tibevrio" ajgoravsa" swvmata ejqnikw'n katevsthse stravteuma eij" o[noma i[dion. La notizia è ripresa da Cedr. hist., (Bekker 690, 12). 38 Sulla politica estera di Maurizio cfr. Bury 1889: II, 83-142; Jones 1964: I, 308-315. Per un’analisi delle guerre combattute durante il regno di Maurizio contro Avari, Slavi e Persiani, cfr. lo studio, condotto a partire dall’opera di Teofilatto Simocatta e degli storici che da lui dipendono, di Whitby 1988. Più in particolare per la difesa del limes Arabicus vd. Shahîd 1995: 394-617.

361

THE LATE ROMAN ARMY IN THE EAST termini della ferma.48 Fin qui è la testimonianza delle fonti. È probabile a questo punto che, rafforzata la difesa sui Balcani e divenuto più sicuro il confine orientale – ed è questa la nostra ipotesi –, Maurizio abbia avuto la possibilità di esaudire le richieste che giungevano da Roma e sia riuscito a inviare un piccolo contingente militare per difendere le mura dell’Urbe. Sempre Teofilatto Simocatta aggiunge un indizio decisivo proprio a sostegno di questa interpretazione. Egli ricorda fra i protagonisti della guerra contro gli Avari un comandante di nome Casto che aveva servito agli ordini di Comenziolo, magister militum per Thracias.49 Nell’estate del 587 ad Anchialo – si badi, la stessa città in cui Maurizio qualche anno dopo convogliò le truppe provenienti dall’Oriente –, Casto venne messo alla testa di uno dei tre corpi in cui era stato diviso l’esercito di Tracia con il compito di sorvegliare i movimenti degli Avari presso il Monte Aimos. In un primo tempo riuscì a conseguire un discreto successo; ma divenuto in seguito più massiccio l’attacco dei nemici, venne sconfitto e fatto prigioniero.50 La notizia della cattura fece scalpore a Costantinopoli e suscitò un diffuso malcontento fra i sostenitori di Casto. L’imperatore fu pubblicamente attaccato con canti di scherno e ritenuto responsabile dell’insuccesso della spedizione. Ma l’insulto non suscitò l’ira, scrive Simocatta a difesa di Maurizio: questi però dovette pagare agli Avari un riscatto eccezionalmente esoso al fine di ottenere la liberazione del prigioniero.51 Simocatta e gli altri cronisti e storici bizantini, per lo più poco attenti nelle loro narrazioni alle vicende della parte occidentale dell’impero, non aggiungono altre notizie sulla sorte di questo generale le cui vicende però possiamo seguire proprio attraverso l’epistolario di Gregorio Magno. Il Casto che difese la Tracia nel 587 può verosimilmente identificarsi con il magister militum che dal 593 al 595 fu al comando delle truppe dei Theodosiaci di Roma.52 Diverse sono, a mio avviso, le ragioni a sostegno di questa identificazione. Casto era stato magister militum vacans prima di giungere a Roma e proprio a questi ufficiali, rimasti temporaneamente senza una comando territoriale e spesso a capo di modesti contingenti di truppe,53 erano conferiti già in età giustinianea compiti speciali in Italia.54 Inoltre egli era inviso all’imperatore e la sua nomina a magister militum

proposito Teofilatto Simocatta, lo storico più importante del regno di Maurizio, in un breve passaggio ripreso più tardi da Teofane e Giorgio Cedreno,44 racconta che l’imperatore, dopo aver concluso la pace con i Persiani, trasferì in gran fretta il suo esercito ad Anchialo, in Tracia, dove gli Avari stavano preparando una nuova irruzione: ta;" dunavmei" oJ aujtokravtwr ej" th;n Eujrwvphn wJ" tavcista metebivbazen ejpiv te th;n ºAgcivalon th;n ejkdhmivan paraskeuavzetai: diegnwvkei ga;r to; ºAbariko;n au\qi" ejqevlein foita'n.45 Non possiamo non restare colpiti dal significato che questo asciutto passaggio può rappresentare per le linee che fino a qui siamo venuti tracciando. Nella sua icastica brevità, la notizia dello spostamento in Tracia di truppe fino ad allora impegnate nella guerra persiana ci consente di fare qualche semplice considerazione e di proporre una ipotesi che, seducente, potrebbe anche rivelarsi fondata. Nel 591 Maurizio firma la pace con i Persiani e trasferisce velocemente il suo esercito in Tracia per rafforzare il confine balcanico. Segue un periodo di breve e relativa tranquillità per l’impero: probabilmente gli Avari stanno ricevendo dai Romani un tributo;46 di certo sono impegnati a difendere i loro confini dall’aggressione dei Turchi.47 Dal suo canto Maurizio non abbassa la guardia e continua a ritenere prioritaria la questione dell’arruolamento. Con un editto emanato nel 592, esclude dalla carriera ecclesiastica e dalla vita monastica chiunque non abbia portato a compimento i suoi obblighi civili e militari: in particolare stabilisce che nessun soldato può farsi monaco prima che siano scaduti i

44 Thphn. chron., A.M. 6082 (de Boor I, 267, 31): eijrhvnh" de; baqeiva" th;n ajnatolh;n katalabouvsh", ºAbariko;" ejpi;; th;n Eujrwvphn ejkuvmaine povlemo". dia; tou'to oJ aujtokravtwr Maurivkio" ta;" dunavmei" ajpo; ajnatolh'" ejpi; th;n Qrav/khn methvgagen. Cedr. hist., (Bekker I, 695, 16). 45 Thphyl. Simoc. hist., V 16, 1 (de Boor 235). Per la traduzione e le note all’opera di Simocatta cfr. Whitby – Whitby 1986. Il trasferimento di truppe bizantine dalle frontiere orientali in Tracia in seguito alla pace persiana è ricordato anche da Seb. hist., c. 16 (Gugerotti 69). Sempre SebƝos riferisce che anche gli Armeni furono spostati in Tracia: hist., cc. 13-14 (Gugerotti 66 e 68), cc. 16-17 (Gugerotti 69 e 71), c. 28 (Gugerotti 81). 46 Thphyl. Simoc. hist., VI 4, 1-3 (de Boor 226): oiJ me;n th'" Sigghdovno" oijkhvtore" ajqrovai" ejfovdoi" tisi; tw'n Sklauhnw'n tou;" povnou" lhi?zontai kai; puri; paredivdosan ta; pro;" nautilivan touvtwn ejpivceira. diav toi tou'to poliorkou'si th;n Sigghdovna oiJ bavrbaroi, hJ de; povli" ej" tou[scaton ajfikomevnh kakou' ijscna;" ejpefevreto swthriva" ejlpivda". ejbdovmh de; hJmevra, kai; oJ Cagavno" toi'" barbavroi" prosevtatte th'" poliorkiva" ajpevcesqai genevsqai te wJ" aujtovn. ejpei; de; to; bavrbaron touvtwn aujthvkoon gevgonen, ajpolimpavnei to; povlisma crusw'n dareicw'n ciliavda" ajpenegkavmenon duvo travpezavn te crusovpaston kai; stolhvn. Le operazione degli Avari iniziarono a Singidunum nel 591 sul Danubio. Essi avevano attraversato il fiume con le navi costruite dagli Slavi, ma dopo una settimana lo Chagan acconsentì a ritirarsi ricevendo 2000 aurei, un tavolo dorato e una sontuosa veste. 47 Thphyl. Simoc. hist., VII 7, 7 (de Boor 256): qevrou" ejnestw'to" kata; tou'ton dh; to;n ejniauto;n oJ pro;" th'/ e{{w/ uJpo; tw'n Touvrkwn Cagavno" uJmnouvmeno" prevsbei" ejxevpemye Maurikivw/ tw'/ aujtokravtori, ejpistolhvn te suntetacw;" ejpinivkia ejnecavratten ejn aujth'/. Thphyl. Simoc. hist., VII 7, 9 (de Boor 257): ejpi; th'/ nivkh/ toivnun eij" mevga ajrqei;" kai; to;n Stembiscavgan suvmmacon poihsavmeno" to; tw'n ºAbavrwn e[qno" katedoulwvsato. Nell’estate del 595 Chagan, re dei Turchi, inviò a Maurizio un’ambasciata con una lettera nella quale si gloriava dei suoi trionfi sugli Avari.

48

Greg. M. epist., III 61 (Norberg 209, 8). Per i problemi di reclutamento di truppe durante il regno di Maurizio cfr. Haldon 1979; Whitby 1995; Carrié 1995. 49 Casto è definito da Simocatta ora locagov" (hist., II 10, 9, de Boor 90) ora taxiavrch" (hist., II 12, 4, de Boor 94). Stein 1949: 814-815, ritiene che locagov" corrisponda probabilmente a tribunus e taxiavrch" a dux. 50 Thphyl. Simoc. hist., II 11, 8-14 (de Boor 90) e II 12, 1-4 (de Boor 93); Thphn. chron., A.M. 6079 (de Boor 257, 11). 51 Thphyl. Simoc. hist., II 17, 5-7 (de Boor 103). 52 Sull’identificazione di Casto, comandante in Tracia, con il magister militum di Roma, cfr. Ferluga – Hellmann – Kämpfer – Ludat – Zernack 1977, s.v. Castus e Martindale 1992, s.v. Castus. Cosentino 1996-2000: s.v. Castus3, ritiene probabile questa identificazione. Pietri - Pietri 19992000: s.v. Castus 4, invece, non prendono in considerazione questa ipotesi. 53 Sul magister militum vacans e la moltiplicazione delle cariche onorifiche in età giustinianea cfr. Stein 1949: II, 428-432. 54 Stein 1949: II, 613.

362

PIETRINA PELLEGRINI: UNA GUARNIGIONE BIZANTINA A ROMA NELL’ETÀ DI GREGORIO MAGNO: I THEODOSIACI di Roma aveva il vantaggio da una parte di allontanare da Costantinopoli un personaggio scomodo, e dall’altra di garantire la carriera di un comandante che aveva un forte e, agli occhi di Maurizio, pericoloso seguito. La corrispondenza di Gregorio Magno del resto testimonia i molti contatti di Casto con l’Oriente. Il papa lo raccomandò a Prisco, patricius Orientis, che nel 588 aveva combattuto contro gli Avari come magister militum per Thracias,55 e lo difese davanti all’imperatore Maurizio dall’accusa di non essersi adoperato fino in fondo per difendere la città di Roma.56 Nel 595, inoltre, Casto presiedette alla distribuzione del soldo alle sue truppe che lo scribon Busa, giunto da Costantinopoli, fece su incarico di Maurizio.57

integrazione della lacuna presente nell’epigrafe conservata nella chiesa di S. Cecilia in Trastevere di Roma: non più numeri [felic(um)] / Theodosiac(orum) – come era stato fatto per analogia col numerus felicum Theodosiacus di Ravenna – ma, probabilmente, numeri [equit(um)] / Theodosiac(orum), secondo una formula impiegata anche in altre iscrizioni militari.60 Fonti Bekker I. 1838 Georgius Cedrenus. Compendium historiarum, Bonn. de Boor C. 1883-1885 Theophanis Chronographia, I-II, Leipzig. de Boor C. 1887 Theophylacti Simocattae Historiae, Leipzig. Brooks E.W. 1952 Iohannis Ephesini historiae ecclesiasticae, CSCO 105, Louvain. Brooks E.W. 1952 Iohannis Ephesini historiae ecclesiasticae, (CSCO 105), Louvain. Ewald P. – Hartmann L.M. 1887-1893 Gregorii I papae Registrum epistolarum, (MGH, Epist. I-II), Berlin. Gugerotti A. 1990 SebƝos. Storia. Traduzione dall’armeno, introduzione e note, Verona. Haury J. 1963 Procopii de bello ghotico, in Procopii Caesariensis Opera Omnia, II. De bellis libri V-VIII, Leipzig. Krusch B. – Levison W. 1965 Gregorius episcopus Turonensis, Historiarum libri X, (MGH, SRM I/1), Hannover. Marini G. 1805 I Papiri diplomatici, Roma. Norberg D. 1982 S. Gregorii Magni. Registrum Epistularum libri I-XIV, (CCSL 140-140A), Turnhout. Reisk I.I. 1829 Constantini Porphyrogeniti Imperatoris De Cerimoniis Aulae Byzantinae, I-II, Bonn. Schoell R. – Kroll G. 197210 Corpus Iuris Ciuilis: Novellae, Berlin. Seeck O. 1876 Notitia Dignitatum accedunt Notitia urbis Constantinopolitanae et Laterculi provinciarum, Berlin. Tjäder J.-O. 1955 Die nichtliterarischen lateinischen Papyri Italiens aus der Zeit 445-700, I-II, Lund. de Vogüé A. 1978-1980 Grégoire le Grand. Dialogues, (SC 251, 260, 265), Paris.

Ora, e qui si chiude la linea del ragionamento che siamo venuti tracciando, questa identificazione ci consente di giungere probabilmente ad una conclusione. Secondo Simocatta nel 587 Casto, mentre si trovava in Tracia, ricevette dal magister militum per Thracias un contingente di 2000 uomini. Dal momento che la Notitia Dignitatum ricorda fra le truppe al comando del magister militum per Thracias anche un numerus di Equites Theodosiaci iuniores, si potrebbe forse riconoscere in questa unità la guarnigione che, al comando di Casto, difese le mura di Roma nell’età di Gregorio Magno.58 Del resto la Tracia rivestì durante il VI secolo un ruolo chiave per il reclutamento e per la difesa dell’Occidente.59 Se si accetta questa lettura, si può proporre una diversa 55 Greg. M. epist., III 51 (Norberg 197, 26): Vir autem gloriosus Castus praeco laudum uestrarum in Romanis partibus exsistit, quem uestrae excellentiae paterna dilectione commendo. 56 Greg. M. epist., V 36 (Norberg 307, 87): Sed de gloriosis uiris Gregorio praefecto praetorio et Casto magistro militum non mediocriter sum afflictus, qui et omnia quae potuerunt fieri nullomodo facere neglexerunt et labores uigiliarum et custodiae ciuitatis in obsessione eadem uehementissimos pertulerunt et post haec omnia graui dominorum indignatione percussi sunt. 57 Greg. M. epist., V 30 (Norberg 297, 8). 58 Haldon 1984: 238, ricorda come Const. Porph. cerim., II 44 (Reisk 663, 2): oJ tourmavrch" tw'n Qeodosiakw'n; (Reisk 663, 15): ajpo; tou' bavndou th'" touvrma" tw'n Qeodosiakw'n, menzioni fra le truppe che nel 949 presero parte alla spedizione contro Creta anche i Theodosiaci, probabilmente da identificare con gli equites Theodosiaci iuniores. Ciò dimostra che questa unità continuò ad essere attiva per molto tempo. 59 Kaegi 1981: 89-101, sottolinea anche come il magister militum per Thracias assunse un ruolo di sempre maggior peso divenendo quasi un antagonista rispetto al magister militum praesentalis. Amory 1997: 286288, ricorda che durante la guerra gotica molti erano i comandanti provenienti dalla Tracia, dall’Illiria e dalle regioni balcaniche. A dimostrazione dell’importanza della difesa del confine danubiano si può ricordare un importante passaggio riportato in un editto della Lex de dioecesi Aegyptiaca promulgata da Giustiniano nel 539. L’editto puniva col trasferimento sulle rive del Danubio e dell’Istro – così era chiamato il Danubio nel suo corso inferiore tra la Pannonia e la Mesia –, quelle unità militari che non avessero assolto ai loro oneri fiscali: Iust., Edict. XIII 11 (Schoell - Kroll 785): parakatasceqhvsontai ga;r au|tai para; tou' kata; kairo;n aujgoustalivou kai; eijskomisqhvsontai tai'" sai'" trapevzai" oijkeivw/ kinduvnw/, kai; dhvmeusin uJposthvsontai oi{ te lamprovtatoi aujtw'n tribou'noi kai; oiJ ejn aujtoi'" prwteuvonte" (oi{per kai; kefalikh;n timwrivan eujlabeivsqwsan), kai; to; pa'n tavgma metasta;n ejk th'" cwvra" ejn toi'" porrwtevrw tou' “Istrou h[toi Danoubivou potamou' tovpoi" metateqhvsetai, toi'" ejkei'se limivtoi" th'" parafulakh'" e}neka proskarterhvson. Per la traduzione in inglese e il commento di questo passo, riportato in realtà come editto 13, cfr. Thurman 1981: 54-55 e più di recente Demicheli 2000: 142.

Bibliografia Amory P. 1997 People and Identity in Ostrogothic Italy, 489-554, Cambridge. Babut Ch.E. 1913 Recherches sur la garde impériale et sur le corps d’officiers de l’armée romaine aux IVe et Ve siècles. In RH 114: 225-260. Bavant B. 1979 Le duché byzantin de Rome. Origine, durée et extension géographique. In MÉFRM 91: 4188. 60

Cfr. ad esempio CIL V 1591 proveniente da Grado: IOHANNIS | MILIS x | MERO x EQVIT | PERSO x IVS | TINIANI x VO | TVMx SOLVIT. Su questa iscrizione cfr. Rugo 1975: nr. 56a, p. 49 e Hoffmann 1961-62. Vd. inoltre AE 1962, 264 e AE 1990, 990.

DE NV

363

THE LATE ROMAN ARMY IN THE EAST Bellen H. 1961 Der primicerius Mauricius. Ein Beitrag zum Thebäerproblem. In Historia 10: 238-247. Bertolini O. 1941 Roma di fronte a Bisanzio e ai Longobardi, Bologna. Bertolini O. 1952 I papi e le relazioni politiche di Roma con i ducati di Spoleto e di Benevento. In RSCI 6: 1-46. Bertolini O. 1968 Appunti per la storia del Senato di Roma durante il periodo bizantino. In Id., Scritti scelti di storia medievale, a cura di O. Banti, I-II, Livorno: I, 227-262. Bognetti G.P. 1967 Tradizione longobarda e politica bizantina nelle origini del ducato di Spoleto. In Id., L’età longobarda, III, Milano: 439-475. Bury J.B. 1889 A History of the Later Roman Empire (395 A.D. to 800 A.D.), I-II, London. Brown T.S. 1984 Gentlemen and Officers. Imperial Administration and Aristocratic Power in Byzantine Italy A.D. 554-800, Roma. Carrié J.-M. 1995 L’État à la recherche de nouveaux modes de financement des armées (Rome et Byzance, IVe-VIIIe siècles). In Cameron Av. (ed.), The Byzantine and Early Islamic Near East. III: States, Resources and Armies, Princeton NJ: 27-60. Christensen A. 1936 L’Iran sous les Sassanides, Copenhague-Paris. Ciampoltrini G. 1989 Contributi per l’epigrafia tardoantica di Firenze. In Epigraphica 51: 246-250. Conti P.M. 1971 ‘Devotio e ‘viri devoti’ in Italia da Diocleziano ai Carolingi, Padova. Cosentino S. 1996-2000 Prosopografia dell’Italia bizantina (483-804), I-II, Bologna. Demicheli A.M. 2000 L’Editto XIII di Giustiniano in tema di amministrazione e fiscalità dell’Egitto bizantino, Torino. Diehl Ch. 1888 Études sur l’administration byzantine dans l’exarchat de Ravenne (568-751), Paris. Durliat J. 1979 Magister militum – 9:7':-1':-9 dans l’Empire Byzantin (VIe – VIIe siècles). In BZ 72: 306-320. Ferluga J. – Hellmann M. – Kämpfer F. – Ludat H. – Zernack K. (ed.) 1977 Glossar zur frühmittelalterlischen Geschichte im östlichen Europa. Seria A. Lateinische Namen bis 900, Wiesbaden. Ferluga J. 1991 L’organizzazione militare dell’esarcato. In Carile A. (ed.), Storia di Ravenna, II/1: Dall’età bizantina all’età ottoniana. Territorio, economia e società, Venezia: 379-387. Fotiou A. 1988 Recruitment Shortages in Sixth-Century Byzantium. In Byzantion 58: 65-77. Goffart W. 1957 Byzantine Policy in the West under Tiberius II and Maurice: The Pretenders Hermenegild and Gundovald. In Traditio 13: 73-118. Goubert P. 1965 Byzance avant l’Islam, II/2: Rome, Byzance et Carthage, Paris. Goubert P. 1961-1964 Les guerres sur le Danube à la fin du VIe siècle d’après Menandre le Protecteur et Théophylacte Simocatta. Actes du XIIe Congrès

International d’Études Byzantines, Ochride 10-16 septembre 1961, I-III, Beograde: II, 115-124. Grosse R. 1918 Die Rangordnung der römischen Armee des 4.-6. Jahrhunderts. In Klio 15: 122-161. Grosse R. 1920 Römische Militärgeschichte von Gallienus bis zum Beginn der byzantinischen Themenverfassung, Berlin. Guillou A. 1969 Régionalisme et indépendance dans l’Empire byzantin au VIIe siècle. L’exemple de l’exarchat et de la Pentapole d’Italie, Roma. Guillou A. 1980 L’Italia bizantina dall’invasione longobarda alla caduta di Ravenna. In Delogu P. – Guillou A. – Ortalli G. (ed.), Longobardi e Bizantini, G. Galasso (dir.), Storia d’Italia, Torino: I, 217-338. Gunnella A. 1994 Il complesso cimiteriale di S. Felicita: testimonianze di una comunità cristiana fiorentina. In Benvenuti A. – Cardini F. – Giannarelli E. (ed.), Le radici cristiane di Firenze, Firenze: 13-32. Haldon J.F. 1979 Recruitment and Conscription in the Byzantine Army C. 550-950. A Study on the Origins of the Stratiotika Ktemata, Wien. Haldon J.F. 1984 Byzantine Praetorians. An Administrative, Institutional and Social Survey of the Opsikion and Tagmata, c. 580-900, Bonn. Haldon J.F. 1995 Administrative Continuities and Structural Transformations in East Roman Military Organisation c. 580-640. In Id., State, Army and Society in Byzantium. London [reprint from Vallett F. – Kazanski M. (ed.), L’armée romaine et les barbares du IIIe au VIIe siècle. Paris 1993, 45-53]: 1-20. Hannestadt K. 1960 Les forces militaire d’après la guerre gothique de Procope. In Classica et Medievalia 21: 136-183. Hornickel O. 1930 Ehren- und Rangprädikate in den Papyrusurkunden. Ein Beitrag zum römischen und byzantinischen Titelwesen, Leipzig. Higgins M.J. 1939 The Persian War of the Emperor Maurice (582-602). Part I: The Chronology, with a Brief History of the Persian Calendar, Washington. Hartmann L.M. 1889 Untersuchungen zur Geschichte der byzantinischen Verwaltung in Italien 540-750, Leipzig. Hoffmann D. 1969-1970 Das spätrömische Bewegungsheer und die Notitia Dignitatum, I-II, Dusseldorf. Hoffmann D. 1961-1962 Der “numerus equitum Persoiustinianorum” auf einer Mosaikinschrift von Sant’Eufemia in Grado. In Aquileia Nostra 32-33: 8198. Isaac B. 1990 The Limits of Empire. The Roman Army in the East, Oxford. Isaac B. 1995 The Army in the Late Roman East: the Persian Wars and the Defence of the Byzantine Provinces. In Cameron Av. (ed.), The Byzantine and Early Islamic Near East. III: States, Resources and Armies, Princeton NJ: 125-155. Jenal G. 1988 Gregor der Grosse und die Stadt Rom. In Prinz F. (ed.), Herrschaft und Kirche: Beiträge zur Entstehung und Wirkungsweise episkopaler und monastischer Organisationformen, (Monographien 364

PIETRINA PELLEGRINI: UNA GUARNIGIONE BIZANTINA A ROMA NELL’ETÀ DI GREGORIO MAGNO: I THEODOSIACI zur Geschichte des Mittelalters, 33), Stuttgart: 109145. Jones A.H.M. 1964 The Later Roman Empire (284-602). A Social, Economic, and Administrative Survey, I-III, Oxford. Kaegi jr. W.E. 1981 Byzantine Military Unrest 471-843. An Interpretation, Amsterdam. Kraemer jr. C.J. 1958 Excavations at Nessana, III, Non-literary papyri, Princeton NJ. Lewin A. 2002 Il Negev dall’età nabatea all’epoca tardoantica. In MedAnt 5: 319-375. Llewellyn P. 1971 Rome in the Dark Ages, London. Markus R.A. 2001 Gregorio Magno e il suo mondo, trad. it., Milano. Martindale J.R. 1992 The Prosopography of the Later Roman Empire A.D. 527-641, Cambridge. Müller B. 1912 Das Heer Justinian nach Prokop und Agathias. In Philologus 71: 101-138. Negev A. 1988 The Architecture of Mampsis, Final Report, I: The Middle and Late Nabatean Period, II: The Late Roman and Byzantine Reports, Jerusalem. Noble Th.F.X. 1995 Rome in the Seventh Century. In Lapidge M. (ed.), Archbishop Theodore, (Cambridge Studies in Anglo-Saxon England, 11), Cambridge: 6887. Palme B. 2004 Theodosiaci Isauri in Alexandria. In Archiv für Papyrusforschung 16:157-163. Pietri Ch. - Pietri L. 1999-2000 Prosopographie chrétienne du Bas-Empire, Roma. Rasi P. 1937 ‘Exercitus Italicus’ e milizie cittadine nell’Alto Medioevo, Padova. Ravegnani G. 1988 Soldati di Bisanzio in età giustinianea, Roma. Ravegnani G. 2002 I corpi dell’esercito bizantino nella guerra gotica. In Medioevo greco 2: 155-175. Ravegnani G. 2004 I bizantini e la guerra. L’età di Giustiniano, Roma. Ravegnani G. 2004a I soldati bizantini in Italia, Bologna. Richards J. 1984 Il Console di Dio. La vita e i tempi di Gregorio Magno, trad. it., Firenze. Rubin R. 1997 Priests, Soldiers, and Administrators: Society and Institutions in the Byzantine Negev. In Mediterranean Historical Review 12: 56-74. Rugo P. 1975 Le iscrizioni dei sec. VI-VII-VIII esistenti in Italia, II: Venezia e Istria, Cittadella. Scharf R. 1997 Constantiniaci = Constantiniani? Ein Beitrag zur Textkritik der Notitia Dignitatum am Beispiel der “constantinischen” Truppen. In Tyche 12: 189-212. Shahîd I. 1995 Byzantium and the Arabs in the Sixth Century, I/1: Political and Military History, Washington D.C. Stein E. 1949 Histoire du Bas-Empire, II: De la disparition de l’Empire d’Occident à la mort de Justinien (476-565), Paris-Bruxelles-Amsterdam. Teall J.L. 1965 The Barbarians in Justinian’s Armies. In Speculum 40: 294-325. Thomson R.W. – Howard-Johnston J. (ed.) 1999 The Armenian History attribuited to Sebeos, Part I:

Translation and Notes, Part II: Historical Commentary, Liverpool. Thurman W. S. 1981 The Thirteen Edicts of Justinian Translated and Annotated, Ann Arbor-London. Whitby M. 1995 Recruitment in Roman Armies from Justinian to Heraclius (ca. 565-615). In Cameron Av. (ed.), The Byzantine and Early Islamic Near East, III: States, Resources and Armies, Princeton NJ: 61-124. Whitby M. 1988 The Emperor Maurice and His Historian. Theophylact Simocatta on Persian and Balkan Warfare, Oxford. Whitby M. – Whitby M. (ed.) 1986 The History of Theophylact Simocatta. An English Translation with Introduction and Notes, Oxford. Zuckerman C. 1998 Comtes et ducs en Égypte autour de l’an 400 et la date de la Notitia Dignitatum Orientis. In AntTar 6: 137-147. Zuckerman C. 1998a Constantiniani – Constantiniaci from Pylai. A Rejoinder. In Tyche 14: 255-258. Zuckerman C. 2002 La haute hiérarchie militaire en Afrique byzantine. In AntTar 10: 169-175.

365

Jews, Rats, and the Battle of Yarmnjk David Woods* Department of Ancient Classics, University College Cork

The so-called battle of Yarmnjk was one of the most important battles in world history. Their victory there allowed the Muslim Arabs to occupy greater Syria and to use it as the launching pad for further expansion into what had once been Byzantine territory. On the other hand, if the Byzantines had managed to defeat them there, this would probably have prevented the emergence of a new empire and a new world–faith. So why did the Byzantines fail so spectacularly at Yarmnjk? What allowed an enemy whom the Byzantines had traditionally despised as mere bandits to inflict such a decisive defeat upon them?1 Answers have varied, but it is generally accepted at present that a rebellion among the Byzantine troops at Yarmnjk fatally weakened them in the face of the Arab threat. The purpose of this paper is to draw attention to overlooked evidence which suggests that the Byzantine army failed because of plague. Furthermore, the discovery that plague severely weakened the Byzantine forces on the eve of the so-called battle of Yarmnjk has important consequences for our understanding of the date and location of this battle. It suggests that we should probably re-date this battle to 638 rather than 636, and that it actually occurred at or near the village of Gabatha (Jaba‘), about 19km southwest of Jerusalem, rather than at Gabitha in the Golan.

river, least of all the Yarmnjk. Similarly, the earliest Syriac accounts refer only to the town of Gabitha, and do not mention the river Yarmnjk.5 Furthermore, writing c.660, the anonymous author of the so-called Chronicle of Fredegar (‘Fredegar’ henceforth) clearly describes the same battle as that described by the two Spanish chroniclers, even if he does not specifically state its location at Gabatha or Gabitha. In contrast, the surviving Muslim sources identify the decisive battle as that fought on the northern bank of the river Yarmnjk.6 Furthermore, although the Byzantine chronicler Theophanes (d.818) and later Syriac sources, such as Michael the Syrian (d.1199) and the author of the Chronicle of 1234, do preserve some account of the battle of Gabitha, they then proceed to present the confrontation near the Yarmnjk as the decisive battle.7 So where was this last decisive battle fought, at Gabatha/Gabitha or on the banks of the Yarmnjk? The current solution to this problem assumes that this battle consisted of a series of engagements and manoeuvres over several days, if not weeks, which drew the combatants southwards from Gabitha (JƗbiya) to the banks of the Yarmnjk.8 There are two problems with this approach. First, and foremost, it ignores the central historiographic issue, the prior question as to the relationship between our sources for this period. The tacit assumption seems to be that the earliest Greek and Latin sources derive their information from the same ultimate source as the later Greek and Syriac sources, but that, for whatever reason, they do not bear quite so good a witness to this source as do these later texts. The idea that the reverse may in fact be the case, that it is the earliest Greek and Latin sources which bear the best witness to their ultimate common source, in this matter at least, while it is the later Greek and Syriac sources which err in this matter, does not seem to have received due consideration. This tendency to prefer the late sources at the expense of the early sources must strike any rational historian as a rather odd way in which to approach the subject.

The standard works of reference confidently report that the decisive battle in the struggle between the Byzantines and the Muslims for possession of Syria and Palestine was the battle of Yarmnjk, and that it occurred in 636.2 Such confidence conceals a welter of confused and contradictory evidence.3 The earliest surviving Latin sources to identify the decisive battle for the possession of Palestine and Syria by name, the Spanish chronicles of 741 and 754, locate it at Gabatha, while the earliest surviving Greek history to refer to it by name, the Breviarium which ‘Nicephorus’ wrote probably in the 780s, locates it at Gabitha.4 None of these mention any

The second problem with this approach is that it simply assumes that which it ought to prove, that is, that the Gabatha or Gabitha described by the earliest Greek,

*The research for this paper was made possible by my receipt of a Research Fellowship from the Irish Research Council for the Humanities and Social Sciences for the academic year 2004-2005. 1 On Arabs as raiders, see e.g. Diod. Sic. II.48; Strabo, Geog. XVI.1.26; Amm. Marc. XIV.4.1. 2 E.g. Ostrogorsky 1968: 111; Kazhdan 1991: 2214; Treadgold 1997: 303. 3 Hoyland 1997 provides an excellent introduction to each of the sources for this period. Space does not allow a full discussion of all the evidence here. 4 Chron. 741, 16; Chron. 754, 11; Niceph. Brev. 20. It is often claimed that, writing c.700, Anastasius the Sinaite refers to the battle at the Yarmnjk, but he actually refers to a battle at the town of Iermycha, southwest of Jerusalem. That is the obvious interpretation based on the actual spelling of the title and the failure of Anastasius to refer to it as a river. See Anastasius, Sermo adversus Monotheletas III.1.86. I do not

believe that the patriarch Nicephorus really did compose the Breviarium traditionally attributed to him, but that is an argument for another day. 5 See Palmer 1993: 3 and 57 for the notices in an early seventh-century chronicle and the Chronicle of ZuqnƯn which mention only ‘Gabitha’. 6 See Donner 1981: 128-148. 7 On Gabitha, see Theoph. Chron. AM 6125, 337; Michael XI.6, 415/420; Chron. 1234, I.243-244. On Yarmnjk, see Theoph. Chron. AM 6126, 338; Michael 11.6, 415-416/420-421; Chron. 1234, I.249-250. 8 E.g. Jandora 1985; Kaegi 1992: 112-114; Nicolle 1994: 61-83. Others, e.g. Stratos 1972: 55-56, distinguish between the battles of Gabitha and Yarmnjk.

367

THE LATE ROMAN ARMY IN THE EAST Latin, and Syriac sources, is identifiable as the Gabitha (JƗbiya) in the extreme northeast of the Byzantine province of Arabia rather than, for example, either the Gabatha (Jebata) southwest of Diocaesarea or that southwest of Jerusalem (Jaba‘), both in the province of Palestina Prima.9 One cannot emphasize enough that none of the earliest Greek, Latin, or Syriac sources provide any firm indication as to the location of the Gabatha or Gabitha which they mention. The purpose of this paper, therefore, is to explain how the Latin chronicles, particularly the Chronicle of 754, may preserve indirect evidence pointing to the identification of this town as the Gabatha southwest of Jerusalem. Whether or not one accepts this, however, they certainly preserve neglected evidence indicating that plague played an important role in severely weakening the Byzantine forces just before the decisive battle.

If we accept Fredegar at face value, it is not clear that there really was a battle at all, at least not in any meaningful sense. On the night before the two sides were expected to engage in battle, the ‘sword of God’ (gladio Dei) struck the army of Heraclius, and 52,000 men died in camp. The result was that the survivors were so dismayed by this apparent sign of divine judgement against them that almost immediately after advancing to battle the following day, they turned and fled. So what really happened? It seems hardly likely that either Fredegar, or any of his predecessors, really believed that an actual sword had appeared and struck down all of these men. This leaves two other possibilities. The first, and least likely, is that Fredegar’s text preserves a confused reference to the nickname of the supreme commander of the Muslim forces in Syria up to and during this battle, KhƗlid b. al-WalƯd, nicknamed Sayf AllƗh ‘Sword of God’.11 This is probably an unfortunate coincidence. The second is that we are dealing with a metaphor. The problem lies in establishing the basis for this metaphor. What sort of phenomenon would a Christian author have interpreted as a sign that God had turned His sword against His people? The answer to this may lie in the Biblical account of how God had allegedly decided to punish King David for taking a census of the people of Israel.12 God sent the prophet Gad to David in order to give him a choice of punishments, and David chose an epidemic:

The Sword of God The first step in the solution to the problems both of the reason for the Byzantine defeat at Gabatha/Gabitha and of the identity of the location of this battle lies in the reevaluation of a neglected detail in the description of this battle by Fredegar. He first describes how the emperor Heraclius raised fresh forces following an earlier defeat, but then launches into a description of the battle itself: He raised a great force throughout the imperial provinces and sent representatives to the Caspian gates, which the Macedonian Alexander the Great had built of bronze above the Caspian sea and had had shut to check invasion by the untamed barbarians living beyond the Caucasus. Heraclius ordered these gates to be opened, and through them poured 150,000 mercenary warriors to fight the Saracens. The latter, under two commanders, were approximately 200,000 strong. The two forces had camped quite near one another and were ready for an engagement on the following morning. But during that very night the army of Heraclius was smitten by the sword of the Lord: 52,000 of his men died where they slept. When, on the following day, at the moment of joining battle, his men saw that so large a part of their force had fallen by divine judgement, they no longer dared advance on the Saracens but all retired whence they came. The Saracens proceeded – as was their habit – to lay waste the provinces of the empire that had fallen to them.10

Gad went to David, told him what the Lord had said, and asked, ‘Which is it to be? Three years of famine? Or three months of running away from the armies of your enemies? Or three days during which the Lord attacks you with his sword and sends an epidemic upon your land, using his angel to bring death throughout Israel? What answer shall I give the Lord?13

duo milia ex militibus Aeragliae in stratum mortui sunt. Cumque in crasteno ad prilium debebabt adgredere, cernentes eorum exercitum milites partem maxema devino iudicio interfectam, adversus Saracinus nec ausi sunt inire prilium. Regressus omnes exercitus Aeragliae ad propries sedebus, Saracini more quo ceperant provincias Aeragliae emperatores adsiduae vastandum pergebant. Ed. and trans. WallaceHadrill 1960: 55. 11 Kaegi 1992: 125, n. 34. The meaning of this nickname was known to Theophanes (Chron. AM 6123, 335) from his translation of the ‘common Syriac source’, and this is an important detail revealing that the author of the ‘common Syriac source’ utilized Muslim historical tradition also. See Conrad 1990: 22-24. 12 It is probably only a coincidence that Heraclius ordered the cubicularius and sacellarius Philagrius to perform a new census sometime during the later years of his life. See Kaegi 1992: 256-257, who dates the census to between late 636 and early 641 on the basis that Philagrius succeeded Theodore Trithyrios as imperial sacellarius when the latter was killed at the battle of Yarmnjk, as he misleadingly calls it, which he dates to 636. This places rather too much trust in the exact wording of Nicephorus (Brev. 20), our only source to describe Theodore Trithyrios as imperial sacellarius. In fact, it is possible that Theodore Trithyrios was only the sacellarius to the magister militum Theodore, the brother of Heraclius, before his promotion to replace him in charge in the east. Hence Philagrius may have performed his census much earlier than assumed, in the early 630s even. 13 1 Chronicles 20.11. Trans. Good News Bible (Glasgow, 1979), 422.

9

See Tsafrir, di Segni, and Green 1994: 127, on Jebata and Jaba‘. Fred. Chron. 4.66: Congregatis undique de universas provincias emperiae nimia multetudinem militum, transmittens Aeraglius legationem ad portas Caspias, quas Alexander Magnos Macedus super mare Cespium aereas fiere et serrare iusserat propter inundacione gentium sevissemorum que ultra montem Caucasi culmenis habetabant, easdem portas Aeraglius aperire precepit. Indique cento quinquagenta milia pugnatorum auroque locatus auxiliae suae contra Saracinus priliandum aemittetur. Saracini duos habentes princepis, ducenta fere milia errant. Cumque castra nec procul inter se exercitus uterque posuissit, ita ut in crastena bellum inirent confligentes, eadem nocte gladio Dei Aeragliae exercitus percotitur: in castris quinquaginta et 10

368

DAVID WOODS: JEWS, RATS, AND THE BATTLE OF YARMNjK The epidemic that followed struck down 70,000 people before God changed His mind and told His angel to stop. David even saw the angel standing in mid-air above Jerusalem, holding His sword ready to destroy the city. Hence this story suggests that Fredegar probably used the phrase ‘sword of God’ in reference to a severe epidemic of some sort. The fact that Fredegar had compared Heraclius to King David earlier in his account of a singlecombat which Heraclius had allegedly fought against a Persian champion reinforces this suggestion.14 Fredegar, or his source, was still drawing parallels in his mind between Heraclius and David when he chose to describe a severe epidemic in this way. But what was the nature of this epidemic? Not only does the Spanish Chronicle of 754 confirm that an epidemic did indeed strike the Byzantine army at Gabatha/Gabitha, but it even provides a solution to the identity of this epidemic.

and he brought back the worshippers of idols to the knowledge of the One God, and bade them declare that Muhammad was his apostle; and his nation were circumcised in the flesh, not by the law, and prayed towards the South, turning towards a place which they called the Kaabah.16 Similarly, writing c.915, the Muslim historian Al-TabarƯ preserves an early Muslim tradition concerning this same dream: When he [Heraclius] reached Jerusalem and performed his worship - with him were his military commanders and the nobles of the Romans – he arose troubled one morning, turning his gaze to the sky. His military commanders said to him, ‘By God, you have arisen troubled this morning, O King.’ ‘Yes,’ he replied, ‘I was shown in a dream last night that the kingdom of the circumcision will be victorious.’ They said to him, ‘O King, we know of no nation that practices circumcision but the Jews, and they are under your control and authority. Send to all over whom you have authority in your lands and command them to behead all the Jews under their control, and be rid of this care’.17

The Dream of Heraclius The second step in the solution to these problems lies in the explanation of the origin of a neglected variant of the story that the emperor Heraclius (610-641) had experienced a dream concerning the rise of the Arab empire, a variant preserved only by the author of the Chronicle of 754. Fredegar preserves our earliest description of this dream:

It is clear, therefore, that these authors all draw on the same ultimate eastern source in this matter at least.

Being well-read he practised astrology, by which art he discovered that, by the will of God, his empire would be laid waste by circumcised races. So he sent to the Frankish King Dagobert to request him to have all the Jews of his kingdom baptized – which Dagobert promptly carried out. Heraclius ordered that the same should be done throughout all the imperial provinces; for he had no idea whence this scourge would come upon his empire.15

The author of the Chron. 754 begins his description of the Muslim invasion of the Byzantine empire with an account of how God had warned Heraclius in his dreams about the coming dangers, exactly as does Fredegar. He sets these warnings in the period immediately following Heraclius’ celebration of his victory over the Persians, and claims that these dreams foretold what would happen if Heraclius did not give proper credit to God for His role in the defeat of the Persians:

The History of the Patriarchs of Alexandria, an Arabic translation and continuation of a collection of the lives of the patriarchs of the Coptic church at Alexandria, preserves a remarkably similar account in a section of its text which was probably written in the early eighth century:

They say that afterward many things pertaining to this event began to come to Heraclius in his dreams as a warning that he would be ravaged mercilessly by rats from the desert. He was also forewarned by astrological readings of the course of the stars. Heraclius, as we have said, subdued the rebellious Persians and restored the imperial territories through his fighting. But seduced, it is said, by the praise of the people who heaped the honour of victory not on God but on Heraclius himself, he feared the rebuke which was being gravely presaged in his recurring visions.18

And in those days Heraclius saw a dream in which it was said to him: ‘Verily there shall come against thee a circumcised nation, and they shall vanquish thee and take possession of the land’. So Heraclius thought that they would be the Jews, and accordingly gave orders that all the Jews and Samaritans should be baptized in all the provinces which were under his dominion. But after a few days there appeared a man of the Arabs, from the southern districts, that is to say, from Mecca, or its neighbourhood, whose name was Muhammad;

16

Trans. Evetts 1907: 492. On the date, see Hoyland 1997: 446-448. Al-TabarƯ, Ta’rƯkh I.1562. Trans. Fishbein 1997: 101. Chron. 754, 6-7: Tunc in sompnis de re huiuscemodi multa ei venire ferunt ex monito et ut a muribus heremi inmisericorditer vastaretur et pro stellarum cursu astrologico premonetur indicio. Hic Eraclius, ut prefati sumus, Persas rebellantes edomuit, imperiales patrias belligerando reformavit. Seductus a laudibus populi non Deo, sed sibi, ut ferunt, honorem victorie exaggerando, increpationem per visum non modicam graviter presagando crebrem expavit. Ed. Mommsen 1894: 336-337. Trans. Wolf 1990: 113. 17 18

14

Fred. Chron. IV.64: ut novos Davit procedit ad bellum. On the popularity of this theme, even in the silver-ware of the period, see Wander: 1973. 15 Fred. Chron. IV.65. Trans. Wallace-Hadrill 1960: 54-55, slightly amended.

369

THE LATE ROMAN ARMY IN THE EAST This account is a more accurate version of events than that preserved by Fredegar in that it clearly distinguishes between the interpretation of dreams and astrology. However, it is unique among all the accounts of Heraclius’ alleged dream in claiming that he received a warning about rats rather than about a circumcised people whom he mistakenly identified as Jews. The author soon repeats his allegation that Heraclius had been warned about rats in his dream when he proceeds to describe the battle of Gabatha itself:

biblical precedent exists. God had apparently warned the king of Egypt about the approach of seven years of plenty followed by seven years of famine by means of two different dreams, the first featuring seven fat cows who were eaten by several thin cows, the second featuring seven full ears of corn which were swallowed up by seven thin ears.21 Alternatively, some might be found willing to accept that Heraclius was warned about both rats and circumcised people in the same dream even. Nevertheless, the simplistic conflation of what seem at first glance to be variant accounts of the same event is never wise, certainly not before one has investigated whether one can explain this phenomenon on the basis of some small misunderstanding during the transmission of the original text. In this case, one solution does offer itself.

After fighting many battles against the Saracens, Theodore, the brother of the emperor Heraclius, withdrew from combat to increase and consolidate his forces for war, at the admonition of his brother, mindful of the prophecy of the rats. But with their throats swelling more each day, such fear spread amongst the Roman legions that when they engaged in battle at the town of Gabata, their army was totally destroyed and Theodore was slain and departed this world. Then the Saracens, firmly confident and free from fear after such a slaughter of noble men, established their rule in Damascus, the most splendid city in Syria.19

One must begin one’s investigation of this problem by realizing that it involves word-play. Heraclius did not actually see any rats in his dream, or a group of circumcised people destroying everything before them. As the History of the Patriarchs of Alexandria states, a voice spoke to him in a dream. Hence everything depends on the exact nature of the words allegedly spoken to him. If one wishes to solve the problem, therefore, one must return to the language in which the earliest form of the anecdote had presumably been composed, Greek. One must now investigate whether there is a term or phrase which could have been used to describe either a circumcised people or rats/mice. Greek uses various forms of the verb peqitÈlmy ‘I cut round about’ to describe the act of circumcision, so it is obvious that some form of this verb probably lies at the heart of this problem. One notes, therefore, that a phrase such as peqitelmËlemoi ‘those being circumcised’, using the present participle of the passive form of this verb, might well be interpreted in reference to Jews, depending on the context.22 But the middle form of this verb may also be used to describe various types of ‘cutting off’ for oneself, such as the conducting of cattle raids.23 In the case of the present participle, therefore, where the middle and passive forms of the verb are identical, ambiguity would be unavoidable, and the context would have to determine the precise meaning of the verb. Furthermore, the verb peqitÈlmy is itself a compound formed from the verb tÈlmy ‘I cut’ and the preposition peqÊ ‘round about’, and the same verb tÈlmy can be used to describe the activities of rats/mice. For example, Aelian uses the compound form ÜpeqtÈlmy to describe the activities of rats/mice who cut off the ears of corn from below in the fields of Egypt.24 It is arguable, therefore, that if the author of the original anecdote had simply described those against whom Heraclius was being warned as peqitelmËlemoi, the nature of these beings would not have been at all clear. The ambiguity would have been

It is important here to point out that the author actually uses the Latin mus which could be used of either rats or mice, and many other rodents also. Neither Greek nor Latin had a specific term for either rat or mouse, and one can only identify to which creature any source actually refers from the context, although the distinction is not always possible even then.20 So why does the author of the Chron. 754 believe that Heraclius experienced dreams concerning rats, while all others who record this event claim that he dreamed about a circumcised race whom he then mistakenly identified as Jews? How could one confuse a description of a swarm of rats/mice with a description of a circumcised people, or vice-versa? Furthermore, how do we decide whether the original author of this anecdote in its earliest form had referred to rats/mice or to a circumcised people? One is tempted to solve this problem by claiming that Heraclius had probably experienced two different dreams, both warning him about same event presumably, but doing so by means of different metaphors in each case. One cannot entirely exclude this possibility, of course, that the original author of this anecdote had claimed that Heraclius had been warned in a dream about rats first, and then about a circumcised people, or vice versa. A 19 Chron. 754, 11: Sicque, multis preliis dimicante contra eos Theodoro Eraclii Augusti germano, monitu fratris presagationem murium reminiscentis ad multiplicandas et colligendas in bello gentes discessit a prelio. Sed cotidie eorum incrassante iugulo ita in Romanis legionibus irruit timor, ut apud Gabatam opidum commisso belligero exercitus funditus lesus et Theodorus neccatus migraret a seculo. Tunc Sarraceni de tanta nobilium strage firmiter certi metu excusso apud Damascum splendidissimam Sirie urbem conscendunt in regno. Ed. Mommsen 1894: 337-338. Trans. Wolf 1990: 114. 20 Liddell, Scott, and Jones 1996: 1155, s.v. ‘R³r’; Glare 1996: 1148. s.v. ‘mus’.

21

Genesis 41.1-33. In general, see Liddell, Scott, and Jones 1996: 1390, s.v. ‘peqitÈlmy’. 23 Hom. Od. XI.402, XXIV.112. 24 Ael. NA VI.41. 22

370

DAVID WOODS: JEWS, RATS, AND THE BATTLE OF YARMNjK Antium among the events of ill-omen in 203BC.28 Writing c.AD78, Pliny recorded the fact that they had gnawed silver shields at Lanuvium as a prediction of the outbreak of the Marsian or Social war (91-88BC), and the fact that they had gnawed the puttees of the general Carbo at Clusium as a sign of his death in 82BC.29 Furthermore, the second century sophist Aelian counted mice among the most prophetic of animals, because their squeaks predicted rain and they always knew to leave an old house just before it collapsed.30 It is possible, therefore, that the original author of the story of Heraclius’ dream was influenced by this classical tradition in his decision to choose the occurrence of a swarm of rats/mice as an event forewarning of disaster. However, the most important reason to believe that the author had intended an ambiguous reference to rats/mice lies in the striking coincidence between an apparent warning against rats first and the description next of an outbreak of bubonic plague during the event which could have been prevented had the warning been properly understood.

maintained by the use of the noun v³kom ‘race’ in reference to these creatures, since it can be used either of man or of any other type of living creature also.25 Hence in the original form of the anecdote, the voice that spoke to Heraclius in his dream had probably warned him against tË v³kom peqitelmËlemym. On the one hand, the author may have intended this to be understood as ‘the race of those who are circumcised’, but he may also have intended it to be understood as ‘the race of those who cut off for themselves’, that is, ‘the race of raiders’. The key point as far as we are here concerned is that it would have been impossible to retain the same ambiguity inherent in such a phrase when trying to translate it into another language. One would inevitably have been forced to favour one possible interpretation over the other. The situation is complicated at this point by the fact that each of the two possible interpretations of the phrase above is itself ambiguous, that ‘the race of those who are circumcised’ could have been used to describe either Jews or Arabs, since both were circumcised, while ‘the race of raiders’ could have been used to describe either rats/mice or hostile Arab tribesmen. My suggestion is that the author had originally intended the phrase to refer to rats, and that his idea was that God had sent the dream to Heraclius not to warn him against rats themselves, but of the disaster that would follow closely upon, and be symbolized by, the occurrence of a swarm of rats. The intention was that Heraclius would be shown to misinterpret the dream by interpreting it as a warning against the Jews rather than about a sign by which he would be able to recognize that disaster was at hand. The problem, it seems to me, is that the author either forgot, or did not know, that Muslim Arabs were circumcised also.26 Hence his language contained an ambiguity that he had never intended. He had intended the ambiguity to be between rats/mice and Jews, but to those who came after him and were far more familiar with the practices of Muslim Arabs, the more obvious ambiguity, if one can speak of such a thing, seemed to be between Jews and Muslim Arabs.

I draw attention at this point to the claim by the author of the Chron. 754 that the throats of the Byzantines were swelling by the day (cotidie eorum incrassante iugulo) at the time of the battle of Gabatha. What does this actually mean?31 One of the main symptoms of bubonic plague was the swelling of the lymphatic nodes in the groin, arm-pit, and neck. It was these enlarged glands, called buboes, which gave the plague its modern name. Writing c.593, the ecclesiastical historian Evagrius preserves a detailed description of its varied symptoms: The misfortune was composed of different ailments. For in some it began with the head, making eyes bloodshot and face swollen, went down to the throat, and dispatched the victim from among men. In others a flux of the stomach occurred. While in some buboes swelled up, and thereafter violent fevers; and on the second or third day they died, with intellect and bodily constitution the same as those who had suffered nothing. Others became demented and put aside life. And indeed carbuncles sprang up and obliterated men.32

There are two reasons to believe that the author of this anecdote had originally intended an ambiguous reference to rats/mice. The first is that destructive behaviour by rats/mice had long been accepted as an important type of ominous event, particularly of death and disaster.27 Writing c.9BC, the Roman historian Livy included the fact that mice had gnawed the gold in the temple of Jupiter at Cumae among the events of ill-omen in 208BC, and the fact that mice had gnawed a gold circlet at

Note the inclusion of the throat among the regions affected. Hence the description of the physical condition of the troops by Chron. 754 suggests that they were suffering from bubonic plague. The next important point concerns the devastating impact of this disease, whatever it was, upon the troops. Even if one cannot accept

25

See Liddell, Scott, and Jones 1996: 1962, s.v. ‘v³kom’. Circumcision had been commonly practiced by pre-Islamic Arabs, especially during the first centuries AD. See e.g. Josephus, Ant. Jud. I.214; Eusebius, Praep. Evang. VI.1-11; Jerome, In Hier. 9.25; Sozomen, HE VI.38.11. One suspects, however, that the Christianization of many Arab tribes during the fifth and sixth centuries AD would have severely curtailed this practice. 27 The second century dream-interpreter Artemidorus identifies mice appearing in dreams as domestic slaves, and claims that dreams of lots of mice, happy and playing, are good because they foretell a happy slave-household which will increase in number (Oneirocriticon III.28). 26

28

Livy XXVII.23, XXX.2. Pliny, HN VIII.221. 30 Ael. NA VII.8, XI.19; VH I.11. 31 No commentator seems to have investigated the significance of this claim. Wolf 1990: 114 does not discuss it in the notes to his translation. Stathakopoulos 2004: 334-350 does not refer to this passage in any of the sources which he cites for outbreaks of plague during the early seventh century. 32 Evagrius, HE IV.29. Trans. Whitby 2000: 231. 29

371

THE LATE ROMAN ARMY IN THE EAST him, and so had clarified that the object of the dream was a swarm of rats rather than a ‘circumcised race’, the authors of the other witnesses to this story seem each to have relied on a translation or edition (probably the same translation or edition) where the translator or editor more correctly retained the ambiguity of his original source, but failed, nevertheless, to understand the real nature or purpose of this ambiguity.

Fredegar’s claim that 52,000 Byzantine troops died in their sleep the night before they were due to give battle, he clearly preserves the memory of a disease that had had a devastatingly high mortality rate. Again, this matches the description of bubonic plague which kills 50-60% of those whom it affects. Finally, one notes the severe impact that the outbreak of this disease had on the morale of the troops as described by Chron. 754. Few, if any diseases, would have struck so much fear into their hearts as bubonic plague. It had been recurring in waves across the Mediterranean basin ever since it first revealed itself in the Justinianic plague of 541. No-one in the Byzantine camp would have been unfamiliar with its symptoms, the ease with which it spread, nor its high mortality rate. Noone would have wanted to stay in camp once they had the slightest reason to suspect that plague had broken out.

It remains to prove that Syria did sometimes suffer sudden explosions in the rat population, that is, swarms of rats. An anonymous Syriac account of the life of Simeon the Stylite (d. 459) preserves a vivid account of a swarm of mice and rats which attacked the people of Aphshon sometime during the first half of the fifth century. This village was apparently situated somewhere near Simeon’s column at Telanissus just east of Antioch in Syria, and its people flocked to him for help:

The coincidence between a dream apparently warning against rats and the fulfillment of this dream by an outbreak of bubonic plague is striking. No-one knew that it was rats which spread the bubonic plague by means of the infected fleas which they carried until the end of the nineteenth century. Hence the original author of the story of Heraclius’ dream could not have deliberately chosen to identify the subjects of the dream as rats and its fulfillment as an outbreak of plague in order that these elements of his narrative would reinforce one another. Unless one possesses a strong faith in coincidence, one is forced to conclude that there is a strong historical element to both elements. This is not to claim that Heraclius did actually dream of rats. Dreams do not come true, and historical accounts of prophetic dreams are no more than literary fiction, clever devices designed to appeal to the imagination of the reader. My point is that it was his knowledge of the occurrence of a real swarm of rats sometime shortly before the battle of Gabatha that led the author of the story of Heraclius’ dream to decide upon that as the subject of the alleged dream. He probably included some notice concerning this swarm of rats in his original history, but as various successors mistranslated and abbreviated his text this apparently unimportant notice was dropped. The important point to note here, however, is that he seems to have been a little too clever for his own good. He used deliberately vague language when describing the contents of Heraclius’ dream in order to hook the reader and keep him guessing concerning its interpretation. Then, when the reader reached his description of the swarm of rats which preceded the fateful battle of Gabatha, all would become clear. Or should have. The problem was that this clever literary device assumed not only that the reader would have the full text before him, but that he would also be intelligent enough to make the connection between the ambiguous dream and the strange swarm of rats. Furthermore, it underestimated the sheer strength of the anti-semitic tradition upon which the author had been playing as the source of the ambiguity. While the author of the Chron. 754 seems to have had access to some translation or edition of this text where the translator or editor had correctly understood the literary device before

They informed his holiness about huge mice and rats who were biting live sheep and eating their insides so that they died. They even began on oxen and beasts so that the animals there, either sheep or asses, would flee, and the rats, fifty or one hundred of them, would run after [one of] them until the animal was exhausted and fell. Then they would eat it. They used to grunt like swine and not run away when men appeared. They even ventured up to small children and would run after them like dogs.33 Fortunately, Simeon gave them some dust from his shrine, and they were able to use this to drive away these vermin. Agapius and Michael the Syrian record a more relevant example, perhaps, when they report that there was a swarm of rats in Syria and Phoenicia in AG989 (AD677/678), that these destroyed the crops, and that a great famine followed.34 One hardly doubts that there must have been many similar such occurrences during the intervening years also, at least at the local level. The Plague of Emmaus The claim that the Byzantine army at Gabatha was destroyed by plague, or at least so weakened both by the outbreak of the disease itself and the desertions that inevitably followed, that it ought never to have attempted to join battle with the Muslims, immediately raises the question as to why we have not learned more about this outbreak of the disease. No source dates any outbreak of plague to the same year as the battle of Gabatha, but the common ancestor of several surviving Christian sources, now known as the ‘common Syriac source’, seem to have recorded an outbreak shortly beforehand c.634, while Muslim sources record an outbreak shortly afterwards c.639.35 If we examine the evidence for this alleged first 33

Syriac Life 61. Trans. Doran 1992: 140-141. Agapius, 492; Michael 11.13, 436/457. 35 For a list of famines and epidemics during the period 284-750, see Stathakopoulos 2004: 177-386, esp. 348-349, who claims that the epidemic of c.634 can be distinguished from that of 638. In contrast, 34

372

DAVID WOODS: JEWS, RATS, AND THE BATTLE OF YARMNjK epidemic c.634, it soon becomes clear that it formed but one part of a larger catalogue of ominous events, and that the author of this catalogue of ominous events had located it at the start of his description of the Arab invasion of Palestine and Syria. This was in accordance with the classical tradition in such matters where catalogues of natural disasters or unusual phenomena were used to signal the approach of political or military disasters also.36 In brief, he arranged his material in order to serve a literary purpose, not to describe an exact chronological sequence of events. The problem is that those who followed the original author of this passage, the author of the ‘common Syriac source’ and his various successors, do not seem to have fully appreciated this fact.

strange, since one would assume that the plague ought to have struck all the various religious and political groups with equal force, Christian as well as Muslim, and that a proportion of senior Christian figures, such as bishops and abbots, ought to have died then also. The only possible explanation is that the plague mentioned by the author of the Greek ancestor of the ‘common Syriac source’ at the start of his account of the Arab invasion of Palestine and Syria was identifiable with the plague whose occurrence he ought to have recorded c.639 had he composed a strictly chronological sequence of events. It is very difficult to make any definitive statements about the plague of Emmaus, since the surviving literary sources do not allow a firm conclusion as to where it began or how it spread.38 It seems to have been confined to Syria and Mesopotamia, but it is not clear why this was so. If we accept the literary tradition at face value, the plague struck the Muslim army while encamped at Emmaus, although it is not at all clear why so large a force as is claimed should have been stationed there at this time, and the caliph ‘Umar ordered Abnj ‘Ubayda to lead his army away from the lowlands of Emmaus to the healthy highlands of al-JƗbiya, that is Gabitha. The route which the army took can be reconstructed by following the tombs of those alleged to have died during the epidemic.39 Indeed, on the face of things, topography and archaeology ought to allow us to reconstruct not only the route itself, but the succession in which the various Muslim officers died as they progressed along this route. For example, the tomb of Abnj ‘Ubayda is situated much further north towards Gabitha than is the tomb of Mu‘Ɨdh b. Jabal, which supports the literary tradition that the latter succeeded the former in command of the army. Unfortunately, while such consistency between the literary and topographical information is initially reassuring, there is a real danger that the literary tradition was itself constructed on the basis of this topography. On the face of things, however, there seems no good reason to doubt the tradition as preserved, that the plague began at or near Emmaus c.639.

If one cannot associate the outbreak of plague that struck the Byzantine army at Gabatha with the alleged epidemic in 634, because this probably did not occur then, one is forced to consider its identification with the notorious plague epidemic of 639 instead, the so-called ‘plague of Emmaus’. One curious feature of this plague is that it has left very little evidence in the surviving Christian sources, whatever their language, but is well attested by the Muslim sources. A collection of extracts from the chronicle which James of Edessa composed in 692 preserves a brief statement that there was a severe plague throughout the districts of Syria in AH18 (12 January 639-1 January 640), Elias of Nisibis (d. 1049) copied the same notice from James of Edessa into his own chronicle, and Michael the Syrian and the anonymous author of the Chronicle of 1234 each preserve short notices about it also. The latter preserves the more detailed notice. He describes how the caliph ‘Umar had personally accepted the surrender of Jerusalem and then, under the heading AH18, adds: After ‘Umar had gone back down to his city the pestilence was unleashed throughout the land of Palestine. Abnj ‘Ubayda, the Arab emir and general, contracted the disease and died at Emmaus, a city of Palestine. His place was taken by Mu‘Ɨdh b. Jabal.37

The Location and Date of the Battle of Gabatha/Gabitha

The contents of this notice, particularly the emphasis on the Muslim personalities, reveal that it derives from a Muslim source. It seems, therefore, that neither the ‘common Syriac source’ nor its ultimate Greek ancestor had described an outbreak of plague in Palestine and Syria that they dated to 639 or thereabouts. This is

I have now demonstrated that the Latin chronicles point to the occurrence of a major epidemic of plague among the Byzantine forces at the battle of Gabatha/Gabitha as the reason for their defeat there. I have also explained that there was only one major outbreak at plague in Palestine and Syria during this period, the so-called plague of Emmaus c.639. The only reasonable conclusion is that it was the plague of Emmaus that destroyed the Byzantine army on the eve of the battle of Gabatha/Gabitha. It seems clear that the plague began within the Byzantine forces, but then spread among the victorious Arab forces

Conrad 1981: 190, n. 48, favours their identification. On the ‘common Syriac source’, see Hoyland 1997: 400-409, and 636, where he includes the outbreak of plague in his attempted reconstruction of the outline of this lost source. 36 See e.g. Dio LVI.24.3-5 (omens before the Roman defeat at the battle of Teutoburg forest in AD9); LXII.1 (omens before rebellion by the British under Boudicca c.AD61). The development of this literary tradition derives from the collection of such data for religious purposes during the early history of Rome. See MacBain 1982: 82-106 for a list of prodigies that often included the items under discussion here, pestilence, earthquakes, and signs in the sky. 37 Chron. 1234, I.255-256. Trans. Palmer 1993: 162.

38 See Conrad 1981: 167-246 for a detailed discussion of the Arabic sources. 39 Conrad 1981: 204-213.

373

THE LATE ROMAN ARMY IN THE EAST as they plundered the camps of the Byzantines and installed their own garrisons wherever the Byzantine forces had been located previously. This gives us a new means by which to approach the problem of the true identity of Gabatha/Gabitha. It ought to be identifiable as the town of that title nearest to where the plague of Emmaus had first struck the Arab forces. If, as Arab tradition seems to suggest, the plague first struck the Arab forces at Emmaus, then it ought to be identifiable as the Gabatha southwest of Jerusalem.

overlooked, that he did meet his commanders much as Arabic tradition relates, but that he met them at the Gabatha southwest of Jerusalem, the town occupied by the victorious commander who had summoned him to Palestine in the first place, rather than at Gabitha. The names of the two towns have simply been confused during the transmission of this tradition. The second question concerns the origin of the tradition that the decisive battle had been fought at the river Yarmnjk rather than at Gabatha. What gave rise to this tradition? Here one notes that Gabatha is situated about 11km (as the crow flies) southeast of the town of Iermycha.45 The parallelism between the names of Gabatha and Iermycha southwest of Jerusalem and the names of Gabitha (JƗbiya) and the river Hieromyces (Yarmnjk) in the Golan, to call them by their classical names, has long been noted, as has the potential of this parallelism to cause confusion among our sources.46 Unfortunately, in the absence of any detailed historiographical study of the relationship between the various sources, the significance of this parallelism has not been fully appreciated. It is my suggestion that the misinterpretation of the name of Gabatha in reference to the more famous Gabitha led in due course to the misinterpretation of the name of the town of Iermycha in reference to the river Hieromyces. One mistake followed naturally upon the other as the decisive battle was relocated into the wrong area altogether. This is not to claim that there was ever necessarily a single tradition that described the location of this battle by means of both toponyms, Gabatha and Iermycha. Instead, the misinterpretation of one tradition originally identifying the site at or near Gabatha may have led in turn to the mistaken amendment of a separate tradition originally identifying the site at or near Iermycha. Indeed, it is arguable that the tradition identifying the site of the battle at Gabatha represents the oldest Christian tradition, while that identifying the site at Iermycha represents the oldest Muslim tradition. This would explain the fact that Muslim sources reveal no knowledge of the battle of Gabatha, while the oldest Christian sources reveal no knowledge of the battle of Yarmnjk as such, only a battle at Gabatha.47 The fact that the later Christian sources –

Two questions now arise. The first concerns the confusion, particularly in the Syriac sources, between Gabatha southwest of Jerusalem and the Gabitha southwest of Damascus. How or why did this occur? Some confusion was always a possibility because of the similarity between the two names, particularly if the original form of this anecdote had failed to provide any other clear indications as to the location of Gabatha, but the real answer to this question lies in the realization that Gabitha in the Golan had always enjoyed a far higher profile.40 The GhassƗnid Arabs had used it as their capital during their period as the chief Arab agents and allies of Byzantines from the early sixth century onwards.41 Its continued importance under the early Umayyads is highlighted by the fact that MarwƗn was hailed as caliph there in 683.42 It would have been natural, therefore, to misread or correct the name of Gabatha to Gabitha, especially if the site in question was being identified as the location of some important political development. One can adduce at least one parallel example for such confusion between the names, although in Arabic rather than Syriac. According to Muslim tradition, the caliph ‘Umar met his commanders at Gabitha (JƗbiya) before he accepted the surrender of Jerusalem.43 Since he had apparently been asked to come to Palestine in order that he might personally accept the surrender of Jerusalem, it is strange that the Arabic sources traditionally locate his meeting with his commanders at Gabitha southwest of Damascus. Why would he have marched so far northwards in order to have to return southwards almost immediately again in order to complete the main purpose to his journey? The obvious answer is that he did not, and this has encouraged some commentators to deny the tradition that ‘Umar visited his commanders in Palestine at this point and personally received the surrender of Jerusalem.44 The alternative possibility has been

45

See Tsafrir, di Segni, and Green 1994: 151, on Iermycha. Eusebius refers to Iermycha three times within his Onomasticon, spelling it differently on each occasion (җ’IeqiloÌh, ’IeqlowÍr, ’IeqilÍh). Jerome adds to the confusion with his translations (Ierimuth, Iermucha, Iermoth). See Steven Notley and Safrai 2005: 102-105. The site of Iermycha, modern Tel Yarmuth, has been the subject of extensive excavation, but the Roman village seems to have been abandoned in the fourth century AD. For a summary of these excavations, see de Miroschedji 1999. The only classical author to record the name of the Yarmnjk is Pliny the Elder when he describes the geography of the Decapolis (NH V.17.74): Hieromice praefluente. 46 See e.g. de Goeje 1900: 60-61 who argues that the confusion of these names led some Arab sources to confuse the battle of AjnƗdayn with the battle of Yarmnjk; also Jandora 1985: 8-21, at 11. The more sceptical may doubt whether these were ever two separate battles to start with. 47 For the earliest Muslim accounts of the conquest of Syria, see Donner 1981: 91-155. He highlights the complex and contradictory nature of the evidence, the fact that Ibn IshƗq (d. 767) and al-WƗqidƯ (d. 823)

40

It is important to realize that this is by no means the only example where the author of the ‘common Syriac source’ seems to have confused two similar toponyms in favour of the more famous. E.g. it is commonly believed that the Arabs penetrated deep behind the Byzantine lines and devastated Amorium in Phrygia c.645 (Michael XI.10, 428-429/441). See e.g. Treadgold 1997: 310. Due consideration of the geographical and military situation suggests that they probably devastated the less famous Anemourium in Isauria instead. I note in passing that one cannot use the Arabic sources on this subject in the manner of Kaegi 1977 unless one can first demonstrate their independence of the ‘common Syriac source’. 41 See ShahƯd 1995, passim. 42 Theoph. Chron. AM 6175, 360; Chron. 1234, I.291. 43 Al-TabarƯ, Ta’rƯkh I.2401. 44 Noth and Conrad 1994: 181.

374

DAVID WOODS: JEWS, RATS, AND THE BATTLE OF YARMNjK Theophanes, Michael the Syrian, the Chron. 1234 – describe the battle of Yarmnjk at length, to the complete omission almost of the battle of Gabatha, suggests that the author of the ‘common Syriac source’ had supplemented his main Christian source at this point with material from elsewhere. Since he certainly used Muslim sources to supplement his account of other events about this period, it is not unreasonable to suspect that he may have derived his account of the so-called battle of Yarmnjk from the same or similar sources.48 In summary, he combined two different accounts of the same battle told from very different perspectives in the mistaken belief that they described two separate battles, and at a point when the Muslim tradition had already suffered severe corruption under the influence of the same Christian tradition with which he was then trying to combine it.

knowledge of this fact, that a Byzantine relief force could not suddenly arrive into view one day, would have affected morale accordingly. All knew that the physical situation of Jerusalem meant that there would be no last minute reprieve. Second, the example provided by the siege and destruction of Jerusalem in 614 was still too fresh in the minds of everyone. Given that the relevant authorities had initially surrendered the city to the Persians then with no resistance, and that subsequent events, the Persian destruction of the city and massacre of its inhabitants after the circus factions managed to take control there, proved that they had been entirely correct in so doing, it beggars belief that there could have been strong support for a determined resistance once more against the Muslims, or that such could have been sustained for two years.51 Finally, even if the real siege of Jerusalem did only begin after the Arabs had inflicted a decisive defeat on the Byzantine field-army, the city had actually faced increased difficulties long before that. Bishop Sophronius of Jerusalem has left two texts which prove that Arab forces already were threatening Jerusalem by 634.52 Regardless of the details of one’s reconstruction, therefore, there can be no doubt that southern and central Palestine had become increasingly insecure for several years before the battle of Gabatha or Yarmnjk, even if one dates it as early as 636. Such insecurity would have affected Jerusalem in two ways. Not only would it have made it difficult for the authorities to supply the city in the normal manner, but it would also have driven a large number of refugees into Jerusalem whose presence would have strained the ability of the city to provide for them even at the best of times. Jerusalem, one suspects, had been placed under severe strain for several years already by the time of Gabatha or Yarmnjk, and would have had no reserves to sustain it for a more sustained siege over another two years.

If my interpretation is correct, and the plague of Emmaus seriously weakened the Byzantine forces on the eve of the battle of Gabatha, then this battle ought to have been fought in 638, about two years later than the date to which modern commentators have traditionally assigned the battle of Yarmnjk, as they call it. The advantage of this redating of this battle, however, is that it solves a longstanding chronological problem that has never quite received the attention that it deserves. I refer here to the strange chronological gap between the battle of Gabatha or Yarmnjk that apparently forced a final Byzantine evacuation of Palestine and Syria, traditionally dated to 636, and the consequent surrender of Jerusalem to the Muslims, traditionally dated to 638.49 Could Jerusalem really have resisted the Muslims for another two years following the news of their decisive victory over the main Byzantine army?50 This possibility strikes one as unlikely for three main reasons. First, Jerusalem was an inland city. It could be much more easily and effectively cut off from outside help than any coastal city, such as Caesarea, for example, which seems to have been the last major city to fall to the Muslims, in 641 apparently. The mere

Conclusion The sequence of events may now be tentatively reconstructed as follows. The Byzantine army was defeated at or near Gabatha, southwest of Jerusalem, in the autumn of 638, after it had been severely weakened by plague. News of this decisive defeat spread quickly. Jerusalem, which had been suffering increased hardship for several years already, began negotiations for surrender almost immediately upon receipt of this news. In the meantime, the caliph ‘Umar had decided to visit his commanders still based at Gabatha, partly in order to supervise the division of the spoils and partly to reassert his authority over them. His presence at Gabatha meant that he was personally able to accept the surrender of Jerusalem in late 638. Some of the Muslim forces may have remained in or about Gabatha during the winter of 638/639, but the bulk of them had probably been relocated in or about Emmaus shortly after the battle at Gabatha in an effort to isolate Jerusalem from the coast,

transmitted a reconstruction of the conquest that dated the battle of AjnƗdayn to July 634, the battle of Yarmnjk to August 636, and the surrender of Jerusalem to 638, while Sayf b. ‘Umar (d.786/809) transmitted a tradition from Abnj ‘UthmƗn YazƯd b. AsƯd al-GhassƗnƯ which reversed the order of battles and dated that at Yarmnjk to 634 and that at AjnƗdayn to 636. Such confusion encourages the belief that little was known for certain about these battles, and that the Muslim accounts of two different battles depended on different oral traditions concerning the same battle really. 48 See Conrad 1990. 49 The evidence for the date of the surrender of Jerusalem is complex and contradictory. The fact that Theophanes specifically states that the Arabs besieged Jerusalem for two years before it surrendered (Chron. AM 6127, 339), combined with the facts that many Arabic sources date the battle of Yarmnjk to 636 and the alleged surrender of the city to the caliph ‘Umar himself to 638, has encouraged the widespread acceptance of these dates. See e.g. Stratos 1972: 81-82, who concludes that it surrendered in February 638. 50 Those who assume that it could have, must then ask themselves why Heraclius did not use the intervening period to send another force in defence of the city. Sahas 1999 even argues that Heraclius allowed his personal hostility towards Sophronius to influence him to abandon Jerusalem.

51 52

375

Sebeos 34 (115-116). Hoyland 1997: 69-71.

THE LATE ROMAN ARMY IN THE EAST Liddell H.G. - Scott R. and Jones H.S. 1996 A GreekEnglish Lexicon, Oxford. MacBain B. 1982 Prodigy and Expiation: A Study in Religion and Politics in Republican Rome, (Collection Latomus 177), Brussels. Mommsen T. 1894 ‘Continuatio Hispana ad Annum DCCLIV’. In MGH Auctores Antiquissimi 11.2: Chronica Minora Saeculi IV-VII, Berlin. Nicolle D. 1994 Yarmuk 636AD: The Muslim Conquest of Syria, (Osprey Military Campaign Series 31), Oxford. Noth A. and Conrad L.I. 1994 The Early Arabic Historical Tradition: A Source-Critical Study, 2nd edn., trans. M. Bonner, (Studies in Late Antiquity and Early Islam 3), Princeton. Ostrogorsky G. 1968 History of the Byzantine State, 2nd edn., trans. J. Hussey, Oxford. Palmer A. 1993 The Seventh Century in the West-Syrian Chronicles, (Translated Texts for Historians 15), Liverpool. Sahas D. 1999 Why did Heraclius Not Defend Jerusalem and Fight the Arabs? In Parole de l’Orient 24: 79-97. ShahƯd I. 1995 Byzantium and the Arabs in the Sixth Century, Washington DC. Stathakopoulos D. 2004 Famine and Pestilence in the Late Roman and Early Byzantine Empire: A Systematic Survey of Subsistence Crises and Epidemics, (Birmingham Byzantine and Ottoman Monographs 9), Aldershot. Steven Notley R. and Safrai Z. 2005 Eusebius, Onomasticon: The Placenames of Divine Scripture. A Triglott Edition with Notes and Commentary, Leiden. Stratos A.N. 1972 Byzantium in the Seventh Century II: 634-41, trans. H. Hionides, Amsterdam. Treadgold W. 1997 A History of the Byzantine State and Society, Stanford. Tsafrir Y. - di Segni L. and Green J. 1994 Tabula Imperii Romani: Iudaea Palaestina. Eretz Israel in the Hellenistic, Roman and Byzantine Periods: Maps and Gazetteer, Jerusalem. Wallace-Hadrill J.M. 1960 The Fourth Book of the Chronicle of Fredegar with Its Continuations, London. Wander S.H. 1973 The Cyprus Plates: The Story of David and Goliath. In Metropolitan Museum Journal 8: 89-104. Whitby M. 2000 The Ecclesiastical History of Evagrius Scholasticus, (Translated Texts for Historians 33), Liverpool. Wolf K.B. 1990 Conquerors and Chroniclers of Early Medieval Spain, (Translated Texts for Historians 9), Liverpool.

since Emmaus lay along the main road from Jerusalem to Joppa on the coast. Infected vermin followed the Muslim forces from Gabatha to Emmaus, and plague broke out in full there during 639. In brief, the defeat of the Byzantines at Gabatha tells us little about the morale or skills of the Byzantine army at this period. The emperor Heraclius was simply unlucky in that when he did finally concentrate his forces in order to inflict one decisive defeat upon the Arabs, they were struck by plague. At a time when an exhausted military establishment was only ever one serious defeat from disaster, this was indeed disastrous. Bibliography Conrad L.I. 1981 The Plague in the Early Medieval Near East. PhD, Princeton University. Conrad L.I. 1990 Theophanes and the Arabic Historical Tradition: Some Indications of Intercultural Transmission. In Byzantinische Forschungen 15: 144. Conrad L.I. 1992 The Conquest of ArwƗd: A SourceCritical Study in the Historiography of the Early Medieval Near East. In Cameron A. and Conrad L.I. (ed.), The Byzantine and Early Islamic Near East I: Problems in the Literary Source Material, (Studies in Late Antiquity and Early Islam 1), Princeton: 317401. De Goeje M.J. 1900 Mémoire sur la conquête de la Syrie, 2nd edn. Leiden. De Miroschedji P. 1999 Yarmuth: The Dawn of City States in Southern Canaan. In Near Eastern Archaeology 62: 2-19. Donner F.M. 1981 The Early Islamic Conquests, Princeton. Doran R. 1992 The Lives of Simeon Stylites, (Cistercian Studies Series 112), Kalamazoo. Evetts B. 1907 History of the Patriarchs of the Coptic Church of Alexandria II: Peter I to Benjamin I (661). Patrologia Orientalis I, 383-518. Fishbein M. 1997 The History of Al-TabarƯ VIII: The Victory of Islam, New York. Glare P.G.W. 1996 Oxford Latin Dictionary, Oxford. Greatrex G. and Lieu S.N.C. 2002 The Roman Eastern Frontier and the Persian Wars, Part II: AD363-630, London. Hoyland R.G. 1997 Seeing Islam as Others Saw It: A Survey and Evaluation of Christian, Jewish, and Zoroastrian Writings on Early Islam, (Studies in Late Antiquity and Early Islam 13), Princeton. Jandora J.W. 1985 The Battle of Yarmnjk: A Reconstruction. In Journal of Asian History 19: 8-21. Kaegi W.E. 1977 The First Arab Expedition against Amorium. In BMGS 3: 19-22. Kaegi W.E. 1992 Byzantium and the Early Islamic Conquests, Cambridge. Kazhdan A.P. 1991 The Oxford Dictionary of Byzantium, Oxford.

376

Cavalry in Late Roman Warfare Hugh Elton Ancient History and Classics Department, Trent University

This article attempts to use a problem in tactical history to pose questions about late Roman warfare. The tactical problem is how cavalry came into contact with enemy, i.e. the charge. Although this article focuses on cavalry in the Late Roman East, much of it should have greater applicability.

Anatolia, the plains and river valleys of Mesopotamia, and the deserts (of various sorts) of South Syria and Arabia. The Romans fought the Sassanid Persians in most areas at one time or another, though most of the efforts against the Persians took place in Mesopotamia. But they also faced Lazi, Albani, and Huns in the north and Arabs and Saracens in the south. Mauricius’ Strategikon made clear the difference between fighting Sassanids who would engage in hand to hand combat, and fighting Turks, Huns, or Avars who preferred to keep at a distance. In both cases, the Romans were advised to defeat the enemy hand-to-hand, but for different reasons. The Sassanids generated an arrow storm before contact, and it was important to minimise the time of exposure to this, whereas the Turks, etc. preferred to wear the enemy down gradually and thus avoiding contact was playing into their hands.6 This complexity suggests that a single view of Late Roman battle can only be a high-level perception.

Before 1976, most discussions of ancient battles by modern writers were simple narratives, based on the perspective of the general. Then in 1976, John Keegan’s Face of Battle was published. This has been very influential; indeed, it often seems to be the only book on non-ancient military history read by ancient historians. The predominant method of discussing battle then switched to the experience of battle, how it felt to be a soldier. Perhaps the best exponent of this has been Hanson, whose The Western Way of War is concerned with hoplite warfare in Classical Greece.1 This focussed on assembling anecdotes from many different engagements to tell a story, hoping in this way to fill the gaps in our evidence. Keegan, however, did not attempt to characterise particular eras of warfare, but only to capture the experience of three specific days in 1415, 1815, and 1916. This is very different from the approach of ancient historians creating a ‘typical’ battle.

The cavalry charge is often seen as the definitive moment in combat for horsemen and as such, provides a good way to examine the problems in describing ancient battle. First, literary examples from the sixth century are discussed. These are compared to a more theoretical example, from the late sixth century, from Mauricius’ Strategikon. Lastly, this can be compared to an early modern manual from the American Civil War. Few Roman historians provide significant details about cavalry combat in the sixth century, while earlier Ammianus Marcellinus, for example, says nothing about the mechanics of combat in the east. Theophylact might be considered typical, when he described an action at the river Nymphius in 582: ‘the Persians made a counterattack when they saw that the Roman cavalry was exhausted by the exertion of the chase’.7 Or Procopius about Satala in 530:

In recent years, a large amount of work has been done on the Roman battle, though little of this has concentrated on the late Roman period. Two approaches can be distinguished. One is the ‘experience’ approach, concentrating on what it felt like to be in battle, e.g. Matthews, Lee, Lendon, and Lenski.2 The other is the ‘mechanics’ approach, e.g. Zhmodikov, Sabin, and McGeer.3 And some works are able to combine the two, e.g. Goldsworthy and Daly.4 There have also been a number of books on various aspects of ancient cavalry; in particular, advances have been made in our understanding of Roman horses and saddles.5

Thereupon Florentius, a Thracian, commanding a regiment of cavalry, rushing (Ðrm»saj) into the enemy’s centre and seizing the [Persian] general’s standard, forced it to the ground.

Many of these approaches, however, by focussing on generalities, minimise the differences between various military actions. To some extent, generalising is necessary, allowing complex situations to be simply understood. But this complexity is also what made late Roman warfare different from warfare elsewhere in time or space. There were three major theatres of operations in the late Roman East, the mountains and valleys of

This suggests that contemporaries were unable to give a clear description of how cavalry fought. It is also obvious that cavalry did not gallop directly into the opposition. Occasionally, however, we do have more detail as when Procopius also described at Satala ‘both sides kept making advances upon their opponents and retiring quickly for they were all cavalry.’8

1

Hanson 1989; Hanson 1994. Matthews 1989: 279-303; Lee 1996; Lendon 2005; Lenski, this volume. 3 Zhmodikov 2000: 67-78; Sabin 2000; McGeer 1995. 4 Goldsworthy 1996; Daly 2002. 5 Dixon - Southern 1992; Hyland 1990; Connolly 1987; Van DrielMurray 1991. 2

6

Mauricius, Strategikon 11 preface. Theophylact I.9.11. 8 Procopius, Wars I.15.15 (Loeb translation cited); cf. I.13.25-28. 7

377

THE LATE ROMAN ARMY IN THE EAST A similar vagueness about combat is found in theoretical literature, like Mauricius’ Strategikon. Mauricius recommended that cavalry regiments form up for battle between 8-10 deep at a maximum and 5 men deep at a minimum. This he contrasts with the standard practice of his day, of all units deploying 10 deep at all times.9 Mauricius also recommended two lines of cavalry units, though it is clear from his arguments that other generals recommended only one row of units.10 Mauricius’ vision of deployment is very clear, but becomes less clear when he describes the approach to the enemy. No distinction is made according to the type of enemy, i.e. both cavalry and infantry are to be approached in an identical fashion. The first description of the attack is in section 3.4, concerned with deploying the regiment for battle.

experience in North America and an observer in the Crimea.13 Cooke distinguishes between three modes, in line, column and ‘as foragers’. The charge as foragers is recommended for use against artillery or for pursuit. For charging in line, Cooke recommends that the unit walk for 20 paces, then trot for 60 paces, gallop for 80 paces, and then ‘charge’. Like Mauricius, Cooke says nothing about the moment of contact with the enemy, but is greatly concerned with order, prefacing the section on charges with a stern admonition: ‘as soon as any confusion is observed, it is necessary to halt and recommence the movement’. He also does not distinguish between charging against cavalry and infantry. Other nineteenth century material also emphasizes control, even in the face of heavy artillery fire, as during the Charge of the Light Brigade in the Crimean War.14

When, within three or four bowshots of the enemy line, the decision is made to close up by the flanks and by the rear, depending on the circumstances, the command is ‘Close Ranks’. They then gradually close up while continuing the advance. The archers open fire, and the whole line together in close order begins the charge [Gk. sumb£llei].11

We can now make some comments on the whole process of the attack. The primary consideration in all types of evidence is that of the importance of order. The skills required to achieve this could only be acquired through training and practice as a unit. The acceleration to contact occurs when the archers have begun shooting. Bowshot is generally reckoned to be about 300 m, so the two lines would be close together. To cover 300 m at a trot (c. 16 kph = 4.5 m/sec) would take approximately 66 seconds. If the enemy were also cavalry units approaching at the same speed, the time to contact would probably be c. 33 seconds. Reaching a full gallop can also be seen as unlikely, as there would not be enough time for the formation to reach this speed. As the two sides closed, archers would begin to shoot. At this range, however, archers were not shooting at individual targets, but simply volleying in the general direction of the enemy. Sighting would also be difficult as the archers were rarely in the front ranks of regiments. For aimed shot at individuals, effective range was less than 100 m.15 In any case, the archers would have had little time to shoot, and rate of shot was probably privileged over accuracy. In all these cases, the precise accuracy of times and distances is less important than the fact that distances and times in combat were short.

The next section (3.5) is concerned with drilling the regiment. At the command ‘Close ranks’, the soldiers close up from the rear for the charge. With the troops marching in close formation, particularly after they have closed in tightly from the flank, the archers open fire and the command is given: ‘Charge’ [Percute=strike]. The dekarchs and pentarchs then lean forward, cover their heads and part of their horses’ necks with their shields, hold their lances up high as their shoulders in the manner of the fair-haired races, and protected by their shields they ride on in good order, not too fast but at a trot, to avoid having the impetus of their charge [Gk. tÍ ÑxÚthti tÁj ™las…aj lit: the sharpness of their riding] break up their ranks before coming to blows with the enemy, which is a real risk. All the archers to the rear should open fire.12 Mauricius then moves on to pursuit formations, avoiding the issue of what actually happened when the two battle lines meet each other. He appears more concerned with control than with speed. The same emphasis is given in the following section, on pursuits. The only idea of what happened after contact is provided by 3.12 suggesting that if the Romans were repulsed, they should retire, then wheel and re-engage. This matches exactly Procopius’ description of the battle of Satala.

Describing the moment of contact was as difficult for contemporaries as it is for us, even though Ammianus, Procopius, and Mauricius had all had extensive combat experience. The moment of contact itself is not discussed, suggesting genuine uncertainty as to what happened. Moreover, the lack of details in the sources makes this a highly theoretical exercise. It can be divided into several possibilities (though the fact that ancient sources do not distinguish between targets may suggest an over schematization on our part). When cavalry charged heavy infantry, they were hoping that the infantry broke and ran. If the infantry held firm, then the cavalry would have had two choices. They could come to a halt and start trying to

Mauricius’ instructions for a cavalry charge can usefully be compared with the 1862 Cavalry Tactics written by Philip Cooke, a US cavalry officer with combat 9

Mauricius, Strategikon II.6. Mauricius, Strategikon II.13. 11 Mauricius, Strategikon III.4 (translations from Dennis, G.T., Maurice’s Strategikon (Philadelphia 1984). 12 Mauricius, Strategikon III.5. 10

13

Cooke 1862. e.g. Brighton 2004. 15 McLeod 1965. 14

378

HUGH ELTON: CAVALRY IN LATE ROMAN WARFARE break into the formation with their spears in hand to hand combat. Or they could veer off and, if equipped with missiles, fire these at the infantry in the hope of breaking their formation. If the infantry broke, or if they were dispersed infantry, then the cavalry rode them down. Against cavalry, the results were different. As both sides met, a swirling, disorganised melee began. We can only rarely provide detailed descriptions. At Callinicum in 531, the Roman infantry formed a foulkon. ‘Many a time after giving up, the Persians would advance against them determined to break up and destroy their line, but they always retired again from the assault unsuccessful’, similar to events at the Hippis in 549. Satala well demonstrates cavalry combat, but we have no detailed accounts of cavalry in battle breaking up infantry, though this did happen at Tell-Beshmai in 502 (at night) and outside Phasis in 556.16

Aurelius Victor described the Alamanni as ‘a people who fight wonderfully on horseback’.19 These superlatives depend on literary accounts which are particularly concerned with individual abilities in riding or weapon handling, like Procopius’ description of the Goth Totila’s cavalry dance before the battle of Busta Gallorum.20 These skills, however, have little to do with the ability to deliver a cavalry charge. Armies, however, function very differently from individuals, as well as differing between themselves. The Roman Empire had a highly developed infrastructure, which included an effective system of providing horses and training troops. Most mounts were supplied by the state, either by purchase or from imperial horse farms.21 The process was administered by the tribunus (later comes) stabuli, a post first attested in the mid-fourth century. He supervised a number of stratores who could reject horses they were offered if they did not meet fixed standards.22 Difficulty in obtaining sufficient horses easily is suggested by two other sources of remounts. Direct levies of horses could occur, while if a cavalry veteran’s son enlisted with one horse, he could join a cavalry unit directly (and with two horses was given the rank of circitor).23 Once acquired, horses were managed. If a horse went lame or fell ill, there was a Roman concept of veterinary medicine.24 Soldiers received a ration allowance (annona), and cavalrymen also received a fodder allowance (capitus) for their horses. The allowance was nominal. Practical estimates of how much food a horse needed vary, but 2.2 kg of grain and 6.8 kg of forage, as well as 36 litres of water per mount seem typical, albeit low. These were minimal estimates, though probably rarely exceeded on campaign.25 Once recruits were equipped, mounted, and organised they participated in an extensive system of exercises, ranging from individual weapon drills up to fullscale army manoeuvres. This would give the Roman army expertise in many of the problems faced in battle, including practice in operating as a formation.26 This expertise was systematically retained as many men served for long periods, e.g. Valerius Iuventinus who served in a regiment of Dalmatae for 20 years in the late third century or Fl. Marcus, who served in the vexillatio Fesianesa for 23 years in the fourth century. Long periods cannot be documented later because of a change in epigraphic habits.27

Each of these battles, Callinicum, Hippis, and Satala, was unique. All of this should be combined to suggest that we should not be trying to create a model of a typical battle, but rather to develop an understanding of the mechanics of battle and the factors which affected combat. It also needs to be connected to an understanding of the nature of Late Roman cavalry forces and of the type of history we are trying to write. How can these tactical thoughts be combined with other areas, e.g. subjective comments on quality, or less often described logistical material? Procopius’ description of mounted archers at the start of Book 1 of the Wars has had a great impact on many modern writers who rarely see Late Roman cavalry as an effective force until the sixth century. Although many units were armed with both bows and lances, other sixth century evidence shows that this was not always the case and that not every trooper was able to shoot.17 Procopius was not describing any technological changes, and other ancient writers were less worried about Roman cavalry. Vegetius in the late fourth or mid-fifth century pointed out that Roman cavalry had improved as a result of contact with Goths, Alans and Huns. He also noted that On cavalry there are many precepts, but since this branch of the military has progressed in its training principles, types of armour, and breed of horses, I do not think there is anything to be gained from books, since the present state of knowledge is sufficient.18

A separate infrastructure existed for campaigning, when horse losses could be enormous, if rarely mentioned by contemporaries. It has been estimated for Napoleon’s

Such comments are hard to interpret. The same is true of the superlatives lavished on many of their enemies, especially the Sarmatians, and Huns. Thus Vegetius makes positive comments about the horsemanship of barbarian cavalry in the fourth century where he praises the emperor’s horsemanship by stating that ‘the nation of the Huns and Alans would like to imitate if it could,’ and

19 Vegetius, de re militari III.26; Aurelius Victor, De Caesaribus 21; Ammianus Marcellinus XXXI.2.6, 20; Agathias V.19.12. 20 Procopius, Wars VIII.31.17-20 21 Elton 1996: 116; cf. for the early Empire, Davies 1969. 22 strator rejecting horses, Ammianus Marcellinus XXIX.3.5; standards, CT VI.31.1 (365/373) 23 CT VII.22.2 (326). 24 Mezzabotta 2000; Doyen-Higuet 1984. 25 Haldon 1999: 287. 26 Rance 2000. 27 ILS 2792, 2783

16 Procopius, Wars I.18.47; VIII.8.31-32; Rance 2004; Joshua the Stylite, Chronicle 277-278; Agathias III.26.8. 17 Procopius, Wars I.1.12-15; Mauricius, Strategikon II.8. 18 Vegetius, de re militari I.20, III.26.

379

THE LATE ROMAN ARMY IN THE EAST campaigns that he lost at least three horses for every human casualty and in the 1812 campaign lost at least 130,000 horses.28 If a horse fell out of service, there was a remount system which meant that many soldiers had a spare horse available and Belisarius is praised by Procopius for his rapid replacement of horses lost in battle.29 A full system of campaign supply existed, which included large supply dumps, like the huge haystack recorded at Batnae in 363. These allowed the army to remain concentrated at all times. Forage and water were the major problems, since they could not be carried in required quantities.30 Replacement of consumable items was also necessary for prolonged campaigning, especially arrows, and javelins. Although individuals carried only 30-40 arrows each, these numbers rapidly mounted up. A proposed requisition for an expedition in 911 included 800,000 arrows and 10,000 javelins for an army of about 34,000 men.31

Strategikon), by political units, there was probably no training. We should thus expect these armies to be more adept at skirmishing than at pitched battles, and we should not expect to find many accounts of them people being able to make effective charges. The large numbers of horses required (because they tired easily) would also have had an effect on operations. The Strategikon thus comments in the case of the Scythians, Turks, etc., ‘they are hurt by a shortage of fodder [boskÁj] which can result from the huge number of horses they bring with them’. Consequently, they avoided fortified camps and instead moved in small scattered groups, concentrating only for battle.35 Nor should we expect to find Saracens or Huns carrying out sieges or long campaigns, since these would require extensive logistical preparation. In 626 the Avar Chagan claimed, very plausibly, that he abandoned the siege of Constantinople after about 5 weeks for logistical reasons.36

Far less can be said about these processes among Rome’s enemies. These need to be divided into two groups, the state of the Sassanid Persians and the less developed powers like the Huns or Turks. Although the Persians only had a large standing army from the sixth century, they had a permanent administration that was able to organise logistics. This is shown in particular by their ability to manage long sieges but also an ability to improvise. In 586, Roman strategy in Mesopotamia before the battle of Solachon revolved around denying the Persians access to water; the Persians in response organised a camel train carrying water. On other occasions, things could go wrong, as at the siege of Archaeopolis in Lazica, where 20,000 horses were lost on account of a shortage of fodder.32 The existence of military manuals also suggests some concept of training, as does the existence of named regiments.33 The Persians could thus undertake long sieges, and could charge home against Roman infantry (even if not always with success).

Thus, the differing infrastructures available to protagonists in the Late Roman East did affect the sorts of battles that were fought. This is true, even if the source material available is rarely sufficiently detailed to understand events at a tactical level, or to provide a detailed knowledge of a campaign’s logistics.37 Detailed examination of weaponry does not do this, though it attracted much attention from both ancient and modern writers.38 Reconstructing any battle, ancient or modern, is a difficult process for many reasons. Success in war is achieved by imposing one’s will on the enemy. This can be achieved in many ways, including but not limited to battle. Although battle was an attractive subject for authors writing within the canons of classical historiography, it does not follow that this typified late antique warfare. On the contrary, the extant manuals focus on avoiding pitched battle wherever possible, in part because battle was unpredictable. Thus the focus on battle by both ancient and modern authors is misleading. But if we follow the manuals in our interpretation, we will look at factors which maximised the strengths of the empire, i.e. logistics, planning, and discipline, and thus stress structural and thus Roman supremacy. Such an interpretation cannot explain how the Persian army was able to sack Antioch in 260 and 540, or besiege Constantinople in 626. Understanding Hun or Avar success is even harder in this world. Neither approach can stand alone. Nor is the use of theory to reconstruct tactical necessities sufficient. In other words, a far more complex approach to understanding ancient war is necessary. If we write only about the face of battle, we miss the face of war.

But other powers had less structure, and this in turn limited their abilities. Thus, even amongst the Huns, veterinary science and horse care were probably less developed. Vegetius’ Mulomedicina suggests that Hun horses were not stabled in winter, but pastured. This supposedly bred hardiness, but would probably also contribute to underfeeding and small-sized horses. To some extent, this could be counteracted by having spare horses (as Ammianus describes for the Sarmatians), though these were probably concentrated among the rich.34 With little concept of organisation (except as suggested by the 28

Elting 1989: 317, 319. Haldon 1999: 142-3; Procopius, Wars VII.1.8; Mauricius, Strategikon V.2. 30 Ammianus Marcellinus XXIII.2.7. 31 Mauricius, Strategikon I.2, XII.B.5; Haldon 2000; De Caer. 657.1015. 32 recent overview, Greatrex 1998: 52-59; Theophylact, II.1.5-6, 2.4; Procopius, Wars VIII.14.44; Sebeos 113-114; Joshua the Stylite, Chronicle 295. 33 Inostrancev 1926; Hamblin 1986: 99-106. 34 Ammianus Marcellinus XVII.12.3. 29

35

Mauricius, Strategikon XI.2. Chronicon Paschale p.725. 37 cf. Trombley 2003. 38 Elton 1996: 107; Lendon 2005: 263; for one example of the ‘hardware’ approach, see Mielczarek 1993 or any of the literature cited there. 36

380

HUGH ELTON: CAVALRY IN LATE ROMAN WARFARE Sabin P. 2000 The face of Roman battle. In JRS 90: 117. Trombley F. 2003 Military Cadres and Battle During the Reign of Heraclius. In Reinink G.J. and Stolte B.H. (ed.), The Reign of Heraclius, Leuven: 241-259. Van Driel-Murray C. 1991 The Roman Cavalry Saddle. In Britannia 22: 33-50. Zhmodikov A. 2000 Roman Republican Heavy Infantry in Battle (IV-II Centuries BC). In Historia 49: 67-78.

Bibliography Brighton T. 2004 Hell Riders, London. Connolly P. 1987 The Roman Saddle. In Dawson M. (ed.), Roman Military Equipment, (BAR S336), Oxford: 7-27. Cooke P. 1862 Cavalry Tactics, Philadelphia. Daly G. 2002 Cannae: The Experience of Battle in the Second Punic War, London. Davies R.W. 1969 The supply of animals to the Roman army and the remount system. In Latomus 28: 429-459. Dennis G.T. 1984 Maurice’s Strategikon, Philadelphia. Dixon K. - Southern P. 1992 The Roman Cavalry, London. Doyen-Higuet A.-M. 1984 The Hippiatrica and Byzantine Veterinary Medicine. In DOP 38: 111-120. Elting J. 1989 Swords around the Throne, London. Elton H.W. 1996 Warfare in Roman Europe, Oxford. Goldsworthy A. 1996 The Roman Army at War: 100 BC - AD 200, Oxford. Greatrex G. 1998 Rome and Persia at War, Leeds. Haldon J.F. 1999 Warfare, State and Society in the Byzantine World, 565-1204, London. Haldon J.F. 2000 Theory and Practice in tenth-century military administration. In Travaux et Mémoires 13: 201-352. Hamblin W. 1986 Sasanian military science and its transmission to the Arabs. Proceedings of the 1986 International Conference on Middle Eastern Studies, Oxford: 99-106. Hanson V.D. 1989 The Western Way of War, Berkeley. Hanson V.D. 1994 ed., Hoplites: The Classical Greek Battle Experience, London. Hyland A. 1990 Equus: The Horse in the Roman World, London. Inostrancev C.A. 1926 The Sasanian Military Theory. In Journal of the KR Cama Oriental Institute 7: 7-52. Lee A.D. 1996 Morale and the Roman Experience of Battle. In A. B. Lloyd (ed.), Battle in Antiquity. London: 199-217. Lendon J.E. 2005 Soldiers and Ghosts, London. McGeer E. 1995 Sowing the Dragon’s Teeth: Byzantine Warfare in the Tenth Century, Washington. McLeod W. 1965 The Range of the Ancient Bow. In Phoenix 19: 1-14. Matthews J.F. 1989 The Roman Empire of Ammianus, London. Mezzabotta M.R. 2000 Aspects of multiculturalism in the Mulomedicina of Vegetius. In Akroterion 45: 5264. Mielczarek M. 1993 Cataphractari and clibanarii. Studies in the heavy armoured cavalry of the ancient world, Lodz. Rance P. 2000 Simulacra Pugnae: The Literary and Historical Tradition of Mock Battles in the Roman and Early Byzantine Army. In GRBS 41: 223-276. Rance P. 2004 The Fulcum, the Late Roman and Byzantine Testudo: the Germanization of Roman Infantry Tactics? In GRBS 44: 265-326.

381

Centuriones ordinarii et ducenarii dans l’armée romaine tardive (IIIe -VIe s. apr. J.-C.)* Sylvain Janniard Université Paris-Sorbonne (Paris IV)

RO[……], princeps, dédie une stèle au Génie de la schole des ordinati.6 Les ordinati peuvent donc être groupés en scholes comme les centurions et il existerait un ordinatus princeps dont il faudrait rapprocher la fonction du centurion princeps des unités auxiliaires.7 En outre, les termes ordinarius et ordinatus doivent renvoyer au même référent sinon l’on ne peut expliquer les abréviations ord., ordi. ou ordin.8

L’apparition du terme ordinarius/ordinatus: A partir du début du IIIe siècle apparaissent dans la documentation épigraphique et papyrologique les termes d’ordinarius et d’ordinatus, seuls ou accompagnés de la mention du centurionat, pour désigner les centurions ou du moins certains d’entre eux.1 L’attestation la plus ancienne se trouverait dans un ordre de réquisition de chameaux de 203 apr. J.-C provenant d’Egypte. L’un des gradés envoyés pour cette mission est appelé ºIrrai'o" Mavlico" et porte le titre d’ojrdina'to".2 Par la suite, la documentation semble prouver qu’ordinarius et ordinatus sont employés comme synonymes de centurion.3 Ainsi, un état par centurie de la cohorte XXe des Palmyréniens conservé pour 218 apr. C. donne, col. ii, l. 21-22:

Certains commentateurs, tels H. Sanders et R. Cagnat, ont pensé que les ordinarii/ti constituaient un groupe particulier de centurions, supérieurs aux autres en grade et assimilables aux primi ordines. D’autres, comme Th. Mommsen et E. Stein, ont proposé de distinguer entre ordinati qui renverrait aux simples centurions et ordinarii qui désignerait les primi ordines.9 Dans un long article de 1940, J. F. Gilliam, reprenant des hypothèses déjà émises par A. von Domaszewski, a réfuté cette thèse. Les centurions de la première cohorte double légionnaire sont appelés primi ordines en raison du surcroît de prestige attaché à leur fonction. Or, nous trouvons aussi des ordinarii dans les unités auxiliaires ce qui interdit de les assimiler aux primi ordines exclusivement légionnaires.10 Bien plus, selon Gilliam, l’inscription d’Alsó-Kosály montrerait aussi une hiérarchie similaire à celle des centurions à l’intérieur des ordinati puisque ce seraient des principes qui auraient fait poser la dédicace au Génie de la schole des ordinati. D’après lui, les termes équivalents d’ordinarius et d’ordinatus proviendraient du sermo castrensis et désigneraient des centurions commandant effectivement des centuries par opposition à des centurions affectés à d’autres tâches.11

(centuria) Marci Muciano co(n)s(ule) ord(inarius/tus) Iul(ius) Marcus et plus loin, col. iv, l. 16-7: (centuria) Antonini Victorino co(n)s(ule) ord(inarius/tus) Domittius Antoninus.4 D’autres textes du même siècle viennent renforcer l’assimilation entre ordinarius/tus et centurion, telle la dédicace de Valerius Crescens, qui porte le titre d’ordinar(ius) princeps vexillationis.5 En Dacie, Aurelius *Je remercie pour leurs remarques Jean-Michel Carrié, Maurice Sartre, Yann Le Bohec et Pierre Cosme. 1 Sur les ordinarii/ti, cf. Stein 1933; Gilliam 1940; Rea 1980. 2 P. Flor. II, 278 r = Daris 1964: 64 = ChLA, XXV, 779, ii, 26 et iii, 2 (un autre ordinatus est attesté en v, 6). Il s’agit d’une réquisition remboursée de chameaux ordonnée par le préfet d’Egypte, envoyée aux stratèges de nome et conservée dans le registre des lettres de la garnison de Babylone/Memphis. La datation est celle d’A. Stein (Archiv für Papyrusforschung 6, 1920: 214-216). Gilliam 1940: 136, n. 41 pense que Malichos servait dans un numerus palmyrenorum attesté à Coptos sous Caracalla (IGR I, 1169 = SB 8810, 10), voir depuis AE 1954: 209. 3 Pour l’ensemble des témoignages d’ordinati/rii antérieurs à l’époque constantinienne, cf. Gilliam 1940: 3-17; Drew-Bear 1981: 109-110; Richardot 1995: 419-420, auquel il convient d’ajouter AE 1991: 1620, AE 1992: 786 = AE 1993: 820, RIU VI, 1355. Pour l’ensemble des témoignages papyrologiques, cf. Rea 1980: 217. 4 P. Dura 98. Dans P. Dura 100 (219 apr. J.-C.), P. Dura 82, col. i, l. 1 et 8 (223-235) et P. Dura 89, col. i, l. 5-6, 11 (239), tous les centurions portent aussi le titre d’ordinati. Il semble en aller de même pour P. Dura 101 (222 apr. J.-C.). 5 AE 1926: 146 (Sadouri, Numidie): Genio Ausum| Valerius Cresce|[n]s ordinar(ius) prince|[ps] u[e]csi[l]ationis| [et] M[a]nil[i]us Felix| [op(tio), una cun com|[mil]itonib[us d(e)d(icauerunt). L’inscription est datée par Carcopino 1925: 139-143, de 240-250 apr. J.-C., époque supposée de l’occupation du fortin où elle fut trouvée. Il s’agirait, toujours selon lui, d’une vexillation d’un numerus palmyrénien. Un autre ordinarius princeps est connu à Anz en Syrie, cf. AE 1940: 211 = IGLS XVI 1299.

6

CIL III 830 = 7631 (Alsó-Kosály, Dacie): …Ge]nio sco|les ordi|natorum suc c(uragentibus) L(ucio) Cilio| Aeliano et| [T]ib(erio) Aurel(io) Ro[…]| princip(ibus) pos(uerunt), restitution de Gilliam 1940: 135. 7 Sur le centurio princeps des cohortes auxiliaires, cf. Domaszewski 1981: 56-57. Dans le cas présent, toutefois, le titre de princeps renvoie au commandement ad hoc exercé par Crescens sur le détachement, cf. Speidel 1981: 7-13. 8 Il ne s’agit pas de nier ici qu’ordinatus ait pu conserver dans certaines occurrences une valeur participiale (e.g. CIL V 7865 = ILS 4664; CIL VI 3603 = ILS 2268). Toutefois, dans la totalité des références que nous utilisons, il est employé comme substantif, marquant l’état qui résulte de l’action accomplie (en l’occurrence, ordinem accipere…). 9 Cagnat 1900: 197; Sanders 1931: 275-279; Mommsen 1913: 376-377; Stein 1933. 10 E. g. CIL VIII 2505 = 18005 (El Kantara, Numidie): Maximo Zabdiboli, Hadri(a)n(o) Palm(yreno), ve[ter](ano)| ex ordin(ato/ario) n(umeri) Pal(myrenorum), uix(it) an(nis) LXXV, Valeria Dulc[is]| uxor et heres eius, praeter quam quot te[s]|tamento cautum est adiecta pecunia [de li]|beralitate sua, marito karissimo f[ecit]. Voir aussi CIL XIII 8208 = ILS 4762 (Aurelius Verecundus, ordinarius/tus Britonum) 11 Gilliam 1940: 135 (Alsó-Kosály) et 143-144 (conclusion générale). L’assimilation entre ordinatus, ordinarius et centurio avait déjà été

383

THE LATE ROMAN ARMY IN THE EAST Un certain nombre d’objections peuvent être faites à l’encontre des interprétations de J. F. Gilliam. En premier lieu, les centurions détachés de leur unité pour des tâches particulières portent généralement un titre spécifique, tels les centurions frumentaires ou régionaires, qui les distingue des centurions restés dans les unités. Les centurions appelés proprement surnuméraires ne sont attestés que par deux inscriptions et un papyrus. Or, l’ensemble des occurrences les montre dans des détachements ou en conjonction avec les frumentaires. Il se peut alors qu’il se soit agi de gradés promus au centurionat loin de leur unité d’origine, à laquelle ils seraient restés attachés administrativement mais en surnombre par rapport aux collèges des centurions de leur unité de provenance. Il est en tout cas légitime de se demander si leur développement au IIIe siècle fut d’une ampleur telle qu’il fallut renommer tout le corps des centurions pour les en distinguer.12 Enfin, si les termes ordinarius et ordinatus avaient eu dès leur apparition le sens générique de centurion, on expliquerait difficilement le besoin de continuer à préciser qu’il s’agissait de centurions et le maintien dans notre documentation jusqu’au VIe s. des qualificatifs centurio, centurio ordinarius et ordinarius.

vexillations formées de leur propre unité et d’autres détachements.15 Le détachement précisément stationné à Alsó-Kosály est sous les ordres d’un centurio princeps vexillationis, Aurelius Ro[……].16 Ce dernier dédie la stèle au collège des ordinati du corps expéditionnaire. L’inscription confirme l’adéquation établie précédemment entre centurion et ordinatus mais l’absence de précision sur le rang d’Aurelius ne nous permet pas d’aller plus loin quant à la nature exacte des ordinarii/ti.17 Des arguments décisifs contre les hypothèses de Gilliam sont cependant venus de deux textes épigraphiques qu’il n’eut pas à sa connaissance. Le premier est l’épitaphe d’Aurelius Gaius qui servit dans le comitatus des Tétrarques dans trois légions successives avant de se retirer vers 303 apr. J.-C. Gaius indique dans l’ordre ascendant les différents degrés de l’optionat qu’il a parcourus. Il fut successivement optio triarius, optio ordinatus, princeps optio puis optio des comites.18 Sa carrière illustre clairement qu’une distinction était dressée entre les centurions ordinati et les autres centurions. Aurelius Gaius fut d’abord adjoint d’un centurion triarius puis d’un centurion ordinatus, ce qui passe dans l’inscription pour une promotion. La possibilité que tous les centurions n’aient pas été ordinarii/ti est confirmée par une seconde inscription encore d’époque tétrarchique, une stèle que Valerius Castus, centurio ordinarius/tus de la légion Ia Iouia Scythica, fait poser pour ses deux filles, en compagnie de son beau-père, Valerius Valerianus, centenarius.19 L’absence de toute précision quant à l’unité de Valerianus invite à faire de ce dernier un centenier de la même légion. On peut donc trouver dans une même unité un ordinarius et un centenarius, ce qui interdit d’assimiler leur fonction. Or, le grade de centenarius a été en usage dans les nouvelles unités issues de la réorganisation militaire de Dioclétien et de Constantin pour, semble-t-il, y remplacer l’ancien centurion.20 Les inscriptions d’Aurelius Gaius et de Valerius Castus montrent ainsi définitivement que le

La dédicace d’Alsó-Kosály à la schole des ordinati ne nous éclaire pas plus sur le sens de ce terme, contrairement à ce qu’estimait J.F. Gilliam. L’abréviation PRINCIP, l. 8, désigne le dédicant. Il s’agit donc d’un nominatif selon le modèle des dédicaces religieuses de ce type.13 Lucius Cilius Aelianus ne peut être compté parmi les dédicants, il est tout simplement l’officier le plus gradé sous les ordres duquel agissent les autres centurions. Les six lettres difficilement compréhensibles l. 7 ne peuvent contenir à la fois le cognomen d’Aurelius et un autre nom. Il convient alors de lire: Ge]nio sco/les ordi/natorum/ suc c(uragente) L(ucio) Cilio/ Aeliano t/ib(uno), Aurel(ius)/ Ro[…]/ princip(s)/ pos(uit).14 Il doit s’agir ici de vexillations orientales, au vu de la forme princips et du génitif grec de schola, sous les ordres d’un tribun. Aurelius Ro[……] n’appartient pas nécessairement à l’unité dépendant directement de Lucius Cilius Aelianus. Des tribuns peuvent commander des

15

IGR I, 623 = ILS 8851 (Publius Aelius Hammonius, vers 240 apr. C.), RIB I, 2042 (Caelius Vibianus). 16 Cf. note 7. 17 L’hypothèse qui ferait d’Aurelius Ro… un princeps de la première cohorte légionnaire ne modifierait pas notre interprétation du terme ordinatus, les primi ordines pouvant être eux aussi considérés comme des ordinati. 18 Drew-Bear 1981: 93-141 = AE 1981: 777 et SEG 31: 1116 (Ada Köy, Phrygie). Drew Bear 1981: 109-111, conclut à juste titre que tous les centurions ne sont pas ordinati mais se range toutefois aux arguments de Gilliam au prétexte que les triarii auraient été des centurions de réserve. Ce statut paraît cependant incompatible avec la position de cette fonction, premier degré de l’optionat, dans la carrière de Gaius. 19 AE 1990: 866 = AE 1989: 641 (Hâršova/Carsium, Mésie inférieure, stèle de marbre blanc, en remploi dans une nécropole des IVe-Ve s.): D(iis) M(anibus)| b(ene) m(erentibus) Castae, quae| uixit an(num) I d(ies) V et| Reginae, quae uixit| ann(os) VII mens(es) V dies| XVIII, filiabus dul|cissimis, Val(erius) Castus| centur(io) ord(inarius) leg(ionis) I Iou(iae)| Scyt(hicae), una cum Val(erio)| Valeriano centen(ario), so|cero suo, posuit. 20 Jérôme Contre Jean de Jérusalem 19, Végèce Mil. II. 8. 8 et II. 13. 45. Cf. Grosse 1920: 117-118, Southern & Dixon 1996: 62.

proposée par Domaszewski 1981: 57, 60, 97 qui attribuait l’origine de la pratique aux numeri. 12 Même scepticisme dans Stoll 2001: 321. Attestations de centurions surnuméraires: CIL V 8278 = ILS 2333, CIL VI 1110, P. Beatty Panop. II, 183, 264, 269, 289. Herz 1999: 177-182, estime qu’il s’agit de centurions sans commandement chargés expressément de la logistique des vexillations ou des légions, en remplacement du primipile. Cependant, leur affectation à la logistique, qui apparaît uniquement dans le P. Beatty Panop. II, n’indique rien sur leur position dans la hiérarchie de l’unité et semble une spécialisation ni pérenne ni exclusive comme le montre, dans le même papyrus, l’emploi de centurions non surnuméraires aux mêmes tâches (cf. e. g. II, 59-60 [Dioscorus, centurio ordinatus], 195, 202). 13 E. g. ILS 2371. 14 CIL III 830 = 7631. On lit l. 5: AELIANO TE? (ligature) et l.6 IB AUREL. Le cognomen d’Aurelius pourrait être Romulianus ou Romuleius. Un Aurelius Romulianus, bénéficiaire de la legio Ia Adiutrix, est attesté en CIL III 11022 = RIU II, 565 (Brigetio, IIIe s.). Pour la forme grecque princips, cf. e. g. AE 1999: 1378.

384

SYLVAIN JANNIARD: CENTURIONES ORDINARII ET DUCENARII DANS L’ARMÉE ROMAINE TARDIVE (IIIE -VIE S. APR. J.-C.) grade d’ordinarius ne correspond pas exactement à ceux de centurion ou de centenarius et que dans la hiérarchie des officiers subalternes les ordinati/ordinarii occupent une position supérieure au moins aux triarii.

hommes (P. Dura 101 = Fink 1971: 2 = ChLA VIII, 356).23 Enfin deux acta diurna, le premier du règne de Sévère Alexandre et le second de 239 apr. J.-C. (P. Dura 82 = Fink 1971: 47 = ChLA VII, 337 et P. Dura 89 = Fink 1971: 50 = ChLA VII, 344) donnent respectivement 924 caligati pour 9 ordinati et 781 pour 6, soit une moyenne de 101 et 130 soldats par centurie.24 Il est très probable que dans le premier des deux acta, trois ordinati soient surnuméraires, portant la moyenne des centuries à 152 milites. Deux autres documents de 218-219 et de 224, qui semblent constituer plutôt des états partiels – peut-être des hommes disponibles au dépôt – permettent de calculer une moyenne de 58 soldats par centurie, ce qui correspond assez bien à une demi centurie à effectif élargi à laquelle la cohorte semble habituée (P. Dura 98 = Fink 1971: 6 = ChLA VII, 353 et P. Dura 102 = Fink 1971: 8 = ChLA IX, 357). Cette dernière évaluation se retrouve aussi sur un papyrus, peut-être d’affectation de tâches, daté de 242-250, comportant deux indications d’effectif de 57 et de 63 hommes pour deux centuries (P. Dura 112 = Fink 1971: 16 = ChLA IX, 367). Diverses hypothèses ont été avancées pour expliquer l’étrangeté de ces chiffres.

Ordinatus et ordinarius dans la documentation papyrologique L’argumentation de Gilliam reposait aussi sur ce qu’il estimait être la pratique des bureaux militaires, pour autant qu’elle nous est transmise par la papyrologie, d’appeler ordinati l’ensemble des centurions d’un même corps. Toutefois les textes qu’il avançait à l’appui de sa démonstration doivent être reconsidérés. Pour le premier d’entre eux, il s’agit d’une liste d’officiers subalternes dressée en 243-244 et conservée sur papyrus peut-être dans l’officium du préfet d’Egypte.21 Sont préservés les noms de cinq décurions d’aile accompagnés de la date de leur promotion à ce poste suivis, sous le chapitre de la cohors III Iturae[o]rum (l. 16), de deux noms supplémentaires précédés de l’abréviation ORDD (l. 17). Le premier éditeur du texte, Henry Sanders, avait proposé pour celle-ci le développement ord(o) dec(urionum) rejeté à juste titre par Gilliam au profit d’ordinati. Néanmoins, le papyrus est incomplet et la liste des ordinati de la cohorte nous manque après le deuxième nom. Rien ne prouve donc que l’ensemble des six centurions de l’unité auxiliaire portait le titre d’ordinati. Il n’est pas exclu en revanche que les ordinati, reportés en premier sur la liste, aient pu être supérieurs en grade à leurs collègues. Les autres témoignages papyrologiques ne s’opposent pas à cette hypothèse à l’exception des archives de la cohorte XXe des Palmyréniens de Doura Europos: trois états d’effectif, par centurie et année d’enrôlement, datés de 218 à 222 apr. J.-C et deux acta diurna datés des règnes de Sévère Alexandre et de Gordien III attribuent aux centurions de l’unité le qualificatif d’ordinatus, à l’exclusion de toute autre dénomination.

J. F. Gilliam y voyait la conséquence d’un détachement permanent de quatre centuries loin de la garnison mais cette explication échoue à rendre compte du terme de numerus purus avant les totaux des acta diurna ainsi que de l’immuabilité du chiffre et de l’identité des centuries manquantes qui n’apparaissent par ailleurs jamais dans le demi-siècle de documentation militaire conservée à Doura. En dernier lieu, avec ses 6 centuries et ses 5 turmes les effectifs de la cohorte sont déjà miliaires. Santo Mazzarino proposait d’y voir l’effet d’une mobilisation militaire accrue dans un contexte d’affrontements récurrents entre Rome et ses voisins parthes puis perses mais cela n’explique pas une moyenne de 120-130 hommes par centurie en permanence, y compris hors de tout contexte

Néanmoins, cette cohorte miliaire montée ne respecte pas le schéma organisationnel attribué d’ordinaire aux corps de ce type. Elle compte six centuries au lieu de dix et celles-ci comportent des effectifs anormalement élevés.22 Un état de 219 apr. J.-C. attribue respectivement 141, [1]46 et au moins 150 hommes à trois centuries (P. Dura 100 = Fink 1971: 1 = ChLA VIII, 355, col. x, 28, xxvi, 9, xxxi, 18). Pour un second état de 222, Robert Fink, à partir du décompte des noms subsistants et de l’espace occupé sur le document par chaque centurie, a estimé l’effectif médian de ces subdivisions tactiques à 120-130

23 R. Fink dans Welles 1959: 31. R. Marichal arrive à des résultats très proches, cf. ChLA VIII: 6-7. 24 Les chiffres de 914 et de 781 ne peuvent renvoyer qu’aux fantassins et non à l’ensemble des soldats de l’unité comme le soutenait J.F. Gilliam (dans Welles 1959: 271-272). Il s’agit en effet d’états quotidiens qui devaient permettre aux autorités militaires les consultant d’avoir une vision immédiate des effectifs disponibles par arme sans avoir besoin de faire de soustraction: il paraîtrait étrange, étant donné la nature de ces documents, qu’aient pu être systématiquement décomptés l’intégralité des officiers subalternes, les principaux gradés, les cavaliers et les méharistes à l’exception de la catégorie la plus nombreuse des fantassins. En admettant même que les calculs aient eu une finalité comptable, comme l’estimait Gilliam, il n’était pas plus compliqué de faire apparaître à part ceux qui percevaient les émoluments les plus bas afin d’obtenir à la fois une base comptable et un document à finalité militaire classique. Quant au terme de milites caligati, c’est précisément dans ce contexte comptable qu’il doit être compris comme renvoyant au simple soldat, c’est-à-dire à celui qui perçoit le stipendium le plus bas, en l’occurrence le fantassin. Ulpien, Dig., III. 2. 2. pr., commentant le sens de dimissus dans l’Edit du préteur, distingue très bien entre le simple soldat (miles caligatus) et le reste de la troupe jusqu’au commandant d’unité (uel si qui alius…). Gilliam confond en outre dans son analyse l’expression générique in caliga et le participe substantivé plus restrictif caligatus.

21

P. Mich. III, 164 = Fink 1971: 20 = ChLA V, 281. Le texte a tout particulièrement bénéficié des lectures de Gilliam 1958. 22 Ps Hygin Mun. cas. 27, Habet itaque cohors equitata miliaria centurias X peditum, equites CCXL, turmas, decuriones, VI; 1, Plena centuria habet milites LXXX (seule mention de l’effectif d’une centurie). Les effectifs des cohortes miliaires sont discutés en dernier lieu par Bartoloni 1996: 58-65, qui admet à la fois les hypothèses de Gilliam et de Santo Mazzarino (commentées infra) pour expliquer l’organisation de la XXe Cohorte des Palmyréniens.

385

THE LATE ROMAN ARMY IN THE EAST d’affrontement comme dans le P. Dura 82. R. Davies, quant à lui, a élaboré une hypothèse plus ingénieuse en supposant dans un premier temps le détachement de quatre centuries de la cohorte au titre de vexillation puis dans un second temps, en l’absence de réintégration du détachement, le retour à un effectif miliaire par gonflement de chaque centurie, ce qui constitue une explication assez compliquée pour arriver au même constat: les centuries de la cohorte ont des effectifs plus importants.25 Il nous semble préférable de suivre plutôt le jugement prudent de R. Fink (dans Welles 1959: 30, “The Dura texts must be allowed to explain themselves”, accepté par R. Marichal, ChLA IX: 11) et de considérer que les six centurions de la cohorte étaient régulièrement à la tête de centuries aux effectifs supérieurs à la normale. Ces centuries alourdies ont même pu valoir à leurs officiers le titre d’ordinatus en lieu et place de celui de simple centurion. Le reste de la documentation papyrologique antérieure à Constantin ne s’oppose pas à cette interprétation car elle ne livre que des témoignages isolés sur la présence d’un ou de plusieurs ordinati dans les unités de la garnison égyptienne.26

298). La place des ordinarii est à nouveau expliquée quelques chapitres plus loin dans l’Epitomé à propos de l’antiqua ordinatio legionis: “mais ceux qui se battent devant et autour du signum et aussi en première ligne, on les appelait principes, c’est à dire les ordinarii et les autres sous-officiers”.29 Végèce indique que dans son ‘ancienne légion’, la première ligne était occupée par les principes. Il fournit ensuite au lecteur l’équivalent contemporain des principes, c’est à dire les ordinarii et d’autres sous-officiers placés eux aussi en première ligne à son époque. Un autre passage de Végèce vient confirmer s’il en était encore besoin le rôle des ordinarii à partir de l’époque sévérienne. Toujours dans l’antiqua ordinatio legionis, il appelle les cinq centurions de la Ière cohorte double des ordinarii: “ainsi les dix centuries de la première cohorte sont dirigées par cinq ordinarii”.30 Certains commentateurs ont voulu disqualifier ces informations au regard de ce qu’ils considéraient comme des incohérences. Ainsi, au chapitre huit du livre deux, Végèce arrive-t-il au total de cinquante cinq centurions pour l’ensemble de la légion. Or, d’après le reste du paragraphe, il ne devrait y avoir que cinquante centurions dans une légion: quarante cinq centurions des cohortes II à IX et les cinq ordinarii de la première cohorte. La solution consiste à distinguer les ordinarii des autres centurions. Pour Végèce, les anciens centurions, assimilés aux modernes centenarii, exercent le commandement dans chaque centurie (singulas centurias curabant, II.8.13 et 14). La légion compte cinquante cinq centuries donc cinquante cinq centurions. La première cohorte ne fait pas exception à la règle: ses dix centuries sont commandées par dix centurions mais cinq d’entre eux, les centurions ordinarii, exercent des commandements particuliers, souvent de plus de cent hommes. La légion compte bien cinquante cinq centurions dont cinq centurions ordinarii qui n’apparaissent pas dans le décompte final, pas plus que le tribun par ailleurs.31

La fonction des ordinarii L’hypothèse semble confortée par l’explication que donne du terme l’Epitoma rei militaris de Végèce. Dressant la liste des dignitates secundum praesentes matriculas, il définit ainsi le grade d’ordinarius: “On appelle ordinarii ceux qui, au combat, mènent les lignes de bataille parce qu’ils sont les premiers”.27 Le rôle des ordinarii est clairement tactique, en rapport avec l’ordre de bataille. Ils sont placés par Végèce à la tête des ordines, or celui-ci emploie toujours ordo avec le sens de rang ou de ligne de bataille et jamais avec celui de centurie. Enfin, leur place est mentionnée: ils sont ‘les premiers’, c’est à dire en première ligne.28 Les ordinarii, à l’époque de l’Epitomé, occupent donc une fonction très importante dans les formations d’infanterie: placés aux premières lignes, ils mènent l’ensemble des ordines au combat. Ce qui explique leur rang dans l’unité, celui de premier officier subalterne en dessous du tribun (Végèce Mil. II. 7. 2, confirmé par P. Mich. X, 592 = ChLA, V,

En dernière analyse, les inscriptions d’Aurelius Gaius, de Valerius Castus et le texte de Végèce nous permettent de comprendre que les ordinarii/ti étaient distincts et

25

29

J. F. Gilliam dans Welles 1959: 27 avec les réserves de R. O. Fink dans le même ouvrage (30). Mazzarino 1971: 61-64 (la cohors serait normalement quingénaire selon lui), R. Davies dans Epigraphische Studien IV, 1967: 109, n. 17. 26 Cf. P. Mich. VII, 450 et 455 r = CPL 132-133 = Fink 1971: 52 = ChLA, XLII, 1213, b 4, P. Mich. VII, 454 = CPL 146 = Fink 1971: 30 = ChLA V, 276, col. ii, 9, P. Oxy. XLI, 2953, P. Beatty Panop., II, 59-60, 190 (légions). 27 Végèce Mil. II. 7. 2 Ordinarii dicuntur qui in proelio, quia primi sunt, ordines ducunt.Quatre manuscrits conservent une autre leçon: Ordinarii dicuntur qui in proelio primos ordines ducunt qui ne modifie pas notablement le sens général de la définition et lie toujours les ordinaires à un commandement dans les premières lignes. C. Lang, dans ses deux éditions du texte, supposait que quia primi sunt était l’interpolation d’un copiste. Cependant, les travaux philologiques de M. Schanz, A. Andersson et P. de Jonge ont montré que, dans la majorité des cas, ce que C. Lang prenait pour des commentaires était des segments originaux. Cf. en dernier lieu, Reeve 1998: 193. 28 Ordo chez Végèce, cf. Blackman & Betts 1989: 518.

Végèce Mil. II. 15. 3, Sed ante signa et circa signa nec non etiam in prima acie dimicantes principes uocabantur, hoc est ordinarii ceterique principales. Pour le placement des officiers subalternes dans les lignes de bataille, cf. Speidel 2000: 475-482; Janniard 2004: 394-395; Wheeler 2004: 170 30 Végèce Mil. II. 8. 6, Sic decem centuriae cohortis primae a quinque ordinariis regebantur. La datation de l’antiqua ordinatio legionis a fait l’objet d’importantes controverses bien résumées par Giuffrida 1997: 65-67, qui se rallie à la datation proposée par H.M.D. Parker et E. Sander: 260-290 apr. J.-C. Il est désormais loisible d’élargir les bornes chronologiques vers l’aval et l’amont comme le montrent l’apparition précoce des grades de centenaire et de ducénaire et le maintien des triarii encore à l’époque tétrarchique. 31 C’était déjà la solution proposée par Grosse 1920: 116. La seule objection que l’on pourrait faire à cette hypothèse est que le triarius prior de la première cohorte équivaut à un simple centurion puisqu’il ne commande que cent hommes. On peut supposer un statut particulier du centurion triarius étant donné le rôle de réserve de sa centurie dans l’organisation légionnaire (Végèce Mil. II.16).

386

SYLVAIN JANNIARD: CENTURIONES ORDINARII ET DUCENARII DANS L’ARMÉE ROMAINE TARDIVE (IIIE -VIE S. APR. J.-C.) supérieurs en grade aux autres centurions, probablement en raison de leur fonction particulière, celle de mener au premier rang l’ensemble des cohortes au combat. La documentation épigraphique et papyrologique n’infirme pas la validité de cette interprétation pour le IIIe s. bien que cette dernière soit essentiellement fondée sur des témoignages d’époque tétrarchique. Il nous semble surtout que la véritable fonction des ordinarii soit à mettre en rapport avec l’apparition du grade de ducénaire.

attestées dans cette place forte de l’Euphrate.35 Tout se passe donc comme si dans le premier tiers du IIIe s. siècle apr. J.-C. apparaissaient deux nouveaux termes pour désigner une ou deux classes de centurions supérieurs aux autres, disposant pour la première de fonctions tactiques spécifiques et toutes les deux d’effectifs plus élevés que la normale. Or, un lien entre les ordinaires et les ducénaires nous semble pouvoir être établi par l’intermédiaire d’un groupe particulier de protectores.

Ducenarii et ducenarii protectores

Le terme de protector est apparu au début du IIIe siècle pour distinguer certains singulares de l’empereur, des préfets du prétoire et des gouverneurs. Gallien en fit un titre honorifique porté par tous les officiers au dessus du grade de centurion lorsqu’ils accompagnaient le prince en campagne.36 Nous trouvons ainsi la dignité de protecteur accolée aux fonctions de primipile et de primipilaire (e. g. AE 1954: 135, IGBulg III/2, 1570), de tribun du prétoire, (CIL III 3126), de préposé (CIL III 3228 = ILS 546) et enfin de préfet de légion (CIL III 3424 = ILS 545). A côté des simples protecteurs, qui en l’absence d’autres précisions doivent être des centurions, existe une catégorie particulière: les ducenarii protectores. Th. Mommsen pensait qu’il s’agissait de ducénaires équestres ayant reçu la dignité de protecteur tandis que O. Seeck à l’inverse y voyait des protecteurs militaires ornés de la ducena dignitas. C’est cette dernière hypothèse que parut confirmer en partie H.-G. Pflaum lorsqu’il montra qu’à partir du règne de Gallien, certains procurateurs équestres bénéficièrent d’une promotion dans la hiérarchie par l’octroi de la dignité de ducénaire détachée de l’exercice effectif d’une fonction de rang réellement ducénaire.37 Néanmoins, il nous semble que cette solution peut difficilement rendre compte de la spécificité des protecteurs ducénaires.

Sa plus ancienne attestation paraît être un graffite du Palais du duc de Doura Europos, construit au plus tard sous Elagabal et abandonné définitivement après la prise de la ville par les Sassanides vers 256, qui pourrait constituer une inscription honorifique en l’honneur de deux personnages et de leurs fonctions respectivement de ducénaire et de centenier.32 D’autres inscriptions de ducénaires peuvent être antérieures à l’époque tétrarchique dont l’épitaphe d’Aurelius Salama, vétéran probable de la cohors Ia miliaria Hemesenorum qui n’est plus attestée à Intercisa après la réorganisation dioclétienne.33 Végèce, toujours à propos de sa description de l’antiqua legio, donne la définition suivante du ducenarius: “De même, le primus hastatus dirigeait en seconde ligne deux centuries, c’est à dire deux cents hommes, on l’appelle maintenant ducénaire”.34 L’assimilation des deux grades semble motivée non par une identité fonctionnelle – le système manipulaire a disparu au IVe s. – mais plutôt par la correspondance numérique des effectifs placés sous les deux officiers. Il semble donc qu’à la fin du IVe s., il ait existé dans l’armée romaine des subdivisions tactiques de deux cents hommes confiés à des ducénaires. Il n’est pas impossible, si l’information de Végèce peut être appliquée aux ducénaires précédents, que de telles subdivisions aient été mises en place au cours du IIIe siècle, non seulement dans les cohortes miliaires comme la cohors Ia Hemesenorum mais peut-être aussi dans les légions, le graffite de Doura pouvant en effet avoir été laissé par l’une des nombreuses vexillations légionnaires

En effet, les procurateurs ornés de la ducena dignitas étaient de rang sexagénaire ou centenaire et pouvaient espérer progresser dans la carrière équestre.38 Or, les protecteurs appelés ducénaires sont des centurions comme semblent le montrer les épitaphes de Marcus Aurelius Processanus ainsi que l’inscription honorifique du corps expéditionnaire de Iulius Placidianus qui place les ducénaires immédiatement après les commandants de détachements.39 Comme nous l’avons vu, les officiers

32 Rostovtzeff 1952: 42-43, n° 952 (pièce n° 8): [[ …]]| et

(ublium) A(ium) Anton(inum)| et ide ducinarias (sic)| et centinarias (sic)| Iar[…]us. Une seconde attestation précoce – 258-259 apr. J.-C., si on admet à la suite de Fr. Dolbeau (REAug 29, 1983 : 30-40, 63, 65) l’authenticité du texte – se trouverait dans la Passio Sanctorum Montani et Lucii. XX. 3: Et cum centenarius diceret notarian sibi datam esse qua contineretur eum fingere. Les mss de la famille a portent ducenarius, ce qui ne minore en rien l’intérêt du texte bien que la lecture centenarius, donnée dans les mss les plus anciens (IXe-XIIe s.) de la famille b, doive lui être préférée. 33 AE 1971: 343 = RIU V, 1201, (Intercisa): D(is) M(anibus)| Aurel(io) Sala|mati uet(erano) ex u|cen(ario) ciui Aedes(itano)| qui uix(it) annis LXX| in his militauit| ann(is) XL. Aussi AE 1914: 88 = AE 1938: 100 (Arþar, Bulgarie, partie inférieure d’une stèle): …] et Aur(elii) Longi|ni u(ixit) a(nnis) VI et Aur(elii)| Bessi u(ixit) a(nnis) V Au|relia Mathae| et Aur(elius) Vrsus| ducenar(ius), b(ene) m(erentibus)| posuer(unt). 34 Végèce Mil. II. 8. 3, Item primus hastatus duas centurias, id est CC homines, ducebat in secunda acie, quem nunc ducenarium uocant.

35

Sur la garnison de Doura, cf. Gilliam dans Welles 1959: 24-26 et James 2004: 16-22, 24-25. L’idée de divisions de cent hommes dans l’armée tardive est défendue aussi par Speidel 2002: 136. 36 Cf. Speidel 1986 (CIL VI 3238 = ILS 2208 et aussi CIL XIII 7535a ; ILIug II, 831 ; IGBulg III/1, 884 = III/2, 1890; AE 1974: 648 ; AE 1979, 448); Christol 1977 contra, sans emporter la conviction, Ibeji 1991: 244-293. 37 Mommsen 1913: 425, Seeck 1905: 1753, Pflaum 1960-1982, II: 950951, Ibeji 1991: 279. 38 Cf. la liste établie par H.-G. Pflaum dans Pflaum 1960-1982, II: 950951, note 15. 39 CIL XI 837 = ILS 2778 = Franzoni 1987: 46 (inscription sur sarcophage de marbre, Modène, fin du IIIe s.): D(is) M(anibus)| M(arco) Aurelio| Processano, u(iro) e(gregio)| ex cent(urione) praet(oriano)| cohort(is) VI,| prot(ectori) ducenario| Bononi Metrodora| coiugi karissimo| cum quo uixit ann(os) XI| m(enses) XI dies XV b(ene) m(erenti); CIL XII 2228 = ILS 569 (Grenoble, plaque de calcaire,

387

THE LATE ROMAN ARMY IN THE EAST supérieurs servant à proximité des princes accolent le titre de protecteur à leur rang, qu’ils continuent d’indiquer en premier. En l’absence de toute mention en ce sens, nos ducénaires ne peuvent être assimilés aux tribuns, préfets ou préposés, comme le montrait déjà l’inscription de Grenoble. Contrairement aux procurateurs, enfin, nos ducénaires militaires ne mentionnent jamais leur titre d’egregius uir à l’exception significative de Processanus qui n’a probablement obtenu l’admission dans l’ordre équestre qu’après son service. Rappelons que jusqu’au milieu du IIIe s. pour les militaires sortis du rang, seul le primipilat ouvrait la voie à l’exercice de fonctions centenaires et que la possibilité effective d’occuper des postes ducénaires était réservée aux primipiles bis ou aux tribuns des unités de Rome. Même dans la seconde moitié du IIIe s., l’octroi de la ducena dignitas supposait encore la possession d’un rang supérieur à celui de centurion y compris du prétoire: Marcus Flauius Florianus n’obtint ainsi la dignité de ducénaire et le titre d’egregius qu’après son probable primipilat, quant à Aelius Aelianus, il dut attendre le rang centenaire et l’octroi d’un brevet honorifique d’ancien protecteur pour en bénéficier.40 Peu de simples protecteurs ont entamé une carrière équestre et aucun des protecteurs ducénaires connus n’a exercé ensuite de charges procuratoriennes effectives.41 Quant à la qualité générale des inscriptions qu’ils nous ont laissées, elle ne va pas non plus dans le sens de candidats aux ordres supérieurs de la société romaine.42

façon étymologique pour désigner un commandement sur deux cents hommes. Un faisceau de témoignages, parmi lesquels l’antiqua ordinatio legionis de Végèce, tendrait à placer en effet l’adoption par l’infanterie romaine d’un ordre décimal et de subdivisions de cent et de deux cents soldats dans le second tiers du IIIe s. apr. J.-C.43 Le titre connu pour la première fois par le graffite de Doura a donc été en quelque sorte officialisé lorsque l’octroi, en principe égalitaire, de la dignité de protecteur à tous les officiers subalternes et supérieurs a obligé dans les faits à distinguer entre les grades selon leur fonction. Lors de la refonte constantinienne de l’armée, le terme de ducénaire, attaché désormais au service immédiat du prince, a été réservé aux nouvelles unités. Ordinarii et ducenarii Or, il nous semble aussi que les centurions qui ont pris le titre de ducénaire lors de leur passage au comitatus princier ne peuvent être que les ordinarii.44 Nous avons vu en effet qu’il s’agissait de centurions plus élevés en grade que les autres et avec une fonction tactique particulière, celle d’être en tête des lignes de combat de l’infanterie. Surtout, comme nous le verrons, l’ensemble de la documentation pour les IVe-VIe s. montre que ducénaires et ordinaires sont organisés et fonctionnent de façon rigoureusement identique, les premiers dans les unités issues de la réorganisation tétrarchoconstantinienne, les seconds dans les troupes de création antérieure.

Il nous semble préférable d’avancer l’hypothèse que le titre de ducénaire accolé à la dignité de protecteur ait pu servir à distinguer un groupe particulier de centurions protecteurs, supérieurs en grade à leurs collègues simples centurions. L’origine du titre est à rechercher dans le sermo castrensis où il fut employé en premier lieu de

L’apparition du nom et de la fonction d’ordinarius/tus correspondrait alors à une modification survenue au IIIe s. dans l’ordre de bataille de l’infanterie romaine qui reposerait désormais, au détriment de l’organisation cohortale, sur un rôle accru des lignes, lui-même explicable par l’introduction et la coordination d’un armement différencié dans les unités.45 La création du rang d’ordinarius a ainsi permis la reconstitution d’un échelon tactique intermédiaire entre la centurie et la cohorte. Si l’on accepte l’assimilation entre ducenarii et ordinarii, ce commandement de deux cents hommes, en revitalisant la structure manipulaire puisque le terme d’ordinarii était utilisé comme synonyme de centuriones

premier semestre 269): Imp(eratori) Caesar[i]| M(arco) Aur(elio) Claudio| Pio Felici Inuicto| Aug(usto) Germanico| max(imo), p(ontifici) m(aximo), trib(uniciae) pote|statis II, co(n)s(uli), patri pa|triae, proc(onsuli), uexil|lationes adque| equites itemque| praepositi et duce|nar(ii) protect(ores) ten|dentes in Narb(onensi)| prou(incia), sub cura Iul(i)| Placidiani, u(iri) p(erfectissimi), prae|fect(i) uigil(um), deuoti| numini maiestatiq(ue) eius. 40 AE 1895: 38 = ILS 4872 = IAq 112 (Aquilée, IIIe s.): Deo Beleno| M(arcus) Flauius| Florianus, u(ir) e(gregius)| a ducena(riis)| quod| princeps leg(ionis)| uouit [primi|pil(us) sumpsit? Aelius Aelianus: AE 1907: 70 = 1915: 74 = ILS 9478, Pflaum 1960-1982, II: 948-952, n° 357. 41 Cf. Chastagnol 1988: 203-204. 42 CIL III 6439 = 3249 (autel, Sirmium): Deo| Sil|ano do|mest[i]|co Aur(eliu)s?| Hatena| pro(tector) duc(enarius); CIL V 5833 (Milan): D ||| M Senni[o Paterno?]| ducenario prot(ectori) […]| qui uixit annis [… m(ensibus)]| VII dies XXIIII, Veria VI[…]| coniugi karissimo et P[ate]|rnius Amicus et Patern[ius]| Pacatus fratres, patr[i]| pientissimo.; CIL XII 2576 (sarcophage, Annemasse): D(is) M(anibus)| et memori(a)e aetern(a)e Tibe|ri(a)e Maxsim(a)e,| Aurelius Roma|nus, protector duc|enarius co(n)iugi incomparabi|li; AE 1900: 42 = CIL III 14165, 1 = IGLS VI, 2844 (Baalbeck): D(is) M(anibus) | Aur(elio) Victr(i) duc(enario) prot(ectori),| uixit annis XXXX, uiuite| felices et nostris profundi|te manis et memores estis uos| nobiscum esse futuros, Aurel(ius)| Baia duc(enarius) prot(ector), fratri pien|tissimo, memoriam institu|it; AE 1964: 256 = ILIug. 1049 (Taurunum, Pannonie inférieure, IIIe s.): D(is) M(anibus)| perpetuae secu|ritati Aur(elia) Bittel|liana ex prouin(cia)| Ital(ia) uix(it) an(nis) XX, Flor|entino filio, Aur(elius)| Florus pr[o]t(ector) duc|[en(arius)] coniu[g]i …| [-

43

Janniard 2000: 68-71. Le Strategikon de Maurice, à plus de trois siècles de distance, considère aussi un tagma de deux cents hommes comme la plus petite subdivision tactique possible, cf. I.4.32 et II.20.1719. 44 Babut 1913: 247, 252 et 1914: 243, 246, 293 proposait une reconstruction assez proche. Selon lui, les ordinarii, assimilables aux primi ordines, auraient pris le nom de protecteurs ducénaires à la fin du IIIe s. et de ducénaires au Ve s. Si nous ne pouvons que souscrire au rapprochement entre ordinaire, protecteur ducénaire et ducénaire, en revanche la démonstration appelle de nombreuses réserves, en particulier en termes de chronologie. Par ailleurs, les prémisses en sont fausses: les ordinaires ne correspondent pas aux primi ordines et tous les centurions ne sont pas devenus protecteurs à la fin du IIIe s. contrairement à ce qu’il estimait. 45 Janniard 2000: 71-73 et Speidel 2002: 130-133, avec la bibliographie antérieure, sur l’importance retrouvée par les lignes en liaison avec l’apparition au début du IIIe s. de phalangarii, de lanciarii et d’archers dans les légions.

388

SYLVAIN JANNIARD: CENTURIONES ORDINARII ET DUCENARII DANS L’ARMÉE ROMAINE TARDIVE (IIIE -VIE S. APR. J.-C.) priores au moins en première ligne, ajoutait une articulation supplémentaire au dispositif tactique et assouplissait ainsi une infanterie en quête de solution face aux modes de combat développés et transmis par les guerriers des steppes.

d’ojptivwn triavre". Cette fonction est un hapax puisque les triarii au Haut-Empire sont commandés par des centurions appelés pili. Aurelius Gaius indique tous les postes qu’il occupa comme optio par ordre croissant d’importance. Celui d’optio triarius est donc le moins élevé de la hiérarchie. Cette situation correspond parfaitement à la hiérarchie de l’“ ancienne ” première cohorte, très certainement généralisable au neuf autres, telle qu’elle nous est fournie par Végèce. Le centurion triarius, au dernier rang, est le centurion le moins gradé de la cohorte. Après ce poste d’ojptivwn triavre", Aurelius Gaius fut wjptivwn wjrdina'to". Le centurion ordinatus occupe donc une position supérieure à celle du centurion triarius. L’inscription permet d’identifier alors la fonction d’ordinarius avec le service dans les principes ou les hastati. Plus précisément, elle confirme l’hypothèse que nous avons émise concernant le sens d’ordinarius/tus: Aurelius Gaius dut être l’adjoint d’un centurion prior des principes ou des hastati avant d’accéder au poste d’adjoint du princeps de la première cohorte. Les centurions de la première cohorte ou primi ordines continuent à se distinguer du reste des officiers subalternes.

Le nouveau grade expliquerait la réapparition des mentions de doubles centuries dans la documentation épigraphique au IIIe s. comme une dédicace à la centurie double du primipile sous les Sévères et la stèle posée par la schole des centuries doubles de la legio VIIIa Augusta sous Gallien.46 Le texte de Végèce devient aussi plus compréhensible grâce à l’assimilation entre ordinaires, centurions priores des premières lignes et ducénaires. Végèce ne connaît pas le système manipulaire qu’il n’expose jamais en raison d’une confusion entre manipulus et contubernium (II.13.7). En revanche l’une de ses sources sur l’antiqua legio devait mentionner la pratique de la double centurie dans la première cohorte légionnaire, ce que Végèce traduit en utilisant le seul équivalent qu’il connaisse à un commandement sur une double subdivision: le terme d’ordinarii.47 La présence d’ordinaires dans les unités auxiliaires, enfin, ne crée plus de difficulté puisque les mêmes réformes tactiques ont dû s’appliquer tant aux cohortes qu’aux légions dans un contexte d’effacement croissant des différences entre ces types d’unités au cours du IIIe s.

L’inscription d’Aurelius Gaius montre donc à la fois la survie des grands principes de l’organisation légionnaire traditionnelle jusqu’au début du IVe s. et corrobore, dans le même temps, la transformation de l’ancienne hiérarchie telle qu’elle est présentée par Végèce, transformation elle-même en rapport avec une modification de l’ordre de bataille entamée peut-être dès l’époque sévérienne où les principes deviennent la première ligne d’une cohorte désormais articulée selon des rangs à l’armement différencié. De surcroît, l’épitaphe atteste de la différence de statut entre les centurions ordinarii et les autres centurions, ce qui pourrait valider l’assimilation entre ordinarii et centurions priores des premières lignes. Néanmoins, l’époque tétrarchique est aussi la dernière à nous fournir des indications sur le rôle des lignes de bataille annonçant une modification de la fonction des ordinarii/ducenarii dans l’armée des IVe-VIe s.

Nous proposons donc la reconstruction suivante pour l’apparition des termes d’ordinarii et de ducenarii: le premier fut utilisé d’abord dans le sermo castrensis puis dans la nomenclature officielle dès l’époque sévérienne pour désigner les centurions priores des premières lignes en liaison avec des modifications tactiques donnant à l’organisation en ligne de bataille une importance nouvelle. L’introduction d’un système décimal dans l’infanterie romaine leur offrit l’appellation concurrente de ducénaire, bientôt officialisée par le service des ordinarii dans les corps expéditionnaires accompagnant le Prince en campagne. Les ordinarii tétrarchiques La Rangordnung de l’armée tétrarchique nous est particulièrement mieux connue depuis la publication de la carrière exclusivement légionnaire d’Aurelius Gaius. Le premier poste de principalis qu’il mentionne est celui

Ordinarii et byzantine

ducenarii

dans

l’infanterie

proto-

Si l’infanterie des IVe-VIe s. continue à agir selon le principe de lignes à l’armement différencié, rien dans notre documentation ne semble indiquer que les articulations tactiques internes des unités aient été organisées en fonction de cette exigence. Néanmoins, comme nous l’avons vu précédemment, le texte de Végèce confie aux ordinarii un rôle capital dans le commandement des lignes de bataille, ce qui rend compte de leur position parmi les premières lignes avec les autres principales. Les équivalences qu’il établit enfin entre les grades disparus de centurion et d’hastatus prior et respectivement les fonctions de centenarius et de ducenarius vont dans le sens du maintien de subdivisions placées sous la responsabilité d’officiers subalternes

46

AE 1978: 504 = AE 1990: 754 (base de statuette en bronze, début du IIIe s., Trèves), Genio (centuriarum duarum) p(rimi)p(ili),| Macal(ius) Auitus,| Vital(ius) Respectus,| ciues Suebi Nicr(etes)| e sens(u) his qui or|dines sequentur| d(ono) d(ederunt) m(erito), cf. Le Roux 1992. ILIug. 272 (Sirmium, stèle, début du règne seul de Gallien): D(is) M(anibus)| Gratius Artilleus,| Clodius Glamosus,| (centuriones priores?) leg(ionis) VIII Aug(ustae)|, bello Serdicensi| desideratis, scho|la (centurionum priorum?) leg(ionis) s(uper) s(criptae), colle|gis bene meren|tibus d(e) s(e) [p(osuit?], cf. Šašel [1992]. Voir aussi CIL XIII 6677a (règne de Maximien), 8150; RIU II, 390 (règne de Philippe). 47 La seule source théorique conservée mentionnant les centuries doubles de la première cohorte légionnaire, à l’exception de Végèce, est le Ps Hygin Mun. cas. 3, Cohors prima causa signorum et aquilae intra uiam sagulariam et, quoniam duplum numerum habet, duplam pedaturam accipiet.

389

THE LATE ROMAN ARMY IN THE EAST combattants, dont l’ordinarius/ ducenarius aurait représenté l’échelon supérieur. Un même principe d’organisation est à l’œuvre dans les tagmata théoriques de l’Empereur Maurice, deux siècles plus tard, puisque les deux ilarchai placés en première ligne sont en fait les hékatontarkai les plus élevés en grade.48

vexillatio Parthusagittariorum de Diospolis (nome Arsinoïte) le grade de ducénaire, fut admis à adorer la pourpre à Constantinople alors qu’il y accompagnait des ambassadeurs Blemmyes (336/338 apr. J.-C.).52 Les textes cités de Jérôme et de la NTh XXI montrent que les ducénaires se rangent ainsi au dessus des centenarii. Il en va probablement de même pour les ordinaires par rapport aux centurions alors qu’une épitaphe de Ladik en Phrygie orientale de la fin du IVe s. révèle que l’on pouvait être promu du grade de campidoctor à celui d’ordinarius.53 Or, les campidoctores occupent une place importante dans les unités d’infanterie tardive, en particulier ils semblent veiller au IVe s. à la cohésion des rangs et leur rôle n’a cessé de s’affirmer au point qu’ils sont considérés au livre XII du Strategikon comme les officiers subalternes les plus importants au combat, en liaison avec leur aptitude à analyser le terrain.54

En terme de rang, dans les unités traditionnelles de l’armée romaine pour lesquelles ils sont exclusivement attestés, les ordinarii viennent immédiatement après le chef de corps.49 Enumérant les dignitates secundum praesentes matriculas des légions, Végèce les place après les tribuns, ce qui est confirmé par un fragment de matricule militaire de la deuxième décennie du IVe s. où le centurio ordinarius est le premier gradé enregistré après le tribun. Cette chaîne hiérarchique est à nouveau respectée sur des briques estampillées datant des réfections valentiniennes d’Aquincum qui mentionnent les trois officiers responsables des travaux: le dux limitis Frigeridus, le tribun de la legio IIa Adiutrix (?) Valentinus et enfin Luppianus, ordinarius probablement de la même légion.50 Il en va de même pour les ducénaires dans les types d’unités issues des réformes tétrarchiques et constantiniennes d’après les témoignages de Jérôme pour la cavalerie de la fin du IVe s. et de la NTh XXI de 441 pour les scholes palatines.51 C’est aussi à partir du rang d’ordinaire ou de ducénaire que l’on pouvait accéder au corps d’état-major des protectores comme le montrent l’épitaphe de Florius Baudio, protecteur orné de la ducena, ancien ordinaire de la legio IIa Italica mort à Spolète lors de la campagne italienne de Constantin ou bien la carrière de Flavius Abinnaeus qui, après avoir servi trente-trois ans et atteint dans la

Les ordinaires et les ducénaires sont regroupés dans leur unité en un collège d’officiers, le koino\n tw=n prwteuo/twn que nous fait connaître un papyrus de l’arithmos d’Eléphantine daté de 578 apr. J.-C. La pratique en est toutefois bien antérieure et ce koinon doit être très vraisemblablement rapproché du terme collectif de principia qu’Ammien Marcellin applique aux principaux gradés des troupes de son temps en dessous du rang de tribun.55 Les ordinaires constituent, d’après le document d’Eléphantine, les premiers officiers de ce collège, supérieurs au reste des priores. Le koinon, en plus de son rôle tactique, exerce des fonctions administratives parmi lesquelles en particulier la tenue de la matricula de l’unité et ce dès l’époque tétrarchique comme le prouve le P. Oxy. XLI, 2953.56 A la tête du

48

Végèce Mil. II.8.3 et 8; II.13. 4-5; Maurice Strat. I.3.17-18; I.5.5-8; II.20.6-7; VII B.17.32-33, cf. Haldon 1999: 109. Pour la présence d’officiers subalternes et de sous-officiers dans les premières lignes, cf. n. 29. 49 Les ordinaires ne se rencontrent que dans les légions et les anciennes unités auxiliaires, cf. CIL V 942 = ILS 2670 = IAq 2779; CIL V 8275 = ILS 2408 = IAq 2736; MAMA I, 306; IGLS XVI 75; Lefevbre 1907: 70. 50 Végèce Mil. II.7.2; P. Mich. X, 592 = ChLA, V, 298; AE 1953: 8. 51 Jérôme Contre Jean de Jérusalem 19: Finge aliquem tribuniciae potestatis suo uitio regradatum per singula militiae equestris officia ad tironis uocabulum deuolutum. Numquid ex tribuno statim fit tiro? non, sed ante primicerius, deinde senator, ducenarius, centenarius, biarchus, circitor, eques, deinde tiro, “ Imagine-toi un quidam dégradé de son rang de tribun par sa faute, dévalant tous les échelons du service équestre jusqu’à celui de recrue, crois-tu que de tribun il devienne immédiatement recrue? Non, mais d’abord primicier, puis sénateur, ducénaire, centenier, biarche, circitor, cavalier enfin recrue ”. La chaîne de commandement biarche, centenier, ducénaire est aussi connue de Jean Lydus De mag. III.2. NTh. XXI (17 avril 441 apr. J.-C.), Imp(eratores) Theod(osius) et Valent(inianus) A(ugusti), Flegetio magistro officiorum. Pr. […] Ideoque suggestionem tui culminis adprobantes u(iris) s(pectabilibus) comitibus scholarum uerberandi regradandiue senatores ac ducenarios licentiam denegamus. 1. […] Domesticos etiam eorum gradibus lege praeterita designatis fieri, non senatores uel ducenarios centenariosue decernimus. […] Dat. XV Kal Mai., Constantinopoli, Cyri u(iro) c(larissimo) cons(ule). Passée l’époque tétrarchique, les ducénaires ne sont attestés que dans les nouveaux types d’unités: auxilia: CIL V 8759 = ILS 2797 = ILCV 499; AE 1977: 806; ILCV 500-501; vexillations: CIL III 14704; AE 1984: 825 = IK 31, 40; ILCV 498; equites: CIL V 8777 = ILCV 522; AE 1903: 291 = ILIug. 1310 = IMS III/2, 53; scholes: CIL VI 32 949 = ILCV 502; AE 1891: 104 = ILCV 497; cuneus: P. Oxy. LXIII, 438; P. Abinn. 42.

52

CIL XI 4787 = ILS 2777 (Spolète, 312): D(is) M(anibus)| Florio Baudioni| uiro ducenario| protectori ex ordinario leg(ionis) II Ital(icae)| Diuit(iensium), uix(it) an(nos) XL mil(itauit) an(nos) XXV, Val(erius)| Vario optio leg(ionis) II Italic(a)e Diuit(iensium),| parenti karissimo m(emoriam) f(aciendam) c(urauit). La datation est le fait de Ritterling, suivi par Scheithauer & Wesch-Klein 1990; P. Abinn. 1, l. 4-8, cf. Barnes 1985. 53 MAMA I, 168 = AE 1928: 159 (bloc inscrit): Fla(ouiva) Mariva, Seleuv|kissa, ajnevsthsa tw'| ajndriv mou Pauvlou| ajpo; kampidouktov|rwn, wjrdenarivou,| mnhvmh" cavrin kai;| a[n tis ejpicirivsi e[s[ta]i| pro\j th\n Triavdan. Deux documents de 493 (P. Lond. V, 1855 + P. Münch. 15) et de 530 (P. Lond. V, 1722) nous apprennent que les gradés de la ‘légion’ de Syène distinguaient les rangs, d’une part d’augustalis (ordinarius) et de campiductor et d’autre part d’augustalis (ordinarius) et de centurio. Treadgold 1995: 91, voit une différence de rang mais non de fonction entre centenarius et ducenarius: “so a ducenarius had the same function as a centenarius or centurion but held a slightly higher rank”. 54 Végèce Mil. III.6.23; Maurice Strat. XII B.8.20; 11; 16.4; 17.53, cf. Wheeler 2004: 168-169 et l’article de Ph. Rance dans ce volume. Ammien signale un tribun du nom de Marinus qui fut promu à ce poste, certainement après un passage dans le corps des protecteurs, depuis le rang de campidoctor (XV.3.10). 55 P. Monac. I, 2 = Wilcken Chrestom.: 470 = P. Münch. 2; Ammien Marcellin XV.5.16 (principiorum uertices); XXII.3.2 et 9; XXV.5.1 (legionum principiis et turmarum) et 8.16 (tribunis principiisque militum); XXVIII.6.17 (numerorum principiis) et aussi Histoire Auguste, Vie des trente tyrans X.7 (a principiis imperator est salutatus). 56 P. Monac. I, 2 = Wilcken Chrestom.: 470 = P. Münch. 2 (enregistrement d’une recrue); P. Oxy. XLI, 2953 (9.9 x 7 cm, 293305 apr. J.-C.): Matrix alae primae| Hiberorum Diocletianae|

390

SYLVAIN JANNIARD: CENTURIONES ORDINARII ET DUCENARII DANS L’ARMÉE ROMAINE TARDIVE (IIIE -VIE S. APR. J.-C.) à une légion.60 Si l’on tient compte qu’une unité de taille équivalente, l’arithmos d’Eléphantine, disposait en tout vers la même époque de huit ordinarii et que l’on comprend chez Végèce une infériorité hiérarchique des flauiales sur les augustales, il n’est pas impossible d’imaginer que les ordinarii se répartissaient pour moitié entre augustales, les quatre plus gradés dont le primicier?, et flaviales avec parmi ces derniers l’adiutor.61

koinon, l’on trouve l’ordinaire avec l’ancienneté la plus grande dans le poste, qui exerce la charge de primicier, sur le modèle connu aussi pour les protectores et les scholes palatines. Il s’agit dans l’unité du premier officier issu du rang et c’est à ce titre qu’il intervient par exemple dans des opérations de réfection d’installations de défense.57 Les ducénaires disposent exactement de la même organisation, mise en place dès l’époque tétrarchique. Une inscription de Mothana (Imthan, province d’Arabie), datée de 306, nous fait ainsi connaître un officier, peut-être d’origine gothique, Laitila, qui se définit comme doukhn(avrio") primavkhro".58 Comme l’a bien montré M. Speidel, il s’agit d’un ducénaire exerçant les fonctions de primicier, ce qui signifie à la lumière de nos textes qu’il est le ducénaire le plus élevé en grade dans le collège des priores de son unité. Les scholes palatines ont, elles aussi, pour principaux officiers des ducénaires dont les primiciers attestés dans notre documentation devaient représenter les doyens.59

Les ducénaires ont une organisation un peu différente. Augustales, flaviales et adiutor ne sont pas attestés dans leur collège en revanche, une partie d’entre eux prend le titre de senator.62 La Novelle de Théodose II réprimant les abus des comites des scholes palatines, en employant le terme au pluriel, interdit d’assimiler senator et adiutor, la dernière fonction étant confiée semble-t-il à un titulaire unique. Nous pouvons plutôt envisager l’hypothèse que les senatores, plus élevés en grade que les autres ducénaires, aient été les équivalents des augustales dans les collèges de priores des nouvelles unités de l’armée tardive.

Outre le primicier, le papyrus d’Eléphantine mentionne la présence d’un adiutor qui doit correspondre à une division fonctionnelle du collège, l’adiutor semblant assurer en effet la rédaction des actes émis par le koinon. D’autres divisions nous sont connues par Végèce qui distingue parmi les ordinarii entre deux groupes, les augustales et les flaviales (II.7.3). L’information est confirmée par un papyrus de Syène de 540, le testament d’un centurion de l’unité locale, où l’un des témoins se désigne comme Flavius Iôsêph, ord(inarios) a(u)goustalios (P. Münch. 8 + P. Lond. V, 1857). Les ordinarii augustales de la legio Va Macedonica étaient regroupés en une schole dans laquelle on pouvait être promu comme le montre aussi un second document papyrologique plus récent d’un siècle. Enfin, trois documents datés respectivement du 18 janvier, du 15 février et du 6 mars 594 permettent de dénombrer au moins quatre augustales dans l’unité de Syène, assimilée

Conclusion A partir du début du IIIe siècle apr. J.-C., certains centurions sont désignés par le terme ordinarius/tus. Dans un contexte d’évolution tactique, il est fort probable que leur nouveau nom reflète la modification de leur fonction, c’est-à-dire le commandement des lignes de bataille (ordines). Il pourrait s’agir en l’espèce d’un remploi de l’ordre manipulaire et le terme d’ordinaire aurait servi dans un premier temps d’équivalent informel au rang de centurio prior dans les lignes des principes et des hastati. Au même moment, apparaît en contexte militaire le titre de ducénaire, détaché de l’octroi de la ducena dignitas et qui doit être mis en rapport avec la modification des effectifs des subdivisions d’infanterie telle qu’elle nous est rapportée par l’antiqua ordinatio legionis de Végèce. Si l’on assimile les ordinarii sévériens à des centurions priores, c’est-à-dire des officiers responsables de deux centuries, il n’est pas

Maximianae Constantianae| Maximianae agent(is) in| castris Thmo[u], Val(erio)| Sarapioni [2] (centurioni) ord(inato). 57 SEG 8: 781 = SB 7425 (Syène, VIe s.): réfection d’une partie du mur d’enceinte par Fl. Konstantinos Erythrios Damianos comes, Theodosios, tribun, Isakios préposé et Pateirênês primicerios. Syria – Princeton III A: 136, n° 237 (Umm el-Jimal, dédicace du castellum tardif, 412-413 apr. J.-C.): Ejpi; Fl(a(ouivou) Pelagivou| ºAntipavtrou tou' lampr(otavtou)| kovm(ito") kai; douko;" ejktiv|sqh oJ kavstello",| spoudv= Ba/ssou| primik[hr(i/ou)……]miva"| e[to[u"] tz’ (th’?), [ijnd(iktiw'no") i]a’. 58 AE 1933: 185 = SEG 7: 1194 (Imtân, Arabie, 306-307 apr. J.-C.): Tou=to to\ mnh=ma| Laitila doukhn(a/rioj)| prima/khroj, ejtw'n| nz’, e[ti sa’. L’inscription a été trouvée à Mothana où la Notice des Dignités place des Equites scutarii Illyriciani, ND Or. XXXVII. 1. Cf. Speidel 1977: 716: “Laitila, who bears an apparently Gothic name, has the rank of ducenarius and the function of primicerius, both pointing to the elite units of the later Roman army”. Cf. aussi la carrière de Flavius Taurinos, attesté comme centenarius du cuneus equitum Maurorum scutariorum d’Hermopolis en 446 apr. J.-C. (BGU XII, 2141) et auquel BGU XII, 2146 de 457 se réfère en tant qu’ancien primicier. 59 Primiciers des scholes palatines: CJ XII.29.2; CIL VI 37276 = ILCV 484 = ILS 9213; CIL VI 31971 = ILCV 485b.

60 P. Ross. Georg. III, 10 (c. 400, Tieio): l’augustalis Pserakos, annonce sa récente promotion dans la schole des augustales de la legio Va Macedonica, A)possxo/lou a)goustali/ou noume/rou kuntanw=n. P. Münch 13-14 et P. Lond. V, 1733 de 594: Fl. Komes fils de Paamios, Fl. Ioannes fils de Victor, Fl. Victor fils de Psabet, Fl. Patermouthis fils de Hatres y sont signalés comme Augustales. Position supérieure des augustales dans les bureaux de la préfecture du prétoire, cf. Jean Lydus De mag. III.6.25; 9, 20. Le rang aurait été créé par Arcadius (Jean Lydus De mag. III.10.14-17). 61 Le chiffre de huit ordinarii, qui se déduit de P. Monac. I, 2 = Wilcken Chrestom.: 470 = P. Münch. 2, paraît très élevé si l’unité stationnée à Eléphantine est toujours, comme à l’époque de la rédaction de la Notice des Dignités, la cohors prima Felix Theodosiana. Une solution consisterait à faire du numerus d’Eléphantine une unité “ milliaire ”, assimilable par sa taille à une légion. 62 Senatores dans les unités tardives: CIL III 14188; CIL V 8752 = ILS 2802 = ILCV 460; CIL V 8760 = ILS 2804 = ILCV 493; CIL VI 32948; CIL VIII 17414; ILS 2796 = ILCV 494; ILS 8883; BGU I, 316. Babut 1913: 248, estimait que senatores étaient un terme générique pour désigner le primicier et les dix premiers officiers de chaque unité.

391

THE LATE ROMAN ARMY IN THE EAST difficile de comprendre comment le sermo castrensis n’eut aucune peine à les appeler aussi ducénaires.

Carcopino J. 1925 Le limes de Numidie et sa garde syrienne d’après des inscriptions récemment découvertes II. In Syria 6: 118-149. Chastagnol A. 1988 La fin de l’ordre équestre: réflexions sur la prosopographie des “derniers” chevaliers romains. In MEFR (Moyen Âge, Temps moderne) 100.1: 199-206. Christol M. 1977 La carrière de Traianus Mucianus et l’origine des protectores. In Chiron 7: 393-408. Daris S. 1964 Documenti per la storia dell’esercito romano in Egitto, Milan. Domaszewski A. von 1981 Die Rangordnung des römischen Heeres, Cologne3. Drew-Bear Th. 1981 Les voyages d’Aurelius Gaius, soldat de Dioclétien. In La géographie administrative et politique d’Alexandre à Mahomet. Actes du Colloque de Strasbourg, juin 1979, Leyde: 93-141. Fink R.O. 1971 Roman Military Records on Papyrus, Cleveland. Franzoni Cl. 1987 Habitus atque habitudo militis. Monumenti funerari di militari nella Cisalpina romana, Rome. Gilliam J.F. 1940 The Ordinarii and Ordinati of the Roman Army. In TAPhA 71: 127-148 = Roman Army Papers. (Amsterdam, 1986): 1-22. Gilliam J.F. 1958 The Appointment of Auxiliary Centurions (P. Mich. 164). In TAPhA 88: 155-168 = Roman Army Papers. (Amsterdam, 1986): 191-205. Giuffrida Manmana Cl. 1997 Flavio Vegezio Renato. Compendio delle istituzioni militari, Catane. Grosse R. 1920 Römische Militärgeschichte von Gallienus bis zum Beginn der byzantinischen Themenverfassung, Berlin. Haldon J. 1999 Warfare, State and Society in the Byzantine World, 565-1204, Londres. Herz P. 1999 Der centurio supernumerarius und die annona militaris. In Laverna 10: 165-184. Ibeji M. 1991 The Evolution of the Roman Army during the third century AD, Birmingham James S. 2004 The Excavations at Dura-Europos conducted by Yale University and the French Academy of Inscriptions and Letters 1928-1937. Final Report VII: The Arms and Armour and other Military Equipment, Londres. Janniard S. 2000 Tactiques, formations et armements de l’armée romaine tardive (IIIe -VIe siècle), mémoire de DEA inédit présenté à l’EHESS sous la direction de M. Jean-Michel Carrié. Paris, 200 pages. Janniard S. 2004 Armatus, scutatus et la catégorisation des troupes dans l’armée romaine tardive. In Y. Le Bohec & C. Wolff (ed.), L’armée romaine de Dioclétien à Valentinien Ier, Lyon: 389-395. Lefevbre G. 1907 Recueil des inscriptions grecqueschrétiennes d’Egypte, Le Caire. Le Roux P. 1992 L’armée romaine sous les Sévères. In ZPE 94: 261-268. Mazzarino S. 1971 La tradizione sulle guerre tra ShƗbuhr I e l’impero romano: ‘prospettiva’ e ‘deformazione storica’. In AAntHung 19: 59-82.

Cette assimilation des ordinarii et des ducénaires rendrait alors peut-être compte de l’onomastique de certains protectores du IIIe siècle. On peut avancer l’hypothèse que les inscriptions mentionnant des protecteurs ducénaires renverraient aux ordinarii revêtus de cette dignité, distingués des autres protecteurs et officialisant ainsi une expression d’abord idiomatique.63 Il paraît en tout cas nécessaire de faire remonter au moins au second tiers du IIIe s. l’apparition des grades tardifs de centenier et de ducénaire. Il se peut en effet que le grade de centenarius ait été introduit dans un premier temps de façon informelle pour distinguer le reste des centurions des centurions ordinarii/ducenarii. Par la suite, lors de la réorganisation constantinienne, les grades de centurio et d’ordinarius furent réservés aux unités d’ancien type tandis que ducenarius et centenarius allaient trouver leur place dans la hiérarchie des nouvelles formations. Dans les troupes tardives, les ordinaires et les ducénaires représentent les officiers subalternes les plus gradés du collège des priores. En plus de leur rôle tactique essentiel, ils assurent le fonctionnement administratif de l’unité, sur le modèle des anciens primi ordines des légions. La collégialité n’interdisait pas le maintien d’une hiérarchie interne stricte, fondée sur l’ancienneté et dont les éléments les plus méritants ne pouvaient s’abstraire que par l’accession au corps des protecteurs.64 Bibliographie Babut E.-Ch. 1913 Recherches sur la garde impériale et sur le corps d’officiers de l’armée romaine aux IVe et Ve siècles. In Revue Historique 114: 225-260. Babut E.-Ch. 1914 Recherches sur la garde impériale et sur le corps d’officiers de l’armée romaine aux IVe et Ve siècles (suite et fin). In Revue Historique 116: 225293. Barnes T.D. 1985 The career of Abinnaeus. In Phoenix 39: 368-374. Bartoloni R. 1996 Alae, cohortes peditatae e cohortes equitatae: loro stuttura e consistenza numerica. In Rivista storica dell’Antichità 26: 53-67. Blackman D.R. & Betts G.G. 1989 A Concordance to Vegetius. Hildesheim. Cagnat R. 1900 Centurio. Dizionario epigrafico di Antichità romane. II Rome: 192-202.

63

L’hypothèse paraît confirmée par la publication récente de l’épitaphe de Maximus, centenarius protector de la legio Ia Italica, SEG 42: 1264 (Laranda en Lycaonie): Ma/cimoj kenth|na/rij prwth/|ktwr musiatiko\j|

a)po\ lhgew=noj| prw/thj I)talikh=j| m(nh/mhj) x(a/rin).

64

Sur le principe de l’avancement à l’ancienneté, en l’occurrence dans la schole des protecteurs domestiques, cf. CTh VI.24.7. Sur l’attachement des troupes à ce mode de promotion, cf. Ammien Marcellin XX.5.7-9. CPL 267 (c. 400) pourrait constituer selon C. Zuckerman le dernier témoignage de protectores issus du rang.

392

SYLVAIN JANNIARD: CENTURIONES ORDINARII ET DUCENARII DANS L’ARMÉE ROMAINE TARDIVE (IIIE -VIE S. APR. J.-C.) Mommsen Th. 1913 Nomina et gradus centurionum. In Gesammelte Schriften VIII (Berlin): 360-384, part.: 376-377. Pflaum H.G. 1960-1982 Les carrières procuratoriennes équestres sous le Haut-Empire romain, Paris. Rea J.R. 1980 Ordinatus. In ZPE 38: 217-219. Reeve M.D. 1998 Notes on Vegetius. In PCPhS 44: 182218. Richardot Ph. 1995 Hiérarchie militaire et organisation légionnaire chez Végèce. In Y. Le Bohec (ed.), La hiérarchie (Rangordnung) de l’armée romaine sous le Haut-Empire, Paris: 405-428. Rostovtzeff M.I. 1952 The Excavations at DuraEuropos. Preliminary Report of the Ninth Season of Work, III, New Haven. Sanders H. 1931 Papyrus 1804 in the Michigan Collection. In G. Hadzsits (ed.), Classical Studies in Honor of John C. Rolfe, Philadelphie: 265-283. Šašel J. 1992 Bellum serdicense. In Opera selecta, Ljubljana: 360-378. Scheithauer A. & Wesch-Klein G. 1990 Von Köln-Deutz nach Rom? Zur Truppengeschichte der Legio II Italica Divitensium. In ZPE 81: 229-236. Seeck O. 1905 Ducenarius. Paulys Real-Encyclopädie der classischen Altertumwissenschaft 5.2, Stuttgart: Cols. 1752-1754. Southern P. & Dixon K. R. 1996 The Late Roman Army, Londres. Speidel M.P. 1977 The Roman Army in Arabia. ANRW II.8: 687-730 = Roman Army Studies. Amsterdam, 1984: 229-272. Speidel M.P. 1981 Princeps as a Title for ad hoc Commanders. In Britannia 12: 7-13 = Roman Army Studies. Amsterdam, 1984: 189-195. Speidel M.P. 1986 The Early protectores and their beneficiarius Lance. In Archälogisches Korrespondenzblatt 16: 451-454, pl. 68-69. Speidel M.P. 2000 Who Fought in the Front?. In G. Alföldy (ed.), Kaiser, Heer und Gesellschaft in der römischen Kaiserzeit, Stuttgart: 473-482. Speidel M.P. 2002 The Framework of an Imperial Legion. In R. Brewer (ed.), The Second Augustan Legion and the Roman Military Machine, Cardiff: 125-143. Stein E. 1933 Ordinarii et Campidoctores. In Byzantion 8: 379-387. Stoll O. 2001 Ordinatus Architectus. Römische Militärarchitekten und ihre Bedeutung für den Technologietransfer. In Römisches Heer und Gesellschaft. Gesammelte Beiträge 1991-1999, Stuttgart: 300-368. Treadgold W. 1995 Byzantium and its Army, 284-1081, Stanford. Welles C.B. 1959 The Excavations at Dura-Europos conducted by Yale University and the French Academy of Inscriptions and Letters. Final Report V, Part 1: The Parchments and Papyri, New Haven. Wheeler E. 2004 The Legion as Phalanx in the Late Empire II. In REMA 1: 147-175.

393

Campidoctores Vicarii vel Tribuni: The Senior Regimental Officers of the Late Roman Army and the Rise of the Campidoctor Philip Rance Berlin

Towards the end of the sixth century, the Strategicon ascribed to the Emperor Maurice outlines procedures for the organisation and deployment of East Roman infantry in battle. Maurice makes the following remarks on the size and command structure of combat units:

of antiquity, significant lacunae and obscurities persist in this field. Maurice’s stipulations regarding the tactical deployment of infantry in the 590s, therefore, provide a convenient retrospective point d’appui from which to review the evidence for the senior positions in the command structure of infantry regiments from the third to late sixth centuries, namely tribunus, vicarius and campidoctor; to examine how these titles relate to other elements of late Roman military titulature, specifically primicerius and ilarch (ÓȜÇȡȤȘȢ), and to identify and account for any changes that these ranks or posts underwent during this period.

‘…form up the units (tagmata) in keeping with whatever the strength of each regiment (arithmos) happens to be. Depending on the amount of available manpower, the regiment forms either one full-size or two more medium-sized units (İóIJİ IJÈȜİȚȠȞ ʌȠȚݲ Û ÐȡȚșȝËȢ, İóIJİ įÌȠ ȝİIJȡȚÍIJİȡĮ). And each of them is to have its own distinctive standard and commanding officer, in some cases the tribunus, a valiant and prudent man, capable of fighting hand-to-hand, but otherwise a vicarius or a campidoctor (ëʌȠȣ ȝÁȞ IJÄȞ IJȡȚȕȠ³ȞȠȞ … ëʌȠȣ įÁ ȕȚțÇȡȚȠȞ â țĮȝʌȚįȠÌțIJȠȡĮ)’.1

Infantry Regiments and Regimental Hierarchies in the late Roman Army Maurice’s discussion of infantry deployment contains the only unequivocal evidence for the subdivision of regiments into separate combat formations on the late Roman battlefield. He requires a full- or high-strength infantry regiment (numerus/ÐȡȚșȝËȢ) to be split into two smaller operational units (IJȐȝĮIJĮ), each with its own standard and commanding officer, one commanded by the tribune, the other by a vicarius or campidoctor. This procedure is not otherwise clearly reported in the sixth century, though an analogous sub-division of regiments for operational purposes can possibly be discerned in the fourth-century army.3 A similar practice may also be recorded on an inscription from Pylai dating to 531 (AE 1995.1427), which Zuckerman has emended to read ‘second bandum of the Constantiniaci’ (įİÌIJİȡȠȞ ȕÇȞįȠȞ ȀȦȞıIJĮȞIJȚȞȚĮț´Ȟ). If this interpretation is correct, the late Latin term bandum (ȕÇȞįȠȞ) refers here to a sub-unit or detachment a named infantry regiment (an auxilium palatinum) rather than, as usually, the regiment itself, though the emendation remains problematic since this usage of bandum is otherwise unknown and in all other sources bandum is used exclusively of cavalry.4 In

This passage is of intrinsic significance because it contains Maurice’s only reference to a vicarius and, moreover, furnishes unique information on the potential combat role of a campidoctor. Scholarly interest in this post, always scarce and for the most part restricted to the slender evidence of the Principate, routinely conceives the campidoctor as a low-grade ‘drill-sergeant’, replete with all the anachronistic connotations this modern parallel might arouse. While this picture may be true of the early third century, none of the later Roman evidence for campidoctores accords with this characterisation; indeed from the mid-fourth century onwards the campidoctor is consistently attested as one of the most important and distinguished figures in an infantry regiment, and increasingly so in the later fifth and sixth centuries. Two hundred years before the compilation of the Strategicon, Vegetius similarly linked campidoctores vicarii vel tribuni as the senior regimental field-officers, and the same association is made in a number of late antique sources.2 The detailed Rangordnung of the late Roman army has attracted much less attention than its first- and second-century counterpart, not least because the more meagre source material is not susceptible to the same level of detailed analysis. Frustratingly little is known with certainty concerning late Roman officergrades, several of which remain little more than titles, though it is a sobering thought that after more than a century of scholarship devoted to the legions of the Principate, one of the most intensively studied institutions

3 E.g. Amm. Marc. XX.4.2; XXXI.10.13, 11.2 reports 300- and 500strong drafts from individual infantry units of unspecified strength. 4 Zuckerman 1995; he convincingly dates this inscription to 531. The objections of Scharf 1997 are successfully rebutted by Zuckerman 1998, but the reconstruction nevertheless remains problematic. A ‘second bandum of the Constantiniaci’ requires the application of the word bandum to a sub-unit or section of a named regiment (as if ‘the second battalion of…’), whereas all other sources, including the Strategicon, treat bandum as a term for the ‘regiment’ itself, synonymous with numerus/ÐȡȚșȝËȢ or tagma; Zuckerman himself (p. 235) notes that this division ‘en compagnies… est fort remarquable’. This would also be a unique instance (where the troop-type is specified) of bandum applied to infantry; Maurice uses the term only of cavalry. For the Gothic origin of bandum see Kramer 1987; 1996: 115-116. Furthermore, since this

1 Strat. XII.B.8.16-20. Text in Dennis 1981: 426 with commentary by Rance forthcoming 2008. The imperial ascription is accepted here. 2 Veg. Epit. III.6.23.

395

THE LATE ROMAN ARMY IN THE EAST contrast to his specifications for the size of cavalry units, Maurice gives no indication of either the paper- or combat-strength of an infantry unit; he merely remarks that, unlike the artificial numerical schema delineated by ‘the Ancients’, the infantry arithmoi of his own day vary in size and that this factor determines deployment.5 The sub-division of large infantry regiments for battle was doubtless an ad hoc expedient dictated by specific military circumstances and the fluctuating combatstrengths of regiments in the field, but it also relates to the continued existence of different types or classes of infantry regiment, originating in different periods, of which the size, establishment and hierarchy differed, but whose distinctive features the Strategicon blurs and generalises under the neutral terms tagma and arithmos. The largely circumstantial evidence for infantry units in fourth- and fifth-century field armies, though far from satisfactory, suggests an establishment strength for auxilia palatina of ca. 700-800 and for legiones of ca. 1,000-1,200. Unfortunately, even fewer data survive in sixth-century sources, and potential variations between different classes of regiment compound the problems of calculation. Definitive solutions remain elusive, but, as a broad generalisation, operational infantry formations of ca. 500 are most commonly attested in the sixth century; though Maurice’s tactical provisions, partly corroborated by earlier practice, alert the modern historian to the possibility that, at least in some cases, these may have been sub-divisions or ad hoc detachments of larger regiments.6

order: tiro, eques or pedes, semissalis, circitor, biarchus, centenarius, ducenarius, senator, primicerius and tribunus. This new regimental hierarchy is listed by Jerome (writing ca. 386-7), in a sequence confirmed by fifth- and sixth-century legislation, and the individual grades are consistently attested in inscriptions, papyri, law codes and incidental references in historical and hagiographical literature from the early fourth to mid sixth centuries.8 The date and origin of this hierarchy are almost wholly obscure: references to most grades begin in the 320s and are soon universal, pointing to a reorganisation by Constantine I after his unification of the empire, but there is evidence that some of these grades were already established under the Tetrarchy, perhaps only in some regional armies, or at least the titles existed, even if they later acquired different functions, their final hierarchical sequence and empire-wide application under Constantine.9 While few of these grades are recorded beyond the reign of Justinian, this apparent terminus may merely reflect the significant reduction in relevant documentation and the decline in the use of Latin after this date, and need not necessarily imply change or discontinuity thereafter. All the evidence indicates that this new regimental hierarchy was restricted to the more recently-created units of the comitatenses, both infantry and cavalry, namely scholae, vexillationes, auxilia palatina and some cunei. This hierarchy and pay-scale were replicated in the clerical grades of various departments of state, which were constituted along

Before proceeding to a discussion of tribuni, vicarii and campidoctores, it is necessary to make some preliminary observations on the late Roman regimental hierarchy, the character and purpose of the Strategicon, and the terminology Maurice employs. By a process that has not been clearly established, a new sequence of regimental grades on a progressive pay-scale was introduced into all newly-raised infantry and cavalry units from the late third or early fourth centuries.7 It comprised in ascending

(1918), though Grosse omits the evidence of Maurice’s Strategicon, believing it, according to the prevailing arguments of RezsĘ (Rudolph) Vári, to be an eighth-century text by an otherwise unknown ‘Urbicius’ (not to be confused with the sixth-century amateur tactician of that name). The same misconception mars the majority of entries in RE. See additional remarks and evidence in Jones LRE 633-4, 674-5; Hoffmann 1969-70: I, 75-81; Haldon 1984: 108-111; 1993: esp. 46-7; 1999: 108111; Richardot 1993: esp. 422-424; Treadgold 1995: 87-96. 8 Jer. Contra Johannem Hierosolymitanum episcopum 19 (PL 23, 370); CJ I.27.2 §§19-36 (534); XII.20.4 (457-470). 9 See Brennan (1998) for important remarks on the regional variation that characterises military developments under the Tetrarchy. Jones LRE 1264 tentatively identifies Valerius Victorinus in AE 1922.72 = AE 1976.631 as the earliest attested biarchus, who died in 324, contra Southern and Dixon 1996: 62, erroneously citing ILS 2805 (which they furthermore misdate to 327; in fact 380-90s); their error is repeated in full by Richardot 2001: 62. However, Speidel 1995 argues that this soldier’s conventional adoption of the imperial nomen ‘Valerius’, distinct from ‘Flavius’ used under Constantine and his successors, assigns this biarchus to Licinius’ army in 324, and he is thus unrelated to any putative reforms credited to Constantine; the response by Woods 1997a is unsuccessful. For still earlier biarchi see e.g. P. Columb. 188.8, a biarchus in the equites promoti legionis II Traianae in the Thebaid in 320, a unit created by Diocletian and a document predating Constantine’s control of Egypt; cf. another possible instance in P. Aberd. 21 (III/IV): ’AȞȠȣȕÊȦȞȚ ȕȚ(…), unit unknown. If the adoptive nomen ‘Valerius’ is a firm indicator of service (and not just of initial enlistment) under emperors from Diocletian to Licinius (284-324), then other biarchi must also be assigned a pre-Constantinian date, e.g. AE 1946.42 with Hoffmann 1969-70: I, 250; 2, 104; see also C.C. Petolescu, ‘Valerius Livianus, Biercus’, ZPE 61 (1985) 238. Speidel 1984 uses the same dating criterion to place a ducenarius (IK-31.40) in a pre-Constantinian context; if correctly dated, the case is unique. It is possible that in some instances the grades in the new ‘Constantinian’ hierarchy may derive their names from older titles or duties (see e.g. circitor with the etymologising explanation of Veg. Epit. III.8.18).

inscription refers to the Constantiaci at Pylai on the coast of Bithynia, very probably the unit’s peace-time station, it is unlikely to relate to the battlefield procedure Maurice describes. If the emendation is correct, therefore, bandum here denotes a permanent or semi-permanent organisational sub-division not an ad hoc operational expedient. 5 For the optimum size of a cavalry bandum or tagma (3-400): Strat. I.4.8-9. 23-5. 32-5; II.20.13-19. Maurice refers to the artificial numerical scheme of ‘the Ancients’ (ȠÚ ÐȡȤĮ²ȠȚ), in this instance the late Hellenistic tactical tradition now represented by Ael. 8.3-9.10, 15.216.3; Arr. Tact. 9-10, 14. Maurice’s immediate source for this material is Aelian’s Tactica Theoria; see Rance (forthcoming 2008). 6 For the thorny question of late Roman unit sizes see Jones LRE 680-2; Elton 1996: 89-90; Coello 1996; Nicasie 1998: 67-74. For sixth-century units of 500 or multiples thereof cf. 1,000 infantry scaled Mt. Burgaon in 535 (Proc. BV II.12.17); two (probably infantry) units totalling 1,000 men were sent to secure Milan in 538 (Proc. BG II.12.26-7); a katalogos of 500 infantry was dispatched to Fiesole in 539 (Proc. BG II.23.2, cf. ibid. I.7.34); a 1,000-strong garrison of Naples in 543 (ibid. III.6.2); a 500-strong katalogos (?) of Tzani operated in Lazica ca. 554 (Agath. II.20.8); 500 ‘guards’ (ijÌȜĮțİȢ) defended Mesembria in 587 (Th. Sim. II.12.6). 7 For the late Roman regimental hierarchy the starting point remains Grosse 1920 esp. 107-138, which reprints with minor revisions idem

396

PHILIP RANCE: CAMPIDOCTORES VICARII VEL TRIBUNI military lines.10 Older units predating ca. 290-300 – namely legiones (regardless of whether later classed as palatinae, comitatenses or limitaneae), cohortes and alae – retained the traditional internal structures and noncommissioned officer-grades they had during the Principate, which they preserved until at least the early seventh century.11 Throughout late antiquity, therefore, there was variation in the establishment, internal structure and titulature of different classes of troops or units created in different periods. The only titles attested in both the old and new hierarchies are tribunus, vicarius, primicerius and campidoctor.

which appears to embody a distinct break between ‘Roman’ and ‘Byzantine’ institutions, but caution should be exercised on this complex issue. Proof of this point is elusive, but the limited evidence points to continuity of some of the older titles and distinctions into the middle Byzantine period.12 Furthermore, the disciplinary code included in Book I of the Strategicon refers to scholae (ıȤȠȜĮÊ), associations or collegia of soldiers of the same grade within a regiment. Such scholae for junior officergrades, which were formalised in the third century, are attested into the sixth century, but are wholly unknown for the new Hellenised hierarchy of the Byzantine army. The mention of scholae in a technical penal procedure within contemporary disciplinary regulations suggests that the Strategicon was written against the background of a high degree of continuity among late Roman officergrades.13

It is possible that at least some of the titles found in the late Roman regimental hierarchy which modern historians have tended to assume were ‘ranks’, equating to strictly-defined functions or unvarying commands of sub-units, were in fact pay-grades (e.g. semissalis) or broader distinctions denoting seniority, entitlements and privileges (centenarius, ducenarius, senator), which qualified the bearer of a particular grade to hold a range of appropriate posts or assume a variety of responsibilities. Certainly there are numerous other regimental posts, titles and specialist personnel attested between the early fourth and late sixth centuries and beyond (e.g. vicarius, campidoctor, decanus (įİțÇȡțȘȢ), mandator, mensor, draconarius, bandifer (ȕĮȞįȠijËȡȠȢ), bucinator), all of which must be located somewhere in this hierarchy and were presumably positions for which one or more of the grades listed above were eligible. With the exception of tribunus, the officer-grades recorded from Constantine to Justinian are missing from the Strategicon, or at least Maurice does not refer to them using these titles. Maurice appears to outline a simpler regimental hierarchy than that attested in previous centuries, and this may correspond to an actual process of simplification and standardization in East Roman military organisation, either as a result of changes introduced in Maurice’s reign or a longer-term evolution over the course of the sixth century which entailed the abolition of some older grades and titles. Scholars have tended to identify in the regimental structure and hierarchy described in the Strategicon a new ‘Maurikian’ army,

In this respect the purpose and idiom of Maurice’s Strategicon require some clarification. Maurice sought to write an elementary instructional manual devoted to training and tactics, principally of cavalry, not a comprehensive or scholarly survey of the Roman army. First, it seems to be the case that Maurice presents a regimental organisation that has been deliberately simplified and standardised for the purposes of his treatise, in short a schematic or idealised ‘model’ regiment which would be broadly applicable to all the different types or classes of unit that might be encountered in a contemporary field army regardless of their individual structural or hierarchical peculiarities. It is to be noted that Maurice limits discussion to the functionally necessary and includes only those officers and personnel who play an operational role in drills, manoeuvres and tactics; certainly all the administrative and logistical grades whose existence is independently

12 The grades senatores (ıİȞÇIJȠȡİȢ, ıȚȞÇIJȠȡİȢ) and ducenarii (as įȠȣțȚȞ(Ț)ÇIJȠȡİȢ = *ducin(i)atores) are attested in some units into the late ninth century, see Stein 1933: 386; Haldon 1984: 138, 288-289, 293-294; 1993: 47; 1999: 111. The term kentarchos (țÈȞIJĮȡȤȠȢ), used by some middle Byzantine authors to designate the commander of 100 men, is possibly a conflation of late Roman centenarius and the standard Greek equivalent ØțĮIJȠȞIJÇȡȤȠȢ; for refs. see Mihăescu 19789: III, 362. 13 Strat. I.8§18 (l. 23). In their current form the disciplinary regulations at Strat. I.6-8 were demonstrably drafted by the same author as the rest of the Strategicon; for argument see Rance (forthcoming 2008). The earlier history of under-officers’ scholae or collegia cries out for new research and much of the relevant bibliography is now outdated, see W. Liebenam, Zur Geschichte und Organisation des römischen Vereinswesens. Drei Untersuchungen (Leipzig 1890; repr. Aalen 1964) 297-309; M. Besnier, Les scholae des sous-officiers dans le camp romain de Lambèse. In MEFR 19, 1899: 199-258; M. Ginsburg, Roman Military Clubs and their Social Functions. In TAPA 71, 1940: 149-156; Le Bohec 1994: 192-193. For evidence of the continued functioning of scholae in the later Roman army: e.g. DRB 5.3; Veg. Epit. II.21 (a rather rigid and antiquarian conception, perhaps inspired by scholae of contemporary civilian bureaux); CPR V.13 + P. Rainer Cent. 165 (395401) as re-edited in Rea 1984: 80, I.7, 82, II.7; Ed. Anast. §8 (SEG IX.356 [AD 501]): ıȤȠȜÂ IJ´Ȟ țȚȡțȚIJËȡȦȞ; CJ XII.37.16§5 (491-518): unusquisque caput scholae, 43.3 (= CTh VII.13.19 [408]); P. Nessana 36.4 (?) (VI); see also PRG III.10.r.22, V.2 (IV/V): ÐʌËıȤȠȜȠȢ, a ‘member of a schola’. I propose to examine this topic in a later study.

10

The new hierarchy is recorded for the agentes in rebus (CJ XII.20.4 [457-470]) and sixth-century ducal officia (CJ I.27.2 §§19-36 [534]); some grades are also attested among the stratores in the department of the comes stabuli and in the administration of fabricae, for which see Jones LRE 578-579, 597-599, 634, 835. 11 Jones LRE 633-634, 662-663, 674-675; Keenan 1990. Contra Richardot 1993: 422-423, nn. 99-100, there is no firm evidence for the new grades in legiones, cohortes or alae. Richardot cites AE 1989, 641; 1990, 866 (early IV) for an apparent centenarius in Legio I Iovia Scythica, which, if correct, would be an unique instance of a legionary centenarius. It is not in fact specified that this man, mentioned as the father-in-law of the principal dedicant and grandfather of the deceased, is a member of the legion; another unit or even a grade in the ordo equester may be meant. The fact that the dedicant is himself a centurio ordinarius is a strong indication of continuity of the older hierarchy in this legio, in line with the unanimous testimony of the evidence. In addition, the mid fifth-century Passio Agaunensium martyrum §8 records a senator militum in a legio, but the historicity of this text cannot be relied upon, see below.

397

THE LATE ROMAN ARMY IN THE EAST attested in the sixth century are omitted.14 Moreover, some of the ostensible ‘ranks’ Maurice mentions (hekatontarch, dekarch, pentarch, tetrarch) may in fact be his own descriptive terms for operational roles or functions rather than the actual titles of officer-grades. Second, the information Maurice provides on structures and personnel is for the most part incidental and framed in a terminology that is partly of his own devising. His outline of the regimental and army hierarchy at the beginning of the Strategicon (I.3-5) is not intended as a description of the contemporary armed forces, as modern commentators have tended to assume (only a reader staggeringly ignorant of military matters would have found these chapters of informative value), but rather aims primarily to lay out the working vocabulary that Maurice will be using in his treatise, in effect a set of cover-all words for units, ranks and posts which will avoid ambiguity in subsequent chapters; hence, for example, such definitions as ‘a comes or tribunus is the leader of a tagma or arithmos or bandum’, or ‘moirarchs, who are also called duces or chiliarchs’.15 These standardised, all-inclusive terms amount to a convenient, all-purpose but partly artificial ‘tactica-speak’ which allowed Maurice to talk in general terms about an army that contained different types of unit with differing internal organization, hierarchies and personnel. An accurate evaluation of the evidence of the Strategicon requires careful appreciation of the cosmetic effects of genre and idiom. It is true, as Enßlin observed, that Maurice is the first author to employ a technical vocabulary that later became standard to Byzantine military literature, but it is very doubtful that the currency of several terms Maurice employs extended outside the milieu of Greek tactical writing or even beyond the pages of the Strategicon.16 Some of the titles and designations used in the Strategicon have all the appearance of genre terms (e.g. ilarch, merarch, tetrarch) and in some instances were possibly coined by Maurice himself (e.g. moirarch, moira); while others appear to be conventional Greek translations of Latin titles (hekatontarch [centurio], dekarch [decurio, decanus]). It is well to remember that Latin remained the language of nomenclature, instruction and command in the East Roman army, in effect the official Heeressprache, up to the 630s. Although the technical content and professional

jargon of the Strategicon tend to invest this Greeklanguage treatise with a degree of official authority, there is little reason to assume that such genre terms were widely or ever used in the ‘real-world’ East Roman army of the 590s, any more than the studied literary idiom conventionally employed by contemporary classicizing historians, and for modern scholars to adopt the vocabulary of Maurice’s treatise as if it were official Roman nomenclature is potentially misleading and erroneous.17 These considerations will be borne in mind in the following study. Tribunus or Comes In the Strategicon the commanding officer of a regiment (IJÇȖȝĮ, bandum or numerus/ÐȡȚșȝËȢ) was a tribunus or comes. This officer poses few problems and will be treated cursorily. The ancient Roman rank of tribunus, previously a designation properly employed only in certain classes of unit, became the generic title for all regimental commanders from the fifth century.18 By the mid fifth century a tribune might also be styled comes, formerly an honorary title or denomination borne by more senior field officers or bestowed on the highest state functionaries, but gradually extended to all tribunes as part of a broader process of proliferation and debasement of late Roman titles and dignities.19 Originally intended as an additional honorific, comes became an alternative designation for tribune, and later the preferred title; note, for example, the sixth-century funerary inscription from Odessus to a dekarch of the ‘bandum of comes Duda’ (ȕÇȞįȠȣ țËȝȚIJȠȢ ǻȠȣįȠȣ).20 When discussing cavalry Maurice always links both terms in the formula țËȝȘȢ òIJȠȚ IJȡȚȕȠ³ȞȠȢ, ‘comes or tribunus’, with precedence given to comes each time. He never employs comes when referring to infantry tribunes, though it is unclear whether this omission reflects a genuine distinction in titulature or merely varying authorial practice in different books of the Strategicon.21 Far more frequently Maurice refers to 17

E.g. Brown 1984: 60. For tribunes in the fourth to sixth centuries: Grosse 1920: 145-150; Jones LRE 640-43; Vogler 1993: 394-399, 402-403. 19 The process of proliferation of titles is obscure, see Jones LRE 527529, 641; Haldon 1984: 120. By the early fifth century tribunes of palatine regiments (scholae) regularly received the title of comes primi ordinis and the attendent dignity spectabilis: CTh. VI.13 (413) = CJ XII.11; Th. Nov. XXI (441) = CJ I.31.3, while even tribunes of ordinary regiments might rank as clarissimus (= ȜĮȝʌȡËIJĮIJȠȢ). Const. Porph. de Cer. I.91 (411.4) reports that the emperor Leo was comes and tribunus of the Mattiari (țËȝȘȢ æȞ țĮÃ IJȡȚȕȠ³ȞȠȢ IJ´Ȟ ȂĮIJIJȚĮȡÊȦȞ) before his accession in 457. Cf. Just. Nov. CXXX.1 (545): țĮÃ IJȡȚȕȠÌȞȦȞ țĮÃ țȠȝÉIJȦȞ. Durliat 1979 charts a similar process of extension and devaluation of titles in the sixth and seventh centuries in relation to magister militum (= ıIJȡĮIJȘȜÇIJȘȢ). For the later development of the rank of tribune in the exercitus Italiae see Brown 1984: 56-58. 20 Beševliev 1964: 60-61, no.89. 21 The twelfth and last book, devoted to infantry, was undoubtedly added later, though still at a very early date and certainly by Maurice himself; see Vári 1906: 71-72; Dennis 1981: 29-30; 1984: xvii-xviii. The overall characterization of Book XII as an afterthought or postscript tacked on to a completed work has been exaggerated, however, and the text of the Strategicon as it has come down to us is the unitary product of a general revision of a twelve-book treatise. It has been asserted that Maurice’s main treatment of infantry at XII.B is a revision of an earlier, 18

14 Maurice’s terminology will be further examined in P. Rance (forthcoming 2008). For regimental administrative staff see Grosse 1920: 127-137, 191-195; Jones LRE 626-627, 665, 672-674; Haldon 1993: 46-47; Ravegnani 1998: 35-36. Maurice does not mention the actuarius, the chief administrative officer of a unit, nor his assistant optiones, nor clerks, usually styled scrinarii or chartularii. This silence led Aussaresses 1909: 48 to doubt the existence of administrative officers. For this administrative hierarchy in the sixth/early seventh century see optio: Miracula S. Anastasii Persae 13 (ed. B. Flusin 1:1423) (describing AD 629-34); P. Ital. 17.28 (early VII); 18-19.53, 68 (early VII); 20.76 (ca. 600); Greg. Dialogi IV.35 (PL 77, 378C); optioprinceps: P. Nessana 19.6 (548); 36.13 (VI); scriniarius: P. Nessana 19.1? (548); actuarius: P. Lond. 1722.57 (530); 1733.75 (594); adiutor: P. Münch. 2.23 (578); 8.50 (540); 13.77 (594); 14.104 (594); P. Lond. 1733.81 (594). 15 Strat. I.3.16, 4.13. 16 W. Enßlin in RE XIV.2 (1930), Mauricius8, col. 2394

398

PHILIP RANCE: CAMPIDOCTORES VICARII VEL TRIBUNI commanding officers of both infantry and cavalry using the general descriptive phrase Û IJȠ³ IJÇȖȝĮIJȠȢ ðȡȤȦȞ, ‘unit commander’, a neutral, non-technical expression similar to the vocabulary favoured by sixth-century classicising historians.22 By the middle Byzantine period the superior status implicit in comes meant that it supplanted the rather technical tribunus altogether.23 In the Strategicon the qualities and duties of a tribune are diverse. He is assumed to be literate and is in receipt of written instructions from his superiors. He reads aloud disciplinary regulations to his subordinates, over whom he exercises summary justice in matters of misconduct and negligence. He has overall responsibility for the training and billeting of his men, and supervises their receipt of requisite rations, mounts and equipment. He personally deploys and leads his unit in both drill and combat, whether it operates in isolation or within a larger formation.24

seventh centuries, though a degree of misconception exists in modern scholarship concerning its character. Contrary to some recent secondary literature, vicarius was not a permanent rank in the regimental hierarchy, but rather a temporary title assumed by one of the senior noncommissioned officers while deputising for an absent tribune, in effect an ‘acting-tribune’; by definition when the tribune was present there was no vicarius.25 In expedient circumstances a vicarius might be temporarily appointed from another unit.26 The formulaic coupling of tribunus vel vicarius (= İóIJİ IJȡȚȕȠ³ȞȠȢ İóIJİ ȕȚțÇȡȚȠȢ) employed in a variety of late Roman texts, including Justinianic legislation, signifies the current ‘officer-incharge’ or ‘officer commanding’, regardless of his rank.27 Jones suggests that increased absenteeism among tribunes may be responsible for the prominence of vicarii in sixthcentury sources, and certainly Justinianic legislation makes provision for an absentee tribune, though given the likelihood of higher levels of literacy among senior noncommissioned officers it might simply be the case that vicarii and former vicarii were called upon to witness documents more frequently than other lower-ranking grades.28 The typical length and frequency of a vicarius’ term of office cannot be determined. The sixth-century dossier of the numerus or legio stationed at Syene contains no reference to a serving vicarius, but between 530 and 594 bears witness to six individuals who style themselves ÐʌÄ ȕȚțĮȡȚ´Ȟ (= ex vicariis), ‘former’ or ‘sometime vicarius’. That in each instance their actual current rank is not specified implies that the distinction of having held this post, however briefly, was of more honorific significance and conferred greater seniority than any non-commissioned grade these men might attain within the regimental hierarchy.29 These were of course

Vicarius, Primicerius and ‘Ilarch’ The highest position open to non-commissioned officers was that of vicarius. Maurice refers to a vicarius (ȕȚțÇȡȚȠȢ) in one instance and only in relation to infantry, though an equivalent position undoubtedly existed in cavalry units (see below for the ilarch). This restricted focus further illustrates Maurice’s practice of omitting titles, posts and personnel which are not essential to his discussion of deployment and tactics; the vicarius is mentioned only because of his potential role commanding a combat formation. The office of vicarius dates to the fourth century and is well attested in the sixth and early perhaps ‘Justinianic’ treatise: see originally Dain 1943: 39 (in reference to XII.B.21 only); repeated in posthumous Dain† 1967: 345; idem† 1968: 130, 132, 134-135; rehearsed by Mihăescu 1967: 404; and recently restated by Dennis 1981: 28-30; 1984: xviii. Space prevents full discussion, but I do wish to stress that the evidence adduced for a Justinianic antecedent to XII.B does not stand up to scrutiny. The character of a self-contained monograph which Dain and Dennis observed in XII.B in fact derives from Maurice’s use of Aelian’s Tactica Theoria (c. AD 106-113) as his literary and conceptual model when compiling this section. Detailed argument will follow in Rance (forthcoming 2008). 22 For cavalry tribunes: Strat. I.3.16-18: ȀËȝȘȢ įÁ ÑıIJȚȞ òIJȠȚ IJȡȚȕȠ³ȞȠȢ Û IJȠ³ IJÇȖȝĮIJȠȢ â ÐȡȚșȝȠ³ â ȕÇȞįȠȣ ÙȖȠÌȝİȞȠȢ; cf. ibid. 4.9-10, 6§2; II.20.5-6. Infantry tribunes: Strat. XII.B.8.19, 22.24. 30, 24.2. 30. Sixthcentury classicising historians never employ tribunus or comes to denote unit commanders, but rather a variety of generic, classicallyprecedented expressions, usually ðȡȤȦȞ or ÙȖİȝÍȞ (see examples assembled by Müller 1912: 103; Grosse 1920: 145-146), but also ȜȠȤĮȖËȢ (e.g. Agath. I.19.3; III.21.6; IV.13.4; V.15.1; Th. Sim. I.15.2; II.2.5, 4.8 [though IJĮȟÊĮȡȤȘȢ at I.14.5], 10.9 [IJĮȟÊĮȡȤȘȢ at II.12.4]); IJÇȖȝĮıȚȞ ÑijİıIJȘțËIJİȢ (Agath. IV.16.5); ʌȡËIJĮȖȠȢ (Th. Sim. VI.8.3). The term IJĮȟÊĮȡȤȠȢ, -ȘȢ is usually reserved for superior field officers or ‘brigade’ commanders (Agath. I.9.4, 11.3, 19.3; II.6.4, 20.7; III.21.6, 22.4, 26.3; IV.13.2, 15.1, 18.1, V.1.3, 2.1, 15.1; Evag. V.9; VI.5; Th. Sim. II.3.1, 4.1, 12.7; V.11.4; VI.8.12; VII.3.6. 8, 12.7). 23 In the passages of the Strategicon reprised in Leo’s Tactica (IV.6, 12) only țËȝȘȢ is given. 24 Written commands and instructions addressed to tribunes: I.6.5; VII.A.4.6; XII.B.pr.6-7, B.24.30-31, and assumed at VII.B.17.41; billeting, rations, mounts and equipment: I.2.3-6. 86-91; VII.A.9; justice: I.7-8; VII.A.4; IX.3.117-21; XII.B.10; deployment and command in combat and drill: I.5; III.1-5, 11; VII.A.2, B.17; XII.B.11, 17.

25 On vicarii generally see Grosse 1920: 149-150; Jones LRE 642-643, 675, contra Haldon 1984: 108-109; Richardot 1993: 419; 2001: 58-59; Treadgold 1995: 88, 95, who all misinterpret vicarius as a permanent rank. Sixth- and seventh-century vicarii: Mal. 256.20-22 (AD 363); Theoph. 335.14-22 (AD 630); P. Masp. 67002(2).12 (567); 67009.r.2022 (VI); 67057(2).6 (566/7); PSI 1366.4; P. Oxy. 1883.2, 11 (504); P. Lond. 1722.53 (530); 1724.83 (578-82); 1727.69, 73 (583-4); 1733.79 (594); P. Münch. 8.47 (ca. 540); 9.111 (585); 10.26 (585); P. Nessana 134.3 (VI). 26 P. Masp. 67002(2).12 (567); possibly 67009.r.20-22 (VI). 27 Veg. Epit. III.4.3: tribunorum vel vicariorum nec non etiam principiorum, ibid. 7: tribunis vicariis principiisque; 6.23: vicarii vel tribuni; 17.1: vicariis comitibus tribunisque vacantibus; cf. SHA Trig. Tyr. 10.4: vicarius tribuni. Official documentation: CJ I.42.2 (512): Û IJȡȚȕȠ³ȞȠȢ â Û IJÂȞ IJÇȟȚȞ ĮÕIJȠ³ ʌȜȘȡ´Ȟ ȕȚțÇȡȚȠȢ; III.13.5: tribuni vel vicarii; XII.37.19.pr.§4: ʌĮȡÇ IJİ IJ´Ȟ IJȡȚȕȠÌȞȦȞ țĮÃ IJ´Ȟ ȕȚțĮȡÊȦȞ țĮÃ IJ´Ȟ įȠȝİıIJÊțȦȞ, ibid. §4: İóIJİ IJȡȚȕȠ³ȞȠȢ İóIJİ ȕȚțÇȡȚȠȢ, 42.1, 51.9. Cf. Const. Porph. de Cer. 426.13-14 (ex Peter the Patrician): Û ’IȠȣıIJ²ȞȠȢ IJȠ²Ȣ ıIJȡĮIJȚ IJĮȚȢ țĮÃ IJȡȚȕȠÌȞȠȚȢ țĮÃ ȕȚțĮȡÊȠȚȢ ÐʌĮȞIJ±ıĮȚ (referring to comitatenses in 518). 28 Just. Nov. CXVII.11 (542): òIJȠȚ IJÄȞ IJȡȚȕȠ³ȞȠȞ, İó Ȗİ ʌÇȡİıIJȚȞ (seu tribunum, si tamen adest), ‘or the tribune, if in fact he is present’; with Jones LRE 675. 29 The same ex vicariis witnessed both P. Lond. 1722 (530) and P. Münch. 8 (ca. 540); one of the two ex vicariis who witnessed P. Lond. 1727 (583-4) also witnessed P. Münch. 9 (585); cf. also the witnesses to P. Münch. 10 (585); P. Lond. 1724 (578-82); 1733 (594). In one instance (P. Lond. 1724) the individual is styled ‘soldier of the legion of Syene and ex vicariis’: ıIJȡ(ĮIJȚÍIJȘȢ) ȜİȖ[(İ´ȞȠȢ)] Ȉ[ȣ]É[Ȟ]Ș[Ȣ] țĮÃ ÐpÄ ȕȚțĮȡ(Ț´Ȟ), a similar formula to Mal. 256.20-22: ıIJȡĮIJȚÍIJȘȢ æȞ țĮÃ

399

THE LATE ROMAN ARMY IN THE EAST ‘deputy tribunes’ serving in garrisons rather than the ad hoc combat expedient Maurice describes.

bureaux to denote a broad range of officers and officials functioning as personal assistant, chief-secretary or chamberlain, whose rank varied according to that of their superior.34 The important role that primicerii played in the day-to-day administrative procedures of the regiment invested them with considerable distinction. The primicerii in the scholae ranked as clarissimi, equivalent in dignity to tribuni in other regiments, and some primicerii could expect promotion to the rank of tribune.35 One sixth-century primicerius is also styled authenta (ĮÕșÈȞIJȘȢ), presumably a mark of his primacy in the regimental hierarchy, though the currency and official status of this designation are unclear.36 The term primicerius itself appears to have been used imprecisely or inaccurately in some unofficial sources, including hagiographical literature, as a generic designation for any senior regimental officer.37

The non-commissioned officer most likely to assume the post of vicarius was the primicerius, the senior noncommissioned officer whose name came first on the regimental muster-roll or matrix.30 This grade is attested from the fourth to early seventh centuries in both infantry and cavalry regiments, in the new post-Constantinian hierarchy and in older-style units; indeed primicerius is the only non-commissioned officer-grade to appear in both classes of unit.31 A primicerius combined a number of functions. He was the regimental ‘lieutenantcommander’, who, unless otherwise engaged or indisposed, ordinarily deputised for an absent tribune in the guise of vicarius.32 It also appears to have been customary for the primicerius to serve as the tribune’s domesticus, a post equating to adjutant or regimental chief of staff, though primicerius and domesticus should not be considered synonymous; primicerius was a noncommissioned grade, domesticus a staff-post he ordinarily held.33 The tribune might choose to assign this job to another senior non-commissioned officer of the regiment, but in some classes of unit custom or legislation reserved this position to the primicerius. Outside the sphere of regimental hierarchy, the term domesticus was more widely used in military and civilian

Maurice does not mention the grade of primicerius by name, but it undoubtedly existed in his day and well into the seventh century.38 As previously noted, Maurice refers to a vicarius only in the context of his potential battlefield role, when a high-strength infantry regiment is divided into two combat formations, each with its own standard. When discussing cavalry units, however, Maurice explains that the second-in-command or deputy to the tribune is the so-called ilarch (ÓȜÇȡȤȘȢ), whom he also defines as ‘the senior hekatontarch’ (Û ʌȡ´IJȠȢ IJ´Ȟ ØțĮIJȠȞIJĮȡȤ´Ȟ). The few details Maurice provides on the function of the ilarch typically relate to his operational role, which is indistinguishable from a vicarius in an infantry unit. The ilarch similarly commands half the regiment if it is ever divided for operational purposes,

ȕȚțÇȡȚȠȢ IJȠ³ ÚįÊȠȣ ÐȡȚșȝȠ³ IJ´Ȟ ȆȡȚȝȠĮȡȝİȞȚĮț´Ȟ. For this dossier see Keenan 1990. 30 The regimental musters were variously called matrix, matriculae, IJÀ ȕȚȕȜÊĮ or țĮIJÇȜȠȖȠȢ. For sixth-century evidence see e.g. Ed. Anast. §§1, 3 (SEG IX.356 [A.D. 501]); P. Ryl. 609 (505); CJ I.31.1, 28.5; XII.33.5§1a, 42.2-3; J. Lyd. de Mag. I.47; III.2, 66; Proc. BV II.16.3; Anec. 24.2-4; Agath. V.15.2; Evag. II.1; Th. Sim. VIII.12.12; Theoph. 300.4-6; P. Münch. 2.8 (578). Maurice calls this document IJÄ įȚĮȖȡÇijȠȞ (Strat. I.2.71); Dennis 1984: 14 omits this phrase. LJS cite IJÄ įȚĮȖȡÇijȠȞ, known only from P. Oxy 127.2 (VI), as a dubious reading, but Strat. I.2. 71 now confirms the contemporary sense of a register of names or property; contra Aussaresses 1909: 66; Haldon 1984: 110; 1993: 46 n.7, who understand an administrative official called a įȚĮȖȡÇijȠȢ (unattested, correctly įȚĮȖȡĮijİÌȢ). 31 For the various usages of the title primicerius see W. Enßlin, ‘Primicerius’, RE VIII.614-24, though some of the evidence he cites for primicerius as a military grade is open to other interpretations; also R. Guilland, ‘Le Primicier (Û ʌȡȚȝȚțÉȡȚȠȢ), le Grand Primicier (Û ȝÈȖĮȢ ʌȡȚȝȚțÉȡȚȠȢ)’, REB 14 (1956) 122-57. For regimental primicerii: Grosse (1920) 120-24, 195; Jones LRE 634, 674-5, 835; Hoffmann 1969-70: I, 79-80, 280; their references should be supplemented by P. Münch. 2.16 (578); P. Nessana 35.2 (VI), and possibly P. Oxy 1513 (IV). 32 Jer. contra Joh. Hierosol. 19 lists a primicerius as second-incommand to a tribune in a cavalry regiment (omitting vicarius because a post, not a grade); cf. Suda Ȉ.2286 (ed. Adler IV.195): ʌȡȚȝȚțÉȡȚȠȢ: Û ʌȡ´IJȠȢ IJ±Ȣ IJÇȟİȦȢ IJȣȤȠÌıȘȢ. In P. Münch. 2.16 (578) a primicerius heads a list of eight ordinarii (or centurions) of the arithmos at Elephantine; cf. similarly P. Nessana 35.2 (VI). The mid fifth-century Passio Agaunensium martyrum §8 records a primicerius commanding a legio (or more probably a legionary vexillation), though see below for difficulties with this source. 33 For domesticus: Grosse 1920: 120-124. Richardot 1993: 419; 2001: 58-59 erroneously equates vicarius and domesticus. For the regimental domesticus in the sixth and seventh centuries: Daci (P. Ital. 18-19.29, 64 [Rome early VII]); Armenii (P. Ital. 23.3 [ca. 700]); Felices Laeti (P. Ital. 24.21 [mid VII]); unnamed: P. Masp. 67040.5 (VI). Later instances of domesticus in the exercitus Italiae may indicate a more generalised usage applicable to a wider range of junior officers, see Brown 1984: 59-60.

34 Th. Nov. XXI.1 (441) = CJ XII.29.1.1 prohibits a tribune of a schola palatina from appointing the three lower grades (senatores vel ducenarios centenariosve) as his domesticus. For domestici in other contexts see Jones, LRE 598, 602-3. 35 CJ XII.29.2.pr. (474) with Haldon 1984: 120, 129-130, 134-135. In the domestici promotion from primicerius to tribune occurred as a matter of course: CTh VI.24.11 (432) = CJ XII.17.2. Amm. Marc. XVIII.3.5 reports a man promoted tribunus ‘from the rank of primicerius’ (ex primicerio) in the protectores, see PLRE I: 935, Valentinus³. 36 CIL VI 32970 (Rome VI): primicerius et aut(h)enta numeri felicum Theodosiac(orum). A later instance is recorded in Codex traditionum Ecclesiae Ravennatis, ed. J.B. Bernhart (Munich 1810) 47 (Rimini VIII?): auctenta numero Ariminensium. 37 The Passio SS. Sergii et Bacchi (AB 14 [1895] 373-95), written in the second half of the fifth century, but following a late fourth-century model, identifies Sergius as ‘primicerius of the schola gentilium’ (ʌȡȚȝȚțÉȡȚȠȢ Üʌ±ȡȤİȞ IJ±Ȣ ıȤȠȜ±Ȣ IJ´Ȟ ȖİȞIJȚȜÊȦȞ) and Bacchus as secundocerius (ıİțȠȣȞįȠțÈȡȚȠȢ) at §§1-2, though the latter grade is otherwise recorded only later and only in the notarii and domestici praesentales (CJ XII.23.7.7; CJ IV.59.1 [473]; XII.7.tit. [474]), 17.4-5). At §3, however, both brothers are ‘the primicerii of our schola’ (ȠÚ IJ±Ȣ ÙȝİIJÈȡĮȢ ıȤȠȜ±Ȣ ʌȡȚȝȚțÉȡȚȠȚ); see Hoffmann 1969-70: I, 282. For date and historical context see Woods 1997b; Key Fowden 1999: 26-28. 38 For the rank of primicerius in the sixth and early seventh centuries: Primi Theodosiani (CIL XI 1693 = ILS 2806 [Florence 547]); Felices Theodosiaci (P. Ital. 16.54 [Ravenna ca. 600]; CIL VI 32970 [Rome VI]); numerus Veronensium (P. Ital. 22.13, 21, 30, 39, 60 [Ravenna 639]); numerus Argentensium (P. Ital. 22.61 [639]); numerus Ravennatis (P. Ital. 23.7 [Ravenna ca. 700]); unnamed units: P. Münch. 2.16 (Elephantine 578); CIL V 1614 (Grado ca. 580); P. Nessana 35.2 (VI).

400

PHILIP RANCE: CAMPIDOCTORES VICARII VEL TRIBUNI while the tribune retains charge of the remaining half.39 Furthermore, Maurice stipulates that on campaign there should be two standard-bearers per regiment, one assigned to the tribune, the other to the ilarch, though in combat only one flag is to be flown. The ostensible purpose of this second or duplicate standard is to frustrate the attempts of enemy spies to calculate the number of units in the Roman army, but it presumably also served a practical function whenever part of the regiment was detached for independent operations under the ilarch, just as the infantry formation commanded in battle by a vicarius must have ‘its own distinctive standard’ (ȕÇȞįȠȞ ÓįȚțÄȞ èțĮıIJȠȞ ĮÕIJ´Ȟ).40 As is often the case, Maurice’s chosen terminology here is not straightforward. The word ÓȜÇȡȤȘȢ is a genre term commonly used in Greek military literature to denote a wide variety of ranks or commands, usually generic or theoretical, but sometimes equating to an actual Latin title, such as decurio, though this is clearly not the case in this instance.41 No officer bearing this Greek title is recorded in the late Roman period, however, or indeed outside the pages of the Strategicon, and the word appears to be an element of Maurice’s simplified ‘tactica-speak’ rather than a real title in use in the contemporary army. Maurice’s generic ÓȜÇȡȤȘȢ of a cavalry tagma is most plausibly interpreted as a primicerius, the grade attested as second-in-command in cavalry regiments from the fourth to sixth centuries. It is possible that the varied denominations and functions that might, in different contexts, be applied to this noncommissioned officer-grade in its various capacities (primicerius, vicarius or domesticus) prompted Maurice to employ this conveniently neutral cover-all designation.42

appearance in the later second century this designation applied to an individual charged with overall responsibility for regimental training and drill on the campus or exercise-ground.44 The origins of this and other more junior posts concerned with training remain obscure. Between the late third century BC and early first century AD the sources furnish little evidence for significant change in the arrangements for military training, which was for the most part delegated to tribunes and centurions. The establishment of the army on a permanent footing and the localisation of units seems to have encouraged a growth in the number and specificity of ‘training-officers’, though the evidence remains slight and several posts are no more than epigraphically-attested titles whose holders’ functions are a matter of scholarly supposition. Furthermore, throughout the army of the Principate the titulature of training-instructors appears to have remained flexible and varied between units of different classes, and there is no clear evidence for a consistent hierarchy or fixed complement of trainers in operation throughout the army.45 The earliest record of a campidoctor dates to the reign of Commodus.46 The emergence of this designation may merely reflect a change of terminology rather than the creation of a new position, whereby the chief ‘training-officer’ came to be denominated campidoctor, whereas formally all personnel with responsibilities in this sphere were generically styled exercitator or bore similarly imprecise labels. If this interpretation is correct, then this titular change occurred only in certain corps. Thus from the reign of Septimius Severus until the second half of the fourth century the title campidoctor itself is firmly documented only in the Praetorian Guard.47 A single exception of a campidoctor in equites singulares, the bodyguard of a provincial governor, in 182, itself the earliest occurence of this title, points to a wider currency, though this irregular instance of a campidoctor in a mounted unit is uncharacteristic of the evidence as a whole (see below). It is probable that functional counterparts of the campidoctor were found in other classes of unit under different titles or guises, such as magister campi or praepositus campi,48 while in other

Campidoctor The campidoctor was an important and well-attested post in a late Roman infantry regiment.43 From its first 39

The ilarch is defined at Strat. I.3.17-18: įİȣIJİȡİÌȦȞ įÁ IJ© țËȝȘIJȚ òIJȠȚ IJȡȚȕȠÌȞz, deputy or second-in-command to a comes or tribunus. In the diagrams at III.1-5 (cf. III.1.9) the ilarch has the same symbol as a hekatontarch, being represented by the letter rho with its numerical function of 100 (men). For division of the unit for operational purposes: IX.3.64-6. 40 Strat. II.20; VII.B.17.32-4, cf. XII.B.8.18-20. 41 For ÓȜÇȡȤȘȢ, ÓȜÇȡȤȠȢ or İÓȜÇȡȤȘȢ see e.g. Polyb. VI.25.1-2; App. Hisp. 43 (= decurio of a turma); cf. Polyb. VI.35.8: ʌȡ´IJȠȢ ÓȜÇȡȤȘȢ (= decurio princeps); Plut. Tim. 31; Arr. Anab. II.7.3 (= generic commander of a cavalry squadron); Asclep. 9; Ael. 23 (= commander of a theoretical troop of elephants); Ael. 19.10-20.2, Arr. Tact. 18.2; cf. Ascelp. 7.2-3 (= commander of a theoretical 64-strong troop of cavalry); Syrianus (= Anon. Byz.), de Re Strat. 15.87, 24.50, 28.39, 29.29. 36 (= the file-leader of a squad of infantry). 42 Haldon 1984: 109 with n.84; 1993: 46; 1999: 109 also identifies the ilarch with the role of vicarius, but is incorrect in the inference that an ilarch (only found in cavalry units) could also be a campidoctor (unique to infantry). He also, wrongly I believe, equates ilarch with the lowerranking grades of ducenarius and senator, rather than primicerius. 43 For campidoctor see Beurlier 1884; Grosse 1920: 126-127; Stein 1933; Domaszewski 1967²: 26; Horsmann 1991: 90-92; Richardot 1993: 420; Wheeler 2004: 168-169. Most older scholarship has recognised that campidoctor is the correct form, and campiductor an orthographic variant; Beurlier attempted to discern an actual distinction of sense and chronology; his arguments are without merit, though see below n.67.

44

For the design, use and legal status of the campus see Davies 1968; 1974; Horsmann 1991: 71-82. The evidence is assembled by Domaszewski 1967: 26-27, 45-46, 4849, 52-53, 59, 61, 77; supplemented and assessed by Speidel 1965: 4145, 55-60; 1978: 28-30; Horsmann 1991: 53-56, 82-102; Le Bohec 1994: 51, 112-113. 46 CIL II 4083 = ILS 2416. 47 CIL V 6886; VI 533 = ILS 2088; CIL VI 2658, 2697, 32536d.1.27, 36847; Ephemera Epigraphica 4:331, no.896.1.II. 48 The post of magister campi is attested in cohors equitata: P. Dur. 83.7 = Fink (1971) 190, no. 48.8: mag(ister) campi; and probably equites legionis: CIL VIII 2562: mag(ister) k(ampi), though see reservations of Speidel 1978: 30-31 n.159 concerning this unit’s identity. It is uncertain whether the post of magister campi is identical to the magister equitum legionis in CIL V 8278 = ILS 2333 (a centurio supernumerarius), and the mag(ister) coh(ortis) in CIL III 10307 = ILS 2540. A praepositus campi in an ala is one interpretation of CIL III 5938 = ILS 2525 (231): p(rae)p(ositus) k(ampi); though k(astrorum) and k(ampidoctor) (by analogy with CIL II 4083 = ILS 2416) have been suggested; see remarks of Speidel 1978: 29, 78-79. On these posts 45

401

THE LATE ROMAN ARMY IN THE EAST armidoctores.55 The titles hastiliarius, apparently a specialist spear-trainer, and armatura, who taught an advanced weapons-drill, probably merely differentiated specialised functions carried out by different doctores.56 An optio campi was possibly also a doctor (armorum) assigned as deputy to a campidoctor or magister campi, though the evidence is uncertain.57 The single instance of a campidoctor sagittariorum, reconstructed from the abbreviation CDS, is doubtful; certainly the application of the term campidoctor to such a specialist training function is inconsistent with the other evidence which points to a supervisory role.58

types of corps, especially élite cavalry units, exercitator continued to be the favoured term for the regimental weapons-trainer and riding-master.49 From the beginning, all these designations referred to a post or function rather than a grade in a hierarchy, and the actual rank of the holders of these posts varied according to the type of unit or the nature of the assignment. Most men recorded as exercitatores, for example, were centurions, but some were evocati.50 The evidence for the title campidoctor, such as it is, suggests that the holders of this post might differ in rank according to context. There are only two explicit items of testimony. First, the earliest record of a campidoctor, an altar dedication from Tarraco (182), refers to a centurion of Legio VII currently seconded to the provincial governor’s equites singulares, in which he simultaneously combined the roles of unit commander and training-officer: praep(ositus) simul et camp(idoctor). This campidoctor, therefore, was a senior centurion, but, as previously observed, this first instance of the term, the only campidoctor recorded in a cavalry unit, is probably atypical.51 The only other witness to the rank of a campidoctor is a funerary inscription to Aurelius Eliaseir, campidoctor of cohors IX praetoria, who is specified as being ‘in the centuria of Marcus’; it follows that Aurelius Eliaseir himself must have ranked below his centurion.52

In the later Roman army, the title campidoctor is attested in both the old-style and post-Constantinian regimental hierarchies, though it is not certain that the roles and seniority of campidoctores in these different classes of unit were identical.59 From the mid fourth century onwards all the evidence points to campidoctor being a post exclusive to infantry units.60 For Vegetius and Maurice the primary role of campidoctores remains that of a regimental drill-master. In the chapters on infantry drill in the Strategicon it is the campidoctor, not the 533 = ILS 2088 records a former coh(ortis) doctor… nunc campidoctor. For doctores armorum: Veg. Epit. I.13.3, 14.2, 15.2; cf. III.pr.7. 55 Armidoctor: CIL VI 5945; AE 1952.153 = 1953.97 (Aquileia): ex evocato et armidoctor leg(ionis) XV. 56 The evidence for hastiliarius is collected by Domaszewski 1967: 5253 (erroneously hastilarius); Speidel 1965: 43-45, 55; 1978: 30 n.159; Horsmann 1991: 91-92, 99, 151-153. Cf. Veg. Epit. I.14.2 for doctores conducting javelin practice. The armatura appears to have combined advanced weapons-drill and tactical manoeuvres, and was taught by specialists also called armaturae. By the later fourth century armatura was a largely festive or ludic remnant of a formally universal training regime attested epigraphically from the Flavian period, though possibly originating in the standardisation of military training under Augustus; for detailed discussion see Horsmann 1991: 92-102; with Rance 2000: 248-254 for the later period. 57 The single instance is P. Hamb. 39.II.(5).8 = Fink 1971: 287, no.76: ȚʌʌİȣȢ ȠʌIJȚȠȞ țĮȝʌȚ İȚȜȘȢ. The unit is ala veterana Gallica (AD 179); this post was probably a deputy to the magister campi or praepositus campi, cf. CIL III 5938 = ILS 2525. Davies (1968) 77 n.53 is wrong to suggest the synonymity of magister campi and optio campi. See also P. Hamb. 39.XX.(59).24 = Fink (1971) 301, no.76: țȚIJĮIJȠȡ țĮȝʌȚ; possibly signifying exercitator, though Davies 1968: 77 n.53 offers other interpretations. 58 CIL VI 31150 (AD 142), frg. C, z.3: c(ampi)d(octor) s(agittariorum). This interpretation is doubted by Speidel 1965: 45; Horsmann 1991: 84 n.148, 91 n.198; Speidel 1994: nr.14. For a doctor sagittariorum cf. CIL VI 3595 (Rome); AE 1971.209 (Madrid), though the context may be non-military, perhaps gladiatorial; see Horsmann 1991: 84 n.151, 91 n.199. Certainly doctores gave archery instruction, cf. Veg. Epit. I.15.2. 59 New-style hierarchy: CIL V 8773 = ILS 2803: Batavi seniores, an auxilium palatinum. Old-style hierarchy: Passio Agaunensium martyrum §8: Thebaei, a legio. P. Lond. 113.5(a).4-5 (498): Transtigritani (cf. ND Or. VII.22 = 58), a legio pseudocomitatensis. P. Münch. 15.17 (593): the legio at Syene (cf. ND Or. XXXI.35 milites miliarenses), a legio limitanea (although routinely styled ÐȡȚșȝÄȢ ȈȣÉȞȘȢ the continued legionary status of ȜİȖİÆȞ ȈȣÉȞȘȢ is confirmed by e.g. P. Lond. 1724.83 [578-82]; P. Münch. 4-5 [581]), 9 [585]). Peter the Patrician apud Const. Porph. de Cer. I.92 (423.7-9D), 93 (429.4A): Lanciarii (in 491 and 518), a legio palatina, on which see below. Unidentified unit: J. Chrys. ep. de exilio suo apud Palladius, Dial. Vitae S. Joannis Chrysostomi (ed. P.R. Coleman-Norton [Cambridge 1928]) 13.28 (= PG 52: 53-54); (ps.-)Phaedrus, fabulae IV.13; CIL VIII 4354 (578-582, ‘Ain el Kasr, Numidia). 60 Treadgold 1995: 95 has campidoctores in hypothetical establishments for infantry and cavalry regiments.

The campidoctor was a regimental drill-master responsible for a unit’s weaponry-training and tactical drill, and whose acquisition of technical expertise was expressed by the titular element doctor.53 While the campidoctor exercised overall supervision of the campus, the actual weapons-instruction was carried out by his subordinates. Again, the evidence is ambiguous and supplies a number of titles, but the most likely explanation is that these weapons-instructors were generically styled doctores or doctores armorum, who were principales on double pay, and who could rise to the post of campidoctor.54 They were probably identical to generally see Domaszewski 1967: 47-49, 53, 59, 98; Horsmann 1991: 84, 87-90. 49 The evidence for exercitatores is collected and discussed by Speidel 1965: 41-43, 59; Horsmann 1991: 86-90. 50 For the ranks holding the post of exercitator see Speidel 1965: 41-43; 1978: 28-29. 51 CIL II 4083 = ILS 2416 with Beurlier 1884: 300 (erroneously citing II 4003); Alföldy 1975: 21, no. 38 (Tf. VII.1); Speidel 1978: 28, 70-72. See above n. 48 for another possible but uncertain instance of p(rae)p(ositus) k(ampidoctor) in an ala. I can see no justification for the reconstruction campidoctor equitum praetorianorum from []R eq(uitum) praetor(ianorum?) in ILBulg 66 = AE 1957.296 (Oescus). 52 CIL VI 2697 (Rome): D(is) M(anibus) Aur(elius) Eliaseir coh(ortis) camppidoctor coh(orte) VIIII praet(oria) (centuria) Marci natio(ne) Pan(nonius) qui vix(it) an(nos) LX e(t) militavit in legione ann(os) X et in praet(orio) ann(os) XXV. To my knowledge, the only discussion is Beurlier 1884: 301. For the practice of distinguishing centuries by the name of their centurions see Speidel 1990. 53 Campidoctores as drill-masters: Veg. Epit. I.13.1; II.7.6 (campigenii = campidoctores), 23.3; III.26.37; SHA Sev. Alex. 53.9; Schol. in Juv. Sat. V.155; VI.261; CGL 2:96.56; 3.353.14: campidoctor: ÛʌȜȠįȚįÇțIJȘȢ. 54 Doctor: CIL III 10516 (= 3566); V 6886; VI 3595; IX 952. A doct(or) a(lae) has been read in CIL VI 3239a, but this interpretation is uncertain, see Horsmann 1991: 84 n.150; Speidel 1994: nr. 641. CIL VI

402

PHILIP RANCE: CAMPIDOCTORES VICARII VEL TRIBUNI tribune, who calls out the commands and who puts the regiment through its paces in a succession of manoeuvres.61 Maurice remarks that during drill or the advance into combat the campidoctor went ahead of his unit (probably on horseback) to inspect the ground, and two mounted campidoctores also accompanied the divisional commander out in front of the battle-line until the engagement began, when they took they places in line beside the divisional standard.62 The title and role of the campidoctor may have persisted in Byzantine infantry units, though certainty on this point is elusive.63 The explicit information Maurice provides about the duties of the campidoctor, and in particular his operating out in front of his unit, partly explains why both Ammianus and Vegetius associated campidoctores with the classical term antesignani, conceived by both authors as troops who fight in front of the standards or ‘front-rankers’, though their usage of the word antesignanus is classicising and non-technical.64

the sources over this period: before the late fourth century all the evidence is epigraphic; thereafter it is entirely literary, though the information supplied by Vegetius’ Epitoma purports to relate earlier practices. It is important to appreciate, however, that none of the later Roman evidence supports the characterisation of campidoctores as low-grade non-commissioned officers. On the contrary, from the fourth to the sixth centuries, where the evidence provides an explicit indication of the status of a campidoctor, he is one of the most important noncommissioned officers in an infantry regiment, and increasingly so towards the end of this period. Indeed, there are strong indications that probably by the fourth century, and certainly by the sixth, the campidoctor ranked third in the regimental hierarchy after the tribune and primicerius, and that he had acquired sufficient seniority to take command of independent operations or detachments. As previously observed, Maurice describes a procedure for sub-dividing high-strength infantry regiments into two smaller units, with the result that some combat formations will be commanded by a tribune, ‘but otherwise a vicarius or a campidoctor’ (ëʌȠȣ ȝÁȞ IJÄȞ IJȡȚȕȠ³ȞȠȞ … ëʌȠȣ įÁ ȕȚțÇȡȚȠȞ â țĮȝʌȚįȠÌțIJȠȡĮ), that is to say the primicerius or the campidoctor. Other sources imply that this was already the case by the fourth century, but the evidence is not clear-cut. Ammianus records campidoctores honoured with statues in full armour at Edessa for their role ‘as front-rankers in brave deeds’ (ut fortium factorum antesignanis) in a large-scale sortie outside Amida in 359. As is often the case, the historian’s predilection for classicising vocabulary and artful avoidance of termini technici renders his usage in this passage uncertain. Elsewhere, Ammianus appears to employ the word campidoctor straightforwardly as the title of a specific grade or position, in reference to a campidoctor promoted to the rank of tribunus vacans.66 In his description of the fighting around Amida, however, the apparent singling out of several regimental ‘drillmasters’ for such exceptional honours seems remarkable, even given what Maurice says about the role played by a campidoctor in guiding his unit prior to combat. It is possible that here Ammianus uses campidoctores imprecisely as a generic descriptive expression for ‘field officers’ in general, just as, by analogy, he elsewhere refers to the ‘officers’ of vanguard units using the vague and non-technical ductores. Certainly Ammianus’ association of the campidoctores in 359 with antesignani is a product of his conscious archaicising and reflects Livian rather than contemporary vocabulary.67

Most modern studies of this subject have restricted themselves to evidence relating to the army of the Principate and uniformly conclude that the campidoctor was a minor figure, who in consequence receives a cursory treatment in standard Anglophone works of reference or is even omitted altogether. The few brief studies of the position and function of a campidoctor present him as a humble ‘drill sergeant’.65 Discussion of campidoctores is complicated by the changing nature of 61 Strat. XII.B.16.3-4: ëIJİ șÈȜİȚ țȚȞ±ıĮȚ, ıȘȝĮÊȞİȚ… Û țĮȝʌȚįȠÌțIJȦȡ; he is apparently the subject of the clause ʌĮȡĮȖȖÈȜİȚ…, ‘he orders…’, which introduces each of the following manoeuvres and the pertinent word of command (ll. 11, 18, 22, 33, 41, 60, 62, 78, 80, 83, 88, 93, 99, 104, 109, 113). In the corresponding treatment of cavalry drill at III.5 Maurice does not clearly specify who is giving the commands, though presumably the tribune (via his mandator). 62 Strat. XII.B.11.4-7. 14-23 (implicitly mounted). This passage is repeated in de Militari Scientia 4, ed. Müller 1880: 118.30 (omitting campidoctores because the author transposes the context to cavalry deployment), 17.52-55. For mounted infantry officers cf. Proc. BG I.28.25. 63 The campidoctor in Byzantine armies is reported only by Leo, Tactica VII.47-48 (ed. R. Vári [Budapest 1917-22] = PG 107: 749 [VII.54-5]); XIV.67 (PG 107: 869). These passages are near-verbatim rehearsals of Maurice’s Strategicon XII.B.11.14-20, though Leo deems it necessary to gloss țĮȝʌȚįȠÌțIJȦȡ for the benefit of his readers with the imprecise òȖȠȣȞ Û ÛįȘȖËȢ, ‘that is to say the guide’. In the absence of independent evidence for the existence of a campidoctor (though see below n.82 for the țĮȝʌȚįȠȣțIJÍȡȚȠȞ, apparently the campidoctor’s flag) it is unclear to what extent Leo reports contemporary military realities; see remarks of Rance 2004: 313-321 on Leo and infantry tactics. 64 Amm. Marc. XIX.6.12; Veg. Epit. II.7.6 (where campigenii = campidoctores), with remarks of Wheeler 2004: 168-169. Ammianus’ equation of campidoctores and antesignani is a non-technical usage, probably intended as a linguistic echo of earlier Latin models, esp. Livy; see below n. 67. Vegetius uses the hapax campigenii as a synonym for campidoctores (the form campigenii, as now Reeve ed. [Oxford 2005] 43, supersedes campigeni of earlier editions); Milner 1996: 38 nn. 5-6 rightly suspects the equation campigenii - campidoctores as naïve etymologising typical of Vegetius’ method. 65 The post is omitted altogether from e.g. Watson 1969; Webster 1985³; Campbell 1994. Of the standard surveys of the Roman army only Le Bohec 1994: 51, 112-113 discusses the campidoctor, defined as ‘a lower grade officer’ (p. 112). This characterisation is shared, with much less excuse, by some Byzantinists, see e.g. Treadgold 1995: 88, ‘a drill sergeant’.

66 Amm. Marc. XV.3.10: Marinus tribunus ex campidoctore eo tempore vacans; see PLRE I: 560, Marinus². For tribunus vacans in Ammianus see also XVIII.2.2 with remarks of Vogler 1993: 397. 67 Amm. Marc. XIX.6.12; cf. XVI.12.20 for ductores. See remarks of de Jonge 1982: 129; Vogler 1993: 399-400; Wheeler 2004: 168-169. J.C. Rolfe (ed.), Ammianus Marcellinus ([LCL] Harvard, Mass. 1935) 1:500 chose to print the orthographic variant campiductoribus (ms. V) at XIX.6.12, but campidoctore at xv.3.10. Editors have traditionally preferred the reading campidoctoribus (see ed. W. Seyfarth [Leipzig 1978] 1:166 apparatus), but in the linguistic context suggested here the variation –duct over –doct is nevertheless apt, and conceivably authorial and intentional. For Ammianus’ usage of the classical antesignani in the

403

THE LATE ROMAN ARMY IN THE EAST Vegetius’ Epitoma exhibits a similar terminological ambiguity. In the majority of cases Vegetius employs campidoctor as the actual title of a contemporary post with distinct responsibilities for training, though typically Vegetius only refers to this role while he bemoans declining in standards.68 In one instance he appears, as Maurice, to link ‘the most experienced campidoctores, vicarii and tribuni’ in a formula equating to ‘officers commanding’, and implying that a campidoctor ranked third in the regimental command structure, but again the usage of campidoctor is equivocal.69 In a subsequent chapter concerning procedures for the fortification of an encampment Vegetius notes that the campidoctores et principes supervise the construction of the rampart, but he immediately proceeds to describe centuriones and tribuni performing this function. It is possible, therefore, that in this passage Vegetius employs campidoctores not as the title of a specific post, but as a vague non-technical synonym for centuriones, and this may even be the case in his earlier reference to ‘the most experienced campidoctores’ (exercitatissimi campidoctores).70 The intermittent imprecision of Vegetius (and perhaps also of Ammianus) in the use of the word campidoctor may be explained by the fact that, as will be discussed below, in late Roman legiones, structured according to the olderstyle regimental hierarchy, a senior centurio (or ordinarius) was indeed the non-commissioned officer mostly likely to assume the post of campidoctor; certainly this class of late Roman unit was of most interest to Vegetius in that it furnished the closest contemporary analogy to his idealised model of the legio antiqua.

period castes doubt on the historicity of this passage, at least as an accurate record of the fate of St. Maurice and the ‘Theban Legion’, and the modern reader should be alert to a practice common among hagiographical writers of embellishing their works with seemingly specific but in fact fictional references to officer-grades and regimental titles in the hope that this spurious circumstantial detail might further authenticate miracles and episodes of saintliness. Nevertheless, while the entire story of St. Maurice and the Theban Legion has been doubted as a fourth- or fifth-century fabrication, the sequence of officer-grades is itself consistent with other late Roman evidence for the new-style regimental hierarchy, and the fifth-century author (or his source) may have misapplied rather than invented these military details.72 This hierarchy receives some corroboration from an incidental reference in an undated, but certainly late antique, prose version of a fable entitled ‘The Ape King’ which was included in later editions of the fabulae of Phaedrus. The author has one of the participants refer to the simian army as comites tui, primicerii, campidoctores, militares officii, where comites means tribuni. Again the sequence tribunus – primicerius – campidoctor appears at least to report a common perception of contemporary senior regimental grades.73 The increased status and prestige of campidoctores are also manifest in the prominent part they played in imperial accession ceremonies in the fifth and sixth centuries, as reported in Constantine VII Porphyrogenitus’ De Cerimoniis, drawing on the lost accession narratives of Peter the Patrician (ca. 500565).74 At the elevation of Leo I in 457, it was Busalgus and Olympius, two campidoctores of unnamed units, who invested the emperor with military insignia, the first laying his torque (ȝĮȞȚȐțȚȞ) on the emperor’s head, the other placing his in Leo’s right hand. An unnamed campidoctor of the Lanciarii (IJ´Ȟ ȁĮȖțȚĮȡÊȦȞ țĮȝʌȚįȠÌțIJȦȡ), a legio palatina, was similarly charged with ‘crowning’ Anastasius with his torque in 491, and in 518 Godila, likewise a campidoctor of the Lanciarii, exercised the same function at the accession of Justin I.75 No campidoctores are recorded at the accession ceremonies of Leo II in 473 and Justinian in 527 since they were both crowned by a senior incumbent, Leo I and Justin I respectively, but it has been suggested that

Modest support for the campidoctor as the third-incommand of an infantry unit is supplied by two other late antique texts. The mid fifth-century Passio Agaunensium martyrum, ascribed to Eucherius, Bishop of Lyons (ca. 434-450), narrates the martyrdom of the officers and men of the ‘Thebaei’, a legio, early in the reign of Maximian (284-305). This document lists the three senior officers of this unit in the sequence primicerius, campidoctor, and senator militum. The primicerius (later saint) Maurice is the most senior officer in the unit, presumably serving as vicarius for an absentee tribune or as commanding officer of a legionary vexillation.71 The evidential value of this passage is again in doubt. The fact that the grade senator does not actually occur in the old-style hierarchy of a legio of any class (palatina, comitatensis or limitanea) or

72 For the date, sources and historicity of this document see van Berchem 1956: 32-33; Bellen 1961; Dupraz 1961; O’Reilly 1978; and most thoroughly Woods 1994. 73 The fable ‘Simius Tyrannus’ is included as Phaedrus IV.13 in B.E. Perry (ed. and Eng. tr.), Babrius and Phaedrus ([LCL] Cambridge, Mass. 1965), but is omitted from the Teubner (L. Müller, 1926) and Budé (A. Brenot, 1924) editions. It is found only in some mediaeval prose paraphrases of Phaedrus, which descend from a lost archetype dating to the sixth century at the latest. For the textual history see Henderson 1999. 74 For fifth- and sixth-century accession ceremonies and their military component see Enßlin 1942; MacCormack 1981: 241-247. 75 Const. Porph. de Cer., ed. J.J. Reiske ([CSHB] Bonn 1829-30) I.91 (411.5-8A); I.92 (423.7-9D); I.93 (429.4ǹ). Wheeler 2004: 168 with n. 111 interprets IJ´Ȟ ȁĮȖțȚĮȡÊȦȞ as the troop-type lanciarii, rather than the regimental title Lanciarii; on this legio palatina see below.

non-technical sense of ‘front-rankers’: XV.7.4; XVI.12.18-20; XIX.6.12; XXV.3.3; XXVI.8.5; XXVII.10.12; XXXI.10.13; cf. postsignani for non-specific ‘rearward ranks’ at XVIII.8.7; XXIV.6.9. For a judicious review of problems arising from these two terms in Roman sources see now Wheeler 2004: 166-170. 68 Veg. Epit. I.13.1; II.23.3; III.26.37; cf. II.7.6 (campigenii = campidoctores). 69 Veg. Epit. III.6.23: praeponendi ergo sunt exercitatissimi campidoctores vicarii vel tribuni. 70 Veg. Epit. III. 8.11: campidoctoribus et principis, but centuriones… et tribuni at ibid. 8.13; see Milner 1996: 76 n.5; 81 n.4. 71 Passio Agaunensium martyrium §8, ed. B. Krusch ([MGH Scr. Rer. Merov. III] Berlin 1896) 32-39.

404

PHILIP RANCE: CAMPIDOCTORES VICARII VEL TRIBUNI Armatus who performed a parallel role at the elevation of Justin II in 565 was also a campidoctor, a proposition strengthened by his on-the-spot promotion to tribune.76 These events permit certain observations. First, the involvement of the campidoctor of the same regiment, the Lanciarii, in the ceremonies of both 491 and 518 suggests that this role may have been a privilege, prescribed by law or custom, enjoyed by particular palatine units. Furthermore, the participation of two campidoctores in 457 hints at the involvement of the formerly-paired legiones the Lanciarii and Mattiari, though the latter unit is not explicitly attested in this role.77 By this date these two regiments had both long been divided into seniores or iuniores, and if this interpretation is correct it remains unclear which two of the resulting four units participated in this capacity, but the limited evidence for unit distributions and common sense marginally favour the Lanciarii seniores and Mattiarii seniores. I know of no explanation for the choice of these two particular regiments, though their status as respectively the senior legio palatina in the armies of the magistri militum praesentales I and II implies that it was a matter of regimental precedence. It may also be significant that Leo I, the emperor for whom this form of ceremony was first formalised at Constantinople, was tribune of the Mattiari immediately prior to has accession.78 Second, the selection of specifically campidoctores, rather than any other representative of the regiment, to play such a conspicuous role in the most important of imperial rituals has not been satisfactorily explained, but while the reasons behind this choice remain obscure it nevertheless further challenges the traditional perception of campidoctores as humble ‘drill-sergeants’.79 Third, it is rarely possible to trace the careers of individual campidoctores, especially given the diminution in the volume of relevant epigraphic data in the later period, but promotion to tribune from the post of campidoctor seems to have been a reasonable expectation from the fourth century. We have already noted the likelihood that Armatus who crowned Justin II in 565 was a campidoctor

afterwards promoted to tribune. A plausible case has also been made for identifying Busalgus (ȕȠÌıĮȜȖȠȢ), the campidoctor of a palatine unit at the accession ceremony of Leo in 457, with Busalbus (ȕȠÌıĮȜȕȠȢ), commander (ÙȖȠÌȝİȞȠȢ ıIJȡĮIJȚȦIJȚțȠ³ IJÇȖȝĮIJȠȢ, presumably tribune) of an unnamed unit stationed at or near Constantinople in 479.80 A yet more illustrious career-path was followed by the campidoctor Godila, who crowned Justin I in 518, arriving at the rank of magister militum within a decade.81 One final note in this context concerns the possibility that the campidoctores were issued with their own standards. Constantine VII elsewhere provides two lists of different types of flag or standard, which include the term țĮȝʌȚįȠȣțIJÍȡȚȠȞ (*campidoctorium?), apparently a kind of emblem or insignia unique to a campidoctor.82 This item is not mentioned in any other text and, if correctly identified, it is not possible to determine whether this was a late Roman phenomenon or a subsequent Byzantine development, nor whether this standard was intended for ceremony, the field or both. Modest circumstantial support for the earlier existence of a *campidoctorium is provided by Maurice’s stipulation that, as previously observed, whenever a high-strength infantry regiment is to be split into two combat formations, one commanded by the tribune, the other by the vicarius or campidoctor, each formation must have its own standard. A case could also be made for the intrinsic utility of a campidoctor being distinguished by some sort of marker when carrying out his primary function as regimental drillmaster, and especially given his position out in front of the regiment during drill and prior to combat. While certainty on this point remains elusive, it is worth noting that only the most important field officers qualified for their own ensign. Since none of the later Roman evidence is compatible with the modern presentation of a campidoctor as a lowgrade ‘drill-sergeant’, it remains necessary to establish the position of a campidoctor in the regimental hierarchy and to account for the considerable increase in the status and responsibilities of campidoctores from the fourth century. In the later Roman sources the rank of a campidoctor is never explicitly stated. We have seen that the evidence from the mid fourth century onwards, such as it is, points to the campidoctor being the third-incommand of an infantry regiment after the tribune and primicerius (as vicarius). It is also highly probable that, as under the Principate, campidoctor was a post or specialist function assigned to one of the senior noncommissioned grades rather than a rank in its own right.

76 Corippus, In Laud. Iust. II.130-136, where the torque is placed around Justin’s neck (inpositi in collo) rather than on his head. See remarks of Hoffmann 1969-70: I, 488; Stache 1976: 271-276; Cameron 1976: 159160; with PLRE IIIA:120, Armatus¹. 77 Enßlin 1942: 272-273, 285; Hoffmann 1969-70: I, 487-488. For the operational pairing of the Lanciarii and Mattiari in the fourth century cf. Amm. Marc. XXI.13.16; XXXI.13.8; Zos. III.22.2; J. Mal. Chron. ed. H. Thurn (Berlin-New York 2000) 253.68-70 (= Magnus of Carrhae, FHG frg. 4.5), with Hoffmann 1969-70: I, 218-222, 324-330. 78 The configuration of the four regiments in the Notitia Dignitatum could support various hypotheses. The Lanciarii seniores and Mattiari iuniores served together under the magister militum praesentalis I (ND Or. V.42, 47); while the Mattiari seniores and Lanciarii iuniores were under the magister militum praesentalis II (ND Or. VI.42, 47). There is strong evidence that in the late fourth century the Lanciari iuniores were based in Lycia; see Hoffmann 1969-70: I, 329; II, 140. For Leo as tribune of the Mattiari: Const. Porph. de Cer. I.91 (411.4). 79 The use of a torque is precedented in the elevation of Julian in 360 when that of a draconarius of the Petulantes was used, cf. Amm. Marc. XX.4.18. MacCormack 1981: 241-243 traces the formalisation of this ad hoc in-the-field expedient into a ‘military coronation’ at Constantinople; the choice of campidoctores, however, remains to be explained.

80

Amm. Marc. XV.3.10 reports a man promoted tribune ex campidoctore by 355. Both Busalgus and Olympius are omitted from PLRE. Hoffmann 1969-70: I, 488 makes the case for equating Busalgus (ȕȠÌıĮȜȖȠȢ) and Busalbus (ȕȠÌıĮȜȕȠȢ) in J. Ant. frg. 211.3 [FHG IV.619]). For Busalbus see now PLRE II:243. For Armatus see n. 76. 81 PLRE II:516, Godilas; IIIA:539-40, Godilas¹. 82 Const. Porph. de Cer. I.1 (11.21C): IJÇ ıțİÌȘ IJ´Ȟ ʌȡȠIJȚțIJËȡȦȞ țĮÃ ıȚȞĮIJËȡȦȞ țĮÃ įȡĮțȠȞĮȡȓȦȞ, ȜÇȕȠȣȡÇ IJİ țĮÃ țĮȝʌȘįȘțIJËȡȚĮ, ȝİIJÇ țĮÃ IJ´Ȟ ȕÇȞįȦȞ; ibid. II.15 (575.19D): țĮÃ IJÇ ȜȠȚʌÇ ıțİÌȘ, IJÇ țĮȝʌȚįȚțIJËȡȚĮ țĮÃ ȜÇȕȠȣȡÇ țĮȓ ıÊȖȞĮ. See Mihăescu 1978-9: [III]361; Haldon 1984: 108-9, 121; 1990: 272-3.

405

THE LATE ROMAN ARMY IN THE EAST This view is affirmed by a late fourth-century funerary inscription from Laodicea Combusta in Lycia (mod. Yorghan Ladik) which commemorates a Paulus ÐʌÄ țĮȝʌȚįȠȣțIJËȡȦȞ ÖȡįİȞĮȡÊȠȣ. This man was possibly serving in the Lanciarii iuniores recorded in this location in this period; in any event the designation ordinarius, the prevailing late Roman synonym for centurio, indicates a unit of the legio class.83 This inscription has prompted differing interpretations. Stein argued that ÐʌÄ țĮȝʌȚįȠȣțIJËȡȦȞ ÖȡįİȞĮȡÊȠȣ indicates promotion ‘from’ the grade of campidoctor to ordinarius, an erroneous assumption which provided the foundation of Stein’s case for excluding campidoctores from this senior grade of non-commissioned officer. In a recent discussion Wheeler rightly doubts Stein’s interpretation, pointing to the high status of campidoctores in other fourth-century sources, and suggests that the epitaph ‘probably indicates which kind of ordinarius’ is meant.84 The meaning of the expression ÐʌÄ țĮȝʌȚįȠȣțIJËȡȦȞ can in be better explained within the context of late Roman titulature. The distinguishing feature of this formula is the use of the genitive plural, equating to Latin ex campidoctoribus. The singular could indeed, as Stein believed, signify promotion ‘from campidoctor’, as in the case of the tribunus vacans promoted ex campidoctore reported by Ammianus.85 The use of ÐʌÄ with the gentive plural, in contrast, is well attested in Greek from the third century, both in the military or civilian spheres, as a formula signifying that the bearer had at some time in the past held a particular post or office of distinction and was subsequently entitled to the continued honorific status and in some cases the attendant privileges.86 We have already noted a significant number of individuals in the legio stationed at Syene who chose to style themselves ÐʌÄ ȕȚțĮȡȚ´Ȟ (= ex vicariis), ‘sometime vicarius’, indicating that they had formerly had the honour of serving as ‘deputy tribune’, an emeritus distinction which obviated the need to specify their current rank. Another instance of a ‘former campidoctor’ (ÐʌÄ țĮȝʌȚįȠȣțIJËȡȦȞ), without further indication of actual rank, is recorded serving in the Transtigritani at Arsinoe in 498.87 These two instances of a ‘sometime campidoctor’, therefore, necessarily imply that campidoctor was a temporary post rather than a fixed rank in a hierarchy, and moreover that it was a position of some distinction which ex-holders would exploit for honorific purposes. The ÐʌÄ țĮȝʌȚįȠȣțIJËȡȦȞ

ÖȡįİȞĮȡÊȠȣ (= ex campidoctoribus ordinarius) on the inscription from Laodicea Combusta demonstrates, quite the contrary to Stein’s thesis, and as Wheeler infers, that in legiones in the late fourth century campidoctor was a post held by ordinarii (centuriones), the senior noncommissioned officers of this class of unit. This conclusion does little to facilitate identification of the rank of campidoctores in the newer-style ‘Constantinian’ hierarchy. No source specifies the rank of a campidoctor in units of this class, though a late fourthcentury funerary inscription from Concordia to a campidoctor of the Batavi seniores who died aged 60 after 35 years’ service provides circumstantial evidence of the veteran status required for this post; certainly this individual had earned his place among Vegetius’ exercitatissimi campidoctores.88 The conjecture of Stein that the post of campidoctor was held by experienced biarchi, one of the most junior non-commissioned grades, is without foundation and demonstrably incorrect.89 Jerome’s listing of the regimental hierarchy is of course of no assistance in this respect, since campidoctor is a post not a rank, and in any case he explicitly describes the grades of a cavalry unit. According to a strict reading of Jerome’s list the non-commissioned grade of senator ranks third after the tribune and primicerius, and, on the basis of the evidence cited above, the little-understood senator is the most eligible grade to be campidoctor.90 This deduction seems to be precluded, however, by the mid fifth-century Passio Agaunensium martyrum, which, as previously observed, names as the three senior officers of the ‘Theban Legion’ a primicerius, a campidoctor, and a senator militum.91 It is of course possible that the author of this hagiographical text errs in this technical detail, and the problems inherent to this particular work and the genre as a whole have already been noted. In any case, no argument can depend on this text and the limited evidence demands a more cautious approach. It is better to draw the provisional conclusion that the post of campidoctor in units with the newer-style hierarchy was open to a range of senior non-commissioned grades, such as senator, ducenarius or centenarius, and possibly even primicerius, perhaps held in combination with the other positions for which this grade was qualified. Such a broad eligibility of grades for the post of senior ‘trainingofficer’ would also correspond to the practices observed under the Principate. Much about the history of the campidoctor remains obscure, and one can only conjecture the process that transformed a rarely-attested and typically low-grade ‘drill-sergeant’ of the late second/early third centuries into a high-ranking non-commissioned officer, junior only to the tribune and primicerius, in the later Roman

83

MAMA I:86, no.168. For the identification of the unit see Hoffmann 1969-70: I, 329; II, 140. For ordinarii see Grosse 1920: 127; Stein 1933; Gilliam 1940; Keenan 1973; 1990: 147-148; Rea 1980. 84 Stein 1933: 379, 386-387; Wheeler 2004: II, 168-169 (erroneously citing I.186 and ÖȡįȚȞĮȡÊȠȣ). 85 Amm. Marc. XV.3.10 86 See the discussion of the analogous civilian honorific formula ÐpÄ ÑpÇȡȤȦȞ, ‘sometime prefect’, by Guilland 1982; Roueché 1989: 23-24. A parallel usage in late Roman honorifics is the granting of retirement status and entitlement to soldiers and officials on the basis of their former holding of a fictive office, such as ex protectoribus, ex comitibus and ex praesidibus, see Jones LRE 104, 526, 530, 637, 744; Haldon 1984: 131. 87 P. Lond. 113.5(a).4-5.

88

CIL V 8773 = ILS 2803, with Hoffmann 1969-70: I, 77; II, 29 n.214. E. Stein, Review of Grosse 1920 in BZ 25, 1925: 386-392 at 388. 90 The limited evidence for the grade of senator is assembled by Grosse 1920: 119-120; Jones LRE 634 with n.57, 647, 674; Hoffmann 1969-70: 75-77, 280, supplemented by CIL VIII 17414. 91 Passio Agaunensium martyrium §8. 89

406

PHILIP RANCE: CAMPIDOCTORES VICARII VEL TRIBUNI army. The link between the small number of campidoctores recorded during the Principate, narrowly restricted in both period and unit-type, and the more regular and widespread attestation of this title from the second half of the fourth century is rather tenuous, but apparent uniformity in the function and duties of campidoctores in both periods points to some measure of institutional and titular continuity. It is possible that a specialised post and title first created in the Praetorian Guard and/or equites singulares were extended to other parts of the army, perhaps gradually via the regular promotional transfer of personnel from these élite corps to provincial postings, as part of a broader current of dissemination and standardization of military techniques from the centre to the periphery.92 While responsibility for weapons-instruction, physical fitness and drill had fallen on tribunes and centurions before the first century, and thereafter was increasingly delegated to a number of specialist instructors, now poorly understood, by the fourth century at the latest overall supervision of these important activities had been concentrated or centralised upon a single ‘drill-master’ uniformly styled campidoctor for each infantry regiment, regardless of the class of unit. This process may have been in operation during the third century (for which the evidence is slimmer than some scholars have assumed), but the introduction of a new regimental hierarchy, without the traditional centurionate, for all newly-created units by the 320s at the latest, probably both accelerated and regularised this development. In units with an older-style hierarchy (legiones) centurions, now more frequently styled ordinarii, occupied the post of campidoctor, possibly in rotation, and thus retained their earlier responsibilities in the sphere of training, but in the newer-style hierarchy the non-commissioned grades that were eligible to hold this post cannot be determined with certainty. It was perhaps as a result of the importance the campidoctor acquired by heading a separate internal hierarchy or team of trainers, as well as the experience and professional expertise required for this appointment, that the status, responsibilities and profile of campidoctores all substantially increased in the later Roman period, and the post of campidoctor gained a new significance as a position of distinction, a stepping stone to the tribunate and, in specific instances, the regimental representative in the most exalted of imperial ceremonies.

Bellen H. 1961 Der Primicerius Mauricius. In Historia 10: 238-247. van Berchem D. 1956 Le martyre de la Légion Thébaine, Essai sur la formation d’une légende, Basle. Beševliev V. 1964 Spätgriechische und spätlateinische Inschriften aus Bulgarien, Berlin. Beurlier E. 1884 Campidoctores et Campiductores. In L. Havet (ed.), Mélanges Graux. Recueil de travaux d’érudition classique dédié à la mémoire de Charles Graux, Paris, 297-303. Le Bohec Y. 1994 The Roman Imperial Army (Eng. tr. of L’Armée Romaine, sous le Haut-Empire [Paris 1989]), London. Brennan P. 1998 Divide and Fall: the Separation of the Legionary Cavalry and the Fragmentation of the Roman Empire. In T.W. Hilliard et al. (ed.), Ancient History in a Modern University 2, Grand Rapids, 2: 238-244. Brown T.S. 1984 Gentlemen and Officers. Imperial Administration and Aristocratic Power in Byzantine Italy A.D. 554-800, Rome. Cameron A.M. (ed., Eng. tr. and comm.) 1976 In Laudem Iustini Augusti Minoris, London. Campbell B. 1994 The Roman Army 31 BC-AD 337: A Sourcebook, London-New York. Coello T. 1996 Unit Sizes in the Late Roman Army, BAR Ser. 645, Oxford. Dain A. 1943 Naumachica, Paris. Dain† A. 1967 (texte mis au net et complété par J.-A. de Foucault), Les Statégistes byzantins. In T&M 2: 317-392. Dain† A. 1968 (complété par J.-A. de Foucault), Urbicius ou Mauricius? In REB 26: 123-136. Davies R.W. 1968 Training Grounds of the Roman Cavalry. In Archaeological Journal 125: 73-100; repr. in idem, Service in the Roman Army, ed. D. Breeze and V. Maxfield (Edinburgh 1989) 93-124. Davies R.W. 1974 Roman Military Training Grounds. In E. Birley, B. Dobson and M.G. Jarrett (ed.), Roman Frontier Studies 1969 ([8th International Congress of Limesforschung] Cardiff 1974) 20-26. Dennis G.T. (ed.) 1981 Das Strategikon des Maurikios, with Ger. tr. E. Gamillscheg ([CFHB 17] Vienna 1981). Dennis G.T. 1984 Maurice’s Strategikon: Handbook of Byzantine Military Strategy, Philadelphia. von Domaszewski A. 1967 Die Rangordnung des römischen Heeres, rev. ed. by B. Dobson (Beihefte der Bonner Jährbucher 14), Cologne-Graz. Dupraz L. 1961 Les passions de S. Maurice d’Agaune: Essai sur l'historicité de la tradition et contribution à l’étude de l’armée pre-dioclétienne (260-286) et des canonisations tardives de la fin du IVe siècle (Studia Friburgensia n.s. 27), Fribourg. Durliat J. 1979 Magister Militum – ıIJȡĮIJȘȜÇIJȘȢ dans l’empire byzantine, Ve – VIIe siècles. In BZ 72: 306320.

Bibliography Alföldy G. 1975 Die römischen Inschriften von Tarraco, Berlin. Aussaresses F. 1909 L’Armée byzantine à la fin du VIe siècle d’après le “Strategicon” de l’empereur Maurice (Bibliothèque des Universités du Midi, fasc. XIV), Paris.

92 For the possible role of the Praetorians as an Ausbildungsschule see Horsmann 1991: 101-102, with Speidel 1978: 51-52.

407

THE LATE ROMAN ARMY IN THE EAST Enßlin W. 1942 Zur Torqueskrönung und Schilderhebung bei der Kaiserwahl. In Klio 35: 268298. Elton H. 1996 Warfare in Roman Europe, Oxford. Fink R.O. 1971 Roman Military Records on Papyrus (American Philological Association Philological Monographs 26), Cleveland, Ohio. Flusin B. 1992 Saint Anastase le Perse et l’histoire de la Palestine au début du VIIe siècle, Paris, 2 vols. Gilliam J.J. 1940 The Ordinarii and Ordinati of the Roman Army. In TAPA 71: 127-48 (= idem, Roman Army Papers [Amsterdam 1986] 1-22). Grosse R. 1918 Die Rangordnung der römischen Armee des 4.-6. Jahrhunderts. In Klio 15: 122-161. Grosse R. 1920 Römische Militärgeschichte von Gallienus bis zum Beginn der byzantinischen Themenverfassung, Berlin (repr. New York 1975). Guilland R. 1982 Études sur l’histoire administrative Byzantine III. L’apoparque – ÐpÄ ÑpÇȡȤȦȞ. In ByzSlav 43: 30-44. Haldon J.F. 1984 Byzantine Praetorians: An Administrative, Institutional and Social Survey of the Opsikion and Tagmata, c. 580-900, Bonn. Haldon J.F. 1990 Three Treatises on Imperial Military Expeditions (CFHB 28), Vienna. Haldon J.F. 1993 Administrative Continuities and Structural Transformations in East Roman Military Organisation ca. 580-640. In Kazanski and Vallett 1993: 43-54. Haldon J.F. 1999 Warfare, State and Society in the Byzantine World 565-1204, London. Henderson J. 1999 ‘Phaedrus’ Fables. The Original Corpus. In Mnemosyne 52: 308-329. Hoffmann D. 1969-70 Das spätrömische Bewegungsheer und die Notitia Dignitatum, Dusseldorf. Horsmann G. 1991 Untersuchungen zur militärischen Ausbildung im republikanischen und kaiserzeitlichen Rom, Boppard am Rhein. de Jonge P. 1982 Philological and Historical Commentary on Ammianus Marcellinus XIX, Groningen. Kazanski M. and Vallet F. (ed.) 1993 L'Armée romaine et les barbares du IIIe au VIIe siècles (Mémoires de l’Association Française d’Archéologie Mérovingienne 5), Paris. Keenan J.G. 1973 An Instance of the Military Grade Flavialis. In BASP 10: 43-46. Keenan J. 1990 Evidence for the Byzantine Army in the Syene Papyri. In The Bulletin of the American Society of Papyrologists 27: 139-150. Key Fowden E. 1999 The Barbarian Plain. Saint Sergius between Rome and Iran, Berkeley. Kramer J. 1987 Ein Gräzismus gotischer Herkunft im Italienischen: bando. In Balkan-Archiv N.F. 12: 197207. Kramer J. 1996 Papyrusbelege für fünf germanische Wörter: ÐȡȝĮȜĮÌıȚȠȞ, ȕÇȞįȠȞ, ȕȠȣȡį´Ȟ, ȕȡĮțÊȠȞ, ıĮijÍȞȚȠȞ. In Archiv für Papyrusforschung 42: 113126.

MacCormack S.G. 1981 Art and Ceremony in Late Antiquity, Berkeley-Los Angeles-London. Mihăescu H. 1967 Prolégomènes à une Édition critique des ȉǹȀȉǿȀǹ-ȈȉȇǹȉǾīǿȀǹ de Maurice-Urbicius. In RESEE 5: 401-417. Mihăescu H. 1978-9 La littérature byzantine, source de connaissance du latin vulgaire. In RESEE [I] 16: 195212; [II] 17: 3-25, [III] 17: 359-383. Milner N.P. 1996² Vegetius: Epitome of Military Science, Liverpool. Ȃüller ǹ. 1912 Das Heer Justinians (nach Procop und Agathias). In Philologus 71: 101-138. Müller K.K. 1880 Ein Griechisches Fragment über Kriegswesen. In Festschrift für Ludwig Urlichs, Würzburg, 106-138. Nicasie M.J. 1998 Twilight of Empire: The Roman Army from the Reign of Diocletian until the Battle of Adrianople (Dutch Monographs on Ancient History and Archaeology v. 19), Amsterdam. O’Reilly D. F. 1978 The Theban Legion of St Maurice. In Vigiliae Christianae 32: 195-207. Rance P. 2000 Simulacra pugnae: the literary and historical tradition of mock battles in the Roman and early Byzantine army. In GRBS 41.3: 223-275. Rance P. 2004 The Fulcum, the late Roman and Byzantine testudo: the Germanization of late Roman Tactics? In GRBS 44: 265-326. Rance P. (forthcoming 2008) The Roman Art of War in Late Antiquity (Birmingham Byzantine, Ottoman and Modern Greek Studies), Aldershot. Ravegnani G. 1998 Soldati di Bisanzio in Età Giustinianea (Materiali e Ricerche, nuova serie 6), Rome. Rea J.R. 1980 Ordinatus. In ZPE 38: 217-219. Rea J.R. 1984 A cavalryman's career, AD 384 (?)-401. In ZPE 56: 79-88. Richardot P. 1993 Hiérarchie militaire et organisation légionnaire chez Végèce. In Kazanski and Vallet 1993: 405-426. Richardot P. 2001 La Fin de l’Armée Romaine (284476), rev. ed.², Paris. Roueché C. 1989 Aphrodisias in late Antiquity, London. Scharf R. 1997 Constantiniaci = Constantiani? Ein Beitrag zur Textkritik der Notitia Dignitatum am Beispiel der “constantinischen” Truppen. In Tyche 12: 189-212. Speidel M.P. 1965 Die Equites Singulares Augusti. Begleittruppe der römischen Kaiser des zweiten und dritten Jahrhunderts (Antiquitas I.11), Bonn. Speidel M.P. 1978 Guards of the Roman Armies (Antiquitas I.28), Bonn. Speidel M.P. 1984 Cataphractarii clibanarii and the rise of the later Roman mailed cavalry: a gravestone from Claudiopolis in Bithynia. In Epigraphica Anatolica 4: 151-156. Speidel M.P. 1990 The Names of Legionary Centuriae. In Arctos 24: 135-137. Speidel M.P. 1994 Die Denkmäler der Kaiserreiter: Equites Singulares Augusti, Cologne. 408

PHILIP RANCE: CAMPIDOCTORES VICARII VEL TRIBUNI Speidel M.P. 1995 A Horse Guardsman in the War between Licinius and Constantine. In Chiron 25: 8387. Southern P. - Dixon K.R. 1996 The Late Roman Army, London. Stache U.J. 1976 Flavius Cresconius Corippis, In Laudem Iustini Augusti Minoris. Ein Kommentar, Berlin. Stein E. 1933, Ordinarii et Campidoctores. In Byz 8: 379387. Treadgold W. 1995 Byzantium and its Army, Stanford. Vári R. 1906 Zur Überlieferung mittelgriechischer Taktiker. In BZ 15: 47-87. Vogler C. 1993 Les officiers de l’armée romaine dans l’oeuvre d’Ammien Marcellin. In Kazanski and Vallet 1993: 389-403. Watson G.R. 1969 The Roman Soldier, London. Webster G. 1985³ The Roman Imperial Army, London. Wheeler E.L. 2004 The Legion as Phalanx in the Late Empire [pt. II]. In Revue des Études Militaires Anciennes 1: 147-75. Woods D. 1994 The Origin of the Legend of Maurice and the Theban Legion. In Journal of Ecclesiastical History 45: 385-395. Woods D. 1997a Valerius Victorinus Again. In Chiron 27: 85-93. Woods D. 1997b The Emperor Julian and the Passion of Sergius and Bacchus. In Journal of Early Christian Studies 5: 335-367. Zuckerman C. 1995 Le įİȪIJİȡom ȕȐmįom ȀȦmıIJĮmIJȚmiĮț´m dans une épitaphe de Pylai. In Tyche 10: 233-235. Zuckerman C. 1998 Constantiniani – Constantiniaci from Pylai: A Rejoinder. In Tyche 13: 255-258.

409

From the Late Roman to the Early Byzantine Army. Two Aspects of Change Oliver Schmitt M. Luther Universität, Halle

The “Roman army in the east” topic has gained some prominence among scholars in recent times. A lot of research has been put in various aspects of Roman warfare in the eastern provinces but nevertheless a number of problems need further treatment and to two of them I will devote the following pages:

Two hundered and fifty years later, after the introduction of the theme system, the picture seems to be quite different. Now we find a largely decentralized Byzantine army in Asia minor, its most important regional corps being those of the themes Anatolikon and Armeniakon, with a third corps stationed in the rear, i. e. the army of the Thrakesion theme. The Balkans no longer possessed an army of its own. There was only one imperial reserve force left near Constantinople, organized in the Obsikion theme.5 Ducates did not exist anymore.

First, I am going to ask if the introduction of the thematic system in the middle of the seventh century was the result of a coherent and logical process, which had its roots in the late Roman army organization of the early fifth century; and

For a long time, the introduction of the theme system was seen as an invention of the emperor Heraclius, who was believed to have created something more or less completely new out of the remnants of the late Roman army, but this view has been reasonably refuted by scholars in the last decades.6 Recent studies however, have emphasized the continuity between late Roman and early Byzantine army organization. The thematic armies are now seen as a mere offspring of the former field armies of the late Roman magistri: the remnants of the praesental forces, it is presumed, became the army of Obsikion, the units of the magister militum per Thracias were transferred to western Asia minor and transformed into the Thrakesion army, the forces of the magister militum per Orientem became the army of the theme Anatolikon and last but not least, the troops of the magister militum per Armeniam - a command introduced by Justinian - were dislocated in north eastern Asia minor to build up the Armeniakon theme.7 But was it quite so simple? To answer this question we have to look at the history of the five late Roman army corps since the fifth century.

second, I will try to explain the role of the cavalry divisions of the late Roman army, which were created during the last decades of the sixth century. In spite of its divided command structure, the late Roman field army (comitatenses) of the early 5th century was to a high degree a centralized force. The larger part of the best units was assigned to the two praesental armies (or army corps), which were stationed in the suburbs and the hinterland of Constantinople.1 Their nominal strength (Sollstärke) can be reasonably estimated at ca. 50,000 soldiers.2 In addition, there were up to three smaller, less important regional armies (or corps): those of the magistri militum per Orientem (up to 21,600 soldiers strong), per Thracias (ca. 19,000 men) and per Illyricum (20,000 men in theory).3 These numbers mean that nearly the half of the field army as a whole had their quarters in or in the vicinity of the capital and there is some reason to presume that in terms of actual strength (Iststärke) even more than half of the field army belonged to the two praesental corps. On the borders, the defensive organisation was based on the ducati, military districts under duces respectively, who had at their disposition a number of units of their own, the so-called limitanei. It is impossible to give the nominal strength of these units, but it seems that the actual strength of many limitaneiregiments was often very low.4

Although evidence for the east Roman army in action is scanty for the 5th century, we can reasonably conclude that the system outlined in the Notitia Dignitatum worked during this period. Without doubt great expeditions were undertaken by units from the praesental armies. An example is the huge force sent to Italy in 424 to reinstall the Theodosian dynasty, which was led by Ardabur, who

1

For the composition of the praesental armies see ND Or. V.27-66; VI.27-69; for the location of their barracks cf. Chron. Pasch. 622.17-18 Dindorf. 2 Estimations for the following numbers are my own (based on Schmitt 2001[2004]: 96-98); I have given here 800 soldiers to each numerus of infantry and 400 soldiers to each vexillatio of cavalry. Slightly different numbers are given by Treadgold 1995: 47-50. 3 For the regional armies see ND Or. VII.24-58; VIII.24-53; IX.18-48. 4 A baffling variety of unit categories exists in the lists of the Notitia: equites, cunei, alae, legiones, cohortes, milites, auxiliares and auxilia for which no information about nominal strength is given by any source. It is obvious that the units classified as legiones, cohortes and alae can no longer compared to the units known from the early principate. For actual strengths of limitanei-units cf. Synesius, ep. 78.2-8 Garzya, who gives 40 men to a unit called Hunnigardi.

5 It’s a pity that no reliable numbers are given for the thematic armies of the seventh century. Whittow 1996: 183 comes to the conclusion that the strength of the thematic armies at this time was well below 100,000 men, which seems quite realistic. 6 For the older view Ostrogorsky 1963: 80-83. Against it see, for example, the arguments of Karayannopulos 1959: passim; Lilie 1984: 27-39; cf. the summary given by Haldon 1993: 3-7; For a not very convincing attempt to resuscitate the theory of Heraclius as the creator of the themes, which can not discussed in detail here see Schmitt 2001: 210-211 with further reading. 7 Treadgold 1995: 23; 1997: 315-316; Whittow 1996: 120-121; Haldon 1999: 73; Lilie 2003: 134-135; cf. already Stein 1919: 133.

411

THE LATE ROMAN ARMY IN THE EAST held the office of magister militum praesentalis.8 Some years later, when his son Aspar was sent to Africa in 431 with an army to fight the Vandals, he probably held the same rank9 and Basiliscus, commander-in-chief of the great Vandal expedition of 468 apparently was praesental magister too and his army was composed of troops from Constantinople.10

lead by Celer, the magister officiorum.18 Procopius tells us - though as a legomenon only - that this was the largest force ever brought together against the Persians,19 an opinion, which is open to some doubts, but certainly it was a exceptionally huge host,20 which must have included a strong corps of foederati, most of them of Germanic origin.21

The defensive wars against the Persians in 421/22 and 441 however were mainly fought by the troops of the magister militum per orientem and the local duces as far as we know,11 but in the battles against the Huns the magister militum per Thracias Arnegisclus was supported by substantial forces from both praesental armies under the magistri militum praesentales Areobindus and Aspar;12 probably troops from the oriental army were also involved.13 But the wars against the Ostrogoths and the Gepids seem to have been fought primarly by the army of Illyricum; the strength of which is given with at least 10,000 men for the year 505.14 It is possible that this regional corps was brought up to substantial strength by units borrowed from the praesental armies.15

It should have become clear that the military system introduced in the early 5th century worked well nearly 100 years later and it focused still on the praesental armies, with the armies of the magistri militum per Orientem and per Thracias assuming a secondary role. Before we look at the transformations of the sixth century, something must be said about the ducatal armies. It was their main task to deal with minor attacks, mostly raids by warbands, but their role was not always restricted to that. Troops from Egypt, the Thebaid and the Cyrenaica took part in the great Vandal expedition of 468, operating successfully,22 which is somewhat surprising, because the army in Egypt is often judged as a mere police force, whose main occupation was to maintain internal security.23 Some thirty years later the dux Palaestinae undertook a counteroffensive against Arab nomads and reoccupied some lost regions on the desert fringe.24 During the Persian War of Anastasius, the duces of Armenia and Mesopotamia operated independently on a minor scale against marauding parties until they were eventually beaten in a field battle against a numerically superior Persian force.25 Although such remarkable fighting qualities may have been in part the result of the fact that at least since the reign of Anastasius the duces had access to regiments detached from the field armies,26 these examples show that we have no reason to assume a general decline of the limitanei during the fifth century.27 We should keep this in mind whenever we think about the fate of the eastern Roman armies in the sixth century.

More detailed evidence for the cooperation of the praesental armies with the troops of a regional magister militum is provided for the war against the SƗsƗnids at the beginning of the 6th century. The main force consisted of the praesental armies under their magistri Patricius and Hypatius. According to our source Joshua Stylites this force alone numbered 40,000 men,16 a realistic figure, indeed. That means that the mass of the units available was mobilized. Areobindus, the magister militum per orientem, had 12 000 soldiers under his command.17 Later, a further corps of unknown strength was added, 8

Olymp. fr. 43.4-5 Blockley; the second praesental magister may have been Ardabur’s son Aspar (PLRE II 165 s.v. Fl. Ardabur Aspar) or, in my opinion more probably, Candidianus (Olymp. fr. 43.11-13 Blockley). 9 Procopius (BV I.3.35) tells us that the east Roman troops in Aspar’s force came from Constantinople. 10 For Basiliscus see Theoph. 115.29-116.2 de Boor; PLRE II 213 s.v. Basiliscus 2; for his army cf. Proc. BV I.6.1. 11 In the Persian war of 421/22 the Roman commander-in-chief was Ardabur, apparently holding the rank of the oriental magister, v. Socr. VII.18.9; cf. PLRE II 137 s.v. Fl. Ardabur 3; for the duces v. Socr. VII.18.25 where a certain Vitianos is named. The war of 441 was lead by the magister militum per orientem Anatolius, v. Proc. BP I.2.12; Marcell. com. s. a. 441.1 writes that Anatolius was supported by Aspar, presumably acting as magister militum praesentalis. For doubts about Aspars’ rank in 441 v. PLRE II 165 s.v. Fl. Ardabur Aspar. 12 Theoph. 102.19-21 de Boor; for Arnegisclus, who was killed in 447 v. Marcell. com. s. a. 447.5; Jord., Rom. 331. For Areobindus v. Nov. Theod. VII.4. 13 Cf. Prisc. fr. 9.3.2 Blockley. 14 Joh. Ant. fr. 213 Müller; Marcell. com. s. a. 479.1-2; 505. 15 The fate of the Illyrian army after 395 remains somewhat of a mystery. Although the ND (Or. IX.18-48) gives it remarkable paper strength for the early fifth century, it seems to have been virtually not in existence during the war against Alaric and his Goths, cf. Cesa 1994: 65-72; Heather 1991: 199-204 and is not mentioned in the wars against the Huns. If Alaric was made magister militum per Illyricum in 397 (cf. Cesa 1994: 73-76, who doubts it; Heather 1991: 205), he apparently had no Roman troops under his command, cf. Claud. 26.535-539 Hall. 16 Josh. Styl. 54; Proc. BP I.8.2. 17 Josh. Styl. 54.

It was Justinian who undertook important steps to reform the military organization of his predecessors. In 528 an entirely new command was introduced, that of the magister militum per Armenian, which controlled from now on the northern part of the eastern front. The office of the dux Armeniae was dissolved and the dux Mesopotamiae had to give up the upper part of his district 18

Josh. Styl. 64; cf. Proc. BP I.8.2. Proc. BP I.8.4. 20 This is proven by the considerable logistical difficulties the Romans had, especially in feeding their troops, v. Josh. Styl. 54; 77; Proc. BP I.8.5. 21 Josh. Styl. 93-94; Scharf 2001: 63-66. 22 Theoph. 117.2-9 de Boor . 23 Cf. Alston 1995: 74-86; see already Maspero 1912: 126; Jones 1964: 662-663. 24 Theoph. 141.9-17 de Boor. 25 Josh. Styl. 51. 26 CJ XII.35.18. 27 This is stressed, in my opinion correctly, by Whitby 1995: 113-114; for the problem of the limitanei and their fighting value see now Greatrex 1998: 34-35 with further reading. 19

412

OLIVER SCHMITT: FROM THE LATE ROMAN TO THE EARLY BYZANTINE ARMY to the newly created magister militum.28 The necessary forces for the new Armenian field army - apparently 15,000 strong in 53029 - were taken from both praesental armies and from the oriental army. According to John Malalas, our most detailed author, the army of the Oriens alone had to provide four regiments (¨riqmoj), which are not further specified by any source, the number of units taken from the praesental armies is not even given. To these troops were added the limitanei, who up to this time had been commanded by the dux Armeniae and some of the limitanei of the dux Mesopotamiae.30 We do not know if these units were considered as part of the field army, the comitatenses, or kept their status as limitanei. In fact, the differences between these categories of troops become increasingly unclear to us during the 6th century. Although the number of soldiers of the oriental army should have been reduced as a result of the aforementioned measures, the magister militum per Orientem Belisar in 530 had ca. 25,000 men under his command during the battle of Daras,31 which is more than double the size of his predecessor Areobindus’ army during the Persian war of Anastasius. But the main reason for this troop-multiplication is all too obivious: Belisar’s army consisted to a great part of the units provided by the eastern duces:32 with him were the soldiers of the ducatus Phoenice Libanensis under their dux Buzes33 and certainly those of the ducates Mesopotamia and Osrhoene.34 In the following year, during the campaign of Callinicum, the situation was similar: again Belisar’s army - at this time 20,000 strong - included to a large part units which belonged to the armies of the eastern duces.35 This heavy reliance on man-power borrowed from the duces had an unwanted consequence for the magister militum per Orientem: he now had to heed to a remarkable degree to the ideas and interests of these generals, a fact more than once lamented by Procopius.36 When the field army of 18,000 men was assembled for the expedition against the Vandals in 533, it consisted chiefly of units of the two eastern magistri militum, i.e. Armeniae and Orientis.37 Some troops may have been added from the remaining pool of the praesental armies.38

It was only consequent, that this force was not commanded by a magister militum praesentalis, but by Belisar, who held - after his dismissal in 531 - for a second time the office of magister militum per Orientem, his second in command being Dorotheos, magister militum per Armeniam.39 Two years later, when the war against the Ostrogoths in Italy began, it was a rather small force that sailed with Belisar to Sicily and the Italian mainland: apart from his own bucellarii40 and 500 Hunnic and Moorish sÚmmacoi there were three regiments of horse, four of foot and 3,000 Isaurians, 7,000 soldiers altogether, very probably units from one or both praesental armies.41 A subsidiary attack was undertaken by the troops of the magister militum per Illyricum Mundus but this force seems to have been rather small.42 We know all too well that things didn’t go according to the Roman emperors’ plans. The expected military leisure walk quickly emerged as a protracted war which was bitterly fought and which demanded considerable efforts on part of the Romans. And it is at this point that difficulties in providing the necessary troops started. Beginning with the setback of Mundus in Damaltia there is a lot of evidence for the growing difficulties of the Romans to assemble an army strong enough to vanquish the Goths and set an end to the war. Already Mundus’ successor Constantinus seems to have had some difficulties in bringing together an, as it comes out, not very large army, although he apparently brought some reinforcements from Byzantium with him.43 When the situation in Italy became critical in the winter of 536/3744 there seemed to have been no sufficiently strong forces at hand to come to the rescue. Instead, mostly rather weak scratch forces were sent in as reinforcements, slowly, one after the other, consisting to a large part of foreign allies and such units that could be spared elsewhere.45 Nearly one and a half year was needed to bring a rather strong army of ca. 7,000 men to Italy to restore the situation, 2,000 of them were Heruls, the rest seems to have belonged to the praesental supression of the Nika riot, see Chron. pasch. 622.15-623.6. For reasons unknown to us these units were not mobilized for the Vandal expedition. It may be that their fighting value was considered too low compared with the regiments from the eastern front, which had fought in the Persian war. 39 Proc. BV 1.11.5-18. 40 Schmitt 1994: 171 A. 208. 41 Proc. BG I.5.2-4. The commanders named by Procopius BG I.5.3) are not known to have taken part in operations on the eastern front before the Expedition into Italy startetd. 42 Proc. BG I.5.2; Mundus apparently attacked with only a part of the Illyrian army, cf. Proc. BG I.7.26-27. 43 Proc. BG I.7.26-27. Constantinus untdertook his expedition by the see, not by the land, as Mundus had done before (see Proc. BG I.7.28), he certainly brought his ships with him from Constantinople. 44 For the date see PLRE IIIA 198 s.v. Belisarius. 45 Proc. BG I.27.1-2: 1600 cavalrymen, mostly Huns, Slavs and Ants; BG I.5.1-2: 3,000 Isaurians, 1,800 cavalrymen, of whom 800 came from Thrace (it is unclear, if this means that they belonged to the Thracian army or if they belonged to praesental regiments stationed in the European vicinity of Constantinople) and 300 further horse of unknown origin. BG 2,13,17: a force of unknown but presumably minor strength under Aratius.

28

Joh. Mal. 358.7-359.17 Thurn; Proc. aed. III.1.1; vgl. I.17.24; cf. Greatrex 1998: 153-154. 29 Proc. BP I.15.11. 30 Joh. Mal. 358.10-359.18 Thurn. 31 Proc. BP I.13.23. 32 Procopius (BP I.13.19-21) tells us that some sÚmmacoi, for the most part Huns and Heruls, fought with the Roman army, but their numerical strength was not very large. Besides these allies there must have been a certain number of Isaurian units at Daras (cf. Proc. BP I.18.5) which probably did not exist during Anastasius’ time (cf. Proc. BP I.18.38-39. 33 For Buzes v. Proc. BP I.13.5. 19. 34 The participation of the dux Mesopotamiae can be reasonably concluded from the fact that Daras was his headquarter (Proc. BP I.22.3); the participation of troops from the Osrhoene is probable because the units from the Lebanon - by far more remote from the theatre of war - were called up. 35 Proc. BP I.18.5-6. 36 Proc. BP I.18.15-25; II.16.6-17; 18.5-26; 19.36-41; cf. I.8.1-5; II.24.13-17. 37 Proc. BV I.11.2-9; cf. Proc. BP I.13.19-21. 38 Cf. Proc. BV I.11.10. There was certainly a number of praesental regiments available, some of them had taken part in the bloody

413

THE LATE ROMAN ARMY IN THE EAST armies.46 It is evident that both armies of the magistri militum praesentales had undergone a process of considerable “milking” since 528. A rough estimation shows that probably far more than 20,000 men had been sent to various theatres of war, mostly to Armenia and Italy.47 What remains striking is the fact that never a praesental army with its magister was sent to war as a coherent force. Sittas, the only magister militum praesentalis whom we know in this period, stood inactively behind in Constantinople during the Gothic war.48

attacked again in 542.57 His vast force of bucellarii, numbering 7,000 cavalrymen, if Procopius can be trusted,58 seems to have been held back in Cilicia to the greater part, maybe as a kind of strategical reserve.59 In 543, after some additional reinforcements had been sent to the east, the strength of the Roman army assembled in Armenia ist given by Procopius with 30,000 men, this number including the regiments of both eastern magistri and of two Armenian duces.60 But the army of the east proved itself to be unfit for larger operations. Instead, its commanders confined themselves to secure some great key fortresses like Hierapolis and Edessa and to some minor raids. Structures of command had become increasingly unclear, recently arrived generals and local duces showed a tendency to act independently with their own troops61 and the resulting confusion was the main reason for the Roman defeat at Anglon in 543.62

The consequences of such a constant weakening of the central reserve forces proved to be disastrous. As the war with the Persian broke out again in 540, there was no army available to challenge the invading SƗsƗnids in open battle or to undertake a counterattack.49 Sittas as magister militum praesentalis and Justinian’s nephew Germanus in an extraordinary command, sent to the east to come to the rescue, came more or less as generals without soldiers.50 The first was killed all too soon in a minor engagement against rebellious Armenians,51 the other saw himself reduced to a helpless onlooker as the king of the Persians attacked in strength.52 The only force at hand was the army of Phoenice Libanensis, 6,000 strong, 53 and this was insufficient even to hold Antioch. In the next year the magister of the east, Belisarius reappeared on the scene, but the situation improved only a little. Certainly, he had brought some troops with him, his bucellarii, some Goths and apparently some units from Thrace54 - these may have been taken from the remaining pool of praesental units - , but again he had to rely heavily on the duces, especially those of Phoenice Libanensis.55 Even with his reinforced army Belisar did not dare to undertake a larger siege56 or to confront the Persian king head on when he

When the situation in Italy again became critical as a result of the Gothic uprising in the early 540s, the old policy of sending in numerically weak reinforcements piecemeal was again taken up.63 So low was the number of available troops, that Belisarius on his return to Italy had to stop in Thrace to undertake a recruiting tour there; the 4,000 men he finally brought together were mostly raw recruits.64 His large corps of private retainers mentioned above had been dissolved in the meantime, when Belisar temporary had lost his emperor’s favor.65 The composition of the large army finally sent to Italy under the command of Narses casts a gloomy light on the situation of the Roman armed forces in the year 552. It consisted to a large part of foreign allies - the Lombard contingent alone numbered 5,500 men - mixed up with what were apparently the last remnants of the praesental armies, bands of private retainers and even Persian prisoners of war.66 Even after the Lombards were sent home the army of Italy remained a rather impressive force numerically; in 555 Narses was able to muster a field force of 18,000 men against the Frankish invaders.67 In the east, emphasis had shifted from Mesopotamia and Armenia to Lazica far in the north from the late 540s onwards and a considerable number of troops was assembled there.68 Commander-in-chief was the magister militum per Armeniam by name, but apparently he was

46

Proc. BG I.13.15-18. Armenia: ca. 10,000 men (15,000 men minus four regiments of the eastern army and certainly some local troops, v. Proc. BP I.15.11; Joh. Mal.. 359.14-18 Thurn); Italy: 10,000 men at last (4,000 men with Belisar; 5,000 with Narses and more than 1,000 other troops, v. Proc. BG I.5.2; II.5.1-2; 13.17-18). Some further units may have been sent to Africa, v. Proc. II.16.1; 19.1. 48 Cf. Proc. BP II.3.8. During the first Persian war of Justinian’s reign, Sittas had acted as commander-in-chief in Armenia and for a short time off the whole east, see Proc. BP I.15.3; 21.3. 49 A number of Vandal prisoners of war had been shipped to the east as reinforcements after 533, but these so-called Vandali Iustiniani remain somewhat of a mystery. Procopius (BV II.14.17) tells us that they were divided between five regiments (kat£logoi) of cavalry but it is in no way clear that these units were newly constituted, as Scharf 2001: 70 note 196 believes. It seems to me far more probable that they were divided between existing regiments (cf. Proc. BV II.14.18) because such a solution is a more practicable one. 50 Iustinian had given Germanus the promise to send him an army, if Procopius (II.6.9; 14-15) is trustworthy here, but this force never did materialize. 51 Proc. BP II.3.8-27. 52 Proc. II.6.9-15; cf. Joh. Mal. 405.66-69 Thurn. The 300 men he brought with him (Proc. II.6.9) were apparently his personal bodyguard. 53 Proc. BP II.8.2. 54 For the reinforcements which Belisar brought with him see Proc. BP II.18.24; 19.15; 32. 55 Proc. BP II.16.16-19; 19.33-34. 56 Proc. BP II.19.1; cf. 18.3-4 47

57

Proc. BP II.21.2-8. Proc. BG III.1.20. For these troops see Schmitt 1994: 171. Proc. an. III.5. 60 Proc. BP II.24.13-17. 61 Proc. BP II.24.17-21. 62 For the defeat at Anglon see now Trombley 2005: 404-411. 63 Proc. BG III.6.9-10; 10.3; 13.20-22; 18.1. 64 Proc. BG III.10.1-3. 65 Most of the bucellarii were given to other officers, v. Proc. an. IV.13, but some of them accompanied Belisar when he returned to Italy, s. Schmitt 1994: 171 note 210. 66 Proc. BG IV,26.8-13. The overall strength of Narses’ army is not given by Procopius. 67 Agath. II.4.10; garrison troops are not included in this number. 68 Procopius (BG IV.13.8) gives the army in Lazica 12,000 soldiers in 552. It may have received further reinforcements in the following years, cf. Agath. II.18.8. 58 59

414

OLIVER SCHMITT: FROM THE LATE ROMAN TO THE EARLY BYZANTINE ARMY only a primus inter pares among other generals.69 The command of the magister militum per Orientem temporally ceased to exist. Belisarius, who hold this post for a third time, was detained in Constantinople since 549 out of suspicion by the emperor, but nevertheless remained in possession of the title.70

there in the next year, presumably brought to strength with units from the Armenian theatre,79 but when this force was routed at Nisibis, a new army had to be built up, which consisted to a large part of allied levies.80 These forces succeeded in restoring the situation in the east, but nevertheless the war was fought for further fifteen years without decision, until finally internal strife in the SƗsƗnid empire turned the overall situation in favor of the Romans. During the war the Roman commanderin-chief mostly seems to have held the title of magister militum per Orientem,81 but the Armenian command did not cease to exist,82 at last by the name. A new command of the magister militum per Lazicam seems to have been constituted,83 it’s troops certainly being taken from the pool of the other two eastern magistri, but we do not know how long this new command existed.

On the Balkans, there was a field army of modest size in Thrace and an even smaller one in Illyricum.71 For the army of Africa, which had received some minor reinforcements during the 540’s72 and for the army of Spain no reliable numbers are given, but they must have been rather small. The praesental armies had vanished from the scene and with them the office of magister militum praesentalis. Only a few units of the scholae, the imperial guards, having lost most of their fighting value because of their long absence from any war, remained in the vicinity of Constantinople. They proved themselves quite useless during the hunnic attack of 55973 and the already aged Belisar had to fight the invaders with a scratch force of some veterans, private retainers and hastily equipped civilians.74 It is evident that the old fifthcentury system of a strong central force and up to three smaller regional field armies was given up during Justinian’s reign. Agathias’ often cited statement that Justinian reduced the Roman armed forces from 645,000 to some 150,000 men during his reign is, of course, wildly exaggerated,75 but it reflects to a certain degree the impression which a later generation got from the results of this emperor’s agressive foreign policy.

Man-power left on the Balkans was insufficent to deal with the growing Avar threat effectively. Supported by Slavic warbands the Avar khagan Baian occupied Sirmium after a protracted siege in 582; it had not been possible to help the beleaguered city with an army of sufficient strength because nearly all the Roman troops were engaged in the war against the Persians.84 In the following years the Avars and Slavs undertook devastating raids into the Balkan provinces,85 although a Roman army of ca. 10,000 soldiers was scratched together after all under the command of Comentiolus. He was given the titel of magister militum praesentalis,86 but his successor Priscus only received the rank of magister militum per Thracias instead.87 It was not before the conflict with the Persians was brought to a successful end that units from the east could be transferred to Europe to fight the Avars with some chance of victory.88

Under Justinian’s successors, things turned from bad to worse. When the war with the SƗsƗnids broke out again it was only with difficulties that the Romans were able to gather a field army. Establishments of units and equipment were obviously at low ebb.76 An army was sent in support of the rebellious Armenians, which had some successes,77 but on the Mesopotamian front the Romans could muster at first no more than 3,000 soldiers according to one source.78 A far larger army was gathered

It is obvious that the five old commands of the magistri had degenerated to mere titles at the end of the sixth century. Two important commands were left, however: Europe and the Orient; in addition there were three more or less isolated local commands: Italy, Africa and Spain, which could not exspect much help from the imperial

69

Cf. Proc. BG IV.9.4; Agath. II.18.8; III.2.7-8; IV.21.1. PLRE IIIA 216-217 s.v. Belisarius 1. 71 It is true that in BG III.29.3 Procopius gives the Illyrian field army alone a strength of 15,000 soldiers, but that is very probably an exaggeration. Cf. Proc. BG III.38.1-3. 5 for a more realistic impression of the available manpower. 72 Proc. BV II.24.1; Coripp. Ioh. I.125-128; the troops consisted to part of new levies (Coripp. Ioh. I.128). 73 Agath. V.15.1-6. 74 Agath. V.16.$1-4; Theoph. 233.20-23 de Boor. 75 V. Agath. V.13.7. A number of scholars has attempted to prove the overall reliability of Agathias’ numbers; see, for example, Jones 1964: 679-684; Treadgold 1995: 43-64, but their arguments are, in my opinion, in no way convincing. 76 Euagr. HE V.8. 77 Sebeos 8 (7 tr. Thomson); According to this source the Romans did not only capture Dvin, but were able to defeat a Persian army of 20,000 men in battle. 78 Joh. Epiph. fr. 3 Müller. It seems quite improbable to me that the Roman army sent into Armenia in 572 was much stronger than the 20,000 men the Persians could muster there in the same year, if Sebeos 8 (7 tr. Thomson) can be trusted. The overall strength of the Roman field army in the East therefore seems not to have exceeded 25,000 70

soldiers at the most, which is considerably less than the 40,000 men in 530 and even the 30,000 in 543. Euagrius’ verdict (HE V.8) that the Romans were in no way ready for war in 572 is certainly somewhat exaggerated, but it cannot be bluntly refused as Stein 1919: 40 does. 79 When the Persian undertook a new expedition into Armenia in 573, the Roman commander-in-chief Johannes had to rely primarly on Armenian, Kolchian, Abasgian and Alan allies (Theoph. Byz. fr. 4 Müller). 80 Evagr. HE V.14. For the war against Persia cf. Stein 1919: 43-46. 81 Theoph. Byz. fr. 4 Müller; Euagr. HE V.8 (Marcianus); Joh. Bicl. s.a. 575.1 (Iustinianus); Theoph. Sim I.9.4 (Ioannes Mystacon); Theoph. 253.26 de Boor (Philippicus); Theoph. Sim. V.8.1 (Narses). 82 Theoph. Sim. I.9.4. 83 Theoph. Sim III.7.8; PLRE IIIB 1091 s.v. Romanus 4. 84 For the military situation of the empire see Men. Prot. fr. 25.2.13-17 Blockley; for the loss of Sirmium v. Pohl 1988: 70-76. 85 Pohl 1988: 76-89. 86 For the Roman troops see Theoph. Sim. II.10.8-9, who writes that only 6,000 men were considered fit for battle by Comentiolus; for his rank see. Theoph. Sim. I.7.4. 87 Theoph. Sim. VI.5.13. 88 See Whitby 1988: 156-165; Pohl 1988: 135-147; 152-162.

415

THE LATE ROMAN ARMY IN THE EAST per orientem, acted the emperor’s brother Theodor first.97 After his dismissal in 634 he apparently was replaced by the imperial sakellarios Theodor, surnamed Trithyrios.98 With him were a certain Baanes and Nicetas, the son of the Persian general and usurpator Sharbaraz, but their titles and exact functions are unknown to us.99 In the west, there remained an independent command, which was held by Bonus in 626 during the Avar attack on Constantinople,100 but it did possess only a few troops of it’s own.101

government. But what counts more is the fact that the empire had now lost it’s ability to fight two great wars at the same time. During the reign of Phocas the situation deteriorated even further. Neither the east nor the Balkans could be successfully defended against the Persians, Avars and Slavs.89 The last resources of the Oriens were squandered in the civil war between Phocas and Heraclius, which ruined to a great part the local troops of Egypt, Syria and Palestine.90 After the remnants of the field army had been beaten near Antioch in 613, the SƗsƗnids were free to ravage and conquest the eastern provinces at will. Ten years were needed to rebuilt an army which was sufficiently strong to undertake a counterattack against the Persians,91 and when the Avars undertook their offensive against Constantinople in the meantime, most of the troops had to be brought back to the imperial city, because no other force was available.92

It should have become sufficiently clear by now that the new Byzantine structures, known as the theme system, cannot be judged as a mere continuation of the late Roman army organization. There were only two large commands left since the end of the sixth century - Europe and East - and only one large army. The offices of the five magistri had become mere titles, apart from the office of the magister militum per Orientem, who often, but not always, acted as commander-in-chief of the East as a whole.

It is true that Heraclius was able to defend the Avars and brought the Persian war to a victorious end. But this victory bore in itself the seed of a disaster yet to come. The units of the one and only existing Roman field army were to a great part needed to provide garrisons for the reoccupied provinces, whose defensive organization rested upon some hastily reconstituted ducates. There were scarcely any troops left to cope with even minor setbacks and this proved to be fatal when the Muslim Arab invasion began.93 With great efforts a strong Roman field army was scratched from the bottom to fight the Muslims, but after it was thoroughly defeated at the battle of al-Yarmnjk in 636 there were no forces left to defend the Near Eastern provinces.94 A new defence system was needed to secure at least Asia Minor.

The creation of the four themes, stationed more or less exclusively in Asia minor meant the dissolution of the independent European command and the splitting of the Eastern command into four. It is well possible that some staff personnel of the old offices was still in existence and could be used to built up the thematic armies of Thrakesion, Anatolikon and Armeniakon, but the regiments for these armies had to be newly assigned. The command and the army of the Opsikion theme had to be more or less a new establishment, a skeleton staff and some units must have been provided by the old magister officiorum, who had had under his command the regiments of the palace guard.102 Finally, the step that the strategos of a theme was given not only military authority but civilian too followed the example of the reconstituted ducates of the oriens after 628.103 The themes themselves had much in common with the former ducates and in fact the theme can be seen as a kind of mega-ducate.

As far as command structures are concerned, Heraclius in person was in command of his forces for the most of the time between 613 and 628.95 That means that the magistri more or less lost their importance during this period. That notwithstanding, there is evidence enough to show that two independent commands did exist during Heraclius’ reign, one in the east and one in the west.96 As commander-in-chief, very probably as magister militum

2. Let us now look at the structural changes in late Roman cavalry organization during the sixth century

89 For the East see Stratos 1968; Kaegi 2003: 39-40; 65-68; for the Balkans see Pohl 1988: 237-248. 90 Stratos 1968: 82-87; Kaegi 2003: 44-57; cf. Howard-Johnston 1999: 14-16; 32-36. 91 Kaegi 2003: 73-114; for the the beginning of the counteroffensive cf. Speck 1988: 101-107; the strength of Heraclius’ army is discussed by Schmitt 2001: 219 note 130; Howard-Johnston 1999: 31-32 and Syvänne 2004: 88. note 4. Syvänne’s estimation seems to me vastly exaggerated. 92 Theoph. 315.11-13 de Boor; cf. Chron. Pasch. 178.20-22 Dindorf. 93 For details see Schmitt 2001: 216-229. 94 Kaegi 1992: 112-146. 95 He was not in command during the Avar siege of Constantinople, v. Chron Pasch. 178.4-11 Dindorf. For the emperor on campaign, especially the risks and inherent disadvantages, see now Whitby 2005: 368-375. 96 Cf. Haldon 1984. 172.

97

Chron. 1234.110. Nic. Brev. 20.9-10 Mango. 99 For Baanes cf. Theoph. 337.3-5 de Boor; Chron. 1234.110. For Nicetas cf. Mich. Syr. 11.6; Chron. 1234.110. 100 Chron. Pasch. 178.10 Dindorf. 101 The number of the defenders of Constantinople in 626 is reasonably estimated at 12000 soldiers at last (cf. Chron. Pasch. 178.21-22 Dindorf; Pohl 1988: 249; Kaegi 2003: 134) of whom by far the larger part seems to have been detached from the main army under Heraclius. 102 Cf. Haldon 1984: 164-182. The corps of the Bucellarii and Optimati may quite well have been attached to the Opsikion theme, because the emperor wished to have these not very strong, but well equipped troops near his capital , see Stein 1919: 133; Treadgold 1995: 99; the Foederati apparently became part of the army of Anatolikon, see Scharf 2001: 133. 103 Schmitt 2001: 213-216. 98

416

OLIVER SCHMITT: FROM THE LATE ROMAN TO THE EARLY BYZANTINE ARMY Although the importance of the mounted soldier steadily increased since the fourth century,104 the organization of the mounted regiments remained more or less the same at least until the middle of the sixth century. The regiment vexillatio or simply numerus in Latin, t£gma, b£ndon or ¨riqmoj in Greek - was the basic unit up to 400 men strong.105 In practice, establishments were often remarkably lower in number, especially with units who had served some time in the field without having received reinforcements.106 No earlier than in the Strategikon of the emperor Maurikios, written at the end of the 6th or in the early seventh century, do some changes in the organization of the cavalry force become clearly visible. The regiment was now assigned to larger units, the moirai and meroi, often compared with brigades and divisions. Unfortunately, it is somewhat difficult to pick up detailed evidence for such a type of cavalry force from other sources. Therefore we have to a great part to rely on the Strategikon itself to get a coherent idea of what did happen to late Roman cavalry organization.

Like the regiments and different from the brigades the cavalry divisions seem to have had proper names: there was a division of the Vexillationes, of the Foederati, the Illyriciani, the Optimati, the Bucellarii.113 The names Vexillationes, Foederati, Optimati and Bucellarii all characterize different categories of troops. The division of the Vexillationes must have consisted of the old regular cavalry regiments (vexillationes) of the late Roman field army, Foederati were mixed units of Romans and Barbarians and with the Optimati and Bucellarii it must have been likewise.114 But there were further, more important differences. In the regiments of the Foederati and especially in those of the Optimati the soldiers possessed armed retainers, who were not simply grooms or camp-followers, but were given a supporting role in battle.115 It is very probable that the soldiers of the Foederati and the Optimati received an additional pay for their retainers from the state.116 That means that a regiment of foederati and of optimati was more expensive than a regular cavalry regiment. With the bucellarii it must have been similar. The Strategikon makes it clear that the bucellarii were especially well equipped soldiers, whose armament and equipment must have been much more expensive than that of other troops117 and who presumably were better paid. It is tempting to conclude, that the assignment of regiments to different divisions above all followed financial aspects, but there remains the problem of the Illyriciani. Did these units belong to a special category defined in terms of pay? I think, the answer is yes. Cavalry units surnamed Illyriciani are found in the Notitia exclusively in the armies of the eastern duces and therefore were part of the limitanei.118 With the growing importance of the ducatal armies for the defence of the eastern provinces these regiments must have assumed a role comparable to that of field army units.119 That notwithstanding, it seems doubtful if these former border regiments were given a pay as high as that of field army units. Evidence for special qualifications or special equipment of the Illyriciani, which would have justified a higher pay, is lacking120 and so we can presume that they received lesser money than their comrades serving in regiments of the other categories.

In the Strategikon, the following picture is given: the largest cavalry formation is the meros or, in modern terms, division, commanded by a merarches. A meros is divided into three moirai (or brigades) commanded by a moirarches and a moira is subdivided into several tagmata (regiments) commanded by tribunes respectively.107 No explicit number is given for the regiments regularly assigned to a brigade or a division, only the rule is written down that a moira should not exceed 3,000 men and a meros 7,000.108 According to the Strategikon, a cavalry division could consist of 14 regiments at least,109 and in one source even a 15th regiment of the Illyriciani-division is mentioned.110 On the average, a regiment is given about 300 horsemen in the Strategikon,111 so a single division of fifteen regiments must have numbered 4,500 soldiers, a strength well within the aforementioned limits. So far all seems clear, but now the problems begin. It is somewhat difficult to believe that a division actually went to battle in a strength of 4,000 cavalrymen or more. Elsewhere, the author of the Strategikon writes that an army of 5,000 men or less should be divided into four divisions (three in the first line of battle, one in the second).112 It is obvious that the actual number of regiments in a division could vary considerably.

The presumption that the divisional names of Illyriciani, Vexillationes, Foederati, Optimati and Bucellarii are reflecting financial and administrative but not tactical structures can be supported by further evidence. First of

104

The role and importance of the cavalry in late Roman armies is open to some debate. For two different viewpoints see Trombley 2005: 391 and note 21, who emphasizes the importance of the cavalry and gives only a minor, more or less purely defensive role to the infantry and Whitby 2005: 360-361 who - in my opinion rightly - concludes that the infantry was not reduced to defensive tasks only. For a similar view see now Syvänne 2004: 277-281. 105 Schmitt 2001 (2004): 97 note 47 with sources. 106 Cf. Maur. Strat. I.4.35; II.20.18-19. 107 Maur. Strat. I.4.3-26. 108 A meros of the first line should consist of three moirai according to Maur. Strat. II.2.10-12. 109 Maur. Strat. III.8.6. 110 Miracula Anastasii Persae 14.1 Flusin. 111 Cf. Maur. Strat. III.2.1-2: 310 men. 112 Maur. Strat. II.4.28-33; cf. II.10-14.

113

Maur. Strat. I.2.10; 3.23-24; III.8.4. On the Optimati, Bucellarii and Foederati cf. Haldon 1984: 96-102; I have argued elsewhere that the proportion of foreigners in the Bucellarii should not be overestimated, Schmitt 1994: 163-164. 115 Maur. Strat. II.6.29-33; 46-48; III.6.7; 7.7. The custom of fighting supported by armed retainers reflects Germanic influence, cf. Caes. Gall. I.48.5-7; Tac. Germ. 13.2-14.2; Proc. BP II.25.27-28; BG IV.26.12; Schmitt 1994: 153-154. 116 For the so-called familiarikÒn see Maur. Strat. I.2.62-64. 117 Maur. Strat. I.2.8-10; 22-27. 118 ND Or. XXII.18-21; XXXIII.16-17; 25-26; XXXIV.18-22; XXXV.15-17; XXXVI.19-22; XXXVII.14-17. 119 Cf. Proc. BP II.8.2; 16.17. 120 Cf. Maur. Strat. I.2.8-21. 114

417

THE LATE ROMAN ARMY IN THE EAST militum.126 Another example may be found in Theophylactus’ report of the Persian campaign in the year 591. According to this source, the general Komentiolos was in command of the right wing of the Roman army and held under his command a cavalry force numbering 1,000 men.127

all, in the Strategikon no further divisional names are mentioned, even if an army consisted of no less than seven divisions; four of them are simply called meros without any further specification, the remaining three being the Vexillationes, the Foederati and the Illyriciani.121 Out of other sources it becomes clear that the regiment remained the basic unit of the late Roman cavalry and that it was used indepentently from larger formations in military operations. The already mentioned 15th regiment of the Illyriciani apparently was a independent garrison unit in Palestine after the reoccupation in 628/29 and not only a part of it’s division122 and the same must have been true with other units like the Scythae and the Voluntarii,123 probably regiments of the Vexillationes or Foederati divisions.

Conclusion The road from the late Roman field army organization to the Byzantine thematic system did not run straightforward, but was considerably twisted. The old system of five field armies had been given up since Justinians’ reign and at the end of the sixth century, there were only two great commands left, Europe and East, with only one large army available. The introduction of the thematic system meant a considerable rift in the army organization. Not only had the regiments to a large part to be newly assigned, what counts more is that he European command was dissolved completely and the Eastern command was split into four sub-units which had more in common with the former ducates than with the old commands of the magistri.

The question why this system of unit-administration was introduced is not easy to answer. Even the point of time is unknown to us, but there are no hints in the sources that it had been established during Justinian’s lifetime. The cause for it’s introduction probably lay with the financial misery of the empire in the years following Justinian’s death, when the coffers of the state were nearly empty to the bottom. In such a situation the idea to tighten the financial administration of the army by making the payments to the troops clearly - and tightly - arranged made some sense.124

With regard to the cavalry divisions of the late sixth century, there existed a difference between the divisions as tactical formations and as administrative units. The former were constituted ad hoc depending on the situation and the number of regiments available, the latter had no tactical role at all but characterized five different categories of troops: Bucellarii, Optimati, Foederati, Vexillationes and Illyriciani with different rates of pay instead.

However, if it is true that the divisions were purely administrative units what is the matter with the moirai, the brigades? And what sense would it make to introduce merarches and moirarches, who had a distinctive tactical role? The problem can be solved easily if we make a strict difference between the division as an administrative unit and the division as a tactical formation and it must be emphasized that these two were not identical. As tactical units, divisions and brigades were certainly ad-hoc formations, which bore no name of their own. To these ad-hoc formations varying numbers of regiments could be assigned, according to the situation and the forces available.125 It was their function to make large, compact groups of cavalry easier to handle by shortening the chain of command. An example is given by Agathias for the battle of Casilinum. In this engagement a apparently rather strong group of cavalry on the Roman left wing was placed under two commanders only, Valerianus and Artabanus, who both held the rank of magister

Bibliography Alston R. 1995 Soldier and Society in Roman Egypt. A Social History, London-New York. Carrié J.-M. 1995 L’État à la recherche de noveaux modes de financement des armées (Rome et Byzance, IVe-VIIIe siècles). In Cameron Averil (ed.), The Byzantine and Early Islamic Near East III. States, Resources and Armies. Studies in Late Antiquity and Early Islam, 1, Princeton: 27-60. Cesa M. 1994 Impero tardoantico e barbari. La crisi militare da Adrianopoli al 418, Como. Durliat J. 1993 Armée et société vers 600. Le problème des soldes. In Vallet Fr. – Kazanski M. (ed.), L’armée romaine et les barbares du IIIe au VIIe siècle. Mémoires publiées par l’Association Francaise d’Archéologie Mérovingienne, 5, Paris: 31-38. Greatrex G. 1998 Rome and Persia at War, 502-532. Arca, 37, Leeds. Haldon J. F. 1984 Byzantine Praetorians. An Administrative, Institutional and Social Survey of the

121

Maur. Strat. III.8.4, 15. Miracula Anastasii Persae 14.1-4 Flusin. 123 Passio Sanctorum Sexaginta Martyrum 303.14, 19-20 Delahaye. 124 Problems in the financing of late Roman armies are discussed by Durliat 1993: 32-33; and Carrié 1995,: esp. 39-44; 50-59, but I do not agree with the latter’s conclusions. There’s a hint for enormous difficulties the state had in paying his soldiers in the last quarter of the sixth century. The great army sent in the east in 574/575 had it’s own financial supervisor , a high-ranking official with special authorities, v. Joh. Eph. HE VI.14; Kaegi 1982: 105-106. 125 Trombley 2005: 391-392. 122

126 127

418

Agath. II.8.3. Theoph. Sim. V.8.1.

OLIVER SCHMITT: FROM THE LATE ROMAN TO THE EARLY BYZANTINE ARMY Treadgold W. 1995 Byzantium and Its Army 284-1081, Stanford. Treadgold W. 1997 A History of the Byzantine State and Society, Stanford. Trombley F. 2005 The Late Roman Practice of War on the Syrian Frontier (A.D. 502-641): Leadership, Infrastructure and Operations. In Meißner B. – Schmitt O. and Sommer M. (ed.), Krieg - Gesellschaft - Institutionen. Beiträge zu einer vergleichenden Kriegsgeschichte, Berlin: 387-416. Whitby M. 1988 The Emperor Maurice and his Historian. Theophylact Simocatta on Persian and Balkan Warfare, Oxford. Whitby M. 1995 Recruitment in Roman Armies from Justinian to Heraclius (ca. 565-615). In Cameron Averil (ed.), The Byzantine and Early Islamic Near East III. States, Resources and Armies. Studies in Late Antiquity and Early Islam, 1, Princeton, 61-124. Whitby M. 2005 War and State in Late Antiquity: some economic and political connections. In Meißner B. – Schmitt O. and Sommer M. (ed.), Krieg - Gesellschaft - Institutionen. Beiträge zu einer vergleichenden Kriegsgeschichte, Berlin: 355-385. Whittow M. 1996 The Making of Byzantium 600-1025, Berkeley.

Opsikion and the Tagmata, c. 580- 900. ȆȅǿȀǿȁǹ ǺȊǽǹȃȉǿȃǹ 3, Bonn. Haldon J. 1993 Military Services, Military Lands and the Status of Soldiers: Current problems and Interpretations. In DOP 47: 1-67. Haldon J. 1999 Warfare, State and Society in the Byzantine World 565-1204, London. Heather P. 1991 Goths and Romans 332-489, Oxford. Howard-Johnston J.-H. 1999 Heraclius’ Persian Campaigns and the Revival of the East Roman Empire, 622-630. In War in History 6: 1-44. Jones A.H.M. 1964 The Later Roman Empire 284-602, Oxford. Kaegi W.E. 1982 Two Studies in the Continuity of Late Roman and Byzantine Military Institutions. In Byzantinische Forschungen 8: 87-113. Kaegi W.E. 1992 Byzantium and the Early Islamic Conquests, Cambridge. Kaegi W.E. 2003 Heraclius. Emperor of Byzantium, Cambridge. Karayannopoulos J. 1959 Die Entstehung der byzantinischen Themenordnung. Byzantinisches Archiv, 10, München. Lilie R.-J. 1984 Die zweihundertjährige Reform. Zu den Anfängen der Themenorganisation im 7. und 8. Jahrhundert. In Byzantinoslavica 45: 27-39, 190-201. Lilie R.-J. 2003 Byzanz. Das zweite Rom, Berlin. Maspero J. 1912 L’organisation militaire de l’Égypte byzantine, Paris. Ostrogorsky G., 1963 Geschichte des byzantinischen Staates. Handbuch der Altertumswissenschaft, XII.1.2. 3. Auflage, München. Pohl W. 1988 Die Awaren. Ein Steppenvolk in Mitteleuropa 567-822 n.Chr., München. Ravegnani G. 1988 Soldati di Bisanzio in età giustinianea. Materiali e Ricerche, N.S. 6, Roma. Scharf R. 2001 Foederati. Von der völkerrechtlichen Kategorie zur byzantinischen Truppengattung. Tyche Supplementband, 4, Wien. Schmitt O. 1994 Die Buccellarii. Ein Beitrag zum militärischen Gefolgschaftswesen in der Spätantike. In Tyche 9: 147-174. Schmitt O. 2001 Untersuchungen zur Organisation und zur militärischen Stärke oströmischer Herrschaft im Vorderen Orient zwischen 628 und 633. In BZ 94: 197-229. Schmitt O. 2001 (2004) Stärke, Struktur und Genese des comitatensischen Infanterienumerus. In Bonner Jahrbücher 201: 93-111. Speck P. 1988 Das geteilte Dossier. Beobachtungen zu den Nachrichten über die Regierung des Kaisers Herakleios und die seiner Söhne bei Theophanes und Nikephoros. ȆȅǿȀǿȁǹ ǺȊǽǹȃȉǿȃǹ 9, Bonn. Stein E. 1919 Studien zur Geschichte des byzantinischen Reiches vornehmlich unter der Kaisern Justinus II. u. Tiberius Constantinus, Stuttgart. Stratos A.N. 1968 Byzantium in the Seventh Century I, 602-634, Amsterdam. Syvänne I. 2004 The Age of Hippotoxotai. Acta Universitatis Tamperensis 994, Tampere. 419

L’occupation militaire tardive dans les oasis d’Égypte. L’exemple de Douch M. Reddé École pratique des Hautes Études, Paris

Il ne me paraît pas inutile, dans le cadre d’un colloque consacré à l’armée romaine au Proche-Orient durant l’Antiquité tardive, de faire le point sur une question dont j’ai traité ailleurs, mais dans des publications dispersées et parfois anciennes (Reddé 1992; 1999; 2004). La nature très hétérogène de la documentation, à la fois archéologique et papyrologique, ne facilite guère, au demeurant, la maîtrise d’une bibliographie complexe que même les spécialistes de l’armée ignorent quelquefois. Une courte synthèse ne sera donc pas superflue.

constitue une exception notable plutôt que la règle (Sidebotham 1994). À l’ouest du Nil, la situation est très différente. G. Wagner (Wagner 1987) avait, il est vrai, prétendu le contraire, et publié pour l’oasis de Khargeh une carte qui présentait un réseau très dense de sites militaires (fig. 1). J’avais, dès 1989, donné un compte rendu assez critique de cette thèse dans le Journal of Roman Archaeology (Reddé 1989), insistant sur la confusion instaurée par le caractère extrêmement massif des constructions en briques crues qui sont la norme dans les oasis et donnent rapidement aux bâtiments, même civils, une allure de «fort». Les prospections de l’oasis, menées à l’occasion des fouilles de Douch, entre 1985 et 1990, m’ont au contraire convaincu que le nombre des fortifications archéologiquement attestées est réduit; j’en ai donné une liste, avec des plans et une description sommaire, dans le Bulletin de l’Institut français d’archéologie orientale (Reddé 1999). Bien qu’ils n’aient pas été fouillés, ces postes peuvent être a priori attribués à l’antiquité tardive, en raison de leur typologie; l’exemple du désert oriental montre toutefois combien la prudence doit désormais s’imposer dans ce genre d’évaluation, puisque des fortins que leur plan ferait normalement attribuer au quatrième siècle, s’avèrent désormais … flaviens (Reddé 2003 b)! Dans certains cas, en particulier à El-Gib et Someira (fig. 2), le doute est permis; ailleurs, comme à Aïn Labakha, une tour plus ancienne est insérée dans le bâtiment existant (fig. 3 et 4) et une fouille serait nécessaire pour en proposer une chronologie fiable. La tour de Taoun el-Hawa n’est pas datée en soi, et elle est proche de dépotoirs céramiques attribués à l’époque du principat (fig. 5). Seul donc le fort d’Ed Deir paraît très certainement attribuable à l’antiquité tardive (fig. 6 et 7). Sa taille (un carré de 74/75 m. de côté), ses grosses tours d’angle massives en font sans doute le lieu de garnison d’une unité assez importante, qui n’est malheureusement pas identifiée par des textes. On songe assez spontanément à l’ala I Abasgorum de la Notitia Dignitatum (Or. XXXI.55), bien qu’Ed Deir soit situé à une vingtaine de kilomètres d’Hibis, capitale de l’oasis, mais où l’on n’a pas trouvé, pour l’instant, de trace d’une fortification antique. Ces différentes observations étant faites, ma carte des installations militaires est infiniment plus réduite que celle de G. Wagner (fig. 8); elle montre en outre que seuls les abords septentrionaux de l’oasis sont surveillés ou défendus, soit au débouché de la piste venant d’Assiout, depuis le nord, soit à celui de la piste d’Esna/Louqsor, au nord-est de Khargeh. Mais qu’en estil de Douch (Kysis), autrefois considéré par G. Wagner comme une fortification du «limes d’Égypte», au sud de l’oasis, en raison à la fois de son enceinte, très imposante, et des très nombreux ostraca militaires tardifs qu’avaient

Commençons par un bref rappel des conditions écologiques qui prévalent dans cette région, isolée au cœur d’un désert absolu, à l’ouest du Nil. À la différence du relief très disséqué de la chaîne arabique, à l’est du fleuve, les oasis sont le fruit d’une abrasion éolienne en creux au sein d’un haut plateau calcaire absolument aride, d’où l’eau — et donc la végétation — est totalement absente. La fertilité des oasis est pourtant réelle parce que les surfaces cultivables sont proches, dans ces dépressions, de la nappe phréatique. L’homme a su en outre, à partir de l’époque perse, exploiter les petits aquifères perchés dans les buttes de grès, grâce à des systèmes de galeries drainantes ou qanat (Bousquet 1996). Tout oppose donc les deux déserts qui encadrent le Nil: à l’est un milieu ouvert, assurément aride, mais où les précipitations d’automne et de printemps entretiennent dans les fonds de vallées une végétation herbacée qui autorise le pastoralisme (Reddé 2003 a); à l’ouest, des régions agricoles fertiles, mais complètement isolées du reste de l’Égypte, et par là même très bien défendues naturellement, car les nomades ne s’aventurent pas volontiers sur les plateaux hyperarides qui les entourent. Cette particularité géographique a d’ailleurs fait des oasis des lieux de relégation politique, notamment pendant les persécutions antichrétiennes (Wagner 1987). Elle explique aussi que l’occupation militaire romaine soit très contrastée de part et d’autre de la vallée. L’intérêt économique du désert oriental repose sur deux éléments très différents: l’exploitation des ressources minérales, d’une part, la relation avec les ports de la Mer Rouge, portes du commerce arabo-indien, d’autre part. Dès le règne d’Auguste, l’armée romaine a donc cherché à contrôler ces régions, accroissant sa mainmise par un réseau de praesidia fortifiés à partir de l’époque flavienne (De Romanis 1996; Cuvigny 2003).1 En revanche, pendant l’Antiquité tardive, la présence militaire y est très mal attestée, et le cas de la forteresse d’Abu Sha’ar 1 La bibliographie sur cette question est désormais considérable, et il n’est pas question de vouloir la citer ici in extenso. On la trouvera dans les deux ouvrages cités en référence.

421

THE LATE ROMAN ARMY IN THE EAST

Fig 1 Carte des fortifications de l’oasis de Khargeh (Wagner 1987: 388, 389)

422

M. REDDÉ: L’OCCUPATION MILITAIRE TARDIVE DANS LES OASIS D’ÉGYPTE. L’EXEMPLE DE DOUCH

Fig. 2 Le fort d’El Gib (cliché M. Reddé)

Fig. 3 Le fort d’Aïn Labakha (cliché M. Reddé)

423

THE LATE ROMAN ARMY IN THE EAST

Fig. 4 Plan du fort d’Aïn Labakha (plan Chr. Braun et P. Deleuze, IFAO, d’après Reddé 1999)

Fig. 5 La tour de Taoun el Hawa (cliché M. Reddé) 424

M. REDDÉ: L’OCCUPATION MILITAIRE TARDIVE DANS LES OASIS D’ÉGYPTE. L’EXEMPLE DE DOUCH

Fig. 6 Plan du fort d’Ed Deir (plan Chr. Braun et P. Deleuze, IFAO, d’après Reddé 1999)

Fig. 7 Le fort d’Ed Deir (cliché M. Reddé)

425

THE LATE ROMAN ARMY IN THE EAST

Fig. 8 Carte des fortins de l’oasis de Khargeh (Reddé 1999) premiers états remontent à la fin de l’époque perse ou au début de l’époque hellénistique, si l’on en croit les datations radiocarbones effectuées sur la paille des briques. Ce complexe religieux est entouré d’un village, dont les vestiges les plus anciens remontent, dans l’état actuel des connaissances, à la fin de l’époque hellénistique ou au début de l’époque romaine, mais des niveaux antérieurs ne sont pas exclus dans des parties du site encore inexplorées. L’habitat semble succéder, en effet, à un autre village et à un autre sanctuaire installés sur le tell voisin d’Aïn Manawir, que les fouilles en cours de l’IFAO permettent d’attribuer à la période perse.2

révélés les premières campagnes de fouilles (Wagner 1983; 1987; O. Douch I à V)? Il faut d’abord résumer ici, très sommairement, l’état actuel de nos connaissances archéologiques (Reddé 2004). Le site occupe une butte naturelle de grès qui domine un piémont limoneux fertile, exploité au moins depuis l’époque romaine, bien que les installations hydrauliques repérées autour du tell ne soient pas datées avec précision et pourraient s’avérer parfois plus anciennes. Le sommet est dominé par un temple en pierre, dédié à Sarapis, et décoré sous Domitien puis sous Hadrien, si l’on en croit les inscriptions hiéroglyphiques de ses parois; il recouvre un sanctuaire antérieur en briques crues, complètement arasé, et non daté intrinsèquement (fig. 9 et 10). L’ensemble est entouré d’une très grosse enceinte de briques crues, qui a connu de nombreuses réfections; ses

2 On trouvera les principales informations sur ce site dans les comptes rendus annuels des travaux de l’Institut Français d’Archéologie Orientale.

426

M. REDDÉ: L’OCCUPATION MILITAIRE TARDIVE DANS LES OASIS D’ÉGYPTE. L’EXEMPLE DE DOUCH

Fig. 9 Vue aérienne de tell Douch (cliché J.-F. Gout, IFAO, dans Reddé 2004: fig. 13)

Fig. 10 Plan schématique du complexe religieux de Sarapis et du kasr de Douch (Reddé 2004: fig. 86)

427

THE LATE ROMAN ARMY IN THE EAST À Douch même, dont le nom grec (Kysis) est bien attesté par une inscription datée de Trajan (SEG 24: 1215), on connaît surtout les niveaux de l’antiquité tardive. C’est vrai notamment des constructions très nombreuses qui ont proliféré à l’intérieur des différentes enceintes du sanctuaire. Les ostraca grecs publiés par G. Wagner et H. Cuvigny sont constitués, dans une large mesure, par des reçus ou des bordereaux de livraisons annonaires; ils révèlent en outre la présence de soldats. Ce sont ces documents qui, associés à la présence d’une grosse «fortification», avaient conduit les premiers fouilleurs à considérer que le «fort» de Douch faisait partie d’un système défensif de l’oasis. Les fouilles que j’ai conduites sur le site de 1985 à 1990 ont permis de montrer, au contraire, que les enceintes autour du sanctuaire avaient pour fonction initiale d’abriter les magasins du complexe religieux, comme il est d’usage en Égypte. Elles n’ont donc pas été construites par et pour l’armée romaine. Leur réutilisation, au quatrième siècle de notre ère, ne constitue qu’un épisode secondaire dans leur longue histoire. Les textes indiquent-ils pour autant la présence à Douch d’une garnison permanente destinée à défendre l’oasis?

avec précision qui est vraiment militaire, et qui ne l’est pas, j’ai compté entre 20 et 30 noms chrétiens parmi les soldats possibles,5 pour 50 à 60 noms païens. Cette proportion est considérable, eu égard à la lenteur du phénomène de christianisation des noms, si l’on suit les positions de R. Bagnall (Bagnall 1986; 1987), elle indique une datation assez tardive dans le courant du IVe siècle. Va dans le même sens la réinterprétation par J.-P. Callu d’un ostracon (O. Douch I.54) d’où se déduit l’équivalence 25 nummi = 100 talents, soit 1 nummus = 4 talents (0,6 myriades) ce qui suggère qu’on se trouve en présence de l’aes 4, frappé à partir de Valentinien (Callu 1989). Le terminus post quem de cet ostracon ne semble donc pas antérieur à 378. Cette argumentation est confortée par le prix de la viande (150 talents la livre, soit 22,5 myriades) qui apparaît dans le même document et se rapproche des prix pratiqués vers la fin du siècle (Bagnall 1985). Ces différents éléments, compte tenu de l’homogénéité globale du dossier papyrologique de Douch, plaident pour une datation entre le dernier quart du quatrième siècle et le premier quart du cinquième. Bien que la céramique associée à ces documents n’ait guère été prise en compte dans les premières campagnes de fouilles, l’examen du matériel issu des mêmes niveaux lors des recherches des années 1985-1990 conforte cette chronologie générale (Reddé 2004). La présence militaire dans les anciens magasins du temple doit donc être limitée à cette période, où le sanctuaire était désaffecté; on ne sait trop exactement quand elle prend fin, mais les témoignages d’occupation humaine sur le site se raréfient, sans trace de destruction manifeste, dans la première moitié du cinquième siècle, malgré la mention d’une invasion de Blemmyes qui auraient dévasté Hibis en 435 (Evagr. h. e. I.7).

Durant tout le principat, les soldats semblent totalement absents de Kysis, si l’on en croit le silence de la documentation. Le premier témoignage connu ne remonte pas avant 302, date à laquelle un cavalier de Denderah achète une chamelle à un exopylite de Douch (P. Grenf. 74). Mais le papyrus dit clairement que l’unité à laquelle appartient l’acheteur est en garnison dans la vallée; par la suite, d’autres Tentyrites seront attestés à Douch ou dans sa région (Wagner 1987: 59; O. Douch 85, 88, 102). On doit donc considérer que ces soldats «montaient» régulièrement dans l’oasis; mais ils n’étaient pas les seuls: G. Wagner mentionne en effet des Thébains (O. Douch III.193, 201), des Ermonthites (Wagner 1987: 60), des Latopolites (Wagner 1987: 43, 59), des Coptites (O. Douch III. 320, 340), des Lycopolites (Wagner 1987: 32, 33, 61), des soldats d’Edfou, d’Asphynis (Wagner 1987: 11), de Psinaula (Wagner 1987: 384), des Saraceni (O. Douch II, 85). De cette liste, parfois inutilement chargée et criticable,3 retenons au moins une chose: les militaires présents dans le sud de l’oasis de Khargeh ne constituaient pas une garnison permanente, ils appartenaient à des détachements temporaires issus des unités de la vallée, parfois de manière concomitante. On est donc en droit de se poser la question de leur fonction effective.

Dans un article récent, J.-M. Carrié a réexaminé différents aspects du dossier papyrologique de Douch, en particulier le terme de portarensis (Carrié 2004). Le mot, attesté à plusieurs reprises6, avait été interprété comme un doublon d’ostiarius par G. Wagner, qui y voyait la sentinelle de garde à la porte du camp (O. Douch I.2021). J.-M. Carrié y reconnaît au contraire, à juste titre, un mot dérivé de porta qui doit être ici considéré comme l’équivalent du grec pylè, «poste de douane». On connaît d’ailleurs à Douch une megalè pylè (O. Douch IV.384385). La qualité militaire ou civile peut en revanche être discutée, aucun argument vraiment dirimant ne pouvant être invoqué en faveur de l’une ou l’autre hypothèse, bien que certains portarenses soient parfois qualifiés de stratiôtai (par exemple O. Douch V.573). Il est certain en revanche que le dossier papyrologique révèle une vraie symbiose entre soldats et pagani, qui se retrouvent mêlés au sein des mêmes listes, et vivent manifestement ensemble. Certains militaires habitent peut-être dans le village, si l’on interprète comme billets de logement les

La chronologie de ce dossier constitue un point non négligeable du raisonnement. L’onomastique du dossier Kysite révèle en effet une proportion importante de noms chrétiens, parfois de seconde génération.4 Même s’il est difficile, en l’absence d’une relecture critique et synthétique de cette documentation complexe, d’évaluer 3

Tous ne sont pas nécessairement des soldats (Reddé 2004: 204). On connaît notamment l’optio Abraam, fils d’Isaac, qui revient assez souvent dans la documentation.

4

5 6

428

Voir toutefois ci-dessous les réserves qu’on peut formuler sur ce point. O. Douch I.41 (complet); abrégé en port. dans I.31 et V.573.

M. REDDÉ: L’OCCUPATION MILITAIRE TARDIVE DANS LES OASIS D’ÉGYPTE. L’EXEMPLE DE DOUCH O. Douch 1 et 1447. D’une manière plus générale, J.M. Carrié pose dans son article la question plus générale de l’interprétation qu’il convient de donner à des appellations jusque là considérées comme militaires (stratiôtai, portarensis, signifer, tesserarius, optio…) mais dont le sens précis peut varier selon le contexte et qui ne sont pas nécessairement tous des soldats de l’armée. Quant aux documents annonaires, ils ne concernent pas tous la distribution de vivres, mais constituent pour certains des reçus de perception de l’annone: selon J.-M. Carrié, il faudrait considérer en ce sens O. Douch I.18, 38, 39, 45, 47 et 50. Dans le détail, assurément, l’interprétation est délicate, d’autant que les premières lectures ont souvent été corrigées après coup par les auteurs eux-mêmes, mais une synthèse historique nouvelle de ces documents fait toujours défaut.

Carrié J.-M. 2004 Portarenses (douaniers), soldats et annones dans les archives de Douch, oasis major. In Balmelle C. - Chevallier P. – Ripoll G. (ed.) Mélanges d’Antiquité tardive: Studiola in honorem Noël Duval. Turnhout: 261-274. Cuvigny H. 2003 Brun J.-P. - Bülow-Jacobsen A. – Cardon D. – Fournet J.-L. – Leguilloux M.-Matelly M.-A. – Reddé M. – Cuvigny H. (ed.) La route de Myos Hormos. L’armée romaine dans le désert oriental d’Égypte. FIFAO 48. Le Caire. Cuvigny H. - Hussein A. - Wagner G. 1992 Les ostraca grecs d’Aïn Waqfa, Oasis de Kharga. DFIFAO 30. Le Caire. De Romanis F. 1996 Cassia, Cinnamomo, ossidiana. Uomini e merci tra oceano Indiano e Mediterraneo. Rome. O. Douch I 1986 Cuvigny H. – Wagner G. Les ostraca grecs de Douch. DFIFAO 24/I. Le Caire. O. Douch II 1988 Cuvigny H. – Wagner G. Les ostraca grecs de Douch. DFIFAO 24/II. Le Caire. O. Douch III 1992 Cuvigny H. – Wagner G. Les ostraca grecs de Douch. DFIFAO 24/III. Le Caire. O. Douch IV 1999 Wagner G. Les ostraca grecs de Douch. DFIFAO 24/IV. Le Caire. O. Douch V 2001 Wagner G. Les ostraca grecs de Douch DFIFAO 24/V. Le Caire. Reddé M. 1989 CR de Wagner 1987. In JRA 2: 281290. Reddé M. 1991 À l’ouest du Nil: une frontière sans soldats, des soldats sans frontière. In Roman Frontier Studies 1989. Exeter: 485-493. Reddé M. 1999 Sites militaires romains de l’oasis de Khargeh. In BIAO 99: 377-396. Reddé M. 2003 a Le paysage de la route. In Cuvigny 2003: 39-49. Reddé M. 2003 b Les fortins du désert oriental d’Égypte et l’architecture militaire romaine. In Cuvigny 2003: 235-262. Reddé M. 2004 Kysis. Fouilles de l’Institut français d’archéologie orientale à Douch, Oasis de Khargeh (1985-1990). DFIFAO 42. Le Caire. Sidebotham St. 1994 Preliminary Report on the 19901991 Seasons of Fieldwork at Abu Sha’ar (Red Sea Coast). In JARCE 31: 133-158. Wagner G. 1986 Le camp romain de Douch (oasis de Khargeh-Égypte). In Studien zu den Militärgrenzen Roms III. Vorträge des 13. Internationalen Limeskongresses, Aalen, 1983. Stuttgart: 671-679. Wagner G. 1987 Les oasis d’Égypte à l’époque grecque, romaine et byzantine d’après les documents grecs. IFAO. BiEtud 100. Le Caire.

Concluons: l’idée d’un poste fortifié pour et par l’armée, dans le cadre d’une défense de l’oasis, ne tient guère. Il n’y a d’ailleurs pas une garnison à demeure, mais des détachements provenant de différentes unités de la vallée, qui ne viennent sans doute pas principalement pour assurer l’ordre, car cette tâche est normalement du ressort du tribun de l’Ala I Abasgorum, qu’on voit venir en tournée au sud de l’oasis dans un ostracon d’Ain Waqfa, près de Douch (Cuvigny-Adel Hussein-Wagner 1992: 79). Ces soldats viennent surtout, à mon sens, pour surveiller les levées en nature et la perception des douanes, normalement opérées par des fonctionnaires civils, avec lesquels ils sont en contact permanent. Les anciens magasins du temple pouvaient à cette occasion retrouver, dans un autre cadre, leur fonction principale d’horreum. Les tâches de l’armée romaine tardive dans l’oasis de Khargeh, l’importance de son dispositif au sein d’un terroir fertile protégé par son isolement même doivent par conséquent être reconsidérées. Tout autant que défensives et militaires, elles étaient administratives, douanières, policières et politiques. Bibliographie Bagnall R. 1985 Currency and Inflation in Fourth century Egypt, supp. BASP V. Bagnall R. 1986 Religious Conversion and Onomastic Change in early Byzantine Egypt.. In BASP 19: 173181. Bagnall R. 1987 Conversion and Onomastics: a Reply. In ZPE 69: 243-250. Bousquet B. 1996 Tell-Douch et sa région (désert libyque, Égypte). Géographie d’une limite de milieu à une frontière d’Empire (avec la collaboration de M. Robin, P. Deleuze, Ch. Braun). DFIFAO 31. Le Caire. Callu J.-P. 1989 Monnaies de comptes et monnaies réelles: l’ostracon 54 de Douch. In ZPE 79: 73-79. 7 J.-M. Carrié ne semble pas retenir cette hypothèse, sans pouvoir pourtant en proposer vraiment une autre. Faute d’alternative crédible, j’en resterai, pour ma part, à l’interprétation de G. Wagner.

429

L’armée romaine d’Afrique de 375 à 439: Mythes et réalités Yann Le Bohec Université Paris IV - Sorbonne

Ayant, dans un précédent article,1 présenté l’armée romaine d’Afrique depuis l’époque de Dioclétien jusqu’à la mort de Valentinien Ier, il m’a paru intéressant de prolonger cette réflexion et de mettre à jour la vieille notice de R. Cagnat.2 La Maurétanie Tingitane sera absente de ce propos car, durant cette période, elle était davantage liée à la péninsule Ibérique qu’à l’Afrique; au demeurant, cette région a été bien et récemment étudiée;3 cette absence ne laisse donc aucun regret. Dans ces conditions, le principal problème vient des sources, en particulier de la Notitia Dignitatum. Il sera intéressant de voir comment les inscriptions et les textes complètent ce document et dans quelle mesure ils permettent de comprendre le succès de l’invasion vandale.

comte d’Afrique7 puis en tant que maître de la milice pour l’Afrique (magister militum per Africam).8 Il disposait d’une armée qui, au moins en théorie, comprenait tous les types d’unités connus chez les Romains, infanterie, cavalerie, alliés et marine. Honorius envoya contre lui Théodose le Jeune avec une armée de Germains, en partie prise sur les effectifs qui se trouvaient en Italie, notamment en Étrurie (légions Herculia, Iouia, Augusta), en Gaule, en Espagne et en Bretagne. Il bénéficia de soutiens sur place: les Africains lui fournirent des archers et des lanceurs de javelots, c’est-à-dire des troupes légères. Gildon appartenait à une grande famille. De même qu’il avait trahi son frère Firmus au profit des Romains, ainsi son frère Maczel le trahit-il au bénéfice de Théodose. L’épopée se termina par une défaite, vers Tébessa, en 397 ou 398.9

La chronologie fournira un cadre utile pour commencer et pour comprendre l’organisation de l’armée romaine d’Afrique et son tragique destin final.

Très peu de temps après le retour au calme, au tournant du siècle, les Austuriens se jetèrent sur les villes de Tripolitaine.10

Une inscription de Tripolitaine non datée,4 qui mentionne un personnage peut-être identifiable avec un correspondant d’Ausone, indique que ce dernier a vaincu des barbares. Si l’hypothèse le concernant se révélait acceptable, on pourrait penser soit au conflit avec les Austuriens qui prit place entre 363 et 365 soit à d’autres troubles qu’il faudrait situer entre 375 et 383.

Et ce n’est pas tout. La province d’Afrique jouissait d’une grande réputation de prospérité, non usurpée comme le prouve l’archéologie. Après avoir pris et pillé Rome en 410, Alaric mena ses Wisigoths dans le Sud de l’Italie avec l’intention de leur faire traverser la mer pour les installer dans cette riche région. Mais les dieux condamnèrent ce projet et ils empêchèrent la traversée de la mer.11 Les barbares remontèrent vers le Nord pour piller Rome de nouveau puis ils gagnèrent, comme on sait, l’Aquitaine et la péninsule Ibérique.

Nous avançons en terrain plus sûr avec la guerre de Gildon.5 Ce prince maure a laissé une réputation de cupidité et de luxure, peut-être en partie injustifiée. Vers 386, il prit une large autonomie par rapport au pouvoir central, et son entreprise présenta un aspect politique. Comme beaucoup d’autres hauts responsables abandonnés dans leurs provinces par le pouvoir central, il voulait davantage de libertés, notamment dans les domaines de la fiscalité et de la sécurité. Il paraît peu probable qu’il ait espéré une réelle indépendance au nom d’un quelconque nationalisme parfaitement anachronique. Il n’est pas davantage assuré qu’il ait pris la défense des donatistes et des circoncellions contre l’empereur;6 dans ce cas, il eût fait un mauvais choix, en s’appuyant sur des groupes en déclin. Quoi qu’il en soit, il dirigea l’Afrique utile pendant quelque douze ans, d’abord avec le titre de

L’année suivante, en 411 donc, s’ouvrit à Carthage la célèbre conférence qui devait consacrer le déclin du donatisme.12 L’armée romaine y prit sa part, puisque la réunion s’ouvrit en présence de trois protecteurs domestiques assistés par trois ducénaires. En 413, le comte d’Afrique Héraclius (ou mieux: Héraclianus) fut menacé de destitution. Pour prévenir le danger, il décida de porter ses coups contre l’empereur Honorius. Il gagna l’Italie; mais, capturé, il fut vite exécuté.13 7

Oost 1962. Grosse 1920: 162-163. Hier. Epist. 79; Modéran 2003: 3. 10 Mattingly 1995:177. Sur ce peuple: Modéran 2003: 211-249 et 279310. 11 Iord. Get. XXX. Romanelli 1959: 632. 12 Carth. a. 411 I. 1 (voir l’édition de S. Lancel). Diesner 1964 b: 4 et 27-28. 13 Hier. Epist. 130; Oros. Hist. VII. 41. 4-12; Hydace 51 (voir 56); Zos. VI.9.

1

8

Le Bohec 2004. 2 Cagnat 1918: 703-766. 3 Euzennat 1989; Villaverde Vega 2001. 4 IRT 565. 5 Claud. 15; Oros. Hist. VII. 36. 2-11 et 13; Zos. V. 11. 2-5. Warmington 1954: 11-12; Romanelli 1959: 603-620; Oost 1962: 27-30; Diesner 1962: 178-186; idem. 1964: 11-13 et 97-102; Burian 1968: 251270; Modéran 2003: 169-171. 6 Thèse de Diesner 1962: 11-13.

9

431

THE LATE ROMAN ARMY IN THE EAST

Fig. 1. Carte de l’Afrique (Tripolitaine). Dessin de l’auteur. Quand arriva le deuxième quart du Ve siècle, l’armée romaine d’Afrique n’avait eu que des adversaires secondaires à affronter, tels les Austuriens, une confédération de tribus semi-nomades qui semblent revenues à des sentiments pacifiques au début du Ve siècle; un texte gravé entre 408 et 423 cite leur nom, mais ne parle pas de leur «furor» comme on l’a écrit.14 Elle avait surtout participé à des conflits internes qui ne l’avaient pas renforcée et n’avaient pas non plus renforcé l’autorité impériale.

charge en temps d’insurrection.18 Mais c’est ici que la situation se complique. D’après la Notitia Dignitatum, le magister peditum praesentalis exerce une autorité directe sur le comte d’Afrique et sur les ducs de Maurétanie Césarienne et de Tripolitaine qui, tous les trois, se trouvent en permanence sur le terrain. Mais il ne fait aucun doute que le comte d’Afrique est le supérieur des deux ducs. Le comte d’Afrique est en effet uir spectabilis depuis environ 320.19 Dans la Notitia Dignitatum, il est appelé tantôt comes rei militaris Africae tantôt comes limitis Africae tantôt uir spectabilis, comes Africae.20 Il réside à Carthage21 et possède des attributions très larges. Dans le domaine militaire, qui est essentiel, il commande l’ensemble de l’armée d’Afrique22 et il mène les soldats au combat.23 Nous verrons qu’il ne commande pas une armée mobile fantôme, mais qu’il est seul à avoir autorité sur des troupes d’élite. Il possède également des attributions judiciaires24 et il intervient beaucoup dans les affaires religieuses, notamment pour pratiquer la charité;25 c’est ainsi que le dernier comte, Boniface, a entretenu une correspondance suivie avec s. Augustin.26 Il

Quand on s’attache aux structures, à l’organisation de cette armée, il n’est pas inutile pour cette période de partir de la Notitia Dignitatum15 en essayant de la confronter aux autres types de sources. Le commandement constitue le premier point à aborder. Comme pour toute l’armée romaine, l’empereur est chef suprême. Mais il ne réside jamais en Afrique. Il délègue son commandement à un magister equitum praesentalis, le maître de la cavalerie «présent» (à la cour),16 qui donc, par définition, est absent, puisqu’il ne vit pas en Afrique lui non plus. Il a directement sous ses ordres le maître de l’infanterie «présent», le magister peditum praesentalis qui présente les mêmes caractéristiques à un niveau inférieur.17 Gildon, qui a porté le titre de maître de la milice, constituait un cas particulier, puisqu’il a exercé sa

18

Grosse 1920: 393-398. Cagnat 1918: 719. ND Oc I. 30 et 32; V. 125 et 128; VII. 140 et 179; XXV. 19. Grosse 1920: 162-163 et 393-398. 21 Cagnat 1918: 721. 22 Cagnat 1918: 718. 23 Aug. Epist.* 11. 12. 24 Aug. Epist.*11.8. 25 Aug. Epist.* 7.1. 26 Olymp. XL. De Lepper 1941. 19 20

14

IRT 480 (Lepcis Magna). Modéran 2003: 165. Dorénavant abrégée en ND Oc, pour Notitia Dignitatum … in partibus Occidentis; Clemente 1968; Hoffmann 1969. 16 ND Oc VI. 41 et VII. 179. 17 ND Oc V. 125. 15

432

YANN LE BOHEC: L’ARMÉE ROMAINE D’AFRIQUE DE 375 À 439:

Fig. 2. Carte de l’Afrique ( Byzacène-Zeugitane-Numidie-Maurétanies, Sitifenne et Césarienne). Dessin de l'auteur. est assisté par un état-major ou officium.27 Des listes de comtes d’Afrique ont été établies.28

La liste des unités qui sont aux ordres de ces officiers est fournie par la Notitia Dignitatum; mais il faut discuter les données de ce texte.35

Le duc de Maurétanie Césarienne, lui, figure dans la Notitia Dignitatum sous le titre de dux limitis Mauretaniae Caesariensis ou encore u(ir) p(erfectissmus), dux et praeses prouinciae Mauretaniae Caesariensis.29 Sa fonction ne semble pas attestée avant le temps de Gratien et Valentinien II; lui aussi possède un officium.30 Le duc de Tripolitaine, u(ir) p(erfectissmus), dux et praeses prouinciae Tripolitanae ou encore dux limitis Tripolitani pour la Notitia Dignitatum,31 ne fit, lui aussi, qu’une apparition tardive: sa fonction n’existait pas avant 363, et elle n’a peut-être été créée que vers 393.32 Il porta d’abord le titre de perfectissime puis celui de spectabilis et enfin, à partir de 410, il fut appelé comte et duc. Lui aussi disposait d’un officium analogue à celui du comte d’Afrique.33 Des listes de ces personnages ont également été publiées.

• Cavalerie d’élite (uexillationes comitatenses) aux ordres du maître de la cavalerie praesentalis (ND Oc. 41 et 53). 1. Equites scutarii (ND Oc. VI. 63; VII. 181). On trouve une unité d’equites scutarii iuniores en Maurétanie Tingitane (ND Oc. VII. 207). Une inscription de Timgad mentionne des scutarii; AE 1946: 42 (Timgad):36 D M S | VAL VINCENTIVS | BIARCVS DE NV[MERO] | ISCVTARIOR[VM]. Le mot numerus désigne soit une unité particulière différente des légions, soit n’importe quel type d’unité. Equites stablesiani Africani (ND Oc. VI. 64 et VII. 182). La ND Oc. VII. 182, les définit comme stablesiani seniores. Une inscription de Sétif mentionne des stablesiani; elle daterait du Haut-Empire à notre avis, mais des auteurs l’ont attribuée au Bas-Empire.37 Equites Marcomani (ND Oc. VI. 65; VII. 183). Equites armigeri seniores (ND Oc. VI. 66; VII. 184). Il existe deux unités d’armigeri (propugnatores), les iuniores et les seniores, parmi les unités palatines (voir plus loin). Deux inscriptions mentionnent des armigeri. La première ne leur donne aucune épithète; elle sera citée plus loin (CIL VIII 9613, de Malliana). L’autre les appelle equites armigeri iuniores (CIL VIII 9255, de Rusguniae, également citée plus loin). Cette dernière est également mentionnée dans la Notitia Dignitatum, mais plus loin. Il faut admettre la dernière théorie sur ces appellations: une unité de iuniores est née de la division d’une unité dite de seniores.38 La présence de ces

Chacun de ces trois officiers supérieurs avait à sa disposition des praepositi limitum. Il y en avait donc pour l’Afrique, pour la Maurétanie Césarienne et pour la Tripolitaine;34 nous verrons que, d’après la Notitia Dignitatum, chacun de ces secteurs stratégiques était divisé en douze secteurs défensifs ou limites, dont le nombre doit être revu.

27

ND Oc. XXV. 37-46. Cagnat 1918: 721. Cagnat 1918: 721-723; PLRE II: 1295-1296. ND Oc. I. 37-38; V. 133-134; XXX. 11. Cagnat 1918: 724. 30 ND Oc. XXX. 20-29. 31 ND Oc. I. 37 et 39; V. 133 et 135; XXXI. 17. Grosse 1920: 179-180 et 410. 32 Hoffmann 1974. Grosse 1920: 179-180. 33 ND Oc. XXXI. 32-41. Cagnat 1918: 723. 34 ND Oc. XXV. 19 (Afrique); XXX. 11 (Césarienne) et XXXI. 17 (Tripolitaine). 28 29

35

Courtois 1955: 80 et 91; Hoffmann 1969 I: 192-193. Leschi 1957 : 234, n° 2; BCTH 1943-1945: 343, n° 2; Le Bohec 2004: 257 37 CIL VIII 8490; discussion dans Le Bohec 2004. 38 Drew-Bear 1977; Carrié - Janniard 2000: 323-324. 36

433

THE LATE ROMAN ARMY IN THE EAST seniores dans la Notitia Dignitatum plaide en faveur de l’existence d’une unité de iuniores qui a pu disparaître ou avoir été oubliée par le rédacteur du document administratif qui, au demeurant, a commis d’autres erreurs, comme nous le verrons. 5. Equites sagittarii clibanarii (ND Oc VI. 67; VII. 185). Ils sont simplement equites clibanarii dans ND Oc. VII. 185. Equites sagittarii Parthi seniores (ND Oc. VI. 68, VII. 186). Les mots sagittarii et Parthi sont inversés dans ND Oc. VII. 186. Sur senior, voir l. 4. Equites primo sagittarii (ND Oc. VI. 69; VII. 188). Equites secundo sagittarii (ND Oc. VI. 70; VII. 189). Equites tertio sagittarii (ND Oc. VI. 71; VII. 190). 10. Equites quarto sagittarii (ND Oc. VI. 72; VII. 191). Equites sagittarii Parthi iuniores (ND Oc. VI. 73; VII. 192). Equites cetrati seniores (ND Oc. VI. 74; VII. 187). La cetra est un petit bouclier de cuir. Comites iuniores (ND Oc. VI. 75, manque en VII). On remarque que cette unité, seulement présente en VI, est la seule de la liste à ne pas être désignée par le terme equites. Equites promoti iuniores (ND Oc. VI. 76; VII. 194). Sur iunior, voir l. 4. 15. Equites sagittarii iuniores (ND Oc VI. 77) et equites scutarii iuniores [comitatenses] (ND Oc. VII. 195). Les appellations diffèrent; il s’agit sans doute d’une erreur imputable à l’auteur du document administratif. Equites cetrati iuniores (ND Oc. VI.78; VII. 193). Sur la cetra, voir l. 12. Equites Honoriani iuniores (ND Oc. VI.79; VII.196). Equites armigeri iuniores (ND Oc. VI.80; VII.198). Cette unité se trouvait sans doute à Rusguniae.39 Equites secundi scutarii iuniores (ND Oc. VI.81; VII.197). ND Oc. VII.197: Equites secundi scutarii iuniores scholae secundae; une schole de bracati est mentionnée sur une inscription de Timgad.40 Des scutaires sont attestés dans la même cité.41 20. Equites stablesiani italiciani (ND Oc. VI.82; VII.180). Le maître de cavalerie «présent», supérieur du comte, commandait donc 19 ou 20 unités de cavalerie, des uexillationes comitatenses, du moins si l’on en croit la Notitia Dignitatum (ND Oc. VI.41 et 53; VII.179). La présence en Afrique de ces unités est explicitement mentionnée par ce document. Les auteurs modernes disent «19 ou 20» parce que ce texte donne deux listes qui différent sur un point: les equites iuniores de l’une ne sont pas mentionnés dans l’autre. • Infanterie d’élite aux ordres du maître de l’infanterie praesentalis (ND Oc V.125).42 Légions palatines, 3 unités (ND Oc. V.144).

Armigeri propugnatores seniores (ND Oc. V.151; VII.142). Des unités d’armigeri sont attestées dans l’épigraphie de Malliana et de Kemellel. CIL VIII 9613 (Malliana):43 MESFLV | AVRELIVS NVC[.]F[.]V | SIERICES ARMIGERI|RVM AVRELI MASVC | AMSRIPP PREFECTVS | FVTIMADI | CVM. AE, 1909: 120 (Kemellel): […]MERI ARMIGERORVM | BOTVM CONPLEBIT DOGS. L. 1, RSAC 42 1908: 209, n° 15: ME(MO)RIA; AE: [… NV]MERI. Cimbriani (ND Oc. V.155; VII.145). Des Cimbriani sont connus à Sétif.44 AE 1984: 940 (Sétif): MEMORIAE FL REGINIA|NI COR CIMB VIX AN | XLIIII MENSE VN DIES | XXIIII ORAS VII AMEN|ESTRAVET AN XVIII ACTARIVS AN IIII CO|NICVLARIVS AN XIIII | MEMORIA FECIT EM|ILIA VLPIOLA DVLCIS[S]IM|A CONIVX. Armigeri propugnatores iuniores (ND Oc. V. 156; VII, 143). Sur les armigeri, voir plus haut. Unité auxiliaire palatine, 1 unité (ND Oc. V. 157). Celtae iuniores (ND Oc. V. 205; VII, 141). Légions comitatenses, 8 unités (ND Oc. V. 223). Secundani italiciani (ND Oc. V. 235; VII. 144). Prima Flauia Pacis (ND Oc. V. 249; VII. 146). Secunda Flauia Virtutis (ND Oc. V. 250; VII. 147). Une inscription de Thelepte mentionne la legio II Flauia Virtutis. CIL VIII 23181 (Thelepte): MARITVS GRATIAS SACTITATI TVE | FORTVNVLA VXOR | VALENTINI CENTVRIONIS | LEGIONIS SECVNDE FLAVIE VIRTV|TIS ET VIXIT AN XL DVCLISSIME VXORI | FECIT QUAE HABVIT PATRE LAOMEDONTE | ET MATRE OLIMPIA IN VRBE SARRA. Cette unité est venue en Afrique au moment de la révolte de Firmus et elle y est restée.45 Tertia Flauia Salutis (ND Oc. V. 251; VII. 148). Flauia uictrix Constantina = Constantici (ND Oc. V. 252; VII. 150). Cette unité ou la suivante, elle aussi, est venue en Afrique au moment de la révolte de Firmus et elle y est restée.46 Secunda Flauia Constantiniana (ND Oc. V. 253; VII. 149 et, en Tingitane, 138). Voir la précédente. Tertio Augustani (ND Oc. V. 254; VII. 151). On a vu, à raison, la IIIe légion Auguste du Haut-Empire sous cette forme de son nom; elle est attestée au moins par une inscription postérieure à l’époque tétrachique. ILAlg I 2766 = AE 1919: 50 (Madaure):47 VETERANORVM MEMORIA FELIX | CAECILI AEMILIANI CONTINENS | NOMEN QVEM PRIMEVM LE|GIONI TERTIE AVGVSTE MILITIA PROBATVM | DEDIT IVVENTAS QVEM POS LABOREM VIRTVTIS | HONESTA MISSIONE MERITVM AD FELICES | ANNOS PROVEXIT SENECTVS CATOLICE LEGI FIDE|LISSIMA MENTE INSERVIENS VIXIT AN|NIS LXXXIII IN PACE | FIDELIS (CHRISME). Fortenses (ND Oc. V. 255; VII. 152). Légion pseudo-comitatensis, 1 unité (ND Oc. V. 256). Constantiaci (ND Oc. V.271; VII. 150, et, en Tingitane, 138).

39 CIL III 9255, et, peut-être, 9613, et AE 1909: 120. Cagnat 1918: 731, n. 1 40 AE 1946: 43, citée plus bas. 41 AE 1946: 42, citée plus haut. Leschi 1957: 233-235; Le Bohec 2004: 257. 42 Cagnat 1918: 729-730.

43

Cagnat 1918: 731, n. 1. Benseddik 1981; Le Bohec 2004: 256. 45 Cagnat 1918: 731, n. 3; Le Bohec 2004: 256. 46 Amm. Marc. XXIX.5.20 et 22. Le Bohec 2004: 256 47 Monceaux P. Un vétéran chrétien à Madaure. In CRAI 1919: 142-149. 44

434

YANN LE BOHEC: L’ARMÉE ROMAINE D’AFRIQUE DE 375 À 439: Le maître de l’infanterie «présent», lui aussi supérieur du comte, commandait 13 unités d’élite, soit 3 légions palatines, 1 unité auxiliaire palatine, 8 légions comitatenses et 1 légion pseudo-comitatensis (ND Oc. V. 125 et 128; VII. 140).48

Praepositus limitis Tubusubditani (ND Oc. XXV.27). Tubusubditanus dérive d’un site dont le nom est connu sous plusieurs formes, Tubusuctu ou Tubusuptu,57 et qui se trouvait en Maurétanie Sitifienne. Praepositus limitis Thamallomensis (ND Oc. XXV.28). R. Cagnat58 pensait qu’il fallait rapprocher Thamallomensis de Thamannuna plutôt que de Turris Tamalleni; nous préférons proposer une similitude entre ce nom et Thamallula, jadis Tocqueville, aujourd’hui Ras el-Oued, en Maurétanie Sitifienne. Praepositus limitis Balaretani (ND Oc. XXV.29). R. Cagnat se demanda s’il ne fallait pas rapprocher ce nom de celui d’un évêché; les sources chrétiennes mentionnent un episcopus Castelliibaritanus; finalement, il rejeta comme peu probable cette hypothèse.59 10. Praepositus limitis Columnatensis (ND Oc. XXV.30). Ici, il n’y a pas d’hésitation possible: il s’agit de Columnata, qui se trouvait en Maurétanie Césarienne.60 Praepositus limitis Tablatensis (ND Oc. XXV.31). R. Cagnat n’avait aucune identification à proposer pour ce site; de manière hypothétique, nous suggérons Tabalati, autre nom de Talalati, site de Tripolitaine aujourd’hui appelé Ras el-Aïn.61 Praepositus limitis Caputcellensis (ND Oc. XXV.32). De nombreuses Cellae se trouvaient en Afrique; nous ignorons où se trouvait celle-ci62. Praepositus limitis secundaeforum in castris Tillibanensibus (ND Oc. XXV.33). R. Cagnat, de manière curieuse, avait envisagé une secunda (legio) qui se serait trouvée à Tillibari, site bien connu de Tripolitaine.63 Nous préférons proposer une autre interprétation: c’est la cohors II Afrorum, bien attestée sur le site de Remada sous le Haut-Empire,64 qui y est restée jusqu’au BasEmpire. Il n’est pas sans exemple que des unités du Principat aient occupé le même emplacement pendant quatre siècles.65 De plus, Tillibari se trouvait en Tripolitaine.66 Praepositus limitis Taugensis (ND Oc. XXV.34). R. Cagnat, en désespoir de cause, a envisagé d’identifier Taug- avec Tucca, ce qui n’a convaincu personne, pas même cet illustre savant67. Dans ce cas, il aurait fallu le placer aux confins orientaux de la Maurétanie Sitifienne. 15. Praepositus limitis Bidensis (ND Oc. XXV.35). Il faut identifier ce site avec Bida,68 en Maurétanie Césarienne. Praepositus limitis Badensis (ND Oc. XXV.36). R. Cagnat, n’a trouvé aucune proposition d’identification pour ce Bad-.69 On peut penser à une assimilation avec

Il est raisonnable de penser que les unités dépendant des maîtres de la milice, résidant à la cour, étaient en fait placées sous les ordres du comte d’Afrique, présent sur le terrain. Ce dernier n’était en théorie le supérieur que des limitanei d’Afrique,49 placés sous la responsabilité de praepositi limitum, qui étaient au nombre de 16 d’après la Notitia Dignitatum; comme nous l’avons dit, il est plus que vraisemblable qu’il avait aussi autorité sur les ducs. Il semble que chaque limes ait correspondu à une unité. En effet, le camp de Remada, les castra Tillibanensia (en réalité Tillibarensia), était occupé par l’ala II Flauia Afrorum, comme nous le verrons. Mais cet unique exemple devrait inciter à la prudence. • Limitanei aux ordres du comte d’Afrique (ND Oc. XXV.19-20).50 Praepositus limitis Thamallensis (ND Oc. XXV.21). Ce secteur défensif tire son nom très vraisemblablement de Turris Thamalleni, actuellement Telmine. Chef-lieu des Nybgenii, ce site appartenait à la Tripolitaine sous le Haut-Empire.51 Praepositus limitis Montensis in castris Leptitanis (ND Oc. XXV.22). Un autre passage mentionne des milites fortenses in castris Leptitanis (ND Oc. XXXI.29). P. Blanchet52 avait pensé à corriger une lettre pour lire Neptitanis; mais il vaut peut-être mieux respecter le texte tant que nous n’avons pas la preuve claire d’une erreur et maintenir Leptitanis. L’auteur de la Notitia Dignitatum semble avoir commis une confusion, et il faut sans doute également placer ce limes en Tripolitaine, avec son centre à Lepcis Magna. Praepositus limitis Bazensis (ND Oc. XXV.23). Le centre de ce limes devait correspondre au lieu dit Ad Badias.53 Praepositus limitis Gemellensis (ND Oc. XXV.24). Deux sites appelés l’un et l’autre Gemellae sont connus, l’un en Numidie et l’autre en Proconsulaire; le premier, qui avait abrité des soldats sous le Haut-Empire, se trouvait au Sud de l’Aurès.54 5. Praepositus limitis Tubunensis (ND Oc. XXV.25). Tubunensis vient très probablement de Tubunae, aujourd’hui Tobna.55 Praepositus limitis Zabensis (ND Oc. XXV.26). Zabensis vient sans doute de Zabi,56 site de Maurétanie Sitifienne qui se trouvait donc sans doute sous l’autorité du duc de Césarienne.

57

Cagnat 1918: 754. Cagnat 1918: 754-755. Cagnat 1918: 755. 60 Cagnat 1918: 755. 61 Cagnat 1918: 756; Trousset 1974: 98-102. 62 Cagnat 1918: 756. 63 Cagnat 1918: 757. 64 Le Bohec Y. Les unités auxiliaires de l’armée romaine en Afrique proconsulaire et Numidie. Aix-Marseille. 1989: 67-70. 65 Roxan M. In BAR. Suppl. S. 15. 1976: 59-72. 66 Trousset 1974: 114-118. 67 Cagnat 1918 : 757. 68 Cagnat 1918 : 757. 69 Cagnat 1918 : 757. 58 59

48

Cagnat 1918: 729-730. CJ I.27.2 § 8 (534). Jones 1964: 205, n. 130. 50 Cagnat 1918: 751; Jones 1964: 205, n. 130. 51 Cagnat 1918: 752; Baradez 1949: 136-137; Trousset 1974: 42-43. 52 Cagnat 1918: 752-753. Baradez 1949: 136-137 (voir 142-143). 53 Cagnat 1918: 753. Baradez 1949: 136-137. 54 Baradez 1949: 136-137. 55 Cagnat 1918: 754; Baradez 1949: 136-137. 56 Cagnat 1918: 754. 49

435

THE LATE ROMAN ARMY IN THE EAST Bazensis (ci-dessus, ND Oc. XXV. 23); la lecture du dernier livre d’Y. Modéran incite à penser à Badès, Vadis, au Sud-Est de l’Aurès, entre Tchouda et Nefta.70 D’après la Notitia Dignitatum (ND Oc. XXV.19 et 20),71 les troupes aux ordres du comte d’Afrique étaient réparties sur 16 limites. Mais il faut lui retirer quatre secteurs pour les rendre au duc de Tripolitaine et six, ou au moins cinq, qu’il convient d’attribuer à son homologue, le duc de Maurétanie Césarienne. On peut toutefois envisager une hypothèse: la Maurétanie Sitifienne (quatre secteurs) aurait été placée, sur le plan militaire, sous l’autorité du comte d’Afrique et pas sous celle du duc de Maurétanie Césarienne. Dans le cas contraire, il ne lui reste que six limites au maximum, car il est possible que limes Bazensis et limes Badensis ne soient qu’une seule et même dénomination.

43 (Timgad): MIMORIA FL | VRSACIO EX | NVMERO PRO|PVNNANTES EX S|CVLA BRACATORVM. Le soldat a connu une promotion du numerus des propugnantes, des fantassins, à la schole des bracati, des cavaliers. Aziz ben Telis (aujourd’hui Benhyahia Abderrhamane): un eques, un cavalier, a laissé une inscription à cet endroit.76 AE 1917-1918: 46 (Aziz ben Telis): MARTININVS VIXIT A|OS XXXIII MILITAVIT AN | EQVES MORDVS ERCV | [.]DICRA NATVS REGION| [.]ISSESA VICV BANNIOLV | […]ORIE PATER ISTRA | […]S FECIT. Propositions de lecture: MORDVS = MORATVS (?). ERCV = (AD H)ERCV(LEM?). DICRA = [I]DICRA (?). ORIE = [MEM]ORIE (?). ISTRA = STRA(TOR?), le strator étant un écuyer. L’inscription est surmontée par un relief; un homme en tunique courte et en manteau fait une libation sur un autel.

L’épigraphie pose d’autres problèmes. Elle attire l’attention sur des centres connus ou elle permet de leur ajouter de nouveaux sites, mais hélas sans apporter beaucoup de précisions chronologiques. Cette lacune explique que plusieurs textes mentionnés ici l’ont déjà été dans l’article que nous avons consacré à l’armée d’Afrique entre 284 et 378.

• Limitanei aux ordres du duc de Tripolitaine (ND Oc. XXXI.17).77 Praepositus limitis Talalatensis (ND Oc. XXV.29 = XXXI.18). Talalati ou Tabalati, actuellement Ras el-Aïn, est un site bien connu de Tripolitaine, que la Notitia Dignitatum a également placé par erreur en Afrique.78 Praepositus limitis Tenthettani (ND Oc. XXXI.19). Il est tentant de faire un rapprochement avec le limes Tentheitanus de l’épigraphie, attesté dès l’époque de Philippe l’Arabe.79 Praepositus limitis Bezerentana (ND Oc. XXXI.20). Bezereos est une forme tardive du nom Vezereos, nom antique de Bir Rhezen.80 Praepositus limitis Tillibarensis (ND Oc. XXV. 33 = XXXI.21). Tillibari est aujourd’hui Remada, également placé par erreur en Afrique.81 5. Praepositus limitis Madensis (ND Oc. XXXI.22). La Notitia Dignitatum mentionne plus loin (XXXI. 30) des milites munifices in castris Madensibus. D. Mattingly place hypothétiquement ce limes dans la région de Dehibat;82 spécialiste de la région, il suppose que la plupart des sites non identifiés se trouvaient dans l’Est de la province,83 sans doute à raison, puisqu’il faut bien admettre une certaine occupation de tout l’espace. Praepositus limitis Maccomadensis (ND Oc. XXXI.23). Il s’agit sans aucun doute de Macomades Syrtis, Mersa Zafran ou Maghmadas, dans l’Est de la province, vers Sirte; ce site se trouvait sur le territoire des Seli, les Psylles de l’époque classique.84 Praepositus limitis Tintiberitani (ND Oc. XXXI.24). On ignore où placer ce limes.85

El-Jem: un centurion chrétien a vécu dans cette ville. CIL VIII 22855 (El-Jem): FLAVI MARTIALIS | CENTVRIO VIX|IT IN PACE ANI|S XXXVI | (chrisme). Thelepte: la légion II Flauia Virtutis, unité comitatensis vue plus haut, a laissé un texte en ce lieu.72 Tébessa: une inscription mentionne le numerus Martensium, absent de la Notitia Dignitatum. ILAlg I 3123 = CIL VIII 16551 et p. 2731 (Tébessa): D M S | VAL VITALIS | MIL EX N | MARTENS | DE GALLES S | VIX AN LV. Caisson, trouvé dans les dépendances de la basilique. Des Martenses sont attestés en Gaule (ND Oc., V.115 et 265). Le S à la fin de la ligne 5 est barré d’un trait: indication d’un grade? Sans doute. La mention des dieux Mânes place ce texte à une période assez haute du Bas-Empire. Madaure: des vétérans chrétiens de la IIIe légion Auguste, autre unité comitatensis vue plus haut, ont fait installer une sépulture collective à leur intention (texte donné plus haut).73 Kemellel: un numerus d’armigeri est mentionné sur ce site, entre Tébessa et Khenchela (le texte a été donné plus haut).74 Timgad: grande ville, où une présence militaire est attestée par des inscriptions du IVe siècle, ce centre ne figure pas dans la Notitia Dignitatum; on y a trouvé une mention d’un numerus de scutaires (texte donné plus haut)75 et d’un numerus de propugnantes, des unités sans doute comitatenses. La présence de comitatenses sur la frontière apporte des informations sur la répartition des unités entre l’arrière et l’avant, sur la stratégie. AE 1946:

76

AE 1917-1918: 46. Le Bohec 2004: 255. IRT 480 et 529. CTh XI.36.33; XII.1.133; CJ I.27.2.1a. Cagnat 1918: 734 et 748; Mattingly 1995: 188-192; Modéran 2003: 258-265. 78 Trousset 1974: 98-102. 79 IRT 880. Cagnat 1918: 750-751; Le Bohec Y. In RD 69/3. 1991: 307330. 80 Trousset 1974: 75-77. 81 Trousset 1974: 114-118. 82 Mattingly 1995: 189. 83 Mattingly 1995: 189. 84 Tissot H. Géographie de l’Afrique. II. Paris. 1888: 226; Cagnat 1918: 750-751; Mattingly 1995: 189; Modéran 2003: 222-223 et n. 38. 85 Mattingly 1995: 189. 77

70

Modéran 2003: 363. Cagnat 1918: 751. 72 CIL VIII 23181 = ILTun. 305. 73 ILAlg. I 2766. Le Bohec 2004: 256. 74 AE 1909: 120. 75 AE 1946: 42. Le Bohec 2004: 256. 71

436

YANN LE BOHEC: L’ARMÉE ROMAINE D’AFRIQUE DE 375 À 439: Praepositus limitis Bubensis (ND Oc. XXXI.25). On ignore où placer ce limes.86 Praepositus limitis Mamucensis (ND Oc. XXXI.26). D. Mattingly place hypothétiquement ce limes dans l’Est de la province.87 10. Praepositus limitis Balensis (ND Oc. XXXI.27). Ce secteur se trouvait peut-être, lui aussi, dans l’Est de la province.88 Praepositus limitis Varensis (ND Oc. XXXI.28). Var- a été rapproché de Base, point de l’Itinéraire Antonin qui se trouvait peut-être dans l’Est de la province.89 Milites fortes in castris Leptitanis (ND Oc. XXXI.29). Ces soldats étaient en garnison à Leptis Magna pour R. Cagnat.90 Voir ND Oc. XXV.22: ils ont été placés par erreur en Afrique. Milites munifices in castris Madensibus (ND Oc. XXXI.30). Ces soldats étaient peut-être installés à Taouargha, au Sud-Ouest de Mesrata.91 Praepositus limitis Sarcitani (ND Oc. XXXI.31). On ignore où se trouvait ce limes.92

• Limitanei aux ordres du duc de Maurétanie Césarienne (ND Oc. XXX.11).98 1. Praepositus limitis Columnatensis (ND Oc. XXX.12). Columnata a déjà été vue avec la province d’Afrique où la Notitia Dignitatum l’a placée par erreur.99 Praepositus limitis Vidensis (ND Oc. XXX.13). Vidensis est mis pour Bidensis, également placée par erreur en Afrique par la Notitia Dignitatum.100 Praepositus limitis inferioris (ND Oc. XXX.14). On ne sait où placer de limes. Praepositus limitis Fortensis (ND Oc. XXX.15). R. Cagnat rapprochait ce nom de Frontensis, qui aurait donné Frenda.101 Nous préférons un rapprochement avec Fortassa, jadis Uzès-le-Duc. 5. Praepositus limitis Muticitani (ND Oc. XXX. 16). R. Cagnat, après avoir envisagé un rapprochement avec le castellum Muteci, y a renoncé pour rester dans une incertitude que nous partageons.102 Praepositus limitis Audiensis (ND Oc. XXX. 17). Ici, point d’hésitation: Aud- renvoie à Auzia, en Maurétanie Césarienne.103 Une inscription de 301, trouvée dans cette cité, mentionnait un praepositus limitis.104 Praepositus limitis Caputcellensis (ND Oc. XXX. 18). Ce site, comme on l’a vu plus haut, se trouvait peut-être en Afrique.105 Praepositus limitis Augustensis (ND Oc., XXX.19). On ignore où placer ce limes.106 D’après la Notitia Dignitatum, les troupes aux ordres du duc de Maurétanie Césarienne étaient réparties entre huit limites;107 il faut ajouter ceux qui ont été indûment attribués au comte d’Afrique, et l’on arrive à un total de onze limites.

Si l’on se fie à la Notitia Dignitatum, les effectifs placés sous les ordres du duc de Tripolitaine étaient répartis entre 12 limites, auxquels il faut ajouter un limes placé par erreur sous l’autorité du comte d’Afrique (Thamallensis).93 Dans cette province, l’épigraphie ajoute surtout des centenaria, fortins dont le nom est lié à la racine signifiant cent; mais on ignore ce qui s’y trouvait en cent exemplaires. Breviglieri (près de Tarhuna): centenare.94 On ne peut lire que la première ligne où se trouve ce mot; le reste du texte n’a apparemment pas de sens. Shemech (lieu-dit Gasr Uadi el-Bir, à 120 km au SudSud-Est de Tarhuna): centenari dans un texte également très difficile à lire.95

L’épigraphie apporte quelques compléments. Sétif: deux inscriptions permettent de retrouver l’une les Cimbriani, une unité palatine (texte donné plus haut),108 et l’autre un gradé, un biarcus; il faut toutefois préciser que cette dernière est datée de 324.109 Rusguniae: des equites armigeri iuniores, appartenant à une uexillatio comitatensis, sont attestés. CIL VIII 9255 (Rusguniae):110 D SANCTO LIGNO CRVCIS CHRISTI SALVATORIS ADLATO | ADQ HIC SITO FLAVIVS NVBEL EX PRAEPOSITIS EQVITV | M ARMICERORVM [I]VNIOR FILIVS SATVRNINI VIRI | PERFECTISSIMI EX COMITIBVS ET COLLAIAL HONESTISSIMA | [FEMINAE] PR[ON]EPOS ELVRI LACONI (hedera?) BASILICAM VOTO | PROMISSAM ADQ OBLATAM CVM CONIVGE NONNI|CA AC SVIS OMNIBVS DEDICAVIT.

Peut-être faut-il ajouter à cette liste les trois tribuns de Bir ed-Dreder,96 les constructions de Ghirza97 et un mystérieux centenarius, mentionné avec son épouse par un texte du IVe ou du Ve siècle. IRT 875 (site inconnu, entre Breviglieri et Tarhuna): FLAVII SEBENTIVS CTN | ET STIDDIN EIVS CON|IVNX HVNC LOCVM | DIDICARVNT O BONVM [I]NI|TIVM NATISQVI FILIIS CONCIL|IVM SALVIS LIBIRIS | CVM FILICITATI TRIVM | FANTI BINAITIR POSSISSV.

86

98

87

99

Mattingly 1995: 189. Mattingly 1995: 189. 88 Mattingly 1995: 189. 89 Cagnat 1918: 750-751; Mattingly 1995: 189. 90 Cagnat 1918: 750-751. 91 Mattingly 1995: 189. 92 Mattingly 1995: 189. 93 ND Oc. XXXI.17. Cagnat 1918: 734 et 748; Mattingly 1995: 188192; Modéran 2003: 258-265. 94 IRT 877. 95 IRT 889. 96 IRT 886. Le Bohec 2004: 259. 97 Le Bohec 2004: 259

Cagnat 1918: 758. Cagnat 1918: 758. 100 Cagnat 1918: 758. 101 Cagnat 1918: 758. 102 Cagnat 1918: 759. 103 Cagnat 1918: 759. 104 CIL VIII 9025. 105 Cagnat 1918: 758. 106 Cagnat 1918: 758. 107 ND Oc. XXX.11. Cagnat 1918: 758. 108 AE 1984: 940. Benseddik, cité par Le Bohec 2004: 256. 109 CIL VIII 8491. 110 Cagnat 1918: 731, n. 1.

437

THE LATE ROMAN ARMY IN THE EAST Malliana: présence d’armigeri (texte donné plus haut).111 Castellum Tingitanum: présence d’un tribun en 406. CIL VIII 9715 (castellum Tingitanum): [ME](alpha, chrisme, omega)MORIAE | [IN]NOCENTIS | […]T[…] IN PACE | [O]CTOBRES DIE SEX|[… A] PR CCCLXVII FV|[…] TRIBVNVS PATER ET | […]E MATER EIVS APVT | [SANCTO]S APOSTOLOS PETRV ET | [PAVLV I]N NOM[I]NE DEI ET CRI|[STI M]EMOR[IAM FECE]RVNT. Portus Magnus: mention d’un praepositus limitis. CIL VIII 9755 (Portus Magnus): [CL]OD LAR|[G]VS 7 PRA|[EP]OSIT L T | B ET D. L. 4: l(imi)t(is?). Arbal: mention de praepositi limitis dont l’un est également procurateur. CIL VIII 9790 = ILS 3251 (Arbal): DIANAE VICTRICI | C IVL MAXIMVS | PROC AVG | PRAEPOSITVS LIMITIS. La mention de Diane place ce texte à une période assez haute du Bas-Empire. CIL VIII 9791 (Arbal): DIANAE AVG | SAC | Q MAXIMVS | PRAEP LIM. Même remarque pour la chronologie que pour le texte précédent.

ait pensé R. Cagnat.119 Le numerus d’Hippone120 et le numerus bis electorum121 seraient, au contraire, plus tardifs, d’époque byzantine. Pour en revenir à la Notitia Dignitatum, nous constaterons que son auteur ne connaissait pas l’Afrique, qu’il n’a pas respecté l’ordre géographique et, surtout, qu’il a mélangé ses fiches. Ces listes austères et ennuyeuses permettent néanmoins d’esquisser un tableau de la stratégie adoptée par Rome en Afrique à la fin du IVe siècle. À partir d’un passage du Code Théodosien, J. Baradez applique à cette organisation les noms de fossatum et de limes.122 Si l’on reporte sur une carte les sites mentionnés, on constate que beaucoup de points se regroupent dans la zone Sud, près du désert: peu de changements sont intervenus depuis le Haut-Empire; on attend surtout un ennemi qui viendrait du Sahara. Le maintien, certes hypothétique, de la présence militaire sur la Seguia bent el-Khrass irait dans ce sens.123 Pourtant, il faut résister à un mythe, celui du chameau. Des auteurs du XXe siècle avaient imaginé que les Berbères auraient constitué d’immenses troupeaux de chameaux qui leur auraient permis de piller plus facilement les sédentaires. Les recherches menées sur le sujet ont montré que cet animal avait été domestiqué et utilisé depuis longtemps et qu’il n’y a eu aucun bouleversement en ce domaine au IIe siècle pas plus qu’au IVe siècle.124 En ce qui concerne l’armée romaine, une partie des effectifs a été placée en arrière, soit pour surveiller les montagnes où vivaient des Maures, soit pour des raisons de logistique, pour se placer plus près des sources d’approvisionnement. Il n’est pas impossible que cette répartition des hommes soit attestée dès l’époque de Domitius Alexander, au début du IVe siècle. Ces troupes au demeurant étaient d’une consternante médiocrité. Dès qu’un problème militaire se posait, il fallait faire venir des renforts d’Europe. Contre Firmus comme contre Gildon, il était nécessaire de faire intervenir des troupes extérieures pour rétablir l’ordre.

Deux inscriptions n’ont pas été prises en compte dans notre enquête précédente, qui était consacrée à l’armée romaine de 284 à 378.112 L’une indique que le comte d’Afrique a fait construire une castram à Cellae au Sud du Bou Taleb, à La Bahira, entre 367 et 375;113 ce terme ne désigne pas nécessairement une construction militaire. L’autre mentionne un centenarium installé en 328 à Bou Atelli, en Grande Kabylie;114 il n’est pas assuré non plus qu’il ait relevé de l’armée car il a été construit par un particulier, à ses frais, sumptibus suis. Par ailleurs, des historiens se sont demandé si des civils et des barbares n’avaient pas été associés à la défense de la région comme ailleurs, notamment en Europe. Trois cas ont été envisagés. Il nous paraît peu probable que des iuuenes115 soient intervenus dans ce domaine, à cette époque; la dernière mention de ce genre d’associations remonte au début du IVe siècle, quand les iuuenes de Saldae (ex-Bougie, aujourd’hui Bejaïa) ont su défendre leur ville contre des ennemis.116 On a ensuite pensé aux gentiles; ils sont certes attestés par le Code Théodosien,117 mais ils n’avaient qu’une mission: effectuer des travaux relevant du génie militaire. Quant aux fédérés,118 les foederati, des peuples étrangers à l’empire et liés par un traité, un foedus, ils sont bien attestés, mais ce statut a été principalement accordé aux Vandales après 435 et 442. D’autres unités ne nous paraissent pas avoir appartenu à l’armée d’Afrique des années 375-439. Le numerus electorum remonterait plutôt au Haut-Empire, quoi qu’en

En fait, ce qui est intéressant dans ces listes, c’est plus ce qui ne s’y trouve pas que ce qui s’y trouve. Un certain nombre de mythes, comme celui du chameau, se sont répandus au cours du XIXe et du XXe siècle, à l’instigation de chercheurs qui n’étaient pas des moindres au demeurant, mais qui n’avaient pas une bonne conception de ce qu’était la guerre dans l’Antiquité. Premier mythe: le comitatus. Nous n’avons pas trouvé en Afrique ce comitatus dont on parle tant depuis Th.

119

AE 1901: 113. Cagnat 1918: 731 et n. 7; Speidel M. P. In AntAfr. 23. 1987: 193-196. CIL VIII 5229. Cagnat 1918: 731 et n. 5. 121 ILAlg I 82. Cagnat 1918: 732. 122 CTh VII.15.1. Baradez 1949: 30 (avec un contre-sens: lire: «sous le VIIIe consulat d’Honorius et le IIe consulat de Théodose»). 123 Jones- Mattingly 1980 (mais que prouve une unique pièce de monnaie?). 124 Demougeot E. In Annales (ESC) 1960: 209-247; Mattingly 1995: 178-180.

111

CIL VIII 9613. 112 Le Bohec 2004. 113 CIL VIII 10937 = 20566. 114 CIL VIII 9010. 115 Jaczynowska M. Les associations de la jeunesse sous le HautEmpire. Wroclaw. 1978: 103-107. 116 Leschi 1957: 349-360. 117 CTh VII.15.1. Cagnat 1918: 745. Baradez 1949: 30 (voir plus loin). 118 Cagnat 1918, mais d’après Th. Mommsen.

120

438

YANN LE BOHEC: L’ARMÉE ROMAINE D’AFRIQUE DE 375 À 439: Mommsen.125 La raison en est simple: le mot comitatus désigne non pas une partie de l’armée mais la cour. Or la cour réside en Europe ou en Orient et pas en Afrique. Certes, des soldats sont dits agentes in sacro comitatu. Il faut comprendre qu’ils étaient détachés auprès de l’empereur en déplacement avec les siens. Le mot comitatus se trouve dans le Thesaurus linguae latinae, et la consultation de cet article montre que cet organe est strictement civil.126 Il a servi de titre à un colloque récent et l’on remarque que les communications sont consacrées aux femmes, aux eunuques, aux sénateurs, …, bref à toutes sortes de personnages;127 seuls sont absents de ce tableau les militaires. Certes, il y en avait, mais en petit nombre; les uns appartenaient à la garde impériale, et ils servaient de garde d’honneur; les autres apportaient leur expérience et leurs conseils à l’empereur.

champs: c’était l’été, précisément parce que les assaillants récoltaient le blé des assaillis. Il aurait fallu choisir. Et les textes de loi qui disent que les soldats limitanei possèdent des terres ne disent pas qu’ils les cultivaient:129 les femmes, les enfants, les colons et les esclaves étaient là pour remplir cette fonction. L’évolution de la religion a déjà abordée dans le précédent article consacré à l’armée d’Afrique. L’abandon du camp d’Alamilliaria aurait permis l’installation d’une basilique chrétienne dans des principia tardifs, ce qui a été contesté.130 D’une manière plus générale, on peut dire que le christianisme a largement progressé au sein des troupes, au détriment d’un paganisme surtout plus discret. Pour l’archéologie, profondément renouvelée pour cette époque, il suffit de renvoyer aux travaux mentionnés dans notre précédent article, car, pour cette deuxième époque, peu de monuments sont connus à la fois par l’épigraphie et l’archéologie. Il est par ailleurs difficile de s’étendre davantage que ne l’a fait J. Peyras sur la question des rapports entre civils et militaires, les premiers assurant surtout à titre d’impôts l’entretien des seconds, comme le prouvent les Codes.131

Deuxième mythe: l’armée mobile et l’armée immobile. Depuis Th. Mommsen, cité plus haut, les historiens pensent que les Romains avaient besoin de troupes mobiles pour courir après des barbares qui menaient des raids en profondeur à l’intérieur de l’empire. Ils recherchaient donc partout désespérément des cavaliers. Mais, en Afrique, on ne compte que 19 ou 20 unités de cavalerie contre 50 unités d’infanterie. Et sans doute les premières possédaient-elles des effectifs moins importants que les secondes. De toute façon, les barbares possédaient peu de forces montées quand ils en possédaient. Durant toute l’Antiquité, l’infanterie lourde resta la reine des batailles, chez les Romains comme chez leurs ennemis. Les raids des Germains en Europe et des Maures en Afrique prouvent seulement que les Romains avaient un problème d’infanterie et pas de cavalerie. D’ailleurs, la Notitia Dignitatum ne dit nulle part que les limitanei n’avaient pas le droit de bouger et que les comitatenses devaient être soumis à un mouvement perpétuel. 128 L’opposition que l’on a cru voir entre des unités mobiles et des unités sédentaires doit être revue; la vraie différence opposait des unités d’élite et des unités de valeur courante. Les corps appelés palatini ne vivaient pas au palais (palatium) et les comitatenses ne séjournaient pas à la cour (comitatus), qui se trouvaient sur d’autres continents. Ils avaient simplement reçu des épithètes d’honneur qui servaient à les distinguer des autres.

Il reste à expliquer pourquoi et comment s’est faite la conquête vandale.132 Quand il s’agit d’histoire militaire, la première question à poser est le pourquoi de la guerre (la ou les causes). Il n’est pas difficile de répondre à cette question.133 Les Vandales voulaient d’une part s’affranchir de la menace que les Wisigoths faisaient planer sur eux dans la péninsule Ibérique et d’autre part conquérir un territoire riche, l’Afrique. Il faut rappeler qu’une guerre, dans l’Antiquité, se faisait en respectant plus ou moins une sorte de droit “ international ” que tout le monde connaissait et admettait. Les Vandales se sont donc conduits comme leurs prédécesseurs, comme leurs contemporains et comme leurs successeurs. Ils ont alors conquis l’Afrique et ils ont forcé les riches habitants à fuir (des textes contemporains rapportent le malheur de certains d’entre eux);134 ils ont pillé, violé, tué, et détruit.135 Les Romains auraient agi de même, et, au cours de leurs multiples entreprises de conquête, ils avaient adopté le même comportement dans le passé. Cette conduite était si banale que les historiens ne rapportent pas les faits de ce genre; il faut chercher dans des documents touchant la vie quotidienne pour en trouver l’écho. Par ailleurs, comme on sait, les Vandales étaient ariens et les Africains n’avaient jamais adhéré à

Troisième mythe: les soldats-paysans. Sur la foi d’une notice de l’Histoire Auguste, écrit qui suscite actuellement plus que des réserves, et sur la lecture de textes de loi ambigus, les historiens ont imaginé que l’État romain avait installé des soldats qui passaient une partie de l’année à travailler la terre et l’autre à combattre, mal au demeurant, ce qui est d’ailleurs bien compréhensible. En fait, la saison de la guerre correspondait au moment où il faut travailler dans les

129

CJ I.27.2 § 8, 534 (des limitanei possèdent des agri); XI.60.3. Lenoir 1986; Duval 1991. 131 Peyras 1992. 132 Courtois 1955; Romanelli 1959: 640-662; Diesner 1964 b: 181-199; Gil Egea 1998. 133 Diesner 1964 b: 166-181. 134 Thdt. Ep. 29, 33-36, 52 et 70. 135 AE 1935: 85 (Altaua). 130

125

Mommsen 1889. TLL 1912: 1793-1797. Comitatus 1998. 128 Contre Cagnat 1918: 715-716; Clemente 1968: 319-343. 126 127

439

THE LATE ROMAN ARMY IN THE EAST cette conception de la nature du Christ.136 Or, au IVe siècle, si les évêques savaient à peu près bien se conduire dans les synodes et les conciles, il n’en allait pas toujours de même au sein du peuple. Il n’était pas rare que des Chrétiens convaincus de leur juste cause aillent expliquer à leurs adversaires qu’ils se trompaient en utilisant le bâton. L’autorité faisait parfois appel à l’armée et les morts s’ajoutaient aux blessés. La conquête vandale est rapportée par Procope et, avec beaucoup de détails affreux, par Victor de Vita; comme ce dernier est un Romain d’Afrique et un nicéen, il était de bonne méthode de passer son récit au gril de la suspicion et de supposer qu’il avait exagéré en parlant de barbares, ariens de surcroît. Pourtant, les victimes n’ont pas toujours tort et, dans le cas de Victor de Vita, force est de constater qu’il avait des raisons de se plaindre quand on sait comment se déroulait une guerre dans l’Antiquité et comment certains chrétiens se comportaient avec leurs coreligionnaires. Une inscription au moins le rappelle, AE 1935: 85 (Altaua): KAL SEM | NNO PRO CCCXC | GLADIO P A BARBAROS. Propositions de lecture: SE(PTE)M(BRIBVS), et P(ERIIT). L’année 390 de la province correspond à 429 de l’ère chrétienne. Il existe donc un autre mythe que nous voudrions contester, celui de Vandales aimables et cultivés.

rendre utiles, de reprendre des hommes libres qui avaient été enlevés par des marchands d’esclaves ou d’organiser des rondes nocturnes pour renforcer une sécurité qui ne régnait pas toujours et partout.145 Dans leur rôle de combattants, en revanche, ils ne réussirent pas même à faire illusion. Des conditions proprement militaires ont peut-être joué: la stratégie de Rome visait surtout à protéger les régions prospères contre d’éventuels assaillants venus du désert. Or les Vandales sont arrivés par l’intérieur et ils ont contourné le système; ils l’ont pris à revers. Assurément, les hommes peuvent se mouvoir et faire un quart de tour pour se disposer face à l’Ouest; il n’en va pas de même pour les fortifications. Impressionnante sur le papier, cette armée était une armée de papier. Les Vandales, y compris les esclaves, ne dépassaient pas les 80000 personnes,146 ce qui représente au maximum 15000 combattants. Leur entreprise peut être décomposée en deux étapes majeures. Ils passèrent le détroit de Gibraltar sans difficulté,147 ce qui éclaire sur l’incompétence des cadres romains. En effet, une armée est toujours très exposée lors d’un débarquement, un épisode toujours difficile et dangereux. Mais rien ni personne ne les dérangea pendant cette opération. À peine à terre, ils auraient commencé les persécutions, montrant leur vraie nature selon Victor de Vita: populus ille crudelis ac saevus.148 Il y eut une première bataille. Boniface la perdit.149 Il y eut un siège, celui d’Hippone, où se mourait s. Augustin.150 Il ne fut finalement pas gagné par les Romains et les Vandales montrèrent qu’ils avaient appris au moins un peu de poliorcétique. Il y eut une deuxième bataille, près de Guelma, en 430.151 Malgré des renforts byzantins, emmenés par Aspar, l’armée d’Afrique fut encore battue. Il ne restait qu’à capituler, ce qui se traduisit par un traité, en 435. Les Vandales obtenaient la Numidie. Mais, dans leur esprit, un traité n’avait pas plus de valeur qu’un simple armistice; il n’était qu’un moyen pour refaire et regrouper des forces. En octobre 439, ils prirent Carthage.152 Un deuxième traité, conclu en 442, reconnaissait leur domination sur la Proconsulaire, la Byzacène et la Numidie orientale.153 Peu à peu, ils s’emparèrent des territoires qui leur manquaient, la Gétulie et le reste de la Numidie. L’Afrique romaine avait vécu.154

En ce qui concerne le prétexte, il a été fourni par le comte d’Afrique, Boniface.137 Des auteurs modernes n’ont pas hésité à parler de sa «trahison», d’autres de sa sottise. Le pouvoir central (Galla Placidia) voulait lui retirer son commandement138 et deux expéditions avaient été organisées pour le chasser de son poste.139 Pour se défendre, il fit appel aux Vandales, espérant sans doute qu’ils repartiraient après l’avoir aidé. Il prit néanmoins conscience de l’erreur qu’il avait commise et il se réconcilia avec Placidia; mais il était trop tard. En théorie, l’armée romaine d’Afrique représentait une force non négligeable: la Notitia Dignitatum, l’épigraphie et l’archéologie montrent son importance. Plusieurs hypothèses ont été avancées pour expliquer sa médiocrité réelle: politique (les intrigues de cour),140 sociale (la révolte de Gildon s’était appuyée sur un mécontentement réel) ou socio-religieuse.141 Il faudrait peut-être ajouter les conditions économiques. Il semble que, depuis longtemps, les soldats avaient perdu l’habitude de recevoir leur salaire avec régularité.142 Mal payés, ils étaient peu motivés et indisciplinés. Dans le meilleur des cas, ils commettaient des exactions au détriment des civils,143 dans le pire des cas, ils se faisaient brigands.144 Dans un rôle de policiers, ils essayaient parfois de se

Cet examen a permis de reposer des vieilles questions et d’en poser de nouvelles. Il nous semble tout d’abord que, pour le pouvoir central, l’Afrique n’a pas représenté une 145

Aug. Epist.* 20.6. Vict. Vit. I.1. Diesner 1964 b: 46. 147 Vict. Vit. I.1; Procop. Vand. III.26. 148 Vict. Vit. I. 2-7. Voir AE 1935: 85, citée plus haut. 149 Procop. Vand. III. 30 (v. 31). 150 Vict. Vit. I.10; Procop. Vand. III.32. Diesner 1964 b: 47. 151 Procop. Vand. III.35. Diesner 1964 b: 49. 152 Vict. Vit. I.4. Diesner 1964 b: 49. 153 Vict. Vit. I.13; Thdt. Ep. 29, 33-36, 52 et 70. Courtois 1955: 323; Diesner 1964 b: 1. 154 Modéran 2003: 541: les Maures passèrent «de la soumission à la révolte». 146

136

Lancel 2000. Procop. Vand. III. 25. Diesner 1964 b: 78-40 et 166-181. De Lepper 1941. 138 Aug. Epist.. 220.7. 139 Aug. Epist.. 229 et 230. 140 Romanelli 1959. 141 Diesner 1964 b: 79-150. 142 Amm. Marc. XXVII.9.1. 143 Aug. Epist..* 20.6. 144 Aug. Epist..* 10.2 et 5. 137

440

YANN LE BOHEC: L’ARMÉE ROMAINE D’AFRIQUE DE 375 À 439: Hoepfner A. 1936. Les magistri militiae praesentales. In Byzantion 11: 483 sq. Hoffmann D. 1969 Das spätrömische Bewegungsheer und die Notitia Dignitatum. ES. 7. Düsseldorf. Hoffmann D. 1974 Der Oberbefehl des spätromischen Heeres im 4 Jh. n. Chr. In Actes du 9e Congrès sur les frontières romaines. Bucarest: 381-397. Jones A. H. M. 1964 The Later Roman Empire. Oxford. Jones G. D. B. - Mattingly D. 1980 Fourth-Century Manning of the Fossatum Africae. In Britannia 11: 323-328. Kotula T. 1972 Der Aufstand des Afrikaners Gildo. In Das Altertum 33: 167-176. Lancel S. 2000. Victor de Vita, témoin et chroniqueur des années noires de l’Afrique romaine au Ve siècle. In CRAI: 1199-1219. Le Bohec Y. 2004 L’armée romaine d’Afrique de Dioclétien à Valentinien Ier. In Le Bohec Y. – Wolff C. (ed) L’armée romaine de Dioclétien à Valentinien Ier. Lyon: 251-265. Lenoir M. 1986 Une martyre près des principia. À propos du camp et de la basilique d’Ala milliaria. In MEFRA 98/2: 643-664. Leschi L. 1957 Études d’épigraphie, d’archéologie et d’histoire africaines. Paris. Mattingly D. 1995 Tripolitania. Londres. Modéran Y. 1989 Gildon, les Maures et l'Afrique. In MEFRA 101/2: 821-872. Modéran Y. 2003 Les Maures et l’Afrique romaine. BÉFAR 314. Rome. Mommsen Th. 1889 Das römische Militärwesen seit Diocletian. In Hermes 24: 195-279. Monceaux P. 1920 Deux victimes des Maures à Madaure. In CRAI: 329-333. Oost S. 1962 Count Gildo and Theodosius the Great. In CPh 57: 27-30. Oost S. 1966 The Revolt of Heraclian. In CPh 61: 236241. Peyras J. 1992 Les questions militaires nord-africaines dans les Codes Théodosien et Justinien. In CT. 46. 161/3: 27-67. Romanelli P. 1959 Storia delle province romane dell’Africa. Rome: 603-620. Trousset P. 1974 Le limes de Tripolitaine. Paris. Villaverde Vega N. 2001 Tingitana en la Antigüedad Tardía. Madrid. Warmington B. H. 1954 The North African Provinces from Diocletian to the Vandal Conquest. Cambridge.

région stratégique très exposée; les Perses et les Germains causaient des soucis bien plus grands aux empereurs. Il fallait toutefois protéger l’Afrique. Hélas, l’auteur de la Notitia Dignitatum, très critiquable pour cette partie de son texte, a mélangé les unités et il a présenté de manière confuse le commandement. Il faut au moins revoir de manière très critique les vieilles notices de R. Cagnat. En outre, plutôt que de concevoir une défense divisée entre des troupes de couverture immobiles et une réserve mobile, ne faut-il pas plutôt opposer des troupes officiellement d’élite et d’autres qui ne le sont pas? Enfin, n’est-il pas temps de repenser la question des Vandales, que Christian Courtois avait décrits avec quelque angélisme? Bibliographie Les abréviations des titres de revues sont empruntées à l’Année Philologique. Baradez J. 1949 Fossatum Africae. Paris. Benseddik N. 1981 Les Cimbriani à Sétif. In BCTH 17 B: 363-369. Burian J. 1968 Die einheimisches Bevölkerung Nordafrikas in der Spätantike. In Altheim F. - Stiehl R. (ed.) Die Araber in der alten Welt. V/1: 170-304. Cagnat R. 1918² L’armée romaine d’Afrique. Paris: 703766. Carrié J.-M. et Janniard S. 2000. L’armée romaine tardive dans quelques travaux récents. In AntTard 8: 321-341. Clemente G. 1968 La Notitia Dignitatum. Cagliari. Comitatus 1998 Winterling A. (ed.) Comitatus. Beiträge zur Erforschung des spätantiken Kaiserhofes. Berlin. Courtois Ch, 1955 Les Vandales et l’Afrique. Paris. De Lepper J.L.M. 1941 De rebus gestis Bonifatii, comitis Africae et magistri militum. Nimègue. Diesner H.J. 1962 et 1964 a Gildos Herrschaft und die Niederlage bei Theveste. In Klio 40: 178-186; 42: 1113 et 97-102. Diesner H.J. 1964 b. Der Untergang der römischen Herrschaft in Nordafrika. Weimar. Donaldson G.H. 1985 The praesides prouinciae Tripolitanae, civil Administrators and Military Commanders. In BAR International Series 274: 165177. Drew-Bear Th. 1977 A late fourth-Century Latin soldier’s epitaph at Nakolea. In HSCP 81: 257-274. Duval N. 1991 La basilique de Bénian (Ala milliaria) est-elle un remploi de principia militaires? In L'Africa romana 8/2: 1079-1089. Euzennat M. 1989 Le limes de Tingitane. Paris. Gebbia C. 1988. Ancora sulle «rivolte» di Firmo e Gildone. In L'Africa romana 5: 117-129. Gil Egea M.E. 1998 África en tiempos de los Vándalos. Alcalá de Henares. Grosse R. 1920. Römische Militärgeschichte von Gallienus bis zum Beginn der byzantinischen Themenverfassung. Berlin.

441