146 46 143MB
English Pages [374] Year 1984
Roman Stone Fortifications Variation and Change from the First Century A.D. to the Fourth
James Lander
BAR International Series 206 1984
B.A.R.
5, Centremead, Osney Mead, Oxford OX2 OES, England.
GENERAL EDITORS A.R Hands, B. Sc., M.A., D.Phil. D.R Walker, M.A.
BAR -s206,1984 :'Roman Stone Fortifications'
©
JAMES LANDER ,
1984
The author’s moral rights under the 1988 UK Copyright, Designs and Patents Act are hereby expressly asserted. All rights reserved. No part of this work may be copied, reproduced, stored, sold, distributed, scanned, saved in any form of digital format or transmitted in any form digitally, without the written permission of the Publisher. ISBN 9780860542674 paperback ISBN 9781407336039 e-book DOI https://doi.org/10.30861/9780860542674 A catalogue record for this book is available from the British Library This book is available at www.barpublishing.com
Acknowledgements This work would not have been feasible but for the enormous labour of excavation and excellent publication by dozens of European archaeologists, mostly members of the Congress of Frontier Studies, whose works are gratefully cited in the Bibliography. I must thank all of the members of my doctoral committee, particularly Dr. Susan B. Downey. Dr. Ronald Mellor made especially helpful comments on historical points, though he is not responsible for any flaws remaining. I am grateful to Dr. James N. Hill for opening the door to new methods of analysis. Encouragement, inspiration and friendship were generously bestowed by Dr. S. Thom~s Parker• I am indebted to Mr. Andrew Vermes for technical help and assistance with material in East European languages, and to Mr. Christopher Berrisford, who made available a word-processor and printer. Most of all I wish to thank my wife, Susan, for her patient support and kindness. To her and to our son, George, this book is dedicated.
iii
Preface While this is a survey rather than an exhaustive study of Roman stone fortifications, some five hundred sites are mentioned in the text and an illustration is provided for most. For the reader's ease and to simplify some cross-referencing, an Index is provided which contains, in addition to certain subject headings, a complete list of sites including variant names crossreferenced. A bold-face page number denotes the location of a plan or illustration. Many fortifications are known by both modern and ancient names. In place of some possibly consistent but certainly awkward system, I normally use the name which is most common. In the example of the legionary fortresses of Deva and Singidunum, the former is commonly known as Chester while the latter is never referred to as Belgrade. Many sites with Arabic names are known in the literature both with and without certain substantives such as Gasr, Ran, Hirbet, Kasr, Khan, Khirbet, Qasr, Qseyr, and Tell. A cross-reference in the Index will direct the reader to the proper name, for example, 'Bshir' rather than 'Qasr Bshir'. And while accents and umlauts for French and German are (I hope) retained, the various marking systems for Arabic have been ignored: transliterations from the Arabic are always haphazard at best, and the vowel-quantity-, ~reathing- and hamzamarkings, while meaningful to Arabists who do not need them, are a wash-out for the rest of mankind. Except where otherwise noted, all dates are of the Christian era. Abbreviations for titles in the Bibliography follow L'Annee philologigue. Finally, for the reader's convenience, an imperial chronology is offered below, followed by a master map indicating the orientation of the sets of five frontier maps after the first page of each chapter. Roman Emperors August us Tiberius Caligula Claudius Nero Galba Otho Vitellius
Vespasian Tit us Domitian Nerva Trajan Hadrian Antoninus Pius
27 B.C.-A.D. 14 14-37 37-41 41-54 54-68 68--69 69 69 V
69-79 79-81 81-96 96-98 98-117 117-138 138-161
Marcus Aurelius Lucius Verus Ccmnodus Pertinax Didius Julianus Septi.mius Severus Cara call a Macrinus Elagabalus Severus Alexander 'Maxi.minus Gordian I Gordian II Balbinus Pupienus Gordian Ill Philip Decius Trebonianus Aenilianus Valerianus Gallienus Claudius Gothicus Aurelian Tacitus
Florianus Probus Carus Carinus NtDnerian Diocletian Maxi.mian Constantius Chlorus Galerius Licinius Cons tan tine the Gr~at Constantius
161-180 161-169 180-192 193 193 193- 211 211-217 217- 218 218-222 222-235 235-238 238 238 238 238 238-244 244-249 249-251 251-253 253 253-260 253-268
Cons tans Julian Jovian Valentinian I Valens Gratian Theodosius Maximus Valentinian
268-270 270-275 275-276
Honorius Arcadius Theodosius
II
276 276-282 282-283 283-285 283-285 284-305 286-305 292-306 293-311 311-323 306-337 337-361
Constantine II
II
II
vi
337-340 337-350 360-363 363-364 364-375 364-387 375-383 379-395 383-387 383-392 395-423 395-408 408-450
! •'
.
&
/) p .:>
(])
,_
.. !
~
"( 8
·-a.
...
!
~j ~
E
! l
c::
•
u.J
C'd
E 0
~
(])
...c:
r-
vii
t
0
Contents Preface
V
Introcluetion
1
Chapter ODe: The Petrified Frontier (c. A.D. 5o-138)
5
I. The Ranans 1 Introduction to Forti fications
II. The Eastern Evidence in a Early Phase Ill. The Early Use of Stone in the West: Claudius to Nerva (41-98) IV. Trajan and Stone Fortification ( 98-117) V. The Hadtianic System (117-138)
5 15 20 30 49
Chapter Two: The System Under Stress (13&-196)
67
I. Changes in the Use of Fortifications under Antoni nus Pi us
(138-161) II. Marcus Aurelius and Ccmnodus: Innovations Following t he Marcomannic War s
67
92 107
Chapter Three: The Sever an Renewal (193-235) I. Changes Throughout the Defensive System II. Mrica and Some Developments in Gate-Tower Desi gn under the Severans Ill • The Use of t he Rounded Gate-Tower in the Severan Period IV. Other Severan Sites in Europe V. The East in the Severan Period VI. Summary
107 114 121 127 132 148 151
Chapter Four: Collapse aad Reconstruction ( 235-337) I. Fortifications Datable to within the Period from Maximinus Thrax to Claudius Gothicus ( 235-270) II. Fortifications Datable to within the Period f r om Aurelian to the Beginning of Diocletian 1 s Reign ( 27D-284) Ill. Fortifications Most Readily Dated to the Peri od of Diocletian and the Tetrarchy IV. Fortifications Most Readily Dated to the Age of Constantine t he Great
ix
158 168 181 193
V. A Study of Notable Attributes of Late Roman Fortifications (284-350) A. Square Towers, 198. B. Rounded Towers, 208. C. Circular Towers, 228. D. Polygonal Towers, 244. E. Fan-Shaped Towers, 246. VI . A Suamary on Late Third and Early Fourth Century Fortifications A. Toward a Chronology of Towers of the Late Third and Early Fourth Centuries, 255. B. Design Changes as Technical Improvements, 257. (1. Towers and Artillery, 258. 2. Barracks Arrangements, 259. 3. Siting and Outworks, 261.) Chapter Five: The Last Frontier (35o-410)
198
252
263
I. The Percepti on of Frontier Requirements by Valentinian I II. The Fortifications of Valentinian I (364-375)
263 270
A. Forts, 270. B. Renovations, 276. C. Fortified Landing-Places, 284. D. Watch-Towers, 289. E. Summary: Valentinian's Response to the Late Four th-century Crisis, 290. Ill. The Collapse of the Defense of the West IV. The Defense of the City of Rome (395-410)
293 293
and his Contemporaries
Conclusion
299
Bibliography
313 313 316
I. Primary Sources II. Secondary Sources
355
Index
Maps The Roman Fmpire, with key to other maps, vii Sites Mentioned in: Chapter One, 6-10 Chapter Two, 68-72 Chapter Three, 108-112 Chapter Four, 152-156 Chapter Five, 264-268
Introduction
The manner in which the Romans employed stone fortifications to support the defense of their Empire changed greatly between the first and fourth centur ies A.D. and also varied from frontier to frontier during that time. The purpose of this study is to examine those changes and variations and to try where possible to explain why they occurred. The method of research has been to collect published information on the design and chronology of indi v idual fortif ications from all frontiers of the Empire, to seek out patterns of development in the design of particular features and in the organization of frontier systems, and to relate these patterns to historical developments. The survey reflects the unevenness of the archaeological data, of research and publicat ion, and, indeed, of the interests and programs of the original builders themselves. In presenting the information I have divided the continuum of development into a r ough pattern of five chapters reflecting what seem to be important phases. The shift from less permanent building materials to construction in stone is examined in the first chapter, along with background information and a view of Hadrian's creation of a 'permanent' frontier for the civ i l ized world. Chapter Two traces the extension and testing of those developments over the next half century. The third chapter examines developments in fortifications as part of larger changes in the Roman military during the Severan period. This era closed with a half century of civil war and invasion, and the sparse evidence of forti fication construction from that period is dealt with briefly in t he beginning of Chapter Four, which then goes on to examine what might be called the first half of 'Late Roman Fortification', mainly the work of t he Tetrarchy and the house of Constantine in their efforts to reconstruct imperial frontiers. The final period of 'Late Roman Fortificat i on', largely the work of Valentinian, is discussed in Chapter Five. Some particularly active emperors, usually with long reigns, provide the focus f or sub-divisions within the first three chapters, on the theory that emperor s guided defense policy of which fortifications were a component. Additional information on circumstances and conditions which would have concerned the emperors is offered here usually as introduction to these subdivi s i ons. In the last two chapters most of the sub-divisions relate to examinations of particular design-features--like tower-shape--in an attempt t o sort out this dynamic period of change. Most of the questions addressed in this study have been raised by scholars in the past. Why did the Romans shift to stone construction? How sudden and even was the transition? Why do 'late' types of fortifications show so many more variations than 'early' types? How do fortification designs re flect '1
frontier policy of individual emperors? What were the purposes of major changes in tower construction? Why do Late Roman fortifications sometimes have barrack-arrangements different from the standard of the Early Empire? To what extent were alterations in the system of defensive architecture centrally planned? How are these developments related to changes in ·the threat to the Empire and in the army's capability to man the frontiers? Is it possible to work out a chronology for fortification-design which could ftm.ction like chronologies worked out for pottery? Though the questions have been raised, the ability to answer them has been limited by the fact that most research has been necessarily parochial. Archaeologists are kept thoroughly busy trying to tm.derstand developments along particular frontiers, like Hadrian's Wall, the Dacian frontier, and the limes Arabicus. To buttress or merely guide their interpretations most scholars at various times draw comparisons with sites on frontiers other than their own, and this effort has been greatly facilitated by the Frontiers Congresses first established by Eric Birley in 1949. Yet questions of the kind mentioned above require a more systematic approach for they pertain to the ftm.ctioning of a system. This examination of fortifications as part of an evolving, empire-wide system has brought me to some conclusions which differ from those of other researchers whose comparative studies in the past have been more in the way of excursions from their main interest and have created as many difficulties as resolutions, especially with regard to third- and fourth-century fortifications. Phrases like 'type diocletian' and 'Valentinianic type' have been used rather loosely, and van Berchem once referred to an imperial 'blue-print' which was perhaps 'mechanically reproduced' tm.der Diocletian (in Bell e.t al. 1962: 21), all of which suggests some easily identified developmental chronology. Von Petrikovits burst this bubble in an important study which concluded that 'any tendency to date late Roman fortification on typological grounds •••(with a few exceptions) is worse than useless ••• ' 0971: 203). Certainly such categories as 'type diocletian' have been abused (Lander 1980), but von Petrikovits' pessimism is also the result of an overly simplistic approach. For example, he rightly criticizes some Danubian scholars who identified an attribute called the 'fan-shaped' tower and declared it to be Constantinian, for many sites thus dated later turned out on better evidence to be Diocletianic or Valentinianic (1971: 183-184). However, the problem was in accepting the 'type' in the first place, for in my research I have fotm.d that there are actually three easily distinguishable categories of the socalled 'fan-shaped' tower, and it turns out that each is in fact clustered within a usefully narrow range of both dates and locations. Another point of confusion has arisen from attempts to explain why the Romans adopted what has been deemed typical 'in late Roman fortifications generally' (Oelmann 1949: · 95): the construction of barracks against the inner face of the fort enclosure wall rather than in the free-standing blocks of earlier forts. Schonberger suggested a tactical purpose behind the change--to protect barracks from in-coming missiles sailing over the enclosure wall (1969: 182). Von Petrikovits concurred and added that the courtyard created by the new arrangement was perhaps intended to provide emergency shelter for refugees and livestock (1971: 202). Gichon, with a Near Eastern perspective, associated the provision of a courtyard with the location of cisterns (in Lander 1980: 1056). MacMullen associated the alteration in barrack-arrangements with the change in the materials used for enclosure walls, from 2
unhealthily damp turf and timber to masonry stone, after which change it took the Roman engineers over a century to adjust to the fact that the new stone enclosure walls could now be exploited for a more economical construction of barracks (1963: 43-44). Luttwak, on the other hand, has referred to the late barrack-arrangement as an economical way of thickening enclosure walls for tactical purposes (1976: 167). Thus for a seemingly simple though important question we see a variety of explanations, each of which may in fact have some validity for certain times and places, but again the present study will find evidence which compels a different explanation: that the change was not only a more gradual and less comprehensive development than is usually implied, but also that it was related above all else to changes in fort-size, which in turn relate to a process of sub-division within the Roman defense forces. I have not investigated deeply the question of where the Romans found the models for their changing designs for fort-plans, gate-dispositions, barrackarrangements and tower-shapes. Whether the Romans were influenced by ancient Israelite fortifications as Gichon is wont to imply (1967: 180; 1972: 44; and 1980: 852), or borrowed from the Greeks directly or by way of their own largely decorative civil fortifications, as von Petrikovits reasonably suggests (1971: 203), or were simply learning by trial and error as must somet i mes have been the case, the speculation on these matters seems far less important than discovering why the Romans ever felt compelled to change their designs at all. There is not much in it for aesthetics, yet tactical considerations can also be obscure. Von Petrikovits suggested that the Romans had to rethink thEdr defenses after the 260's when the Goths who were attacking Greece and Asia Minor showed some unwonted ability in pressing a siege with sophisticated techniques (ib. 193). However, Luttwak has argued convincingly t hat it was the very fact that Rome's enemies, at least in Europe and North Africa, could not usually boast of any real siege-capability that allowed the Romans to adopt an overall strategy based on the surviveability of isolated strong-points (1976: 134-136). The evidence I have collected supports Luttwak's thesis and forces attention away from purely tactical considerations to more strategic and even socio-economic problems faced by the Roman state. Thi s also helps to deflate t he view that changes in the Komans' own tactical ski lls, particularly in the use of artillery, served significantly to explain changes in fortification-design, especially tower-shape. This view was much overblown in the past, as some researchers are now discovering (Welsby 1982: 63-66) . In this study the comparisons of dated to undated sites increases information on developments in design and also allows broader and perhaps less risky conclusions about strategic planning under the various emperors. One new conclusion reached here is that while Valentinian may have continued the occupation of many existing forts in the hinterland behind the Rhine, no substantial hinterland sites were part of his vast program of fresh construction. This opposes the views of Staehelin (1948: 294), Schleiermacher (1951: 178-183), and Schonberger (1969: 182) among others and supports Luttwak's view that whenever a defense-in-depth policy 'showed any signs of enduring success, it was promptly abandoned' (1976: 132), or, in Valentinian's case, little was done to promote it. While this view seems to me far more consistent with other evidence about Valentinian's fortification and frontier policy, I support my argument by re3
dating certain key sites, notably Alzey and Kreuznach (though the argument can stand without them). This means that a more detailed examination-rare in this survey-is offered to explain why I feel that the datings which most authorities have adopted are less probable than those I propose. In other circumstances I make similar cases regarding certain features at a few sites such as Castell Collen, Dumer, and Visegrad. This may seem a Procrustean effort to make 'facts' to fit a pre-ordained scheme, but the evidence and arguments will have to speak for themselves. In general I have been a willing slave to the evidence provided and conclusions reached by three or more generations of dedicated European archaeologists specializing in winning information from the earth. It is perhaps because of the immediate and often pressing need for such work that none of the many scholars who are more amply qualified has carried out the type of study I have attempted. This broad survey approach is inevitably vulnerable for its lack of depth in many specific areas, but the long tradition of concentrating attention on individual sites, provinces or frontiers has provided only the foundation for a study which can begin the investigation into questions about Roman fortifications as part of a defensive system for an Empire.
Chapter One The Petrified Frontier (c. A.D. 50-138)
I. The Romans' Introduction to Fortifications Among civilized mankind's earliest constructions are fortifications protecting property, cities, states and empires. Historians and archaeologists can research through fifty centuries of fortifications, which are often the most prominent traces in a sometimes trackless past. In the middle of that time period and for ten centuries of their own the Romans often found need to build fortifications. Indeed, in the first of his one hundred and forty-two books on Roman history Livy tells that, according to the 'more common' legend, Romulus and Remus had their fatal argument over Remus' jeering act of jumping over his brother's half-built walls. Romulus, in a fit of rage, killed Remus and threatened, 'So perish whoever shall overleap my battlements.' He then went on to fortify the Palatine hill. A defensive earth barrier and a later stone wall surrounding the seven hills which became Rome were built respectively in the fifth and fourth centuries B.C.--not in the sixth, as the misnomer 'Servian' implies. The wall was substantial, 3.6 m. wide at base and fronted by a ditch 29.6 m. wide and 9.0 m. deep. The wall may have been a response to the Gaulish sack of the city in 386 B.C. It was a success thereafter: Hannibal never attempted a siege, and the 'Servian' wall, with some repairs and modifications, would remain in use until late in the first century B.C. (Todd 1978: 13-20). Throughout the republican period the Romans simultaneously absorbed other peoples' fortifications and built their own. Some of the finest creations of Greek fortification fell to the Romans as they expanded southward into Magna Graecia (Winter 1972; Lawrence 1979). The Romans repaired and rebuilt some Greek constructions, as at Paestum (Krischen 1941: 19ff.), and they had already in the fourth century B.C. built ashlar walls around their first mari- . time citizen colony at Ostia (Meiggs 1960: 22). By the time the Republic ended, the Romans were fully a part of the ancient tradition of civil fortification, though their overall attitude may have differed somewhat from that of the Greeks. Aristotle had believed that 'the strongest wall will be the truest soldierly precaution' and that 'those who have their cities surrounded by walls may either take advantage of them or not, but cities which are unwalled have no choice' (Politics 1331.1-9). Strabo reported that the Romans in the early Empire took the more critical view that 'men should defend the walls, not the walls the men' (234). The distinction may lie in the fact that a Greek city always had much to fear in the military capabilities of other
1 .
Jnchtu t hi l
Roman Britain
Q
.() Mel•~~f~ ' •si!•
j)onto.H
back to the 340's; and the evidence from Irgenhausen shows the same coin series as Wilten, though Gilles still dates it to the second half of the fourth century 0976: 446-447). The scattering of the legio I Martia's stamped bricks has given that legion a reputation as a 'Baubataillon' (E. Birley, in Swoboda 1972-73: 191, n. 48), and the presence of that legion must help to explain the coincidence of this isolated cluster in Raetia of late Roman forts with square projecting towers. Whether the legionaries actually built the other forts is unknown, though their tile-stamps did show up at Zurich (en. 13.12106). Whether these small forts are Tetrarchic or late Constantinian, and whether their builders were of the same generation as the builders of Kaiser-Augst, what is most intriguing is the possibility that the legio I Martia was actually-if perhaps not deliberately-introducing an Eastern style of fortification which would in fact be adopted only rarely in the West. Indeed, the instances grow more numerous as one moves eastward. On the Lower Danube have been found, on the one hand, a collection of Tetrarchic building inscriptions recording mostly repairs to existing forts (Kolendo 1966: 139-154) and, on the other hand, a collection of small forts with common features which resemble some of the Eastern and Raetian sites just discussed. Unfortunately, the two collections do not overlap, and apart from these distant analogies there is no firm evidence for dating any of these Lower Danubian sites (Gudea 1974: 177-179). Gornea, Ravna and Dierna (fig. 207), which are situated within a 20 km. radius along the Danube in Lower Moesia, are all square and of similar size (41.5, 40, and 35 m. square). Dierna has the widest wall (2.1 against 1.5 m. for the other two) and the most outwardly-projecting towers. Ravna has been much rebuilt and its original gate- arrangements are unknown, though it probably had only one gateway) certainly it has the smallest towers (6.2 m. wi de, compared with 9 and 9.2). The masonry of all three forts consisted of alternating rows of stone and brick. Gornea and Dierna were located on the north bank of the Danube, as was Putinei (fig. 208), which measures 40 by 100 m. (ib. 178). Its walls were 1.6 m. wide and had the same stone-and- brick masonry as the other forts. Not far from Ravna, in the estuary valley where the Porecka river flows northward into the Danube, the Romans built a long 2 m. wide rampart across the river and its valley , and on t he south side of the rampart was placed a fortification roughly 60 m. square (fig. 209). Its corner towers are 8 m. square, and like the other sites (except Dierna) the towers project only partially. Finally, about 300 km. east of this cluster of sites, the fortifications of Nova Cerna (fig. 210) are situated on the south bank of the Danube near Durosturum. Here are an outer and inner circuit, and though the latter has a length of only 35 m., it seems to possess at least one interval-tower, while its two cornertowers project in two different ways. Ivanov supposes that the two circuits are contemporary (1974: 68-69), but Scorpan is surely right to doubt this (1980a: 87). Excavations have so far gleaned third-century pottery and some coins which date back no further than Constantine (Miltev 1977: 357). Although Dierna, where a civil settlement pre-existed the fortification, yielded coins as early as Probus, it is interesting that, as at Nova Cerna, the earliest coins found at Putinei and Porecka Reka were issued by Constantine (Gudea 1974: 178 and Petrovif 1977: 262). Also, Petrovit made a strong comparison of Porecka Reka with Wil ten, judging both to be fortified supplydepots for the annona militaris, and Wilten, as we have seen, may date back as far as the 340's (Petrovit 1977: 265).
206
Innsbr uck · Wdten
Figure 206: Wilten, Irgenhausen and Schaan
0
0
0
0
0
0
0
0
0
0
0
0
a
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
c:;
0
0
0
0
0
0
0
c
0
0
0
0
0
0
0
0
0
0
0
0
0
0
lrgenhousen
Schoon
Figure 207: Gornea
Ravna
Dierna
\
' 50 m.
0
1:1 000
In North Africa, there are two small clusters, both in Nmnidia, of known late Roman forts with projecting square towers. We know that uprisings and incursions plagued this region in 292 and that in 297 the Augustus Maximian had personally to lead a campaign which employed detachments drawn from three legions, including - the II Herculia (Eutropius 1.23; Rachet 1970: 254-255). Maximian's cult image was Hercules (Veget. Epit. de ~ mil. 1.17), and it is probable that Maximian's victorious forces re-established Roman control by constructing new forts like Aquae Herculis and Aqua Viva (above, figg. 177 and 182). These square forts, located 30 km. apart, have barracks built against the enclosure walls which are barely a metre wide and constructed of mud-brick on a stone foundation. The corner-towers, about 6 m. wide on both forts, only partially project outward, and the interval-towers are smaller in both cases. Aqua Viva is much larger ( 86 m. square versus 52), and seems to replace a smaller stone fort nearby at Djebel Mechaieb (Baradez 1949: 297). The large fort at Zerbaret et-Tir (fig. 211), 114 by 158 m., also apparently replaced a smaller fort. In its tower- and gate- arrangements, it wall-width and mudbrick construction, Zerbaret closely parallels the other two forts. The interval between the towers exceeds 45 m. (ib. 298-299, pl. 125b). All of the features which are common to Zerbaret, Aqua Viva and Aquae Herculis are also found at Seba Mgata (fig. 212), except that it is a parallelogram in shape. At 60 by 84 m., Seba Mgata is just one quarter the size of Zerbaret et-Tir. About 70 km. southeast of this cluster of forts with square towers is another made up of Bourada (above, fig. 189), Drab Souid and es-Senam, all within a 20 km. radius and situated to replace the nearby Hadrianic fort of Gemellae. Bourada is, as we have seen, dated by inscription to the end of Constantine's reign, while three lamps bearing Constantinian monograms- were discovered at Drab Souid (Guey 1939: 205). Bourada measures 77 by 80 m., esSenam 80 m. square, and Drab Souid is reported to be of similar size; all were constructed of mud-brick (Fentress 1979: 84-85). The chief variation one can note in the use of square towers on late Roman fortifications is that some project fully and others do not. Of the former, seventy per cent are located in the East or Egypt: its seems to be an essentially Eastern characteristic brought west, perhaps by the new Tetrarchic legions. Most of the sites in the West reveal a continuance of the tradition of building towers which at least partially project internally. In general, the square tower is rare in the West in the late third/ ear1y fourth century and is found only in a few pockets. B. Rounded Towers The towers which are here called 'rounded' are those which have a curving end-wall but are not circular since they have two side-walls which are more or less parallel. Three definable sub-divisions can be distinguished, and the first of these will be called 'rounded/ square-backed' towers, described by von Petrikovits as 'rectangular towers with rounded face which did not project totally, but were so bonded into the wall that their rearward side looked like a reinforcement of the wall' (1971: 199). He added that the temptation to seek a narrow time- period for such a distinctive type must be resisted since the known examples seemed to range widely in date, from Diocletian to Valentinian. This objection is still valid, though it may be modified slightly. 208
.·
.....
Figure 20.9: Porecka Reka
Figure 208: Putinei
0
50 m.
1:1000
Figure 210: Nova Cerna
--· - I
- i
20~ ·
.
Figure 211: Zerbare t et-Tir
OL....--t....-~o...--· ~.
100 m.
1:2000
Figure 212: Seba Mgata 0
100 m.
1:2000
Figure 213: - - - Beauvais
o_--'----::-7"~)1 : 5000
Of the sixteen sites which are discussed below, it should be noted that none are found in Africa, Egypt; or the East, and those in Britain are only coastal reconnaissance posts. Only three examples come from the Danube frontier (all Pannonian .and situated within a radius of 100 km.). The main Rhine frontier offers only one such site, Alzey, but Raetia, again, yields a cluster of six sites, and Gaul has at least four towns with such towers, which indeed may be models for the rest. Three of the sites in Gaul are within a radius of 30 km. Beauvais (fig. 213), Senlis (fig. 214), and Soissons each have interval-towers which project outwardly 3 to 3.6 m., inwardly only 0.8 to 1.3 m., and which have about 6 m. of width. Other features of lay-out and construction have led Johnson to suggest that all three are the work of the same builders, or at least builders using the same plan, and he believes construction took place shortly after 285 (1973: 210-217). Carcassonne in southern Gaul seems to share many of the same features, but there is no useful dating information (Butler 1961: 29-30). Alzey (above, fig. 191), a fort near Mainz yet still 25 km. west of the Rhine, is important not only because it has been excavated but also because three other Upper Rhineland forts, Burg bei Stein (above, fig. 170), Horbourg (fig. 215) and Yverdon (fig. 216) have been closely compared to Alzey in terms of their rectangular or rhomboid plans, their 2.7-3.0 m. walls, their circular corner-towers, and their rounded/square-backed interval-towers. Burg bei Stein was, as we have seen, probably constructed around 294. Yverdon and Horbourg, giving little clue themselves, have been dated to Diocletian on analogy with Burg bei Stein or to Valentinian on analogy to Alzey. However, the Valentinianic dating of Alzey is far from certain (the see the discussion below, pp. 276-280). Of three other Raetian forts with such towers, Arbon (fig. 217) has been associated with the Tetrarchy to the extent that the Notitia Dignitatum states that its garrison was a cohors Herculeae Pannoniorum (occ. 35.34); Zurzach (fig. 218) is thought to belong to the first half of the fourth century, though there are at least two phases here in the two adjacent sites (von Petrikovits 1971: 212); and Zurich (above, fig. 205), as we have seen, may date to Constantine or his sons. These three irregularly shaped forts have only 15-18 m. between the towers, unlike the 32-35 m. spacings seen at Alzey, Horbourg and Yverdon. In Pannonia, the town of Scarbantia was located 50 km. south of the Danube at a point where a northbound route diverged before reaching Vindobona and Carnuntum. It is believed that the town was fortified (fig. 219) and a cohort placed there around 304 (above, p. 193>· The towers are basically the same as Alzey's and the lay-out of the town walls resembles that seen at Beauvais: , the builders seemed inclined to plan a military-style rectangle, but they then shortened the perimeter by lopping off the corners. At Castra ad Herculem (above, fig. 178) the rounded/ square-backed towers vary in size and do not bond into the enclosure wall, and so they are thought to be part of a secondary phase of construction (Soproni 1978: 46-48, 102). The primary phase is represented by the two towers on the east end of the fort, which are ordinary rounded towers bonded to the enclosure wall but not projecting behind it. If the fort's name suggest~ that the fort was begun during the Tetrarchy, the secondary phase is simply later, and not necessarily Valentinianic, as Nagy proposes without citing evidence (1971: 150, n. 14). 211
Figure 214:
0
1 : 5000
Figure 215 : Hor bourg
0
lOO m.
1:2000
Figure 216 : Yverdon
212'
' ·. ' -,
\
I I I
>~ / 1\ ~>/ , _______.... ( / // /
1:1000
'
/
',\~
0
\
•
/
I
I 1
,
~ ', , ...., / /> , \\ .....
41.00
Figure 269: Ran el-Qattar
0
50 m.
1:1000
__
.....
IOssv'
----------------
..
...~~
::::====~===---
•·
'=~--... .......,_ -----------------~
Figure 270: Zwentendorf, splayed fan tower
10 m. I
1:200 250
Figure 271: Campona
~ :
,. ~ .
~:
~:
Figure 272: Intercisa
~! .."'=" ;
: i
..
"I : ' :a.. ;
251
construction of the reign of Constantine (1959: 133), but later it became clear that at least three building phases could be seen in the west gateway of the fort (1978: 58-59) (fig. 273). The earliest phase included small rounded towers which are different in plan from those rounded towers found at intervals along the enclosure, which must almost certainly be a secondary wall. I believe that the simple rounded tower near the gateway is a remnant of a fortification which was largely replaced-towers and walls-at a later date, perhaps during the reign of Valentinian. It was during this time that a watch-tower located 3 km. southwest of Visegrad was dedicated by the Legio I Martia in 372 (Soproni 1967: 138-143). Probably to be included among sites with splayed fan towers are two other Pannonian forts which, according to Mocsy, possessed 'fan-shaped' towers: Azaum and Celemantia ( 1962: 637 and 642). The stonework of both forts dates back to the second century, and both were certainly occupied into Valentinian's reign, as indicated by tile-stamps used in reconstructions at both sites (Soprani 1978: 159 and 164). I have not seen plans for either site, but presume the towers were splayed fans. Something very like a splayed fan tower can be seen in Upper Moesia at the long-used fort · of Drobeta (fig. 274), which underwent at least three major phases of construction. There is a destruction level which dates to the late third century, and during some subsequent phase of reconstruction Drobeta not only received fan towers but also three large rectangular bastions to restrict the number of entrances (see also p. 171 above). Similar rectangular bastions are also known at Visegrad (fig. 273) and Zwentendorf (fig. 233), and more rounded bastions were added at Intercisa (fig. 275) and Campona (fig. 271). In all cases the purpose was the same, to close off some of the numerous entrances on these old forts now viewed as security risks. If Drobeta's bastions are contemporary w'ith those at the other sites mentioned, and if Drobeta's splayed fan towers are contemporary with it's bastions, then the fan towers may well be Valentinianic. Finally, in Britain, the towns of Ancaster and Godmanchester received as additions to their third-century walls solid projecting bastions which protected angles in the curtain and which have been called 'fan shaped'. In both cases the bastions are fourth-century in date, though their dating remains imprecise and their relationship to the development of fan shaped towers on the Danubian forts is unknown and presumably tenuous (fig. 276) (Todd 1975; Green 197 5; and Brit. 197 5: 251). VI. A Summary on Late Third and Early Fourth-century Fortifications The evidence in this chapter confirms the historical information that Diocletian and Constantine put in motion massive programs for fortification construction. There was a need for new types of forts in new locations so that the restored frontiers would also provide a defense-in-depth to protect civilian centers and road systems in the event of new incursions. Yet the uniformity does not extend beyond this recognition of need. First, there are such disparities as the fact that timber fortifications were still constructed in the Lower Rhine at Oudenburg, Heumensoord, Asperden, Villenhaus and Goudsberg, even in the later third and early fourth centuries. And in late Roman Britain forts were_still being laid out with rounded corners and no projecting 252
Figure 273: Visegrad, - : .. \
three-phased gateway 30 m.
0
1:600
Figure 274: Drobeta
0
100 m.
1:2000
=--=---:.-'11''11"==--y
Figure 275: Intercisa, late bastion
10 m. I
1:200
253
Figure 276: Ancaster, n.w. angle
Godmanchester, s.e. angle
~
~
~
Town -:;::.--- d i t c h
~
,
l
;---
,,
i ,·
I'
I I
. '
:
'
!
t. .. · ~.":" .... / I ... •
,' '
{
.
'
/
1:2000
I
Figure 277 : Burgle bei Gundremmingen
0
50 m.
1:1000 .:.::.·~ "':'"·:_;..~.-, r. er r r 1 r. r ,
1
.
;·7-~
-254
towers in County Durham and Yorkshire (Piercebridge and possibly Newton Kyme and Elslack) and along the Saxon Shore. Second, the above survey of attributes would seem to suggest that the specifics of fortification-design were never established on an Empire-wide basis. Twelve different categories and sub-divisions of tower shape have been distinguished, and if for the moment we divide the imperial frontiers into four sectors--Britain and the Rhine, the Danube, Egypt and the East, and North Africa--we discover that only one of the twelve tower shapes is found in all four sectors: not surprisingly, it is the traditional square tower, though now, of course, projecting outward to a varying degree. Four tower shapes can be found simultaneously in only two sectors, and five tower shapes are not known outside a particular sector. The clustering appears to be more particular still: the majority of polygonal towers are found in a single province, Britain. Lower Moesia has a majority of the 'large fan' towers and nearly a majority of all U-shaped towers. All 'simple rounded' examples come from Pannonia; all 'external circular' interval-towers from the Middle Danube; most 'splayed fan' towers from Pannonia or Upper Moesia. Apparently all of the late sites in North Africa had square towers only and were located in two clusters of apparently quite different dates. It is interesting that the nine late fortifications known to exist along that stretch of the strata Diocletiana which runs between Damascus and Palmyra show more than a single tower type. Only the two largest forts, Dumer and Han al-Manqoura, possessed interval-towers (U-shaped) along with 'large fan' towers at the angles. The other seven forts are in two groups: the northernmost four possessed 'small fan' towers (though in the case of Han al-Basiri this is mainly a guess) and the southernmost three possessed square cornertowers. Here there seem to be two building programs separated perhaps- by a brief period of time. Or perhaps a garrison stationed in Palmyra built the northern sites and the southern ones were built by a unit based in Damascus or Bostra. Another possibility is that the distinction is ftmctional: if van Berchem's correlation of seven of the sites with locations and tmits listed in the Notitia Dignitatum is correct, it may be significant that thte northern sites seem at one time to have been garrisoned mainly by ala tmits, while the southern sites held mainly cohortes (1952: 13-16; Not. Dign., or. 32).
A. Toward a Chronology of Towers of the Late Third
and Early Fourth Centuries It is worth elaborating a fact occasionally noted in this chapter:
that some of the twelve categories of tower shapes are discovered as interval- and angle-tower combinations. A large number of sites with U-shaped intervaltowers also possess 'large fan' corner-towers (Libida, Abrittus, Capidava, Troesmis, Dinogetia, Han al-Manqoura, Dumer, Lejjun and Udhruh). Some of these are among the nine sites pos~essing both U-shaped towers and a phrourion. At Contra .Aquincum and Iovia, 'simple rounded' interval-towers are found in conjunction with 'small fan' corner-towers. At Horbourg, Yverdon, and Zurzach, the 'rounded/square-backed' interval towers are found with circular towers at the angles. Most forts with 'splayed fan' towers had undergone reconstruct ions which included rectangular bastions restricting gateways. Several late Roman sites reveal late designs replaced by even later designs, and these provide the basis for some attempt at seriation. If this 255
evidence is combined with external dating evidence, it is possible to achieve nearly absolute dating for several categories of tower shape, though exceptions persist. First of all, several forts possessed the 'simple rounded' interval-towers which were later replaced, or supplemented, either by 'rounded/ square-backed' towers (Castra ad Herculem), U-s haped towers (Tropaeum Traiani), external circular towers (Iovia), or splayed fan shaped towers (Visegrad). Moreover, this category of 'simple ro•mded' towers is nowhere known to be posterior to any of the other eleven categories--unless possibly at Qasr Qarun, whose rounded towers may be secondary to the rectangular ones. As noted above, the first three sites seem genuinely datable to Diocletian, as does Contra Aquincum, whose simple rounded towers were never replaced. And since Iovia and Contra Aquincum show simple rounded towers in conjunction with 'small fan' shaped corner towers, the four Syrian forts which happen to lack interval towers while possessing 'small fan' corner towers may add circumstantial support for this dating since they are located on the strata Diocletiana. The 'simple rounded' towers at Tropaeum Traiani were begun sometime after 288/289; Contra ~uincum was built in 294; and the Syrian forts may date to shortly after 298. Of the four tower shapes found in construction phases secondary to that of the 'simple rounded' tower, the U-shaped tower is most similar in design, incidence, and apparent chronology. At Tropaeum Traiani, the U-shaped towers were a secondary revision of a still tmfinished plan. And just as 'simple rounded' towers are often associated with 'small fan' corner-towers, so Ushaped interval-towers are often to be found with 'large fan' corner-towers. All four shapes occur seemingly together along the strata Diocletiana. The cluster of sites with U-shaped towers in Lower Moesia (including Tropaeum Traiani) seem to belong to a building program begun between 293 and 308, and those in Egypt seem to date between 295 and 306. The primary 'simple rounded' towers at Castra ad Herculem are probably Diocletianic, and that fort's 'rounded/square-backed' towers are secondary, but how much later is unknown. Elsewhere the dating for rounded/ square-backed towers in some Gallic town walls has been noted as shortly after 285; one Raetian example is dated to 294, while five others are thought to be later on evidence and within contexts which set them apart. Scarbantia's towers may date to 304; extremely large versions of this tower at Tokod and extremely small ones at Scarborough are well-dated to Valentinian. No firm conclusion can be drawn from this evidence, though one interpretation is that the 'rounded/ square-backed' tower is primarily Diocletianic, but later builders occasionally followed the model with noticeable adjustments: perhaps the straddling effect of this tower shape was structurally useful for stabilizing relatively narrow walls (like those at Tokod and Scarborough) or walls with long curtains between the towers (like the Gallic town walls). Of our three sub-divisions of circular towers, only the 'external circular' type is to be found in a secondary relation to the 'simple rounded' and 'small fan' towers at Iovia. Apart from this evidence, 'external circular' towers
1Mil estones near the Syrian forts date 293-303, according to Poidebard (1934: SO).
256
remain undated here as elsewhere. Circular towers of the 'straddling' and 'semi-circular' sub-divisions seem mainly to be Constantinian, though circular corner- towers are found in association with 'rounded/ square-backed' towers not only at Alzey, Horbourg, Yverdon, Arbon and Zurzach, which seem to be Constantinian, but also at Burg bei Stein, which is certainly Diocletianic. The last example of a tower shape which shows itself in a secondary relation to 'simple rounded' towers is at Visegrad, which has 'splayed fan' angle towers in its secondary or tertiary phase. While such towers are never found on forts as part of the primary construction, Visegrad is unique in that a major part of its enclosure wall seems also to have been reconstructed at the time when the 'splayed fan' towers were first employed. Whether this indicates that Visegrad was a prototype or final evolution is impossible to say. The bulk of the evidence, none of it firm, suggests that such towers may date as late as Valentinian. Therefore the notion that all 'fan shaped' towers are Constantinian seems quite wrong; in fact, most 'small fan' and 'large fan' towers are most likely Diocletianic, and 'splayed fan' towers are probably Valentinianic. Square tower s a r e found on many forts which ar e indisputably dated to Diocletian, but also on several forts which are just as indisputably Constantinian. As we shall see, square towers also appear on forts which were built by Constantius II or Valentinian. Finally, the category of polygonal towers seems t o be Tetrarchic, though the examples · are few and most come from Britain, where, indeed, a Severan example at Risingham seems to have established a provi ncial tradition, reinforced by the purely decorative transformation of York ar ound the tim~ when that fortress became the base of operations for the second most powerful man in the Tetrarchy, Constantius Chlorus.
B. Design Changes as Tactical Improvements Though the i s sue of the extent to which design changes were the result of tactical considerati ons wi ll be placed in a broader perspective in the Conclusion of this s t udy, some particular points about developments of the third and early four t h centuries must be brought out at this point. On an imperial scale, the fortification building programs of Diocletian and Constantine appear to be less systematic than we might have expected judging from the apparent need and from the contemporary references. Nearly two centuries earlier Hadrian's constructions, realizing his conception of a preclusive linear system of defense at a time relatively free from stress, seem more uniform by far. Undoubtedly many forts of 'traditional' design were kept in use without major modifications by Diocletian and Constantine. Ammianus Marcellinus reports that Julian, as Caesar, re-occupied in 357 a Rhineland fort which had been built by Trajan and which had only recently been overrun by the Alamanni; Julian saw to its renovation 'with mural artillery and strong apparatus' (17 .1.12). And in Britain, not only did Hadrian's Wall function into t he fourth century with relatively little a1 teration to the defensive design of its forts, but even during the Tetrarchic period of great innovations in tow er design newly constructed forts like Piercebridge could still be laid out in the old style, with rounded corners and a lack of projecting towers, but also with a greater wall-width.
257
Some of the changes in fortification design seem more co~etic or psychological than defensive. York's late defenses were powerful-looking, yet were only a facade for one side of the fortress. The defenses of Luxor, around an ancient temple, seem excessive for the region. Some developments in internal arrangements, especially the lay-out of the principia, suggest some propagandistic fusion of the military cult of the insignia and emperor-worship (MacMullen 1963: 43-48; Fellmann 1976: 190- 191 and 1979: 52-53). On some frontiers innovation may have been stunted for lack of either threat or changes in deployment. Britain's Saxon Shore forts were innovative and date to a period just before or during Diocletian's reign; but after that, new fortification in Britain is largely a matter of minor repair, with an eccentric employment of polygonal towers. In North Africa, tower-design never shook loose from the dominance of the traditional square shape. 1. Towers and Artillery The different categories of tower shape have been shown to be more regionalized than not, and there is little to be found of any developmental relation. True, square towers no longer monopolize the field, but neither do they become extinct. A development, early in Diocletian's reign perhaps, of using 'simple rounded' towers at intervals was soon found unsatisfactory along the Danube where these towers were replaced in some places by U-shaped and circular towers. No one seems to have said definitively that one shape was better than another and therefore should be employed everywhere in the Empire. A change in the Romans' use of artillery may account for some deve1oP.lllents in tower design. It is clear that early in the Empire's history Roman engineers had conceived of towers as surveillance points: when they wanted artillery platforms they built them specially and named them so, as with the ballistarium at High Rochester in 220 (RIB 1280-1281). But a change occurred at some time before the middle of the fourth century, by which time the Roman military had widely adopted as a defensive weapon a one-armed stone-throwing catapult which was much greater in size than the arrow-firing or stone-shooting ballistae which were standard during the early Empire (Marsden 1969a: 188198). Although two centuries earlier Trajan's famous engineer from the East, Apollodorus of Damascus, had referred to a one-armed stone-thrower (ib. 1969b: 249), such machines are not mentioned again until the time of Ammianus Marcellinus, Vegetius and the Anonymous author of De Rebus Bellicis, all writing in the second half of the fourth cent;ury. Marsden is uncertain when the onearmed stone-thrower became so important to the Roman army, and therefore to our sources, but the weapon itself gives clues. First of all, it is a defensive weapon almost exclusively: it is large, heavy, and rarely mobile, unlike the ballistae so extensively used during the Republic and early Empire and no doubt still in use in the later period. So, large stone-throwers are likely to have been adopted at a time when at least one part of the army had a permanently defensive role, like the Tetrarchy's limitanei. Secondly, the onearmed stone-throwers were best located on a platform both large and stable. The frame of such weapons was massive: a modern reconstruction based on ancient descriptions weighed two tons (Payne-Gallwey 1958: App. p. 18). The recoil, according to Ammianus, would break up whatever surface the machine stood upon, so a buffer of turf or brick was necessary; and to snap the trigger the master artilleryman used a hammer, probably with a long handle, and needed space so he could stand well to the side ( 23.4.5-6). One preferred 258
to give the machine-known by uncomforting names like 'scorpion' and 'wild ass'-plenty of room. Ammianus relates how an improperly loaded machine let loose a ricochet which smashed a technician's body into unrecognizable bits (24.4.28). So, such weapons would need large towers, whose height and projection, moreover, would substantially increase the range of the device. Though such weapons had some anti-personnel value-and, again, Ammianus describes how a Gothic attack was halted by the effect of one boulder flying into the enemy's midst, hitting no one but stunning all (31.15.12)-a major use for such weapons was to break up the siege-equipment of attackers ( 19.7.67). Evidence from Vegetius can be interpreted to suggest that a unit of ten or eleven men was needed to maintain and operate each machine, and it appears that the technicians who built such machines were, like the engineers who constructed the forts themselves, under the orders of the praefectus castrorum (Veg. 2.10). In short, the size of the machines, the space needed for operation, the stress of their recoil, and the desire to improve their range-all called for larger projecting towers. One may say, despite numerous exceptions, that in general larger and more greatly projecting towers are to be found in the East and along the Danube than on the Rhine, in Britain or in North Africa. The tower categories which predominate in the West are often only partial in their projection: 'rounded/square-backed', circular 'straddlers', semi-circular, and polygonal towers. The categories most favoured in the Lower Danube region and in the East include the fully-projecting 'external circular', 'U-shaped', and various 'fan-shaped' towers. The divergence of preference no doubt results in part from the greater threat posed by Rome's enemies across the Danube and Euphrates. It may be that Roman forces in the East had a deeper reliance on defensive measures, including fortifications and artillery, than did the troops in the West. Ammiantis tells an interesting tale of troops recently brought to the East from Gaul and now trapped, along with Eastern veterans, by the Persian siege of Amida in 359: the Western troops were useless in countersiege operations and preferred to sally forth, though they suffered heavy casual ties before the commandant refused to let them out again (19.5.2). One should add that whether artillery was a major factor or not, projecting towers were tactically important in allowing defenders to fire-with bows, spears or rocks-on assailants close to the walls. We have already noted that it was in the vicinity of gateways and blind angles that towers were first given major projection as early as the second century. In short, towers were no longer simply for surveillance, . and tower-projection, which began for structural reasons and continued for tactical reasons quite separate from the use of artillery, reached its apogee when and where the use of heavy catapults made this necessary. 2. Barrack Arrangements One development often mentioned in connection with late Roman fortification is the construction of barracks not in the traditional blocks, but in rows attached to the enclosure wall (Oelmann 1949: 95). I referred in my Introduction (pp. 2-3) to the surprising variety of explanations for this development which have been offered recently by different scholars. While these explanations suggest a fairly comprehensive change consciously wrought, the evidence compels certain q~lifications: 259
1. Construction of barracks against walls begins before the 'late' period of fortification building. 2. The development is a much more prominent feature m the East and North Afri ca than in Britain or on the Rhine. 3 . The feat u re i s associated almost exclusively with 'small' forts and fortlets. 4 . The number of examples is considerably reduced if one distinguishes between sites where the structures against the walls are certainly barracks and sites where they serve some different purpose. Tisavar in North Africa, dating between 182 and 192, and the Severan sites of Qasr el-Hallabat and Uweinid in Arabia, are early examples of small sites with barracks built along the line of the enclosure. Other sites in Arabia and Syria which are 'late' and which suggest such arrangements-Qasr Bshir, Deir el- Kahf, the upper fort of Muhattet el-Haj, and most of the 'bans' along the strata Diocletiana-are 'small', with side-lengths measuring between 40 andn 65 m. But the larger sites in the East, like Dumer, Lejjun and Udhruh, which are contemporary with the smaller sites but have side-lengths between 180 and 250 m, do not have barracks built against their enclosure walls. Significantly, sites of intermediate size, like Da'janiya in Arabia and Qasr Qarun in Egypt, wi th side-lengths between 80 and 100 m., have some accommodations against the walls and some in the traditional block-formation. In comparison with desert sites, a smaller proportion of European sites has yielded clear evidence about internal arrangements in their primary phase. Most new constructions of the 'late' period in Europe still seem to possess the open roadway- the traditional via sagularis-between the enclosure wall and the location of other structures, especially barracks. This is clear at Deutz, Haus Burgel, Kaiser-Augst, Zurzach, Castra ad Herculem, and Lympne, and seems to be the case also at the road-forts of Bitburg, JUnkerath and Neumagen. Some earlier sites which are known to have undergone considerable reconstruction in the late period did not in the end receive new barrack arrangements : the wall s are still free at Intercisa, Drobeta, Campona and Zwentendorf. On some l ate sites where structures lean on the enclosure wall these ar e not barracks but granaries (Lejjun, Betmauer, Lorenzburg, Wil ten and perhaps Schaan), bath-houses (Boppard and Schaan), or something as yet unidentifi ed (Famar s, Richborough, High Rochester, Ilkley and Bewcastle). Barracks can be found built against the walls of small European forts-like the two clusters of guadriburgia in Raetia and on the Lower Danube (Schaan [perhaps], Irgenhausen, Gomea, Ravna, Dierna, Putinei, and Porecka Reka) which so resemble Eastern sites and which have side-lengths measuring only between 35 and 65 , m. (except Putinei, an odd 40 by 100 m.). Not mentioned in this list are five other examples of European sites with barracks built against their enclosure walls, but these arrangements seem to date later than Diocletian and Constantine. Burgle bei Gundremmingen (fig. 277) is, even by Raetian standards, a peculiar-looking site: von Petrikovits is not sure whether to call it a frontier fort or a road- fort (1971: 188), and coin evidence suggests a construction date in the reign of Constantius II (Bersu 1964). Cuijk, Alzey, and Kreuznach are late forts located along the Rhine: they have barracks built against their enclosure walls, but they are far larger than most of the Raetian and Lower Danubian sites with such arrangements. However, it will be argued below that the barrack arrangements in these three forts are the result
260
of massive reconstructions and are not part of the original Tetrarchic or Constantinian plans for these forts. Finally, Altrip (fig. 279), which certainly had barracks built against its enclosure walls from its beginning, is only half the size of Cuijk, a third the size of Kreuznach and Alzey, and is attested as a construction of Valentinian. One can conclude that for late Roman fortifications with side-lengths over 100 m., the feature of barracks against walls did not become standard anywhere in Diocletian's or Constantine's Empire, yet for smaller forts such arrangements had become the norm. What has changed is not simply the Roman attitude toward thickening walls, avoiding enemy missiles, economising on construction materials, or making room for refugees--though these could be secondary factors. What has changed is the average size of Roman forts. Early stone forts, even for the quingenary cohorts, were rarely smaller than 1.0 ha., yet in response to a gradual process referred to earlier as the sub-division of the Roman army, frontier fortifications decreased in size so that a very large minority, and perhaps a majority, ~f new construct ions in the late Roman period are less than 1.0 ha. in size.
3. Siting and Outworks The provision of defenses for locations behind the frontiers has a long history, and in the early Empire much of the army in the East had regularly been garrisoned in major towns behind the limes, while in the West legionary bases were usually a considerable distance behind the auxiliary frontier forces. Many cities had been fortified before the third century for reasons of pride as much as protection, and after the Marcomannic Wars and the civil walls of Marcus Aurelius and Septimius Severus, towns across the Empire, and notably in Britain, girdled themselves in stone. But with the attacks of Saxons, Alamanni, Goths and Persians between 230 and 270, town fortifications increased rapidly. More troops were garrisoned in towns and along the roads and rivers linking civilian suppliers and military consumers. In the early Empire one or two ditches provided the main outworks for frontier forts, and in the late Empire fort-builders took greater advantage of a hill's slope or a river's bank to provide natural defenses for their sites. But al so in the late period the narrow berm and the one or two ditches 'were replaced by wide, flat-bottomed ditches and wide berms' (von Ietrikovits 1971: 197). The myriad of examples can be reduced to an average : in most cases these berms stretched between 8 and 15 m. from the curtain wall, and the ditches were 2.Q-4.0 m. deep and 5-15 m. wide. Von Petrikovits explains that the new style in outworks was intended 'to keep the enemy's siege engines and artillery ;yay from the wall' (1971: 197), but Luttwak adds the important qualification that probably the Persians were the only enemy who knew how to employ engines and artillery, and that the barbarians elsewhere could at best 1an areal sizes for forts in the first and second centuries, see: (Rhineland sites) Baatz 1974: 25; (Hadrian's Wall forts) Breeze and Dobson 1978: 4445; and (sites in Wales) Jarrett 1%9: 150-152).
~e lack of excavation of Eastern sites means that the main evidence for ditches at such sites comes from Poidebard's photographs (1934). These suggest that many desert forts simply had no ditches, while a few clearly did. 261
improvise a battering ram 0976: 135-163). Wider berms and ditches may have been intended to enable small garrisons to fend off more easily just such attacks. More likely, however, the berms were extended simply because towers were now projecting farther from the wall. And the complex of small ditches which had multiplied over the decades would, if simply duplicated at a greater distance from the fort, do nothing more than provide assaillants with cover behind the ditches' baulks over which the defenders would no longer have a clear view. It was far more sensible, and easier, to excavate a single broad ditch, which provided the defenders with a better field of fire, and which, in the right terrain, would confront the enemy with a muddy approach. These are worthwhile measures, whether or not the enemy has artillery or rams. It is interesting that the late Roman fort at Piercebridge in Britain had a wall 3 m. wide and a ditch 8.7 m. wide, separated by a berm 7 m. wide, though the berm can not be explained as providing support for projecting towers since the fort has none ( Scott 1977: 32-33). The mystery was solved in 197 9 when the berm was found to be covered with lilia, neatly cut rectangular pits which would have been filled with sharpened stakes and covered to deceive an attacker. Rough Castle in Scotland is the only other site in the Empire known to possess this type of anti-personnel device, which, in fact, is described for us by Julius Caesar (BC 7 .73).
262
Chapter Five The Last Frontier (337-410) As testimony to the overall success of Diocletian's and Constantine's consolidation of t he Empire and its frontiers one can offer the fact that Rome's enemies took surprisingly little advantage of the civil wars and usurpations which di srupted the two decades following the death of Constantine the Great in 337. The Persians pressured Armenia throughout this period, but would overcome Roman resistance only during the confusion of Julian's retreat and death in 363. Between 352 and 358 Alamanni and Franks crossed the Rhine, and Quadi and Sarmati breached the Danube, yet all had been driven back before Julian turned hi s attention to the East in 361. However, the failure of his Eastern campaign and the mercifully brief chaos of Jovian's seven-month reign seem to have stimulated Rome's enemies to new efforts. The joint reign of Valentinian (364- 37 5) and his brother Valens (364-37 8) witnessed barbarian attacks on nearl y every frontier, and Valentinian in particular is responsible for meeting this threat with a new conception of frontier defense. I. The Perception of Frontier Requirements by Valentinian and his Contemporaries Although Valenti nian's effort would mark the last program for defending the Empire when it was still a complete entity, the f i rst step in his politicostratigic system was a full sharing of power wi~h his brother Valens--a step probably taken willingly, but also at the behest of the general staff (Am. Marc. 26.4.1- 4). The fact that Diocletian's Tetrarchy had failed to preserve political stability did not negate the strategic sense of dual emperorship. And for a solution to the chronic problem of succession Valentinian would rely forthrightly on t he dynastic principle. Such eclecticism is evident also in Valentinian' s frontier policy, which combined a construction-program as concentrated at Diocletian's with a system of linear reconnaissance like Hadrian's and a certain aggressiveness reminiscent of Trajan, or, more likely, of Julian. Yet it was present circumstances which inspired Valentinian, and apparently others, to seek new solutions. A certain Anonymous wrote an essay known to us as De Rebus Bellicis, which was addressed to an emperor who was probably Valens (Cameron 1979: 1-7; cf. Thompson 1953: 2-4). Half of the work provided descriptions of weapons, some novel and some ancient but with clever modifications. The descriptions were sketchy and relied heavily on accompanying illustrations, but this entire half of the work seems like sugar-coating for the strong medicine which the Anonymous prescribed for the ailing Empire in the rest of the essay. It was not simply the barbarian threat which debilitated the Romans, Anonymous argued, but an economy weakened either directly through the government's heavy taxes, poor monetary policies, and wasted largesses to soldiers and bureaucrats, or indirectly through the government's failure to restrain corrupt administrators, to re-codify the law, and to force those most able to bear the cost to
263
1 Roman Britain
Ho.l t•~ tcr•
HoultllooWS
l t -$ttcl
.o
ID
~---",/
.tiloMttu·s
Sites Mentioned in Chapter Five
265
3
The Danube Frontier
, ()
•
.-t~s
/()(J
J;jfplf'tef~'J / ~()
Sites Mentioned in Chapter Five
~"" )f()-€..5 ;4,_
,.. /
J11/er/1J r/
N
/
/
/
J(
...: = t CI.J1J'? O J1 t4_
~"?MO"'fl'.4
'\ I
S u,P.Lrro l"'
I \ /
' ', /
/
?fo.ef;~
/
-
/
I
,..
I·Ucl