Studies on Roman Pottery of the Provinces of Africa Proconsularis and Byzacena (Tunisia): Homage À Michael Bonifay 1887829768, 9781887829762

Citation preview

JOURNAL OF ROMAN ARCHAEOLOGY? JRA® SUPPLEMENTARY

ISBN 1-887829-76-8 ISBN-13: 978-1-887829-76-2

SERIES NUMBER

76

ISSN 1063-4304 (for the supplementary series)

Copyright © 2009 Journal of Roman Archaeology, L.L.C.

Printed by Thomson-Shore, Dexter, Michigan

JRA® and Journal of Roman Archaeology are registered trademarks of Journal of Roman Archaeology, L.L.C.

This and other supplements to the Journal of Roman Archaeology may be ordered from: JRA, 95 Peleg Road, Portsmouth, RI 02871, U.S.A.

Telephone (+USA) 401 683 1955

telefax (+USA) 401 6831975 e-mail: [email protected] Web site: JournalofRomanArch.com

Reprints of all JRA works available at nominal cost by sending a message to [email protected] Permission to copy may be obtained only direct from JRA, by e-mail, letter, fax or phone.

Kindly note that the Copyright Clearance Center (USA), the Copyright Licensing Agency (UK), and other national Reproduction Rights Organizations are not permitted to authorize copying or to collect fees for doing so.

STUDIES ON ROMAN POTTERY OF THE PROVINCES OF AFRICA PROCONSULARIS AND BYZACENA (TUNISIA) HOMMAGE

À MICHEL BONIFAY

edited by J. H. Humphrey with contributions by P. Bes, K. Carr, M. G. Fulford, J. Lund, M. Mackensen, J. T. Pena, J. Poblome, J. J. Rossiter, D. L. Stone, & R. S. Tomber

PORTSMOUTH, RHODE ISLAND 2009

ADDRESSES

OF CONTRIBUTORS

P. Bes

[email protected] Icrates Project, Katholieke Universiteit Leuven, Blijde-Inkomststraat 31 Bus 3314, BE-3000 Leuven

K. Carr

[email protected] Department of History, Portland State University, Portland, Oregon

M. G. Fulford

[email protected] Department of Archaeology, University of Reading, Reading

J. Lund

[email protected]

Collection of Classical and Near Eastern Antiquities, The National Museum of Denmark, Frederiksholms Kanal 12, DK-1220 M. Mackensen

Köbenhavn

K.

[email protected]

Institut für Vor- u. Frühgeschichtliche Archäologie und Provinzialrömische Archäologie, Ludwigs-Maximilians-Universität, München J. T. Peña

[email protected] Department

of Classics, 7233 Dwinelle

Hall #2520, University

of California,

Berkeley J. Poblome

[email protected] Sagalassos Archaeological Research Project, Katholieke Universiteit Leuven, Blijde-Inkomststraat 21 Bus 3314, BE-3000 Leuven

J. J. Rossiter

[email protected] Department of History and Classics, University of Alberta, Edmonton, Alberta

D. L. Stone R. S. Tomber

[email protected] Department of Classics, Florida State University, Tallahassee, [email protected] Department of Conservation and London WC1B 3DG

Scientific

Research,

The

Florida

British

Museum,

TABLE OF CONTENTS Editor’s preface 1. Introduction: Michel Bonifay’s Etudes

Michael

Fulford

with previously published reviews of M. Bonifay's Etudes

12

Roberta Tomber, Ariane Bourgeois, Paul Tyers 2. Technology and organisation of ARS ware production-centres in Tunisia Michael Mackensen

17

3. The forming and slipping of African Sigillata: evidence from the Palatine East assemblage

45

J. Theodore

Penta

4. Methodological constraints affecting the precise dating of African Red Slip Ware

65

John Lund

5. African Red Slip Ware on the move: the effects of Bonifay’s Etudes for the Roman East Philip Bes and Jeroen Poblome

73

6. North African lamps: Bonifay and beyond Jeremy Rossiter

93

7. Strong local production in Tunisia: supplementing Bonifay from the case of Leptiminus Karen Carr

105

8. Supplying Rome and the empire: the distribution of stamped amphoras from Byzacena David Stone

127

9. Imitation and diffusion: the case of Dressel 30 in Egypt Roberta Tomber

151

}

M sigillée D2 Oudhna

Sidi

Jdidi

Zahruni

Keay

e

o

25-

Spatheia

Afr.!IC-Keay 25.1 Pupput

-Keay

35-55-56-57-62

Dressel30

"Spatheia" tardifs

[9d

S

Sidi Khalifa Hergla

x LHadrumetum| Afr. Hl sigillée C Sidi

Marzouk Tounsi

Zegalass À Afr IIA? IIB var. Sid Saad Henchir es-Srira u Sufetula (Sbeitla)

Cillium (Kasserine)

.

e

Leptiminus| p

(Kairouan) o

Afr. 1 1B var.

Sebkhet Sidi el Hani Afr. IIA? IIB ni AfR

Afr.t-ll Keay 25-62-61 ee REeY

Keay 62

Ksour Essaf M AL Sullecthum | Afr. -tI Keay 25

(dont Hr ech Chekaf)

Thysdrus (El Jem) e

Hr Ben Hassine & Hr ech ChoggafA

Afr.1 f.

Keay 25.2 Dressel 2/4

. Thelepte

Sidi Aich

Majoura A

Keay 8B

Afr. I-II Keay 25 ?

Junca Keay 59/8B

Al Oued el Akarit Afr.I-lID Keay 25 Tacape (Gabés)

A Zitha Tripol. Dressel 2/4 Ateliers d'amphores : À A : repéré, non repéré * Neapolis

: ville antique

© (Kairouan) : ville actuelle : Sebkha

Ateliers de céramique sigillée :

Bb:

:: repéré, non repéré

100 km

Location of workshops in Africa Proconsularis and Byzacena, by M. Bonifay (= id., “Que transportaient donc les amphores africaines," in E. Papi [ed.], Supplying Rome and the empire [JRA Suppl. 69] p. 15, fig. 5). Triangles mark amphora workshops (black triangles = located; open triangles = unlocated); squares mark sigillata workshops (black squares = located).

Preface The publication in 2004 of Michel Bonifay's Etudes sur la céramique romaine tardive d'Afrique (Archaeopress; British Archaeological Reports $1301, Oxford; ISBN 1 84171 6451 0; £52) represents a watershed in the study of the Roman pottery of what is today Tunisia, not least for pottery found outside the capital of Carthage. Other scholars working in Tunisia immediately recognized its utility for work at their own sites, while scholars working in other parts of the Roman Mediterranean world were quick to grasp its relevance in view of the massive scale on which pottery products (and their contents), not least African Red Slip wares and amphoras, were exported by sea in both the High and the Late Empire from what would become Tunisia. This was quickly highlighted by three reviews which have appeared, reprinted here on pp. 12-16 by kind permission of their respective authors and publishers. Bonifay has continued to produce important articles since his monograph appeared. I take the oppportunity of highlighting two: "Ceramic production in Africa during late antiquity: continuity and change,” in L. Lavan, E. Zanini and A. Sarantis (edd.), Technology in transition A.D. 300-650 (Late Antique Archaeology 4 [2006]; Brill, Leiden 2007) 143-58; and "Que

transportaient donc les amphores africaines?" in E. Papi (ed.), Supplying Rome and the empire (Proc. Conf. Siena 2004; JRA Suppl. 69, 2007) 8-31. The latter article included two important figures ("Localisation des ateliers" and "Contenu des amphores africaines: tableau récapitulatif des hypothéses") which are reproduced here because of their usefulness and relevance to

the contents of the present collection.

|

Michel Bonifay’s monograph has thus provided a clear stimulus and a fortunate opportunity for several other scholars to develop and expand upon certain aspects of his work, adding to the picture that he has drawn their own ongoing, and largely still unpublished, research. The present collection hardly seeks to match the scale and scope of Bonifay's remarkable achievement, but intends simply to further the debate by addressing particular topics touched upon by him. A good many scholars currently engaged in work on N African pottery were invited by the editor to contribute to a collection which took its inspiration from Bonifay's work but was intended neither to constitute a critique of his work nor a comprehensive update with addenda and corrigenda. Different scholars have responded to the invitation in different ways,

according to the stages of their own research, but the variety of their contributions itself reflects the multifaceted nature of Bonifay's own work and his interests in themes which extend well beyond the shores of North Africa. The present project was facilitated by the kindness of Paul Naish at the publisher Archaeopress, Oxford, who made available at a favourable price multiple copies of Bonifay's book for the use of the present authors and several other scholars who had originally hoped to contribute. J. H. Humphrey

Type

Poix

Tripolitaine II

eo

Leptiminus Il

*

Arguments

archéologiques

Analyses

chimiques

| Absente du Monte Testaccio; utilisation dans un

Contenu possible

Datation

-

salsamenta? vin?

fin ler - milieu Iles. +

-

salsamenta?

fin ler - Ile s.

chais à vin( ?) en Libye

| Absente

du

Monte

Testaccio;

atelier littoral;

morphologie: pointe creuse Hammamet 1

*/o | Seule variante E présente au Monte Testaccio

-

vin? Variante E: huile?

| fin Jer (?) - Ile s.

Hammamet 2

e/o | Absente du Monte Testaccio

-

vin?

! | lile - Ve s.

Hammamet 3 Carthage EA IV Leptiminus I

€/O | Associée à des couvercles percés (celliers) eo ? |Résidus de poissons dans un exemplaire; morpho-

-

vin? salsamenta? salsamenta?

Ve - VIle s. Ier s. fin ler - Ile s.

Ostia XXIII Ostia LIX Uzita Pl. 52, 10

? | Morphologie: proche du type Africaine I. ? |Inscriptions peintes: OLEVM e/o | Absente du Monte Testaccio; morphologie: pointe

-

huile? huile? salsamenta?

fin ler - milieu Ile s. fin ler - milieu Ile s. fin ler - Ile s.

logie: pointe creuse

creuse Tripolitaine I

O

|Présente au Monte Testaccio

-

huile

ler s. - milieu Ile s.

Tripolitaine III

O

| Présente au Monte Testaccio

-

huile

milieu Ile - IVe s.(?)

Africaine I

©

Traces de lipides

huile

milieu Ile - IVe s.(?)

|Présente au céramique

Monte

Testaccio;

bouchons

en

Africaine II A

*

| Rare au Monte Testaccio; bouchons en liege _

Africaine H B

?

| Absente du Monte Testaccio

Pas de trace de lipides | salsamenta, vin?

-

?

Ille s.

milieu Ile - Ille s.

Africaine II B «pseudo-| tripolitaine»

0

| Ateliers continentaux

-

huile?

[le s.(?)

Africaine IC

*

| Ateliers littoraux; étiquettes de plomb mention-

-

salsamenta?

milieu [lle - IVe s.

nant le terme officina ou avec la représentation d'un trident; résidus de poissons? (Cabrera III); présente sur la fabrique de salaisons de Nabeul Africaine II D

e

Type «station 48 place corporations»

des]

[contra : inscription peinte OLEI] _

Poix ®

Arguments

archéologiques

| Bouchons en liège

Pas de trace de lipides | salsamenta, vin? Analyses

chimiques -

Contenu possible salsamenta, vin?

Pas de trace de lipides | vin? salsamenta?

milieu Ile - IVe s. Datation Ile - milieu Ile s.

Keay 25.1

@

Keay 25.3

.

Keay 25.2

@

| Noyaux d'olives

spatheion 1

©

| Noyaux d'olives (Dramont E); résidus de poissons | Pas de trace de lipides | conserves d'olives, vin? | finIVe- milieu Ve s.

spatheion 3 (miniature)

*

| Bouchon en céramique/en liege

Keay 3B « similis »

Pas de trace de lipides | vin? salsamenta? Pas de trace de lipides | conserves d’olives, vin?

(Tarragone)

Keay 27 Keay 36

IVe s. IVe s. | finIVe- milieu Ve s.

salsamenta? -

vin?

fin Vle-

le)

-

huile?

Ves.

e/o ejo

-

? ?

fin IVe - Ve s. fin IVe - Ve s.

Keay 59

o

Keay 8B

O

Keay

35A

[e]

Keay 35B

@

Viles.

-

huile?

fin IVe - milieu Ve s.

| Ateliers continentaux

-

huile?

milieu Ve - milieu VIe s.

-

huile?

Ve s.

| Résidus de poissons ; présence sur les fabriques de

-

salsamenta?

Ve s.

salaisons du cap Bon

Keay 55-56-57

e/o

Keay 62

eio

Pas de trace de lipides | ?

Traces d'huile

?

deuxiéme moitié Ve s.

Vie s.

(réutilisation ?) Keay

61C

Keay 61

Schóne-Mau

XXXV

et|

O

-

huile?

fin VIe - milieu VIle s.

?

-

?

Vile s.

-

vin

ler - milieu Iles.

*

| Morphologie

seudo Dressel 2/4 Dressel 30

®

{Morphologie

-

vin

Ille - [Ve s.

Amphores globulaires

?

| Morphologie

-

vin?

Vile s. +

Contents of N African amphoras: table recapitulating the different hypotheses, by M. Bonifay (= id., “Que transportaient donc les amphores africaines," in E. Papi [ed.], Supplying Rome and the empire [JRA Suppl. 69] fig. 8 on pp. 23-24). In the second column the black circles indicate presence of pitch, the open circles an absence of pitch.

Introduction: Bonifay’s Etudes Michael Fulford Michel Bonifay’s study of the Late Roman pottery manufactured in that area of Africa Proconsularis and Byzacena which also corresponds with the territory of modern Tunisia represents a major landmark, as much for the study of ceramics of the Roman world as for the economy of those provinces and their relations with their neighbours around the shores of the Mediterranean.

With

"Africa",

Bonifay

embraces

ceramics

for which

there

is either

direct,

primary evidence of manufacture in Tunisia in the form of kilns and/or waste material, or epigraphy represented by amphoras stamped with the abbreviated name of the originating city, such as Neapolis or Hadrumetum, or secondary evidence such as fabric and typology which can be linked to the output of known production centres. With "late", Bonifay follows J.

W. Hayes! in embracing pottery manufactured between the 2nd c. A.D., when African pottery, particularly amphoras, lamps and table wares, began to circulate in quantity in the W Mediterranean, and the 7th c. A.D., when

that extensive, maritime distribution faded away.

What is particularly remarkable and important about Bonifay's book is its breadth and inclusivity. For so long we have been used to working with individual categories of African ceramic production — the lamps, the red-slipped wares, the amphoras, the cooking or kitchen wares, and so on, and with an approach to the study of pottery which was driven, in the first place, by the fine art market and material such as the lamps and finer, decorated red-slipped wares — pieces which were readily appreciated for their artistic merits and subsequently collected or displayed. So, one mark of its landmark status is that Etudes embraces all types of ceramic material manufactured in Proconsularis and Byzacena: amphoras, sigillatas, lamps, domestic pottery (la vaisselle), miscellanea, including figurines (céramiques diverses) and building materials (céramiques architecturales). In addition to wheel-thrown domestic pottery,

Bonifay

also

includes

the

handmade

wares

(céramique

modelée),

such

as the

calcite-

tempered pottery first characterised petrographically by D. P. S. Peacock/ for which a Tunisian origin can be attested. That it has been possible to advance knowledge of African ceramics so rapidly (it is less than 40 years since publication of Hayes' ground-breaking Late Roman pottery) is a reflection of both the increase in the number of excavations around the Mediterranean producing Africanmanufactured wares, but also of the concentration of work undertaken at Carthage as part of the UNESCO-sponsored 'Save Carthage' campaign which saw its peak of activity during the 1970s and 1980s. Although earlier work by A. Carandini and colleagues had flagged the importance of African cooking and other domestic wares in the assemblages recovered from excavations in Ostia, it was the Carthage work which brought these materials to greater prominence, particularly for the later Roman (4th-5th c.), Vandal (Sth-early 6th c.) and Byzantine (6th7th c.) periods, with the establishment of new typologies and dated sequences. Although some common parlance was adopted in the reporting of the pottery excavated from Carthage, such that comparisons could more easily be made across different sites and excavations, so much of the work of different teams was being done in parallel that there was not the opportunity to establish a single typological and fabric model for Carthage. The characterisation of the clays and temper used in the various wares rested almost entirely on Peacock's petrological analyses, which were focused principally on the amphoras, the tablewares and the domestic wares, including the handmade wares.? As far as ceramics produced in Africa were concerned, the

petrology gave further foundation to the differentiation between different regions of production in Africa Proconsularis previously proposed on the basis of typology and superficial assessment of ware and surface treatment of the vessels. J. W. Hayes, Late Roman pottery (London 1972). 2

D. P. S. Peacock, “Petrology and origins," in M. G. Fulford and D. P. S. Peacock, Excavations at Carthage:

Hie British Mission vol. 1,2. The Avenue du Président Habib Bourguiba, Salammbo. The pottery and other

ceramic objects from the site (Sheffield 1984) 6-28. 3

Peacock ibid. 11.

10

Introduction: Michel Bonifay's Etudes

Thus, a major contribution of Etudes is that it brings the Carthage work together with other recent research undertaken across Tunisia to produce unified series of characterisations and typologies. Whereas prior to 2004 it had been necessary to refer to several typologies for each of the major categories of pottery — lamps, red-slipped wares, amphoras and domestic wares — all of these are now embraced by Etudesin one volume. Scholars can now refer to Bonifay's typology not only of the major categories, but also of the minor, such as the handmade, calcitic wares. Of course, these new typologies do not in any way minimise the massive contributions of scholars, many still active, on those individual categories, and it will probably take a generation for some to set aside, for example, J. W. Hayes' typology of the red-slipped wares, or S. P. Keay's of the amphoras,? so ingrained are these in their memory. But so fast-moving is the development of our knowledge of ceramics produced in Tunisia that Etudes, too, will surely require extensive revision or replacement within about a generation. This point is underlined by the fact that one volume has now captured the ceramic production of the larger part of a single province, based to a very great extent on the evidence of pottery consumption rather than on pottery production. While Etudes documents the evidence for the different production centres for each of the classes of ceramics in its Part 1 (which accounts for less than 20% of the whole), we are reminded how much of that evidence derives

from surface collection rather than systematic investigation of workshops and kilns through extensive excavation as well as contextual survey and surface collection. The prospect of undertaking such work is daunting, if not terrifying, given the prospect of how much material might be recovered. Indeed, M. Mackensen’s study of the late ARS production from El Mahrine in the Bagrada valley in the north of Tunisia? gives an insight into the volume of information that can be derived from the survey and cataloguing of the production of a single group of related workshops. Equally, with a focus on evidence recovered more from the coastal cities and settlements of northern and central Tunisia than from inland (where the majority of the citations are from production sites) or the south, we should not forget the contribution that systematic study of the ceramics of the consumer towns and settlements from those areas will bring in due course. In the individual examples which are used to illustrate a type, reference throughout Bonifay's volume is made to little more than a dozen or so Tunisian sites, and a considerable number of parallels are drawn from settlements and wrecks outside of Africa. Inevitably, as more research is undertaken, it will be difficult to resist a regional approach to the study of the ceramic production of Africa Proconsularis and Byzacena; but that is for the future. Hints as to how the various production centres impacted on the economy of the provinces are summarised in Part III on pp. 449-52, with fig. 254. The export of the northern production of El Mahrine and Oudhna is largely confined to the north, and there is a corresponding, local concentration of consumption of the wares (dominated by amphoras) produced in and around the

ports of Hadrumetum, Leptiminus and Sullecthum. However, the sigillata production of the central and southern centres, located more

than 100 km

inland, of Sidi Marzouk,

Djilma and

Sidi Aich reaches a range of coastal destinations between (as well as beyond) Sullecthum and the border with Tripolitania (Libya). This reminds us how, even when production was at a smaller distance, up to about 100 km, from Mediterranean ports, it was still possible for that production, once it reached the coast, to circulate widely around the Mediterranean. Arretine production of tablewares is another case in point. Even though the wide distributions of African ceramics (other than amphoras) around the Mediterranean was undoubtedly facilitated by the export of grain and other foodstuffs and by the amphora-borne commodities, there were also precedents, such as at Arretium, for specialised, inland manufacture of pottery making a major impact on Mediterranean markets. A significant exception to this rule is the production of the major manufactory at La Graufesenque some 100 km from coastal Narbo. While this South Gaulish tableware does circulate in the Mediterranean, particularly in the West, its principa! centres of consumption lay to the north in N Gaul, Germany and Britain. 4

5

S. J. Keay, Late Roman amphorae in the western Mediterranean (BAR S196, Oxford

1984).

. M. Mackensen, Die spätantiken Sigillata- und Lampentöpfereien von El Mahrine (Nordtunesien). Studien zur nordafrikanischen Feinkeramik des 4. bis 7. Jahrhunderts (Münchner Beiträge zur Vor- und Frühgeschichte 50, 1993).

Introduction: Michel Bonifay’s Etudes

11

The massive contribution of Etudes is that it gives both the excavator without specialised knowledge of ceramics as well as the ceramicist a very robust series of typologies. These will be of enormous help in the study of ceramic assemblages from outside of Tunisia, particularly in the discrimination of the total African contribution in an assemblage from that of the local, or of other imported, sources. It is already clear from the analysis of pottery assemblages from major west Mediterranean ports such as Tarragona, Marseilles and Ostia/Portus that African cooking wares and other domestic pottery play a significant röle alongside lamps, red-slipped wares and amphora-borne containers. However, P. Reynolds’ study of the pottery from settlements in the hinterland of Alicante in south-east Spain® gives an indication of the contribution of African ceramics to rural societies located up to 50 or more kilometres inland. As quantitative approaches to the study of ceramics continue to be adopted, and aided by this single handbook (and surely Part II of Etudes should be re-published as such) on African ceramics, we can look forward to much more holistic as well as nuanced assessments of the róle of African production in the Mediterranean region. It is time to be able to move on from the qualitative distributions of the kind espoused by Hayes in 1972 but still perpetuated, as with the examples reproduced in Part III (e.g., figs. 252-56). Even at the qualitative level, however, there can be profound changes to perception. When Hayes published his Late Roman pottery it seemed that the distribution of the latest forms of ARS was skewed to the E Mediterranean and to Constantinople, but subsequent work has shown this not to be the case (fig. 252); the forms are found around the shores of the whole Mediterranean and beyond. Along with the study of the composition of the cargoes found in shipwrecks — a direction indicated by Bonifay in his Part III: Diffusion, 452-54, fig. 255 — we will also be in a better position to disentangle the contribution of direct trade among the inter-connected communities of the W Mediterranean between Africa and Gaul, Hispania, Italy, etc. from that of redistributive networks.

What Etudes is not, is a study of all Late Roman pottery from N Africa. In this regard it differs from, say, P. Tyers' provincial survey of all the principal types of pottery found in Britain,” whether imported or made within the province(s). Wares imported into Africa, except insofar as they bear on African issues, such as amphora production in Mauretania Caesariensis or Tripolitania, are scarcely considered (e.g., 452). In fact, imported pottery, mostly comprising amphoras from the E Mediterranean, but also including some eastern red-slipped wares as well as cooking wares from both the E and the W Mediteranean, is very much in the minority in the Late Roman period. However valuable a source of evidence imported pottery is, in order to understand more fully the contribution and vibrancy of the economy of Africa Proconsularis and Byzacena it is necessary, of course, to look at the evidence of consumption beyond the shores of Africa itself. In conclusion, we cannot do better than agree wholeheartedly with Bonifay's own concluding remarks. He notes that much of the research to date has been undertaken by scholars working from outside Tunisia and the Maghreb more widely, and that it is time for new work to be taken forward locally, particularly in Algeria where so little ceramic research has been undertaken in recent years. This requires the co-operation not only of scholars working in the Maghreb, but of the entire community of those working around the Mediterranean. A North African problem is also a Mediterranean one.

6

. P. Reynolds, Settlement and pottery in the Vinalopó Valley (Alicante, Spain) A.D. 400-700 (BAR $5588, Oxford

7

1993).

P.Tyers, Roman pottery in Britain (London 1996).

12

Introduction: Michel Bonifay’s Etudes Previously published reviews of M. Bonifay’s Etudes

1. Roberta Tomber, "Alive and well: the state of Roman pottery studies,” Antiquity 80 no. 307 (March 2006) 218-20. Reprinted by kind permission of Antiquity Publications Ltd. [the review also treated J Ma. Gurt i Esparraguera, J. Buxeda i Garrigos and M. A. Cau Ontiveros (edd.), LRCW I. Late Roman coarse wares, archaeometry, which is omitted here]

cooking

wares

and

amphorae

in

the

Mediterranean:

archaeology

and

"This volume, the thesis of Michel Bonifay, represents approximately 20 years of experience with Roman pottery throughout the Empire, but particularly North Africa and Gaul. The breadth of this study is immense and it will become a standard reference for North African ceramics. It is exceptional in that it brings together production and consumption sites, the latter from throughout the Empire. North Africa is the starting point for Bonifay's seminal work on the late Roman (here defined as second to seventh centuries A.D.) pottery from modern-day Tunisia. As the author notes,

since the UNESCO Campaign to Save Carthage in the 1970s, the pottery of North Africa has frequently been synonymous with the pottery of that city. Subsequent work throughout the country, by diverse international teams, including those with whom Bonifay has been associated, has rectified this, by providing a more even coverage of Africa Proconsularis. Importantly, this work has included numerous production sites particularly for amphorae, and sigillata, but also coarse wares and lamps. The study is presented in three parts: Production, Typology and Chronology, and Diffusion; it flows from the first two data-rich components to the third contextualised section. The scope of this work is huge, calling on all sources of evidence, from epigraphic to ethnographic. The first two parts provide a fabric and form characterisation for the production sites. Scientific analysis is summarized and referenced to more detailed petrographic analysis by Claudio Capelli. Here, as elsewhere, the bibliography alone forms an important resource. These kiln fingerprints provide the framework for identifying the distribution of North African products throughout the Empire, from which Bonifay has illustrated the shifting locus of ceramic exportation from Tunisia through time. Clearly, this work is underpinned by the ability to reliably differentiate kiln products from the different regions of Tunisia, not only from sherd material, but also working back from publication.

Bonifay has the experience to achieve this. The relationship between form/fabric and kilns is complex and there is bound to be some overlap in the fabric of different workshops within a region. Therefore those with less firsthand experience of the material need to be aware of the complexity when assigning sherd material to individual kilns.

Bonifay has successfully sketched an historical narrative for the exchange of North African pottery, as others before him, setting it within the broader context of African agriculture. Departing from the long-held view that North African amphorae were primarily for the transportation of oil, he favours salsamenta and wine as the more commonly exported products, with oil in smaller

quantities. As in all aspects of his work, these conclusions are based on a range of evidence, including the location of the kilns, excavation context and inscriptions. Two additional strands of evidence, not always available, were used to investigate contents: the identification of pitch or

other linings on the interior vessel walls and the description of stoppering and decanting methods. Pitched vessels are equated with the transport of wine, while the stoppering /decanting techniques are related to the viscosity of the different products, thus useful for inferring contents. The much-debated question of whether amphorae determined the exportation of other vessel types (mainly finewares) forms the subject of another interesting discussion; here again Bonifay presents an alternative vision, that fineware distribution is instead related to that of wheat. While

Bonifay recognises that additional evidence for both amphora contents and agricultural and maritime installations may change these interpretations, they provide fresh insights into the role of

Africa within the Empire. In his conclusions, Bonifay calls for a closer dialogue between ceramic specialists and historians. The book demonstrates the contribution that ceramic-based studies can make to broader

narratives. Since detailed aspects of fabric and typology may be of interest to a limited audience only, pottery specialists should ensure that their conclusions reach the widest possible readership. At the same time, the dissemination of raw data remains vital to re-interrogate data in the light of

new hypotheses. This dissemination can be achieved through a variety of media, including publica-

tion, online resources such as the Archaeological Data Service at York (http:/ /ads.ahds.ac.uk), or CDs to supplement publication. By publishing widely, in formats that incorporate varying levels of detail, specialist study will be integrated into the mainstream and reach the wide audience it deserves."

Introduction: Michel Bonifay’s Etudes 2. Ariane Bourgeois, Revue des Etudes Anciennes 108.2 (2006) 805-8. permission of the Direction of the Revue des Etudes Anciennes, Bordeaux.

13 Reprinted

by kind

"La céramique romaine d'Afrique du nord est un domaine complexe, en particulier en Tunisie. Son étude est en plein développement depuis plus de trente ans, gráce à des chercheurs de toutes les nationalités. Elle a été longtemps menée à partir de données extérieures à l'Afrique, fouilles de lieux de consommation et d'épaves, musées et collections diverses dans toute la Méditerranée, et elle est

maintenant analysée et classée sur place dans les provinces romaines d'Afrique, aprés des prospections ou dans les sites étudiés, habitats, nécropoles, zones de production. Aprés un lent démarrage axé surtout sur la typologie (Lamboglia, Hayes, Atlante I), il y a maintenant un nombre impressionnant de publications qui font une large place aux amphores, céramiques fines, vaisselles utilitaires et lampes de ces régions. Le livre de Michel Bonifay est alors une magistrale synthése critique, riche d'apports novateurs essentiels, fondés sur une longue expérience personnelle. En effet sa participation à des prospections sur les cótes tunisiennes et à d'importantes fouilles franco-tunisiennes — sans oublier d'autres sites antiques au Liban, en Égypte, en Albanie, en France —, lui a fourni une documentation considérable et nouvelle, de premiére main, lui permettant ainsi de mettre un peu d'ordre dans un domaine oü les publications ne tenaient pas toujours compte des classifications et chronologies antérieures et en proposaient souvent de nouvelles, Ce gros livre est trés clair et aisé à utiliser, gráce à ses nombreux dessins et photos, tableaux et cartes, gráce à une bibliographie considérable d'environ sept cents titres, dont les plus récents sont de 2004, voire «à paraitre» (p. 491-517). Il embrasse toute la production céramique de la province augustéenne de Proconsulaire, les amphores, les sigillées claires, les vaisselles culinaires, communes

et modelées, les lampes, et méme des figurines, des moules (a patisserie?) et des matériaux architecturaux. Il envisage et compare tout cela dans une progression en trois parties inégales, la production (p. 7-85), la typologie et la chronologie (p. 87-441), enfin la diffusion (p. 442-486). Pour étudier la production, le livre fait le point sur les ateliers d’où provenaient les différents objets. Sauf exceptions, ils ont été longtemps négligés, car on en connaissait trés peu, et la rareté des timbres de potiers était une autre source d’incertitude. Durant les deux dernières décennies, des prospections, avec des découvertes d'outils, par exemple les casettes indispensables lors de la cuisson pour donner à la sigillée claire sa couleur si caractéristique (p. 60-66), ont permis de combler partiellement cette lacune, mais aucune officine, sauf

à Oudhna, n'a été vraiment fouillée. Pourtant,

fondamentale est l'identification de certains centres de production importants, par exemple El Mahrine, qui a fourni en abondance de la sigillée claire D et des lampes sigillées à Carthage. Comme dans les autres provinces romaines, beaucoup de ces ateliers ont fabriqué de tout. Enfin, outre les aspects techniques de la fabrication, outre les catégories et les formes produites, ces localisations permettent des analyses physico-chimiques, notamment pétrographiques, très utiles pour préciser l'approvisionnement de sites lointains et pour différencier les séries de céramique africaine, en apparence assez homogènes (p. 523-525, pl. I-III: microphotographies des différentes catégories depâtes). La deuxiéme partie, essentielle, est le coeur du livre, car elle renouvelle et clarifie la typologie, en

ne faisant jamais fi des travaux antérieurs. Elle s'efforce aussi, pour chaque forme ou plutót type, de cerner la chronologie. D'ailleurs les diverses catégories et leurs types sont présentés dans l'ordre de leur apparition, quand c'est possible, avec l'évolution, la durée et les variantes de chacun, en distin-

guant les formes classiques, les formes locales ou régionales, et les formes tardives. À juste titre, il est souligné qu'il faut envisager tout le territoire provincial, sans se focaliser sur Carthage, dont les données, plus fournies à cause de son importance économique et politique et gráce aux nombreuses fouilles menées là depuis plus d'un siécle, ne doivent pas étre surévaluées au détriment des autres grandes villes et des autres régions. Les fiches des types successifs d'amphores, de sigillée africaine, de céramiques culinaires, communes et modelées, de lampes, sont présentées selon un schéma trés clair (autres dénominations

dans les typologies précédentes, résumé critique de ce qui est connu, bibliographie, description des particularités techniques de la forme, de la päte et de l'engobe; ensuite énumération de quelques exemples, inédits ou avec la référence à une publication; enfin, chronologie proposée). Chacune est assortie d'une abondante illustration, photographique pour les lampes et les décors, graphique pour les profils d'objets, méme si l'échelle varie parfois sur la méme figure ou méme est indiquée par le chiffre du diamétre supérieur. Ainsi les amphores (p. 87-153), aux types numérotés de 1 à 66, sont illustrées par les figures 46 à 83, les 95 types de sigillée (p. 153- 210) par les figures 84 à 111, les 38 types de céramique culinaire (p. 210-244) par les figures 112 à 131, les 74 types de céramique com-

mune (p. 245-303) par les figures 131 à 169, les 6 types de céramique modelée (p. 303-310) par les figures 170 à 174, enfin les 85 types de lampes (p. 311-430) par les figures 175 à 241. L'apport le plus neuf pour la sigillée claire est ce qui concerne les objets tardifs des VI*-VII? siècles, voire VIII*.

14

Introduction: Michel Bonifay’s Etudes En effet les niveaux supérieurs des fouilles ou les habitats précaires de cette période rendent souvent difficile la datation des objets sigillés, de qualité plus disparate qu'avant, au vernis médiocre, à la morphologie moins standardisée, dégradation complétée par l'abandon partiel de la cuisson en casette. Or les données nouvelles des sites tunisiens de Jdidi, Nabeul, Pupput, bien connus de M. Bonifay, ou celles de la Crypta Balbi à Rome, lui ont permis d'affiner la compréhension de l'évolution et la chronologie. Un autre probléme compliqué et irritant est celui des nombreux bols à listel, fabriqués dans beaucoup d'endroits dés le début de l'Empire, aussi bien en céramique fine (p. 178 sq) que commune (p.249-260): la diversité des mortiers sigillés Hayes 91, déjà soulignée par ce dernier, est telle qu'elle entraine ici la subdivision en huit types (types 48-53, 78, 94). Enfin, la typologie des lampes actualise et précise des publications qui restent fondamentales, dont certaines déjà anciennes. La longue durée des productions africaines, dés le II? siécle, celle de certains modeles

omniprésents comme les formes Deneauve VII et VIII, Atlante VIII et X, la diversité des ateliers à grande diffusion (El Mahrine, Oudhna, Henchir es Srira) mais aussi de tout un tissu de petites officines locales, la pratique du surmoulage de plus en plus fréquente avec le temps, les innovations ponctuelles à commercialisation limitée, ont encore rendu la situation complexe: la tentative de clarification est d'autant plus méritoire, notamment pour les innombrables lampes en sigillée (p. 353417), appréciées dans tout le monde romain et imitées dans des lieux aussi improbables que Treves (p. 461). Aux VI*-VIF* siècles, on voit aussi l'appauvrissement du genre. La troisiéme partie aborde avec précaution la diffusion: la céramique africaine, témoin des activités économiques. Cette question a reçu depuis longtemps des réponses diverses, faussées par la difficulté d'interprétation des comptages et statistiques indispensables en céramologie, notamment dans le domaine des amphores. Or on méconnait souvent le contenu et la contenance de ces dernières, et les datations sont souvent trés larges et donc floues. L'établissement de cartes de diffusion,

surtout si l'on veut à tout prix y intégrer les faits politiques et géographiques des V*-VII* siécles, a également toutes les chances d'étre faux. D'autre part, alors que la documentation littéraire et épigraphique sur les échanges entre provinces est fort pauvre, les données archéologiques, en Afrique et dans le reste de l'Empire, sont de plus en plus nombreuses et précises: avec étonnement, on constate que l'approvisionnement en vaisselle africaine de Carthage est moins diversifié que celui de Marseille, à la méme période, et aussi que les importations de vaisselles extérieures sont rares en Afrique, bien qu'on ait étudié des épaves chargées d'amphores et de céramique africaines au large de la Provence, preuves d'un commerce interprovincial actif de denrées alimentaires. Une autre source d'erreur est le maintien de l'idée reçue que tout, en Proconsulaire, repose sur la production d'huile,

trés dynamique depuis le II* siécle, que toutes les amphores africaines étaient destinées au transport de cette denrée, et donc que les vaisselles de méme origine trouvées un peu partout dans le monde romain, notamment en Italie, accompagnaient des cargaisons d'huile. En réalité, on connait en Orient

des vases africains sans amphores de méme provenance, ou bien en Afrique méme il y a des amphores poissées produites sur place, impropres pour l'huile mais utilisées pour contenir du vin ou des salsamenta, dont on a fouillé plusieurs «usines à salaisons» (Nabeul). En outre, si beaucoup d'ateliers d'amphores sont dans des régions oléicoles, d'autres officines ont d'autres localisations. ourtout, il ne faut pas oublier l'implantation de certains centres de production de sigillée claire et de céramique commune importants dans des zones céréaliéres riches, fournissant le blé de l'annone toujours exigible par Rome puis Constantinople jusqu'à la fin de la domination byzantine en Afrique, blé qui ne laisse pas de traces archéologiques, ce qui expliquerait en partie l'abondance et la permanence des exportations de vaisselles à époque tardive. Enfin, l'histoire des exportations d'amphores, celle de la production et de la diffusion des vaisselles, permettent de déceler les fluctuations de la vitalité des principales régions économiques, celle de Carthage et celle d'Hadrumète, avec une relative stagnation au début de la domination vandale (p. 481), puis un renouveau. En tout cas on a des preuves de contacts commerciaux entre l'Afrique et le royaume franc, avec la Catalogne, la Ligurie, Rome, etc, encore au VII? siècle et au début du VIII, loin du vieux cliché d'un déclin

généralisé à cette période. Pour finir, il faut bien faire quelques remarques critiques, surtout formelles. Ainsi, p. 4, la carte fig. 1, dissocie bizarrement Toulouse et la Garonne, et les tracés du Rhóne et du Rhin sont inexacts; p. 453, sur une autre carte, figure 255, les noms d'épaves sont en frangais, ceux des cités antiques en

latin, et on est surpris de lire cargoes, en anglais, au lieu de cargaisons (d'amphores), méme si le fond de carte provient d'un article en anglais; quelques fiches typologiques, présentant des exemples parfois disparates, n'ont pas de numéro de type (p. 165 et fig. 89, pour les formes Hayes 82-84-85; p. 203 et fig. 108: le type 79 precede une fiche «autres variantes de types à listel», de forme Hayes 92, sans grand rapport avec le type 79, et le type 80 est l'objet de la fiche suivante ; p. 221 et fig. 118, les «autres formes» n'ont pas non plus de numéro de type et présentent trois profils différents; etc). Dans d'autres cas, si les formes «prolifiques», Hayes 61, 91, 181..., sont à juste titre longuement

Introduction: Michel Bonifay’s Etudes

15

développées et dissociées en plusieurs types, d'autres modéles assez fréquents à Carthage comme les bols Hayes 80/81 ont ici une place trés réduite. Quelques photographies sont peu nettes (p. 194, décor du vase n? 12), alors que, en revanche, celles des lampes rendront de grands services. Par ailleurs, le type 64 en céramique commune, p. 293, fig. 163, n'est peut-étre pas à sa place dans ce livre, puisque «cette cruche ne paraít pas originaire de la cóte orientale de la Tunisie» et que les quatre exemples proposés ont tous été trouvés sur la rive nord de la Méditerranée. Je me demande aussi s'il était utile, dans un livre à la matière déjà si riche, de traiter des matériaux de construction,

des pesons de filets de pêche, alors que je regrette la brièveté des lignes et des images consacrées aux carreaux architecturaux, si particuliers (p. 439). Mais ce sont des pécadilles sans grande importance, qui ne doivent pas faire oublier qu'il s'agit d'un ouvrage remarquable et fondateur, dés maintenant indispensable pour les historiens et les

céramologues travaillant en et sur l'Afrique romaine, et également dans toutes les régions de l'Empire romain."

3. Paul Tyers, Britannia 38 (2007) 375-76 (reprinted by kind permission of the Society for the Promotion of Roman Studies, London).

"Bonifay lays out his stall in the opening pages of this volume. Geographically he is concerned with the pottery produced within the modern boundaries of Tunisia, and Late Roman here is mid2nd to late 7th c. A.D. The principal vessel types are amphoras and red-slip wares (sigillée africaine in Bonifay's terminology), and while neither of these could be described as common finds on Romano-British sites, they are both exported widely around the Mediterranean and will be familiar to anyone who has worked with Roman pottery in that area. Bonifay's principal experience has been on sites in Tunisia and southern France, particularly Marseille, where these exports are both abundant and of great variety. However, he is quite explicitly seeking to re-orientate the study of these vessels towards their production and to examine in detail the regional character and distribution of the various categories within Tunisia, and then apply this improved understanding of the sources to the exported material. The first section (Production) presents the evidence for the location of kiln sites and details of

the manufacturing processes and raw materials. Much new information has evidently been collected during a survey of the Tunisian coastal regions and fieldwork for the archaeological map of Tunisia, which has resulted in the identification of a number of new production centres. Combined with earlier published work (notably by Mackensen, Peacock et al.) and supported by a program of petrological analyses by G. Capelli, these threads are drawn together to provide a picture of the history of production. The shifting focus of red-slipped ware production through its long history is particularly well laid out. The second and by far the longest section (Typologie) describes the vessels themselves, starting with the amphoras, followed by the red-slipped wares, culinary wares, handmade wares, lamps, and concluding with a short section on ceramic building materials. The material is presented as a type series (so Ampliore types 1 to 66, Sigillee types 1 to 95, and so on). The catalogue and illustrations are extensively cross-referenced with sub-headings in the text and annotations on the illustrations to pre-existing type series (so Bonifay's Sigillée type 1 is equivalent to Hayes type 2, etc.), and the ‘traditional’ names for most types are used in the introductory and concluding discussions. Nevertheless, references to Bonifay's types have already been spotted 'out in the wild' at pottery conferences in France, so these type series seem destined to become (yet) another layer in the burgeoning ceramic nomenclature, despite Bonifay's own comments on p. 156. The many illustrations are drawn from sources on both sides of the Mediterranean and the descriptions contain valuable hints on distinguishing forms and, where possible, sources. At the end of the volume are 3 pages of colour photographs of fresh sherd breaks at a scale of 2 : 1 illustrating

the principal regional fabrics. Although the reproduction of these is good, larger images would be preferable if they are intended to aid identification in the field or when using a binocular microsope.

Summary descriptions of the wares are to be found in the Production section, and the more detailed petrological descriptions will doubtless be found in the numerous publications referenced from there.

The concluding section (Diffusion) tackles the issue of the value of pottery studies as an indicator of wider economic and historical patterns. Can one indeed write history from pottery, or just the history of pottery? In contrast with some writings over the last 30 years or so, Bonifay is somewhat gloomy on the prospects here and presents several examples where new information on the source or

16

Introduction: Michel Bonifay’s Etudes chronology of some type or other has undermined the foundations of a neat historical link. At one point he even suggests that it might be necessary to revisit the sherds of some of the major published assemblages in the light of new evidence, and re-classify and re-quantify them. One can only hope that they have not been cleared out of the museum stores to make space for something else. The story of the identification of the contents of African amphoras is summarized on pp. 463-67, with the many perils and pitfalls highlighted. Bonifay suggests that relatively few of these jars were used to transport olive-oil, which had at one time been considered to be their major content. Rather than any amphora-borne commodity, he suggests that wheat exports are the major driver of the African economy. The concordance between Bonifay's classification and those of Hayes, Keay and others is presented in a table near the end of the volume, cross-referenced to the principal descriptions and illustrations of the types in the Typologie section; but other than this there is no index. This is

perhaps unfortunate for a 525-page book, as anyone seeking to unravel the full story of a particular type or assemblage which runs across all three sections of the work is left without any obvious means of doing so. Bonifay himself recognizes that the next advances in the study of African pottery will be made on-line rather than on paper, so perhaps a start could be made here by distributing on a CD, or by some other means, more and larger images of the various fabrics, along with some means of tracking the many valuable insights presented in this work. Michel Bonifay is to be congratulated on this work, which is based on a thesis presented at the University of Paris. It summarizes what is known (and what is not), and lays out a framework for

the future study of the material. Its influence will be widespread far beyond the boundaries of the production area, given the wide distribution and long history of these vessels."

Technology and organisation of ARS ware production-centres in Tunisia Michael Mackensen Without any doubt the Etudes sur la céramique romaine tardive d'Afrique by M. Bonifay constitute an opus magnum of outstanding importance for knowledge and the present state of research on different categories of Roman pottery manufactured in the N African Maghreb. Because of his continuous work for more than two decades on various archaeological projects in Tunisia (Nabeul, Oudhna, Pupput, Sidi Jdidi)! and in numerous other projects all over the Mediterranean,

from

Marseille

to Alexandria,

the

author

is one

of

the specialists

of Late

Roman pottery and so supremely qualified to write a summarizing report on mid-Roman (late 1st to late 3rd c.) and Late Roman

(early 4th to 7th c.) pottery produced

in N Africa (the latter

includes the Vandal [A.D. 429-533] and early Byzantine periods [533-670/698]). The red slip ware (ARS), cooking wares and amphoras from N Africa were widely distributed, not only to Italy, but even to the more distant northwestern and eastern provinces of the empire.? The chronological range covered by Bonifay is the same as that of the handbook Late Roman pottery by J. W. Hayes, which, as far as the systematic classification of types of ARS ware goes, is not yet replaced. Bonifay's emphasis is upon pottery produced in Tunisia from the 4th to the 7th c., a time-span

for which he has a vast amount of information at his disposal.

It may also be remarked that Bonifay's book, which is written in French? will be warmly welcomed and widely used in the Maghreb. There will not be any difficulties in the reception of its research results or for the additional classification of categories. The general topics of his study are production, typology and chronology, as well as distribution and the economic implications. He discusses not only the priority of comprehensive research on the production centres of ARS ware, but also the assignment of different fabrics, described and defined by A. Carandini, J. W. Hayes, and others.* Of particular importance for the reconstruction of supra-regional trade patterns of ARS ware and the observable changes in them is the macroscopic classification of specific fabrics characteristic of the various groups of fine red-slipped table-ware, which can be connected with chemically-defined reference groups (which are based on material from identified potteries), and reliably attributed to production centres in northern and central Tunisia? In the following contribution I intend to assess selectively some of the views on the production of ARS ware. In the light of recent archaeometric results and archaeological evidence hitherto unpublished, I shall concentrate on rare potters' tools with graffiti, specific N African kiln furniture, and some technological aspects, to offer possible conclusions for the organisation of the ARS ware production-centres. 1. ARS production sites in Tunisia (fig. 1) In the systematic research carried out thus far on Late Roman production-centres of ARS ware and lamps in N Tunisia, most of the archaeological data were obtained from either a series of field surveys at El Mahrine (1981-92,

1997-99),

Henchir

el-Biar

(1987)

and

Bordj

el-

1

Cf. Bonifay 2004, 1-3 fig. 1.a-c.

2 3

[bid.1. Inthe past many substantial papers and monographs published in English, German or Italian were often virtually ignored because of an obvious language barrier.

4

Ibid. 45-65 fig. 22; Ostia I, 28-37; Hayes 1972, 289-92; Atlante I, 19 f., 52 £., 58-60, 141 f., 144, 147 f. and 156 f.

5

Cf. Mackensen and Schneider 2002 (3rd-7th c. ARS); Brun 2004 (Oudhna); for the methodological

approach, the requirements, and the results of the investigation of mid-Roman ARS ware, cf. now Mackensen and Schneider 2006 (2nd to late 3rd/early 4th-c. ARS).

M. Mackensen

18

CD groupe D1

Carthage

Borj el-Jerbi El Mahrine groupe D2

Hr el-Biar

productions D locales

Oudhna ? /N

uu

Zahruni

Neapolis

ateller X

A.

Sidi Khalifa

groupe C

A

Hadrumetum

Chougafiya A

| (Kairouan) o

Leptiminus

E

Sullecthum Thysdrus (El Jem) e.

Ksar el-Guellal

ducti procuc ons

A e Sufetula (Sbeit

continentales

Acholla

A Thelepte

Sidi Aich vw

À

vA Thaenae e

lunca

Tacape (Gabès) Cro

7 CR D

aus

CAS

St

ovK

Bir Oued Farah

RA

Ateliers de céramique sigil lée : ^ L.»

: non repéré

A ?

A‘ A'

:connu par l'archéologie 23

:douteux

* Neapolis

**-.

A

: ville antique

© (Kairouan) : ville actuelle wx

: Sebkha

groupe C 0 L-

: productions 100 km |

Fig. 1. Production centres of African red slip ware in Tunisia (after Bonifay 2004, 46 fig. 22).

Technology and organisation of ARS ware production-centres

19

Djerbi (1997-99, 2002) south of Thuburbo Minus/Tebourba,® or from excavations of a workshop with kilns installed in the abandoned Baths of the Laberii at LIthina /Oudhna, as well as from further field surveys in the northern periphery of that city (1994-96).7 The final report on the two most important mid- and Late Roman potteries at Henchir elGuellal near Djilma and Sidi Marzouk Tounsi near El Ala, both in central Tunisia and surveyed by a British-Tunisian team in 1984-85,? sites where production of plain and appliquédecorated, high-quality ARS ware and lamps had already started in the early 3rd c. is still

awaited.!? At present the complete range of plain and decorated ARS vessels and lamps, coarse ware and ceramic building-materials, as well as an overall view of the two production sites and their structural organisation, is not easy to obtain because the results for different groups of material are scattered in a number of publications.!! From its foundation in the early 3rd c., the production-centre at Sidi Marzouk Tounsi, which had probably developed as a vicus circa villam (i.e., a wealthy villa surrounded by a potterymaking village of c.25 ha), had been inhabited until the middle of the 6th c. by highly specialised craftsmen. They produced not only technically-refined, plain thin-walled ARS vessels, but also elaborately stamp- and relief-decorated ARS tableware.!? What is missing is a synthesis, even if one based only on the material available from the surveys carried out at Sidi Marzouk Tounsi and on ARS vessels, tools and moulds which were re-attributable to this production-centre on the basis of archaeometric analyses. Bonifay deals with

these two production

sites (as well as with some

others of minor

importance) and the ARS vessels in fabric C (CL?) in only four paragraphs. Without going into the details of the typology and chronology of these highly important and, in terms of their craft skills, very innovative workshops which were working to a high level of quality, his discussion of the end of ARS production in the Early Byzantine province of Byzacena is

necessarily brief.!* In the present state of research, the mid-Roman ARS ware in fabric A (Al, A? and Al/2), which was most probably manufactured in an (unlocated) production-centre in N Tunisia, and

which in the early 3rd c. was contemporary with central Tunisian C! ware, is fairly difficult to assess — in particular, the appliqué-decorated A1/? ware. For various reasons Bonifay supports production of the A!/? ware at Oudhna, but no archaeological evidence yet exists for this possibility. Neither the coarse-grained fabric and thick, glossy slip (which is similar to the 6

Mackensen

1993, 1998a, 1998c and 2004a; Mackensen and Schneider 2002. The surveys at El Mahrine

and Bordj el Djerbi, in particular, I had the chance to carry out in connection with ongoing work on Late Roman pottery from the excavations of the German Archaeological Institute at Carthage. 7 .J Barraud et al. 1998. Preliminary report: Peacock, Bejaoui and Ben Lazreg 1990, 66-81 figs. 5-9, 13-14. 9

Cf. Mackensen 2006b, 113 f.; Mackensen and Schneider 2006, 166 f. fig. 2; 174-77 Table 3; 183-85 fig. 11

with conclusive pottery evidence. 10

Cf. recently Mackensen and Schneider 2002, 131-34 figs. 5-6; 151-55 figs. 22-24; Mackensen 2004a, 132-

39; Mackensen and Schneider 2006, 166 fig. 2; 171 fig. 5.2-10; 183-85 fig. 11 with the main ARS forms of the 3rd c.; Mackensen 2006b, 111-21.

11

Cf. Mackensen and Schneider 2002 and iid. 2006, both papers containing some new survey material and

12

various fragments of ARS vessels in collections and museums which could be attributed to the sitespecific chemical reference groups of these potteries by means of chemical analyses (for each site there is a complete bibliography); cf. also Mackensen 1993, 33 f. and 37; 2005a, 2006a and 2006b. Cf. Peacock, Bejaoui and Ben Lazreg 1990, 68, 74 and 83; Mackensen and Schneider 2002, 130-34 figs.

5-6, 141, 151-55, figs. 22-24; iid. 2005 and 2006; Mackensen 2006b; 1998a, 26-30; 1998b; 2003a; 2003b; 2004a,

132-39

figs. 2-5; 2004b; 2005a; 2005b; 2006. Much

depends

also on a personal

knowledge

of

these remote sites which are difficult to locate in the countryside, or on a knowledge at least of the

material from the British-Tunisian survey. 13

14

Mackensen 2005b, 317 and 2006b.

Bonifay 2004, 50 f. mentions Hayes form 84 which was imitated by the local potteries producing ARS ware in C? fabric.

15

Ibid. 47 f.

20

M. Mackensen

6th-c. D? fabric from Oudhna) nor the numerous appearances of appliqué-decorated ARS forms

Hayes 24, 35 and 38 in A!/? fabric in deposits of the early 3rd c. from the Capitolium at Oudhna,!$

are sufficient to prove it, as Bonifay himself emphasizes.

Decisive are the new

chemical analyses of the A!/ A? group, which do not match closely enough the D? reference group from Oudhna to postulate the production of A1/A? ware somewhere in the Roman colonia of Uthina.!? When presenting typical examples of ARS production-centres in an urban or rural context, Bonifay selected sites which are situated in NE Tunisia. He favours particular productioncentres in the region covered by some of his own projects, ones therefore more easily accessible to him. He concentrates on the Late Roman workshops at Uthina/Oudhna, the potteries in the periphery of Pheradi Maius/Sidi Khalifa, those at Gemellae/Sidi Aich, and even a site at Sidi Zahruni near Nabeul which is of only local or restricted regional importance.!? ARS production in D? fabric at Oudhna (probably in the kilns located in the N periphery of the urban settlement) had already begun in the middle or second half of the 4th c., much earlier than the excavated late 5th- and 6th-c. workshop with three kilns in the Baths of the

Laberii.? But the range of forms and the repertoire of stamp types (Hayes style E[I])?° of this latter workshop seem to be somewhat limited: they are hardly comparable with the great number of Late Roman ARS forms, site-specific stamp types, and decoration patterns produced in D! fabric at the large rural production sites El Mahrine, Henchir el-Biar, and Bordj el-Djerbi near Tebourba?! which all started ARS production around A.D. 320/330. Another important (though still unlocated) N Tunisian production-centre manufacturing

ARS forms in D? fabric was suspected in the region between Zaghouan, El Fahs and Oudhna.?? These so-called D? potteries were re-named "atelier X" by Bonifay (fig. 1).? During the first half of the 6th c. this large atelier produced ARS vessels decorated with elaborate central figural stamps with additional subsidiary motifs (Hayes style E[II]),^* a style of decoration which is unknown either at the Late Roman D? potteries at Oudhna? or at the C? potteries at

Sidi Marzouk Tounsi.?$ In his comprehensive account of Late Roman pottery from Tunisia it was not possible for Bonifay to deal with all of the ARS forms and their decorations. Nevertheless, it is surprising that the most famous category of relief-decorated ARS bowls and plates is not taken into further consideration either during the discussion of ARS production-centres or in the typology

chapter.’ This category consists of the appliqué-decorated ARS forms in €? fabric such as a variety of bowls Hayes 51 A/B, 52 B, 53 A, 54 and 55, as well as the mould-made reliefdecorated plates Hayes 56, 89 A and the rare 82 B var. in C* fabric, all produced at Sidi 16 17

Ibid. 48; 162 fig. 87.1-5. Recently for the plain A! and A? ware and the appliqué-decorated A!/? ware cf. Mackensen and Schneider 2006, 168-73 Table 1; for Oudhna, cf. iid. 2002, 158 Table 5.

18

Bonifay 2004, 53-57 figs. 24-26. For Pheradi Maius see now Ben Moussa 2007.

19

Ibid. 55; Brun 2004, 243; cf. Barraud et al. 1998, 140 fig. 7 (kilns 4-7).

20

For the stamp-types, cf. Gauckler 1897a, 456 pls. 8-9; Mackensen 1993, 27-32 figs. 4.1-30 and 5.1-9; for the ARS forms, cf. Barraud et al. 1998, 149 fig. 11; Mackensen and Schneider 2002, 129 f. fig. 4; Bonifay

2004, 181 fig. 96.1, 2, 4, 5; 191 fig. 102.1-2 (A[III] /E[1] and E[I]). 21

Mackensen

1993, 1998a, 30-33, and 1998c; Mackensen

and Schneider 2002, 125-27 fig. 3, 143-49 figs.

18, 20.1-11. 22

Mackensen 1993, 441-48 and 457 f.; id. 1998a, 33-36; Mackensen and Schneider 2002, 128 and 149 f; cf.

23 24 25 26

the commentary by Bonifay 2004, 50 on the heterogeneity of the chemical reference group D; in particular concerning the ARS forms and stamp types of the 4th and 5th c. there are difficulties so long as the site is not known and surveyed. Bonifay 2004, 46 fig. 22, and 49. Mackensen and Schneider 2002, 149 f. fig. 21.2-8. Cf. Barraud et al. 1998. Cf. Mackensen 2003b, 107 f. fig. 2.3 pl. 7 with examples of rare figural stamp motifs.

27

Bonifay 2004, 189 (s.v. décors).

Technology and organisation of ARS ware production-centres

21

Marzouk Tounsi.?? These forms are known not only from Carthage and various sites in Tunisia? and Algeria, but were also widespread over much of the empire, being reported even from Egypt and Syria.?? What is really needed for each ARS production-centre, either surveyed or excavated, is a documentation of the potters' tools and kiln furniture such as the protective equipment for firing ARS ware, the so-called cylindrical saggars (see below), along with the range of vessel-forms and lamps (including wasters and discoloured sherds) and their frequencies in the material available. No less important is the range of decoration and relief-decoration motifs (noting differences in size and details), either appliqué or mould-made, and of the stamp types as well as their decoration patterns?! with reliable catalogues, drawings and photographs. Good macroscopic descriptions of the respective fabrics are also required. Other urgent methodological requirements include comparable quantitative chemical analyses (Wavelength Dispersive X-Ray Fluorescence [WD-XRF]) with major and trace elements of each productioncentre, be it of local, regional or supra-regional

these demands:

importance.**

Up to now, only two sites meet

El Mahrine?? and, with certain reservations, the preliminarily-reported Late

Roman pottery workshop in the Baths of the Laberii at Oudhna.“ 2. Technological aspects of the production of ARS ware Potters' tools: the so-called pugilla with graffiti (see catalogue on pp. 40-43) Many tools used in a pottery were of organic material (e.g., hardwood from olive trees) and therefore can rarely be proven in archaeological contexts, except in an arid zone. Besides these, bone shapers and pins, as well as shapers of polished stone and metal spatulas, stili and pipes of different diameter, were noted in excavations of Roman pottery workshops. potteries in N Africa, however, many of the tools were formed of clay.

In Late Roman

Already in 1896 P. Gauckler detected several tools in a Late Roman workshop which had been installed in the abandoned Baths of the Laberii at Oudhna. Publication included not only plaster moulds for the production of ARS lamps of type Hayes II B/Atlante X Ala and Ale (with one or two opposite nozzles) but also some clay tools (mostly oval) for polishing (socalled pugilla), with a graffito ante cocturam on one of the two concave sides. Some of these pugilla show a cross-monogram or a potter's name (Renati), while on others the owner or tenant of a workshop is quoted in a graffito (ex of(ficina)/abis[...]). There are also two punches, one with a graffito Per. ?? Characteristic oval tools were recorded by L. Poinssot and KR. Lantier as surface finds from Bordj el-Djerbi, one of them with a 4-line graffito, [...]mi/[...] felici/ter bono

tuo/T...]emmeni, probably an invocation of blessing on the potter's work.% The most important evidence came to light during regular excavation lum Tidditanorum /Tiddis in 1957: some 250 tools among them a punch bearing a graffito using na/Feli/cis (sic). So far this exceptional source

of a workshop in a potter's quarter at Castelof different form and function were recorded, both capital and cursive letters (e)x osici/ for the knowledge of pottery technology in

28

Cf. Mackensen and Schneider 2002, 131-34 fig. 5 and 151-55 figs. 22-24.

29

Cf. Mackensen

30 31

Hayes 1972, 83-91 and 136 f. Cf. Mackensen 1993, 441-57 with a detailed analysis of the stamped ARS ware from northern central Tunisian production centres. Cf. recently Mackensen and Schneider 2006 for mid-Roman ARS ware; iid. 2002 for Late Roman ware. Mackensen 1993 and 1998c; for the chemical reference group see Mackensen and Schneider 2002, 27 and 134-37; with 41 analysed samples in particular see 143-45 nos. 1-41 fig. 18.1-20; 156 Table

32 33 34

35

2004a, 152-57.

Barraud et al. 1998; cf. also chemical analyses and reference-groups:

and ARS 1251.

Mackensen and Schneider 2002,

140 f. and 150 f., Table 5; Brun 2004. Gauckler 1897a, 454-56; id. 1897b, pl. 43 = Mackensen 1993, 28 fig. 3; 77 nos. 1-7; cf. also a further pugillum(?) at Carthage with a graffito ex oficina Abedonis, but of uncertain provenance: Gauckler

36

1897a, 455 n.1; Mackensen 1993, 78. Poinssot and Lantier 1923, Ixxv f.; Mackensen

37

Lassus 1958, 261 fig. D.34; Guéry 1968, 275-79 figs. 7-9; Berthier 2000, 126 fig. 32.

1993, 474.

22

M. Mackensen

Roman Africa has been studied only selectively in short preliminary reports and in a recent

monograph on Tiddis.?5 In the course of the surveys I conducted at El Mahrine and Henchir el-Biar since 1981 and at Bordj el-Djerbi since 1997, various potters' tools were documented, analysed and discussed. Several fragments of mushroom-shaped tools with a handle, as well as so-called pugilla, bear graffiti in cursive script or in capital letters (amongst these a fragmentary two-line cursive

graffito combined with a cross-monogram??) or they simply show a palm-branch or a star.*° Apart from some kiln furniture (see below) and plaster moulds for lamps of different types, there are almost no potter's tools known from the British-Tunisian survey of the central Tunisian production-centres.?! Nevertheless, some material came to light in particular at Sidi Marzouk Tounsi, where the site has been destroyed through unauthorised clandestine digging organised by the local population since c.1970 in search of saleable antiquities.4? These activities resulted in most of the archaeological evidence (such as numerous plaster moulds for lamps of type Hayes II A and plaster moulds for appliqué decoration or for relief-decorated plates and dishes of forms Hayes 56 and 89 A) ending up in museum and private collections in Tunisia and Europe, as well as in N America. Some of the punches, models and moulds can now be re-attributed because of an archaeological analysis of the relief- and stamped decoration, as well as on the basis of chemical analyses and a comparison with the reliable reference group. In his Etudes, Bonifay also discussed interesting technological aspects of the production of plain ARS ware. The use of some of the familiar types of potter's tools for polishing the surface

of already dried and slipped vessels was confirmed.** He publishes

a mushroom-shaped tool

with a graffito Vitalis in cursive from the pottery workshops which were surveyed in the northern periphery of Oudhna, and a pugillum with the graffito Sapli in capital letters from Sidi Jdidi.4> Here it seems reasonable to mention again the pugillum type which I first discussed in 1993,4 as there are now unpublished specimens available which provide decisive evidence concerning the provenance of some of these pugilla with graffiti (cf. cat. 1 [below] nos. 1-6) by means of chemical analyses recently carried out by G. Schneider (Berlin), and by means of comparison with the known reference groups for the more important ARS production-centres.^? The objective of this small archaeometric project was to re-attribute these pugilla to ARS production-centres. These tools are of special interest as rare epigraphic documents. Because of 38

Lassus 1958, 252 plan, 255 and 261

39

109-32 figs. 30-33. Mackensen 1993, 71 figs. 14.3 and 16.2.

40

41

42 43

figs. D-E; cf. Mackensen

1993, 24, 68 f. fig. 17.1-2; Berthier 2000,

Mackensen 1993, 68-87 figs. 14-16 and 23; id. 1998c, 437 f. fig. 5.1-2; Mackensen

and Schneider 2002,

143 nos. 1-2 fig. 18.1-2 and 147 figs. 19.3 and 20.1. Peacock, Bejaoui and Ben Lazreg 1990, 73 fig. 9.a-d; 78 fig. 12.e-f; cf. Mackensen 1998b, 361-70 figs. 6.114 and figs. 7-9 (samples-tablet and clay punches); id. 2004a, 140 fig. 5.6 (mushroom-shaped tool). The provenance of the mushroom-shaped tool is known as it was found on the surface at Sidi Marzouk Tounsi; the samples-tablet and the punches could be re-attributed because of the stamp types which are specific for this production centre. Cf. Peacock, Bejaoui and Ben Lazreg 1990, 68; Mackensen 1998a, 26-30. Mackensen and Schneider 2002, 151 nos. 1-5 figs. 22.8, 23.1-3 and 24.1-4;

158 Table

6; Mackensen

2004b, 801-4 figs. 7-8; id. 2005a, 169-81 fig. 1 pl. 10.1; id. 2006a. 44

45

Bonifay 2004, 60.

Ibid. 59 f. nos. 1-2 fig. 29.1-2; for Vitalis cf. Kajanto 1965, 274; for the activities in the workshop in the Baths of the Laberii and

46 47

in the periphery of Oudhna

cf. Barraud

et al. 1998,

161

fig. 18.34; for the

production centre in the periphery of Sidi Khalifa, cf. Pavolini and Tortorella 1997, 266 figs. 4-5. Cf. Ben Moussa 2007, 191 fig. 68 (Pheradi Maius); 218 f. Mackensen 1993, 74-83. Mackensen and Schneider 2002. A small pugillum with scratched-in palm-branches on both sides (Mackensen 1993, 78 no. 4, figs. 20.2 and 22.4) has now been analysed (sample no. 288, lab. no. V123) and can be attributed to Sidi Marzouk Tounsi because of its high titanium content.

Technology and organisation of ARS ware production-centres

Fig. 2. Sidi Marzouk Tounsi, central Tunisia. Potter’s tools, so-called pugilla. Scale 2 : 3. 1-5 with graffiti; 6 with stamped decoration (after Hayes style D). See Catalogue p. 41.

23

CHEMICAL COMPOSITION OF ARS POTTERS’ TOOLS (SO-CALLED PUGILLA) AND POSSIBLE KILN-PROPS FROM THE CENTRAL TUNISIAN PRODUCTION CENTRE SIDI MARZOUK TOUNSI

À

(analyses by WD-XRF of samples ignited at 880°C with Lo.i.=loss on ignition, major elements given as oxides, normalized to a constant sum of 100%) cat. no. sample no. lab. no.

SiO,

TiO; ALO,

FeO,

MnO

MgO

CaO

Na;

K,O

PG

V

Cr

Ni

Zn

Rb

TABLE 1. PUGILLA WITH GRAFFITI (1-6) AND WITH STAMPED DECORATION 1 2 3 4 5 6 7

213 222 221 271 220 174 287

H903 H951 H950 T499 H949 G726 V122

67.59 65.96 69.46 61.75 62.59 62.89 66.78

1.28 1.35 1.43 1.51 1.54 1.56 1.46

19.20 20.06 17.90 23.78 22.15 2322 1846

6.38 6.86 6.55 7.73 7.55 6.92 7.63

0.000 0.018 0.019 0.023 0.020 0.017 0.032

1.23 1.27 0.90 1.17 1.21 1.42 1.63

1.27 1.43 0.88 1.30 2.11 0.81 1.04

0.39 0.52 0.42 0.30 0.39 0.41 0.42

229 2.40 2.26 2.27 228 2.55 2.44

0.13 0.17 0.19 0.17 0.19 0.21 Q.11

118 94 94 111 104 106 104

106 107 92 133 115 124 105

25 29 23 40 29 20 20

68 70 56 77 79 81 106

92 92 86 96 102 101 118

Sr

Zr

Ba

(Ce)

loi.

273 249 236 328 323 225 291

461 439 525 302 363 351 407

544 504 485 418 440 466 499

127 101 104 137 134 170 136

1.29 1.10 0.93 1.54 1.18 1.70 1.36

219 315 286 270 246

362 360 395 334 370

484 493 452 534 532

130 100 94 103 110

2.06 1.66 1.25 1.53 1.58

(7)

TABLE 2. POSSIBLE KILN-PROPS WITH STAMPED DECORATION (2-6) 2 3 4 5 6

293 291 292 289 290

V128 V126 V127 | V124 V125

64.16 64.59 63.71 62.08 65.75

1.41 1.51 1.58 1.60 1.48

21.97 21.18 21.06 23.33 20.62

6.87 7.19 8.22 7.79 6.94

0.017 0.021 0.020 0.021 0.021

1.36 1.27 1.12 1.29 1.16

0.79 1.08 1.15 0.98 1.07

0.63 0.52 0.43 0.34 0.46

2.68 2.45 2.47 2.44 2.35

0.10 0.18 0.24 0.14 0.16

120 100 109 105 91

125 107 128 146 114

36 28 35 32 28

74 73 73 82 79

93 93 92 113 97

means and standard deviations for reference group Sidi Marzouk Tounsi calculated from 23-25 samples (after rejecting statistical outliers) SMT

mean +S

64.70 2.10

1.50 0.11

21.20 1.30

6.87 0.40

0.021 0.003

1.33 0.24

1.21 0.25

0.41 0.08

2.38 0.17

0.190 0.060

107 14

116 11

29 10

73 10

99 7

299 65

351 48

487 48

128 22

1.49 0.34

|

Technology and organisation of ARS ware production-centres

25

the lack of stamps carrying the name of the of{flicina on mid- and Late Roman ARS vessels (see

below) and because of a lack of ostraca and papyri, their graffiti are the only source of information on the organisation of Late Roman pottery workshops in N Africa. The technical term pugillum, meaning parvus pugnus (little fist), was first identified by J. W. Salomonson on one of these oval tools*? which normally have two concave sides (rarely, one concave and one convex side). The length measures between 4.0 and 7.0 cm or between 9.5 and 13.5 cm, but even within these two groups there is no average or standard size. A typological classification according to the size and basic form (concave and/or convex sides) did not produce

convincing results.4? The most important pugillum is the one with a 4-line cursive graffito published by Salomonson (figs. 2.3, 4.2); the provenance assigned was "Tunisia" but without any knowledge of the region or the actual findspot.?? Despite the general difficulty of deciphering such a graffito, Salomonson gained all the important information and on palaeographical criteria suggested a date within the second half of the 5th and first half of the 6th c?! The graffito reads: ex officina Quod/vultdei Tzacunis/Cresce(n)s pugi/l(Dum fecit The double consonants in the words officina and pugillum were reduced to one, which seems to be quite usual. The owner or perhaps the officinator with the Christian-theophoric name Quodvultdeus mentions his filiation just with the father's name in the genitive without adding f(ilius), a formula which is common in N Africa, as is the usage of Christian-theophoric names.

Tzacunis is the undoubtedly the autochthonous, non-Romanised name of his father.?? Quodvultdeus was probably not the owner but the tenant and responsible senior potter of this officina ^? t was thought that Cresce(n)s was an assistant ("Werkstattgehilfe") in the workshop? but I have suggested that the literate Crescens who produced the pugillum and inscribed the graffito in good cursive ante cocturam in the clay might have been a potter who worked with an individual contract and certain responsibilities in the workshop of Quodvultdeus.?? Recently the provenance of this pugillum was identified by means of chemical analysis as Sidi Marzouk Tounsi. The result of the analysis perfectly fits the reference group with a very high titanium (TiO?) content which is specific to this major central Tunisian ARS production-

centre (Table 1 no. 6, sample 174).% A further pugillum with two concave sides shows a 3-line graffito pugillu(m)/Victoris / Magni in Late Roman cursive (figs. 2.4, 4.4).°” It is the second known potters’ tool mentioning the ancient technical term. The two common personal names, both in the genitive, can be interpreted as the names of an officinator Magnus and a potter Victor. There is a third pugillum with a 2-line graffito ex of(f)ici/na Magni in cursive (figs. 2.5, 4.3).°° The identity of the name and, in particular, the results of the chemical analysis with the characteristic high titanium content of these two tools emphasize the probability that both pugilla belonged to the workshop of a certain Magnus at the potters' village at Sidi Marzouk Tounsi (Table 1 nos. 4-5, samples 271 and 48

Salomonson

49

For the group of pugilla and the differentiation of two types (type 1 with two concave sides; type 2 with

1982, 348 and 353.

a concave and a convex side), cf. Mackensen

50 51 52 53 54 55 56

1993, 74-83 figs. 18-21, 22.1-2 and

22.4-5; for a second

potter's tool with graffito with the technical term pugillum, cf. ibid. 78 type 1 no. 2 figs. 20.1 and 22.1; Mackensen 1998c, 438 fig. 5.1-2; id. 2004a, 138 f. fig. 5.1-5. Salomonson 1982, 343-56, especially 345 figs. 1-2. Ibid. 354 f. Ibid. 351 f.; cf. M. Khanoussi and L. Maurin, Mourir à Dougga. Recueil des inscriptions funéraires (Ausonius Mémoires 8, 2002) 463 no. 1165 (Zocunis). Mackensen 1993, 474. Salomonson 1982, 352. Mackensen 1993, 83 and 474 f. Mackensen 1993, 78 figs. 18.1 and

19.1; id. 2004a, 138 f. fig. 5.1; Mackensen

and Schneider 2002, 151

no. 1 fig. 24.1; 158 Table 1 (SMT 1). 57 58

Mackensen 1993, 78 no. 2 figs. 20.1 and 22.1. Salomonson 1982, 357 n.6 fig. 2; Mackensen 1993, 78 no. 3 fig. 18.2 and 19.2; id. 2004a, 139 fig. 5.2.

26

M. Mackensen

^N ^

/

é

»

4

*

;

|

:

-

3:

|

vie

à

i

.

3

pe

3l |i

5

a— |

m

|

j Ix

4 .

:

13

utt

*

,

^

4

p

x! ag i ^

fF,

|

:

;

e»

te V

Fig. 3. Sidi Marzouk Tounsi, central Tunisia. Potter's tool, so-called pugillum with graffito. Scale 2 : 3. See catalogue p. 40.

220). Despite the missing archaeological context for these two pugilla, a more general dating in the 5th and the first half of the 6th c. can be entertained on the basis of palaeographical criteria. For two smaller pugilla no findspot was indicated but a (central) Tunisian provenance seemed probable. Both show graffiti in capital letters with the names Paulinus (fig. 2.1) and Nar[.]

(fig. 2.2), which might be completed to Nardus.?? The lettering of Paulinus, the diminutive of Paulus, was carried out rather carefully — in contrast to Nar[.] — and some of the letters show characteristic rounded serifs; however, a dating seems difficult (probably 4th or [first half of the] 5th c.). Both names are known in N Africa in late antiquity.9 The pugillum of Paulinus was partly blackened in a fire. The results of the chemical analysis of these two pugilla confirm a provenance in the workshops at Sidi Marzouk Tounsi (Table 1 nos. 2-3, samples 222, 221). Another Hayes 175 (Hayes 173 rarely also

small pugillum shows, instead of (style D) (figs. 2.6, 4.5).61 Stamps of and 174) on C? dishes Hayes 82 and with cross-monogram stamps which

a graffito, a small stamped fish similar to type small fish showing scales or even the backbone 84 are normally combined with lentoids or birds, can be dated to the mid- and second half of the

5th and first half of the 6th c.f? The chemical analysis and the high titanium content (TiO?) confirms an attribution of these stamp types of style D to the production-centre at Sidi Marzouk Tounsi (Table 1 no. 7, sample 287). Of particular interest is an extraordinarily large pugillum, hitherto unpublished and without recorded provenance (except "probably Tunisia"). It shows a 2-line graffito in well-spaced and carefully-written capital letters with characteristic oblique serifs (figs. 3, 4.1); also significant is the bifurcated L at the beginning of the name Lucitela. The average height of the let59

Mackensen 2004a, 139 figs. 5.3 and 5.5.

60

Cf. Kajanto 1965, 244 and 336.

61 62

Cf. Hayes 1972, 220 f. and 256 fig. 47f. Ibid. 256; cf. Mackensen 1998b, 361 figs. 4.5 and 5.7.

Technology and organisation of ARS ware production-centres reat

Hu

ay

E

E

^ * "

LA

*

e

C E b

27

, "1

|

M

"d

* a

v

3

>

"

LS

|

E

%

>

|

!

»

a

,

LI

Fig. 4. Sidi Marzouk Tounsi, central Tunisia. Potter's tools, so-called pugilla. 1-4 with graffiti; 5 with stamped decoration (after Hayes style D). Scale 1 : 1

ters is 0.6 cm. The ductus of this graffito and the ordination of the letters is reminiscent of monumental inscriptions of the first third of the 3rd c., in particular of the Severan period, and a similar dating for this pugillum seems likely. When the writer had to correct the spelling of Rogatinni by providing a second n, he simply overwrote the i with the letter n. The graffito reads: ex of(ficina Rogatinni/Lucitela Zranei The name Rogatin(n)us, the diminutive of Rogatus, which is one of the most common names in N

Africa,€? is almost exclusively recorded in the N African provinces.“ The name Lucitela is undocumented so far in this region; it might be a diminutive of Lucitas,9 while Zraneus is one of the typical autochthonous

names,

in this case Lucitela's father's name

in the genitive. Roga-

tinnus might have been either the owner or the tenant of the workshop (in any case, the officinator of the pottery) and Lucitela was probably a potter of the Rogatinnus workshop. The results of the chemical analysis of this pugillum confirm a provenance from the pottery village at Sidi Marzouk Tounsi (Table 1 no. 1, sample 213). The dating of the mid- and Late Roman graffiti on potters' tools on the basis of palaeographical criteria is difficult, yet now there is epigraphic evidence of the first period of the

production-centre at Sidi Marzouk Tounsi, founded in the first quarter of the 3rd c.96 From the time of the assumed Severan dating of the pugillum of the workshop of Rogatinnus (figs. 3, 4.1) with its literate potter Lucitela, son of Zraneus, we have good reason to postulate that this officina was already active from the beginning of the production of plain and appliquédecorated red slip vessels of C! fabric. The pugillum of Paulinus (fig. 2.1) may be dated to the Late Roman period, whereas three pugilla with graffiti which mention the officinae of Quodvultdeus and Magnus (figs. 2.3-5, 42-4), in characteristic late cursive, probably belong to the second half of the 5th and first half of the 6th c. — i.e., the prolific late period of the produc-

tion-centre at Sidi Marzouk Tounsi, which came to an end by the mid-6th 63 64

c.°” These 5 pugilla

Cf. Lassère 1977, 285-89. Kajanto 1965, 297.

65

Cf. Kajanto 1965, 173.

66 67

Mackensen and Schneider 2006, 166 f. fig. 2; Mackensen 2006b, 173 f. Forthe end of the production centre at Sidi Marzouk Tounsi, cf. Mackensen 1998b, 370 with n.77; id.

2003b, 108.

28

M. Mackensen

M 050m

rm

_

[A GE

— —

M

SS

_

=

=

>

N= ‘= C -ME

E Z

NE

o]

|

N=

=

|

Ss

a

b

C

——À d

Fig. 5. Possible reconstructions of firing stacks of ARS vessels of form Hayes 99 with protective kiln furniture (cylindrical saggars) (after Bonifay 2004, 63 fig. 31.a-d).

with graffiti, all of them now attributable to Sidi Marzouk Tounsi, are rare epigraphic docu-

ments illustrating the identity and labelling of special tools$? as practised at the highly specialised workshops in central Tunisia's chief ARS ware production-centre. Cylindrical

saggars: protective

kiln furniture

These potter's tools, all of fired clay and mostly red-slipped, are not alone in being regionally specific, for the thick-walled cylindrical saggars with an almost horizontal, broad rim and no bottom/floor (“casettes de cuisson" or "zylindrische Brennhilfsmittel"), and which are

designed to be stacked, are also typical of N African (Tunisian) fine-ware potteries. They protected the red slip ware during firing in the kiln against fumes and soot and are known from all

the important ARS production-centres such as Sidi Marzouk Tounsi,/? Henchir es-Srira,”° Henchir el-Guellal near Djilma/7! El Mahrine (fig. 7.5, figs. 8-9), Henchir el-Biar and Bordj el-

Djerbi (fig. 10),7^ and now also from Oudhna and Sidi Zahruni.? To judge from the frequency of fragments of these saggars on the surface of rural production sites in NE Tunisia,”“4 they were essential for firing ARS ware and protecting it from damage in kilns caused by faulty firing. These rarely published saggars from Sidi Marzouk Tounsi and El Mahrine were drawn (with

sections) in the normal position with their rims on top.” 68

Cf. clay punches with rare graffiti: a punch with a graffito Per in capital letters from Oudhna (supra n.35), another one (type Hayes 289 A) with a graffito in capital letters Abmoe from Sidi Marzouk Tounsi (now confirmed by chemical analysis): Mackensen 1998b, 369 figs. 6.13 and 9.2; a further punch without findspot but probably Sidi Marzouk Tounsi (similar type Hayes 294 var.) with a graffito of a Christian-theophoric name i(n)speseindeo in italics mentioned by Mackensen 1993, 478 with n.68; for the name Spesindeo cf. Kajanto 1965, 217. Cf. further pugilla with ornamental or figural graffiti, probably from central Tunisian ARS production centres: Mackensen 1993, 78-80 nos. 5-6 figs. 21.1-2,

22.2 and 22.5; id. 2004a, 140 fig. 5.4. 69

Peacock, Bejaoui and Ben Lazreg 1990, 68 and 70 figs. 7.1 and 8.a.

70

Ibid. 76 fig. 12.a-b.

71

Ibid. 81 fig. 14.a-n.

72

73 74 75

Poinssot and Lantier 1923, Ixxvi; Mackensen

1993, 88-94 fig. 24 and 291 no. 8 fig. 101.5; id. 2004a, 142

f. figs. 6-7. Barraud et al. 1998, 162-65 figs. 19-20; Bonifay 2004, 60-65 fig. 30.1, 4, 6 and 8. (Cf. the numbers given for El Mahrine by Mackensen 1993, 88 Table 1. Cf. nn. 68 and 71.

Technology and organisation of ARS ware production-centres

29

Bonifay suggested four different possibilities for arranging ARS vessels in such a saggar (fig. 5). Even without having archaeological evidence at his disposal, he favoured his hypothetical proposal with the saggar upside down (fig. 5c); here the almost horizontal rim was used as base or foot and the wide opening was on top of the saggar (diam. 24 cm, h. 25 cm). Type- and size-identical vessels were piled up with foot on floor in the usual way."6 According to Bonifay, the dried ARS vessels were put into cylindrical saggars outside the kiln and then the filled

saggars were transported

into the firing chamber." The idea of carrying piles of highly

fragile, thoroughly dried ARS vessels, already set up in a saggar, from the drying sheds to the kiln does not seem very convincing. More probable is that dried vessels were piled up singly in the kiln itself, most probably fixed at crucial points with insulating material (i.e., clay pads of different size) or even with small potsherds where necessary for the stability of the stacked fine ware. The usage of clay pads or clay strips in N Tunisian potteries could be demonstrated at El Mahrine where overfired dishes and bowls showed sintered clay on the underside of the

wall and the bottom (fig. 6.1-2).8 After about a dozen vessels had been piled up, the stack would have been covered with one of the protective cylindrical saggars with the horizontal rim in top position. The complete process of loading a large kiln (furnus) was doubtless rather time-consuming. Very interesting is the calculated number

of vessels for one filling of a kiln of the size

preserved in the Baths of the Laberii.”? The quantity of vessels and saggars depends on the internal height of the kiln (1.7-1.9 m) and, of course, normal-sized bowl Hayes form 99, Bonifay reckons to the composition of the kiln load, his calculation size-identical vessels of the same form (cf. fig. 5),

on the with a seems since

size of the vessels. On the basis of a total of 2160-2880 bowls. In relation to be somewhat theoretical with its he has already refered to the well-

known heterogeneity of kiln loads.9 It is even more difficult to judge how the loading of the kiln was actually performed by the firing master who was responsible not only for the arrangement of the vessels, including the cylindrical saggars, but also for the important process of heating the kiln, firing the vessels, and cooling down the kiln again (which lasted at least several days). As usual, it is not very easy to reconstruct the conditions and the course of different working procedures in a Late Roman fine-ware pottery in Africa Proconsularis or Byzacena. But in connection with the discussion of cylindrical saggars and firing processes in production-centres of ARS ware, there is more, partly unpublished evidence available from my field surveys conducted at El Mahrine and Bordj el-Djerbi in 1997-99 which needs to be taken into consideration. Saggars from El Mahrine, which differ in diameter from height of 16.5 cm (observed max. height 21.5 cm [cf. Oudhna with the same type of ARS vessel: there is one large, rather survey of 1981 (El Mahrine site I) which is composed of three and

67/El

Mahrine

2.2, 4.2 and

9.1

c.20 to 55 cm, with an average 25 cm]), were not always filled striking fragment from the field dishes of forms Hayes 59 B, 61 A

of different size, sintered

together,

heavily

deformed,

partly glazed and completely discoloured (fig. 6.1).?! The dishes piled up in a saggar had melted because of poor regulation of the temperature in the kiln and consequent overfiring. In an almost completely preserved saggar (El Mahrine site IIa) noted during the field survey of 1986, there were still 6 deep dishes/bowls of form Hayes 67/El Mahrine 9.2 and 9.3 of different diameter and height with partly blurred stamped decoration, with one identical stamp 76 77

78 79

80 81

Ibid. 61 and 63 fig. 31.c. Barraud et al. 1998, 163; Bonifay 2004, 61.

Cf. overfired dishes and bowls with insulating material on the bottom: Mackensen 1993, 176-80 figs. 42.2, 3 and 5b; fig. 43.3-6. Barraud et al. 1998, 163-65; Bonifay 2004, 61.

Cf. for details M. Bonifay in Barraud et al. 1998, 140-46 figs. 2-6 (kilns 1-3) especially 165 for the mixed composition of the firing load with different products (ARS vessels, lamps, statuettes, coarse ware). Mackensen 1993, 176-80 figs. 42.4 and 43.4; id. 2004a, 144 fig. 8.4; Mackensen and Schneider 2002, 143

fig. 18.1.

30

M. Mackensen

KAMA KAA KA AAG AL —_——

| uere eene

Vn

J

=|

4

T

Hl m

SALES

FANS

2

de

TN

7

|

4 Fig. 6. El Mahrine, NE Tunisia. Scale 1 : 3. ] waster with sintered ARS vessels of forms Hayes 59 B, 61 A and 67/El Mahrine 2.2, 4.2 and 9.1; 2 Hayes 67/E] Mahrine 9.2 with sintered insulating material on lower wall;

3-4 reconstruction of piling ARS vessels in cylindrical saggars based on the evidence from El Mahrine (cf. figs. 6.1, 7 and 8.1) and combined with possible kiln-props from Sidi Marzouk Tounsi.

Technology and organisation of ARS ware production-centres

31

NU —

1

—

7

)

oo

2

M

_

3

—

|

M

X

EDAM 4

-

E

—-

a

RCez

=

os

eer

-

5

Fig. 7. El Mahrine, NE

Tunisia, deposit M

Ila-86/169-175

with 6 different ARS

vessels in a cylindrical

saggar. Scale 1 : 3 (after Mackensen 1993, 292 fig. 101.1-5). 1-2 Hayes 67/El Mahrine 9.3; 4 Hayes 76 A/El Mahrine 10.1; 3 Hayes 67/El Mahrine 9.2; 5 cylindrical saggar.

type alternating with a different stamp type (Hayes style A[II] and A[III]/El Mahrine I.2/1 and I.3/1), and one dish of form Hayes 76 A/El Mahrine 10.1 (fig. 7.1-4). These 7 vessels, with diameters between 24.5 and 36.0 cm and heights between 4.9 and 6.4 cm, were piled upside down in the saggar with its rim horizontal (its normal position; diam. 36 cm, h. 21 cm; fig. 7.5), with

32

M. Mackensen

their bottoms on top.#? But in a saggar 21 cm high there was space for at least a dozen vessels of form Hayes 67 or 76, even if they were not size-identical (cf. fig. 6.3-4). It is unlikely that it was only because of a slight discolouration and a spotty slip on the inside of the vessels that what amounted to almost half of the whole load (together with its saggar) was separated out as rejects and thrown on the rubbish dump; there must have been a more severe damage — for instance, cracks or complete breakage. Already these two examples of the combination of different ARS vessels (figs. 6.1, 7.1-4) demonstrate

forms been same type-

that, at least at El Mahrine

in the 4th and

first half of the 5th c., various vessel

of different size were piled up in one and the same cylindrical saggar. This might have the exception to the rule, but there is no clue to the ratio of different vessels in one and the saggar. However, it seems reasonable not to assume that all saggars had to be filled with and size-identical vessels.

There is one further interesting fragment of a cylindrical saggar (diam. c.44 cm), with horizontal rim and with two small circular stamps on it, from the field survey carried out at El

Mahrine I in 1999 (fig. 8.1). The vertical wall and the rim were heavily deformed during overfiring of the kiln; the top of the rim was partially glazed and carried traces of insulating material in the form of lengths of a clay strip. The characteristic rims of two deep dishes/ bowls of form Hayes 67/El Mahrine 9.1 were sintered to the interior vertical wall of the saggar, although the wall and bottom of these ARS vessels are missing. There were also traces of sintered third and fourth (unclassifiable) vessels which had broken off. The groove on the upper side of the rim and the hooked rim of the two best-preserved rims are characteristic of form Hayes 67/ El Mahrine 9.1 and indicate the original position in the saggar; but it is far from clear that the saggar was orientated upside down, as in Bonifay's proposal (fig. 5.c). If, however, these dishes of form Hayes 67 were piled up with their bottoms orientated upwards and the rims downwards, the lowest dish was placed either on top of the horizontal rim of another saggar in the lower position or on the floor of the kiln (fig. 6.3-4). For structural reasons and for stabilising a pile placed directly on the firing floor, the lowest dish probably needed a kiln-prop as support (see below for several stamp-decorated examples from Sidi Marzouk Tounsi [fig. 11.1-6]). The cylindrical saggar with its horizontal rim on top could then be put as a protection over the pile of ARS vessels, as suggested by the evidence of the existing fragment (as reconstructed on fig. 6.3-4, but omitting the necessary insulating material between single dishes). A second deformed fragment of a cylindrical saggar from El Mahrine I8* provides us with additional information about stacking these saggars in a kiln. The horizontal rim of a cylindrical saggar is sintered upside down, but slightly slipped out of its correct position, on top of the horizontal rim of a second saggar (fig. 8.3); both saggars measure 37 cm in diameter. Between the two fragments there is a thin layer (c.5 mm) of clay which was used as insulating material. Obviously the saggars were sometimes stacked up in this rather unexpected way as well, and different orientations of the piled-up saggars were applied. As far as the question of positioning a cylindrical saggar with the rim upwards or upside down in the kiln is concerned, a further argument needs to be considered. Already in 1923 L. Poinssot and K. Lantier had observed graffiti on the horizontal rims of saggars from Bordj elDjerbi.5 L. Maurin and J. Peyras mentioned two graffiti [ex] oficlina...] and [ex] oficina Titi) [...] on the rims of two saggars from the same site.?6 From the surveys carried out at El Mahrine in 1981-92, only two rim fragments with stamps of single circles and a single capital letter are 82

Mackensen 1993, 89 and 291-93 fig. 101.1-5; cf. Bonifay 2004, 61 who speaks of a cut-out bottom

83 84

instead of a horizontal rim. Cat. 2.1 no. 2. Cat. 2.1 no. 4.

85

86

Doinssot and Lantier 1923, Ixxvi.

Maurin and Peyras 1973, 34.

Technology and organisation of ARS ware production-centres

BS -

233

^x X

Hayes

67

Fig. 8. EI Mahrine, NE Tunisia.

1-3 deformed and sintered cylindrical saggars (survey 1999). Scale 1 : 3.

known (fig. 9.2).87 Much better evidence exists from the central Tunisian centres Sidi Marzouk Tounsi, Henchir es-Srira and, in particular, Henchir el-Guellal near Djilma, where many fragments of saggars were published. On their rims occur either complete or abbreviated names,

written in capital letters or cursive,99 and ornamental and figural stamps such as concentric cir87 88

Mackensen 1993, 88 figs. 24.3 and 24.5. Peacock, Bejaoui and Ben Lazreg 1990, 70 fig. 8.a and 79 fig. 14.g-n.

34

M. Mackensen

I-—

Fig. 9. EI Mahrine, NE Tunisia. Scale

| : 3.

1-2 saggars with graffiti on horizontal rim (survey 1984 and 1992); 3-5 cylindrical saggars, partly with sintered insulating material (survey 1999).

cles, palm branches and birds,? which are partly known from stamped ARS vessels. Bonifay observed that saggars at the workshop graffiti on the horizontal rim.?°

in the Baths of the Laberii at Oudhna

rarely show

In the meantime, more evidence from El Mahrine and Bordj el-Djerbi can be provided. From the first site there is a fragment with a 5-pointed star (fig. 9.1), scratched onto the rim ante 89

Ibid. 76 fig. 12.a-b and 79 fig. 14.a-f.

90

Barraud et al. 1998, 162 fig. 20.38-39. Cf. Ben Moussa 2007, 188-90, fig,. 67.

Technology and organisation of ARS ware production-centres

35

Fig. 10. Bordj el-Djerbi, NE Tunisia. 1-4 cylindrical saggars, partly with graffiti on horizontal rim (survey 1997-99). Scale | : 3. cocturam,?! and one mentioned

above with two circular stamps on it (fig. 8.1).?? But most of the

rim fragments of saggars from El Mahrine (sites I and IIa-c) do not show any signs or letters. The rim of a large fragment of a cylindrical saggar from Bordj el Djerbi shows the graffito in capital

letters Ex of(f)icin[a ..] (fig. 10.1); the part bearing the name is missing,? and on each of two rim fragments of other saggars only three letters in cursive are partly preserved (fig. 10.3-4)?4 91 92 93 94

Cat.2.1 Cat.2.1 Cat. 22 Cat. 22

no. 5. no. 2. no. 1; Mackensen 2004a, 141 fig. 7.1. nos. 3-4; BD 98/B1.

36

M. Mackensen

In his preliminary report on the survey of central Tunisian production-centres, D. P. 5. Peacock pointed out that the graffiti on the rims of cylindrical saggars, mainly observable at Henchir el-Guellal near Djilma but in small numbers also at Henchir es-Srira and Sidi Marzouk Tounsi, were necessary to identify the work of individual potters after firing. He assumed that "only the top of a saggar pile would have needed labelling in this way", which gives a hint towards communal firings of at least two or a few individual potters (or work-

shops).” It seems reasonable to mark the horizontal rim of a saggar when stacked in the normal position (and not upside down). Only then the explanation put forward by Peacock makes sense and the stacks of saggars could be attributed to a potter or a workshop (officina) according to the inscription on the rim. In this case the ARS vessels of different potters using one and the same large kiln at a single site could be sorted out and identified. This was obviously the case at the rural production-centre at Bordj el-Djerbi, where some saggars were marked with ex of(f)icina [...] and [ex] of(f)icina TI...] and other illegible cursive graffiti (fig. 10.1 and 10.3-4). However, at nearby El Mahrine, where two large pottery sites with kilns (I and IlIa-c) were identified at a distance of c.400 m from each other,% on the saggars no graffiti with (abbreviated) names but only a few signs or single letters are known. Similarly limited seems to be the number of saggars with graffiti from the workshop in the Baths of the Laberii at Oudhna. Surprisingly, at the large potter's village at Sidi Marzouk Tounsi, where several contemporary workshops with kilns and all the other necessary installations are to be expected, just one saggar with a graffito Sitali or Stalt in capital letters was published in the preliminary report. In contrast stands Henchir el-Guellal near Djilma with many partly-abbreviated names scratched into the rims of the saggars ante cocturam. Perhaps the observed differences reflect the organisational structure of the production-centres (e.g., the number and size of workshops and kilns).

2.3 Possible kiln-props

with

stamped

decoration

Small distance pads or props, which were probably needed when piling up ARS dishes and bowls in the kiln, have not been identified either in the ARS production-centres at El Mahrine and Bordj el-Djerbi or in the workshop in the Baths of the Laberii at Oudhna. But some objects can now be published (cf. cat. 3, nos. 1-6), all lacking graffiti but decorated with stamped

ornamental and figural motifs of Late Roman stamp styles B, C and D

(after Hayes).? These

might have been used as kiln furniture and, in particular, as props or distance pads, as is shown in the two reconstructions (fig. 6.3-4) of piled-up ARS vessels in cylindrical saggars. These props measure between 2.8 and 4.3 cm in height and between 2.7 and 4.4 cm in diameter. A fragment of this postulated type of kiln furniture, a small cylindrical prop (h. 3.3 cm) with circular base and top (diam. 4.1 cm) having concave sides (fig. 11.1), is known from Sidi Marzouk Tounsi. Both base and top are decorated with stamped palm-branches with short stem

of type Hayes 9 (style B); originally two identical branches were stamped on each side’? A second specimen

(h. 3.9 cm, diam. 3.7 cm) is fully stamp-decorated

(fig. 11.2), with even

the

concave sides decorated with two rows of small, dart-shaped palm-branches with short stem of type Hayes 6 (style C)? between them runs a horizontal line of three concentric circles of type

Hayes 26 (style B);!9° the stamped decoration of top and bottom is identical and shows 8 stamps, each of three concentric circles of type Hayes 26 (style B) in an outer band surrounding a central rosette-like stamp which is known as relief-decoration on the shoulder of lamps of type

Hayes II A/Atlante X Ala from Sidi Marzouk Tounsi.!9! A third specimen (h. 4.2 cm, diam. 3.6 cm) shows an identical stamped decoration on both ends (fig. 11.3): the central stamp is an elaborately decorated square with a diagonal cross with alternating dot and angle lines similar to 95 96

97

Peacock, Bejaoui and Ben Lazreg 1990, 79. Mackensen 1993, 45-51 fig. 10, and 461 f.

Forstyles B, C and D and their dating, cf. Hayes 1972, 219-21.

98 99

Cf. Hayes 1972, 231 fig. 39.b. Ibid. 231 fig. 39.j; cf. also Peacock, Bejaoui and Ben Lazreg 1990, 72 fig. 8.q.

100

Hayes 1972, 235 fig. 39.q.

101

Peacock, Bejaoui and Ben Lazreg 1990, 73 fig. 9.d.

Technology and organisation of ARS ware production-centres

37

Fig. 11. Sidi Marzouk Tounsi, central Tunisia. Possible kiln-props with stamped decoration (after Hayes style B, C and D). Scale 2 : 3.

type Hayes 96 (style D);!° at each corner of the small square there is a diagonal line of three stamped large dots and in the angles are two small concentric circles. A fourth specimen (h. 4.3 cm, diam. 4.4 cm) is stamped on both ends with a cross-monogram (fig. 11.4) similar to type Hayes 289 A (style D) with small hooks at each end of the arms, and under the left and right arm respectively there is a Greek A (upside down) and w;1% the edges of one end are damaged. The surfaces of the stamp-decorated ends of a fifth specimen (h. 3.2 cm, diam. 3.7 cm) are heavily damaged. Two of the three different stamps on each side are partly preserved (fig. 11.5): there is a four-petalled rosette, which is known as relief-decoration on the shoulder of

lamps of type Hayes II A/Atlante X Ala from Sidi Marzouk Tounsi,'™ a small cantharus with line- and dot-patterns similar to type Hayes 271 (style D), and a wild animal to left, with ribs, legs, mane and broad tail indicated (probably a badger).1% A sixth, rather small specimen (h. 2.8 cm, diam. 2.7 cm) shows on one end a stamp of a small rosette of 8 rounded

petals (fig.

11.6) and on the other three identical but blurred stamps which partly overlap. For these objects (except for the fragment which was picked up at Sidi Marzouk Tounsi [fig. 11.1]) no findspot is known. The stamp types used for decoration of fig. 11.2-5 can be attributed

to styles B, C and D (after Hayes), which are known from Sidi Marzouk Tounsi.!° [n addition, the chemical analyses (Table 2 on p. 24, nos. 2-6, samples 289-93), measured recently by G. Schneider, show a very high content of titanium (TiO?) which is specific for the production102

Hayes 1972, 247 fig. 44.f.

103 104

Ibid. 273 fig. 54.a. Peacock, Bejaoui and Ben Lazreg 1990, 73 fig. 9.g.

105

Hayes 1972, 270 fig. 45.1.

106 107

Cf. ibid. 253 fig. 48.b-c. Peacock, Bejaoui and Ben Lazreg 1990, 70 and 72 fig. 9; cf. Mackensen 1993, 449-51 and 454 f.

38

M. Mackensen

centre of Sidi Marzouk Tounsi.!95 They indicate the likely provenance of these possible kilnprops from Late Roman workshops at Sidi Marzouk Tounsi. 3. Organisation of ARS-ware production-centres in NE and central Tunisia [In the present state of research, and following a series of conventional surveys (at that time without geophysical investigations), at important ARS production-centres such as El Mahrine and Bordj el-Djerbi south of Tebourba (NE Tunisia) and Sidi Marzouk Tounsi and Henchir elGuellal near Djilma southwest of Kairouan (central Tunisia) (fig. 1), only limited conclusions can be reached regarding the organisation and infrastructure of these rural pottery-making

villages of differing sizes. ? All were established in areas of extensive production of olives and, in the region south of Tebourba, of corn too.!!? No buildings or installations of single workshops (officinae) with potters' wheels, kilns, wetting tanks, drying sheds, store-rooms, and dwelling-houses of the potters' families (perhaps also of necessary assistants) have yet been excavated at the above-mentioned mid- and Late Roman production-centres. The only exception is the workshop and kilns of the late 5th and 6th c. in the former Baths of the Laberii at Oudhna,!!! but at this site no information is available on other installations of this workshop or the living-conditions

of the potters (figuli). For the isolated site I at El Mahrine,

which,

to

judge by the area of the scattered wasters and discoloured sherds, measures c.60 x 75 m (0.45

ha),!? one could expect 3-4 workshops. The known graffiti on the rims of saggars (figs. 6.3-4, 7.5 and 8-10) from most of the rural production-centres (see above) and even from the workshop

in the Baths of the Laberii, either

with ex of(f)icina [...] or with abbreviations of names or just ornamental signs, point to shared use of large kilns for their output of ARS vessels by several smaller workshops and probably to

communal firings which were run by a special firing master.!!? This would explain the marking of the saggars for the identification of ownership in a kiln used by several workshops.!* In contrast to early and mid-Roman sigillata produced in Italy and the NW provinces, ARS ware does not show stamps with individual potters’ names combined with of(f)icina or fe(cit) at the centre of the vessel's floor. Notable in this context are the illiterate planta pedis stamps

on late 1st/early 2nd c. dishes of form Hayes 4 A in A! fabric (fig. 12).!45 Exceptional are a onehandled jug (lagoena) in the Rómisch-Germanisches

the 3rd c., which bears a graffito ante

Museum

Koln, dated to the first half of

cocturam in cursive ex of{fjicina/Viluti,l6 and

the

graffito (ante cocturam) in capital letters ex of(f)ici(na)/Navigi on the back side of a redslipped, relief-decorated and two-handled jug of the Navigius workshop,’ now attributed by chemical analyses to Henchir es-Srira (central Tunisia).1!5 The graffiti on the so-called pugilla (figs. 2-4) comprise the only epigraphic evidence which can be used in a discussion about ownership or tenancy of individual workshops of the 108

Cf. Mackensen and Schneider 2002, 130-34, 141 and 158 Table 6; iid. 2006, 174-77 Table 3.

109

Cf. for the size of these production sites the scatter of discoloured sherds, wasters and kiln furniture: for El Mahrine I (c.0.45 ha) and II (c.1.8 ha) and Bordj el-Djerbi (c.2.4 ha), cf. Mackensen 1993, 45-51 fig. 10;

110

Mackensen and Schneider 2002, 143 and 147; for Henchir el-Guellal near Djilma (c.4 ha) and Sidi Marzouk Tounsi (c.25 ha), cf. Peacock, Bejaoui and Ben Lazreg 1990, 69 fig. 5, and 76. Peacock, Bejaoui and Ben Lazreg 1990, 83; Mackensen 1993, 483-86; Bonifay 2003, 119.

111

Barraud et al. 1998, 140-46 figs. 2-6.

112

Mackensen

113 114

For the social status and the probable identity with the conductor or tenant of the pottery, cf. Mees 2002, 238 and 304. For the central Tunisian production centres, cf. Peacock, Bejaoui and Ben Lazreg 1990, 79; Mackensen 1993, 479; further Peacock 1982, 125 f.; Strobel 1987, 100-13 especially 105; Mees 2002, 304 and 31315.

1993, 45-49 fig. 10.

115 116.

Hayes 1972, 25. - Coll. K. W. Salomonson 1976, 126 no. 547 pl. 53.1-2; Atlante I, 76 pl. 134.4.

117

Cf. Salomonson 1976, 168 f. no. 622 pl. 652. The various signatures of the 3rd-c. workshop of Navigius and related potters are not mentioned in this context.

118

Cf. recently Mackensen and Schneider 2006, 178 f. Table 5.

Technology and organisation of ARS ware production-centres

39

Fig. 12. Mid-Roman ARS dish (fabric A!) Hayes 4 A with illiterate planta pedis stamp and triple-rouletted circle. | - at scale of 1: 2; 2 - at scale of | : 1.

3rd-6th c. The pugilla with modest 2- to 4-line graffiti, which all come from Sidi Marzouk Tounsi, are datable on the basis of palaeographical criteria. The graffiti mention either a workshop with the phrase ex of(f)icina combined with single names in the possessive case, or just a single name in the possessive or nominative case, or else an abbreviated name. In fairly good and practised, homogeneous handwriting (esp. figs. 2.3 and 3)!?? three of these graffiti identify two individuals each, two of them quoting their filiation with an autochthonous father’s name: approximate dating of graffito 1st third of 3rd c. 5th c. and 1st half of 6th c.

workshop and officinator ex of(f)icina Rogatinni ex of(f)icina Magni [ex of(f)icina] Magni

2nd half of 5th c. and 1st half of 6th c.

— ex of(f)icina Quodvultdei (f) Tzacunis

further persons Lucitela (f) Zranei Victor

Cresce(n)s

These graffiti have been interpreted as implying that rural pottery workshops with the necessary installations such as kilns, potter's wheels, and so on, were leased from a private landowner (probably a possessor) and located on his fundus, by individual potters (figuli) as responsible entrepreneurs (maybe conductores figlinae) or else that they were even the property of economically independent officinatores, probably senior potters of the workshop. The craft

was practised at the same site for generations in the same families,"? but we can only speculate on the initiatives for the foundation of new ARS production-centres — for instance, in the early 3rd c. at Sidi Marzouk Tounsi or around 320/330 at El Mahrine, Henchir el-Biar, and Bordj el-

Djerbi on the vast fundus of an anonymous possessor.!?! Rogatinnus, Magnus and Quodvultdeus were either the tenants (conductores) or even the owners of such an officina; but a decision about which of those is not possible on the available epigraphic evidence from Sidi Marzouk Tounsi. It is not clear whether

the well-known

contracts for the tenancy of Egyptian

amphora

potteries from the 3rd to 6th c., discussed by various authors,!? can be used as a reliable and even relevant model for N African fine-ware potteries at El Mahrine!# or Sidi Marzouk Tounsi 119

Cf. the graffito (E)x osici/na/Feli/cis (sic) of a punch from the Tiddis workshop: Guéry 1968, 275-79 fig.

120

7; Berthier 2000, 126 fig. 32. Cf. now Ben Moussa 2007, 219-22. Cf. Mackensen 1993, 474 f.

121

Ibid. 479 ff. especially 484-86; for the foundation of Bordj el-Djerbi, cf. Mackensen and Schneider 2002, 125 and 147. A cautiously entertained connection between central Tunisan ARS production centres and the foundation of El Mahrine, south of Tebourba (including a migration of some specialists), could not be proven: Mackensen 1993, 469; cf. Bonifay 2003, 120 = 2004, 479 f.

122 123

Cockle 1981; Strobel 1987, 92-97 and 106-13; Mees 2002, 209 ff. especially 238-60. Cf. already Mackensen 1993, 471-73 and especially 484-86; generally, Mees 2002, 234; for criticism, cf.

40

M. Mackensen

as well; nevertheless, the papyri do report on the differentiated commitments potteries and the relations between owner (possessor) and tenant (conductor).

in leased

The literate individuals Cresce(n)s and Victor who produced two of the pugilla (figs. 2.3-4, 4.2 and 4.4), were probably free potters (and not just assistants or Werkstatigehilfen) working in the officina of Quodvultdeus or Magnus. The wage which had to be paid (merces placita) by the conductor / officinator was paid on the basis of a special contract (contractum) which specified the exact number of ARS vessels which had to be produced; but neither of them indicates his position and relation to the conductor /officinator of the workshop nor his special craft skills and professional competence. Yet without archaeological investigation of the workshops it is impossible to decide if a single workshop was run by an individual senior potter as officinator, with the help of the male members of his family (including the juvenile children), or whether (depending on the degree of the workshop's size and specialisation) it had to be reinforced by further qualified potters. Furthermore, the issue of specialisation of particular workshops towards qualitatively demanding ranges of open and closed ARS vessel-forms and various applied techniques of decoration remains unresolved. But here the fact already postulated by J. W. Salomonson in 1962 and only recently confirmed by evidence should be mentioned: potters of Late Roman workshops at Sidi Marzouk Tounsi had direct access to various precious relief-decorated silver bowls and plates, as well as to con-

sular ivory diptychs such as the one of the consul of the year 408, Anicius Auchenius Bassus.l# They could use these objects, or just part of their decoration, for producing clay moulds.!* This might have been made possible by an aristocratic family, perhaps the possessor fundi. However, it is not clear if a member of the distinguished senatorial family of the Anicii, who possessed estates in the region of Uzappa/Ksour Abd el-Melek,U6 was personally involved.!?? The known examples of directly moulding silver vessels suggest that the practice was probably the exception and not the rule at Sidi Marzouk Tounsi. It may be argued that in Africa Proconsularis the conditions of tenure, rent and property in a fine-ware production-centre such as Sidi Marzouk Tounsi during the 3rd c. were not totally different from those prevalent in the 4th to 6th c., but details are far from clear. We have to expect that in the Late Roman period even nominally free potters had probably been tied to the rural production-centre as members of the plebs rustica (humiliores), but they might have changed workshop at the same site, taking with them personal punches (stamp types) or

moulds and other specific potter’s tools.!? FROM

CATALOGUE OF POTTERS' TOOLS AND KILN FURNITURE ARS WARE PRODUCTION-CENTRES IN NORTHERN AND CENTRAL

TUNISIA

1. Pugilla with graffiti from Sidi Marzouk Tounsi, central Tunisia 1. Pugillum with 2-line graffito in capital letters ex ofíf)icina Rogatinni/Lucitela Zranei. L. 14.5 cm, w 9.6 cm. Analysis sample no. 213, SMT (Table 1). Coll. K.W. (figs. 3 and 4.1).

Unpublished. 2. Pugillum with graffito in capital letters Paulinus. Analysis sample no. 222, SMT (Table 1).

L 6.9 cm, w 4.6 cm.

124 125

Pavolini and Tortorella 1997, 269 f. Forthe correct identification of the consul, see Spier 2003. Mackensen 2004b, 791 ff. (793 nn. 16-17 with references to the fundamental work of J. W. Salomonson), 797-99 figs. 2-4, and 803 f.; id. 2005a, 177-81 figs. 1 and 4 pls. 10.1 and 14.23.

126

Salomonson

127

Mackensen 2004b, 804. Cf. now for the Anicii in central Tunisia van den Hoek 2006, 212-15.

128

Cf. the evidence of rare stamp identities at the two different sites I and II at El Mahrine: Mackensen 1993, 475-79 fig. 127 and especially 478 for rare graffiti on punches quoting ex of(f)icina and a name in the possessive or just a name in the nominative. Cf. supra nn. 68 and 117, as well as Mees 2002, 299 f. and 302.

1973, 17 n.50; Mackensen 2004b, 794.

Technology and organisation of ARS ware production-centres

41

Coll. V.Q. (fig. 2.1). Ref.: Mackensen 2004a, 139 fig. 5.3. 3. Pugillum with graffito in capital letters Nar[.]. L 6.2 cm, w 3.9 cm. Analysis sample no. 221, SMT (Table 1). Coll. V.Q. (fig. 2.2). Ref.: Mackensen 2004a, 139 fig. 5.5.

4. Pugillum with 2-line graffito in cursive ex of(Pici/na Magni. L 9.4 cm, w 6.0 cm. Analysis sample no. 271, SMT (Table 1). Rómisch-Germanisches Museum Koln, inv. no. KL 604 (former coll. Karl Lóffler) (figs. 2.5 and 4.3). Refs.: Salomonson 1982, 344 fig. 2; 357 n.3; Mackensen 1993, 78 no. 3 figs. 18.2 and 19.2; 474 f.; id. 2004a, 139

fig. 5.2. 5. Pugillum with 3-line graffito in cursive pugillu(m)/Victoris/Magni. L 6.3, w 4.1 cm. Analysis sample no. 220, SMT (Table 1). Coll. V.Q. (figs. 2.4 and 4.4). Ref.: Mackensen 1993, 78 no. 2 figs. 20.1 and 22.1; 474 f.

6. Pugillum with 4-line graffito in cursive ex of(f)icina Quod/vultdei Tzacunis/Cresce(n)s pugi/l(l)um fecit. L 10.4 cm, w 6.5 cm.

Analysis sample no. 174, SMT (Table 1). Former Coll. Arch. Inst. Rijksuniversiteit Utrecht, Inv. no. B5 79.9; now Coll. V.Q. (figs. 2.3 and 42).

Refs.: Salomonson 1982, 348-53 figs. 1-3; Mackensen 1993, 78 no.1 figs. 18.1 and 19.1; 474 f.; Mackensen and Schneider 2002, 134, 151 no. 1 fig. 24.1, 158 Table 6, SMT 1; Mackensen 2004a, 139 fig. 5.1.

7. Pugillum with stamp type Hayes 175 var. (style D). L 62 cm, w 3.9 cm.

Analysis sample no. 287, SMT (Table 1). Coll. V.Q. (figs. 2.6 and 4.5). Unpublished. 2. Cylindrical saggars from El Mahrine and Bordj el-Djerbi, NE Tunisia 1. El Mahrine, sites I and IIa-c (mainly survey 1999) 1. Fragments of an almost completely preserved saggar with horizontal rim with traces of insulating clay on interior wall and on underside. H 21.5 cm, rim diam. 24.5, inner wall diam. 39 cm.

Musée de Carthage, storeroom DAI Karthago excavation, inv. no. M Ila-86/175 (fig. 7.5). Ref.: Mackensen 1993, 29] no. 8 fig. 101.5; id. 2004a, 142 fig. 6.2.

2. Fragment with horizontal rim with insulating clay stripe and two stamps of single circles on top; heavily deformed and discoloured vertical wall (7.5YR 7/6-5YR 6/6; surface 2.5 YR 3/0-4/4). On inside two rims of Hayes 67/El Mahrine 9.1 (10YR 4/3) sintered to the wall. Rim diam. 31 cm, inner wall diam. 39 cm. As no. 1 (above), inv. no. M 1-99/B7 (fig. 8.1). Unpublished.

3. Fragment of vertical wall with insulating clay on underside; heavily deformed and discoloured exterior wall (2.5YR 5/8-10R 4/0) with graffito (ante cocturam) [...]oi in capital letters. As no. 1 (above), inv. no. M 1-99/B5 (fig. 8.2). Unpublished. 4. Fragment with horizontal rim with insulating clay and, on top of its rim, sintered rim fragment of a second saggar, but upside down; surface of lower saggar with cracks, both are discoloured (10R 4/0-2.5YR 4/0). Rim diam. 25.5 cm, inner wall diam. 37 cm. As no. 1 (above), inv. no. M 1-99/B6 (fig. 8.3).

Unpublished. 5. Fragment of horizontal rim with graffito (ante cocturam) in form of a 5-pointed star. Rim diam. 16.5 cm, inner wall diam. 27 cm.

As no. 1 (above), inv. no. M IIb-92/B1 (fig. 9.1). Unpublished.

6. Fragment of horizontal rim with graffito (ante cocturam) of a single capital letter B. Rim diam. 16.5 cm, inner wall diam. 27 cm. As no. 1 (above), inv. no. M 1-84/26 (fig. 9.2). Ref.: Mackensen 1993, 94 no. 6 fig. 24.3.

7. Fragment with horizontal rim, heavily deformed, discoloured (2.5 YR 5/4-10R 4/0) and partly melted

42

M. Mackensen

surface and insulating clay pads. Rim diam. 23 cm, inner wall diam. 34 cm. As no. 1 (above), inv. no. M I-99/B4 (fig. 9.3).

Unpublished. 8. Fragment with horizontal rim and quadruple zigzag line on discoloured exterior wall (2.5YR 4/2). Rim diam. 25 cm, inner wall diam. 38 cm. As no. 1 (above), inv. no. M [-99/B3 (fig. 9.4).

Unpublished. 9. Fragment with horizontal rim and single zigzag line on discoloured exterior wall (2.5 YR 5/0) with traces of insulating clay patches and stripes on exterior wall and rim. H 17.0 cm, diam. 27 cm, inner wall diam. 39 cm As no. 1 (above), inv. no. M [-99/B1 (fig. 9.5). Ref.: Mackensen 2004a, 142 fig. 6.1. 2. Bordj el Djerbi (survey

1997-99)

1. Fragment with horizontal rim with graffito (ante cocturam) Ex of(ficin[a..] in capital letters and traces of insulating clay on interior and discoloured exterior vertical wall (2.5YR 4/0-10R 4/4). Rim diam. inner wall diam. 27 cm.

16.5 cm,

Musée de Carthage, storeroom DAI Karthago excavation, inv. no. BD 99/B1 (fig. 10.1). Ref.: Mackensen 2004a, 141 fig. 7.1. 2. Fragment with horizontal rim and insulating clay on discoloured exterior vertical wall, traces of sintered clay on the interior wall as well (2.5YR 4/0-10R 4/0). Rim diam. 22 cm, inner wall diam. 33 cm. As no. 1 (above), inv. no. BD 99/B3 (fig. 10.2).

Ref.: Mackensen 2004a, 141 fig. 7.2. 3. Fragment of horizontal rim with graffito in cursive (ante cocturam) [...] ...[...]. As no. 1 (above), inv. no. BD 97/B1 (fig. 10.3). Unpublished. 4. Fragment of horizontal rim with graffito in cursive (ante cocturam) [...] ...[...]. As no. 1 (above), inv. no. BD 98/B1 (fig. 10.4). Unpublished. 3. Possible kiln-props with stamped decoration

1. Fragment of cylindrical prop with narrowed midriff and identical stamped decoration on round ends. Diam. 4.1, h 3.3 cm.

Stamped decoration of 2 slightly dart-shaped palm-branches with short stem type Hayes 9 (h 2.2 cm), style B. Sidi Marzouk Tounsi, central Tunisia (in 2002) (fig. 11.1). Unpublished. 2. Cylindrical prop with stamped decoration around the body and on round ends. Diam. 3.7 cm, h 3.9 cm. Two rows of small dart-shaped palm branches with short stem (h 1.3 cm) type Hayes 6, style C, between them a horizontal line of 11 stamps of 3 concentric circles (diam. 0.7 cm) type Hayes 26, style B. On round ends,

identical decoration of outer circle of 3 concentric circles type Hayes 26, which surround rosette-like stamp (diam. 1.3 cm). Analysis sample no. 293, SMT (Table 2).

Coll. K.W. (fig. 11.2). Unpublished. 3. Cylindrical prop with identical stamped decoration on round ends. Diam. 3.6 cm, h 4.2 cm. In the centre, decorated square with diagonal cross and alternating dot and angle lines (1.0 x 1.0 cm) similar to type Hayes 96, style D; at the corners, diagonal lines of 3 stamped dots; between the lines, 2 concentric circles (diam. 0.7 cm) similar to type Hayes 22.

Analysis sample no. 291, SMT (Table 2).

Coll. K.W. (fig. 11.3). Unpublished. 4. Cylindrical prop with identical stamped decoration on round ends. Diam. 4.4 cm, h 4.3 cm. Cross-monogram with decoration of fine dots and small hooks at end of each arm, A (upside-down) under left

arm and w under right arm (height 2.9 cm) similar to type Hayes 289 A, style D. Analysis sample no. 292, SMT (Table 2). Coll. K.W. (fig. 11.4). Unpublished.

Technology and organisation of ARS ware production-centres

43

5. Cylindrical prop with identical stamped decoration on round ends. Diam. 3.7 cm, h 3.2 cm. Four-petalled rosette in double outline (diam. 1.2 cm); cantharus with line and dot-patterns (h 2.0 cm) similar

to type Hayes 271, style D; small animal to left with low legs, ribs, mane and broad tail, perhaps a badger(?). Analysis sample no. 289, SMT (Table 2).

Coll. V.Q. (fig. 11.5). Unpublished. 6. Cylindrical prop with stamped decoration on round ends. Diam. 2.7 cm, h 2.8 cm. 8-petalled rosette. Analysis sample no. 290, SMT (Table 2). Coll. V.Q. (fig. 11.6). Unpublished. Acknowledgements | am greatly indebted to the Institut National du Patrimoine, Tunis, and in particular to A. Ennabli (Musée de

Carthage), then Directeur des Monuments et Sites d'INP, for kind permission to survey the ARS productioncentres south of Tebourba in the context of ongoing work on Late Roman pottery from the German excavations at Carthage. I also owe sincere thanks to P. Zanker (DAI Rom) for funding a pottery campaign at Carthage in the autumn of 1999, during which the survey material from El Mahrine and Bordj el Djerbi was documented;

the sherds were drawn by M. Gschwind

(München).

Most of the other drawings and photo-

graphs were prepared by B. Kohler, the illustrations were digitalised by S. Peisker (both Universitat München). I wish to express my gratitude as well to the Gerda Henkel Stiftung (Düsseldorf) for funding an interdisciplinary archaeometric project on ARS ware and inscribed potters' tools. The chemical analyses were carried out in splendid cooperation by G. Schneider (Freie Universitat Berlin). [ would like warmly to thank R. Haensch (Kommission für Alte Geschichte und Epigraphik of the DAT) for advice and discussion and S. S. Frere and J. H. Humphrey for improving the English version. Bibliography Atlante | = A. Carandini et al., Atlante delle forme ceramiche |. Ceramica fine romana nel bacino mediterraneo (medio e tardo impero) (EAA, Roma 1981). Barraud, D., M. Bonifay, F. Dridi and J.-F. Pichonneau

1998. "L'industrie céramique de l'Antiquité tardive,"

in H. Ben Hassen and L. Maurin (edd.), Oudhna (Uthina). La redécouverte d'une ville antique de Tunisie (Ausonius Mémoires 2).

Ben Moussa, M. 2007. La production des sigillées africaines. Recherches d'histoire et d'archéologie en Tunisie septentrionale et centrale (Col. Instrumenta 23, Barcelona).

Berthier, A. 2000. Tiddis cité antique de Numidie (Mémoires Académie des Inscriptions et Belles-Lettres NS 20). Bonifay, M. 2003. "La céramique africaine, un indice du développement économique?," AnTar 11 (2003) 113-

28. Bonifay, M. 2004. Etudes sur la céramique romaine tardive d'Afrique (BAR $1301, Oxford) Brun, C. 2004. "Détermination d'origine par fluorescence X de quelques exemplaires de l'ensemble de céramiques du IVe s. p.C. découverts dans une citerne du capitole d'Uthina (Tunisie)," in H. Ben Hassen and L. Maurin (edd.), Oudhna (Uthina), colonie des vétérans de la XIIIe légion (Ausonius Mémoires

13) 236-44.

Cockle, H. 1981. "Pottery manufacture in Roman Egypt. A new papyrus," JRS 71, 87-97. Gauckler, P. 1897a. "Oudna," BAC 1897, 454-58. Gauckler P. 1897b. "Moules et instruments de potier," in F. du Coudray

La Blanchère et P. Gauckler, Cata-

logue du Musée Alaoui (Paris) 252-54. Guéry, R. 1968. "Notes de céramique," BAAlg 3 (1968) 272-81. Hayes, J. W. 1972. Late Roman pottery (London).

Hayes, J. Kajanto, Lassére, chute

W. 1980. A I. 1965. The J.-M. 1977. de Carthage

supplement to Late Roman pottery (London). Latin cognomina (Comm. Soc. Sci. Fennica 36.2). Ubique populus: peuplement et mouvements de population dans l'Afrique romaine de la à la fin de la dynastie des Séveres (146 a.C.-235 p.C.) (Etudes d'antiquités africaines).

Lassus, J. 1958. "Tiddis," Libyca 6 (1958) 251-64.

Mackensen, M. 1993. Die spütantiken Sigillata- und Lampentópfereien von El Mahrine (Nordtunesien). Studien zur nordafrikanischen Feinkeramik des 4. bis 7. Jahrhunderts (Münchner Beitráge zur Vor- und Frühgeschichte 50).

Mackensen, M. 1998a. "Centres of African red slip ware production in Tunisia from the late 5th to the 7th century," in L. Sagui (ed.), Ceramica in Italia: VI-VII secolo (Bibl. Arch. Med. 14) 23-39.

44

M. Mackensen

Mackensen, M. 1998b. “New evidence for Central Tunisian red slip ware with stamped decoration (ARS style D),” JRA 11 (1998) 355-70. Mackensen, M. 1998c. "Arbeitsgeráte aus dem spátantiken Tópfereizentrum von El Mahrine (Nordtunesien), RómMitt 105 (1998) 431-39. Mackensen, M. 2003a. “Production of 3rd century sigillata A/C (C1/2) or 'el-Aouja' ware and its transition to sigillata C? with appliqué decoration in central Tunisia," RCRFActa 38, 279-86. Mackensen, M. 2003b. "Datierung und Provenienz einer spátantiken figürlichen Punze für nordafrikanische Sigillata. Zur Spatphase der Feinkeramikproduktion in Sidi Marzouk Tounsi (Zentraltunesien)," BayVgbi 68 (2003) 101-8.

Mackensen, M. 2004a. "Produzione e diffusione della ceramica sigillata africana nella Tunisia centrale e settentrionale dalla metà del III secolo alla metà del V secolo d. C.," in M. de Vos (ed.), Archeologia del territorio. Metodi, materiali, prospettive. Medjerda e Adige: due territori a confronto (Labirinti 73) 131-60. Mackensen, M. 2004b. "Tonpatrizen und Vorlagen figürlicher Darstellungen auf spátantiken nordafrikanischen Sigillataplatten der Form Hayes 56," KéinJb 37, 791-804. Mackensen, M. 2005a. "Tonabformung eines reliefverzierten MetallgefaBes des 3. Jahrhunderts aus dem zentraltunesischen Tópfereizentrum Sidi Marzouk Tounsi," BayVgbl 70 (2005) 169-82.

Mackensen, M. 2005b. "Der ‘Triumph der Venus’ auf spátrómischer Sigillata (C^) aus der nordafrikanischen Provinz Byzacena," in B. Brandt, V. Gassner and S. Ladstatter (edd.), Synergia. Festschrift für F. Krinzinger IT (Wien) 311-20. Mackensen, M. 2006a. "Gipsmatrizen für Appliken zentraltunesischer Sigillata des 3. bis Mitte des 5. Jahrhunderts," Bay Vebl 71 (2006) 177-96. Mackensen, M. 2006b. "The study of 3rd-century African red slip ware based on the evidence from Tunisia," in D. Malfitana, J. Poblome and J. Lund (edd.), Old pottery in a new century. Innovating perspectives on Roman pottery studies. Atti del Convegno Int. di Studi, Catania, 22-24 aprile 2004 (Monografia dell'Istituto per i Beni Archeologici e Monumentali, C.N.R. 1) 105-24. Mackensen, M. and G. Schneider 2002. "Production centres of African red slip ware (3rd-7th c.) in northern and central Tunisia: archaeological provenance and reference groups based on chemical analysis," JRA 15 (2002) 121-58. Mackensen, M. and G. Schneider 2005. "Chemical analyses of red-slipped roof-tiles from Bir Ftouha,” in S. T. Stevens, A. V. Kalinowski and H. vander Leest, Bir Ftouha: a pilgrimage church complex at Carthage JRA

Suppl. 59) 371-78. Mackensen, M. and G. Schneider 2006. "Production centres of African red slip ware (2nd-3rd c.) in northern and central Tunisia: archaeological provenance and reference groups based on chemical analysis," JRA 19 (2006) 163-90. Maurin, L. and J. Peyras 1973. "Uzalitana, la région de l'Ansarine dans l'Antiquité," CahTun 19 (1971 [1973]) 11 ff. esp. 33 f. Mees, A. W. 2002. Organisationsformen rómischer Tópfer-Manufakturen am Beispiel von Arezzo und Rheinzabern unter Berücksichtigung von Papyri, Inschriften und Rechtsquellen (Monog RGZM 52.1-2). Pavolini, C. and S. Tortorella

1997, "Le officine di El Mahrine,

il libro di M. Mackensen e

lo stato attuale

della ricerca sui centri di produzione della ceramica africana," ArchCl 49 (1997) 247-74. Peacock, D. P. S. 1982. Pottery in the Roman world: an ethnoarchaeological approach (London). Peacock, D. P. S., F. Bejaoui and N. Ben Lazreg 1990. "Roman pottery production in central Tunisia," JRA 3 (1990) 59-84. Poinssot, L. and R. Lantier 1923. "El Mahrine- établissements agricoles et églises," BAC 1923, Ixxiv-Ixxviii. Salomonson, J. W. 1973. "Kunstgeschichtliche und ikonographische Untersuchungen zu einem Tonfragment der Sammlung Benaki in Athen," BABesch 48, 3-82. Salomonson,

J. W.

1976.

"Rómische

Keramik

aus Nordafrika,”

in P. La

Baume

and

J. W. Salomonson,

Rômische Kleinkunst Sammlung Karl Loffler (Wiss. Kat. Rom.-Germ. Mus. Koln III) 120-87. Salomonson, J. W. 1982. "Litterae africanae'. Ein Tonfragment mit kursiver lateinischer Inschrift in der archáologischen Sammlung der Utrechter Universitat,” in J. den Boeft and A. H. M. Kessels (edd.), Actus. Studies in honour of H. L. W. Nelson (Utrecht) 343-93. Spier, J. 2003. "A lost consular diptych of Anicius Auchenius Bassus (A.D. 408) on the mould for an ARS plaque," JRA 16 (2003) 349-54. Strobel, K. 1987. "Einige Bemerkungen zu den historisch-archáologischen Grundlagen einer Neuformulierung der Sigillatenchronologie für Germanien und Ratien und zu wirtschaftsgeschichtlichen Aspekten der rómischen Keramikindustrie," MBAH 6.2, 75-115, van den Hoek, A. 2006. "Peter, Paul and a consul: recent discoveries in African Red Slip Ware," ZAntChrist 9

(2006)

197-246.

The forming and slipping of African Sigillata: evidence from the Palatine East assemblage J. Theodore Pefia Although several recent studies — including the remarkable book by M. Bonifay, the publication of which served as the stimulus for the organization of the present volume — have added a great deal to our knowledge of workshop sites at which African Sigillata (African red slip ware) was produced, there remain substantial gaps in our understanding of the techniques employed for the manufacture of this family of high-quality tablewares. The present contribution is intended to help mitigate this situation by documenting some of the techniques employed for the forming and slipping of African Sigillata. It draws on observations made by the author during his analysis of the African Sigillata from the Palatine East excavations in the

center of Rome.! Previous discussions of the forming and slipping of African Sigillata

Before I consider the evidence provided by the materials in the Palatine East assemblage, it will be useful to summarize the views previously expressed regarding the techniques employed for the forming and slipping of the various classes of African Sigillata. In his seminal work Late Roman pottery (1972), J. W. Hayes put forward the view that a large portion of vessels in

African Sigillata A were produced by being thrown free-hand and then turned on their exterior surface to fashion the base.?By contrast, he believed that effectively all of the vessels belonging to African belonging to African mounted on a potter's have been employed

Sigillata C and Sigillata D and wheel (properly for the forming

C/E, and also probably a substantial portion of those E, were manufactured by being thrown inside a mould known as a jolly) in a procedure similar to that known to of relief-decorated Italian and Gallic Sigillata.?

Hayes made the following observations in support of the proposition that African Sigillata C and C/E were manufactured by being thrown inside a wheel-mounted mould: 1. The manufacture of certain forms, specifically Hayes Forms 48-50, could not have involved the turning of their bases and lower walls, as these forms' broad floors could not have been subjected

to the forces

involved in this procedure without collapsing. 2. Thethin walls of uniform thickness of these same forms would have been difficult to produce by turning. 3. The fact that these forms have either a ring foot consisting of a tiny ridge or a shallow, inset base is compatible with their manufacture in a wheel-mounted mould, as this feature may have been designed to facilitate the removal of the vessel from the wheel-mounted mould.

4. The small facets present on the exterior surface of these vessels in the area of the lower wall may have been produced by facets on the interior surface of a wheel-mounted mould.* 5. The scratches present on the exterior surface of these forms may have been produced "by small particles adhering to the jolly as it rotated" (294). 6. The contrast between the smooth surface of the lower portion of the exterior of these forms and the distinctly rougher surface in the area underneath the rim is compatible with their manufacture in a wheel-

mounted mould, with the smooth area representing the area where the vessel had been in contact with the wheel-mounted mould, and the rough area the area where the vessel had not been in contact with it.

7. The fact that these forms are slipped either on the interior surface only or on the interior surface and just the upper portion of the exterior surface is compatible with their manufacture in a wheel-mounted mould, as this may be the result of the practice of applying the slip by pouring it into the vessel while it was still

1

For the Palatine East excavations, see Hostetter et al. 1994. For the Palatine East pottery assemblage

and the methods employed for its study, see Ikáheimo 2003; Pena 1999 and 2007. 2

Hayes 1972, 292. Ibid. 293-95. For a more recent affirmation of this view, see id. 1997, 59-60.

4

Hayes’ views regarding the origin of these facets appear to have been inconsistent: in his general discussion of the appearance of African Sigillata C and his discussions of 4 of the most common forms in this class (his Forms 48-50 and 52), he refers to them as "turning marks" (Hayes 1972, 65, 67, 72, 76 and

290).

46

J. T. Pena in the mould, with the slip covering only the interior surface and — in cases where the vessel had shrunk in the rim area as it dried and pulled away somewhat from the mould — the upper part of the exterior surface.

Hayes cited the following three points as evidence

that African Sigillata D and E were

manufactured by being thrown inside a mould (jolly): 1. Certain forms, specifically Hayes Forms 58-61, have either a ring foot consisting of a tiny ridge or a shallow, inset base that might have been designed to facilitate the removal of the vessel from the mould (jolly). 2. Certain forms, specifically Hayes Forms 67 and 68, have a “clean-cut” (angular?) exterior that is suggestive of manufacture in a mould (jolly). 3.

Although these vessels have a fairly coarse fabric, their exterior surfaces are smooth, perhaps the result of

these having been pressed against a smooth surface, such as that of a mould (jolly).

Examples of African Sigillata D and E frequently have small facets and scratches on their exterior surface similar to those that appear on examples of African Sigillata C and C/E, and are also generally slipped only on the upper portion of their exterior surface; Hayes may also have had these characteristics in mind when he suggested that these two classes were manufactured by being thrown inside a mould (jolly). It should be pointed out that Hayes acknowledged that a significant weakness with his argument regarding the use of wheel-mounted jollies for the manufacture of these classes of African Sigillata was the fact that no jollies in ceramic or plaster (the materials normally employed for the manufacture of these items) had been reported from any of their known

production sites.® In the 35 years since the publication of Hayes’ book there has been no systematic and comprehensive effort to revisit the question of the methods employed for the forming and slipping of African Sigillata. For example, L. Sagui, A. Carandini, S. Tortorella, and E. Tortorici, in

their highly influential treatments of African Sigillata C, C/E, D, and E in Atlante 1 (1981), were content simply to note Hayes' view that these classes were manufactured by being thrown

inside a mould (jolly), offering no further discussion of the methods employed to form and slip the vessels belonging to them.? J. Schuring, in her study of the African Sigillata recovered in the excavations at San Sisto Vecchio in Rome (1988), concluded that in at least some cases vessels belonging to African Sigil-

lata A, A/D, C, and D were formed by being thrown inside a mould (jolly).? This view was founded largely on the opinion of a professional potter, L. Jakobs, whom Schuring consulted on the question. Jacobs, having examined both the sherds at Schuring's disposal and profile drawings of complete examples of the various forms represented, concluded that the vessels' generally thin walls and the 'somewhat carinated' (angular?) profiles of several forms were compatible with throwing inside a mould (jolly). The specific forms that Jacobs flagged as having a profile suggesting that they had been thrown inside a mould were: Hayes Forms 3, 6, and 9 in Sigillata A, Hayes Forms 27, 31, and 32 in Sigillata A and/or A/D, Hayes Forms 45A and 50 in Sigillata C, and Hayes

Forms 58, 59, 61, and 91 in Sigillata D. The set of materials from San

Sisto Vecchio also contained three sherds displaying a seam in the wall that Schuring took to be evidence that the vessels to which they belonged had been formed by rolling out multiple sheets of paste and then placing these inside a mould (jolly). These included a sherd preserving the rim and upper/middle wall from a Hayes Form 31 in Sigillata A/D, a sherd preserving the rim and upper/middle wall from a Lamboglia Form 40bis in Sigillata C, and a sherd preserving a portion of the floor of an unidentified form in Sigillata D. Interestingly, Schuring makes no

reference to surface attributes as evidence for the techniques employed for the forming of African Sigillata, except to note that fragments of examples of the Lamboglia 40bis in Sigillata C bore marks indicating that the vessels to which they belonged had been turned?

5 6 7 8 9

Ibid. 295. Ibid. 293. Sagui 1981, 58; Carandini and Tortorici and Tortorella 1981c, 119. Schuring 1988, 34-35. Ibid. 34 n.40.

1981a, 117; Carandini

and Tortorella

1981b,

78; Carandini

The forming and slipping of African sigillata

47

Y

0

10cm

|

|

Fig. 1. Ceramic items from El Mahrine workshop site that probably served as forming tools. l. Fragment of ring that probably served as chuck. Upper diam. 18.0; lower diam. 24.0; H. 13.9 (Mackensen 1993, 70 Abb. 14.1); 2. Oval plate (pugillum) that probably served as rib. L. c.12; w. c.8; max. th. c.2 (Mackensen 1993, 72 Abb. 15.6); 3. Reconstruction of mushroom-shaped implement that probably served as smoothing tool. Diam. 14.0; reconstructed h. c.7.6 (Mackensen 1993, 72 Abb. 15.4).

M. Mackensen, in his 1993 study of the materials (including Sigillata D, North African Cookware and lamps) recovered in surface investigations at the pottery workshop at El Mahrine, considered various aspects of the techniques employed for the forming of the vessels manufactured at this establishment and of African Sigillata more generally.!? On the basis of an appliqué bearing a depiction of a potter working at a wheel occurring on an example of Sigillata C, he argued that African Sigillata was thrown on a single wheel that the potter propelled by means of a stick.!! Mackensen concluded that the surface attributes (including grooves, bands, ridges, and facets) of the examples of Sigillata D from El Mahrine indicated

that these vessels had been formed by being thrown freehand and then turned." On the basis of this evidence, he explicitly rejected Hayes’ view that a sizeable portion of African Sigillata

was formed by being thrown inside a mould (jolly).? As part of his discussion, Mackensen evaluated various ceramic items recovered at El Mah-

rine and other African Sigillata workshop sites that might have served as forming tools.!! These included the following: *

a fragment of a large ceramic ring formed by throwing on a wheel from El Mahrine that he identified as a chuck — that is, a device that served to attach a vessel blank to a wheel for turning (see below) (fig. 1.1);

*

various small, oval plates with two concave sides or one concave and one convex side from El Mahrine

10 11 12 13

Mackensen 1993, 63-87 and 174-75. Ibid. 64-67. Ibid. 69 and 174-75. Ibid. 174-75.

14

See also Mackensen 1998, which documents additional forming tools recovered during surface collection carried out at El Mahrine subsequent to his completion of the manuscript for his book of 1993. Id. 1993, 69, 70 fig. 14.1, 509 A.1. Mackensen uses the term Abdrehhilfe (‘turning accessory’) rather than Donsel (chuck) to refer to this item.

15

48

J. T. Peña and other workshop sites (explicitly termed a pugillum in prae cocturam graffiti inscribed in some examples) that he identified as a rib employed for the turning of vessel blanks (see below) and perhaps

also for their initial throwing (fig. 1.2);16 *

various fragmentary items (some hollow, some solid) from El Mahrine and other workshop sites, termed

‘mushrooms’, that consist of a broad, lentoid head with wear on its lower surface and a short, upright, cylindrical handle attached to the center of its upper surface that he identified as probable polishing or

smoothing tools (fig. 1.3); and *

afragmentary ceramic item probably from a workshop site somewhere southwest of Kairouan that he identified as the handle of a needle tool or similar implement.!?

In the volume that constitutes the focus of this collection of papers Bonifay has little to say regarding the methods employed for the forming and slipping of African Sigillata.!? Although he briefly notes the views of Hayes and Schuring (see above) regarding the use of jollies, he makes no mention of Mackensen's opposing view. Bonifay does state that some African Sigillata vessels bear surface marks indicating that they were formed in a mould rather than thrown freehand. By way of example, he illustrates a fragment of an example of Atlante Form 17.17 in Sigillata A on which the exterior surface of the wall is characterized by the presence of alternating raised and depressed areas radiating outward from the foot that he believes

indicate that the vessel to which it belonged was manufactured in a mould.” He also suggests that a small number of large, open vessels recovered at the Sigillata D workshop at Oudhna, one of which he illustrates in a profile drawing, might have functioned either as a throwing template or bat (see below), or in some undefined way as a throwing accessory?! The analysis of the African Sigillata from the Palatine East The pottery assemblage from the Palatine East excavations contains a large amount of African Sigillata, including a substantial quantity of Sigillata A, C, and D, and more modest

amounts of Sigillata A/D, C/E, and E (Table 1).? Although this set of materials includes fragments of a large number of vessels belonging to a wide array of forms, it is less than ideal for efforts to infer forming and slipping techniques on account of the high degree of brokenness and low degree of vessel completeness that characterize most contexts. In addition, as the Roman-period sequence ends around the middle of the 5th c., the forms manufactured in Sigillata D from this point onwards (e.g., Hayes Forms 99 to 109) are represented by only a very small TABLE 1 AFRICAN SIGILLATA IN THE PALATINE EAST ASSEMBLAGE Class

Weight of sherds (gm)

Afr. Afr. Afr. Afr. Afr.

Sig. Sig. Sig. Sig. Sig.

A A/D C C/E D

(rims)

(bases)

7,708 308 14,216 2,660 48,126

329 12 582 86 785

161 5 315 44 465

275

10

0

564 73,857

9 1,803

5 995

Afr. Sig. E

Other TOTAL

16

| Number of vessels | Number of vessels

Id. 1993, 74-83. For additional examples, see id. 1998, 437-38, 438 fig. 5; Bonifay 2004, 59, 60 fig. 29.1;

Mackensen in this volume. 17

Mackensen 1993, 69-74, For additional examples see id. and Schneider 2002, 146 fig. 19.3, 148 no. 1;

Bonifay 2004, 59-60, 60 fig. 29.2-3; Mackensen and Schneider 2006, 188 no. 1, fig. 13.1. 18 19

Mackensen 1993, 74, 75 fig. 17.3. Bonifay 2004, 59-60.

20 21 22

Ibid. 59 fig. 28b. Ibid. 59 fig. 28a. Forthe final publication of the Palatine East pottery assemblage, the African Sigillata will be presented according to the various fabrics that can be distinguished within these general groupings.

The forming and slipping of African sigillata

49

number of residual sherds from mediaeval contexts. The number of examples of these forms is too small to permit any comprehensive and systematic treatment of the techniques employed for their manufacture, and they are not taken into consideration here. The evaluation of the techniques employed for the forming and slipping of the vessels in question drew on three different kinds of evidence: 1. Information regarding the methods employed by contemporary craft potters to manufacture vessels generally similar in form and dimensions to those attested in African Sigillata. 2. The attributes of the surfaces of sherds; and

3. Seams visible in the breaks at the edge of sherds. Contemporary craft potters usually form vessels similar to those attested in African Sigillata by first throwing a vessel similar to the desired vessel in size and shape, though with somewhat

thicker walls in the area of the rim and/or base (here referred

to as a blank), and

then finishing this by the operation known as turning. The blank may be thrown either directly atop the wheelhead or atop a removable disk (known as a bat) attached to the wheelhead. In order to form a medium or large vessel, a lump of hydrated paste roughly equal to the amount required to form the blank is centered on the wheelhead and the blank thrown from this. Small vessels are frequently formed by a somewhat different technique known as throwing from the hump. In this technique, a lump of plastic paste equal to the amount required to form several blanks is centered on the wheelhead, pulled up into a shape of a tall cone, and a succession of blanks thrown from the top of this. After forming, the blank is removed

from the wheel either

by slicing it free with a cutting implement (consisting of a length of wire held between two toggles) in cases where it was formed directly atop the wheelhead or thrown from the hump, or simply by detaching the bat in cases where it was formed atop one of these devices. The use of a bat is particularly advantageous for the forming of large and/or thin-walled vessels, as this permits the freshly thrown blank to be removed from the wheel with little risk of distortion or collapse.

Once removed from the wheel, the blank is allowed to dry partially and then remounted on the same or a different wheel for turning. In order to turn the partially dried blank, the potter mounts it on the wheelhead either inverted or right-side-up in centered position, sets the wheel in motion, and employs a tool with a more or less sharp cutting edge (usually a rib, a bladed tool, or a loop tool) to slice away (turn off) slivers of paste, thinning and shaping the rim, wall, floor, and/or base to obtain the desired form.?? Potters

typically collect turned-off

slivers of paste, returning these to the slop bucket for recycling.** For turning, it is essential that the blank be sufficiently but not too dry. In general, a blank is dry enough for turning if the potter can hold it by its rim in inverted position without causing the rim and wall to distort and/or can press on its bottom without causing this to flex. If the blank is allowed to become too dry, the cutting implement will dig into the paste rather than slicing through it, leaving a surface disfigured by linear or transverse gouges. A blank can be mounted on a wheel for turning by either of two methods. In the first of these, it is anchored directly to the wheelhead by means of a thin sheet, a coil, or multiple pads of plastic clay. In the second method, it is set either over the top of, or inside of, a holding device (known as a chuck) mounted atop the wheelhead. Chucks, which come in two varieties (ring-

shaped and cone-shaped) may consist either of plastic clay or a static material, such as ceramic. Although chucks made of plastic clay must be kept moist, they are advantageous in that they can be modified in accordance with the specific form and dimensions of the blank and are

less likely to mar the blank's surface than are chucks made of a static material. 23

For turning the

For a general discussion of turning, see Clark 1999, 60-62. For detailed explanations (illustrated by step-by-step photographs) of the methods employed for turning various vessel forms similar to those attested in African Sigillata, see Barbaformosa

1999, 46-53, 71-74, 80-81, 89-93, 95-97, 101-5 and 128-

30. For the wide variety of tools that contemporary potters employ for turning, see Whitford and Wong 1986, 66-81; Czysz 1982, 319 Abh. 28. 24

Barbaformosa

1999, 56-57.

Fig. 2. Photographs showing potter engaged in turning blanks with the aid of chuck in plastic clay. l. Potter employing loop tool to turn exterior surface of cup blank set over cone-shaped chuck (Barbaformosa 1999, 72 fig. 12); 2. Potter setting plate blank over ring-shaped chuck for turning of exterior surface (ibid. 101 fig. 9); 3. Potter using rib to turn interior surface of plate blank set inside ring-shaped chuck (ibid. 104 fig. 14).

Fig. 3. Detail of surface of vessel in leather-hard state thrown and turned by the author. Upper area, which is unturned, displays fine striations produced by smoothing with sponge. Lower area, which was turned with loop tool, displays compacted surface with facets, drag marks, and pitting.

exterior of small, open forms, such as cups or bowls, a tall cone-shaped chuck is employed, with the blank placed over the top of this in inverted position (fig. 2.1). For turning the exterior of large open forms, such as dishes or plates, or the interior of both large and small open forms, a substantially wider, ring-shaped chuck is employed, with the vessel blank either placed over the top of this in inverted position for turning of the exterior (fig. 2.2) or set inside it in rightside-up position for turning of the interior (fig. 2.3). For closed forms, such as jugs, the blank can be set inside a

tall, ring-shaped chuck in inverted position for turning the base and lower wall,

and in right-side-up position for turning the neck and shoulder. Chucks are advantageous in that they allow vessel blanks to be mounted on the wheel quickly and easily (permitting, for example, some forms being turned with the aid of a ring-shaped chuck to be switched from inverted

to right-side-up

position,

or vice versa, with

a minimum

of time

and

effort)

and

provide support for the vessel blank during the turning procedure — which helps to prevent its distortion or breaking. The first of these two considerations is particularly important in cases where a potter wishes to turn several vessels of more or less identical shape and dimensions, as

the same chuck can be used over and over again with, in the case of chucks made of plastic

The forming and slipping of African sigillata

51

clay, little or no modification. The turning of a vessel generally imparts one or more distinctive attributes to its surface, and the presence of any of these on the surface of an ancient sherd may indicate that the vessel to which it belonged was turned (fig. 3). First, the pressure that the cutting implement exerts against the paste compresses it and aligns its clay mineral component, producing a surface that is generally more smooth and compact than an unturned surface. Second, the edge of the cutting implement may snag bits of aplastic material present in the paste, either pulling these along the surface to produce horizontal striations (often with an aplastic grain embedded at one end) known as "drag marks”, or plucking these out of the paste altogether, leaving a more or less densely pitted surface. Third, since potters engaged in turning produce a curve in the vessel profile by progressively altering the angle at which they hold the cutting implement as they move it up or down over the surface of the blank, this operation produces a surface that is marked by a series of distinct horizontal facets, the width of which is determined by the frequency with which the potter changed the angle of the implement. Although in some cases these facets are later smoothed or scraped away with the aid of a tool such as a rib, they are sometimes allowed to remain. Scraping operations of this kind may also produce drag marks and/or pitting of the surface. Fourth, the mounting of a blank on a chuck may produce a horizontal band of compressed paste on its surface, corresponding to the area where the blank was in contact with the chuck. As turning involves cutting into the blank with a more or less sharp implement, this technique permits potters to produce distinctly sharp angles in a finished profile, and the manufacture of vessel forms with a profile characterized by sharp angles may well (though does not necessarily) involve turning. Vessels made in a mould generally display a smooth surface free of attributes of the kind just described. The examination of the breaks at the edge of sherds sometimes reveals seams in the vessel wall. In the case of vessels thrown on a wheel, these represent either points where the wall was folded over onto itself during the forming of the vessel blank (e.g., to produce a thickening in the rim area, a rim flange, and so on) or points where a separately formed piece was added to the vessel (e.g., the addition of the rim/neck to the body of an amphora). In the case of vessels formed in a mould, these represent the juncture between two separate sheets of paste that were pressed into or thrown inside the mould. The evaluation of the surfaces and breaks of the African Sigillata from the Palatine East revealed that, with but a single exception, the manufacture of the forms for which it was possible to infer the set of operations employed for their forming involved a significant amount of turning on their exterior surface and, in many cases, on both the interior and exterior surfaces.

While it is theoretically tured by being thrown followed by the turning have sacrificed some of

possible that the blanks for some or all of these inside a mould (jolly), the practice of throwing of one or both surfaces would have made little the principal advantages to be obtained from the

forms blanks sense, use of

were manufacinside a mould in that it would wheel-mounted

moulds” while retaining their not insubstantial disadvantages.2 It thus seems all but certain that the blanks for these forms were thrown freehand. The one exception is a rectangular dish/platter in African Sigillata D that was formed by pressing sheets of paste into a free (rather than wheel-mounted)

mould.

By considering one representative example of each of the three major classes of Sigillata represented (A, C, and D), it is possible to illustrate the basic operations employed for the forming and slipping of the bulk of the African Sigillata vessels present at the Palatine East. The first to be considered is an example of Hayes Form 9B, a bowl in Sigillata A that was produced over the period c.150/160 to 210/220.27 This form, which ranges between 13 and 17 cm in diameter, has an up25 26

Eg. the ability to manufacture forms with extremely thin walls of uniform thickness; the ability to manufacture vessels with a complex form by means of a substantially automatic process. Eg.theobligation to manufacture and maintain a large number of wheel-mounted moulds; the limiting of

the number of examples of a specific form that could be manufactured at any one time to the number of moulds (jollies) available for that form; the relatively slow drying-time associated with blanks thrown inside a mould (jolly). 27

For his form, see Hayes 1972, 35-37; Carandini and Tortorella 1981a, 27.

52

10 cm

Fig. 4. Profile drawings of 4 examples of African Sigillata from Palatine East described in text. 1. Hayes Form 9B (Palatine East no. 6978) in Sigillata Al (NE Tunisia); 2. Hayes Form 52B (Palatine East no. PE 191) in Sigillata C3 (Central Tunisia) or D1 (El Mahrine); 3. Hayes Form 61 A/B (Palatine East no. PE 6694) in Sigillata DI (NE Tunisia, El Mahrine);

4. Rectangular dish/platter with decoration of beads (nos. PE 2786/2787) probably in Sigillata D1 Tunisia, El Mahrine). Dashed lines represent seams between sheets of paste.

(NE

Fig. 5. Exterior surface of lower wall/base of Palatine East no. PE 6978 (Hayes Form 9B). curved wall with a rounded rim, a sloping floor, and a ring foot. There are always two decorative grooves on the exterior surface of the wall a short distance below the rim, and the vessel is usually slipped on all surfaces. This example, Palatine East Accession Number 6978, consists of three joining sherds that preserve the entire vessel profile (rim diam. 5, foot diam. 5; H. 3.8) (fig. 4.1). All preserved surfaces of the vessel are covered with a glossy, light red slip. The entire interior surface and the exterior surface of the wall are pimply (fig. 5). The foot has a pimply outer face with a curved transition to the lower wall, a smooth resting surface with drag marks and pits, and a smooth inner face with an angular juncture with the area that it encloses. The area enclosed by the foot is smooth, with pronounced gouges, drag marks, and pits.

The forming and slipping of African sigillata

53

Fig. 6. Upper surface of rim flange/interior surface of wall of Palatine East no. PE 191 (Hayes Form 52B). The sequence of operations involved in the forming and slipping of this vessel can be reconstructed thus: 1. A blank with a disk-shaped base was thrown (possibly from the hump) in upright position on a potter’s wheel. 2. The blank was removed from the wheel and allowed to dry partially. 3. The partially-dried blank was mounted on a wheel in inverted position (probably by being set over a cone-shaped chuck). 4. Asharp implement was employed to turn off the bottom of the disk base and to turn out its interior to form the ring foot. 5. À pointed implement was employed to cut the two decorative grooves on the exterior surface of the upper wall. 6. The vessel was removed from the wheel. 7. The vessel was slipped on all surfaces (probably by being dipped into a large container of slip). It cannot be excluded that the cutting of the decorative grooves on the exterior surface of the wall (Operation 5) preceded the turning of the base (Operation 4), or, indeed, that these were cut while the freshly thrown blank was still on the wheel (thus following Operation 1). The second vessel is an example of Hayes Form 52B, a bowl in Sigillata C that was produced over the period c.300 to 400.78 This form, which ranges between 11 and 23 cm in diameter, has a broad, flanged rim, a curved wall, and a tiny ring foot. There is a single decorative groove or a pair of grooves on the upper surface of the rim flange near its outer edge and a single decorative groove or a pair of grooves in the floor. There are from 3 to 6 (most often 4) appliqués equally spaced around the upper surface of the rim flange. The vessel is slipped on its interior surface, including the upper surface of the rim flange, and sometimes also on the upper

portion of its exterior surface, including the underside of the rim flange and the upper wall. This example, Palatine East Accession Number 191, consists of 6 joining fragments that preserve the entire vessel profile (rim diam. 15-16, foot diam. 4.5; H. 4.9) (fig. 4.2). The upper surface of the rim flange and interior surface of the wall and floor are covered with a thick, glossy, red slip. The vessel has a pair of decorative grooves on the upper surface of the rim flange and another pair of decorative grooves in its floor. The edge of one appliqué of unclear subject is preserved on the upper surface of the rim flange. The upper surface of the rim flange is smooth, while the interior surface of the wall and the floor are covered with tiny, closely-spaced, horizontal

28

For this form, see Hayes 1972, 76-78; Carandini and Tortorici 1981b, 162-63.

54

J. T. Pena

Fig. 7. Underside of rim flange/exterior surface of wall/foot of Palatine East no. PE 191 (Hayes Form 52B). ridges (fig. 6). There is a small horizontal band on the interior surface of the wall c.1.5 cm below the rim in which these striations have been pressed flat, leaving a somewhat irregular surface. The exterior surface of the vessel is divided into 4 distinct zones: the underside of the rim flange, the upper/middle wall, the lower

wall, and the base (fig. 7). In the first of these, the surface is covered with small pimples and bears widelyspaced, tiny, concentric ridges. The outer edge of the rim flange is covered with thin, matte slip, and there is one irregular drip of thin, matte slip that runs along the middle of the flange’s underside. In the second, the gently curved surface is very smooth. The boundary between this zone and the first consists of a sharply defined edge, above which the surface of the first zone drops away ever so slightly. In the third, the somewhat

more sharply curved wall has a very smooth surface that is covered with small horizontal facets. The boundary between this zone and the second is indistinct, with the uppermost facets of the third zone blending into the continuous surface of the second zone in some areas. In the fourth, the foot is covered with tiny facets and grooves, while the area enclosed by it has a very smooth surface. The sequence of operations involved in the forming and slipping of this vessel can be reconstructed as

follows: 1. A blank was thrown (possibly from the hump) in upright position on a potter’s wheel. 2. The blank was removed from the wheel and allowed to dry partially. 3. The partially-dried blank was mounted on a wheel in inverted position by being set over a cone-shaped chuck. 4. A sharp implement was employed to turn the exterior surface, cutting out the tiny ring foot and thinning and shaping the lower wall and also probably the middle and upper wall. 5. A hard implement was employed to scrape smooth the surface of the middle and upper wall and the area enclosed by the ring foot. 6.

The partially finished blank was removed from the wheel.

7. The partially finished blank was mounted on a wheel in right-side-up position (probably by being set inside a ring-shaped chuck). 8. A hard implement was employed to smooth the upper surface of the rim flange. 9. A pointed implement was employed to cut the decorative grooves in the upper surface of the rim flange and in the floor. 10. At least one, probably 4, and perhaps as many as 6 mould-made appliqués were attached to the upper surface of the rim flange by means of a slurry.

The forming and slipping of African sigillata

55

Fig. 8. Exterior surface of wall of Palatine East no. 6694 (Hayes Form 61 A/B) (composite view). 11. 12.

The vessel was removed from the wheel. The interior of the vessel was slipped (probably by pouring slip inside the vessel, tipping and turning it to coat the entire interior surface and the upper surface/outer face of the rim flange, and then pouring off the excess).

It cannot be excluded that the smoothing of the upper surface of the rim flange (step no. 8) and the cutting of the decorative grooves in it and in the floor (step no. 9) were carried out while the newly-formed blank was still on the wheel, nor must the vessel have been mounted on the wheel when the appliqués were attached (step no. 10). If one assumes both of these alternative possibilities, then there is no need to assume that the vessel was mounted on the wheel in right-side-up position (step no. 7) for the execution of these three operations following the turning of its exterior. Although the absence of any indications on the exterior surface that it was mounted on a wheel for turning in right-side-up position may be taken as supporting this interpretation, it should be noted that exceedingly few examples of African Sigillata from the Palatine East assemblage display surface indications of being mounted on the wheel for turning in either right-side-up or inverted position. The third is an example of Hayes Form 61A/B (or 61 transitional), a dish in Sigillata D that was produced c.400 to 425.7? This form, which ranges between 22 and 41 cm in diameter, has a low, upcurved or slightly incurved wall, a broad floor, and a shallow inset base. The rim has a flat outer face that is sharply defined at its lower edge, resulting in a distinctive triangular profile. There is usually a single decorative groove or pair of grooves at the edge of the floor and a group of 2-5 decorative grooves closer to the center of the floor that enclose a complex composition of stamped decoration. The interior of the vessel and usually some portion or all of the exterior surface of the wall are covered with slip. This example, Palatine East Accession Number 6694, consists of two joining fragments that preserve the entire vessel profile except for the central portion of the floor /base (rim diam. 33, base diam. 20; H. 4.3) (fig. 4.3). The interior surface, the outer face of the rim, and the upper half of the exterior surface of the wall are

covered with a matte, red slip. The juncture of the floor and interior surface of the wall is marked by a distinct edge that is situated directly above the step that defines the outer edge of the base. Beyond this edge, the interior surface of the wall drops away from the floor in a curve that is effectively parallel to that of the

exterior surface of the wall. There is a single decorative groove in the lower wall immediately outside the edge of the floor. The flooris flat. Its central portion presumably bore a composition

29

For this form, see Hayes 1972, 105; Mackensen

1993, 321; Bonifay 2004, 167.

of stamped

decoration

Fig. 10. Potter at Shadiwal, Gujrat (Pakistan) employing mushroom-shaped ceramic implement to smooth base of dish (Rye and Evans 1976, 240 pl. 39.b). enclosed by multiple grooves. The interior surface has three distinct zones: the wall, a horizontal band on the wall c.0.5 cm below the rim, and the floor (fig. 9). The surface of the wall is smooth save for the band just

referred to, which is slightly rough. The surface of the floor is slightly, though noticeably, smoother than that of the interior surface of the wall; the area displaying this distinctive smoothing extends outward beyond the edge of the floor to a point just beyond the groove in the lower wall, somewhat flattening the slight burr raised by the cutting of this element. The exterior surface has 4 distinct zones: the outer face of the rim, the rim/wall juncture, the upper/middle wall, and the lower wall/base (fig. 8). The outer face of the rim is

smooth. The rim/wall juncture is covered with tiny facets. The surface of the upper/middle wall is slightly

pimply, with medium facets. The surface of the lower wall/base is smooth and continuous, with occasional pits that are concentrated in the area immediately inside and outside of the step that defines the outer edge of the base.

The forming and slipping of African sigillata

57

The sequence of operations involved in the forming and slipping of this vessel can be reconstructed as follows: 1. A blank was thrown in an upright position on a potter's wheel (possibly atop a bat). 2. The blank was removed from the wheel and allowed to dry partially. 3. The partially dried blank was mounted on a wheel in inverted position (probably by being set over a ringshaped chuck). 4. A sharp implement was employed to turn the exterior surface. This involved cutting into the wall to form the lower edge of the rim, thinning and shaping the upper/middle and presumably also the lower wall, and presumably also thinning and cutting out the inset base. 5. A hard implement was employed to scrape smooth the surface of the lower wall and the base. 6. The partially finished blank was removed from the wheel. 7. The partially finished blank was mounted on a wheel in right-side-up position (probably by being set

inside a ring-shaped chuck). 8. A hard implement was employed to smooth the outer face of the rim and to shape and smooth the interior

surface of the wall. 9. A pointed implement was employed to cut the decorative juncture with the floor. 10. Ahard implement was employed to smooth and perhaps [11. A pointed implement was employed to cut two or more the floor.] (12. Stamps were employed to impress a complex decorative

groove in the interior surface of the wall near the compress the floor. decorative grooves enclosing the central area of scheme in the central area of the floor.]

13.

The vessel was removed from the wheel.

14.

The interior of the vessel and the upper portion of its exterior were slipped (probably by dipping the vessel into a large container of slip).

Operations 11 and 12 have been placed in brackets, as it is not certain that they were carried out for the vessel here being described. The pugilla from El Mahrine and other Sigillata workshop sites are roughly the right size and shape to have served for the shaping and smoothing of the interior surface of the vessel's wall (step no. 8), in a procedure presumably similar to that depicted in fig. 2.3. The diameter of the head of the mushroom-shaped implements from El Mahrine and other Sigillata workshop sites, the slightly convex shape of the lower surface of this element, and the presence of wear on this surface all seem compatible with the use of these objects for the smoothing or smoothing /compression to which the vessel's floor was subjected (step no. 10). Perhaps worth noting in this connection is the fact that traditional potters in some parts of Pakistan employ ceramic implements of generally similar size and configuration to smooth the walls and floors of ves-

sels? (fig. 10). This operation may have been undertaken with a view to rendering this part of the vessel more resistant to the warping to which the floors of broad-based vessels would have been susceptible during the final drying and firing phases. The distinctive morphology of the interior surface of the lower wall and floor

of this vessel (i.e., the presence of a distinct outer edge of the floor, beyond which the interior surface of the wall drops away in a curve matching that of the wall's exterior surface) occurs in many other examples of

Hayes 61 and in many examples of other broad-based forms in Sigillata D (e.g., Hayes Forms 58, 59, 62), and it may well be the result of the joint use of the pugillum and the mushroom-shaped implement to perform these two operations. The practice of cutting a groove in the floor or lower wall aligned more or less directly above the step defining the outer edge of the base (step no. 9) is attested with many examples of this and other broad-based forms in Sigillata D, and it may well have had some connection with these two operations. For

example, this feature might have served to indicate the area within which the potter was free to smooth/ compress the floor without damaging the step.?! As previously noted, the Palatine East assemblage contains fragments of one example of a rectangular dish /platter in Sigillata D that was formed in a free mould. It may prove useful to include a comparable description of this vessel. The form in question, not previously recorded, has a broad,

flanged

rim

decorated

30

Rye and Evans 1976, 33, 60, 222 pl. 21.g, 240 pl. 39.b, and 281 pl. 80.

31

It seems possible that the aligning of the edge of the floor with the step that defined the edge of the base was undertaken with a view to producing sets of vessels that could be set one atop the other, with the step of the upper vessel seated over the outer edge of the floor in the lower vessel, so as to yield a stable, compact arrangement that would have facilitated the stacking of unfired vessels for firing and/or of finished vessels for transport. (Efforts to reproduce such an arrangement using both profile drawings and actual sherds from the Palatine East did not, however, yield satisfactory results.)

58

J. I. Pena

"

ARIAS

a m

Fig. 12. Exterior surface of same.

The forming and slipping of African sigillata with a swag of beads in relief on its upper surface, a low, steep wall, and a

59

flat base.?? This example, Pala-

tine East Accession Number 2786/2787, consists of 4 sherds, three of which join (fig. 4.4). When complete, the vessel to which they belong stood 3.5 cm high, with one of its 4 sides measuring at least 20 cm long. The interior and exterior surfaces of the sherds are covered with a glossy light red/red slip that has flaked away from nearly the whole of the exterior surface except the area immediately adjacent to the rim (figs. 11-12). There is a slightly recessed band along the inner edge of the rim flange, and a large, slightly recessed rectangular panel at the center of the floor. The outer face of the rim flange is flat and somewhat irregular. The center of the floor, which is not preserved, may well have borne a composition of stamped decoration enclosed by multiple grooves. The interior surface is smooth, as is the very limited area of the exterior surface

that is preserved. There are small, somewhat irregular furrows along the inner edge of the swag of beads, the outer edge of the recessed band along the inner edge of the rim flange, and the junction between the interior surface of the wall and the floor. From seams visible in the breaks it is evident that this vessel was formed from at least three separate sheets of paste. The locations of these seams are indicated in the profile drawing of the piece. One sheet (no. 1) corresponds to the exterior part of the entire preserved portion of the vessel, the second (no. 2) corresponds to the upper surface of the rim flange from its outer edge to the exterior of the recessed band along its inner edge, and the third (no. 3) corresponds to the interior surface of the wall and the portion of the floor outside the recessed panel at its center. Although the portion of the vessel made up of no. 1 may have consisted of a single sheet of paste that covered the entire interior of the mould, the portions of the vessel made up of no. 2 and no. 3 were each probably built up from 4 separate sheets, one for each side of the vessel. The forming of this vessel thus probably involved the cutting out and positioning of 1 large and 8 small sheets of paste. The appearance of the swag of beads on the upper surface of the rim suggests that it was probably executed in barbotine, rather than moulded.

The specific sequence of operations involved in the forming and decorating of the vessel can be reconstructed as follows: 1. Sheet 1 was pressed inside a rectangular mould. 2. Sheet 2 was laid atop the outer edge of Sheet 1 to form the upper surface of the rim flange, with this operation carried out once for each of the vessel's 4 sides. 3. Sheet 3 was laid atop the inner surface of Sheet 1, abutting the inner edge of Sheet 2, to form the inner surface of the wall and the outer portion of the floor, with this operation being carried out once for each

4.

of the vessel's 4 sides. Asharpimplement was used to trim the outer edges of Sheets 1 and 2 (presumably it was run along the

5.

A slip trailer was employed to produce a swag of barbotine beads along the upper surface of the rim.”

outer edge of the mould) to form the outer face of the rim. 6.

Apointed implement was employed to impress a furrow along the inner edge of the swag of beads to highlight this element, to impress a furrow along the contact between Sheets 2 and 3 at the outer edge of the recessed band along the inner edge of the rim flange to seal this juncture, and to impress a furrow along the juncture between the inner surface of the wall and the floor to sharpen this feature. [7. A pointed implement was employed to cut two or more decorative grooves enclosing the central area of the floor.] [8. Stamps were employed to impress a complex decorative scheme in the central area of the floor.] 9. The vessel was removed from the mould. 10. The interior and exterior surfaces of the vessel were slipped (probably by dipping the vessel into a large container of slip). Operations 7 and 8 have been placed in brackets as it is not certain that they were carried out for the vessel

in question. The fact that the slip on the exterior surface flaked away nearly everywhere except in the area immediately adjacent to the rim is a point of some interest, as it sheds additional light on the manufacturing process. When the vessel was still seated in the mould, the evaporation of the pore water that it held would have proceeded more rapidly on the side facing away from the mould (i.e., the vessel's interior surface) than on the side in contact with it. This differential would have been less pronounced in the area immediately adjacent to the rim, where the water could have evaporated via the outer face of the rim and the small gap that would have opened up between the exterior surface of the vessel and the mould, where the former pulled

32

Fora rectangular dish/platter in Sigillata D that has a flanged rim with a straight row of beads along the outer edge of its upper surface, a low, steep, scalloped wall, a flat base, and a complex composition of stamped decoration at the center of the floor, see Carandini and Tortorella 56); Mackensen 1993, 357-59 (Hayes Form 47.1-2).

33

For slip trailers, see Whitford and Wong 1986, 157-65.

1981b, 92 (Hayes Form

60

J. T. Pena

away as the vessel dried and shrank. The pattern in the preservation of the slip on the exterior surface suggests that, at the time that the slip was applied, the exterior surface in the area farther from the rim had not yet dried thoroughly, resulting in a poor fit between the vessel and the slip, with the latter only weakly bonded to the former and susceptible to flaking. This suggests that, after being removed from the mould, the vessel was not allowed to dry thoroughly before it was slipped. Discussion

The approach employed for the forming of the example of Hayes Form 9B in Sigillata A considered above is representative of that utilized by the workshop(s) that produced this class for the manufacture of the small to medium-sized bowls (Hayes Forms 3, 6, 8, 9, 14, 16, 17) that represent the bulk of production. This involved the throwing of a blank with a small disk base,

followed by the restricted use of turning on its exterior surface, including the turning out of the interior of the disk base to form a ring foot and, for some forms, the accentuation of the angle or

the refining of the curve in the wall.*4 The approaches employed for the forming and slipping of the example of Hayes Form 52B in Sigillata C and the example of Hayes Form 61A/B in Sigillata D are representative of those employed for the manufacture of most of the more commonly attested forms in these two classes,

as well as those attested in Sigillata C/E and E. This involved the throwing of what was presumably a flat-based blank, followed by the extensive use of turning on its exterior surface, including shaping the wall and cutting out the base. The base generally took the form of a flat base, an inset base, and a tiny ring foot or a double ring foot with a low, triangular profile (as on

Hayes Forms 45 and 46 in Sigillata C, Hayes Form 67 in Sigillata D, and Hayes Form 68 in Sigillata E). This approach, which involved the turning off of paste over a large portion of the exterior surface, though at no point to any great depth (in contrast with the turning out of a ring foot, as with Hayes Form 9B), may have originated with the central Tunisian workshops

responsible for the manufacture of Sigillata C and A/D that commenced the production of these classes at some point during the late 2nd or early 3rd c. It should be noted, however, that it was

also employed by for the forming of period. Dishes and that manufactured proach related to

the N Tunisian workshop(s) responsible for the manufacture of Sigillata A two common dish forms (Hayes Forms 27 and 31) during roughly the same other forms with broad bases loomed large in the output of the workshops Sigillata C and A/D, and it may well be that the adoption of this apthe requirements of the manufacture of forms of this kind. Specifically,

while it entailed the removal of paste over a large portion of the exterior surface, this did not amount to a great deal of material, given the shallow depth of turning. The turning of the exterior surface of blanks could thus have been carried out fairly rapidly and easily, with little likelihood of ruining the blank by cutting too deeply into its wall or base, while generating only a very small amount of turned-off paste, thereby keeping to a minimum the paste lost through turning and the time and effort required for potter or assistant to collect turned-off slivers of paste for recycling. It seems possible that the blanks for dishes and other broad-based forms in Sigillata C, C/E,

and D were thrown on bats in order to minimize the risk that they would distort or collapse when removed from the wheel. The advantages to be gained from this approach would have been particularly notable for the workshops that produced Sigillata C and C/E, since the

blanks for these classes presumably had notably thin walls. This practice would have subjected workshops to some of the disadvantages associated with the use of wheel-mounted moulds, in that these establishments would have been obliged to manufacture and maintain a large

number of bats, while the number of examples of these forms that they could manufacture at any one time would have been limited to the number of bats on hand. The fact that disk-shaped objects that might have been employed as bats have not been reported from N African workshop sites strongly suggests that, if bats were employed for the manufacture of these forms, 34

With some examples of these forms it appears that a different approach was employed, involving the

throwing of a flat-based blank, followed by the turning out of a broad cylinder at the center of the base and the insertion of a separate strip of paste along the edge of the resulting cut to form a ring foot.

The forming and slipping of African sigillata

61

they were manufactured of some material not normally preserved in the archaeological record,

wood being perhaps the most likely possibility.# The blanks for dishes and drical chuck for the turning ceramic ring from El Mahrine for the turning of blanks for manufactured at El Mahrine exterior surface, while

other broad-based forms were presumably mounted on a cylinof their interior and/or exterior surface. While the fragmentary identified as a probable chuck is too small to have been employed forms of this kind, the blanks for various medium-sized forms may have been placed over this item for the turning of their

blanks

for various

small

forms

manufactured

there may

have

been

placed inside it for the turning of their interior surface. The cluster of 4 closely-spaced grooves on the exterior surface of this item immediately below its mid-point may have served as guidelines that facilitated the positioning of blanks for the turning of their exterior surface. The rarity of ceramic chucks of this kind from N African workshop sites suggests, however, that turning was normally carried out with the aid of chucks (both ring-shaped and solid) made of plastic clay.?® Various forms of evidence indicate that, for the manufacture

of Sigillata A/D,

C, C/E, D,

and E, blanks were regularly subjected to turning while they were still somewhat plastic. Most indicative in this regard is the fact that, for several of the broad-based forms belonging to these classes,

the central portion

of the floor/base

is bowed

downwards,

reaching

a point

somewhat lower than the edge of the base. This is presumably the result of the application of downward pressure in the course of the smoothing / compression of the floor that was carried out in connection with the turning of the interior surface. From this it can be inferred not only that the blank was still flexible when this operation was performed, but that the turning of the exterior surface was undertaken prior to that of the interior surface, and that for the turning of the interior surface the blank was mounted in a ring-shaped chuck, with the edge of its base seated somewhat above the surface of the wheelhead, rather than in contact with it. That the blank/vessel was still somewhat plastic even at the conclusion of the turning process can be inferred from the fact that the production of the stamped decoration that appears in the floor of many forms in Sigillata D and E (an operation executed after the turning of the exterior and interior surfaces)

could

not have

been

carried

out if the floor of the vessel had

reached

a

leather-hard state. This is also demonstrated by Hayes Form 594, a flanged dish in Sigillata D with vertical flutes in the exterior surface of the wall.” The production of the vertical flutes (an operation carried out subsequent to the turning of both the exterior and interior surfaces)

involved pressing an elongated implement with a curved surface into the exterior surface of the wall. In most examples this produced a more or less pronounced bulge in the interior surface of the wall opposite the point where each of the flutes was impressed. It may be possible to draw connections (if only in a tentative fashion) between the methods employed for the forming and slipping of African Sigillata and the circumstances that governed the development and/or operation of that industry. In the case of Sigillata A, we can note that the approach employed for the turning of the exterior surface of the bowl forms comprising the bulk of production required less labor than that employed in the manufacture of Italian and Gallic Sigillata, which generally involved the substantial thinning and shaping

of the exterior surface of the wall and the cutting out of both sides of the ring foot.?? While by itself this probably represented only a very small economy, the producers of African Sigillata A may have been able to combine this with other economies, both large and small, to produce vessels of satisfactory quality in a relatively efficient manner. In Tunisia, where oleoculture probably represented an unusually large component of the agricultural sector, these other economies may have included the employment of seasonally-underemployed agricultural workers at 35

For ceramic disks from Gallic Sigillata workshop sites at Montans and La Muette/Lyon that likely functioned as bats, see Desbat 2004, 150 fig. 25.

36 37

38

For ceramic disks from a workshop site at Rue de la Chapeau Rouge, Lyon, that might have functioned as chucks for the turning of thin-walled ware, see Desbat 2004, 151, 152 fig. 29. Forthis form, see Hayes 1972, 96-100; Carandini and Tortorella 1981b, 82-83; Mackensen

For the methods employed for the turning of Gallic Sigillata, see Czysz 1982, 310-21.

1993, 317.

62

J. T. Pena

low wages for tasks requiring unskilled and semi-skilled labor, such as the digging and transport of clay, the preparation of paste, and the collection of fuel consisting of olive prunings and olive pressings (probably available at token or no cost). Over the course of the 2nd c., the

workshop(s) that manufactured Sigillata A abandoned the practice of executing bands of decorative chattering on the interior of some bowl forms and the exterior of others, while at

the same time shifting to the use of a paste that contained a distinctly denser and coarser aplastic component. The relationship between these two developments is unclear, in the sense that it is uncertain whether the shift to a coarser paste rendered the execution of chattering problematic, or whether the abandonment of chattering freed up producers to employ a coarser

paste.? Whatever the relationship, this development may represent an extension of efforts on the part of producers to realize economies in the manufacturing process, perhaps in the context of opportunities for expanding the size of their market. The shift to a coarser paste may have involved either the exploitation of one or more new clay sources, perhaps as a consequence of the exhaustion of the traditional source or sources, or the streamlining of the paste preparation process, with less effort made to remove the coarse aplastic component from the clay. Also worth noting is the fact that the coarser paste was suitable for the manufacture of utilitarian wares, cookwares, and amphorae. The adoption of this paste for the manufacture of high-end tablewares would thus have permitted a workshop to turn out a broad array of products using a single, all-purpose paste, the preparation of which required relatively little labor. The other classes of African Sigillata are noteworthy for the lack of attention the potters paid to finishing the exterior surface. Producers generally made little or no effort to smooth away turning facets on the exterior surface of the wall and they adopted an indifferent and

inconsistent approach to the slipping of the exterior surface. Although the approach employed for the turning of the exterior surface was different from that employed for the manufacture of bowls in Sigillata A, it would also have entailed relatively little labor. While these approaches might have been tied to the suite of forms (dishes and other low, broad-based

forms

that had large interior and upper surfaces that were readily visible to users, while presenting little in the way of visible exterior surfaces) that represented the most substantial component of production in these classes, they may to some extent represent an extension or even an intensification of the economizing efforts associated with the manufacture of Sigillata A. Conclusions

An evaluation of the African Sigillata from the Palatine East excavations indicates that the manufacture of the vast majority of vessels in Sigillata A/D, C, C/E, D, and E through at

least c.450 involved the freehand throwing of a blank followed by the turning of its exterior surface or of both its exterior and interior surfaces. The limited use of turning in the manufacture may have reflected an effort on the part of potters to:achieve economies in the manufacturing process. The use of moulds was restricted to the manufacture of a very limited number of vessels belonging to specialty forms, such as rectangular plates/platters, that could not be thrown and turned on a wheel. There is no evidence that the manufacture of these classes involved the use of wheel-mounted moulds in a process similar to that involved in the manufacture of reliefdecorated Italian Sigillata and Gallic Sigillata. Barring the discovery of new evidence to the contrary, scholars should discard the view that it did. 39

The fact that some examples of African Sigillata A from the Palatine East assemblage manufactured in the coarser fabric have decorative chattering suggests that the shift to a coarser paste may have been the independent variable, and the abandonment of chattering the dependent variable. From examination of

these vessels, it is evident that the coarseness of the paste rendered it difficult to execute neat, sharplycut chattering, resulting in vessels that may well have been regarded as unsatisfactory from an aesthetic point of view.

The forming and slipping of African sigillata

63

Bibliography Barbaformosa

1999. The potter's wheel (Hauppage, NY).

Donifay, M. 2004. Etudes sur In céramique romaine tardive d'Afrique (BAR S1301, Oxford). Carandini, A. (ed.) 1981. Atlante delle forme ceramiche I. Ceramica fine romana nel bacino mediterraneo (medio e tardo impero) (EAA Suppl. 2, Rome). Carandini, Carandini, Carandini, Carandini, Carandini,

156-63.

A. A. A. A. A.

and and and and and

S. S. S. E. E.

Tortorella Tortorella Tortorella Tortorici Tortorici

1981a. 1981b. 1981c. 1981a. 1981b.

"Produzione "Produzione "Produzione "Produzione "Produzione

A," in Carandini 1981, 19-52. D," in Carandini 1981, 78-117. E," in Carandini 1981, 119-22. C/E," in Carandini 1981, 117-18. C? e C4 decorata a rilievo a matrice," in Carandini 1981,

|

Clark, K. 1999. The potter's manual (3rd edn., Edison, NJ).

Czysz, W. 1982. “Der Sigillata-Geschirrfund von Cambodunum-Kempten: ein Beitrag zur Technologie und Handelskunde mittelkaiserzeitlicher Keramik,” BerRGK 63, 281-345.

Desbat, A. 2004. “Les tours de potiers antiques," in M. Feugére and J. C. Gerold (edd.), Le tournage, des

origines à l'an Mil. Actes du colloque de Niderbronn, octobre 2003 (Monografies Instrumentum 27) 137-54. Hayes, J. W. 1972. Late Roman pottery (London). Hayes, J. W. 1997. Handbook of Mediterranean Roman pottery (Norman, OK) Hostetter, E. et al. 1994. "Excavations of the Late Roman complex on the NE slope of the Palatine Hill (198991),” in Rome papers (JRA Suppl. 11) 131-81. Ikáheimo, J. 2003. Late Roman African cookware of the Palatine East excavations, Rome: a holistic approach (BAR 51143, Oxford). Mackensen, M. 1993. Die spätantiken Sigillata- und Lampentöpferein von El Mahrine (Nordtunisien) (Münchner Beiträge zur Vor- und Frühgeschichte 50). Mackensen, M. 1998. “Arbeitsgerate aus dem spätantiken Tópferzentrum von El Mahrine (Nordtunisien),”

RómMitt 105, 431-39. Mackensen, M. and G. Schneider 2002. "Production centres of African red slip ware (3rd-7th c.) in northern and central Tunisia: archaeological provenance and reference groups based on chemical analysis," JRA 15, 121-58. Mackensen, M. and G. Schneider 2006. "Production centres of African red slip ware (2nd-3rd c.) in northern and central Tunisia: archaeological provenance and reference groups based on chemical analysis," JRA 19, 163-90. Pena, J. T. 1999. The urban economy during the Early Dominate: pottery evidence from the Palatine Hill (BAR 5784, Oxford) Pena, J. T. 2007. "The quantitative analysis of Roman pottery: general problems and the methods employed for the analysis of the Palatine East assemblage," in E. Papi (ed.), Supplying Rome and the empire (JRA Suppl. 69) 153-72. Rye, O. and C. Evans 1976. Traditional pottery techniques of Pakistan: field and laboratory studies (Smithsonian Contributions to Anthropology 21, Washington, D.C.). Sagui, L. 1981. "Produzione C," in Carandini 1981, 58-78.

Schuring, J. 1988. "Terra sigillata Africana from the San Sisto Vecchio in Rome," BABesch 63, 1-48. Whitford, P. and G. Wong 1986. Handmade potter's tools (New York).

Methodological constraints affecting the precise dating of African Red Slip Ware John Lund African Red Slip Ware is probably the best-documented Roman fine ware of the Mediterranean thanks to eminent research by J. W. Hayes along with scholars such as N. Lamboglia, J. W. Salomonson, L. Anselmino, C. Pavolini, M. G. Fulford, and M. Mackensen.! Now

M. Boni-

fay has joined this select group with his Etudes sur la céramique romaine tardive d'Afrique, which provides a clearer and more comprehensive picture of the ceramic industries of Africa Proconsularis and Byzacena (transport amphorae, table wares, plain wares, terracotta lamps)

than had previously been available.? In 2008, wishing to advance our knowledge of Roman and late-antique tablewares, Bonifay, together with M. Ángel Cau and P. Reynolds, organized an Exploratory ECREA/ESF Workshop in Barcelona on ‘Late Roman Fine Wares: solving problems of typology and chronology’. ? Its goal was to review the typology and dating evidence for African Red Slip Ware, Phocaean Red Slip Ware, and Cypriot Red Slip Ware. Amongst other things, the resulting publications will comprise a new Conspectus of these Late Roman fine wares complete with revised dates. The present contribution was inspired by my participation in this stimulating enterprise, and may be thought of as a kind of prolegomena to it. Whereas the Barcelona Workshop focused on the chronology of the specific forms of these wares, my aim here is to discuss certain methodological issues — using African Red Slip Ware forms as an example — that are relevant to the precision with which we can ascribe dates to any artefact type. Archaeologists are familiar with these issues, but they are rarely discussed in connection with the chronology of ceramic fine wares, despite the implications they have for the degree of precision with which we can assign dates.* As I cannot here consider all the issues involved, I will concentrate on four that seem to be the most pressing. Point of departure In his pioneering publication Late Roman pottery, J. W. Hayes presented a new typology for

African Red Slip Ware together with suggested dates for most forms and their variants.? He established these through analysis of contexts in which the forms in question were associated with other datable artefacts — a common enough procedure in archaeology.? Contexts from the Athenian Agora supplied "a large proportion of the available dating evidence", together with data from sites such as Antioch, Dura-Europos, Emporio

(Chios), Corinth, Istanbul, Kóln, Abu

Mena, and Karanis.” Hayes often relied on deposits with associated coins,’ but he declined to "speak of sealed groups" because he recognized that much of the material came from dubious excavations, preferring instead the term “probable association of finds"? 1

Among the principal publications are Hayes 1972 and 1980; Lamboglia 1958 and 1963; Salomonson

2

Schneider 2002 and 2007. Bonifay 2004.

1968

3 4 s 6

and

1969;

Anselmino

and

Pavolini

1981;

Fulford

1984;

Mackensen

1993;

Mackensen

and

The Workshop (held on 5-9 November, 2008) was supported by the Institució Catalana de Recerca i Estudis Avangats and the European Science Foundation. I take this opportunity of thanking the organizers for inviting me to participate. See Millett 1987 for a broader discussion based on Romano-British pottery. Hayes 1972. For succinct summaries of this method, see, e.g., Peacock 1982, 161-65 and passim; Biers 1992, 53-54 and

passim; Orton et al. 1993, 189; Millett 2000, 54-55; Caple 2006, 211. Cf. Hayes 1972, 1-2 and 1980, 481-83; Peacock 1989, 161-62; Fentress 1998.

Hayes 1972, 24 (Form 3), 30 (Form 6), 37 (Form 9), 41 (Forms 14 and 15), 42 (Form 16), 48-49 (Form 24), and so on. For appraisals of the relationship between coins and pottery, see Rotroff 1997; Poblome 2008 9

with references. Hayes 1972, 2.

66

J. Lund

The scholars who have subsequently discussed the chronology of African Red Slip Ware have followed the same methodology, adding new dating evidence — for instance, from contexts such as the Crypta Balbi or the Palatine East in Rome.!? Alternative dating methods (e.g., those based on scientific analyses) have had no relevance to the dating of African Red Slip Ware;!! the same goes for seriation, a technique commonly applied for ordering assemblages and pottery types in prehistoric archaeology, although it has been applied successfully to Sagalassos Red Slip Ware.!? The fact that Hayes' chronology has largely stood the test of time testifies to the soundness of his approach.!? My aim is not to question this time-honoured methodology, but to focus on four methodological issues that seem to set certain limitations on our ability to establish a precise start and terminal date for each ARS form. 1. What are we dating?

In 1993, C. Orton, P. Tyers and A. Vince noted that At least two definitions of the date of an artefact are in use: (i) the date at which it was made, (ii) the range of dates within which artefacts of its type were commonly in use.!? J. Poblome later observed that Pottery can, in fact, be dated in at least three ways: when it first came on the market, when it was

most popular, and when it became part of the stratigraphy in which it was found. The two last possibilities need careful distinguishing, for even if the vessels found in a particular stratigraphical unit date from the third century AD, for instance, this does not necessarily imply that the unit

originated in that century.P? These statements raise the issue of what is implied by assigning a date to a ceramic form: its period of manufacture? marketing? use (in the widest sense of the word)? discard? or systemic context?16 Orton and his co-authors preferred the first option, but stressed the need for scholars

to clarify which definition they use. Yet few of those who have discussed the dating of African Red Slip Ware have addressed this issue. One exception is M. Mackensen, who clearly thinks in terms of the date of production since he speaks about the Herstellungsbeginn and Produk-

tionsende of specific forms in his book on the manufacturing centre of El Mahrine.! If we widen the focus to include other kinds of Roman table wares, B. Hartley and B. Dickinson state, in their introduction to a new corpus of names on Gallo-Roman terra sigillata, that in In dating the potters we have attempted always to give the range within which the pots were made.!®

Their carefully researched survey of the dating evidence shows that they too base their dates on context associations between pottery and other datable artefacts, but one searches in vain for

an explication of how they derived dates of the manufacture of the pots from such evidence. By contrast, C. M. Wells, in the introduction to the Conspectus formarum terrae sigillatae italico modo

confectae (1990), wrote: Where dates are given, they are those of the find-context, not necessarily of manufacture.!?

10

Cf. Sagui 1998; Pena 1999; Arena et al. 2001.

11

Cf. Mackensen 1993, 385-96.

12

Cf. Peacock 1989, 163: “seriation or sequence dating has been little used on Roman pottery but offers

considerable scope for ordering material". For Sagalassos Red Slip Ware, see Groenen and Poblome 2003; van de Velden et al. 2008, with references.

13

This was confirmed by the review of the chronological evidence undertaken during the Barcelona Workshop.

14

Orton et al. 1993, 185.

15

17 18

Poblome 1999, 272; cf. also Steuer 1998, 136-42: "Die entscheidende Fragestellung: Was soll eine chronologische Angabe aussagen?" Pefia 2007, 7, defines systemic context as "a situation in which it [an artifact] is involved in a human behavioral system," in contrast to its archaeological context. Mackensen 1993, 433-35 and passim. Hartley and Dickinson 2008, 4.

19

Wells 1990, 2; see further Roth-Rubi 1990.

16

Methodological constraints affecting precise dating of ARS Ware

#7

As a matter of fact, it seems evident that, as long as we derive our dates of individual African Red Slip Ware forms from their occurrences in contexts of use and discard, then it is inherently

impossible for us to date anything but the time of their use and/or discard.?? Nearly all of the examples of dating pottery manufacture given by Orton, Tyers and Vince relate to documentary

sources or other direct evidence about pottery production in later historical periods?! Such information is not available for African Red Slip Ware, and we have no dated contexts from Tunisian kiln sites which might allow us to pinpoint the production periods of the artefacts produced there.?? 2. Implications of the theory of pottery life cycles In 2007, J. T. Pefia launched a new model of the life-cycle of Roman pottery which is not without chronological importance.”? It distinguishes between the following phases in the ‘life’ of a pot: manufacture,

distribution,

prime

use, re-use, maintenance,

recycling, discard, and

reclamation.?? This represents an ideal sequence, of course, because a piece of pottery might have shattered and been discarded during any of the first 5 phases. Kiln sites will yield evidence for the manufacture phase but it cannot be assumed a priori

that all pottery found at such a site was produced there.? Finds in shipwrecks are of prime importance as far as the distribution phase is concerned,*° even if one must be careful to exclude possible personal possessions of crew-members and intrusive finds of later periods;^ shops with

pottery for sale are also mentioned in connection with the distribution phase.”® The next two phases are particularly relevant to the present query. Pefia defines prime use as "the use of a vessel for the application or applications for which it was manufactured", and re-use as "the use of a vessel or a vessel part for some application after the conclusion of its use for its prime-use application". He notes that the "division into prime use and reuse ... is to a certain extent problematic", and that the distinction works best with transport amphorae.? In the case of fine wares, it is certainly difficult at times to distinguish prime use from re-use of Pena’s type A (i.e., "involving an application similar to the vessel’s prime-use application without any physical modification to it") and B (i.e., "involving an application different from

the vessel's prime-use application without any physical modification to it"). Indeed, the instances of re-use of fine wares (including African Red Slip Ware) quoted by Pefia all comprise deliberate modification of the vessel, and thus adhere to his type C?! Recycled fine wares are

likewise 20

recognizable

through

the modifications

they have

undergone.” Finally,

Pefia

Cf. Millett 1987.

21

Orton et al. 1993, 187-88.

22

The kilns excavated at Oudhna do not seem to have yielded independent dating evidence: cf. Barraud et al. 1998. Otherwise, the ceramic kiln sites are known only from surveys: Peacock et al. 1989 and 1990; Mackensen 1993, 382; Ben Moussa 2006 and 2007, 132; Nancef 2007.

23 24 25 26 27

Cf. Pena 2007, 6-16. CA ibid. Ibid. 32-35. Parker 2008. Cf. Pefia 2007, 35-38 and, for a prime example, Millett 1993. For wrecks containing African Red Slip Ware, see Parker 1992, 128 no. 268, 137-38 no. 292, 140 no. 301, 168 no. 375, 169 no. 376, 177 no. 398,

180 no. 415, 186 no. 434, 219-20 no. 524, 236-37 no. 578, 262 no. 659, 267 no. 671, 270-71 no. 682, 27980 no. 708, 329 no. 873, 334 no. 880, 345 no. 912, 383-84 no. 1030. Wreck

sites published after 1992

will be enumerated in the Conspectus of ceramic contexts to be published by the Barcelona conference. 28

Pena 2007, 36.

29 30 31

Ibid. 8-0. Ibid. 10-11. Ibid. 197-208. Moreover, in a review R. Tomber (JRA 21 [2008] 500) noted that Pena's "flow diagrams

show the majority of vessels being discarded after prime use, with less than c.5% re-used before or after 32

this stage". Cf. Pena 2007, 253.

68

J. Lund

envisages four types of behaviour in the Roman world that would "generate significant amounts of pottery that was disposed of by Discard”: 1, 2. 3. 4.

Pottery workshops. Wholesale/storage and retail facilities for pottery. Wholesale/storage and bulk retail facilities for wine, oil and/or fish products. Residences and other loci where food was regularly stored in moderate quantities, prepared, and consumed ...*?

It is indicative of the complexities involved that ceramic workshops as well as wholesale/ storage and retail facilities were also singled out as sources for the manufacture and distribu-

tion phases.?* This model implies that we must distinguish between these phases when analyzing contexts with ceramic vessels if these are to serve as chronological pointers. Yet the brief summary given above shows the difficulties involved in doing this, except in the cases of re-use (type C) and recycling, which involved a physical modification of the vessels; but these phases are largely irrelevant from a chronological point of view. By contrast, the most untainted chronological evidence surely comes from what M. Schott referred to as "life assemblages" — i.e.,

“sets of vessels recovered in use-related contexts". Such contexts are unfortunately rare,” except for wrecked ships, which belong to Pefia's distribution phase.

3. How to date the introduction of an African Red Slip Ware form? D. A. Spratt developed a model to track innovative processes in which these were broken down into 6 phases: (a) discovery, (b) invention, (c) development,

(d) investment,

(e) produc-

tion/distribution, and (f) obsolescence.*° He noted that "the duration of each phase may differ

from case to case, but all innovations pass through all of these stages".? S. Rotroff applied this model to Athenian mouldmade bowls and concluded that Stages B and C were of very short duration, but [...] there was a surprising lag in time before stage E — profitable production — was achieved. That at least seems to be the message of the small number of fragments found in Agora deposits dating within the first 40 years of production.”

Another example of this is the time lag between the emergence of Italian Sigillata in the West

and its popular appearance in the E Mediterranean.?? There is no reason to think that African Red Slip Ware was any different. Also, it is unlikely that a novelty would have hit the markets at exactly the same time everywhere. Since there seems to be a tendency for "both the number of find-sites and quantities

to diminish with increasing distance from a production centre”, one might assume (though there is no proof of this) that a novel type was first offered for sale close to its place of manufacture and only in the event of a successful reception by local consumers distributed to more distant markets after an unspecified time lag.# If this is correct, then datable contexts located close to the presumed source may be more reliable indicators of the introduction of a given type than more distant ones. More to the point, nearly all of the independently dated contexts used to date African Red Slip Ware forms provide termini post quos for the closure of the said contexts, and it is in most cases impossible to determine how long any object might been used before ending up there (5 33 34

Ibid. 272. lbid. 291-99.

35 36

Ibid. 18-19; for site formation processes, see further Hingley and Willis (edd.) 2007. Spratt 1982 and 1989.

37 38

Rotroff 2006, 368. Ibid. 373.

39 40

See Poblome et al. 2004, xii-xiii and passim. Peacock 1989, 167.

41

As noted by Peacock (1989, 167), this ideal distribution pattern is often "disturbed by factors such as the function of the find-site, topography and method of transport, or the situation of the markets and marketing methods".

Methodological constraints affecting precise dating of ARS Ware years, 10 years, more?). We have little exact knowledge about the use life of ceramic fine wares. Pefia suggested that "many vessels in this category remained in use for a relatively short period of time, perhaps on the order of no more than a year", yet he cites African Red Slip Wares from Rome and Spain with traces of wear that are suggestive of prolonged use. In the case of Italian-type terra sigillata from the Villa of Livia at Primaporta, A. Klynne seems to reckon with an average use-life of 10 years for plates and 5 years for drinking vessels,?? whereas J. Poblome's assessment was that Sagalassos Red Slip Ware had a use life of about 25 years.“ An even higher figure (41 years) assumed for finds from Haltern may be excessive.*? No

doubt conditions varied from one place to the next, depending, for instance, on how easy it would have been to replace broken pots. In any event, we have no way of knowing at which point in its use life a piece of pottery ended up in the archaeological context we rely on for dating purposes. 4. How to ascertain the end date of an African Red Slip form? If we are right in thinking that it is only possible for us to date the use and/or discard of a given form (and not its period of production), it becomes impossible to assign it a universal end date, since its use life surely ceased at different points in time at different places. Also, some allowance must be made for special cases such as heirlooms or artefacts preserved for centuries because of their religious or other connotations.4ó H. Steuer has produced an interesting illustration of the imponderables involved in combining the time of production of items of jewelry and weaponry with that of its circulation, acquisition, and eventual deposition in a tomb upon the owner's death. He concluded that the theoretical time-frames involved might vary between less than a generation and more than a

century.? Tomb groups have played a relatively small rôle in establishing the chronology of African Red Slip Ware, but those contexts that have done so often comprise a large number of

residuals, which tend to confuse the issue even further.*? In practice, it is rarely possible to distinguish between residual finds and objects with an exceptionally long use life. A possible way around this might be to construct separate chronologies for each African Red Slip form for each geographical region or Roman province. It seems possible to do so in the case of Roman lamps, and there is no reason why something similar might not be attempted for other artefacts.*? A pattern deviating from that of the source region might help to distinguish residual finds, even if it is also possible that products remained longer in use farther from the source area. Conclusions

Artefact dating is based nal and external evidence caution when ‘copying and logies are neither absolute interpretation which make that: 42 43 44 45

on a pattern of interpretation, which needs to take all other interinto account. This implies that third parties should exercise great pasting’ the results.?? M. Millett's observation "that pottery chrononor precise but are the result of a process of generalization and them inherently fuzzy’?! is confirmed by the above, which suggests

Pefia 2007, 58-60. Klynne 2002, 36-38. Poblome 1996, 85. Von Schnurbein 1982, 133. Cf. Klynne 2002, 37-38.

46

Dark 1995, 77.

47

Steuer 1998, cf. 141 fig. 4.

48

To quote but one example, pottery of the 4th c. A.D. accounted for only 16% of the ceramic material found in the foundation trench of the Arch of Constantine at Rome: Zeggio and Rizzo 1998, 146.

49

Cf. Lund 1991, 283-86 and 288-90.

50 51

lowe this observation to J. Poblome. Millett 2000, 54; see further Lucas 2005, 95-113.

70

J. Lund

1. So long as the chronology of artefact types (such as ARS forms) is determined on the basis of their occurrences in contexts of use and/or discard, we cannot date anything but their use and/or discard — not the production as such. 2. The clearest chronological evidence comes from "sets of vessels recovered in use-related contexts". Unfortunately, such contexts are rare. But pottery (except for personal possessions of crew members) found in uncontaminated shipwrecks certainly belongs to only one of Peña's phases, that of distribution, and finds from such contexts should therefore be more central to our attempts at establishing the chronology of ARS forms (and other artefact types) than they have been in the past. 3. Nearly all dates for ARS Ware are based on contexts providing us with termini post quos (of the contexts), but there is no way of knowing how long a given artefact might have been in use before ending up there; the terminus post quem therefore cannot tell us with precision

when the type emerged.?? If the use life of tableware pottery lies between 1 and 25 years, then the date at which it was originally quem of the context by between 1 and 25 the future record the presence (or absence) wear (much as numismatists observe the

acquired may equally predate the terminus post years. This implies that archaeologists should in of traces of cut marks or other signs of prolonged wear of a coin) on pottery from such contexts, in

order to determine whether it was new or old at the time of its deposition.?? 4. It is possible (though as yet unproven) that vessels found in datable contexts located close to their presumed geographical source may be more reliable indicators of the introduction of a given type than more distant ones. 5. It is unrealistic to believe that we can determine a precise end-date for the use of a given form, since it certainly did not go out of use everywhere simultaneously. This contribution is not intended to question the general validity of current (or future) chronological systems for dating African Red Slip Ware; it intends merely to define the inherent methodological limits to any such scheme. Even if it is acknowledged that the dates suggested for African Red Slip Ware forms are nothing but approximations, this should not deter us from continuing to refine the chronology. True, absolute precision is a goal that may never be reached, but the uncertainty shrinks with each new find in a dated context. This is precisely what is demonstrated so brilliantly in M. Bonifay's Etudes sur la céramique romaine tardive d'Afrique, which is currently our most reliable guide to the chronology of African Red Slip Ware — and much else besides. Acknowledgements I wish to thank Morten Axboe and Peter Pentz (Prehistoric Department of the Danish National Museum) and Jeroen Poblome (Sagalassos Archaeological Research Project, Katholieke Universiteit Leuven) for helpful comments and criticism. Bibliography Anselmino, L. and C. Pavolini 1981. "Terra sigillata: lucerne,” in Atlante delle forme ceramiche 1. Ceramica fine romana nel bacino mediterraneo (medio e tardo impero) (EAA, Roma) 184-207. Arena,

M. 5. et al. (edd.) 2001.

Roma

dall'antichità al medioevo.

Archeologia

e storia

nel Museo

Nazionale

Romano Crypta Balbi (Milano). Barraud, D. et al. 1998. "L'industrie céramique de l'Antiquité tardive," in H. Ben Hassen and L. Maurin (edd.), Oudhna

(Uthina). La redécouverte d'une ville antique de Tunisie (Ausonius

Publications

Mémoires

2, Bordeaux) 139-67. Ben Moussa, M. 2006. "Remarques sur la céramique de surface de Hr. Bloul," in F. Bejaoui (ed.), Actes du 4eme Colloque int. sur l'Histoire des Steppes tuntsiennes (Sbeitla 2003) (Tunis) 115-41. Ben Moussa, M. 2007. La production de sigillées africaines. Recherches d'histoire et d'archéologie en Tunisie septentrionale et centrale (Col-lecció Instrumenta 23, Barcelona). Biers, W. R. 1992. Art, artefacts and chronology in classical archaeology (New York).

52

Cf. Dark

1995, 67-71

and

76-78. Cf. Rotroff 2006, 363:

“a precise

date

for the introduction

of the

moldmade bowl will probably always elude us". 53

See,e.g., Walker 1997, 24; for a good illustration of cut marks on pottery, see Peña 2007, 59 fig. 4.2.

Methodological constraints affecting precise dating of ARS Ware

71

Bonifay, M. 2004. Etudes sur la céramique romaine tardive d'Afrique (BAR $1301, Oxford).

Caple, C. 2006. Objects: reluctant witnesses to the past (Abingdon). Dark, K. R. 1995. Theoretical archaeology (London). Ettlinger, E. et al. 1990. Conspectus formarum terrae sigillatae italico modo confectae (Bonn). Fentress, E. 1998. "Preface," in Sagui 1998, 5-6. Fulford, M. G. 1984. "The red-slipped wares," in id. and D. P. S. Peacock (edd.), Excavations at Carthage: The British Mission vol. 1.2. The Avenue du Président Habib Bourguiba, Salammbo: the pottery and other ceramic

objects from the site (Sheffield) 48-115. Gowlett, J. A. J. 2006. "Archaeological dating," in J. Bintliff (ed.), A companion to archaeology (Oxford) 197-

205. Groenen, P. J. F. and J. Poblome 2003. “Constrained Correspondence Analysis for seriation in archaeology applied to Sagalassos ceramic tablewares," in M. Schwaiger and O. Opitz (edd.), Exploratory data analysis in empirical research (Berlin) 1-30. Guidobaldi, F., C. Pavolini and P. Pergola (edd.) 1998. I materiali residui nello scavo archeologico (CollEFR

249). Hartley, B. P. and B. Dickinson 2008. Names on terra sigillata. An Index of makers' stamps & signatures on Gallo- Roman terra sigillata (Samian Ware) vol. 1 (À to Axo) (BullInstClassStudLon 102-01).

Hayes, J. W. 1972. Late Roman pottery (London). Hayes, J. W. 1980. Supplement to Late Roman Pottery (London). Hingley, R. and S. Willis (edd.) 2007. Roman finds: context and theory (Proc. conf. Durham; Oxford). Klynne, A. 2002. The Prima Porta Garden Archaeological Project. Terra sigillata from the Villa of Livia, Rome (Uppsala). Lamboglia, N. 1958. "Nuove osservazioni sulla 'terra sigillata chiara'," RStLig 24, 257-330. Lamboglia, N. 1963. "Nuove osservazioni sulla ‘terra sigillata chiara’,” RStLig 29, 145-212. Lucas, G. 2005. The archaeology of time (Themes in Archaeology; London). Lund, J. 1991. "Towards a better understanding of the production pattern of Roman lamps," Acta Hyperboren

3, 269-95. Mackensen, M. 1993. Die spätantiken Sigillata- und Lampentöpfereien (Münchner Beitráge zur Vor- und Frühgeschichte 50).

von

El Mahrine

(Nordtunesien)

Mackensen, M. and G. Schneider 2002. "Production centres of African Red Slip Ware (3rd-7th c.) in northern

and central Tunisia. Archaeological provenance and reference groups based on chemical analysis," JRA

15, 121-58. Mackensen, M. and G. Schneider 2006. "Production centres of African Red Slip Ware (2nd-3rd c.) in northern and central Tunisia. Archaeological provenance and reference groups based on chemical analysis," JRÀ

19, 163-90. Millett, M. 1987. "A question of time? Aspects of the future of pottery studies," BullInstArchLon 24, 99-108. Millett, M. 1993. "Samian from the sea: Cala Culip shipwreck IV," JRA 6, 415-19.

Millett, M. 2000. "Dating, quantifying and utilizing pottery assemblages from surface survey," in R. Francovich and H. Patterson (edd.), Extracting meaning from ploughsoil assemblages (The Archaeology of Mediterranean Landscapes 5; Oxford) 53-59. Nancef, J. 2007. "Nouvelles données sur l'atelier de poitiers de Henchir Ech Chekaf (Ksour Essef, Tunisie)," in M. Bonifay and J.-Ch. Tréglia (edd.), LRCW 2. Late Roman coarse wares, cooking wares and ampliorae in the Mediterranean. Archaeology and archaeometry II (BAR 51662 ii; Oxford) 581-91. Orton, C., P. Tyers and A. Vince 1993. Pottery in archaeology (Cambridge). Parker, A. J. 1992. Ancient shipwrecks of the Mediterranean & the Roman provinces (BAR S580, Oxford).

Parker, A. J. 2008. “Artifact distributions and wreck locations: the archaeology of Roman commerce,” in R. L. Hohlfelder (ed.), The maritime world of ancient Rome (Ann Arbor, MI) 177-96. Peacock, D. P. S. 1989. Pottery in the Roman world (London). Peacock, D. P. S., F. Béjaoui and N. Ben Lazreg 1989. "Roman amphora production in the Sahel region of Tunisia," in Amphores romaines et histoire économique: dix ans de recherche (CollEFR 114) 179-202. Peacock, D. P. S., F. Béjaoui and N. Ben Lazreg 1990. "Roman pottery production in central Tunisia," JRA 3,

59-84. Peria, J. T. 1999. The urban economy during the Early Dominate. Pottery evidence from the Palatine Hill (BAR S784, Oxford).

Peña, J. T. 2007. Roman pottery in the archaeological record (Cambridge). Poblome, J. 1999. Sagalassos Red Slip Ware: typology and chronology (SEMA 2, Leuven). Poblome, J. 2008. “Sherds and coins from a place under the sun. Further thoughts from Sagalassos,” Facta 2, 193-212.

72

J. Lund

Poblome, J. et al. (edd.) 2004. Early Italian sigillata. The chronological framework and trade patterns (BABesch

Suppl. 10). Roth-Rubi, K. 1990. “Datierung,” in Ettlinger ef al. 1990, 39-43.

Rotroff, S. I. 1997. “Coins and stratigraphy,” in Sheedy and Papageorgiadou-Banis 1997, 8-16. Rotroff, S. I. 2006. "The introduction of the moldmade bowl revisited. Tracking a Hellenistic innovation,"

Hesperia 75, 357-78. Sagui, L. (ed.) 1998. Ceramica in Italia: VI-VII secolo. Atti del Convegno in onore di John W. Hayes (Firenze). Salomonson, J. W. 1968. "Etudes sur la céramique romaine d'Afrique. Sigillée claire et céramique commune de Henchir el Ouiba (Raqqada) en Tunisie centrale," BABesch 43, 80-145. Salomonson, J. W. 1969. "Spátrómische rote Tonware mit Reliefverzierung aus nordafrikanischen Werkstátten. Entwicklungsgeschichtliche Untersuchungen zur reliefeschmückten Terra Sigillata Chiara 'C'," BABesch

44, 4-109.

Sheedy, K. A. and C. Papageorgiadou-Banis (edd.) 1997. Numismatic archaeology — archaeological numismatics (Oxford). Spratt, D. A. 1982. "The analysis of innovation processes," JArchSci 9, 79-94. Spratt, D. A. 1989. "Innovation theory made plain," in S. E. van der Leeuv and R. Torrence (edd.), What's new? A closer look at the process of innovation (London) 245-57. Steuer, H. 1998. "Datierungsprobleme in der Archäologie,” in H. Beck, D. Geuenich and H. Steuer (edd.),

Ruineninschriften als Quellen interdisziplinärer Forschung (Berlin) 129-49. van de Velden, M, P. J. F. Groenen and J. Poblome 2008. "Seriation by constrained correspondence analysis: a simulation study,” Computational statistics and data analysis (2008). See doi: 10.1016/j.csda.208.08.020.

von Schnurbein, S. 1982. Die unverzierte Terra Sigillata aus Haltern (Bodenaltertümer Westfalens 19). Walker, A. S. 1997. "Excavation coins: the use and misuse of numismatic evidence in archaeology," in Sheedy

and Papageorgiadou-Banis 1997, 17-26. Wells, C. M. 1990. "What is the Conspectus? A note to the reader," in Ettlinger et al. 1990, 1-2.

Zeggio, S. and G. Rizzo 1998. "I materiali residui come indicatori della storia di un sito: il caso della fossa di fondazione dell'Arco di Constantino," in Guidobaldi, Pavolini and Pergola 1998, 125-48.

|j

African Red Slip Ware on the move: the effects of Bonifay’s Etudes for the Roman East Philip Bes and Jeroen Poblome Following a brief introduction, this paper is intended to assess the impact of Michel Bonifay's Etudes sur la céramique romaine tardive d'Afrique (2004) on ceramological studies in the Roman East and, more specifically, on the aims and methodology of the ICRATES Project housed at the Katholieke Universiteit Leuven.! His revision of part of the typo-chronological framework of African Red Slip Ware (hereafter ARS) as originally laid down by J. W. Hayes in 1972 and 1980, in particular, is prone to affect the interpretation of the collected evidence for ARS in the Roman East. This collected evidence for ARS will be presented and considered based on Hayes' original typo-chronological framework, on the one hand, and by combining those of

Hayes and Bonifay/ on the other. We conclude by discussing some methodological and interpretative issues. Roman red-slipped pottery Red-slipped pottery of Roman date has attracted the attention of scholars since the late 19th c. Notwithstanding some early landmarks (e.g., F. O. Waagé's work on the material from Antioch on the Orontes?), it seems that the 1970s formed a turning point in the conceptual and thematic approaches toward Roman red-slipped pottery. They were significantly affected by J. W. Hayes’ seminal Late Roman pottery (1972)! and his Supplement to Late Roman Pottery (1980)? which

are still being widely

used

today; in Hayes’

words,

"the publication of Late

Roman Pottery in 1972 may be said to have marked the close of the initial phase of the study of the later fine wares of the Roman Mediterranean, pioneered by scholars such as Lamboglia and Waagé". In fact, Tunisia (and especially Carthage) became the subject of intensive archaeological investigation not long after the publication of Late Roman pottery, a situation Hayes dealt with in the Supplement. Hayes himself studied much of the Roman pottery excavated at

Carthage, as did J. A. Riley and M. Fulford, followed by R. Tomber’ and others. More recent investigations in Tunisia's non-coastal zones have examined the manufacture of ARS.® Pottery

studies over recent decades have resulted in a scholarly agreement that sees Tunisia (or Africa Proconsularis / Byzacena and adjoining regions) as one of the major pottery manufacturing areas of the Roman Mediterranean based on sometimes integrated archaeological and archaeometrical research. Bonifay's Etudes? is the most recent landmark in this field. The ICRATES Project In addition to its róle as a dating tool and, to a lesser extent, as a vehicle for illustrating the

import-export 'balance' of a deposit, site, or area, Roman red-slipped pottery receives growing attention from a variety of other research angles, amongst which is the conviction that Roman pottery has the potential of contributing to our understanding of the workings of 'the' Roman 1

The ICRATES (Inventory of Crafts and Trade in the Roman East) Project of the Katholieke Universiteit Leuven is directed by J. Poblome and supported by FWO Research Grant G.0788.09. All dates in this

paper are A.D., unless otherwise mentioned. All graphs present absolute numbers.

E.g., Hayes 1976 and 1977; see also, e.g., Riley 1981, Fulford 1984 and 1994, Tomber 1988. E.g., Peacock et al. 1990, Mackensen 1993, Slim et al. 2008.

Bonifay 2004.

U) aA

Id. 1980. Ibid. xiii.

on

. Bonifay 2004, 87: “la nomenclature proposée en filigrane pour chaque grande catégorie de mobilier, [...] n'est pas une classification mais un simple index destiné à faciliter la liaison entre le texte et les figures." Waagé 1948. Hayes 1972.

Oo

UO

2

74

P. Bes and J. Poblome

economy." After all, the wide dispersion of a restricted number of red-slipped tablewares with generally well-known provenances and typo-chronological frameworks cannot but reflect something of a connected socio-economic and socio-cultural Mediterranean.!! In light of the publication tradition of Roman tablewares developed during the 20th c., the time was deemed ripe for an evaluation of its strengths and weaknesses, as well as implicit scholarly paradigms. The general aims of the ICRATES Project are to study, analyse and explain patterns of production, distribution and consumption of Roman material culture. The first phase?? of the project concentrated on the distribution and consumption of tablewares in the Roman East by collecting the majority of the published record for the E Mediterranean on a sherd-by-sherd basis in a relational database-system. To this end, aspects such as fabric identification and recognition (including provenance) and a typo-chronological framework needed to be as accurate as possible. The publication tradition was thus deconstructed to yield a relational database that now includes over 25,000 records, with data collection and processing ongoing. The collected data contains most published excavation sites and survey regions and derives from all the modern countries of the E Mediterranean (Greece, Turkey, Syria, Lebanon,

Israel, Jordan, Egypt and Libya). For processing and visualizing, the method devised by E. Fentress and P. Perkins?? is used;

despite some methodological drawbacks, it still serves our purposes.!^ The results were laid out in a Ph.D. dissertation, A geographical and chronological study of the distribution and consumption of tablewares in the Roman East, submitted by one of us to the Katholieke Universiteit Leuven in 2007. As categories of Roman tablewares are thought to reflect the direction and intensity of economic exchange, studying, analyzing and interpreting the variation (notably the geographical, chronological and quantitative changes) in these categories aimed to reconstruct patterns of exchange. More importantly, it focused on an understanding of the factors that shaped, influenced and altered the system and nature of economic exchange, as far as the

collected evidence could be taken. The second leg of the ICRATES Project, which began in January 2009, moves within and beyond the ceramic dimension by focusing on the conceptual, emic context of the typo-chronological evolution of tablewares in the Roman East, investigating which factors directed the evolution of tableware. The ICRATES database holds a substantial number of entries of ARS ware, a category of tableware that was widely dispersed in the East from the 3rd but especially from the 4th until the 7th c. All the ARS ware entered into the database thus far is classified (directly or indirectly) according to J. W. Hayes' typo-chronological framework. However, in light of Bonifay's study in which the date ranges of a number of Hayes' Forms are revised, we have an opportunity to make a comparison between the graph that was based on Hayes' classification, on the one hand,

and

a combinationP

of Hayes'

and Bonifay's classifications, on the other.

Although the differences between the two graphs are not (except in one part) dramatic, it seems worthwhile to pay some attention to them as it touches upon certain issues of methodology, as well as aspects of Late Roman economic history.

10

11 12

E.g., Tomber 1993 and 2004; McCormick 2002.

Horden and Purcell 2000. This research was supported by the Fund for Scientific Research-Flanders (FWO-Research Project G.0152.04). The ceramic focus was on the 4 main Eastern sigillatas, Italian sigillata, Candarlı Ware, and the three major Late Roman Red Slip Wares (ARS, CRS, and PRS). In addition, a wide variety of other categories of tablewares is collected in the database: e.g., Italian thin-walled wares, Gaulish

sigillata, (early imperial) lead-glazed wares, Koan/Knidian bowls/cups, Egyptian and Tripolitanian Red Slip Wares, and so on. 13 14

Fentress and Perkins 1988; Fentress et al. 2004. Fora first survey of the aims, the methodology, and some preliminary results, see Bes and Poblome 2006.

15

Bonifay did include some forms of the 1st to 4th c. Although quantitatively these do not always appear to have played an important róle in the E Mediterranean, their dispersion can be used as a tool in an attempt to reconstruct (major) lines of exchange. Bonifay did not redate all post-4th c. forms.

African Red Slip ware on the move: the effects for the Roman

East 75

African Red Slip Ware ARS ware is one of the better-known red-slipped tablewares of the Mediterranean. Despite its central Mediterranean origin, it is found in quantity at eastern sites, not only in large urban centres of the littoral but at interior regions, including sites that seem to have been of non-urban status.!® Despite important research having been carried out and certain exceptions notwithstanding, the

general

situation

for the study

of tableware

in the

E Mediterranean

is far less

developed than for ARS ware. As also seems to be the case with Italian Sigillata,!® the picture for ARS ware appears to grow ever more complex in light of new excavations and surveys, archaeometrical analyses, and other evidence. This growing complexity, however, enables scholars to consider the regional manufacturing infrastructure within the context of the agricultural and artisanal development of Africa Proconsularis/Byzacena, which offers, in theory, a better understanding of the dispersion of ARS ware within a diachronic socio-economic and socio-cultural framework.!? Moving beyond stating that an object travelled from A and ended up at B in order to reconstruct the spheres of producers, distributors, merchants and consumers, and the ways in which these parties interacted, such aspects should contribute to a better understanding of the workings of the Roman economy (or economies?). ARS

ware in the ICRATES

database

A total of 3,660 individual entries of ARS

ware has been collected in the database.

Most

entries could be identified by form, yet a fair number are (also) identified by stamped motif.^? This figure, of course, is dwarfed by what the output of the manufactories of Proconsularis/ Byzacena and the resulting distribution must have been. Yet although this realization is essential, the broad geographical and chronological scope of the ICRATES Project can still offer good possibilities. Our approach plays down the impact of individual publications (a selective catalogue, for example, or one that is chronologically limited). The collected evidence from the 4th to the 7th c. reveals patterns on both regional and macro-regional scales and the evolution of exchange patterns and the economy, and it offers tentative clues for formulating modified or alternative hypotheses. The potential for interpretation derives not from the database alone, which is basically a static collection of data, but from a comparison of quantitative, geographical and chronological developments of the different categories of tablewares. In this paper, however, the focus will solely be on ARS ware! The trends to be discussed are based on the evidence as it was used by P. Bes in his associated doctoral research, in addition to evidence collected since.?? New chronologies: shaking hands? This is the place neither to present a full overview of the published record that is included

in the database?” nor to discuss the evidence in detail by region. Instead, the data will be taken as a whole with, where appropriate, reference to observations

made

in the

original

research.

16

E.g., Orssaud 1980, Rossiter and Freed 1991, Harper 1995, Slane 1997, Rautman 2003, and Lund 2006.

17

E.g., Sagalassos Red Slip Ware: see Poblome 1999; id. et al. 1998.

18

E.g., Poblome et al. 2004; Oxé, Comfort and Kenrick 2000.

19

Unfortunately, few publications that pertain to the E Mediterranean have made use of the A-E fabric classification. For a synthesis and discussion of the regional production framework in Proconsularis/

20

When a vessel is identified by form and stamped motif, it is not counted twice. Both parameters are included in the ICRATES database. For a discussion and interpretation of the distribution of the three major red slip wares of the Late Roman period in the E Mediterranean, see Bes and Poblome 2008. This newly entered data was relatively small in quantity and did not alter the original conclusions. In addition to three unpublished collections of pottery (from Greece and Jordan), some 320 publications have been entered at the time of writing.

Byzacena, see Bonifay 2004, 45-65.

21 22 23

FIG. 1. THE DISTRIBUTION OF ARS PER 15-YEAR INTERVAL WITH AND WITHOUT CYRENAICA IN ABSOLUTE NUMBERS AND BASED ON HAYES’ TYPO-CHRONOLOGY FOR THE EASTERN MEDITERRANEAN BETWEEN A.D. 200 AND 700

250 — |

225

- W-

ARSW with Cyrenaica | ARSW without Cyrenaica

200

dm \

oauio[qog '[ pue seg ‘4

ii.

X 1

100

2

MS

M

75 =

50

V

à ,

4

zx

—.-* X d

T

4e

N

=”

:

4 P

25

a

|

a

sm 0)

|

uL.

mn, S

zen

AD

|

1

AD

1

AD

if d

T

T

AD

f

AU

1

AD

1

i

1

AD

|

)

= AD

cd

|

nom

AD.

:

|

|

AL

;

i

|

AD

I

AD

1}

I

[

AD

1

4

AD.

T

|

AD

|

|

. AD

195-210 225-240 255-270 285-300 315-330 345-360 375-390 405-420 435-450 465-480 495-510 525-540 555-570 585-600 615-630 645-660 675-690

| i

AD

I. RAW DATA.

ARS with Cyrenaica ARS without Cyrenaica

195-210 109.89 6.8

210-225 146.23 7.4

225-240 171.28 27.76

240-255 180.24 37.33

255-270 155.79 37.49

270-285 106.4 37.91

285-300 87.44 43.75

300-315 88.79 46.62

ARS with Cyrenaica ARS without Cyrenaica

AD 315-330 80.22 49.9

330-345 59.68 52.44

345-360 89.79 82.20

360-375 152.49 142.72

375-390 154.04 144.99

390-405 140.08 131.11

405-420 113.54 104.72

420-435 100.22 94.17

AD 435-450

450-465 111.49 98.38

465-480 92.83 70.12

480-495 85.56 54.25

495-510 101.91 65.31

510-525 154.04 110.98

525-540 158.07 122.04

540-555 141.45 114.13

570-585 190.65 159.27

585-600 207.34 175.05

600-615 227.8 175.9

615-630 185.82 137.73

630-645 645-660 131.52 7 101.3 89.64 72.22

ARS with Cyrenaica ARS without Cyrenaica

107.86 99.95

-X

AD 555-570 ARS with Cyrenaica 155.2 ARS without Cyrenaica 126.3

| ARS with Cyrenaica ARS without Cyrenaica

AD 675-690 14.11 11.56

690-705 9.13 7.48

660-675 15.11 11.76

LL ]S€'] UEWONM OY} 10] $Pajjo Jy} :2A0UI OU} UO aTeEM dIIS pay ue»ugy

FIGURE

78 The

P. Bes and J. Poblome collected

evidence,

which

excludes

entries

west

of a line

drawn

from

Berenice

to Niko-

polis?! and is based upon Hayes’ classification, is represented in fig. 1. Such graphs (not commonly used by most other scholars?) can capture major diachronic developments. Although the following paragraphs will rely heavily on ARS ware as a tool capable of explaining patterns of exchange, we need to add some perspective. Economically, Proconsularis/ Byzacena had other 'cards up its sleeve' that were of much greater value to the area's economy proper. Olive oil, grain and, to a lesser extent, wine are regarded as its surplus products. Yet if the 'secondary cargo hypothesis' is thought to be a structural element of Roman Mediterranean

trade," the distribution of ARS ware could be considered as proxy evidence, mirroring the lines along which contacts took place diachronically. In reality, however, intra-and inter-regional trade will have been extremely complex structures. Before we move to a comparison between this graph and the one that shows the evidence using Hayes' and Bonifay's classifications combined, we may point out some aspects which underscore the most important observations. First, the strong peak for the mid-3rd c. is largely

accounted for by the evidence from Berenice.?? By omitting the evidence from Berenice, a more ‘realistic’ picture is created for the East (Berenice might not be seen as truly ‘eastern’ geographically, but perhaps rather as an intermediary between West and East — see the uninterrupted line in fig. 1). Evidently, Berenice received ARS ware earlier and in much greater quantities than anywhere else in the East. The reason for adding the evidence from Berenice is to draw attention to the following matter. It shows that an interpretation may be heavily influenced by a single publication regardless of the historical, political and economic significance of the site in question. It can be argued that the typo-chronological framework used, and perhaps even the methodology used, can equally shape our thoughts in ways that need not always be correct. The strong peak on the graph is followed by a strong decrease in the quantity of ARS ware — also accounted for by the evidence from Berenice.? Perhaps Berenice (and possibly Cyrenaica as a whole) was the easternmost limit of the core dispersion of ARS ware until after the earlier part of the 3rd c., when this limit shifted to incorporate the East more generally. After all, some ARS ware was already present in the East earlier in the 3rd c. and before (e.g., at Dura Europos).?! Perhaps Cyrenaica was able to market an acceptable surrogate for ARS ware because outgoing trade from Africa Proconsularis/Byzacena (a trade that included ARS ware) became more focused elsewhere.?? If, from now on, the dotted line is considered (which means excluding Berenice), we find a

strong surge of ARS ware in the East around the mid-4th c., which is traditionally associated

with the founding of Constantinople in 330.? Although the actual 'founding' of Constantinople probably did not contribute to the change in the distribution of ARS ware in direct terms, it does designate a major shift in the orientation of economic and other patterns that had developed during the first two to three centuries A.D. With the empire's capital being located more centrally and at a great distance from Rome came a new pivotal point in administrative, political, military, diplomatic, and economic terms. It is in this context that an explanation for a reorientation in the distribution of ARS ware during the second half of the 4th c. needs to be placed. The phenomenon, however, seems to be short-lived. Toward the end of the 4th c. quan24

Some 1,200 entries of the c.25,000 in the database derive from sites in the W Mediterranean (predominantly Carthage).

25 26 27

J. Lund, however, makes ample use of such graphs: e.g., Lund 1995 and 2006; see also Malfitana 2002. Bonifay 2003, 2004, and 2005. Fulford 1987; Parker 1992.

28

Kenrick 1985.

29

Note, however,

30 31 32 33

that at Berenice "finds attributable to the fourth and early fifth centuries were very

limited and not a single coin of the fifth century was found": Kenrick 1985, 3. Reynolds 1995, 108-9. At Ostia, N African pottery "rose dramatically" between the early 2nd and early 3rd c., from c.20 to 85%, a trend that continued throughout the 4th c. Cox 1949. Tripolitanian Red Slip Ware may have partly fulfilled this röle in Late Roman times: Kenrick 1985, 387. Cameron 2005.

African Red Slip ware on the move: the effects for the Roman East 79 tities were dropping again,“ a decrease which continued until the late Sth c., though with differing intensities. The Vandal incursions of the 430s (Carthage fell in 439) are quite often mentioned as the factor that caused a decline in ARS ware moving around the (eastern) Mediterranean,” yet it seems as if this decrease began earlier by several decades, which suggests that the Vandals need not have been the sole factor. Perhaps they influenced the manufacture of ARS ware only slightly. Yet if a shrinking quantity of ARS ware flowed into the E Mediterranean (the distributional aspect), what about the actual production output? Another point to consider is that the manufacture of Phocaean and Cypriot Red Slip Wares began by the late 4th c., according to present knowledge.*° Perhaps these partly replaced ARS ware on eastern markets, while the peak in the second half of the 4th c. was a unique phenomenon reflecting the new geo-political situation in which cultural and economic exchange could flourish. Factors influencing the dispersion of a category of Roman tableware could be functioning at different levels. From the late 5th right into the early 7th c. we find a clear and strong upsurge of ARS ware flowing into the E Mediterranean — much more pronounced than the peak which is (tentatively) linked to the effects of the foundation of Constantinople. This, the strongest peak, may have been a side-effect of Justinian's reconquest which brought certain areas (such as Proconsularis/Byzacena) within the sphere of control and under the umbrella of taxation (i.e., taxa-

tion in kind, for which N African grain is a possible candidate?/). This would mirror the shifting political-economic situation in the Mediterranean; but there is a dilemma, for the graph leaves no doubt that several decades prior to Justinian's reconquest (which returned Proconsularis/Byzacena to the empire in 533-534) ARS ware was on the rise again. Perhaps other factors were at play that promoted the renewed dispersion of ARS ware, with Justinian’s reconquest simply giving it another push in the ‘right’ direction. Finally, in the course of the 7th c. the dispersion of ARS ware steadily dropped, although production seems to have continued into the 7th, possibly even into the early 8th c.?? Because of the long tradition and the different methodological approaches towards pottery processing and study, the collected evidence is liable to be skewed, but it seems unlikely that the pattern attested is to be explained solely by such irregularities in the published evidence. While the published record is uneven for certain areas of the East, the research admittedly did suggest that southern Greece, for example, received a broader range of forms, including some rare(r) ones. During

the second and

third peaks, as well as in the intervening period, a rela-

tively larger share of ARS ware flowed into the area. Probably this was the result of southern Greece’s geographical location between Proconsularis/Byzacena and Constantinople, which

allowed it to profit from these economic and other ties, even in ‘bad’ times.?? It may be useful also to illustrate this regional evidence, again without Cyrenaica (fig. 2). Here the major trends of fig. 1 return. Again, one publication dominates the graph, in this case

the recent overview for Cilicia by L. Zoroÿlu;# in addition, the quantities for a number of regions may be insufficient to compare on equal terms. Attention may be drawn to several differences. The first is the period around the mid-4th c., when ARS ware increased overall (see

34 35

Fulford 1984, 113. Hayes 1972, 423; Waage

36

Phocaean Red Slip Ware: Hayes 1972, 323. Cypriot Red Slip Ware: Hayes 1972, 371; but see also Meyza 2007, 43-44, on the possible continuation between the earlier Cypriot Sigillata and the later Cypriot Red Slip Ware.

37

Bonifay 2003, 2004, and 2005.

38

Fora more detailed discussion of the collected evidence, including that for Cypriot and Phocaean Red Slip Wares, see Bes and Poblome 2008. This observation is supported by the collected evidence from the survey of ancient Tanagra, and possibly also that of Koroneia (Boeotia), directed by J. Bintliff (University of Leiden) and Bozidar Slapsak (University of Ljubljana). Both authors of this contribution are involved in processing and studying the Late Hellenistic to Late Roman pottery. For preliminary results see, e.g., Bes, Poblome and

39

1948, 56-57; Abadie-Reynal

1989, 150.

Bintliff 2006; Poblome, Ceulemans and De Craen 2004-5.

40

Zoroglu 2005.

08

FIG. 2. THE DISTRIBUTION OF ARS PER 15-YEAR INTERVAL IN THE EASTERN MEDITERRANEAN BETWEEN A.D. 200 AND 700, IN ABSOLUTE NUMBERS, BROKEN DOWN PER REGION.

E Cyprus

M Levant-Interior South

A

IIl Levant-Coastal South 150 gs

|

A

[]ILevant-Interior North Levant-Coastal North

DER

T$

Aegean-East E

Crete

:

|

^

[] Aegean-North

EC

LR d

eee

be

, |

NW

Pau as

el à

:

Me

See

u... mM

u. NR m

(AMI

25

ee

A"

q

amr

;

mec

nes

zu.

zu

Zune. don

zu

SES

mS

i".

ug

EEE

WM

»"

PASS T

D

=

nee

=

"

es

a

.

"ww

\

"eere ta

À

| A

1

:

ae

=

nn

ates

=

"

:

er

ces

S

ea

=

ne

eJ =

f mm

N

om

"

"e"

xe

e

w

ER

A

a"

"

NNNM

M

u

en

ns

—

2

X

R^ x

=

"d

1

"en m

Q

&

ne

‘on

LAN

©_—

EEE Rs

BE ww

"e.

a

2

ree

"

"

" u.

(o

SE

u.

,

p

ERBEN

"

i

"

NN NN

cers

,

p 7

s t

e

N

jo

ne =

ge

t.

SPP

i" ae

RER BER BER

D x

ern >.

MT

,

"

n'a

IV

zu

nn RE "a"

N

NT

ee

wwNN

n.

d: :

=

SS

:

:

Bey

T e

Qu

:

«i

Ley

a

os

100

50

oS AME

B ^cecan-Southwest

c

7

^

Asia Minor-Cilicia [25M

23

$

nun

.

v

=

EEE 0)

T

T

T

T

et

T

T

T

j

T

T

al

T

AD AD. 2 AD AD: AD © AD AD AD AD AD^ AD. AD ADAD AD 195-210 225-240 255-270 285-300 315-330 345-360 375-390 405-420 435-450 465-480 495-510 525-540 555-570 585-600 615-630 645-660 675-690

AD 195-210 Aegean - north Aegean - southwest Crete Aegean - east

0.5 0.15 1.76 0.12

Asia Minor - Cilicia

1.32

Levant - coastal north

1.52

210-225

225-240

0.76 0.2 1.75 0.36 0.85 2.3

2.19 6.1 2.87 2.46 2.7

0.16 0.87

255-270 2.13 9.24 3.8 3.11 3.59 4.49 1.96 2.22 3.31 1.71

270-285 2.03 9.72 4.04 3.11 3.55 3.77 1.96 2.62 3.4 1.64

285-300 2.25 10.86 4.95 3.23 4.11

4.08 1.32 1.5 2.34 1.22

240-255 2.53 9.07 3.73 3.11 3.59 4.68 1.96 2.22 3.31 1.71

390-405 7.35 28.19 14.86 2.27 36.21 9.46 9.88 6.57 8.85 4.92

405-420 9 21.87 11.94 2.41 33.56 9.22 2.82 3.04 6.46 2.04

420-435 6.18 17.13 8.34 2.25 35 7.27 4.32 3.42 6.48 1.66

435-450 5.22 19.04 8.49 2.96 36.02 7.04 5.1 4.05 7.67 1.8

570-585

585-600

6.57 30.18 11.3 6.76 51.21 9.1 16.62 3.89 11.41 5.51

6.36 29.33 14.34 11.27 59.34 7.85 15.46 4.2 10.95 9.02

600-615 5.24 24.03 13.18 10.84 64.97 4.24 13.75 6.37 15.69 10.58

615-630 4.3 15.43 11.31 10.33 49.96 3.63 9.45 5.14 13.45 9.31

Levant - interior north

Levant - coastal south

0.13

Levant - interior south

0.1

Cyprus

0.85

AD 375-390 Aegean - north 6.49 Aegean-southwest 31.8 Crete 16.28 Aegean - east 2.2 Asia Minor - Cilicia

Levant - coastal north

37.63

9.87

Levant - interior north 13.41

Levant - coastal south Levant - interior south Cyprus

8.34 9.95 6.37

AD 555-570 Aegean - north Aegean-southwest Crete Aegean - east

4.23 25.89 6.34 1.38

Asia Minor - Cilicia

45.77

Levant - coastal north

7.63

Levant - interior north 14.07 Levant - coastal south

3.04

Levant - interior south

9.75

Cyprus

3.81

DATA. 315-330 2.73 13.19 7.49 1.65 6.03 3.57 1.58 5.59 4.05 2.05

330-345 2.83 13.74 8.65 1.04 8.27 4.07 0.99 5.54 3.33 2.16

345-360 3.83 21.41 11.2 1.38 10

360-375 5.78 30.32 16.16 2.2 36.11

3.93 1.96 4.62 3.97 1.85

300-315 2.37 12.04 5.62 2.15 4.66 3.48 1.9 5.31 4.15 2.17

6.25 7.81 8.34 5.22 4.5

8.96 13.26 10.44 9.71 6.93

450-465 4.34 18.52 7.87 2.33 32.6 7.59 7 4.39 8.87 2.78

465-480 3.48 13.7 5.1 1.76 16.04 6.38 6.12 4.88 8 2.22

480-495 2.26 9.33 2.79 1.14 8.55 7.25 5.72 5.47 6.82 1.91

495-510 4.67 8.04 3.49 0.97 14.63 8.89 5.86 5.7 7.01 1.85

510-525 7.79 16.93 8.3 1.02 35.68 10.16 6.25 6.6 7.87 4.61

525-540 3.59 21.42 8.06 0.83 48.71 10.25 5.96 3.59 10.13 4.79

540-555 2.97 23.57 5.1 1.04 49.55 6.88 7.88 1.84 7.88 3.33

630-645 2.21 9.46 9.82 9.52 28.51 2.75 3.13 3.74 9.56 7.55

645-660 1.96 8.37 8.42 8.92 24.31 2.1 2.24 2.39 4.26 6.95

660-675 1.02 0.61 0.34 0.34 6.8 0.27 0.85 0.34 0.17 1.02

675-690 1.02 0.51 0.34 0.34 6.8 0.17 0.85 0.34 0.17 1.02

690-705 0.66 0.33 0.22 0.22 4.4 0.11 0.55 0.22 0.11 0.66

18 jseq UEWON JU} I0J SPoFJa ay} ‘AOÛ ay} uo aleM dis pay ueougy

FIGURE 2. RAW

82

P. Bes and J. Poblome

fig. 1). On a regional basis, however, a distinction can be made between the Aegean-Southwest geographical classification (i.e. southern and central Greece) and the other regions, the latter displaying virtually no difference compared to the preceding decades. Second, in the period of the later 4th and early 5th c., Cyprus presents a slight increase, whereas the quantities of ARS ware for all other regions decreases; however, we may need to keep in mind the relatively small quantities available for Cyprus. Thirdly, it can be observed that, in addition to the higher quantities published, the Aegean-Southwest displays an increase around the mid-6th c., whereas most, if not all, other regions remain the same or decrease slightly. We need to remain on the lookout for specific publications that may skew the data-set. Nevertheless, taken as a whole, the observed differences do appear significant enough not to be ignored. The third graph (fig. 3), which shows the evidence both including and excluding Cyrenaica, is based on a combination of the typo-chronological frameworks of Hayes and Bonifay. Hayes' classification remains the basis, yet in cases in which Bonifay suggested new date ranges his dates supplant those of Hayes (this is facilitated by the fact that Bonifay largely retained Hayes' classification). It does not appear necessary to discuss fig. 3 at length since most of the general trends seen in fig. 1 are repeated here. Figure 4 integrates figs. 1 and 3, namely those lines presenting the developments of ARS ware (excluding Cyrenaica) based on Hayes' classification, on the one hand, and those based on the combined classifications of Hayes and Bonifay, on the other.*! Two

important

differences between

figs. 4 and 3, however,

may

be noted.

First, the overall

course of the graph is more fickle and the highs and lows are less pronounced. Second, the period of the second half of the 5th and the early 6th c. is worth remarking. The first difference concerns the generally more fickle course of the graph that combines Hayes and Bonifay (the dotted line in fig. 4): it shows several highs and lows that are much less pronounced compared to the evidence based solely on Hayes’ classification (the uninterrupted line in fig. 4). This is the case for the later 4th c., and especially the later 6th and early 7th c. But the general trend remains: an increase in the quantity of ARS ware. This is also the case (the second important difference) for the second half of the 5th and early 6th c. Here the two lines clearly diverge. This diversion is significant; more importantly, the two lines do not show

the same develop-

ment. If the Hayes line is followed, it shows a considerable decrease, Bonifay line does not and (read with caution) shows a slight increase.*?

whilst the Hayes-

What caused these differences? This is an important consideration both for methodological and interpretative purposes. The answer lies simply in Bonifay's redating of a number of Hayes’ forms. Bonifay uses much evidence from Tunisia itself, as well as from other sites in the W Mediterranean, the archaeological and stratigraphical details of which do not figure in his Etudes in detail. The element of redating can be clearly seen in the first part of the graph up to the later 4th c., where both lines virtually run parallel. The reason for this is that Hayes' forms of the 1st to 4th c. were either not redated by Bonifay or, when they were, they are quantitatively negligible. For example, Hayes Form 50 (variants A and B) is commonly found at eastern sites, yet Bonifay did not propose a new date (nor type). Other Hayes' forms were redated, such as 3 (Sigillée Types 1 and 2),* 8 (Sigillée Type 3), 14 (Sigillée Types 5 and

7),5 27 (Sigillée Type 13),47 31 (Sigillée Type 11),48 and 32 (Sigillée Type 25),* but these are represented and recorded only in very small numbers and hence only have a modest impact on our data-set. 41

Figures 3 and 4 are based on the same data-set.

42 413

Fulford 1984, 114. Bonifay proposed new date ranges for Hayes’ Forms 8A-B, 14A-C, 15, 16, 17B, 18, 26, 27, 31, 32, 58A-B, 61A-B, 67, 74, 79, 81, 83, 88, 90A, 91A-D, 99A-C, 104A-C, 105, 108, and 109.

44

Bonifay 2004, 154 and 156, fig. 84.

45

Ibid. 156, fig. 84.

46

Ibid. 157-59, fig. 85.

47 48 49

Ibid. 158-59 and 162, fig. 85. Ibid. 157-59, fig. 85. Ibid. 162-63, fig. 88.

African Red Slip ware on the move: the effects for the Roman East s3 However, if the later 4th to 7th c. 1s observed in detail, the date ranges of a number of Hayes' forms that occur regularlv at eastern sites were modified by Bonifay. More important, it appears, is the recognition that certain forms (Bonifay's Sigillée Types) were subdivided. For example, the very common Hayes Form 67 (Sigillée Type 41), dated by Hayes between c.360 and 470, is thought to have occurred between c.350 and 500, an extension of roughly 3-4 dec-

ades. Similarly, Hayes Form 91 (variants A-D) was originally dated from c.425 to c.650. Bonifay regards this quite common form (Sigillée Types 49-50, 52 and 54) as attributable to the 5th to 7th c., thus extending it by some 60-70 years?! Another example is Hayes Form 99 (with variants A-C), one of the most common forms of ARS ware in the E Mediterranean and originally datable between c.510 and 620. Bonifay considers a much lengthier period for his Sigillée Type 55A-C, from the late 5th to the late 7th c. — an extension of the date range by about a

century.?? Also, Hayes Form 104 (with variants A-C), originally dated between c.530 and 625, is now thought to have a broader date range: c.475 to 675 (Sigillée Type 56A-C).°° Finally, Hayes Form 105, another very common form dated between c.580 and 660, was subdivided by

Bonifay and given a longer date range (Sigillée Type 57A-C, c.575-700).°4 The technicalities of the date ranges aside, the suggested longer ranges for some of the most common forms or Sigillée Types implies that, by using the method of Fentress and Perkins, their quantities are ‘smeared’ out over broader chronological intervals. This appears to explain the first difference in fig. 4. The second difference, the divergence between the two lines during the second half of the Sth

and

earlier

6th

c., also

appeaxs

to relate

to what

is discussed

in the above

overview

concerning Hayes’ forms and their counterparts in Bonifay. The 5 forms (or Sigillée Types) mentioned above are not chosen without reason: these 5 are the most common forms included in

the database, and it should be noted thatthe (extended) date ranges of Forms 67, 99 and 104? now include the later 5th c. — roughly the period when ARS ware was at its lowest in Hayes'

graph.” Methodologically this resulted in a quantitative shift of ARS ware in the graph towards the second half of the 5th c., thereby downplaying Hayes' typology line.

the prominent decrease in the

Implications and discussion Bonifay's Etudes has wide implications for the way scholars look at the manufacture, dispersion and consumption of (later) Roman pottery from Proconsularis/Byzacena. As illustrated above, the methodological and interpretative implications for the ARS ware evidence as collected within the ICRATES Project are equally significant. To some extent the interpretation of pottery stands or falls with a typo-chronological framework. Thus broader interpretations may need to be reconsidered following revision of a particular typo-chronological framework (or part thereof). Given the considerable differences between the two lines in fig. 4, are we to rethink the economic history of Late Roman Proconsularis/Byzacena based on the collected evidence alone? Are we to move from full crisis to moderate quantitative output and distribution? As ARS ware is only one part of the picture, the answer is surely no. Yet the differences noted do offer some points for discussion. First of all, the method of Fentress and Perkins used here, despite providing a general framework for our approach, has certain drawbacks. First, unidentified entries (i.e., those only identified by fabric and not by type or stamp) are omitted from the graph. A

50 .51 52 53 54 55 56

Ibid. 171-73, fig. 92. Ibid. 177-79 and 181, fig. 95. Ibid. 180-81, fig. 96. Ibid. 181-83, fig. 97. Ibid. 183-85, fig. 98. This goes for the early variants A of both Forms 99 and 104, and in case a fragment could only be identified to the general form. Fulford

(1984,

109-10) already noted that Forms

suggested in Late Roman pottery.

104 and

105, amongst

others, started earlier than

FIG. 3. THE DISTRIBUTION OF ARS PER

15-YEAR INTERVAL

rs

WITH AND WITHOUT CYRENAICA

IN ABSOLUTE NUMBERS AND BASED ON A COMBINATION OF HAYES' AND BONIFAY'S TYPO-CHRONOLOGIES, FOR THE EASTERN MEDITERRANEAN BETWEEN A.D. 200 AND 700

180

|

| r\

160

—$—

ARSW

Hayes & Bonifay (with Cyrenaica)

- #-

ARSW

Hayes & Bonifay (without Cyrenaica)

hm.

140

m

120

JJ (D

T M

100

un

CU

m

A.

|

+

xe

©

80

3

E

60

(D

40

Gel

LES

„ ne"

"M

, u

.N

I i]

iD

AD.

1

AD.

AD

AD

r

AD

AD

AD:

T

AD

An

AD

AD

195-210 225-240 255-270 285-300 315-330 345-360 375-390 405-420 435-450 465-480 495-510 525-540

AD

"AD

DD:

AD

555-570 585-600 615-630 645-660 675-690

195-210 95.55 6.35

210-225 133.50 8.02

225-240 167.16 29.15

240-255 171.97 38.49

255-270 139.32 37.95

270-285 139.90 39.45

285-300 147.27

300-315 91.56

45.61

46.58

AD 315-330

330-345

345-360

360-375 129.38

375-390 131.30

390-405 125.50

405-420 114.99

420-435 107.33

119.48

122.12

116.61

106.55

ARS Hayes & Bonifay (with Cyrenaica)

78.89

49.57

104.76

ARS Hayes & Bonifay (without Cyrenaica)

46.24

41.42

95.23

AD 435-450 450-465

465-480

ARS Hayes & Bonifay (with Cyrenaica) ARS Hayes & Bonifay (without Cyrenaica)

ARS Hayes & Bonifay (with Cyrenaica) ARS Hayes & Bonifay (without Cyrenaica)

ARS Hayes & Bonifay (with Cyrenaica) ARS Hayes & Bonifay (without Cyrenaica)

120.02 107.29

121.53 105.87

136.57 110.73

AD 555-570 570-585

585-600

113.32 97.89 AD 675-690 86.62 70.50

131.11 111.11 690-705 55.95 45.25

145.06 121.94

98.38

480-495

495-510

510-525

525-540

540-555

156.93 122.63

136.84 101.60

124.47 88.82

146.77 119.49

121.23 105.31

600-615

615-630

630-645

645-660

660-675

156.17

136.97

116.20

123.50

111.30

97.28

16431 168.38 131.57 , 128.88

8 JSC ueWON OU} 10J SJD9JJ9 JY} :9AOÛ aU} UO aleM di[S pay ueougy

AD ARS Hayes & Bonifay (with Cyrenaica) ARS Hayes & Bonifay (without Cyrenaica)

DATA

c

FIGURE 3. RAW

FIG. 4. THE DISTRIBUTION OF ARS PER 15-YEAR INTERVAL WITHOUT CYRENAICA IN ABSOLUTE NUMBERS AND BASED ON BOTH HAYES’ TYPO-CHRONOLOGY AND A COMBINATION OF HAYES' AND BONIFAY'S TYPO-CHRONOLOGIES, FOR THE EASTERN MEDITERRANEAN BETWEEN A.D. 200 AND 700

D

180

—tr— ARSW Hayes (without Cyrenaica) =

160

"-

140

ARSW

Hayes & Bonifay

:

(without

Cyrenaica)

:

TITRE

A

120

J UJ

Rx LA

, *

D

=)

100

Sr

80

:

1

HL —

"

ge C o

=ae

60 ,

-m

40

20

()

e

=

m

/

|

i

|

À

|

|

T

AD AD AD AD AD AD AD AD AD AD AD AD AD AD AD AD AD 195-210 225-240 255-270 285-300 315-330 345-560 375-390 405-420 435-450 465-480 495-510 525-540 555-570 585-600 615-630 645-660 675-690

195-210 6.80 6.35

210-225 7.40 8.02

225-240 27.76 29.15

240-255 37.33 38.49

255-270 37.49 37.95

270-285 37.91 39.45

285-300 43.75 45.61

300-315 46.62 46.58

330-345 52.44 41.42

345-360 82.29 95.23

360-375 142.72

390-405 131.11 116.61

420-435 94.17

119.48

375-390 144.99 122.12

405-420 104.72

ARS Hayes & Bonifay (without Cyrenaica)

AD 315-330 49.90 46.24

ARS Hayes without Cyrenaica ARS Hayes & Bonifay (without Cyrenaica)

AD 435-450 450-465 99.95 98.38 105.87 107.29

465-480 70.12 110.73

480-495 54.25 122.63

495-510 65.31 101.60

510-525 110.98 88.82

525-540 122.04 119.49

540-555 114.13 105.31

ARS Hayes without Cyrenaica ARS Hayes & Bonifay (without Cyrenaica)

AD 555-570 126.30 97.89

570-585 159.27 111.11

585-600 175.05 121.94

600-615 615-630 175.90 137.73 131.57, 128.88

630-645 89.64 123.50

645-660 72.22 111.30

660-675 11.76 97.28

ARS Hayes without Cyrenaica ARS Hayes & Bonifay (without Cyrenaica)

AD 675-690 11.56 70.50

690-705 7.48 45.25

AD ARS Hayes without Cyrenaica

ARS Hayes & Bonifay (without Cyrenaica)

ARS Hayes without Cyrenaica

106.55

98.38

/8 JSCJ ueuloy JU} IOJ S129JJ9 JU} :9AOUI aU} uo 318M dIIS pay ueoujgy

FIGURE 4. RAW DATA

88

P. Bes and J. Poblome

methodology has been developed to include such data for the Boeotia Survey, which offers opportunities to include all collected data.°” Second, it does not take into account the fact that vessels may have been more popular within their date range. Although knowledge of this seems to be slight, if not largely absent, for tablewares in the Roman East, based on a general lack of well-defined, possibly closed deposits, we consider the production and distribution rate of a vessel during its date range to have been constant. Further, it is possible that the dispersion of ARS ware was dissimilar in different parts of the East. Contextual analysis of welldefined deposits could contribute here,’ as does seriation?? Well-defined and quantified deposits could also contribute to economic interpretations regarding the dispersion of pottery. What was Bonifay Etudes really aimed at as far as the chronology of ARS ware is concerned? Originally, Hayes, as well as his predecessors such as Waagé, tackled questions about typology and chronology (and provenance) using evidence from the “consumer end’ of the line: Une des particularités de la céramique africaine est que sa typologie et son cadre chronologique ont été établis hors d'Afrique, sur des sites consommateurs souvent très éloignés des sites producteurs. Part of the evidence also came from excavations that were conducted in different (scientific or

stratigraphic) frameworks. In the words of Hayes: It must be stressed at this point that the conclusions presented in this volume are of a provisional nature, and will no doubt require modifications as more evidence of a precise nature becomes

available. [...] The forthcoming publications mentioned above, along with those of a number of excavations currently under way, should go far towards solving the many dating-problems which exist at present. One of the chief aims of this book is to provide a typological framework for these

future studies. Bonifay’s work is clearly one of those “future studies”, and one that focuses on the heartland of ARS ware. Here lies the first important yet chiefly theoretical consideration. There should be no doubt about the merit of Bonifay's approach (not a classification but "un simple index" of ARS ware) or about the work in general. However, could the chronology of manufacture and the chronology of dispersion and consumption not have been (substantially) different? In other words, some time may have passed before vessels that were produced were actually sold and traded overseas; further, this may also have played out differently in differing regions. Though often regarded as a seasonal activity, pottery manufacture need not even have been a job carried out at an appointed time or with a similar output each year; these goods were not perishable, and could be stored until they could be ‘marketed’. It does not necessarily imply direct supply on brisk demand. The exchange sequence itself provided further opportunities for vessels to be held in stock; and as far as the consumer is concerned, use and discard behaviour

should be considered with regard to the recent model offered by J. T. Pefia and its ramifications

for the interpretation of the archaeological (pottery) record.9? These and other issues need to be considered in future research. For the economic

interpretation of ARS

by a political-historical framework

ware, the collected data was

(tentatively)

framed

focused, admittedly, on the major events and develop-

ments such as the annona between Proconsularis/Byzacena and Rome, the founding of Constanti-

nople, the Vandal occupation of Proconsularis/Byzacena, and Justinian's reconquest. Probably these events shaped only some of the lines of exchange, though archaeological evidence from Monte Testaccio™ and Ostia echo the growing importance to Rome of Proconsularis/Byzacena (and Tripolitania). In turn, Rome as well as other major harbours acted as centres of redistribution. The axis Proconsularis/Byzacena—Rome/Ostia—Alexandria may well have been one of the paths along which ARS ware was dispersed in the E Mediterranean, later with Constan57 58

Poblome, Ceulemans and De Craen 2004-5. — Putzeys 2007. Deposits only or mainly dated by tablewares are less suited for such purposes.

59 60

Orton, Tyers and Vince 1993, 189-96; Groenen and Poblome 2003. Bonifay 2004, 87.

61

Hayes 1972, 2.

62 63 64

Bonifay 2004, 1-2 and 87. Pefa 2007. Ibid. 300-6.

African Red Slip ware on the move: the effects for the Roman

East 39

tinople as a major pulling factor since, from 330 onwards, D. Pieri visualizes a major shift by which the Egyptian grain now supplied Constantinople, whilst N African grain was shipped to Rome; but in years of famine or crop failure Rome could call on Egypt, as well as other regions.® Under normal circumstances this division of the grain supply system would be operational.° This has recently been challenged by Bonifay, who does not deny that “dès avant 330, le blé d'Égypte a été détourné de Rome pour alimenter la future ville de Constan-

tinople”.° In explaining the substantial decrease of ARS ware in the East following the Vandal conquest, he conjectures whether ARS ware may not have supply lines of African grain. This premise is based on the scarcity products besides ARS ware during late antiquity (except for the were not uncommon in the East between the late 5th and 7th c.). The this hypothesis is that Proconsularis/Byzacena served next to Egypt erable quantity of grain up to the Vandal conquest, and probably also

been shipped along the of other African ceramic so-called spatheia, which important consequence of as a supplier of a considat later times.

The material needs to be looked at from different perspectives, for the context of manufacture (as well as that of dispersion) is crucial in understanding the regional framework of production, but it is also important to incorporate regional trajectories of the many eastern regions. In this regard, evidence is available concerning the growing of agricultural crops, diachronic patterns of rural and urban settlement, and the artisanal framework of the production of ARS ware, amphorae, cooking wares and lamps — all the object of considerable attention from Bonifay.9? With regards to dispersion, it is clear that improved methodologies, such as fabric identification and full quantification, have much to contribute. Furthermore, we need to accept the complexity of intra- and interzregional exchange patterns and the fact that it may be possible to reconstruct and explain them only up to a certain level. Multi-pronged efforts that move beyond the traditional scope of archaeology can also contribute to understanding the economic position of Proconsularis/Byzacena in late antiquity. The ICRATES Project provides other ways of looking at the meaning of Roman tablewares, but it cannot be conducted in isolation: Bonifay's Etudes clearly forces us to rethink our data, both in the ways we approach it and in the ways we interpret it. Acknowledgements The research was carried out within the framework of FWO projects G.0421.06 and G.0788.09, the Belgian Programme on Interuniversity Poles of Attraction (P6/22), and the 2007/02 Concerted Action of the Flemish Government. Bibliography Abadie-Reynal, C. 1989. "Céramique et commerce dans le bassin Égéen du IV? au VII? siècle,” in V. Kravari, J. Lefort and C. Morrisson (edd.), Hommes et richesses dans l'Empire byzantin I. IV*-VI* siècle (Paris) 143-

59. Bes, P. M. 2007. A geographical and chronological study of the distribution and consumption of tablewares in the Roman East (Ph.D. diss., K. U. Leuven).

Bes, P. M. and J. Poblome 2006. “The ICRATES platform. A new look at old data. Integrating published evidence on Early Imperial tablewares from Greece," in D. Malfitana, J. Poblome and J. Lund (edd.), Old pottery in a new century: innovating perspectives on Roman pottery studies (CNR-Istituto Beni Archeologici) 141-65. Bes, P. M. and J. Poblome 2008. “(Not) see the wood for the trees? 19,0004 sherds of tablewares and what we can do with them," in S. Biegert (ed.), RCRFActa 40, 505-14.

Bes, P. M., J. Poblome and J. L. Bintliff 2006. “Puzzling over pottery. Thespiae, Tanagra and methodological approaches toward surface ceramics," in D. Malfitana, J. Poblome and J. Lund (edd.), Old pottery in a new

century: innovating perspectives on Roman pottery studies (CNR-Istituto Beni Archeologici) 339-45. Bonifay, M. 2003. "La céramique africaine, un indice du développement économique?" AnTar 11, 113-28. Bonifay, M. 2004. Etudes sur la céramique romaine tardive d'Afrique (BAR 51301, Oxford).

65 66

67 68

Pieri 2005, 148. Rickman 1980, 198-99.

Bonifay 2003, 120. Id. 2003, 2004, and 2005.

90

P. Bes and J. Poblome

Bonifay, M. 2005. "Observations sur la diffusion des céramiques africaines en Méditerrannée orientale durant l'Antiquité tardive," in F. Baratte et al. (edd.), Mélanges Jean-Pierre Sodini (Travaux et Mémoires

15) 565-81. Cameron, A. 2005. "The reign of Constantine, A.D. 306-337," in A. K. Bowman,

P. Garnsey and A. Cameron

(edd.), The Cambridge Ancient History XII. The crisis of empire A.D. 193-337 (Cambridge) 90-109. Fentress, E. and P. Perkins 1988. "Counting African Red Slip Ware," in A. Mastino (ed.), L'Africa romana V (Sassari) 205-14. Fentress, E., 5. Fontana, R. B. Hitchner and P. Perkins 2004. "Accounting for ARSW:

Sicily and Africa," in S. E. Alcock and J. F. Cherry (edd.), Side-by-side survey. studies in the Mediterranean

world (Oxford)

fineware and sites in

Comparative

regional

147-62.

Fulford, M. G. 1984. "The red-slipped wares," in M. G. Fulford and D. P. S. Peacock, Excavations at Carthage: the British Mission 1.2. The. Avenue du Président Habib Bourguiba, Salammbo: the pottery and other ceramic objects from the site (Sheffield) 48-115. Fulford, M. G. 1987. "Economic interdependence among urban communities of the Roman Mediterranean," World Arch 19, 58-75. Fulford, M. G. 1994. “Quantification of the pottery," in M. G. Fulford and D. P. S. Peacock, Excavations at

Carthage. The British Mission IL, 2. The Circular Harbour, north side. The pottery (Oxford) 97-114. Groenen, P. J. F. and J. Poblome 2003. "Constrained Correspondence Analysis for seriation in archaeology applied to Sagalassos ceramic tablewares," in Exploratory data analysis in empirical research. Proc. 25th Conference Gesellschaft für Klassifikation e.V., Munich, 2001 (Heidelberg) 90-97. Harper, R. P. 1995. “The pottery," in id., Upper Zohar. An Early Byzantine fort in Palaestina Tertia (Oxford) 21-33. Hayes, J. W. 1972. Late Roman pottery (London). Hayes, J. W. 1976. "Pottery: stratified groups and typology," in J. H. Humphrey (ed.), Excavations at Carthage 1975 conducted by the University of Michigan vol. I (Tunis) 47-108. Hayes, J. W. 1977. "Pottery report, 1976," in J. H. Humphrey (ed.), Excavations at Carthage 1976 conducted by the University of Michigan vol. IV (Ann Arbor, MI) 23-98. Hayes, J. W. 1980. Supplement to Late Roman Pottery (London). Horden, P. and N. Purcell 2000. The corrupting sea. A study of Mediterranean history (Oxford). Kenrick, P. M. 1985. Excavations at Sidi Khrebish Benghazi (Berenice) Ill,1. The fine pottery (Libya Antiqua Suppl. V). Lund, J. 1995. "A fresh look at the Roman and Late Roman fine wares from the Danish excavations at Hama, Syria,” in H. Meyza and J. Mlynarczyk (edd.), Hellenistic and Roman pottery in the Eastern Mediterranean: advances in scientific studies (Acts of the II Nieborów Pottery Workshop; Warsaw) 135-61. Lund, J. 2006. "Ceramic fine wares from the 4th century BC to the 7th century AD,” in L. W. Serensen and K. Winther-Jacobsen (edd.), Panayia Ematousa I (Danish Institute at Athens Monog. 6.1) 182-230. Mackensen, M. 1993. Die spatüntiken Sigillata- und Lampentöpfereien von El Mahrine (Nordtunesien). Studien zur nordafrikanischen Feinkeramik des 4. bis 7. Jahrhunderts (Munich). Malfitana, D. 2002. “Eastern terra sigillata wares in the eastern Mediterranean.

Notes

on

an

initial

quantitative analysis," in F. Blondé, P. Ballet and J.-F. Salles (edd.), Céramiques hellénistiques et romaines. Productions et diffusion en Méditerranée orientale (Travaux de la Maison de l'Orient Méditerranéen 35) 133-57. McCormick, M. 2002. Origins of the European economy: communications and commerce, AD 300-900 (Cambridge). Meyza, H. 2007. Nea Paphos V. Cypriot Red Slip Ware (Warsaw). Orssaud, D. 1980. "La céramique,” in J.-P. Sodini et al., ^ Déhés (Syrie du Nord). Campagnes I-III (1976-1978). Recherches sur l'habitat rural," Syria 57, 234-66. Orton, C., P. Tyers and A. Vince 1993. Pottery in archaeology (Cambridge). Oxé, A., H. C. Comfort and P. M. Kenrick 2000. Corpus Vasorum Arretinorum (Antiquitas 41, Bonn).

Parker, A. J. 1992. Ancient shipwrecks of the Mediterranean and the Roman provinces (BAR S580, Oxford). Peacock, D. P. 5., F. Bejaoui and N. Ben Lazreg 1990. "Roman pottery production in central Tunisia,” JRA 3, 59-84. Pena, J. T. 2007. Roman pottery in the archaeological record (New York). Pieri, D. 2005. Le commerce du vin oriental à l'époque byzantine (Ve-Vlle siècle) (Bibliothèque Archéologique et Historique 174). Poblome, J. 1999. Sagalassos Red Slip Ware: typology and chronology (SEMA II, Brussels). Poblome, J., A. Ceulemans and K. De Craen 2004-5. "The Late Hellenistic to Late Roman ceramic spectrum of

Tanagra," BCH 128-29, 561-69.

African Red Slip ware on the move: the effects for the Roman Poblome, J. et al. 1998. "Sagalassos Red Slip Ware. The organization of a manufactorv," MBAH

East 91 17, 52-64.

Poblome, J. et al. (edd.) 2004. Early Italian Sigillata. The chronological framework and trade patterns (Babesch Suppl. 10). Putzeys, T. 2007. Contextual analysis at Sagalassos. Developing a methodology for classical archaeology (Ph.D. diss., K. U. Leuven).

Rautman, M. L. 2003. A Cypriot village of late antiquity. Kalavasos-Kopetra in the Vasilikosvalley JRA Suppl. 52). Reynolds, ford).

P. 1995.

Trade in the western

Mediterranean,

AD

400-700:

the ceramic evidence

(BAR

5604, Ox-

Rickman, G. E. 1980. The corn supply of ancient Rome (Oxford). Riley, J. A. 1981. "The pottery from the cisterns 1977.1, 1977.2, 1977.3," in J. H. Humphrey

(ed.), Excavations

at Carthage 1977 conducted by the University of Michigan vol. VI (Ann Arbor, MI) 85-124. Rossiter J. J. and J. Freed 1991. “Canadian-Turkish Monde Classique 25 (1991) 145-74.

excavations at Domuztepe,

Cilicia (1989),”

Echos du

Slane, K. W. 1997. “The fine wares,” in S. C. Herbert (ed.), Tel Anafa IL, i: The Hellenistic and Roman pottery JRA Suppl. 10.2) 247-406. Slim, L. et al. 2008. “Les fabriques romaines de salaisons de poissons a Néapolis (Nabeul, Tunisie): travaux 1999-2005," in J. Napoli (ed.), Ressources et activités maritimes des peuples de l'Antiquité (Les cahiers du littoral 2, Boulogne-sur-Mer) 203-22. Tomber, R. 5. 1988. "Pottery from the 1982-3 excavations,” in J. H. Humphrey (ed.), The circus and a Byzantine cemetery at Carthage vol. 1 (Ann Arbor, MI) 437-528. Tomber, R. S. 1993. “Quantitative approaches to the investigation of long-distance exchange,” JRA 6, 142-66.

Tomber, R. S. 2004. “Polarising and integrating the Late Roman economy: the role of Late Roman amphorae 17," Ancient West & East 3.1, 155-65. Waage, F. O. 1948. “Hellenistic and Roman

tableware of North Syria,” in id. (ed.), Antioch on-the-Orontes

IV.1: Ceramics and Islamic coins (Princeton) t-60. Zoroglu, L. 2005. "Roman Fine Wares in Cilicia: an overview," RCRFActa 39, 243-48.

North African lamps: Bonifay and beyond Jeremy Rossiter Lucerna vel lampas sine oleo accensa fumare potest, fetere potest, lucem habere non potest! Á pottery or bronze lamp lit without oil can smoke and smell but it cannot shed light.

Background The study of pottery lamps in Roman Africa still owes much to the pioneering work of J. Deneauve, whose 1969 catalogue of the lamps in the Bardo and Carthage museums created the first significant typology of Greek and Roman lamps, both local and imported, found at sites in Tunisia.? Since the appearance of his landmark study, N African lamp studies have

progressed along a linear path, marked by the periodic appearance of new publications dealing with lamp collections from a variety of sources. In the first place, there have been further studies based on collections of lamps located in N African museums, such as A. Ennabli's catalogue of late-antique lamps in the Carthage museum and E. Joly's catalogue of the lamps in the Sabratha museum.’ There have also been a number of new studies based on collections of N African lamps located in museums outside N Africa, such as D. M. Bailey's catalogue of N African lamps in the British Museum or C. Trost and M.-C. Hellmann's catalogue of N African lamps in the Bibliothèque Nationale.* Such museum catalogues are of value, not least since they tend to provide the most complete examples of different lamp types, but the frequent lack of context for the lamps and the consequent lack of meaningful dating evidence is a limitation. In addition to museum catalogues, there have been several recent publications of lamps (more usually, fragments of lamps) found during excavation and survey at different N African sites. These include the publications of lamps found by the different teams working at Carthage as part of the UNESCO-sponsored program chiefly from the mid-1970s to mid-1990s,? as well as reports on the lamps from the Raqqada necropolis, from kiln sites at El Mahrine, and from recent excavations at Uthina (Oudhna).f There have also been several reports on lamps from excavations in Libya, in particular Sidi Khrebish (Berenice) in Cyrenaica and the 3rd-c. military site at Bu Ngem in Tripolitania.’ Such site reports tend to provide the best dating evidence for the lifespan of certain types, although the difficulties of dating lamps from 1

Augustine of Hippo, Sermo 107.2 [= PL 39 col. 1958]. It is probable that the word lampas here means a metal lamp and the word lucerna a ceramic lamp: cf. Juv., Sat. 3.285: multa praeterea flammarum et aenea lampas; Colum., RR 12.18.5: ferreae lampades.

J. Deneauve, Lampes de Carthage (Paris 1969). A. Ennabli, Lampes chrétiennes de Tunisie (Paris 1976); E. Joly, Lucerne del Museo di Sabratha (Rome

1974). 4

5

D. Bailey, A catalogue of the lamps in the British Museum 178-202; C. Trost and M.-C. Hellmann, Lampes antique III. France) (Paris 1996). — Eg, L. Anselmino, “Le lucerne rinvenute negli scavi della problemi tipologici e cronologici," in Actes. Colloque sur

III. Roman provincial lamps (London 1988) Lampes chrétiennes (Bibliotheque nationale de missione archeologica Italiana a Cartagine: la ceramique antique. Carthage, 1980 (Tunis

1982) 157-70; J. Rossiter. "Lamps from the late-4th to early-5th century deposit," in J. H. Humphrey (ed.), The circus and a Byzantine cemetery at Carthage, vol.

I (Ann Arbor, MI 1988) 529-48; K. Knowles,

"The

lamps," in M. G. Fulford and D. P. S. Peacock (edd.), Excavations at Carthage, The British Mission, vol. IL2. The circular harbour, north side. The pottery (London 1994) 23-41; B. Bechtold and K. Schmidt,

"Lampen," in H. Niemeyer, R. Docter and K. Schmidt, Karthago. Die Ergebnisse der Hamburger Grabung unter Decumanus Maximus, Teilband II: Die Funde, II (Mainz 2007) 594-616.

6

A. Ennabli, A. Mahjoudi and J. W. Salomonson, La nécropole romaine de Raqqada (Tunis 1973); M. Mackensen, Die spütantiken Sigillata- und Lampentopfereien von El Mahrine (Nordtunesien) (Munich 1993); H. Ben Hassen and

7

L. Maurin,

Oudhna

(Uthina). Colonie de veterans de la XIIIe légion. Histoire,

urbanisme, fouilles et mise en valeur des monuments (Bordeaux 2004) 144-45 and 231-32. D. M. Bailey, Excavations at Sidi Khrebish, Benghazi (Berenice), vol. 3, part 2. The lamps (Libya Antiqua Suppl. 5, Tripoli 1985); R. Rebuffat, "Lampes romaines à Gholaia (Bu Njem, Libye)," in Th. Oziol and R. Rebuffat, Les lampes de terre cuite en Méditerranée. Des origines à Justinien (Lyon 1987) 83-90.

94

J. J. Rossiter

archaeological contexts are considerable (it is more common to find lamps used to date archaeological contexts, rather than the other way round). Much rarer are ‘synthetic’ studies which draw on a broad range of material published in museum catalogues and site reports to forge new, comprehensive typologies of the lamps of certain key periods or from certain key regions of N Africa. The most important examples in the last 30 years are L. Anselmino and C. Pavolini's major study of pottery lamps from Tunisia? and J. Bussiére’s wide-ranging classification of pottery lamps from Algeria.? The former is now widely used as the definitive typology of the ARS lamps produced in Byzacena during the Late Roman, Vandal and Byzantine periods, while the latter, recently augmented by a second volume, offers a much-needed typology of lamps of all periods from Punic to Byzantine found in Algeria. To these essential studies can now be added M. Bonifay's work on lamps from Tunisia, which forms a significant part (pp. 312-430) of his Etudes sur la ceramique romaine tardive d'Afrique. The aim of Bonifay's lamp report is twofold. First, it serves as a forum for the classification of the lamps found during recent excavations by French and Tunisian archaeologists mainly in the Pupput necropolis but also at Sidi Jdidi, Oudhna and Nabeul; second, it uses the new material as a starting point to create a new typology of the different kinds of lamp produced in Tunisia during the Roman and post-Roman eras. In creating a new typology, Bonifay is quick to acknowledge his debt to previous scholars, in particular Deneauve and Anselmino-Pavolini, who provided the basic framework on which he builds his new, more refined classification, but

there are some important differences and limitations. First, unlike Deneauve, Bonifay concentrates almost exclusively on lamps made in N Africa. There is little discussion here either of lamps imported into Tunisia (mainly from Italy) during the first Roman century or of the more difficult question of how to distinguish between those imported lamps and their earliest locally-made imitations. There are also regional limitations. The territory covered is roughly the same as that covered by Anselmino and Pavolini's Atlante volume — the part of N and central Tunisia which includes Carthage and its hinterland, as well as the Cap Bon peninsula and coastal sites as far south as Sousse. Lamps from other parts of Tunisia remain less well documented, although new studies of material from sites such as Leptiminus (Lamta) are beginning

to correct this imbalance.!® The material included by Bonifay is classified and discussed with meticulous thoroughness, and it is the results of his new classification and of the ways in which it builds on previous classifications that deserve close attention. At the outset, Bonifay outlines his criteria for typological analysis. Unlike Deneauve, whose typology rests heavily on nozzle forms, Bonifay emphasizes the need to look at a broader set of type markers, including nozzles, handles, dimensions, decoration, and bases. Nevertheless, the first part of his typology leans heavily on Deneauve's catalogue of lamps from Carthage, which means that nozzle forms still remain the dominant criterion. N Áfrican brown-slipped lamps (2nd-3rd c. A.D.) In its early stages the N African lamp industry depended on the importation and copying of

Italian models.!! The first 'Roman' lamps produced in N Africa were copies of Italian imports, 8

L. Anselmino

9

orientale. AHante delle forme ceramiche I: ceramica fine romana nel bacino mediterraneo, medio e tardo antico (Rome 1981) 184-207. J. Bussière, Lampes antiques d'Algérie 1 (Monographies Instrumentum 16; Montagnac 2000); id., Lampes antiques d'Algérie II. Lampes tardives et lampes chrétiennes (Monographies Instrumentum 35; Montagnac 2007).

10

and

C. Pavolini,

"Terra

sigillata:

lucerne”

in Enciclopedia

dell'arte antica,

classica

e

For Lamta (Leptiminus), see J. W. Hayes, “Lamps,” in L. M. Stirling,, D. J. Mattingly and N. Ben Lazreg, Leptiminus (Lamta), report no. 2. The East Baths, cemeteries, kilns, Venus mosaic, site museum, and other

studies (JRA Suppl. 41, 2001) 202-5; T. Allen in D. Stone et al., Leptiminus (Lanta), report no. 3 (2010 forthcoming). 11

Fora good discussion of the beginnings of the lamp industry in N Africa, see Bussière 2000 (supra n.9)

141-48.

North African lamps: Bonifay and beyond

95

initially, it would appear, made by African lamp-makers operating ‘under license’ from large Italian producers. Soon these African ‘branch workshops’ (succursales) started to produce lamps of original design, although the debt to Italy remained strong. Bonifay begins his typology with a discussion of various types of early N African lamps which illustrate this evolution from an import industry to a local N African industry. These include a small group of lamps classified as "lampes africaines de tradition italique”, lamps which closely imitate imported Italian models. The great majority of the lamps in the first part of his typology, however, belong to the two 'classic' Romano-African lamp types identified by Deneauve as his Types VII and VIII. Bailey refers to these 'classic' African lamps as "Standard African Loeschke Type VIII” lamps, describing them as "lamps with buff bodies with brown slips”;!? in general this description is valid, although it should be noted that, while many of these lamps do indeed have a dark (‘chocolate’) brown slip, others have a coating which varies in colour anywhere from red-brown to orange-brown. The clay is generally fine and light brown to ‘buff’ in colour. African

lamps

based on Italian models

The earliest lamps in Bonifay's typology (his Types 1-3) are a group of lamps made in the ‘Italian tradition’ which come from 2nd-c. tombs excavated at Pupput. Type 1 are ‘ear’ lamps (lamps with side lugs) of Deneauve Type VG. Types 2 and 3 are lamps of related form with a decoration of globules on the shoulder. Bonifay remarks on the similarity of some of these Type 2 and 3 lamps (especially his Type 2, no. 7) to lamps of Bailey's Italian Type K, which are common in Italy in 2nd- to 3rd-c. contexts.!? The question then arises whether these lamps are in fact African lamps made in imitation of Italian lamps, or Italian imports. In a number of places (e.g., 313, 321, 325), Bonifay admits

to uncertainty; some of the lamps included here, he sug-

gests, are probably imports from Italy rather than locally-made pieces. The argument usually rests

on

visual

observation

of the

fabric,

which

in the case

of Italian

lamps

is distinctly

micaceous. À more scientific approach (compare, for example, the clay analysis of a group of N

African lamps reported by Bailey) might ultimately be the only way to resolve such issues." ‘Classic’

Romano-African

lamps

Tackling the much larger category of Deneauve Type VII and VIII lamps, Bonifay offers some significant reworking of the old typology. Deneauve's extensive class of Type VII lamps is here reclassified as 6 sub-types (Bonifay Types 4 to 9) based on 6 variations of the form found among the lamps from Pupput (also referred to by Bonifay as Pupput Types 1 to 6). Sub-type 1, of which Bonifay illustrates no fewer than 25 examples from Pupput, constitutes what can be considered the 'standard' African Type VII lamp. The Pupput material provides useful new dating evidence for this type, most examples of which are dated by context to the 2nd c. Other Type VII sub-types vary from the standard form in size and in details of decoration. Sub-type 5,

which Bonifay, borrowing from Bussière,!* quaintly terms "lampes galettes” (shaped like a cake), are distinguished by their large size and relatively flat discuses. Bonifay includes a number of lamps in this group, ranging in date from the early 2nd to the early 3rd c. Sub-type 6, which consists of a single lamp from Pupput, is questionable as a sub-type; it comes from a tomb of mid- to late 3rd-c. date, much later than other lamps from this group. Deneauve's Type VIII lamps receive similar treatment from Bonifay, with 4 new sub-types proposed (Bonifay Types 10 to 13), illustrated mainly by examples found in tombs at Pupput. The main distinction between sub-types is the style of the shoulder decoration, which uses one of four motifs: garlands of olive (sub-type 1) or laurel (sub-type 2), grape vines (sub-type 3), or globules (sub-type 4). Under sub-type 3 Bonifay includes some unusual variants: a lamp from

12 13

Bailey (supra n.4) 179-80. Id. A catalogue of the lamps in the British Museum II. Roman lamps made in Italy (London 1980) 246-53, especially Q1116; E. Fabbricotti, "Osservazione sulle lucerne a perline," Cenacolo 4 (1974) 23-30; J. Rossiter, “The lamps,” in C. J. Simpson (ed.), The excavations at San Giovanni di Ruoti Il. The small finds

(Toronto 1997) 89-101, Group 4. For similar lamps from Algeria, see Bussiére 2000 (supra n.9) p. 277, 14

nos. 755-82. Bailey (supra n.4) 477-79.

96

J. J. Rossiter

Fig. 1. Lamp fragment from Carthage (Deneauve Roman Type X; Bonifay Type 15 variant) (photo K. Fodor). Fig. 2. Unpublished lamp from Carthage (Deneauve Roman Type VIII; Bonifay Type 32) (photo author) Fig, 3. Unpublished lamp from Carthage (Atlante Form VIIID; Bonifay Type 50) (photo author). Fig, 4. Unpublished lamp from Carthage (Atlante Form VIIIC1/X; Bonifay Type 61 variant) (photo author). Fig, 5. Unpublished lamp from Carthage (Atlante Form XD4; Bonifay Type 67) (photo author). Fig. 6. Lamp from San Giovanni di Ruoti, S Italy (cf. Bonifay Type 71) (photo author). Fig. 7. Unpublished lamp fragment from Bir Ftouha, Carthage (photo P. Reynolds).

15

Bussiére 2000 (supra n.9) 99-100.

North African lamps: Bonifay and beyond

97

Deneauve’s catalogue (no. 1013) which illustrates a form of Type VIII lamp common at Carthage, and a lamp from Pupput which is said to be an Italian imitation of an African Type VIII lamp. Again, the tombs at Pupput provide useful dating evidence for some of these lamps. The earliest examples of sub-type 1 come from a tomb (tomb 615) of late 2nd- to early 3rd-c. date; an example of sub-type 4 is from a tomb dating to the second half of the 3rd c. In general, Bonifay rejects earlier proposals of a 2nd-c. date for these Deneauve Type VIII lamps in favour of a 3rdc. or later date. Certainly lamps of this type were in common use during the Severan occupation of the fort at Bu Ngem, dated to the years between 201 and 259/263.16 For his next group of classic 'African' lamps ("lampes à becs traingulaires": Bonifay Types 14-16 = Deneauve Type X) Bonifay draws heavily on Deneauve's catalogue of Carthage lamps. Of the 36 examples of the group catalogued here, 24 are from Carthage and only 4 from Pupput, suggesting that this type of lamp was a product of Carthaginian workshops. Very few of these lamps are securely dated, although one of the Pupput examples (Bonifay Type 15, no. 6) is from a tomb dating to the mid- to late 3rd c. A wide range of variants is noted, especially among the many examples of sub-type 2, to which can now be added a recently-discovered fragment from

Carthage (fig. 1).!7 A number of examples of these Type X lamps are known from sites in Algeria. Bonifay suggests that these may be the result of surmoulage, but, in the absence of clear proof of that, simple imitation by local workshops would seem just as likely. The whole question of surmoulage is one which needs further investigation. The re-use of lamps to make new moulds is an unsurprising manufacturing trick, but finding unequivocal evidence of it is more difficult. Bussiére uses lamp-makers' marks to demonstrate surmoulage in the case of certain Italian lamps reproduced in N Africa, but the spread of lamp types by this method from one region of N Africa to another remains poorly documented.!5 At the end of the first part of his catalogue, Bonifay deals briefly with a number of locallyproduced lamp types which have been documented from Raqqada and from sites outside Tunisia such as Setif, Sabratha and Bu Ngem. To these he adds a few lamps from Nabeul and Pupput which he considers to be products of a previously unidentified 'Cap Bon' lamp workshop. The story of these local N African lamp workshops has yet to be fully told. One suspects this is only 'the tip of the iceberg' and that future studies of lamps from the many small towns and regional centres scattered throughout Roman Africa will reveal a complex variety of locally-made lamp types, at present only hinted at in the literature. African buffware lamps (4th-5th c.) Derivatives

of classic African

lamps

One of the important developments in N African lamp studies since Deneauve's day has been the recognition that a typological distinction needs to be made between the 'classic' N African lamp types of the 2nd to 3rd c. (Deneauve's Types VII and VIII) and their later (4th- to 5th-c.) derivatives. The need for this distinction has been argued before, but now for the first time Bonifay gives it a thorough and definitive treatment. He divides the various Deneauve Type VII and VIII derivatives into 7 different sub-types, three for Type VII (Bonifay Types 25 to 27) and four for Type VIII (Bonifay Types 28 to 32). As usual, he uses examples from his main research sites (Pupput, Oudhna, Nabeul) to illustrate these different sub-types, adding a few examples from Carthage where lamps of these types have been found in large quantities in Late Roman contexts.!? An unpublished lamp from the Canadian Theodosian Wall excavations there offers a rare variant of Bonifay's Type 32; the shoulder is decorated with clusters of raised dots or globules interspersed with a series of raised arrows (fig. 2). 16 17

Rebuffat (supra n.7). ].J. Rossiter in S. T. Stevens, M. B. Garrison and J. Freed, A cemetery of Vandalic date at Carthage (JRA

Suppl. 75, 2009) pp. 245-46 and no. 86. 18

Bussiére 2000 (supra n.9) 98.

19

Anselmino (supra n.5) 160; Rossiter (supra n.5) 536-38.

98

J. J. Rossiter

Although a few of the Pupput pieces are said to be from tombs of 4th-c. date, most of the lamps included here are from late 4th- to early 5th-c. contexts. What is less clear is where these lamps were made. Were Carthage, Nabeul and Pupput supplied from the same source? Or

did they each have their own local workshops producing lamps of the same type??? Other

buffware

Bonifay identifies kin lamps’), which

lamps a number of variants of the so-called "lampes à cótes de melon" ('pumpwere

a common

product

of N African

lamp

kilns in the 4th and

5th c.

(Bonifay Type 33). Numerous examples have been documented at Carthage and in Algeria?! Unlike the Late Roman globule lamps discussed above, the source of these Late Roman ‘pumpkin’ lamps is not certain. They appear to be a distinctly N African product, rarely found elsewhere.?? An early 3rd-c. lamp found at Carthage (Deneauve no. 907) offers a possible precedent, but it is a rare piece of unknown context. Bonifay illustrates the type with examples from Pupput,

Thuburbo

Maius,

Sidi Jdidi

and

Nabeul.

Again

the question

remains

whether

these

pieces are from a common source or whether (more probably) they derive from numerous local kilns copying models produced by one or two or the larger urban manufacturers. African red-slipped lamps The beginning of ARS

lamp production

The next part of Bonifay's typology deals with ARS lamp production in N Africa. As he is

quick to point out, this is essentially a reiteration of Anselmino and Pavolini's 1981 typology, but with some refinements and with the addition of new material from his own fieldwork. There are two important innovations: one is the approach taken to the less-well-documented ‘early’ ARS lamp types; the other is the attempt made to localize ARS lamp production on the basis of decoration. As with the Atlante typology, the early ARS lamp forms (Atlante Forms I-VII = Bonifay Types 37-42 "types precoces") are treated here (353-58) much more sparingly than are (358-427) the ‘classic’ ARS forms (Atlante Forms VIII-XII = Bonifay Types 43-81). This is not the fault of the author but merely a reflection of how much relative data for these two classes of lamps is available from museum collections and excavation reports. Any attempt to build up a definitive picture of early ARS lamp production in N Africa faces a serious deficit of evidence. What Bonifay offers is a useful working hypothesis which others, with access to additional evidence, can put to the test in the future. It is generally accepted that most Atlante Forms I-VII lamps were produced by workshops in central or 5 Tunisia. The pieces included in Bonifay's typology are nearly all from sites in this area, including El Jem, Sbeitla and Kasserine. Since the number of published

lamps from such

sites is limited, the picture which emerges is inevitably tentative, with lamp types and variants based on rare and scattered examples. For example, Bonifay's Types 38 and 39 (Atlante Forms IV and V) are represented here by just 5 pieces, two of them small fragments from Bonifay's own survey at En Nadour on Tunisia's E coast. The chronology of these early ARS lamp types is not well defined, again owing mainly to the small number of finds from dated contexts. The earliest ARS lamps included here (Atlante Form I = Bonifay Type 37) are rare ARS imitations of Deneauve Type VIII lamps, for which Bonifay suggests a 3rd-c. date based on two fragments from dated contexts at Nabeul. Other types (Atlante Forms IV-VI = Bonifay Types 38-41) are even less securely dated, with most of the examples listed by him coming from unstratified contexts. A comprehensive treatment of these early ARS lamp types remains one of

the greatest desiderata in N African lamp studies. 20 21 22

Forthe excavation of a kiln producing buffware lamps in Carthage, cf. BAC 1943-45, 362 and 417. Bussiére 2000 (supra n.9) 119-20; Rossiter (supra n.5) 540. Bussière 2000 (supra n.9) 119, n.110, points to a few examples found in Italy, but whether these were of

Italian or N African manufacture is not clear.

North African lamps: Bonifay and beyond

99

‘Classic’ ARS lamps The ‘classic’ ARS lamp types (principally Atlante Forms VIII and X) are much more fully represented. Partly because of their widespread circulation in the Carthage area, large numbers of these lamps have been recovered, and examples are to be found in most museum collections. As noted by the authors of the Atlante typology, this allows for a much more comprehensive analysis of forms and variants. Bonifay's stated aim is not to develop a new typology of these classic lamps, but rather to refine the existing typology with particular emphasis on decoration and what this may tell us about provenance and date: je ne propose pas de nouvelle classification mais un aménagement de celles qui existent déjà, en essayant avant tout de préciser les zones de production et la chronologie (358).

In general Bonifay's typology of ARS lamps adheres closely to the Atlante typology, taking into account the various modifications made to it by subsequent studies, especially the work by M. Mackensen

on the lamp

kilns at El Mahrine.

That is not to say, however,

that Bonifay's

work is without innovation. There are a number of significant changes to the existing typology, not least in the organization and identification of various classes. In place of the Atlante’s cumbersome lettering system (Form VIIIA1a, VHIA1b etc.), Bonifay opts for a much simpler numbering system (Types 43, 44, and so on). On occasion the sequence of types is re-arranged (Atlante Form VIIIAla = Bonifay Type 45, Atlante VIIIB = Bonifay Type 43). Atlante Form

VIII lamps

The basis of Bonifay's classification is, in the first place, regional. Thus, lamps of Atlante

Form VIII are here classified under three broad 'production area' headings: central Tunisia (Byzacena) (Bonifay Type 43 = Atlante Form VIIIB); central and N Tunisia (Bonifay Types 4447 = Atlante

Forms

VIIIA1a-c,

VIIIA2a-b,

VIIIC1a-b,

VITIC2a-b);

N

Tunisia

(Bonifay

Types

48-50 = Atlante Forms VIIIC1c-e, VIIIC2c-d, VIIID). The central Tunisian origin of the first of these groups has long been argued, not least on the basis of finds from the kiln site at Henchir

es-Srira.? Bonifay adds several examples from Nabeul coming from contexts dating between the mid-4th and mid-5th c. The second and largest group is harder to classify. Production of lamps of all four types in this group is attested (or strongly implied) both in central and in N Tunisia. Examples of the first three types were found at El Mahrine.?^* Dating of this group of lamps remains problematic. According to Mackensen, production of these lamps begins in the

mid-4th c. Bonifay argues for a somewhat later date, noting that lamps of Types 45 and 46 found at Nabeul come from securely dated contexts of the first half of is made up primarily of lamps of Atlante Form VIIID, a distinctive discus decoration and a reduced (or non-existent) shoulder. Bonifay decoration, the most common being a scallop or a star; a lamp from Wall excavations at Carthage provides a further unusual variant, occupying the centre of the scallop (fig. 3).2° Also included

in this group

are lamps

of Atlante

Form

the 5th c. The third group type of lamp with a full lists 8 variants of discus the Canadian Theodosian with a figure of Venus

VIIIC1c,d,e

and

VIIIC2c,d,

which

Bonifay regards as precursors to Form VIIID. Again, dating remains an issue. Examples of lamps

of Atlante Form VIIID found at Carthage have been dated mainly to the second half of the 5th or early 6th c. Mackensen, however, dates examples of lamps of Atlante Form VIIID8 found at

El Mahrine to the second half of the 4th c.2° At issue here is the similarity between lamps of Atlante Form VIIID8 and the buffware 'pumpkin' lamps of Deneauve Type IXB - Bonifay Type

33. Bonifay argues for the need to make a clear distinction between the former and ARS imitations of the latter, which are almost certainly earlier.?? 23

D. Peacock, F. Bejaoui and N. Ben Lazreg, "Roman pottery production in central Tunisia,” JRA 3 (1990) 59-84.

24

Mackensen (supra n.6) figs. 32, no. 7, 33 no. 1, 34 nos. 6-7.

25 26 27

Canadian Carthage Excavations CC2 No. L-018. Mackensen (supra n.6) 150. Examples of ARS ‘pumpkin’ lamps are rare. As an illustration, Bonifay (2004, 370, n. 228) notes a find

from the recent French excavations at Lepcis Magna. The only other example I know of was found at Carthage in the Canadian Theodosian Wall excavations: CC2 no. L-188 (unpublished).

100

J. J. Rossiter Atlante Form X lamps

Lamps of Atlante Form X are classified in a similar way, first by region and then by decoration. As Bonifay notes, earlier studies have tended to emphasize one or other of these criteria, but rarely both. Among the earlier studies which Bonifay singles out for comment are those by Hayes (1972) and Peacock (1984), the first laying the foundation for the identification of regional differences in the origin of Form X lamps, the second suggesting the importance of

decorative motifs as a tool for distinguishing types.?® In sorting these classic ARS lamps into new groups, Bonifay aims to incorporate the lessons of both earlier studies: on the one hand, he maintains a strict regional division (Type 53 to 60 = central Tunisia; Types 61 to 67 = N Tunisia);

on the other, he uses details of decoration (variations of discus design and shoulder motif) as a means for further typological distinction. But these are not the only criteria for distinguishing

types. Differences in size, quality of moulding, and fabric also contribute to differences of classification. Ultimately it leads to a highly refined, if somewhat uneven, typology, since the criteria for classification are not uniform. For example, the difference between Bonifay's Types 54 and 55 is mainly one of overall size (which in turn affects the kind of discus images employed); Type 56 is characterized by the size of shoulder motifs; whereas Type 57 is distinguished on the basis of fabric and quality of moulding. (The numbering system here demands careful

attention,

not

least because

it re-uses

the

old

letter-based

system

of the

Atlante

typology in conjunction with Bonifay's own new numerical system.)

Central Tunisian products are divided into 5 'groups' identified both as lamps of Atlante X Group C1 to C5 and, more simply, as lamps of Bonifay Types 53 to 60. For each of these types Bonifay lists a large number of variants distinguished primarily on the basis of discus and shoulder decoration. Most of the examples he uses are drawn from the standard catalogues, many of them coming from Carthage, although a few examples from Italy and other parts of the Mediterranean world are also included. Type 55 is especially common outside N Africa — perhaps, Bonifay suggests, because the lamps were relatively small and easy to transport in quantity. Dating of these central Tunisian products is well known in broad outline but remains problematic in detail. As Bonifay notes (371 and 382), examples of these lamps from securely dated contexts in their region of production are rare. Dated examples tend to be from N Tunisia or from

outside Africa. Type 54 no. 86, from Marseilles, is from a context dated to 425-450; Type 56 no. 20, from Carthage,

is dated

to the mid-6th c. A S Italian copy of Type

55, no. 28 (from Car-

thage) was found at the villa site at San Giovanni di Ruoti (Basilicata) in a context securely dated to 490-535.2? For Atlante Form X lamps of N Tunisian origin (Types 61-67) Bonifay adopts a similar approach, using a variety of criteria for purposes of classification. These lamps are divided into three sub-groups, the first (Types 61-63) being hybrid forms of Atlante Forms VIII and X, the second (Types 64-65) "imitations serviles" (utilitarian copies) of central Tunisian lamps of Atlante Form X, and the third (Types 66-67) distinct N Tunisian products. Lamps of Types 61-63 are rare, represented in Bonifay's catalogue by only 12 examples, mostly from Carthage. A further example, from the Canadian Theodosian Wall excavations at Carthage, is illustrated

here (fig. 4).9 By contrast, Types 65 and 66 are represented by nearly 60 lamps each, offering a wide range of variants based mainly on decorative style, especially the different types of

shoulder decoration. In this respect Bonifay adheres closely to the typological distinctions proposed by Chapman, Davies and Peacock in the British Carthage Avenue Bourguiba report.?! Lamps of Type 66 were produced at Oudhna and El Mahrine, as well, Bonifay argues (391), as at other (unknown) kiln sites in N Tunisia. Oudhna is also suggested as the probable source of 28

J. W. Hayes, Late Roman pottery (London 1972); S. Chapman, S. Davies, and D. Peacock, "The lamps,” in

M. G. Fulford and D. P. S. Peacock, Excavations at Carthage. The British Mission vol. 1.2. The Avenue du 29 30

Président Habib Bourguiba, Salammbo. The pottery and other ceramic objects (Sheffield 1984) 232-41. Rossiter 1997 (supra n.13) 96-97, no. 63. Canadian Carthage Excavations CC2 no. L-189.

31

Supra n.28.

North African lamps: Bonifay and beyond

101

lamp no. 12 of Type 67 which comes from the British harbour excavations at Carthage. An almost identical lamp (fig. 5) was found at the Canadian Theodosian Wall site at Carthage. The majority of dated examples of Type 65 and 66 lamps belong between the late 5th and mid 6th c. Bonifay's catalogue includes several lamps from the excavations at Sidi Jdidi, from contexts dating to the late 5th c. A near-complete Type 66 lamp (cf. Bonifay's Type 66 no. 50)

came from a mid-6th c. context in the ecclesiastical complex at Bir el Knissia (Carthage); it appears (392) that these lamps do not occur in contexts of the mid- to late 5th c. A somewhat later date is proposed for lamps of Type 67, which Bonifay places between the late 6th and mid-7th c. Bonifay's final group of classic ARS lamps are what he terms "lampes tardives", lamps produced in the 6th and 7th c. in a style copying the standard Atlante Form X lamps of previous centuries, though generally of inferior quality. Within this broad category he identifies three different types of lamps: 1) lamps with 'expanded' decorative motifs ("contours saillants") (his Type 68); 2) lamps of very worn appearance ("décoration estompé") resulting from repeated surmoulage

or the use of old moulds (his Type 69); and 3)

lamps with 'incised' relief decoration (his Type 70).

What is important here is the evident abandonment in the later 6th c. of earlier techniques of lamp-making. Bonifay argues that by the end of the 6th-c. lamp-makers in N Africa were not engaged in the business of making new plaster moulds for their products, but were either re-

using old moulds or using old lamps to make new clay moulds.?? This change to clay moulds is most obvious when looking at lamps of Type 70, which range in date from the late 6th to the late 7th c. or even later. Among those used by Bonifay to illustrate this incising technique are a number of lamps and a fragmentary clay mould found at Rougga, southeast of El Jem. Study of these incised Byzantine lamps is, as Bonifay notes, still 'embryonic': the type in general needs further study. Bonifay's typology of Form X lamps concludes with a small but interesting group of lamps (his Type 71) which have the form and appearance of ARS lamps but were produced, either by surmoulage or with the incising technique, in local buffware fabrics. He points to similarities between this group of lamps (especially his Type 71 no. 9 from Rougga) and lamps from Sicily. Similar ‘imitation ARS’ lamps have been found in significant numbers at sites in Algeria (e.g., Hippo, Tipasa, Tiddis, Timgad)

and

in 5

Italy. (e.g., San Giovanni

di Ruoti

[fig. 6] and

Or-

dona?4). Bussiere’s recent work in particular has demonstrated the importance of local lamp production at select sites in Algeria in the 6th and 7th c. As research on lamps extends further into the more remote areas of N Africa, these locally-produced lamp types are likely to enter more

into the mainstream of N African lamp studies. The Rougga examples noted by Bonifay may be as late as the 8th c.?? Byzantine 'pot-lamps'

As a final addition to his typology, Bonifay offers a new assessment of a distinctive group of N African lamps which have often been referred to in the literature as ‘Vandal’ or ‘ink-pot’ 32

J. Rossiter, “Lamps from Unit 4000," in S. Stevens, Bir el Knissia at Carthage; a rediscovered cemetery church. Report no. 1 (JRA Suppl. 7, 1993) 178-80, no. 29.

33

On this change of technique, see also M. Bonifay, "Ceramic production in Africa during late antiquity: continuity and change," in L. Lavin, E. Zanini, A. Sarantis (edd.), Technology in transition A.D. 300-650

34

(Leiden 2007) 151-52. Bussiére 2007 (supra n.9) 45-55 lists 3162 examples of sites in Algeria. For examples from $ Italy, see Rossiter "Présentation de l'ensemble des lampes découvertes de 1974) 7-101, nos. 398 and 881. The lamp illustrated here

locally produced ‘imitation ARS’ (supra n.13) 97-99, Groups 7 and 8; 1962 a 1971," in J. Mertens, Ordona (fig. 6), although found in Lucania,

lamps from C. Delplace, IV (Brussels is probably

a Campanian product; it is worth comparing to Bussiere lamp C 1760 (from Djemila). 35

According to Bonifay (417), they derive from post-Byzantine contexts dated by R. Guery to the 8th c.

102

J. J. Rossiter

lamps.?6 In form they are quite different from the moulded lamps which form the bulk of the typology. Elsewhere I have used the term 'pot-lamp' to describe them, since their basic form is that of a wheel-made pot. Bonifay emphasizes that these lamps are almost certainly not Vandal in date: datable examples from Carthage are nearly all from 6th- to 7th-c. contexts. A newly-published lamp fragment from a cemetery site at Carthage offers a variation of the

usual 'pot-lamp' type with a simple decorative pattern on the rim.” Where do we go from here? Bonifay's new typology sits squarely on the shoulders of earlier typological studies of N African lamps and, as such, represents an important step forward in a linear progression of studies whose primary interest is the form and decoration of pottery lamps. Broader contextual questions, such as where in archaeological sites lamp finds have been most prolific, or where and how, according to contemporary writers, lamps are most likely to have been used, are not

raised. In this respect Bonifay's study is very different from the recent study by H. Eckardt of

lamps in Roman Britain.?? Eckardt's interests range beyond the study of lamps as typological artifacts: questions of lamp distribution within archaeological sites, the relationship between pottery lamps and other kinds of lighting devices, the use of lamps in religious ritual, and the art-historical significance of lamp iconography are all discussed. Bonifay's approach is more one-dimensional, focusing almost exclusively on the appearance of the lamps and, to a lesser extent, on the manufacturing processes used to create them. The result is a thorough typology, thoroughly annotated, which eclipses all previous classifications of the same material. How useful is it to a fieldworker? The response is likely to be mixed. On the one hand, anyone needing to check the typological placement of a complete N African lamp (which in an excavation context is a relative rarity) will find Bonifay's work most helpful. On the other hand, the more common need to allocate small fragments of pottery to a particular lamp type will prove much more difficult: the sheer number of types (85) identified by Bonifay, not to mention the multiplicity of variants within each type, is daunting, and in many cases the differences between types are so minor that only a complete or almost complete lamp could be satisfactorily placed into the typology. For convenience, smaller lamp fragments will probably continue to be classified according to the broader divisions set out in the Atlante. Two major areas of investigation remain wide open: one is the study of lamps from regions lying outside the core territory of Roman N Africa (Carthage and its hinterland); the other, the study of lamps which date outside the core periods of Roman, Vanda! and Byzantine settlement in N Africa. Geographically, the study focuses on material recovered from sites in NE and central Tunisia. Bonifay introduces a good amount of new material but it too tends to be from sites in the same limited area (Nabeul, Pupput, Sidi Jdidi). A major leap forward will come when lamp studies start to move beyond this region. There are already signs of change. Bonifay signals imminent publication of lamps from the forum excavations at Rougga;? lamps from recent work at Lamta

(Leptiminus)

are receiving new

attention;

and lamps from Jerba

have been briefly treated.*! Outside Tunisia, J. Bussiére’s major catalogue of Algerian material has set new standards for the study of lamps in that region.? Unfortunately, there is still no

36

Eg. Trost and Hellmann

37 38 39

lamps," in Fulford and Peacock (supra n.28) 240. For examples of these lamps from Algeria, see Bussiere 2000 (supra n.9) 126-27, Type EVI.2; X. Delestre, Hippone (Aix-en-Provence 2005) pl. XXII, no. 211-12. Rossiter (supra n. 17) no. 9 on p. 255. MH. Eckardt, MHluminating Roman Britain (Montagnac 2002) R. Guéry and M. Bonifay, "La ceramique antique," in M. Euzennat and H. Slim, Rougga I. Le Forum et ses abords (Rome forthcoming). T. Allen (supra n.10).

40 41

(supra n.4) 124-25, "lampes

tournees

dites vandales”;

Peacock,

“Vandal

E. Fentress, A. Drine and KR. Holod, An island through time. Jerba studies vol. 1. The Punic and Roman

periods (JRA Suppl. 71, 2009) 252-54, referring to Attic and Punic lamps found during the recent survey. Roman lamps are mentioned briefly (but not illustrated) on 215-16, 219-221, 227 and 231. 42

Bussière 2000 (supra n.9).

North African lamps: Bonifay and beyond

103

equivalent for lamps from W Libya: although reports exist for Sabratha and Bu Ngem, lamps

from Tripoli (Oea) and Lepcis Magna remain poorly known.? When lamps from these and other sites are fully studied and incorporated into the mainstream of N African lamp studies, the overall picture will change and the dominance of N and central Tunisian producers may be shown to be less than is normally supposed. As for the chronological scope of Bonifay's work, it ends where, traditionally, one would expect it to end, in the final years of Byzantine rule in N Africa. He refers in several places to lamp production of the 7th c., occasionally even to lamp types which remained in circulation into the 8th. Rarely, however, does he mention Islamic Africa. Yet pottery lamps continued to be used, and manufactured, for many centuries after the collapse of Byzantine power. Recent excavations at Bir Ftouha on the edge of Carthage have produced new evidence for a variety of ceramic lamp types in use in the late 10th and early 11th c. (fig. 7).44 But what about the centuries in between? What kind of lamps were in use in the region in the 8th and 9th c.? At the moment those centuries remain frustratingly obscure. I began with a quote from Saint Augustine of Hippo, in which the bishop uses the image of an unfilled oil lamp as a metaphor for an ungenerous patron. Augustine continues the analogy with a further image of the patron's house filled with numerous lighting devices: Nam si aliquis infinitas divitias possideat, et non habent charitatem, sic est, quomodo si habent plures lucernas et multas lampades, pinguedinem non habentes.*?

For if a man possesses infinite wealth but lacks generosity, it is as if he has numerous pottery lamps and many bronze lamps which lack fuel.

There are many other places in Augustine's writings where lamps are mentioned, sometimes

metal lamps belonging to churches, sometimes pottery lamps used in houses.f6 If we can believe a remark made by Augustine's biographer Possidius, the lamps in Augustine's own house were

simple clay objects; the only hint of luxury in the house was a set of silver spoons.?’ Of course we cannot know for certain what type of pottery lamps these were. Nevertheless, the kind of meticulous classification represented by Bonifay's new study (and equally by Bussiére's new publications of lamps from Algeria) allows us to narrow down the possibilities to a few well-

dated types.*? The study of N African lamps has advanced a long way since the pioneering work of Deneauve, but there is much more to be done, especially on lamp types and lamp production in regions well away from the Carthage area; lamps in other parts of Tunisia, and indeed in other parts of N Africa, need to be studied with equal intensity. At the same time, we need to move

beyond the study of lamps simply as typological artifacts, remembering that they were valued household objects which played an important róle in the daily — and nightly — life of the N African population. [ am grateful to Nichole Sheldrick for her help in preparing the lamp illustrations.

43 44

ForELibya, in addition to D. M. Bailey's landmark study of the lamps from Benghazi (supra n.7), there is a short catalogue of the material from Tolmeita: E. Fabbricotti, Catalogo delle Iucerne di Tolemaide (Cirenaica) (BAR $962, Oxford 2001). S. T. Stevens, A. Kalinowski and H. vanderLeest, Bir Ftouha: a pilgrimage church complex at Carthage (JRA Suppl. 59, 2005) 519; J. Rossiter, P. Reynolds and M. MacKinnon, "A Roman bath-house and a

group of Early Islamic middens at Bir Ftouha, Carthage,” Karthago (forthcoming). 45 46 47

August. Sermo 107.2 (= PL 39 col. 1958). Eg., Ep. 3.1 (= PL 33 col. 0061): legi enim litteras tuas ad lucernam iam cenatus. Contra Cresconium 3 (= PL 43 col. 0513): ibi protulit Silvanus capitulatam argenteam et lucernam argenteam. Possidius Calamensis, V. Sancti Aurelii Augustini 22 [= PL 32 col. 0052]: cochlearibus tantum argenteis utens, caetera vasa quibus mensae inferebantur cibi, vel testea, vel lignea, vel marmorea fuerunt ('he only

had spoons made of silver; the other vessels in which food was served at his table were made of pottery, 48

wood, or marble’). At the time Augustine was writing, ARS lamps of Atlante Form VIII (= Bonifay Types 43-51; Bussiere

Types EVIII, 7-30) were in wide circulation. Bussiere lists 74 examples from the museum at Hippo. There was also a local lamp industry at Hippo, for which see Bussiere 2007 (supra n.9) 171-77, Types EXI, 47-55.

Strong local production in Tunisia: supplementing Bonifay from the case of Leptiminus Karen E. Carr Introduction

The pottery from Leptiminus is not yet fully studied: to date, only the pottery from the East Baths and some interim reports are published. The pottery from the survey portion of the project will shortly appear in Leptiminus 3, but as yet very little of the stratified (excavation) material from the 1st to 4th c. A.D. has been prepared for publication. Because the survey material was unstratified, and because the East Bath stratigraphy begins with the 5th c., there is currently limited chronological data available.! Thus, while we cannot yet make many comments on the dates proposed by Michel Bonifay, drawing upon the studies already done at Leptiminus by the late J. Dore, J. Moore, T. Allen, as well as my own, does allow us to describe some previously unrecognized forms and to point out some distribution patterns.? As Bonifay writes (487), "la céramique africaine ... présente à la fois une homogénéité très affirmée et de forts particularismes régionaux". Here the principal emphasis will be laid on the extremely local production of many forms. Nabeul has very different pottery from Leptiminus; even Mokhnine,

Henchir ech Chekaf and Uzita, all much

closer to Leptiminus,

yield

different pottery forms, and different proportions of similar forms. Despite the widespread trade in pottery all across Tunisia, both inland and along the coast, and despite the enormous international trade in pottery, both exported from and imported to Tunisia, and despite Leptiminus' status as an important producer of export pottery, most of the pottery used in and around the town consisted of what was not exported. As Bonifay states (449), "ces objets avaient été avant tout produits pour les Africains, pour leur approvisionnement en denrées de premiere nécessité, pour leur table, pour leur éclairage". Much of this locally-used pottery was also locally manufactured, in forms unique to this town or that. Because of this multitude of unique local forms, comprehensive

catalogues

of Tunisian

forms,

such

as Bonifay's, will inevitably

become more and more clogged with wares that were produced only in one small area, never left that area, and are perhaps of interest only to archaeologists working in that area. Nevertheless, many N African wares that were never exported across the Mediterranean found wide markets along the N African coast or inland toward the desert, and these will reward further

study. The new forms of pottery thus far found at Leptiminus and not mentioned by Bonifay, or whose identification needs adjustment in light of the finds from Leptiminus, include slipped tablewares, lamps, cooking wares, and common or unslipped bowls and basins, each of which will be considered here (I do not intend to address the flagons or amphorae). As always, I rely on the pottery classification system for Leptiminus devised by the late J. Dore. I have endeavored to adhere to his system, even though his loss has made the pottery work inestimably more difficult than it would have been with his continued expert help.? At Leptiminus, the surface survey project directed by D. J. Mattingly and D. Stone with N. Ben Lazreg included an urban survey, which covered 125 ha atop the ancient town and its suburban districts, and a rural survey, which covered a further 10 km? of the town's immediate

hinterland. The rural survey noted several small habitation sites, including rural sites 501 and 502, both farmsteads occupied from the 1st to 7th c. Rural site 520 is another possible farmstead

1 2

All dates cited are A.D. Iam grateful to J. Dore, T. Allen and J. Moore for allowing me to make use of their as yet unpublished contributions to Leptiminus 3, and to J. Moore for use of her unpublished concordance of the Leptiminus pottery.

3

Among other issues, I lack illustrations for some subtypes and variants that Dore describes. I have filled in a few of the gaps with some of Dore’s preliminary drawings that were available to me, but for others the listed comparanda will have to suffice.

106

K. E. Carr

with 2nd-7th c. material. Rural site 522 might have been a small village; the material again starts in the 1st c. but is most abundant in late antiquity (Stone and Carr, both in Leptiminus 3).

I was responsible for the pottery and lamps from the rural survey, including the rural sites, while J. Dore was responsible for the pottery from the urban survey, and T. Allen is publishing the lamps from the urban survey (all appearing in Leptiminus 3). N. Ben Lazreg supervises the site for the Institut National du Patrimoine and in his own right has carried out an important series of excavations at cemeteries, baths and domestic structures with some remarkable mosaic floors. The joint excavations directed by N. Ben Lazreg first with J. H. Humphrey and then with L. Stirling and S. T. Stevens have investigated a variety of different sites around Leptiminus, five of which are mentioned in this chapter. Site 10 was a cemetery in use between the late 2nd and early 4th c. (see Leptiminus

1 and 2). The East Baths (Site 1) were built in the

late 2nd c. and re-used for industrial purposes in late antiquity (Leptiminus 2). Site 290 included a number of industrial kilns dating from the 1st to the late 3rd c. Site 270, not far from Site 290,

was probably a clay-processing plant associated with the same industrial pottery production and in use during approximately the same period. Finally, Site 304, lying between Site 290 and Site 270, was a burial complex which was in use between the late 2nd and the 5th c. (Ben Lazreg et al. 2006). Dore has published the pottery from Site 10 (Leptiminus 2). I am publishing the pottery from Site 270 and the slipped tablewares and casseroles from Site 290, while J. Moore is responsible for the unslipped pottery from Site 290 and the pottery from Site 304. Slipped tablewares As Bonifay remarks (69), Leptiminus did not produce African Red Slip fine tablewares for

export, so the finds from Leptiminus will add little to our (incomplete) information regarding the production sites of these wares. The Leptiminus forms that have sometimes been described as African Red Slip (henceforth ARS) were actually slipped cooking casseroles rather than tablewares, and will be dealt with here among the casseroles. We confirm Bonifay's statement that no true ARS tablewares emerged from the Leptiminus kilns so far excavated, and we can add that none of the lamps found seems to have been produced locally (Carr, Leptiminus 3; Allen, Leptiminus 3). On the other hand, a number of local imitations of ARS tablewares were produced in the Leptiminus kilns. Without having any locally-produced ARS forms, we can still venture some comments regarding the distribution patterns at Leptiminus of ARS produced elsewhere. Indeed, a rather

distinctive pattern emerges from the survey material. It is clear that very little ARS tableware and very few lamps came to Leptiminus before about 200. For no obvious reason there was then a substantial increase in both the amount of imported ARS tableware and the number of lamps. The increased numbers of both continue through the 4th into the 5th and 6th c., and even into the 7th c., to the very end of the production of ARS ware when glazed pottery was developed in c.800. The paucity of slipped tablewares at Leptiminus in the Ist and 2nd c. contrasts with the abundance of these forms in the same period at Carthage and other more northern Tunisian sites. Presumably this is mainly because the early ARS forms were produced further north. We do not yet know the manufacturing site of ARS A wares, the first wares to be produced, but Bonifay (47) plausibly suggests the Mejerda river valley or perhaps somewhere along the Algerian coast. He writes (451): "la vaisselle semble circuler avec moins de facilité [que les amphores] sur le territoire africain ... Ainsi, les categories de sigillées africaines apparaissent paradoxalement plus varies à Marseille qu'à Carthage".

If it was difficult or expensive to acquire the early forms of ARS tablewares as far south as Leptiminus, despite its location ask why this did not stimulate forms. Beginning in the late 1st tion elsewhere, Leptiminus had the 2nd c., if not earlier, these

on the coast where ships might easily bring cargoes, we might a local industry producing local imitations of the early ARS c., at just about the same time as the beginning of ARS producindustrial kilns in operation (Site 290). From the beginning of kilns produced both good-quality ARS cooking casseroles and

ordinary unslipped tablewares. What would have been simpler than to produce imitations of

Strong local production in Tunisia: the case of Leptiminus

— 107

ARS tablewares too? Yet the Leptiminus kilns do not seem to produce imitations of ARS tablewares until they have more access to ‘true’ ARS forms a hundred years later. By the beginning of the 3rd c., those same Leptiminus kilns did begin to produce imitations of ARS tablewares. These local forms are not mentioned in Bonifay's catalogue, but they merit a mention here for two reasons: first, in order to help complete a catalogue of Roman pottery from Tunisia, and second, to illustrate the problems inherent in producing such any such catalogue. In whichever corner of the ancient world one works, one finds innumerable locally-produced forms,

differing from each other only in small ways, and sometimes only in the fabric. Rather than categorizing smaller and smaller distinctions, ceramicists must begin to ask themselves what their purpose is in devising classification systems (cui bono?), and what the detailed classification of local forms can tell us either about the pottery industry or about the local sites themselves. One might be tempted to surmise that when the inhabitants of Leptiminus were at their

most prosperous in the 2nd and 3rd c. (the height of the kiln activity), they used metal or glass tablewares and lamps, and that the rise of imported ARS tableware corresponded to a decline in the pottery industry there and perhaps a more general impoverishment of the town as a whole. According to this view, economic conditions will have forced people to buy pottery to eat from rather than metal or glass. It will be interesting to see whether there is a corres-

ponding decrease in the finds of metal and glass at about this time. The earliest ARS tablewares to arrive at Leptiminus in significant quantities were still probably manufactured, as Bonifay says, mainly to the north of Leptiminus. In the 2nd and 3rd c., Leptiminus seems to have obtained its high-quality slipped tablewares mainly from Sidi Khalifa, south of Hammamet

near the coast, and from Oudhna.

The earliest of these forms are

the slipped bowls ARS 31 and 32. Both seem to begin production around A.D. 200. ARS 31-Boni-

fay sigillata 11 is, according to him (157), not common in Byzacena and probably produced further west. There were 58 pieces of ARS 31 from the rural survey, however, and about 15 from the urban survey, so perhaps this form is more common in Byzacena than Bonifay supposed. These earliest imports in quantity seem also to have inspired the earliest imitation ARS forms in a local Leptiminus fabric. This is ARS 27/31/62=Dore’s F AF 12 02, which may date as

early as the first part of the 3rd c. The form is a wide shallow bowl with a small foot, possibly related in some way to the not dissimilar ARS cooking casserole 181 that was produced at Leptiminus at the same time and in the same kilns (see below). Perhaps confirming a pent-up demand, even this first venture in imitating ARS ware is remarkably common at Leptiminus: (nearly 300 examples from the urban survey alone). These vessels imitate forms whose manufacturing locations remain unknown but are probably to be sought in N Tunisia or Algeria. But we cannot be certain that these vessels appear as early as all that since the pieces are too fragmentary to be sure exactly what form they represent. If these tablewares are not so much ARS 27 as ARS 62 for the most part, then they date later, to the end of the 4th or beginning of the 5th c., when ARS ware was at its most common at Leptiminus, and the presence of imitations would be less interesting. These ARS 31 bowls are soon followed by ARS 50 (there is no corresponding number in Bonifay), which began production around A.D. 230. Like ARS 31, ARS 50 is a wide shallow bowl, again not too different from the cooking casserole ARS 181 that was a principal product of the Leptiminus kilns. At least some of the ARS 50 bowls were probably produced in central or S

Tunisia, as the examples at Leptiminus are in a C/E ware typical of that region, with a lustrous dark slip on both the interior and the exterior (see Dore, Leptiminus 3). C ware is probably produced southwest of Kairouan, at Sidi Marzouk and other sites, while

E ware seems to come

from even further south. As with the earlier ARS 31, there is a locally-produced imitation of ARS 50 at Leptiminus (not mentioned by Bonifay), of which 112 sherds were found in the urban survey. Both ARS 31 and ARS 50 appear at Leptiminus well before the date when its known kilns cease production near the end of the 3rd c. Possibly these early ARS imitations were produced in the kilns that we have excavated, though I have not seen any misfires to confirm it, so perhaps they were produced elsewhere in the territory of the town. Like the 5th-c. ARS

108

K. E. Carr

imitations at Nabeul, from a rural kiln at Sidi Zahruni, could these imitations have come from outside the main urban area?

There is considerably less imported slipped tableware at Leptiminus in the 4th c. than there was in the 3rd c. — or at least much less of it reached the suburban periphery. No imported slipped tableware from the 4th c. turned up in the rural survey. On the other hand, Dore

found a good deal (146 sherds) of ARS 58a among the material from the urban survey. ARS 58a is a slightly deeper wide bowl with a flat everted flange rim. Of the ARS 58a bowls from the urban survey, he determined that about 2076 of the sample was fabricated in C ware, the rest in E ware. These wares are probably produced in central or S Tunisia. Thus these 4th-c. imported slipped tablewares seem to continue the trend from the later 3rd c. for imported ARS tablewares to reach the town from central or S Tunisia, rather than from the north.* During the 4th

c., there is no evidence of local imitations of slipped tablewares. It is not yet clear whether or how this fact, or the overall decline in imported ARS tablewares in this period, is related to the end of production at the known kilns. The amount of imported slipped tableware picks up again in the early 5th c. with some ARS 61A-Bonifay sigillata 37 (65 sherds from the urban survey) and ARS 67-Bonifay sigillata 41 (39 sherds from the urban survey). Once again these are wide, shallow bowls, but now with more complex rims: ARS 61 has a vertical or slightly incurved rim, flattened on the outside so that it has a more or less triangular profile, while ARS 67 has a broad horizontal rim that

rises in two stages. But these ARS 61 and ARS 67 bowls do not originate from either of the two production centers that supplied Leptiminus in earlier centuries. The 5th-c. ARS forms come instead from a

third location. Both ARS 61 and 67 were probably manufactured in NW

Tunisia,

west of Carthage. M. Mackensen has proposed that ARS 61A originated in the valley of the Medjerda.? ARS 67 is also a northern form from El Mahrine and probably from other northern production centers too.? It seems most likely that both ARS 61A and 67 would have traveled to Leptiminus by ship along the coast. A fair amount (138 sherds from the urban survey) of a variant of ARS 67 known as ARS 68 (no number in Bonifay) also appears at Leptiminus. ARS 68 is roughly contemporary to ARS 67 and similar except in having a slightly different, more compressed rim. ARS 68 bowls may also

originate in the northwest of Tunisia, though Bonifay (51, 199) notes variants of this form in the sigillata E fabric from S Tunisia as well, and a local imitation made at Sidi Jdidi near Hammamet. 5maller quantities of ARS 91-Bonifay sigillata 49, 50, 51, probably issuing from northern production zones, also appear at Leptiminus, 51 pieces in the urban survey, 10 in the

rural survey. Some at least were in a Carthage D ware (see Dore, Leptiminus 3). Like the other ARS tablewares imported to Leptiminus, ARS 91 is another wide, shallow bowl, but now with a hooked flange below the plain rim, as is often the case with 5th- and 6th-c. African bowls. From the later 5th c. there were 54 sherds of ARS 87-Bonifay sigillata 45 from the urban survey. The production sites of ARS 87 are still poorly understood but Bonifay suggests (175) that it may also be a NW Tunisian form (a northern variant of the central Tunisian form ARS 82), or even Algerian. ARS 87 is

a somewhat

deeper bowl with a thickened rim that varies in

its precise shape; it is a precursor of the much more common form ARS 88. Thus most or all of the ARS tablewares found at Leptiminus and dating to the 5th c. seem to come from northern production centers. As in the 4th c., there do not appear to be local imitations of any of these forms at Leptiminus. Much as one might like to draw some connection to the dramatic political changes that visi-

ted N Africa in the 5th c., it is difficult to see any correlation. While one might suppose that the Vandal invasion of N Africa and capture of Carthage in 439 would interrupt local trade in tablewares

rather than facilitate it, we

find a vibrant

trade in tablewares

between

the Car-

thage area and Leptiminus all through the 5th c. As S. Gutiérrez warned in the case of Visi4 5 6

Carr 2007, 598. Mackensen 1993, 321. Mackensen 1993.

Strong local production in Tunisia: the case of Leptiminus

109

gothic Spain, "le commerce et les relations politiques peuvent suivre des voies différentes" (Gutiérrez Lloret 1998, 557 and n. 43, quoted at Bonifay 449). With the beginning of the 6th c. the production centers feeding ARS tablewares to Leptiminus changed again. Both the urban and the rural surveys yielded large quantities of ARS

88-Bonifay sigillata 46, a very shallow, wide bow! with a thickened rim. Bonifay (175) sees ARS 88 as being produced at Sidi Khalifa, near Hammamet to the north. Perhaps not coincidentally, this is the same center from which Leptiminus imported most of her slipped tablewares in the 2nd and 3rd c. ARS 88 was also being produced at the small workshop of Chougafiya, even closer to Leptiminus, and this may account for the large quantities of ARS 88 found at Leptiminus (Bonifay 49). Perhaps this explains why the fabric of most of the ARS 88 found at Leptiminus is a coarse sandy brick red, while Bonifay (175) describes the Sidi Khalifa production as fine and orange. Alternatively, this could be another example of local imitation of an ARS form, but if so there were no wasters (misfired sherds). The large bowl ARS 99-Bonifay sigillata 55 also predominates among the slipped tablewares of the 6th c. at Leptiminus. ARS 99 bowls differ from the earlier bowls in being much narrower and deeper and having a heavy rolled rim. The fabric is coarse, with a thin matte red or orange slip. Their main production center is certainly Oudhna, which was one of Leptiminus' old supply centers in

the 2nd and 3rd c. A fair amount (99 sherds come from the urban survey) of ARS 103 also appears at Leptiminus. ARS 103, another deep bowl with a thickened rim, may also have been produced

at Sidi Khalifa (Bonifay 203), but again there may well be some local imitations. Thus, all of these 6th-c. forms will have been imported to Leptiminus from the same centers as in the 2nd and 3rd c., with not much coming either from the south-central centers or from the northern centers to the west of Carthage. If we want to look for the effects of the Byzantine reconquest of 533, the shift from northern sources (ARS 61A, 67, possibly 68 and 87) to sources in the Gulf of Hammamet (ARS 88 and 99) may be one.

With this revival of the older trade pattern, local imitations start up again. As at other sites along the coast of Byzacena (Bonifay 205), at Leptiminus there is a good deal of the

locally-made imitation ARS bowl 99/103, namely Bonifay sigillata 84=Dore F AF 42 2.7 From the urban survey came 112 examples of this local imitation of ARS 99/103 (Dore, Leptiminus 3). The local imitation has a thin slip, with brush striations or scratches on the interior. The later 6th c. and 7th c. do not show any decline in the amount of ARS tablewares present at Leptiminus; on the contrary, slipped tablewares continue to be imported in about the same numbers as in the 5th c. There are substantial quantities (331 sherds from the urban survey, 77 pieces from the rural survey) of ARS 105-Bonifay sigillata 57. Again, ARS 105 bowls are wide and very shallow, with a thickened knob on the rim. The fabric is coarse red-orange or orange,

with a thin red slip. J. W. Hayes dates these late bowls to 580/600-660+, and Bonifay would probably extend the form at least to 700. There are also 19 sherds of ARS 109=Bonifay sigillata 60 from the urban survey; production of this late form (another shallow bowl, this time with a

plain rim) probably extends into the early Islamic period. Thus imports of slipped tablewares to Leptiminus ended only with the end of all ARS production, and the end does not appear to be due to any slacking of demand on the Leptiminus side. Bonifay does not suggest a production center for ARS 109. The production center of the common form ARS 105 remains uncertain, but he proposes (185) that at least some variants originated at Henchir ech-Chekaf near Ksour Essaf, on the coast south of Leptiminus. Again in the late 6th, as in the 4th c., the source of the imports shifted from north of Leptiminus to the south, perhaps indicative of a re-orientation of Leptiminus' connections towards Lepcis Magna and Tripolitania. One waster resembling 105A from Rural Site 522 indicates local production of an unslipped local variation; there are also 7 unslipped examples of ARS 105B-Bonifay sigillata 57B from the rural survey, though no wasters. These unslipped local imitations of ARS 105A and 105B, not mentioned in Bonifay, represent new forms, even if not very surprising ones.

7

Bonifay et al 2002, 135; Dore 2001, fig. 1.64, no. 39, "local".

110

K. E. Carr

Lamps

Just as no ARS tablewares seem ever to have been produced at Leptiminus, no lamps seem to have been produced there either. The Roman lamps found at the site were all imported from other Tunisian production zones or (in the case of some early lamps) from overseas. We still do not know much about the production centers of earlier Tunisian lamps, but it is instructive to compare what we do know about the sources of the lamps with the town's sources of contemporary ARS slipped tableware. For the later period when we are better informed, the comparisons become more interesting. The lamps collected with the survey material fall roughly into three groups. The first contains lamps of Deneauve VIIA and B and VIII and of Bailey types O, P and Q. These were made between roughly A.D. 100 and 300, some imported from Italy, the coarser ones made in Tunisia. From the rural survey 12 lamp fragments fell into this group (Carr, Leptiminus 3), and about 50 from the urban survey. From the urban survey, 15 lamp fragments definitely belonged to the Deneauve VIIIB type with a heart-shaped nozzle and a garland on the rim (-Bonifay lamp 11) (Allen, Leptiminus 3). Bonifay (80) sees some of these Tunisian lamps as originating in the factories of Byzacena while others are from NW Tunisia. Presumably the ones from Leptiminus ought to belong to the Byzacena group. He does not propose a production site for the Deneauve VIIIB lamps, but their relative numbers at Leptiminus may argue for a production site in Byzacena. As there is little in the way of imported slipped tableware from this period, and the inhabitants of Leptiminus seem to have been using locally-made, unslipped bowls and dishes, it seems reasonable to think that they may also have been using locally- or at least regionally-made lamps. A second group contains early ARS ware lamps or transitional ARS forms such as Atlante IIV and VII (=Bonifay lamp types 38 and 42). This group begins production in the late 3rd c.,

probably in central Tunisia (Bonifay 313). This corresponds well with the 4th-c. imported tableware ARS 58A which also originated in central Tunisia. As with the ARS, there is not much in the way of lamps that can be dated to the late 3rd or 4th c. from the rural survey, which yielded only 2 lamp fragments in this group (Carr, Leptiminus 3), but the urban survey yielded more than 50 fragments (Group 4 in Allen, Leptiminus 3).

The third group of lamps found in our survey material is of the form Atlante XA /Hayes Type II. The large majority of the lamps fall into this group. The urban survey yielded 19 examples of Hayes II lamps with a heart-shaped motif on the border, and over 100 Hayes II lamps with other similar motifs (Group 6 in Allen, Leptiminus 3). From the rural survey there were another 40 lamps in this group (Carr, Leptiminus 3). They date to c.400-700. Thus three centuries of late antiquity yielded about 160 lamps or fragments, while three centuries of the High Empire yielded only half that number. As with the ARS tablewares, so with the lamps: about twice as much fineware and about twice as many lamps was being imported to Leptiminus in late antiquity as during the High Empire. Cooking wares The kilns excavated thus far produced, in addition to amphorae, mainly slipped and unslipped cooking casseroles. The merchants of Leptiminus exported these cooking casseroles and their lids both within Tunisia and across the empire. This is therefore the area where we might expect the Leptiminus excavations to have the most to contribute to a general catalogue

of Roman pottery from Tunisia and to Roman pottery studies in general. Indeed, we can propose several new forms or subtypes and some chronological refinements for forms that have previously been recognized. Most of the cooking wares produced in the Leptiminus kilns were made between A.D. 100 and 300. They consist largely of slipped ARS 181 and 183/184 casseroles and their slipped ARS 182 and ARS 185 lids. As Bonifay says, "les formes 181-182 et 184 sont omniprésentes en Byzacene

... et, de maniére générale, sur tout le littoral oriental de la Tunisie" (67);

"il apparaît que la production de céramique culinaire de Leptiminus était particulièrement importante: Hayes 181, 182, et 184" (69). For the ARS 181 casseroles (the most common products of the excavated kilns), we can support Bonifay's identification of a straight-sided variant (fig. 1.1) (Bonifay cookware 4/Hayes

\

\ €

Strong local production in Tunisia: the case of Leptiminus — 11

2

4C

3

1

D

| 5

Ut

Zu

—

6

i 7

8

9

22

10 11

)

|

4

— 13

5cm 12

—=

m

Fig. 1. Casseroles. 1972, 200/Atlante

I, 215). In the urban survey

110 sherds

of this straight-sided variant were

found, and 26 more in the rural survey. The fabric is red and somewhat coarse, with limestone inclusions. There is a thin orange slip, sometimes with a yellow band. This straight-sided variant of the ARS 181 casserole was common in the material from the kiln site 290 where it was produced. While the survey material cannot contribute to questions of chronology, the

contexts in which this form is found in excavation at the site confirm Bonifay's notion that the straight-sided variant should be contemporary with the curved ARS than earlier as has been

thought

(cf. Dore,

Leptiminus

181 casseroles, rather

2, 77; id., Leptiminus

3). The

strati-

graphy of the kiln excavations is not yet fully studied, but at present the production of the straight-sided variant of ARS 181 casseroles seems likely to be contemporary with the other early variant, Bonifay cookware 5A (see below). Both forms seem likely to begin production in the last quarter of the 1st c. This would move the start of production of the straight-sided variant of the ARS 181 casserole somewhat earlier than Bonifay suggests.

The Leptiminus excavations and surveys yielded many examples of other variants of the ARS 181 casserole that are listed in Bonifay's catalogue. The first variant, Bonifay cookware 5A (fig. 1.2), has a very slightly curved rim; there are 3 examples from the rural survey. It probably begins production in the last quarter of the 1st c., together with the straight-sided

112

K. E. Carr

variant (see above). The fabric is coarse, orange or beige, with thick red slip to cover any imperfections in the surface. There were a few examples (one from the rural survey) of another variant of the ARS

181 casserole with a rather more curved

rim (Bonifay cookware

5B). The

fabric is generally light orange, with a matching slip. This form might have been produced mainly in the first half of the 2nd c. Everywhere at Leptiminus there were a great many of the classic ARS 181 casseroles-Bonifay cookware 5C, with the C-shaped profile. Well over 1000 examples of the classic ARS 181 casserole were found in the urban survey, and 125 in the rural

survey. The start date of Bonifay's cookware 5C should probably be placed in the late 2nd c. He suggests an end date in the first quarter of the 4th c. but my guess, based on the apparent date of abandonment of the kiln site 290, is that it will be earlier, at the end of the 3rd c. Wasters of all three of these types, as well as the straight-sided variant, were found at the kiln site,

indicating that all were produced there. A

late version

of the

ARS

181

casserole

with

vertical

wall=Bonifay

cooking

ware

5D,

which he dates to the second half of the 4th c., appears in 29 examples from the urban survey. There is little 4th-c. material from the rural survey, and no examples

of this late variant. The

late 4th-c. date falls after the closing of the site 290 kilns, but it was probably

produced

somewhere nearby within the territory of the town. À still later variant of the ARS 181 casserole, Bonifay's cookware 38, has an in-turned rim, beveled inside (fig.1.5). Some pieces have an exterior groove just below the rim, some have a slip band on the exterior (Dore, Leptiminus 3). Some of these late ARS 181 casseroles (if that is what they are) are roughly slipped on the interior and over the rim, while others are entirely unslipped. Bonifay dates this form to the 6th and 7th c. There were 59 examples of this type from the urban survey and 113 from the rural.5 Refinements in the dates of the earlier variants may be expected when the stratigraphy of the kiln site is published. This form was exported all over Tunisia and the Mediterranean region, and continued in production into the 7th c., as Bonifay says (244). In short, casserole production in the region of Leptiminus evidently collapsed, for unknown reasons, around A.D. 300, but revived around 400. The other main type of ARS ware casseroles produced in the Leptiminus kilns was ARS 183/184. These cooking vessels are smaller than ARS 181 casseroles, with a strongly up-tilted, thickened flange rim, often with a lid locator on the inside of the rim. The fabric is generally

sandy and red; the casseroles are slipped on the exterior only. The ARS 183 casseroles are distinguished from the ARS 184 ones mainly by their generally larger size and the absence of

slip. These ARS 183/184 casseroles fall into roughly the same period (2nd and 3rd c.) as the ARS 181 casseroles, which was when the kilns at Site 290 were active. These casseroles are divisible into many sub-types, some slipped, others not, as Bonifay indicates (his cookware 7A, 7B, 7C, 7D, 15-17; cf. Carr, Leptiminus

3, nos.

63-64,

66-72).

The

line between

slipped

and

unslipped types is sometimes blurry: the coarsest of the slipped types are not far off from the finest of the cooking ware types. For instance, Dore's early ARS 184 casserole variant F AF 4 20, as he suggests (Leptiminus 3), is the fineware equivalent of Sabratha 58, but the two forms are

otherwise very similar and the fabrics are nearly identical. It is difficult to say which of the many sub-types of the ARS 183/184 casseroles correspond to deliberate distinctions and which may be accidental. Close analysis of the stratigraphy

from the kiln site may shortly enable us to chart chronological distinctions among these subtypes, and perhaps to detect that different types were produced in different kilns at Leptiminus. Meanwhile, some chronological distinctions can be proposed. If we take the variants in chronological order, Dore's form F AF 4 21 may be an early rare variant of the ARS 184 casserole. Dore's casserole F AF 4 21 is the slipped equivalent of his Sabratha casserole 46.2750, which is dated at Sabratha to the second half of the 2nd c. The slip on the F AF 4 21 variant, 8 _ Unfortunately it had not yet been identified at the time of Dore's report on the East Baths, although the late-antique material from those baths might have been expected to include some examples. From the East Baths the bowls with inturned rims (Dore, Leptiminus 2, nos. 57-58) may be related to this form; they

come from contexts of the late 6th c., and appear to have been locally manufactured. 9

. Hayes 1972, 203.

Strong local production in Tunisia: the case of Leptiminus — 113 however, resembles that of a South Gaulish Samian, perhaps indicating a date in the late 1st c. (Dore, Leptiminus

3). Another

candidate

for an early variant

of tne ARS

184 casserole

is

Dore's C CA 5 00, a small casserole similar to his Sabratha casserole 51 found in contexts of the

late 1st and early 2nd c. Dore's C CA 5 00 was not common anywhere at Leptiminus, but it appears in a context provisionally dated to the 2nd c. at Site 270, the clay-processing structure near the kiln site. The small, fine casserole Dore F AF 4 28/02, which seems to be a slipped version of C CA 7 00 with the same out-splayed flange rim, may be a somewhat later ARS 184 variant that dates to the 2nd and 3rd c. The fabric is gritty orange-brown, the slip a smooth orange; the form was common in the urban survey. The unslipped Dore C CA 7 00 itself (= Bonifay cookware 7C) was very common (325 sherds) in the urban survey (fig. 1.6), though less common (33 sherds) in the

rural. It finds a parallel in Dore's Sabratha casserole 59, dated between the early 1st c. and the late 2nd c. Dore's F AF 4 31 seems to be a larger version of the same general form and so might follow the same chronology; its fabric is gritty or brown, with a brown exterior slip. For a somewhat later variant of ARS 184 casseroles, we might look to the unslipped casserole Dore C CA 6 00 (fig. 1.7). Dore's C CA 6 00=Bonifay cookware 7B, which is possibly to be conflated with Dore's C CA 14 00 (fig. 1.8), seems to date to the mid-2nd to mid-3rd c. based

upon the stratigraphy at Site 270, or to the second half of the 2nd c. based on parallels to

Sabratha casserole 58. Dore's C CA 6 00 was quite common in both the urban (253 examples) and the rural survey (58 examples). The slipped version, Dore's F AF 4 30, might thus also be dated to the mid-2nd to mid-3rd c. Dore (Leptiminus 3) describes the rim of F AF 4 30 as "similar to some

of the Tripolitanian

coarseware

casseroles,

the ones

I think

of as late in the Libyan

Valleys series". The fabric of his F AF 4 30 variant is typically hard sandy or brown, with a brown exterior slip and an unslipped interior. Another small, fine casserole, Dore's F AF 4 24, a slipped version of his C CA 10 00 (fig. 1.9),

is also probably to be dated between the mid-2nd and mid-3rd c. based on the provisional stratigraphy worked out at Site 270 (but see below). C CA 10 (=Bonifay cookware 7 late types?) was not uncommon in either the urban survey (58 sherds) or the rural. The fabric is sandy redbrown, with a smooth red-brown slip. Even the unslipped form, Dore C CA 10 00, is only just one step away from being slipped. This variant finds a parallel in Dore's Sabratha 42 of the 2nd c. Yet another variation of the ARS 184 casserole is Dore’s F AF 4 29, a medium-sized version with a lid locator on the outer face of the rim; the overall section of the rim is triangular (Dore,

Leptiminus

3). His slipped subtype F AF 4 06 seems to be related to his Sabratha 64, which

dates there to the early 3rd c. Thus most of the variants of ARS 183/184 at Leptiminus seem to date to the 2nd and 3rd c.

A later variation on the ARS 184 casserole, however, may be the unslipped casserole Dore C CA 8 00 (fig. 1.10), which seems to be related to the earlier Sabratha 59 but has a shorter rim and more rounded contours. Its date is uncertain. There were no examples of Dore C CA 8 00 at our Site 270 despite the fact that 669 examples came from the urban survey. As the stratigraphy at Site 270 peters out around the beginning of the 4th c., this may be an indication of a later date for this form. Alternatively, Dore's C CA 10 00 (fig. 1.9), dated early above, might

be adjusted to this later date based on a relationship with Fulford and Peacock's casserole 4, dated to the 3rd or 4th c. at Carthage, together with Dore's C CA 9 00 (a smaller version of his C CA 8 00) and his C CA 11 00 (fig. 1.11). These may be related to the Sabratha casserole 67 and/or to Uzita (Van der Werff) pl. 37 no. 13. As the casserole Sabratha 67 dates to the 4th c., Dore's C CA 10 00 might also prove to be a late variant of ARS 184 casseroles. It now seems likely that Dore's classification (Leptiminus 3) distinguishing between his C CA 11 00, C CA 12 00 (fig. 1.12), and C CA 13 00 (fig. 1.13) should be collapsed into a single variant of the ARS 184

casserole, although there are not many examples of any of these from the urban survey. In my own report on the pottery from the rural survey, the casserole Dore C CA 12 00 appears as a

separate undated variant of his C CA 7 00, with only one sherd to represent it, and C CA 13 is considered equivalent to Bonifay's cookware 15 (ARS 183), with 26 sherds, dated A.D. 275-400.

All of these variants are larger versions of Dore's C CA 7 00. As C CA 7 00 may be an early variant of the ARS 184 casserole, these may be somewhat later versions of the same variant.

114

K. E. Carr Most of the ARS 183/184 casserole types from Leptiminus thus date to the 2nd or 3rd c. There

is, however, one slipped subtype of the ARS 184 casserole, Dore's F AF 4 07, which is similar to his Sabratha 70 casserole. The latter is dated to the 4th or even 5th c., and on that basis the F

AF 4 07 variant might also fall in the late-antique period (Moore, Leptiminus

3). This sub-

type would then postdate the closing of the excavated kilns, and would have to have been produced at some different kiln site in the town's territory. No wasters of this form were found at our kiln site; on the other hand, the form is not found in either the urban or the rural survey — usually the best source of 4th and 5th c. material at the site. Lastly, Dore's F AF 4 22 and F AF 4 23, a minor subset of the slipped ARS 183/184 casseroles, may have functioned more as cups or bowls than as cooking vessels. These are small, fine vessels, in a sandy orange brown fabric, with a smooth, slightly lustrous slip on the outside face. Dore's F AF 4 23 resembles his Sabratha 49, which dates to the second half of the 2nd c. (Dore,

Leptiminus 3). Yet there are only a few examples of this form from the urban survey, and none from the rural survey. As casseroles were used for cooking stews and similar dishes, they all needed lids. It is thus no surprise to find that the Leptiminus kilns produced many different sizes and shapes of lids to fit the many different casseroles produced there. However, the most important suggestion regarding cookware lids to emerge from our project is that these casserole lids cannot have served only as lids since they vastly outnumber the casseroles to which they have been thought to correspond. Bonifay does not mention this issue, but it is evident at Leptiminus. The most plausible solution seems to be to envisage these casserole lids serving also as plates, especially since plates are otherwise conspicuously absent from the Leptiminus assemblage. But questions remain: if the casserole lids were also serving as plates, why not produce them in a finer ware, more suitable for the table? Why not make plates entirely different from lids, in different sizes? The apparent versatility of the casserole lids ties in with the multi-purpose use of the smaller ARS 184 casseroles, reminding us that our mapping of the pottery may not correspond very well to the mapping inside the heads of the ancient potters who made them, or to the mapping of the women who cooked with them or the families who ate from them — nor do those mappings all need to have been the same. It seems likely that the categories lid versus plate, cookpot versus tableware, were not so firmly distinguished in the minds of the ancients

as they are in the minds of modern students of their pottery. The ARS 185 casserole lids from Leptiminus include several variants — Bonifay cookware types 6A, 6B, 6C, 6D, with 6C being the most numerous (figs. 1.3, 1.4). From the rural survey

there are 37 examples of the Bonifay 6D lid (Carr, Leptiminus 3, no. 58). If Bonifay is correct in dating this type to the 4th c., and it is the lid for the straight-sided Bonifay cookware 5D, then the greater frequency of this type may indicate that some of Bonifay's 5D casseroles were missed, or that these 6D lids were being used more as plates than as casserole lids. On the other hand, the Bonifay 6D lids may have also served as lids for some other, as yet unrecognized form. We know that the ARS 185 lids, usually thought of as the lids for the ARS 184 casseroles, were also used for the equally common Pupput 1 stewpots (see below). We know this from funerary contexts where these Pupput 1 stewpots appear with the ARS 185 lids still in position on top of them (Moore, Leptiminus 3, citing evidence from Leptiminus Site 304 and elsewhere.) Further small variations, not noted by Bonifay, have been noticed at Leptiminus, though it is too soon to say whether they will prove to have a chronological or some other significance. One common variation of the ARS 185 casserole lid is flat conical, with a small roll at the top of the lip and line-burnishing; these are mostly rather small and may be the lids for the smaller casseroles Hayes 184.1. From the 1st to the 3rd c., Leptiminus' kilns also produced smaller quantities of other cooking casseroles. Some of these casseroles seem to have followed extended developmental trajectories. For example, the locally-made Leptiminus casserole better known as Sabratha 37 (Carr, Leptiminus 3 no. 74; C CA 50 00 in Dore's classification) (fig. 2.1) seems to have begun production

in the late 1st c. and then perhaps evolved into a small version of Sabratha 13 (Dore's C CA 21 00) with a more vertical flange rim, which at our Site 290 Moore

(Leptiminus

3) places in the

late 1st and early 2nd c. This small version of the Sabratha 13 casserole in turn may have evol-

Strong local production in Tunisia: the case of Leptiminus

+

T

115

)

1

2

y 3

/ 4

5

Ó

6

_

7

:

5

5cm

a

|

oe

-

8 Fig. 2. Casseroles.

ved into what is often called the Pupput 1 casserole (=Bonifay cookware 20; Sabratha type 42, 44, 46, 48 casseroles; Carr, Leptiminus 3, nos. 75, 76, 78; Dore, Leptiminus 3,

C CA

1-4, with 461

examples from the urban survey) (figs. 2.2-3) in the 2nd and 3rd c., before ending the series when the kilns at our Site 290 shut down around the late 3rd c. Bonifay does not mention the early variant as Sabratha 37, but it seems likely from the Leptiminus material that this is the early form of his cookware 20. Sabratha 37=early Bonifay cookware 20 casseroles were found in the fill of Kiln F at our Site 290, and these casseroles were one of the principal products of that kiln complex. Moore dates Kiln F to the 1st c., so it seems probable, based on this and the dating of Sabratha

13, that the start date of Bonifay's cookware

20 should

be moved

earlier than 100,

perhaps as early as A.D. 50-75. Beginning in the mid-2nd c., we can subdivide the Pupput 1 casserole into finer categories. There are many variants in different sizes — my type 76 alone can be subdivided into at least 5 sub-variants — which may or may not have functional or chronological significance. It is to be hoped that the stratigraphic information from the Leptiminus kiln site will soon permit us to

make finer chronological distinctions. Three of the sub-forms of the Pupput 1 casserole seem to date to the mid-2nd to the mid-3rd c., judging from the preliminary stratigraphy at our Site 270. These sub-forms are Dore's C CA 2 00=Sabratha 44, Dore's C CA 3 00=Sabratha 46, and Dore's C CA 4 00 and C CA 33 00 (a variant of C CA 4 00), which both correspond to Sabratha

48. A similar form, C CA 4 01, corresponds to Sabratha casserole 49, which dates to the second half of the 2nd c. Dore's C CA 1 01 also corresponds to a Sabratha form dated to the second half of the 2nd c. On the other hand, Dore's C CA 1 00=Sabratha type 42 casserole seems be a later successor, dating to the mid-3rd to 4th c. to judge from the provisional stratigraphy at Site 270. Another casserole with an "orlo bifido" rim as Sidi Jdidi 4 (=Bonifay cookware 31; Carr, Leptiminus 3, no. 65; Dore's C CA 23 00) (fig. 2.4) is produced

over-fired examples Kiln F to the 1st c., probably be moved ware 31 casserole

at Leptiminus, with wasters and

found in the fill of Kiln F at Site 290. Again, if Moore is correct in dating then Bonifay's date for his cookware 31 casserole beginning in 200 should considerably earlier. There is also one possible waster of Bonifay's cookfrom our Rural Site 501, which yielded a good deal of 1st c. material.

116

K. E. Carr

Bonifay's cookware 31 casserole seems to have a similar production run to the Pupput 1 casserole sequence, beginning in the 1st c. and continuing until our Site 290 kilns shut down around the late 3rd c. According to Bonifay, his cookware 31 casserole itself continued to be produced until 400, but perhaps at another kiln site in the territory of Leptiminus. A possible much later suc-

cessor to this type is Dore's C CA 22 00 (fig. 2.5), with a taller, more complex rim. This appears in the East Baths (Leptiminus 2, 83 no. 3), and Moore's comparanda (Hayes 1978, fig. 8 no. 11 etc.) suggest a 7th-c. date based on the coins associated with those comparanda. Even during the 2nd and 3rd c. when the Leptiminus kilns were producing huge numbers of casseroles, there are a few casserole forms found at Leptiminus but apparently not produced there. We do not know why this is. Were there specialized uses for those casseroles? Could the imported casseroles compete on price despite the transport costs? Whatever the reason, what was true of amphorae — examples of Leptiminus' Africana I oil amphora have been found at Sabratha, where olive oil was produced — seems to be true in reverse of casseroles.!? For instance, the 2nd-c. casserole often known as Pupput 2 (=Bonifay cookware 21; Carr, Leptiminus 3, no. 79) (fig. 2.6) was found at Leptiminus in the rural survey in a number of examples, but there

is no evidence of local production. The casserole is a deep, rounded cooking pot with an upright rim and a lid locator on the inside of the rim; its fabric is generally red with a gray surface, or gray throughout. As Bonifay predicted, other casserole forms, despite being commonly exported around the Mediterranean, hardly appear at Leptiminus. The generally common casserole ARS 23, for example, is not so common at Leptiminus (18 sherds from the urban survey, a few more from the rural). This is a slipped, relatively fine casserole (Bonifay cookware 1; Dore's F AF 9 01 and F AF 9 02) with an outward-sloping wall and a slightly rounded bottom; the rim is often folded to the interior. Hayes dated ARS 23 to the 2nd and the first part of the 3rd c.; Bonifay (211) extends those dates in both directions, starting it in the Flavian period and extending it even as far as the late 4th c. He describes ARS 23 casseroles as grouped "sur les sites de la cóte nord, entre le cap Bon et la frontiére algérienne." ARS 23 casseroles are also rare at Uzita and at Nabeul, but common at Carthage and in Algeria (Bonifay p. 67). Presumably these casseroles were being manufactured primarily in N Tunisia or N Algeria, and their rarity at Leptiminus is related to the rarity of contemporary ARS tablewares also originating at northern production sites. Bonifay further suggests (67) that the common casserole ARS 197 and its lid ARS 196 (his cookware 10 and 11) were also made in NW

Tunisia, because their findspots correspond closely

to those where ARS 23 is found and the chronology overlaps. ARS 197 is a deeper casserole with a rounded bottom and a thickened rim with a lid locator on the inside. The exterior is frequently blackened (Hayes 1972, 209). Apparently confirming Bonifay's suggestion, relatively few pieces of the casserole ARS 196/197 were found in either survey or excavation at Leptiminus: as he says (225), the form is "assez peu diffusée au sud de Nabeul”. The closing of the kilns at Site 290 around the end of the 3rd c. seems to have entailed that cooking pots began to be imported to Leptiminus in larger quantities and greater variety from elsewhere in Tunisia, and even from abroad. A wide, handled cup like the Carthage casserole Fulford 4 (Dore's C CA 15 00; Carr, Leptiminus 3, no. 81) arrived. There is just one example from

the urban survey, where Dore agrees that this is "possibly a later Carthage type", but 49 were found in the rural survey. The form dates perhaps to the 3rd-4th c., and at Leptiminus is possibly to be envisioned in the latter half of that range, only after the kilns ceased production. A different possible identification for the form comes from Moore, who sees Dore's C CA 15 00 as identical to his C SB 25 00, a wide, handled cup she compares to Carthage Salammbo 29 (residual in a pre-A.D. 425 context) and the Carthage Circus Late Roman Dish 1 (probably residual as well), but the date, and the underlying hypothesis, would remain the same in

either case. Both the urban and the rural survey at Leptiminus also produced a few examples of a casserole resembling Bonifay's ,

10

4

"Autres formes" Byzacena b (his p. 222). Dore suggests a date

Keay, 1989, fig. 13, 226; Arthur 1982, fig. 1; Bonifay 451.

Strong local production in Tunisia: the case of Leptiminus

|

| >

2

? N

1 Uem M

11

cy

— 117

; |

|

u

12

Fig. 3. Bowls.

in the 3rd-4th c. for this casserole (Dore C CA 30 00b; Carr, Leptiminus 3, no. 83) (fig. 2.8) based on comparisons with the Fulford and Peacock type 5.3 casserole from Carthage. By the 5th c., production of a local late casserole (Bonifay cookware 38) had begun, and it was no longer necessary to import casseroles from elsewhere. Bowls

Two main types of bowls were produced at Leptiminus; we may call them Uzita 1 and Uzita

2 after the town just inland from Leptiminus where they were first identified. Both are described in Bonifay's catalogue, but he does not mention Leptiminus as a production spot. Although these bowls were produced at Uzita, as he says, they were equally certainly produced at Leptiminus. The Uzita 1 (=van der Werff 1982, fig. 7, form 1b; Bonifay common ware 1; Dore's

C SB 13 00, 14 00, 15 00, and possibly 17 00 and 18 00; Carr, Leptiminus 3, no. 97) (fig. 3.1) is a very simple hemispherical bowl with a slightly projecting flange rim. The fabric is generally

sandy reddish-brown, with a dark gray exterior face and sometimes the creamy surface that probably comes from salty water being used in the manufacturing process (whether intention-

118

K. E. Carr

ally or not). Based

on evidence

the late 1st c., just about the form continues into the 2nd form shows some variation: Bonifay specifically mentions

from Carthage,!!

Bonifay dates the earliest Uzita

1 bowls

to

time that the Leptiminus kilns were starting up production. The and the first half of the 3rd c.!? By the late 3rd or early 4th c. the some bowls have a bifid rim, while others have a triangular rim. an example of the triangular rim published by Peacock's survey

from the workshop of Dahar Slima on the edge of Leptiminus.? Dore made no attempt to date the Uzita 1 bowl, but it is very common on both the urban and the rural survey. At our Site 270, the Uzita 1 bowl is fairly common, occurring in contexts of the 2nd to the early 4th c. according to the provisional stratigraphic analysis. It seems most probable that production at Leptiminus of the Uzita

1 bowl

ends with the closing of the Site 290 kilns in the late 3rd c. However,

a

similar bowl with an inverted rim folded back to an everted flange flipped upward at the end (Dore's C SB 18 01; Carr, Leptiminus

3, no. 95) (fig. 3.2) may be a late variant of the Uzita

1

bowl; it may have been produced at some other local kiln site. A similar form has been found at

Henchir ech-Chekaf!^ and these bowls might come from there or be related to that production. The bowl C SB 18 01 is absent from our Site 270, where the stratigraphy peters out around the beginning of the 4th c., increasing the probability that it is a late form. The Uzita 2 bowl, while also clearly produced at Leptiminus in the kilns of Site 290, and again a long-lived form, is much less common on both the rural (5 sherds) and the urban survey (30 sherds)

than the Uzita

1. As with Uzita 1, the earliest version of the Uzita 2 bowl, with

the rim first convex and then concave, definitely begins in the late 1st c. A.D., being derived

from earlier Greco-Roman bowls.* The fabric of the Uzita 2 bowl (=Bonifay common ware 2; Dore's C SB 3 00, C SB 3 01, C SB 3 02; Dore, Leptiminus 1, 145, nos. 40-41; Carr, Leptiminus 3, no.

89; Sabratha 213, 214) (fig. 3.3) is generally red and sandy, typical of the Leptiminus production. Again, Bonifay was unaware that the Uzita 2 bowl was produced in the Leptiminus kilns as well as at Uzita. Moore (Leptiminus 3) mentions wasters and over-fired examples of the Uzita 2 bow! from the fill of Kiln F at our Site 290, demonstrating local production in the late 1st c. and confirming Bonifay's suggestion of a Flavian start date for the form. Two of the exemplars of this form from the rural survey too are somewhat over-fired (Carr, Leptiminus 3). The main period

of the Uzita 2 bowls

is the 2nd c., and Bonifay (245) does not see this form

continuing past the end of that century. Nevertheless, the Leptiminus excavations did yield some possible variants of the Uzita 2 bowl that may date as late as the 4th c. At our Site 270,

the Uzita 2 bowl variant Dore's C SB 3 01 (fig. 3.4), with a more pronounced carination and a heavier, more triangular rim, was found in a context probably of the 4th c. However, sherds of this variant may be residual in this context, as the structure at Site 270 appears to have been abandoned in the late 3rd c. Dore equated his C SB 3 01 with Sabratha 213 (Leptiminus 1, 145), which he dates to the 2nd c. B.C., but his early date is only because, as Bonifay says (245),1° the

Uzita 2 bowl emerges from an older Greco-Roman rim form. At least 5 other types of bowls probably manufactured at Leptiminus were not listed in Bonifay's catalogue. Most of them can be dated to the 2nd or 3rd c., the main production period of our Site 290 kilns. The earliest of these, Dore's C SB 2 00 (fig. 3.5), is a simple bowl with a

plain inturned rim; the fabric is sandy red-brown, usually with a creamy surface. It may be identified with a similar bowl in a similar fabric from Uzita (Uzita pl. 44 nos. 1-2). Despite its simplicity, C SB 2 00 is not extremely common (16 examples from the urban survey, 14 from the rural). While the survey material cannot be dated, the context at our Site 270 hints at a date for C SB 2 00 between the mid 2nd and the end of the 3rd c. Unfortunately, the corresponding Uzita forms were collected in an unstratified context.!” Production of Dore's C SB 2 00 at our Site 11

Hayes 1976a, fig. 17, D1: Late Roman Buff Ware, Bowl type 1?; Ortisi 1999, fig. 12, no. 253-59.

12 13

van der Werff 1982, 142. Peacock, Bejaoui, Ben Lazreg 1989, fig. 22, nos. 27-28.

14 15 16

Nacef 2007, fig. 4 no. 32, etc. Riley 1979, 329-30. Ibid.

17

van der Werff 1982, 394.

Strong local production in Tunisia: the case of Leptiminus

— 119

290 is suggested by a greenish over-fired example from there (Moore, pers. comm.), as well as by some over-fired and poorly fired examples from Site 270 and the rural survey. A second example of a bowl that was probably manufactured at Leptiminus but is not mentioned in Bonifay's catalogue is Dore's C SB 1 00 (as Uzita 25 no. 8; Carr, Leptiminus 3, no. 99) (fig. 3.6), a plain bowl with a plain rim, usually with a groove just below the rim. It seems to date to the 2nd to 4th c. based on the apparent context at our Site 270, but it may be residual here since at Uzita the date seems to be Flavian, based on the other material found in the same level.!8 There are also parallels to Sabratha 156, which Dore dated between the Ist c. B.C. and the 1st c. A.D. Dore's C SB 1 00 is very common on the rural survey, less common on the urban. Evidence for local manufacture of this form is based on several greenish examples from the rural survey (FVNs 77968, 80040, and 80751) and several other poorly-fired pieces. This form may have been produced in our Site 290 kilns or at some other local kilns. Also beginning in the 2nd c., Dore's C SB 6 00 and C SB 7 00 (Carr, Leptiminus 3, no. 100;

compare Uzita 20 no. 31 or 22 no. 24) (fig. 3.7), were both probably produced at Leptiminus, although perhaps not in the kilns of Site 290. These are small bowls with in-turned thickened rims; C SB 7 00 is distinguished from C SB 6 00 mainly by a groove on the exterior below the rim. The fabric for both is red and sandy, a typical Leptiminus fabric, usually with a creamy surface. Dore (Leptiminus 3) suggests that the large number of sherds from Field 52 in the urban survey may indicate production débris; there is also one possible waster from Rural Site 502 (Carr, Leptiminus 3). At Site 270, C SB 6 00 and C SB 7 00 bowls both appear in contexts provisionally dated to the 2nd and 3rd c. Beginning probably in the 4th c., Leptiminus yields a complex sequence of related bowls that seems to stretch into the 7th c. This series, beginning with Dore's C SB 4 00, all seem to be produced at Leptiminus, and all are absent from Bonifay's catalogue. All of the bowls in this

sequence have a sandy, red-brown or brown fabric with a smoothed surface and occasionally a blackened rim. The C SB 4 00 series is not yet well understood, but it seems to begin with the small bowl C SB 4 00 (fig. 3.8). Dore's C SB 4 00, not very common anywhere at Leptiminus, is a small bowl with a flaring wall and slightly up-tilted flange rim with a lid locator on the top surface. Moore (Concordance, forthcoming) suggests that this bowl may find a parallel in the locally-produced casserole Dore C CA 30. If the parallel is valid, the date of C SB 4 00 should fall in the 3rd c. At our Site 270, C SB 4 00 seems to fall into the 2nd-4th c. The bowl Dore C SB

22 00 (fig. 3.9) may be a later variant of the same type. This is a bowl with a thick, slightly up-tilted flange rim. The wide up-tilted flange rim resembles the rims on Uzita 3 basins (see below) and may be related to them; this would suggest a date similar to that of the Uzita 3 basins, in the 4th c. Dore C SB 22 00 bowls do not appear in the Leptiminus rural survey nor at Site 270, an absence that may support the 4th-c. date since there is little 4th-c. material in either collection. Examples were found in a 7th-c. context at the East Baths (Dore, Leptiminus 1, p. 85 no. 20 and fig. 1.66 no. 20), but it may be residual there. A small hemispherical bowl with a flange rim, Dore's C SB 20 00 (fig. 3.10), quite finely made, may continue his C SB 4 00 series. There are a number of examples of C SB 20 00 from the urban survey but none from Site 270 or the rural survey. Their absence from Site 270 tends to suggest a late date. Moore (Concordance, forthcoming) suggests a late parallel for it in the Carthage Salammbo dish/lid 2 (late 4th to late 5th c.) or lid 3 (c.525-575/600). Indeed, C SB 20 00 was found in the East Baths of Leptiminus in a late 6th-c. context (Dore, Leptiminus 1, p. 94 no. 106). If this bowl form is late, then probably so is Dore's C SB 21 00 (fig. 3.11), as they are in

the same fabric and have similar shapes. Moore sees a parallel for C SB 21 00 in the basin Bonifay's common ware 38 (see below), which would date the former around A.D. 550-700. An example of C SB 21 00 from the East Baths of Leptiminus derives from a context of c.450, which

would be too early for Bonifay's common ware 38. I am inclined to think that the similarity in rim shape does not make this bowl into a basin, and that we should perhaps consider C SB 21 00

to be a late bowl in Dore's C SB 4 00 series, paralleling but not identical to Bonifay's common 18

Ibid. 344-45.

r

EL

Vos

|

M

«€

4

A,

Ä

13

—

D

2m

14 Fig. 4. Bowls, mortaria, basins.

ware 38. The variant Dore C SB 21 01 (fig. 3.12), found in a context of the last part of the 5th c. at the East Baths (Dore, Leptiminus 1, p. 95 no. 115), should presumably fall into the same series. A larger version of C SB 20 00 in a similar fabric, which Dore terms C SB 23 00, probably

dates to about the same time, as it was found at the East Baths in an early 7th-c. context (Leptiminus 1, p. 87 no. 21 with fig.1.66 no. 21), where it may be residual. Indeed, it seems hopeless

at this stage of affairs to try to establish a chronological ordering for these subtypes; we can hope that further stratified finds will clarify the development of the series. Local production

of the C SB 4 00 series is indicated by one greenish example from the rural survey (from field 1425) and one probable waster from our kiln site 290 (FVN 88605) (Moore, pers. comm .). Four other currently undated bowls, not mentioned in Bonifay's catalogue, were probably also locally produced. The first of these (fig. 4.1) is a large bowl or basin with a flanged rim that sweeps outward in a shallow and generally horizontal curve (Dore's C LB 27 00; Carr, Leptiminus 3, no. 86). The fabric is red-brown, sometimes gray, and often with a creamy surface. Two over-fired examples from Rural Site 522 and field 1360 in the rural survey indicate local production. The second (fig. 4.2) is a small bowl in a pinkish-brown fabric with a blocky triangular everted flange rim (Carr, Leptiminus 3, no. 90), possibly locally produced to judge by the poor firing of some examples found in the rural survey. A small bowl (fig. 4.3) in a dark red

fabric with an everted uptilted flange rim (Carr, Leptiminus 3, no. 94) also appears to have been locally made, to judge from a single waster found in the rural survey. Fourthly there is a bowl (fig. 4.4) with a flaring rim thickened and turned back to the interior (Carr, Leptiminus 3, no. 96) in the coarse sandy red fabric typical of Leptiminus, often with a creamy surface; this too yielded several possible wasters in the rural survey.

Strong local production in Tunisia: the case of Leptiminus

— 12

A minor bowl type found at Leptiminus and not mentioned in Bonifay’s catalogue is Dore’s C SB 19 00 (fig. 4.5), a less common plain bowl with an "orlo bifido” rim. The fabric is red-brown, sometimes with a yellow surface. It finds parallels in Sabratha 153 and perhaps the Carthage Circular Harbour dish 1. This form may be derived from earlier forms, as Dore dates the Sabratha parallel to the first half of the 1st c.; at our Site 270, however, it comes from apparently mid-2nd to late 4th-c. contexts. It was found in our Site 10 excavations as well as at Site 270 (Dore, Leptiminus 1, 147), but was not found in the rural survey. There is at present no evidence of local manufacture for this type, whose scarcity and fabric in any case suggest that it may be imported to Leptiminus. Finally, the Leptiminus project has yielded 4 new bowl forms not included in either Bonifay's catalogue or Dore's classification and for which we have as yet neither dates nor evidence of local production. The first is a shallow bowl (fig. 4.6) with a flat everted flange rim (Carr, Leptiminus 3, no. 91). The fabric is dark red-brown with gray inclusions which suggest that the form is imported to Leptiminus; some examples have a creamy surface. There is also a smaller, deeper

bowl

(fig. 4.7) with an everted flange rim (Carr, Leptiminus

3, 92). Its

fabric is red and fine, with red inclusions, and a creamy surface. The third form has a flat everted flange rim sagging in the center, is in a coarser pinkish-brown fabric with a creamy surface (Carr, Leptiminus 3, no. 93), and might even be placed among the casseroles. Of the fourth form (fig. 4.8),

a bowl with a horizontal flange rim (Carr, Leptiminus 3, no. 105), we have just 4

examples; it comes in a red fabric with a creamy surface, which might be local. Mortaria

Since most of it was collected in the course of surface survey, the nature of the material studied thus far makes it difficult to comment on the mortaria from Leptiminus. The sherds recovered in the survey did not yield many pieces large enough to exhibit both rims and the basalt grits on the bottom of the bowls. As far as can be determined at present, Leptiminus has yielded very few mortaria that seem to date earlier than the end of the 3rd c., and none of them are locally produced. Near the end of the 3rd c. Leptiminus began to produce its own mortaria (Bonifay common ware 9-Dore's C LB 17 00; Carr, Leptiminus 3, no. 111). Dore's form

(fig. 4.9) is a flanged mortarium with a wide drooping rim, fairly deep. The fabric is orange- to maroon-brown,

sometimes

with

a brownish

creamy

surface.

Bonifay's

idea

(252)

that

this

mortarium might have been produced at Leptiminus amongst other central and southern sites in Byzacium finds confirmation in the presence of an over-fired example from field 1113 in the rural survey. A similar flanged mortarium

(Bonifay common

ware

10; not in Dore's classification; Carr,

Leptiminus 3, no. 112), perhaps descended from the above but with a beaded heel to the rim, may date to the 4th c. The fabric is coarse red or red-brown. There is no evidence for local manufacture of this mortarium. Its dating subsequent to the closing of the excavated kilns might be thought to argue against local manufacture, but its similarity to the preceding form in both form and fabric makes it worth considering whether this mortarium too may have been produced at or near Leptiminus. A third type (fig. 4.10) of flanged mortarium from Leptiminus, Dore's C SB 8 00, does not appear in Bonifay's catalogue. It has an in-turned rim and a short horizontal flange just at the rim's in-turn. There remains some doubt as to whether it is a mortarium and not just a bowl, as Dore (Leptiminus 3) suggests. To judge from the sandy red-brown fabric and sometimes greenish creamy surface, it was probably produced at Leptiminus. Moore (Concordance, forthcoming) suggests that it should be dated late, perhaps related to the Libyan Valleys’ ULVS no. 118 and/or Carthage Salammbo bowl 35, seeing the form as possibly residual in the early to mid7th c. This mortarium is absent from the rural survey, but not uncommon in the urban survey (34 sherds).

As with the casseroles, during the 4th and 5th c. Leptiminus seems to have imported mortaria from the north, from the area around the Gulf of Hammamet. From the 5th c. there is a deep mortarium with a wide flange below the rim (Dore's C SB 11 00=Bonifay common ware 13; Carr, Leptiminus 3, no. 113), which seems to be imported from around Nabeul (Bonifay 255).

122

K. E. Carr

Although there is only one example of this mortarium from the rural survey (fig. 4.11), there are 16 from the urban survey. Its fabric is fairly fine, red, with a red surface. Bonifay dates the form beginning in the first half of the 5th c., not mentioning that it reached Leptiminus. Dore compares it to his C SB 8 00 discussed above (fig. 4.10), but it is not clear whether he thinks the two forms are related; I would incline to think that they are not related, given the differences in fabric. Around A.D. 500, however, local production of mortaria at Leptiminus resumed, and apparently in the 6th and 7th c. Leptiminus once again produced its own mortaria. One of these late mortaria, not mentioned in Bonifay's catalogue, is a locally-produced flanged mortarium (Dore's C SB 9 00/10 00; Carr, Leptiminus 3, no. 114), which is sandy orange, sometimes with a creamy surface. Mortarium C SB 9 00 (fig. 4.12) has a slightly thickened and in-turned rim, with a very small flange low down on the exterior wall. This mortarium was fairly common in both the rural (36 examples from the rural survey, with a waster from Rural Site 522) and the urban survey (38 examples). It is comparable to Fulford and Peacock's fig. 77 type 14. Mortarium C SB 10 00, while closely related to the other, is a shallower subtype with a more pronounced flange, comparable to Bonifay common ware 12A, and on that basis probably dating sometime around the late 5th to the second half of the 6th c. C SB 10 00 is less common than the other; it did not appear in the rural survey, and was not common in the urban survey. Basins

Basins are larger ceramic vessels that might have been used to wash dishes or clothes or even babies, for cleaning the stones out of beans, or for short-term

storage. Because

of their

larger size, heavier weight and utilitarian purpose, basins generally were not traded extensively around the Mediterranean, or even around N Africa. The basins found at Leptiminus fall into three main types, all of which find parallels at the nearby inland town of Uzita and each of which follows a very long developmental sequence. From their original identification at that site by J. van der Werff, the three main types of basin from Leptiminus are referred to as

Uzita 1, Uzita catalogue, but and, as with Leptiminus as

2, and Uzita 3. The sequence for each of these types is laid out in Bonifay's the Leptiminus material can add some details to the development of the types the bowls, also allow us to add that they were all as certainly produced at they were at Uzita.

According to Bonifay (263), based on stratigraphic data from Pupput and Nabeul, the Uzita 1 basin (Bonifay common

ware 20; van der Werff 1982, pl. 68; Dore's C LB 13 00 and C LB 6 00;

Carr, Leptiminus 3, nos. 119 and 120) makes its first appearance in the beginning of the 2nd c. and goes out of production around the end of the 3rd c. Dore might begin the Uzita 1 form rather earlier than Bonifay does, based on a parallel for his C LB 6 00 at Sabratha (Sabratha 51) that starts there in the early 1st c. (Dore, Leptiminus

1, 148). The early Uzita 1 variant (fig. 4.13) is

a shallow basin made oblong by adding hand-made plaques between the halves of a round wheel-turned basin. The basin has a projecting flange rim with a lid locator on the top face. The fabric of the Uzita 1 basin is very variable, but at Leptiminus it is generally the usual red and sandy Leptiminus fabric, with the typical creamy surface on the exterior. There were 59 examples of Dore's C LB 13 00 basin from the urban survey and 20 from the rural, while 14 examples of Dore's C LB 6 00 came from the rural survey. This form also appeared at the kiln

site 290 (Moore, Leptiminus 3).1? One example of Dore's C LB 6 00 from Rural Site 520 is a waster with a brown, bubbled surface, indicating local production. We do not know whether, like the casseroles, bowls and mortaria, the Uzita 1 basin goes out

of production when the kilns from Site 290 cease production in the late 3rd c. However, in the 4th and 5th c., the Uzita 1 basin does start to be made (fig. 4.14) in a darker red, coarse sandy fabric (Bonifay common

ware 23; Carr, Leptiminus 3, no. 129). On some vessels the creamy sur-

face is still present but thin; on other examples there is no creamy surface at all. Either basins 19

In her Concordance (forthcoming), however, Moore suggests that C LB 6 00 may instead be a hand-made oval basin.

Strong local production in Tunisia: the case of Leptiminus

123

—. 23

M

TS

) 6

5cm -——— mm

Fig. 5. Basins.

for use at Leptiminus were being made at another nearby kiln site, as yet undiscovered, or they were being made as far away as Uzita. At present there are no wasters of this form from Leptiminus. The early Uzita 2 basin (Bonifay common ware 21; van der Werff 1982, pl. 45, no. 4; Dore’s C LB 1 00, C LB 2 00, C LB 4 00; Sabratha 277; Carr, Leptiminus 3, no. 122) is a very deep round or

oval basin, with a wide, strongly up-tilted flange rim with a convex top face. Some examples (Dore's C LB 1 00) have a groove on top of the flange (fig. 5.1). Dore's C LB 2 00 differs from his

C LB 1 00 in lacking the groove. The fabric in both of these variants is sandy red-brown or orange-brown, often with a creamy surface. Some poorly-fired pieces from the rural survey may indicate local manufacture. The Uzita 2 basin may begin production at Leptiminus a little later than the Uzita 1 basin, around the middle of the 2nd c. There is confirmation of production of the Uzita 2 form at Leptiminus kiln site 290 (Moore, Leptiminus 3). In the rural survey there were 36 examples of Uzita 2 basins, and more than 200 in the urban survey. On the other hand, the Uzita 2 form does not appear at Leptiminus site 270, despite the fact that Bonifay dates it to the 2nd c. and the first half of the 3rd c. We

also have a

local variant (fig. 5.2) of the Uzita

2 basin with a short flange at the point where the rim meets the wall (128 in Carr, Leptiminus 3, no. 128).

124

K. E. Carr By the 4th c., the Uzita 2 basin (Bonifay common

ware 21 late variant; Carr, Leptiminus

3,

no. 127) returns to a more horizontal flange rim, thickened on the end (fig. 5.3). It is now in a coarse red fabric, poorly fired, with a thin creamy surface. Bonifay dates this form to the 4th and 5th c. Another variant (fig. 5.4) of the same Uzita 2 basin (Bonifay common

ware 24; Carr,

Leptiminus 3, no. 130) (fig. 5.4) has a knob at the end of the everted rim and a sharp carination where the rim meets the wall; the fabric is light red and sandy, with a creamy surface. Also from the 5th c. comes a thicker, coarser variant of the Uzita 2 basin (Bonifay common ware 25; Carr, Leptiminus 3, no. 131), perhaps made at Sullecthum rather than Leptiminus. The fabric is coarse red, some

having

a creamy

surface.

Yet another

late variant

(fig. 5.5) of the Uzita 2

basin (Leptiminus 3, no. 134; possibly related to Bonifay's common ware 27) is creamy all the way to the core, with narrow grooves on the exterior wall. This form seems to date to the second half of the 5th c. while later variants of the Uzita 2 basin reach the end of the 6th c. A third kind of basin, which Bonifay suggests calling Uzita 3, first appears in about the middle of the 3rd c. and does not enjoy as long a run. The Uzita 3 basin (van der Werff 1982, pl. 45, no 3; Dore's C LB 3 00; Bonifay common

ware 22; Carr, Leptiminus

3, no. 123) is, like the

Uzita 2 basin, very deep. The Uzita 3 basin (fig. 5.6) has a curved wall and a wide, nearly vertical flange rim with a very convex top face. The fabric is sandy and variable in color, sometimes brick red, sometimes pink, sometimes creamy all the way through to the core. A greenish cream surface on one example from the East Baths (Leptiminus 2, p. 85 no. 19) may indicate local manufacture. The above chronology is that of Bonifay, and we cannot confirm the chronology of this basin from the unstratified survey material, while it is absent from Site 270 and Site 10; at the East Baths it must be residual in a context of the late 6th c.

Bonifay also suggests that there is another important late basin from Byzacena specifically which corresponds to his common ware 38. The rim of this basin resembles that of the ARS 105 tableware bowl, which is certainly common at Leptiminus. Perhaps some of the coarser 105 rims should really be considered rims for this type of basin. A few less important basins, produced at Leptiminus but not mentioned by Bonifay, may also merit a brief mention. One is a simple plain basin thickened at the rim (Carr, Leptiminus 3, no. 116; not present in Dore's typology) (fig. 5.7) which seems to be datable to the 1st c. to judge by the preliminary stratigraphy at Site 270. The fabric is a sandy red with small black inclusions and a thick creamy white surface. Ten examples came from the rural survey. Epilogue Bonifay's conclusion that we should beware of over-generalizing Tunisian pottery is of the

utmost importance. Most of the pottery produced and used in Tunisia was made locally and used locally. Thus it tends to resist generalization. On the other hand, Bonifay is also correct to say that we need to find ways to organize and systematize all of the local types by recognizing affinities and inter-relationships among the various local types. Equally important, we should resist making facile connections to historical or political events known more from literary texts than from archaeological data. Political events rarely lead directly to economic catastrophe, or vice versa, and the ceramic history of Leptiminus does not march in step with any identifiable political events. Finally, we should always remember that pottery, even the best pottery such as ARS ware, is not the luxury material of the ancient world. Pots were cheap and low-status. When people wished to demonstrate status, they preferred glass or metal tableware. Despite

the attractiveness

of some

status on those who bought them. Thus and 3rd c. could just as well be taken as glass and metal tablewares. Likewise, wares, we do not know whether to take

ARS

tablewares,

they cannot have conferred

much

the dearth of ARS tablewares at Leptiminus in the 2nd a sign of properity — its inhabitants could afford to use when in the 4th c. they began to import ARS tableit as an indication of economic decline, or the opposite.

Acknowledgements

I wish to acknowledge the invaluablé assistance of the several project directors N. Ben Lazreg, J. H. Humphrey, D. J. Mattingly, L. Stirling and D. Stone, and the ceramic studies of M. Bonifay, J. Moore, A. Opait, and especially the late J. Dore, some of whose drawings I have employed here.

Strong local production in Tunisia: the case of Leptiminus

125

References Allen, T. forthcoming. "The lamps," in Leptiminus 3. Arthur, P. 1982. "Amphora production in the Tripolitanian Gebel," Libyan Studies 13, 61-72. Atlante = A. Carandini et al., Atlante delle forme ceramiche I (Rome 1981). Ben Lazreg, N., 5. Stevens, L. Stirling and J. P. Moore 2006. "Roman and Early Christian burial complex at Leptiminus (Lamta): second notice," JRA 19, 347-68. Bonifay, M. 2004. Etudes sur la céramique romaine tardive d'Afrique (BAR $1301, Oxford). Bonifay, M., C. Capelli, T. Martin, M. Picon, L. Vallauri 2005. "Le littoral de la Tunisie: étude géoarchéo-

logique et historique (1987-1997): la céramique," AntAfr 38-39 [2002-3] 125-204. Carr, K. E. in L. Stirling et al. 2001. "Preliminary report from Leptiminus, Tunisia," ECM 40, 170-224. Carr, K. E. 2007. "Late Roman coarse wares and cooking wares from the rural survey at Leptiminus, Tunisia,” in M. Bonifay and J.-C. Tréglia (edd.), LRCW

II. Late Roman

coarse wares, cooking wares and

amphorae in the Mediterranean: archaeology and archaeometry (British Archaeological Reports $1662. ii, Oxford) 597-602.

Carr, K. E. forthcoming. "The pottery from the Rural Survey," in Leptiminus 3. Dore, J. 1989. "The coarse pottery," in J. Dore and N. Keay, Excavations at Sabratha, 1948-1951, II: The finds, Part 1 (Society for Libyan Studies, London). Dore, J. with R. Schinke 1992. “First report on the pottery," in Leptiminus 1, 115-56.

Dore, J. 2001. "The major pottery deposits following the disuse of the East Baths," in Leptiminus 2, 75-99 and "Pottery," in Leptiminus 2, 205-7.

Dore, J. forthcoming. "Fine, coarse, and cooking pottery from the Urban Survey,” in Leptiminus 3. Fulford, M.

1994. "The cooking and domestic wares",

in id. and

D. P. S. Peacock

(edd.), Excavations

at

Carthage, The British Mission, 11,2: The circular harbour, north side: the pottery (Oxford) 53-75. Gutiérrez Lloret, S. 1998. “Il confronto con la Hispania orientale: la ceramica nei secoli VI-VII," in L. Sagui (ed.), Ceramica in Italia, VI-VII secolo. Atti del Convegno in onore di John W. Hayes (Florence) 549-67. Hayes, J. W. 1972. Late Roman pottery (London).

Hayes, J. W. 1976. "Pottery: stratified groups and typology,” in J]. H. Humphrey (ed.), Excavations at Carthage 1975 conducted by the University of Michigan (Tunis) 47-123. Hayes, J. W. 1978. "Pottery report —1976," in J. H. Humphrey (ed.), Excavations at Carthage 1976 conducted by the University of Michigan, vol. 4 (Ann Arbor) 23-98. Keay, N. 1989, "The amphorae," in J. Dore and N. Keay (edd.), Excavations at Sabratha 1948-1951, vol. Il. The

finds (Tripoli) 5-67 and 80-83. Leptiminus 1 = Ben Lazreg, N. and D. J. Mattingly (edd.), Leptiminus (Lamta): a Roman port city in Tunisia. Report no. 1 (JRA Suppl. 4, 1992) 115-56. Leptiminus 2 = Stirling, L. M., D. J. Mattingly and N. Ben Lazreg, Leptiminus (Lamta) Report no. 2. The East Baths, cemeteries, kilns, Venus mosaic, site museum, and other studies (JRA Suppl. 41, 2001).

Leptiminus 3 = Stone, D., D. J. Mattingly and N. Ben Lazreg, Leptiminus (Lamta), Report no. 3. The field surveys (JRA suppl. forthcoming 2010). Mackensen, M. 1993. Die spätantiken Sigillata- und Lampentöpfereien von El Mahrine (Nordtunesien). Studien zur nordafrikanischen Feinkeramik des 4. bis 7. Jahrhunderts (Münchner Beitráge zur Vor- und Frühgeschichte 50). Moore, J., with K. Carr and A. Opait, forthcoming. "Products of the kiln complex at Site 290," in Leptiminus 3.

Moore, J. with J. Dore forthcoming. "Table of concordance for Leptiminus pottery publications," in Leptiminus 4. Nacef, J. 2007. "Nouvelles données sur l'atelier de potiers de Henchir ech Chekaf (Ksour Essef, Tunisie)," in

M. Bonifay and J.-C. Tréglia (edd.), LRCW2: Late Roman coarse wares, cooking wares and amphorae in the Mediterranean:

archaeology and archaeometry, vol. 2 (British Archaeological

Reports 51662, ii, Oxford)

581-90.

Ortisi, S. 1999. "Ein frühkaiserzeitlicher Keramikkomplex aus einem Schóptbrunnen in Insula E 177 West in Karthago," RómMitt 106, 439-93. Peacock, D. P. S. 1984. "Seawater, salt and ceramics," in M. G. Fulford and D. P. S. Peacock (edd.), Excavations at Carthage, The British Mission, 1,2: The Avenue du Président Habib Bourguiba, Salammbo: The

pottery and other ceramic objects from tlie site (Sheffield) 263-64. Peacock, D. P. S., F. Bejaoui and N. Ben Lazreg 1989. "Roman amphora production in the Sahel region of Tunisia," in Amphores romaines et histoire économique (CollEFR 114) 179-222. Riley, J. A. 1979. "The coarse pottery from Berenice," in J. A. Lloyd, Excavations at Sidi Khrebish-Benghazi (Berenice) II (Libya Antiqua suppl. 2) 91-46.

126

K. E. Carr

Stirling, L. M., D. Stone et al. 2000. "Roman kilns and rural settlement: interim report of the 1999 season of the Leptiminus Archaeological Project," EMC/Classical Views 44, n.s. 19.2, 179-224.

Stone, D. forthcoming. "Rural Survey gazetteer," in Leptiminus 3. Van der Werff, J. H. 1982. Uzita. Vondstenmaterianl uit een antieke Nederzetting in Midde-Tunisie (Utrecht).

Supplying Rome and the empire: the distribution of stamped amphoras from Byzacena David

L. Stone

Economic models and N African amphoras Since N African amphoras came to be recognized as prominent imports at many Mediterranean sites, historians and archaeologists have debated the nature of the economy with growing intensity. The influential school of thought associated with M. has largely been concerned with the testimony of ancient authors about economic

western Roman I. Finley values,

especially concerning business transactions and the management of estates.! Regarding trade, Finley and those who

agree with him have supported

the idea that much

of the exchange

which took place was 'redistributive' in nature, and did not conform to modern expectations of rational economic behavior. His position was formed in opposition to the modernizing conclu-

sions about trade reached by M. I. Rostovtzeff in his Social and economic history of the Roman Empire (1957).In particular, Finley argued against the existence of a ‘market economy’ in which numerous goods were shipped for purchase and consumption over long distances.? C. R. Whittaker suggested that long-distance commodity movements could even be explained by the desire of wealthy aristocrats to ship goods from a distant estate to supply other properties they owned? When this school of thought did pay attention to particular products, such as those from Byzacena, it tended to assert that quantities had been significantly overestimated by overzealous investigators. A number of scholars who object to Finley's ‘minimalist’ view of the Roman economy have deployed several arguments against it, including evidence for technological change, the presence

of

a market

economy,‘ and

the effects of taxation as a stimulus

to trade.” A whole

range of goods has been invoked to demonstrate far greater levels of exchange than Finley acknowledged and far greater integration of the Roman economy.? The propositions Finley and others made now appear substantially weakened. Many scholars are now sidestepping this debate and moving on to other concerns such as “consumerism” and the possibility that the

Roman empire harbored multiple "economies". N African exports have at times been touted as evidence of the movement of substantial quantities of goods over long distances but, with few exceptions, they have not figured significantly in the generation of theories about the Roman economy.’ Below I offer a case study of African exports which strikes a balance between minimizing and maximizing points of view and argues that the line drawn between them is far from clear. It also integrates new ideas about the Roman economy. I examine the products of three key ports in the region of Byzacena,!® a large flat plain along the E coast of Tunisia, which has been recognized as one of the major sources of amphoras transported within the empire from c.A.D. 150 to 400. It utilizes the 200-plus stamped amphoras which have been found outside Byzacena but which can be directly associated with the three port towns of Leptiminus, Sullecthum, and Hadrumetum as a basis for asking three questions: What was the distribution of stamped amphoras from each port? Can we detect overall patterns of shipping routes or trade from Byzacena in these finds or does

WS

NH

m

Finley 1985; Whittaker 1989. Finley 1985, 34. Whittaker 1985. E.g., Jongman 1988.

PON

FD

A

Greene 2000. Temin 2001. Hopkins 1980 and 1996. Harris 1993; Papi 2007; Woolf

D =

1992.

Exceptions are Carandini 1970; Hitchner 1993; Mattingly 1988; Mattingly et al. 2000. 0

Cf. Desanges 1963; id. 1992.

128

D. L. Stone

the distribution of amphoras from each port reflect different patterns? and, How distribution patterns support or challenge current models of the Roman economy?

do the

Since the late 1960s, studies of the production and export of foodstuffs from N Africa have

led in many directions.!! They have dramatically increased the knowledge of material from pottery production sites!? and have directed attention to the study of olive oil, wine, and fishsauce production areas.’ Their impact has encouraged archaeologists to conduct field surveys where economic issues have emerged as critical factors in regional development over the long-

term!4 and it has promoted the development of archeometric analyses of ceramic fabrics and vessel contents.!? They have also caused consideration of practical issues of how the export operations worked.!° The most important of these efforts is the synthesis found in M. Bonifay’s

thorough study of N African pottery," which has certainly provided the foundation on which all other studies of N African amphoras will be constructed. Its emphasis lies on the typology, manufacture, and the determination of vessel contents.

One area which, while not neglected, has been relatively under-explored is the study of N African products at the sites to which they were distributed.!® Analysis of distribution was a

feature of some early studies,!? but it has had a reduced róle in recent years in comparison to studies at the production sites themselves. There is now more to be learned from the distribution patterns of N African amphoras and the contexts in which they were found. I will argue that both redistributive and market forces accounted for the movement of stamped amphoras from Byzacena, the former being important

for their distribution to the center of the empire,

the latter being of greater importance in their distribution beyond the capital. I also argue that the evidence for exchange of fish products (garum, liquamen, salted fish) in these amphoras is significant, both at Rome and in the W Mediterranean; this provides further support for the presence of a market economy, since the State did not requisition these fish products as part of the annona. It not only emphasizes the importance of fishing in the economy of Byzacena, but also the development of a market for these products in the wider empire. Firstly, common

I consider

the source of African amphoras,

the stamps

and stamping

practices

in Byzacena, and the contents of these amphoras. Secondly, I discuss the distribution

patterns of amphoras from the three major ports of Byzacena. Thirdly, I consider what factors may have most influenced these distribution patterns. Sourcing N African amphoras The most accurate and methodologically rigorous way to study the production and distri-

bution of amphoras is to combine fabric and typological analysis, since this allows examination of the greatest number of sherds or whole than those which were stamped. Much indicate the distribution of unstamped locations of Punic and Roman amphoras

pots. The number of unstamped vessels is much greater has been done through form and fabric analysis to amphoras. For instance, distribution maps chart the made in Tunisia but found elsewhere in the Mediterra-

nean.^ For Byzacena, however, D. P. S. Peacock and R. Tomber suggested two decades ago that it was not possible to distinguish the place of manufacture easily, due to the relatively homogeneous sedimentary geology of this region;?! even thin-section analysis was incapable, they 11

The most important early studies were Zevi and Tchernia 1969; Carandini 1970; Hayes 1972.

12

Mackensen 1993; Stirling and Ben Lazreg 2001.

13 14

Mattingly 1988; Hitchner 1993; Brun 2003; Slim, Bonifay, and Trousset 1999; Wilson 1999. Hitchner 1988 and 1990; Dietz, Sebai and Ben Hassen 1995; Fentress 2001; Stone 2004; Stone, Matting-

15 16 17

ly, and Ben Lazreg forthcoming; Barker et al. 1996. Garnier 2007; Formenti and Joncheray 1995; Peacock and Tomber 1989. Pena 1998; Godfrey 2004. Bonifay 2004a; see also id. 2007.

18 19 20 21

See Bonifay 2004a, 443-62 for comments on distribution. Zevi and Tchernia 1969; Manacorda 1977; Panella 1993; Keay 1984. E.g., Peacock and Williams 1986, 157; Peacock 1986, 110; Manacorda Peacock and Tomber 1989, 292.

1977, 262; Carignani 1986, 274.

The distribution of stamped amphoras from Byzacena

129

felt, of allowing investigators to draw many distinctions among the fabrics. Potters exploiting the similar clay, limestone, and sand beds near the three ports produced

fabrics that had few

distinctive features, apart from the fine quartz and calcareous inclusions which characterized the pottery of the entire region. In publishing amphoras from Carthage, D. P. S. Peacock referred to products from Byzacena as "Central Tunisian Ware" without further specification, and

many have followed suit when studying amphoras from Byzacena.?? Recent research is changing the picture. In a number of studies, M. Bonifay and C. Capelli, with the assistance of other archaeologists, have argued that it is possible to distinguish

among the fabrics from different production centers in Byzacena. They claim that, while the distinctions are subtle, one can determine the origin of amphoras from Byzacena based on le rapport relatif des minéraux de fer ou de calcaire diffusés dans la matrice (couleur de la páte), dans ce dernier cas, la fréquence du quartz et des microfossiles dans cette fraction, le pourcentage, les dimensions moyennes et maximales, les degrés d’assortiment et de rotondité des composants

dégraissant (essentiellement quartz, roches calcaires, fossiles) et des vacuoles.”

On this basis they have assigned Keay 35B amphoras from the Dramont E shipwreck at SaintRaphaél to the workshop of Sidi Zahruni and Keay 25.1 from the Héliopolis 1 wreck to Sullecthum.^* Bonifay has also tentatively suggested Sullecthum as the origin of several am-

phoras found in a cemetery at Pupput.?? Some studies from other sites to which African amphoras were exported have also distinguished fabrics from different ports in Byzacena.?6 Archaeometric evidence has also been gathered. Instrumental Neutron Activation Analysis (INAA) has been carried out on pottery from both Sullecthum and Leptiminus. It enabled investigators to propose that amphoras from the Plemmirio B wreck near Syracuse originated in Sullecthum. However, INAA has rarely been performed on amphoras exported from Byzacena,

and is thus of limited use here.?? The findings of Bonifay and Capelli and the results of INAA are so recent that they have yet to be widely applied. Much more remains to be done in the analysis and publication of amphoras before the data would allow a different sort of assessment than the one undertaken here. In order to explore the distribution of amphoras from ports in Byzacena with the evidence currently available, at present one has to rely on the analysis of amphora stamps since they can be correlated with specific ports. Stamps and stamping practices on amphoras from Byzacena Stamped amphoras produced at Leptiminus, Hadrumetum, and Sullecthum belong mainly to the well-known ‘African’ amphora series: the 'Africano piccolo’ and 'Africano grande’, or Africana I and IL?5 They date from the middle of the 2nd to the 4th c. The practice of stamping was not regularly employed at Leptiminus, Hadrumetum, and Sullecthum in this period and it appears to have been even less frequently practiced in Byzacena at other periods when amphoras were produced between Punic to Byzantine times. What does distinguish the amphoras from Byzacena from other stamped Roman amphora forms found in the Mediterranean is the inclusion of the name of a port in some of the stamped texts. Some of the stamped 22 23 24 25

Peacock Bonifay Bonifay, Bonifay

1984, 14 and 17-18; id. 1994, 44. 2004a, 26-29. Capelli and Long 2002; Capelli, Ben Lazreg and Bonifay 2006. 2004a, 13.

26 27

E.g., Fabrics 2j (Sullecthum) and 2k (Leptiminus) in Pena 1999, 185. Twostudies have performed INAA on pottery from Byzacena: Taylor, Robinson and Gibbins 1997; and Sherriff et al. 2002. In the case of the amphoras from the wreck Plemmirio B, researchers had access to

pottery from Leptiminus and Sullecthum for INAA, but lacked ceramics from other important kiln sites such as Hadrumetum and Thaenae. Thus they determined the source of the amphoras from the wreck as Sullecthum, but could not "absolutely fix" it (Taylor, Robinson and Gibbins 1997, 20).

28

Inthisand the following two paragraphs, I am indebted to a section of the Leptiminus stamped amphora report initially composed by D. J. Mattingly. I thank him, but note that he should not be held to agree with the present text as I have modified it substantially from its original form.

130

D. L. Stone

amphoras from Leptiminus include LEP to indicate the source of origin, some stamped ‘African’ amphoras from Hadrumetum include HADR, and others from Sullecthum include ASVL. The

inclusion of a port's name on Roman amphora stamps is attested for only a few ports.?? Stamps or graffiti from earlier and later periods have been found on amphoras from Byzacena but they tended not to include the name of the port. In fact, the texts naming ports appear almost

exclusively on Africana I and II amphoras (cf. Tables 1-3 below)?! Additional such as the name

of an individual

(often in the tria nomina

format,

information,

but also abbreviated

in

other ways) appearing on stamps, can sometimes be used to identify the production center.” The practice of stamping African amphoras was arguably undertaken to ensure 'quality control'. That is to say, by affixing a stamp to an amphora at the time of its production, the potter was guaranteeing the solidity and durability of the container, as well perhaps as the

accuracy of the measurement of the contents it was destined to hold.* It is not certain to whom the initials on the amphora stamps refer. One hypothesis holds that the initials on stamps identify the manager (officinator) of the workshop at which the amphoras were produced^^ Another is that they indicate the officinator of the facility in which the amphoras were

filled. Even though no stamps are known to mention the contents of the amphoras, several scholars have posited a connection between the two. D. Manacorda and C. Panella suggested that the stamps included the names of the ports in Byzacena in order that the port towns could exercise control over the olive oil produced in their countrysides and destined for the supply of

Rome.” P. Reynolds thought the purpose of the stamps was “to distinguish exports of amphorae containing goods of private estates from municipal goods collected (e.g. as rents on municipal

land)

for the cities of Hadrumetum,

Sullecthum,

and

Leptis Minor",

in particular

that "they refer to the collection of oil (for the annona) by curatores olearii".?7 In other words, the stamps served as a visual means to mark cargoes belonging to imperial estates, private estates, and the annona, when these were carried in the same ship.*® When potters in Byzacena did stamp amphoras, they tended to place the stamp on the neck, just below the rim. Only a small number of examples have stamps on the handles, toes, or rims. A study of 142 stamped amphoras found by the Leptiminus Archaeological Project? identified stamps in the following positions: neck (117), handle (4), toe (4), rim (7), unidentifiable (10). In

general, most amphoras appear to have received a single stamp, although the fragmentary nature of many vessels makes this difficult to confirm. The percentage of amphoras stamped during production at these port towns is unknown but it was much lower than the percentage 29 30

31

Variants of LEP, HADR, and ASVL were also employed. Other N African ports are also mentioned on stamps. The most common are Tubusuctu and Neapolis. A stamp reading TOP has been suggested to have originated in Byzacena at Taparura; likewise, a stamp reading TH has been suggested to have originated at either Thapsus or Thaenae (Bonifay 2004a, 15). Amphoras produced in Beirut contained the stamp BER (Berytus): Reynolds 1997-98, 49 and 59-63. On the use of toponyms on Roman amphoras, see Manacorda and Panella 1993, 59-60. ‘It is rare to find the name of the port stamped on an amphora that is not an Africana I or II. Of the 142 stamped amphoras found during the recent survey and excavations at Leptiminus, 32 contained the letters LEP or a variant; many were fragmentary, but it was possible to identify 14 of the 32 by type: 1 Africana I, 2 Africana IIC, and 11 Africana IID (Stone, Mattingly and Opait, forthcoming).

32 33

Examples include LEPMI/BSCD, Cf. Manacorda 1993, 38-39.

FANFORT/COLHADR,

and DON/ASVL.

34

Manacorda 1977, 203. Regarding stamps on Dressel 20 amphoras from S Spain, the more common view is that the initials indicate the officinator of the workshop at which the amphoras were produced: Liou

and Tchernia 1994, 137-45. Another suggestion is that these initials refer to the owner of the oil at the time the amphora was filled: Remesal Rodríguez 1998, 190.

35

Gibbins 2001, 322.

36 37

Manacorda and Panella 1993, 60. Reynolds 1995, 47.

38 39

lbid.85. Stone, Mattingly and Opait, forthcoming.

The distribution of stamped amphoras from Byzacena

131

employed at some production sites elsewhere (e.g., those in the Guadalquivir valley}. Further, since the percentage of amphoras stamped at each of our three production sites will have varied, one cannot use the number of stamps recovered to estimate which town was the most important producer and exporter. The practice of stamping in the Roman world frequently occurred at times when investment in agricultural production, pottery manufacture, and shipping was significant; thus the practice was prominent in Republican Italy, as well as Early Imperial Gaul and Spain, in addition to Byzacena in the 2nd and 3rd c.*! One should not expect the practice to relate directly to the extent of investment in agricultural production, pottery manufacture, and shipping, as the desire of the State to organize trade and control taxation may also have stimulated it. The contents of amphoras from Byzacena What did the stamped amphoras which can be associated with the ports of Byzacena contain? Jt seems clear is that the stamps themselves do not inform us about the contents. Scholars have explored the issue of what Africana amphoras transported through contextual studies of findspots, textual analysis of dipinti, and gas chromatographic analysis of residues left in container walls. These studies have caused several revisions of the communis opinio. Forty years ago scholars emphasized the growth of the African economy in connection with the export of

olive oil in amphoras.?? Later, the detection of an impermeable pitch lining the interior of many amphoras in the Africana II series led to residue analyses which suggested that they carried fish products and wine more commonly than initially thought, and olive oil less fre-

quently.43 This makes sense, as pitch dissolves in oil and alters its taste. Some scholars stated that any amphora lined with pitch did not contain olive oil, yet subsequent analyses have shown that some pitch-lined amphoras did contain olive oil. It now appears that, while pitched amphoras would probably not have contained olive oil intended for cooking and consumption, they may have carried oil destined for lighting, perfumes, or hygenic uses, in which

cases the pitch would not have negatively affected the product.** The most recent and detailed study shows that Africana IA, Africana IB, and pseudo-Tripolitanian Africana IIB amphoras contained

olive

oil, while

Africana

IIA,

IIC, and

IID were

for salted

fish

products?

and

possibly also for wine,* but the last word has probably not been said. Methodology For this study, I investigated several sources and publications of amphora stamps. The main reference collection is the CEIPAC (Centro para el Estudio de la Interdependencia Provincial en

la Antigüedad Clasica) Corpus. A number of early studies on the distribution of N African amphoras also contained valuable information.*® Several other publications list stamped amphoras from Byzacena which had been found in various western regions and provinces.# The 40

Estimates of the frequency of stamping N African amphoras have ranged as high as 12%, but most investigators regard this number as too high: cf. Gibbins 2001, 322; Mattingly 1988, 55; Peacock, Bejaoui and Ben Lazreg 1989, 200-1; Rodriguez-Almeida 1977, 218-19. It appears that stamping was

more common in the 3rd than the 2nd c. 41 42

43 44

45

Manacorda 1993, 51-52. Zevi and Tchernia 1969; Carandini 1970; Panella 1993.

Ben Lazreg et al. 1995. Bonifay 2007.

Iusetheterm “fish products” to indicate the range of conserved fish goods that were shipped, including garunt, liquamen, and whole fish.

46

On contents, see Bonifay 2007. Wine has received less attention as a product exported from Byzacena, although it may have been important (Brun 2004). A connection between fish-salting tanks and amphora

47 48

49

production has been established at both Leptiminus and Sullecthum. Fish-products appear to have been the dominant product exported in these amphoras: Bonifay 2004a, 472-73 and 478. CEIPAC (http:/ /ceipac.gh.ub.es). Zevi and Tchernia 1969; Manacorda

1977.

Keay 1984; Mayet 1978; Ribas 2004; Rodriguez-Almeida 1984; Remesal Rodríguez 2007.

132

D. L. Stone

RTAR (Recueil de timbres sur amphores romaines) database is an additional resource.?? [ also benefited from helping to publish amphora stamps from the survey and excavations conducted by the Leptiminus Archaeological Project.?! The information gathered from the various publications was collected in Tables 1-3 and represented on maps (figs. 1-3). Leptiminus More amphora stamps are known from Leptiminus than any other African production site. This is largely a result of the field survey and excavations of the Leptiminus Archaeological Project. As of this writing, about 150 amphora stamps have been found in the port town itself, while 79 from Leptiminus have been identified outside Byzacena. Many of the stamps were found during excavations of a kiln site ($290) in the E suburbs of Leptiminus, while others were

located during a large-scale survey of the town.?? A smaller number of stamps (7) was collected during a survey of pottery production sites carried out in the done at Leptiminus, it has proven possible to associate with bear an abbreviation of the town's name (generally LEP), but (HONOR, MARI, SECUND, and {swastika}ZY).?* Since these production areas at Leptiminus, the present analysis includes

Montmaurini

Vienne Fos (2) Nice

Matarol

Istres

aMárseille

\Giglio

Turris Rome | Libisonis76) Minorca (2) Ostia (19)

eTróia

Valenciae

Elche

e Barbate

(X9 E^

.%

N

(Mallorca (7) Hippo Regius ,

e

,

4 ziiae

.

1980s.% Due to the detailed work the town not only stamps which also several stamps which do not stamps can be securely linked to examples of these stamps when

Leptiminus (150%

Pupput N

9

»" > 2

.

Mediter Cea.

C

Q o

«i

Volubilis (3) Gaza (4)

mo

N 900

1000 km

Fig. 1. Distribution of stamped amphoras from Leptiminus.

50 51 52 53 54

RTAR (http://publications.univ-provence.fr/rtar/). The author wishes to thank D. J. Mattingly, A. Opait and J. Moore for drawing his attention to several bibliographic references consulted here. This material will appear in Leptiminus report no. 3 (Stone, Mattingly and Ben Lazreg forthcoming). Peacock, Bejaoui and Ben Lazreg 1989. Evidence for these will be thoroughly discussed in Stone, Mattingly and Opait, forthcoming.

The distribution of stamped amphoras from Byzacena TABLE

133

1

STAMPED AMPHORAS FROM LEPTIMINUS FOUND ELSEWHERE Stamp

Type

Findspot

Reference

AN.LEP

Af IID

Montmaurin

Fouet 66a)

AN.LEP

Af I/II

Ostia

Rodriguez Manacorda

Almeida 1984, 270; CIL 1977, 193, 200 (no. 66c)

AN.LEP

Af IID

Gaza

Manacorda

1977,

1969, 235; Manacorda

1977, 193, 200 (no. XV

3387;

193, 200 (no. 66b); Ben

Arieh

1974, 94, pl. XXIX.3. AN.LEP

Turris Libisonis

Villedieu 1984, 187

AVSPI/LEPTI

Af IID

Marseille

Manacorda 1977, 193, Tchernia 1969, 199-200

AVSPI/LEPTI BSCD/LEPMI

Af Af IIB

Volubilis Ostia

Mayet 1978, 384 Manacorda 1977, 193, 199-200 (no. 63b); Zevi & Tchernia

BSCD/LEPMI BSCD/LEPMI

Af IID Af IID

Rome Ostia

BSCD/LEPMI C.LEPT/SCD

Af I/II Af IIB

Ostia Fos

C.LEPT/SCD

Af IID

Ostia

199

(no.

60);

Zevi

&

1969, 190

Manacorda 1977, 193, 199-200 (no. 63b) Manacorda 1977, 193, 199-200 (no. 63b); Calza 1920 n.8 Manacorda 1977, 193, 199-200 (no. 63b) Manacorda 1977, 193, 199 (no. 56); Zevi & Tchernia

1969, 204

Manacorda

1977, 193, 199 (no. 56); Calza

1920

n.7

C.LEPT/SCD CLE/ARA/P

Af II Af HD

Rome Ostia

CLE/MAX

Af IID

Fos

CLE/MAX

Af IID

Ostia

Remesal Rodriguez 2007, 222 n. 641a Manacorda 1977, 193, 199 (no. 57); Calza 1920 n.28 Manacorda 1977, Tchernia 1969, 204 Manacorda

1977,

193, 193,

199

(no.

58);

Zevi

199 (no. 58); Calza

&

1920

n.30 CLE/MAX

Af I/II

Istres

Manacorda

1977, 193, 199 (no. 58); Benoit 1953,

CLE/MAX

Af I

Majorca (Cabrera 1)|

288 Cerda 1994, 299; Ribas 2004, 1074

CLST

Af II

Volubulis

Manacorda

384 Manacorda 3432.

1977,

193,

199 (no. 59b); CIL

XV

193,

199

XV

1977, 193, 199 (no. 59a); Mayet

1978,

CLST

Af I/II

Rome

CLST

Af I/II

Ostia

Manacorda

1977,

Ostia

3432. Manacorda

1977, 193, 199 (no. 59d); Calza 1920

CLST

Af ID

(no.

59c);

CIL

n.27 CLST

Af HD

Vienne

Godard

CLST C.M.C LEP COLE/GETV

Af II

Barbate Rome Ostia

Bueno Serrano 2003, 21-22 CIL XV 2633 Manacorda 1977, 192, 199 (no. 55); Calza 1920 n. 1

Rome |

Manacorda 1977, 192, 198-99 (no. 53); Rodriguez Almeida 1984, 270; CIL XV 3385

De Nicolas 1979 n. 68; Ribas 2004, 1073 Calza 1920 n. 3

COL.LE/CATVLIO

Af IID | Af I/II

1995, 315

COL.LEP COL.LEP

AfII Af IID

Cala N’Forcat Ostia

COL.LEP/CAS

Af IID

Ostia

Manacorda

Rome

4 Manacorda 1977, 192, 198 (no. 49); Rodriguez Almeida 1984, 270; CIL XV 3382.

COL.LEP/CAS

Af 1/0

1977, 192, 198 (no. 49); Calza 1920 n.

134

D. L. Stone

Stamp

Type

Findspot

COLLEP/C.I.S

Af I/II

Rome

Reference

Manacorda

1977, 192, 198 (no. 50); Rodriguez

Almeida 1984, 270, CIL XV 3383 CJOLLEF?/C]LAUDI

Af IID

Turris Libisonis

Villedieu 1984, 187

US/OPTATUS COL.LEP/EHD

Af II

Tróia

Maia

COL.L.EP/FELIC

AfIID

Ostia

COL.L.EP/FELIC

AfIID

Rome

Manacorda 1977, 192, 198-99 (no. 51); Calza 1920 n. 2 Manacorda 1977, 192, 198-99 (no. 51); Rodriguez Almeida 1984, 270, CIL XV 3384b. Marec 1961, 191

COLLEP/[F]ELIC

Hippo Regius

COL.L.EP/[FEL |IC

Af II

COLEP/FELIX

Af I/II

Mallorca

(Cabrera

1)|

Rome

1975; Keay

1984, 123

Cerdà 1994, 299; Ribas 2004, 1073 Rodriguez

Almeida

1984,

270,

CIL

XV

3384a;

54);

Rodriguez

Manacorda 1977, 192, 199 (no. 52) COLL/FRTS

Af I/II

Rome

Manacorda

1977,

192,

199

(no.

CJOLL/FIRTS

Af IID

Turris Libisonis

Almeida 1984, 270, CIL XV 3386 Villedieu 1984, 187

COL.L.EP/M.A.I

Af IID

Turris Libisonis

Villedieu 1984, 187

COL.L.EP/M.A.I COL.L[/M.A[ HONOR HONOR

Af Af IID Af II

HONOR

Rome Valencia Pupput Minorca (Cavalleria)| Rome

HONO/RATI LE[PMI

Af IIB AfIID

LEP

Af IIB

LEP

Giglio [Mallorca (Cabrera 3)|

Remesal Rodriguez 2007, 221 n. 640a Marquez & Molina 2005, 308 Bonifay 2004a, 12-13 De Nicolas 1979, 70; Ribas 2004, 1075 CIL XV 3469; Callender 1965 n. 716

Celuzza & Rendini 1992, 127 Bost et al. 1992, 140

Rome

Manacorda

Turris Libisonis

Villedieu

[Mallorca (Sant Jordi}! Ostia

1977, 193, 200 (no. 64) 1984, 187

LEP/IUL/TIM LEP/ROSAR

AfIID Af IID

LEP /ROSAR LEP /ROSAR

Af IID Af

Pupput Rome

Cerda 1994, 299; Ribas 2004, 1074 Manacorda 1977, 193, 200 (no. 65); Calza 1920 n. 6 Bonifay 2004a, 14-15; Bonifay 2004b, 28-29 Remesal Rodriguez 2007, 222 n. 642a

LEPMI

Af IA

Ostia

Manacorda

LEPMI/BSCD

Af 11C?

Gaza

Manacorda

1977, 193, 199 (no. 63a) 1977,

193,

199-200

(no.

63b);

Ben

Arieh 1974, 94, pl. XXVIII.4 LEPMI/BSCD

Af II

Gaza

LEPMI/BSCD

Af IID

Turris Libisonis

LEPMI/BSCD LEPMI/BSCD LEPMI/BSCD LEPMI/BSCD

AfIID Af I/II Af I Af II

LEPMI/BSCD

Af II

Torremolinos (Malaga)

Serrano Ramos et al. 1993, 210

LIEPM/DOM[ LEPMI/DOMFE

Af ITD Af IID

Mataró Ostia

Keay 1984, 122; Manacorda 1977, 193 (no. 62) Manacorda 1977, 193, 199 (no. 62); CIL XV 3449; Rodriguez Almeida 1984, 271; Zevi & Tchernia 1969, 181-82

LEPMI/DOMF

Af I/II

Rome

LIEPT LEPT

|Mallorca (Cabrera 3)| Rome Mallorca (Cabrera 1)} Minorca (Cavalleria);

Ostia Ostia

Arthur & Oren 1998, 199 Villedieu

1984, 187

Bost et al. 1992, 140 CIL XV 3591 Cerda 1994, 299 De Nicolas 1979 n.71

Manacorda

1977, 193, 199 (no. 62); CIL XV 3449;

Rodriguez Almeida 1984, 271; Zevi & Tchernia 1969, 181-82 Calza 1920 n. 64. Manacorda 1977, 193, 199 (no. 61); Calza 1920 n. 5.

The distribution of stamped amphoras from Byzacena Stamp

Type

Findspot

Reference

LEPT

AFT/TI

Elche

Marquez

Villora

Reynolds

1993, 252

Vidal

AfII

MARI

Af IIA

Ostia

RTAR 2 (1988) 267; Manacorda (no. 24)

MARI

AF IID

Volubilis

RTAR 2 (1988) 267; Mayet corda 1977, 191, 197 (no. 24)

Af IID

MARI/LEPTI

{palm} | Af IID

SECVND TITTOR/ISLEP

| Af IID

{swastika} ZY, YZ | Af IB {swastika} (retro) or

Mallorca

(Cabrera 3)|

Molina

MAR

MARI

Mallorca

&

(Cabrera 3);

135

2005,

311;

Bost et al, 1992, 140; Ribas 2004, 1075 1977,

191,

197

1978, 385; Mana-

Bost et al. 1992, 141; Ribas 2004, 1075

Nice

Bonifay

Rome

CIL XV 3171

2004a,

12-13

Uzita

Van Der Werff 1982, 184

Rome (Quirinal)

Bird et al. 1993, 91, fig. 20.251, 97 fig. 21.5, 98 n.5.

YS {swastika} (retro)

they are found outside Byzacena. Comparable fieldwork at Sullecthum allows the inclusion of similar material, but this is not the case at Hadrumetum, where kiln sites have not yet been

located. A number of other stamps discovered at the Leptiminus kiln sites contain anepigraphic symbols such as concentric circles. Such circles have been found on vessels from Byzacena at a number of Mediterranean sites, but in the absence of detailed fabric analysis it is impossible to determine whether they originated at Leptiminus or another site. In particular, the presence of anepigraphic circles in stamps found at kiln sites at Sullecthum makes it too problematic at present to identify the origin of particular anepigraphic stamps, although further work could

clarify the situation.” Rome and Ostia have the highest numbers of stamped amphoras from Leptiminus. The total number of Leptiminus stamps identified at Ostia is 19, and the same number applies to Rome.

Since the total number of all Leptiminus stamps found other than at Rome, Ostia, and Leptiminus is 41, just under half of the stamps (48%) found outside Leptiminus come from the capital and its chief port. The largest concentration of Leptiminus stamps elsewhere (11) is in the Balearic Isles. Six stamps were excavated at Turris Libisonis on Sardinia. Several stamps have been found in southern France (6), along the coast of Spain (4), at Volubilis (3), and at Pupput

(2). Distribution of Leptiminus products in the E Mediterranean was limited, to judge from the

paucity of stamps recorded there (only in the Gaza Strip”). It was possible to identify the type of 40 stamped amphoras originating at Leptiminus.?? We may divide them into: oil amphoras fish-product or wine amphoras

1 Africana IA 1 Africana IIA

Of these, 1 Africana IA, 1 Africana HA

4 Africana IIB; 1 Africana IIC

and 2 Africana IIB amphoras

33 Africana IID. were

found

at Rome

or

Ostia, while 2 Africana IIB amphoras and 1 Africana IIC amphora were found elsewhere. The Africana IID amphoras were divided between Rome and Ostia (14) and elsewhere (19). Sullecthum

Archaeological research at Sullecthum has included extensive surveys, which have discovered and recorded several kiln sites and amphora stamps.”® Fish-salting tanks, where garum or other fish products may have been bottled, have yielded other stamps.?? A total of 25 stamped 55

See ibid. for a preliminary discussion of the differences in anepigraphic stamps.

56

Ben Arieh 1974, 94; Arthur and Oren 1998, 199.

57

Theauthor relied primarily on identifications made by others in publications of these amphoras. In the case of amphoras found at Turris Libisonis and on the Cabrera 3 wreck, he made the identifications based upon published drawings. Peacock, Bejaoui and Ben Lazreg 1989; Lavoie 1989. Foucher 1970.

58 59

136

D. L. Stone

.

Marseille

Empüries e

Nice €

Monaco

NE .

.

.

eCosa

"

|j

Turris Libisonis (2

Santa Pola

J

stia (16)

Q Elche

ome

>

q?

e

ŒMéllorca (5) Pupput (2)

Sultecthum (259)

\

Malt“28°

AR

»

Athens

À Plemmirio

7° 20 CX

Mediterranean

er

Sea

eBeirut Gaza

* Shipwreck identified as carrying amphoras from Sullecthum

0 L-

500

1000 km

—MEN

Fig. 2. Distribution of stamped amphoras from Sullecthum.

amphoras have been identified at Sullecthum. These discoveries have allowed Peacock and others to paint a picture of a number of kiln sites ringing the town, as well as others located in

the near hinterland, with a shift toward rural production over time.9? Due to the detailed examination of Sullecthum, it has proven possible to associate with the town not only stamps which bear an abbreviation

of the town's name

(generally ASVL or ASYLL), but also several

stamps which do not (MVI, NAVIG, PAS, and TERTI). Since these stamps can be securely liked to production areas at Sullecthum, the present analysis includes examples of these stamps when

they are found outside Byzacena.®! Outside Byzacena, the highest number of amphora stamps has been found at Rome (62) and Ostia (16). At Monte Testaccio alone 50 of these stamps have been found, including 46 in the

recent Spanish excavations.°? The Rome-Ostia finds represent 80% of the total of 98 stamps. Of the remainder, 5 were found in the Balearic Isles, 2 at Pupput, and 2 at Turris Libisonis. Along the coast of Spain, a further 4 Sullecthum stamps have been identified (including one from an undocumented underwater provenience). One each derived from Marseille and Malta. Sullecthum stamps have also been located at three sites in the E Mediterranean: the Athenian agora (1), Beirut

(1), and

Gaza

(1). In addition,

investigators

have

plausibly

INAA and thin-section analysis three shipwrecks with amphora Héliopolis 1; Plemmirio B; and Monaco A.

identified

through

cargoes from Sullecthum:

60

Peacock, Bejaoui and Ben Lazreg 1989, 199-201.

61 62 63

Comparable fieldwork exists at Leptiminus but not Hadrumetum. Anepigraphic stamps from Sullecthum have not been considered. Cf. Remesal Rodriguez 2007. Joncheray 1997, 149; Bonifay 2004a, 453.

64

Taylor, Robinson and Gibbins 1997, 14-18; Gibbins 2001, 326.

65

Bonifay 2004a, 454.

The distribution of stamped amphoras from Byzacena TABLE STAMPED AMPHORAS Stamp

Type

AELEOR/ASYL

Af II

ASVL [AJSVL/...

2

FROM SULLECTHUM FOUND ELSEWHERE

Findspot Mallorca

Reference (Cabrera

1)

Cerda 1994, 303 f9a

Rome (Testaccio) Pola

(Villa

Remesal Rodriguez 2007, 217 n.616

Af IIB

Santa

Palmeral) Mallorca (Cabrera 1)

de

El | Marquez

Villora & Molina

Vidal 2005, 303

../A...(ASVL?)

Af II

ASYL...

AfIIB

| Rome (Testaccio)

Rodriguez Almeida 1981, 21

[AJSVL/ATA ASYL/CAEL

Af AfI/D

Rome (Testaccio) | Marseille (Punta Debie)

Remesal Rodriguez 2007, 216 n.613 Liou 1975, f12

| Turris

137

Libisonis

Cerdà 1994, 303 f9c

ASVL/CATI

AfIID

ASVL/CAT

Af

ASVL/CA[T ASVL/CAT[ AS[/CA[ ASVL/]JEN ASYLL/FELIC ASVLL/FELIC ASVLL/FELIC ASYLL/FELIC ASYLL/FELIC ASYILL]/FEL[IC] A[SYLL ]/ F[ELIC] ASVLL/FELcIC

Af Rome (Testaccio) Af Rome (Testaccio) Af Rome (Testaccio) Af Rome (Testaccio) AfI/II | Rome (Esauiline) AfI/II | Rome (Esquiline) Af I/II | Rome (Testaccio) Af Rome (Testaccio) Af Rome (Testaccio) Af Rome (Testaccio) Af Rome (Testaccio) Af IB Ostia

ASVL/FVSCI

Af

Rome (Testaccio)

Remesal Rodriguez 2007, 215 n.607b

ASYL/]VSCI

Af

Rome (Testaccio)

Remesal Rodriguez 2007, 215 n.607a

ASVL/L.A.D [AS]VLLE/[VS ASVLL/.MAR ASYL/NAVI

Af Af AfIID Af IB

ASYL/NAVI

Af I/II | Rome (Monte della Gius- | CIL XV 3390d,5

ASYL/NAVI

Af1/II

| Rome (Esquiline)

CIL XV 3390d,6

ASVLL/NREP

AfI/IL | Rome (Esquiline)

CIL XV 3390e,7

ASYL/NREP ASYL/NREP ASY[L]/NRFV ASYL/NRFVS ASVL/OT.FEL ASVL/SATVR ASYL/SATVR [A]SYL/SILVA ASYL/SILVA ASVL/SILVA]

Af Rome (Testaccio) Af Rome (Testaccio) Af Rome (Testaccio) Af Rome (Testaccio) Af I/II | Ostia Af I/II | Ostia Af Rome (Testaccio) Af Rome (Testaccio) Af Rome (Testaccio) AfIIA | Spain*

Remesal Rodríguez 2007, 212 n.602a Remesal Rodríguez 2007, 212 n.602b Remesal Rodriguez 2007, 212 n.603a Remesal Rodriguez 2007, 212 n.603b Manacorda 1977, 200 n.84 Manacorda 1977, 200 n.96 Remesal Rodríguez 2007, 216 n.611a Remesal Rodríguez 2007, 216 n.612a Remesal Rodríguez 2007, 216 n.612b Márquez Villora & Molina Vidal 2005, 297

ASYIL/TEREN

Af IB

Ostia

Manacorda

ASYIL/TEREN

Af T

Ostia

Manacorda

1977, 200 n.97b

ASYL/.T.F

AfI/I

| Ostia

Manacorda

1977, 200 n.86

ASYL/.T.F. ASYL/.T.F.

Af Af

Rome (Testaccio) Rome (Testaccio)

Remesal Rodríguez 2007, 213 n.604a Remesal Rodriguez 2007, 213 n.604b

ASYL/.T.F.

Af

Rome (Testaccio)

Remesal Rodriguez 2007, 213 n.604c

Rome (Testaccio)

Rome (Testaccio) Rome (Testaccio) | Ostia Rome (Museo Nazionale

Villedieu 1984, 189 Remesal Rodríguez 2007, 211 n.600a

Remesal Rodríguez 2007, 211 Remesal Rodriguez 2007, 211 Remesal Rodríguez 2007, 211 Remesal Rodriguez 2007, 216 CIL XV 3390a,1 CIL XV 3390b,2 CIL XV 3390b,3 Remesal Rodríguez 2007, 214 Remesal Rodriguez 2007, 214 Remesal Rodríguez 2007, 214 Remesal Rodríguez 2007, 214 Manacorda 1977, 200 n.82a

n.600b n.600c n.600d n.614

n.606a n.606b n.606c n.606d

Remesal Rodriguez 2007, 211 n.599 Remesal Rodriguez 2007, 217 n.615a Manacorda 1977, 200 n.89 | Manacorda 1977, 200 n.90a

Romano) tizia)

1977, 200 n.97a

138

Stamp ASYL/.T.F. ASYL/.T.F. [ASY]L /[.T.]F. ASYL/T.F. AS[YL]/.T[.F]

D. L. Stone

Type Af Af Af Af Af

Findspot Rome (Testaccio) Rome (Testaccio) Rome (Testaccio) Rome (Testaccio) Rome (Testaccio)

Reference Remesal Rodriguez Remesal Rodríguez Remesal Rodríguez Remesal Rodríguez Remesal Rodriguez

[AS]YL / ]F.[

Af

Rome (Testaccio)

Remesal Rodríguez 2007, 213 n.604h

CLARI/ASVLL

Af

Rome (Testaccio)

Remesal Rodríguez 2007, 214 n.605a

2007, 2007, 2007, 2007, 2007,

213 213 213 213 213

n.604d n.604e n.604et n.604f n.604g

CLARI/ASVLL

Af

Rome (Testaccio)

Remesal Rodríguez 2007, 214 n.605b

C[LARI]/ASVL[L} [CLAR]I/[ASVL]L

Af Af

Rome (Testaccio) Rome (Testaccio)

Remesal Rodriguez 2007, 214 n.605c Remesal Rodriguez 2007, 214 n.605d

DON /ASVL

AfIID

DON/ASVL

AF II

DON/ASVL DON/ASVL ...JDON/...]ASVL

? Turris Libisonis Af I/II | Athens (Agora) Af I/II | Ostia

| Mallorca

(Cabrera

3)

Rome (via Gesü e Maria)

CIL XV 3390¢,10

Villedieu 1984, 189 Manacorda 1977, 200 n.78a Manacorda 1977, 200 n.79

FELIC/ASYLL

AfIID

FELIC/ASYLL

AfI/II

IVLIAE/ASYL

Af Il

Ostia

Manacorda

L.E.D/ASVL L.E.D/A[SJVL MAVI/ASVL

Af Af Af II

Rome (Testaccio) Rome (Testaccio) Mallorca (Cabrera 1)

Remesal Rodriguez 2007, 212 n.601a Remesal Rodríguez 2007, 212 n.601b Cerda 1994, 303 f9b

MVI

Af I

Ostia

Manacorda

MVI

AfI/II

NAVIC PARIAT/ASVL [PIARIAT/ASVL PARIA[T]/ASVL

| Ostia

Bost et al. 1992, 139

Manacorda

| Rome (Testaccio)

| Emptiries

Af IB Af Af Af

Ostia Rome (Testaccio) Rome (Testaccio) Rome (Testaccio)

1977, 200 n.83a

CIL XV 3390c,4

Keay

1977, 200 n.99

1977, 196 n.1; Panella 1983 f.39

1984, 102 f. 38.1; Almagro

204 CIL XV Remesal Remesal Remesal

3493 Rodríguez 2007, 215 n.608a Rodríguez 2007, 215 n.608b Rodriguez 2007, 215 n.608c

PAS

Af IIA

| Beirut

Reynolds

PAS

AfIIA

| Nice

Bonifay 2004a, 13, f3.6

|

Cosa

1997-98, 82, £203

PAS

AfIIA

PET.IAL/ASVL

Af I/II

Bruno & Scott 1993, 186, f46.5

P.FELI/ASVL QVIN/ASYL OVIN/ASYL

Af IIB | Gaza AfI/U | Malta AFI/II | Rome (Esquiline)

Ben Arieh 1974, tXXIX,3 Manacorda 1977, 200 n.93b CIL XV 3390f,8

OVIN/ASYL

AfI/II

CIL XV 3390£,9

QVIN/ASYL QVIN/ASYL QVIN/ASYL ROG/ASYL

Af Af Af Af

| Ostia

Manacorda

| Rome (via Gest e Maria)

Rome Rome Rome Rome

Basch

(Testaccio) (Testaccio) (Testaccio) (Testaccio)

Remesal Remesal Remesal Remesal

1977, 200 n.92

Rodriguez Rodriguez Rodríguez Rodríguez

ROG/ASYL

AfI/II

| Rome (Testaccio)

RVFVS/ASYL

AfI/II

| Rome (in hortis Torlonia)

RVFVS/ASVL

AfI/I

| Ostia

SVL SVL

AfIIA | Pupput AfIIA | Pupput

Bonifay 2004b, 28 Bonifay 2004b, 28

SVL/.T.F

Af IB

Manacorda

TERTI

AfI/II

| Rome

CIL XV 3536

TERTI/ASYLL

AfIID

| Ostia

Manacorda

TERTIA/ASYLL

AfIID

| Mallorca (Cabrera 3)

Bost et al. 1992, 141

‚T.F/ASYL

AFI/II

| Ostia

Manacorda

Sabratha

2007, 2007, 2007, 2007,

215 215 215 216

CIL XV 3293 | CIL XV 3134 Manacorda

1977, 200 n.95a

1977, 200 n100 1977, 200 n.98 1977, 200 n.87

n.609a n.609al n.609b n.610a

1952,

The distribution of stamped amphoras from Byzacena

139

Stamp

Type

Findspot

Reference

VIAT/ASVLL

Af

Rome (Testaccio)

Remesal Rodriguez 2007, 226 n.661a

VIAT/[ASVLL] VICTO/RINI/ASVLL,

Af AfIIB

Rome (Testaccio) | Elche (La Alcudia)

Remesal Rodriguez 2007, 226 n.661b Márquez Villora & Molina Vidal 2005, 305

*Unknown underwater findspot

It was possible to identify the type of 21 stamped amphoras originating at Sullecthum. We may divide these into: oil amphoras 2 Africana I (not sub-typed) 5 Africana IB 4 Africana IIB; fish-product or wine amphoras 4 Africana IA 6 Africana IID. Seven of the Africana I and IIB amphoras were found at Rome and Ostia; four were located elsewhere. Of the Africana IIÀ and IID, 2 were found at Rome and Ostia, 6 elsewhere. Hadrumetum

Since mediaeval and modern Sousse overlies Hadrumetum, archaeological investigations have been limited and have not considered pottery production areas. All identifications of

stamps from Hadrumetum have been made on the basis of stamps which include the lettering CHADR, COLHADR,

CH, HDMT, or a variant. Stamps which do not include the town's name but

could be associated with local kiln sites have not been reported at Hadrumetum.96 The distribution of the 42 stamps from Hadrumetum is strongly weighted toward Rome and Ostia. The number of Hadrumetum stamps identified at Ostia is 13, at Rome 12. Stamps at the

e Lectoure

| Minorca

Airola

Turris Libisonis

lannutri Rome (6)

à

stia (13)

>

@ * Tharros G mallorca (2)

o? eo KT

Reggio

SS

Pupput

mie

i Thamusida

Hadrumetum (4)C

© Ÿ > 2

Mediterranean Sea

a

Q 2:

vc

500 1000 km _EEE Fig. 3. Distribution of stamped amphoras from Hadrumetum.

66

A small number of stamps have been found in excavations at Hadrumetum; they do not offer any new stamps which can be directly correlated with production in the town, and thus were excluded from consideration here. See CIL VIII 62 and 10477; Foucher 1964, 108.

140

D. L. Stone TABLE 3 STAMPED AMPHORAS FROM HADRUMETUM

FOUND ELSEWHERE

Stamp

Type

Findspot

Reference

CHD/AE

Af

Rome (Testaccio)

Remesal Rodríguez 2007, 221 n. 638a

CHDR/AED

Af II

Airola

Manacorda

CHDR/AED CHAD/C.ARFR CHDRAE-HDR]/AED] CHDR/CFVC CHADR/MFFOR CHIVL

Af Af ? Af Af Af

Ostia Rome (Esquiliine) Tunisia* Ostia Ostia Minorca (Port de Sanitja)

Manacorda 1977, 198 n.43b CIL XV 3379 Gauckler et al. 1902, 76 n. 12 CIL XV 3380 Manacorda 1977, 198 n.36 Ribas 2004, 1075

CLAVSI/FHADR

Af

Rome (Testaccio)

Remesal Rodríguez 2007, 221 n. 639a

CLHDR/FLC

Af IID

Ostia

Manacorda 1977, 198 n.44a; Tchernia 1969, 192, f16b

COLHADR/HERCR[

Af IID

Turris Libisonis

[COL]/[(H?)AD] COLLVMI/COLHADR FAFO FANFO.../COLHAD FANFORT/COLHADR

Af II Af I/II

Mallorca (Cabrera 1) Rome Pupput Rome (Testaccio) Rome (Monte della Giusti-

FANFORT/COLHADR FANFORT/COLHADR

Af I/II AfI/II

FANFORT/COLHADR FANFORT/COLHADR FANFORT/COLHADR FANFORT/COLHADR FANFORT/COLHADR

Af I/II Af T/II Af I/II AfIA Af IID

Ostia Ostia Rome (Testaccio) Ostia Ostia

CIL CIL CIL Zevi Zevi

FANFORT

Af I/II

Lectoure

Zevi

I/II I/II I/II IB I

Af I/II Af I/II

1977, 198 n.43a

Zevi

&

Villedieu 1984, 187

Cerdà 1992, 06c CIL XV 3377 Bonifay 2004b, 28-29 CIL XV 3375c,8 | CIL XV 3375a,1

zia)

Rome (villa Martinori) CIL XV 3375a,2 Rome (cemeterio quodam | CIL XV 3375a,3 Romano)

/COLHADR?

XV XV XV & &

3375a,4-1 3375a,4-2 3375a,5 Tchernia 1969, 185, f9 Tchernia 1969, 190, f14a

&

Tchernia

1969,

183,

n.10;

Manacorda 1977, 198 n.34g FANFORT/COLHADR

Af IID

Turris Libisonis

Villedieu 1984, 187

FANFORT/CO]LHADR

Af IID

Turris Libisonis

Villedieu 1984, 187

FANFORT/COLHADR [FANF(O)./COLH[...

Af IID Af Il

Turris Libisonis Thamusida

Villedieu 1984, 187 Callu et al. 1965, 544; 1977, 198 n. 34i RTAR 2 n.1320 Cerdà 1992, 10c CIL XV 3375b,6 CIL XV 3375b,7

Manacorda

FANFOR[T]/COLH[ADR] | Af I/II FANFORT/COLHADR Af II FANFOR/COLHADRV Af I/II FANFOR/COLHADRV Af T/II

Tharros Mallorca (Cabrera 1) Rome (Baths of Diocletian) Rome (Esauiliine)

HDMT/SDL

Ostia

Manacorda 1977, 198 Tchernia 1969, 181

(H)MCAS(S)/FHAD(R) Af IID IVNIOVINTI/COLHADR | Af I/II

Giannutri Rome

Celuzza & Rendini 1991, 105 CIL XV 3376

PAVL/F.H

?

Turris Libisonis

Villedieu

PAVL/FH

Af TA

Ostia

Zevi & Tchernia 1969, 192, f16e

PAVLF/HADR

Af TI

Ostia

CIL XV 3378

QSM/CH

Af IID

Ostia

Manacorda 1977, 198 n.46; Tchernia 1969, 192, f16a

SEC.D/F. HD

?

Turris Libisonis

Villedieu 1984, 187

SEC.DI/F.HD

Af IID

Ostia

Manacorda

SEC.DI/F.HD

Af I/II

Reggio Calabria

Manacorda 1977, 198 n.47b

*Unknown findspot

AFIID

n.45;

Zevi

&

1984, 187

1977, 198 n.47a

Zevi

&

The distribution of stamped amphoras from Byzacena

141

Ü

4

q e"

Sg

J

.