10. Problems and Possibilities in Comparutive Survey: A North African Perspective

David L. Stone

In the last 25 yearc, has witnessed a boom in field HISTORY OF SURVEY RESEARCH IN TUNISIA survey archaeology. The success ofregional survey pro- Landscape archaeology has a long and distinguished jects in the northern Mediterranean, together with a history in North . Surveys of the countryside were govemment-sponsored program to inventory all archaeo- among the earliest investigations undertaken in the l9th logical sites in Tunisia, has encouraged archaeologists to century. The Atlas alchëologique de la Tunisie, compiled begin l9 programs of surface survey in this country. With using reports made by French archaeologists, soldiers, the explosion ofthis research, survey investigations now and surveying teams working in the late l9th and early form one of the main bodies of evidence for Tunisia's 20th centuries, was a formidable achievement of archaeo- rural landscapes, and significant patterns of landscape logical site recording (Babelon et al.1892-1913;Cagnat exploitation are emerging from thege studies. et al. l9l4-32). Each sheet plotted sites in a 20- by 30-km The benefit ofsuch an extended focus ofattention on region at the scale of l:50,000 and located all the known the countryside is a vast databank of information on rural archaeological sites, but described only a selection ofthe sites; the challenge for archaeologists comes in discerning major ones in an attached explanatory text. For its time, how to evaluate that information. It is easy to be pessimistic this was an in-depth assessment of the rural landscape, about the data. The emphasis of recent suwey work in this even though many of these sites were no more than 'dots part of the Meditenanean has fallen heavily on recording on maps,' smaller sites were inevitably overlooked, and previously identified sites and monuments, rather than on the quality ofrecord-keeping for the 60 individual sheels finding new sites or undertaking rigorous, systematic varied tremendously. An archaeological atlas with a similar investigations of the countryside. Such a methodological structure was compiled at the same time in Algeria (Gsell yet orientation has not enabled archaeological research in l9l1). Tunisia to play a role in understanding issues of social Despite this impressive early research, very little survey organization, nor has it led many ofthe projects conducted work was carried out between l9l0 and the late 1970s. there to interest a wide audience of archaeologists. Com- The only landscape studies to take place in this period parative quantitative analysis of the collected data is occurred around Sufetula, on Cap Bon, and along the feasible on a limited basis, when ideally it would be southem frontier ofRoman expansion. These three projects desirable to use much more frequently. Nonetheless, the primarily involved recording known sites. Near Sufetula, basic quantitative data from fieldtsurveys are available in the Brathay Survey studied Roman-period farms within a sufficient depth that they can used to construct a þe l0-km radius ofthe town to understand regional economic 'landscape history'. Even if that history is legible in general development (Addyman 1962: Addyman and Simpson rather than specifìc terms at the present time, a good 196ó). On Cap Bon, researchers investigated the spatial argument can still be made for the usefulness of a com- organization of Punic settlements and cemeteries, most of parative study of the survey data in their current state. which were located near the coast (Acquaro et al. 1973; Based on the results compiled to date from those l9 field Barreca and Fantar 1983). On the southern border ofthe survey projects undertaken in the last 25 years within , the Limes Tripolitanus Survey examined Tunisia, this study will demonstrate that comparative the purpose of the forts, barriers, and outposts erected analysis ofsurvey data clearly indicates long-term changes there (Trousset 1974). in the countryside, while recognizing the problems as well Modern archaeological f,reld survey research is a de- as the possibilities inherent in this endeavor. velopment of the last quarter-century, when interest in landscape studies re-emerged vigorously. Beginning in the late 1970s and early 1980s, l9 survey projects, many "t

Problens and Possibilities in Comparative Survey: A North African Perspective 133

Survey Publications

Carle archéologique BenBaaziz1992 Bir Mcherga Maurin and Peyras '1991 El Meknassi Ben Baaziz 1991a Gafsa Annabi 1991b; Ben Baaziz 1991b Mididi Ben Baaziz 1966 Oued Cherita Annabi 1991a Oued Sejnane Ghalia l992 Rohia Ben Baaziz 1986 Sidi El Hani ' Ben Baaziz 1988 Sousse , Annabi 1988 Sraa Ouertane Ben Baazlz 1986 Carthage Survey Greene 1983a; 1983b; 1984; 1986; 1992; forthcoming; Greene and Kehoe 1995 'l Coastlines Survey Ben Lazreg et al. 1995; Bonifay et al. 1 992; Chelbi et al. 1995; Paskoff ef a/. 991 ; Paskoff and Trousset 199'l; Trousset 1992 Dougga Survey De Vos 2000 Jerba Survey Fentress 2000; 200'l Kasserine Archaeological Survey Hitchner 1988; 1989; 1990; 1995; Neuru 1987 Leptiminus Archaeological Survey Mattingly 1992; Mattingly et al.2000: Stone ef a/. 1998 Segermes (Project Africa Proconsularis) Die|z et al. 1995: ørsted et al. 2000

Sahel Pottery Survey Peacock et a/. 1 989: 1 990; Peacock and Tomber 1 989 Sufetula-Masclianae Survey Barbéry and Delhoume 1982 Tablel0.l BibliographyoftheprincipalpublicationsofrecentfieldsurveyprojectsinTunisia.(Mostproiectsarelisted undel the name by which they refer lo themselves. The Segermes Survey is also known as Project Africa Proconsularis. The 'Coastlines Survey' is a shortened version of theProgramme tuniso-français d'étude du littoral de la Tunisie. 1¡¡ låe case of the Sufetula-Masclianae Survey and the Sahel Pottery Survey, the name has been assigned. The Carte archéologique includes multiple regional investigations, listed individually in lhis table, which are largely comparable to the other survey projects.)

of which are still ongoing, have generated a large and THE SURVEY DATABASE growing database of information on rural settlement At present, Tunisia must be acknowledged as having one patterns. In this respect, Tunisia has lagged behind some of the most comprehensive regional archaeological data- of its Meditenanean counterparts, such as Italy and Greece, bases in the Meditenanean. The accelerated pace at which where survey work between the 1960s and 1980s led to held surveys have started her€ is remarkable, although the the testing of new theories and methodologies (McDonald rapid growth of suwey as a research methodology in a and Rapp 1972; Renfrew and Wagstaff 1982; Cherry number of regions of the Mediterranean emerges clearly 1983); indeed, the impetus provided by successful endeav- from other papers in this volume. ors in the northern Mediterranean is partially responsible The 19 recent survey projects fall within most of the for the application of survey techniques in Tunisia. As main geographic a¡eas ofthe country. They have occurred another catalyst, one might point also to the inauguration on the north coast, in the upper and lower Tell, the Sahel, the Carte nationale desosites archëologiques et des of the high and low Steppes, and on the island ofJerba in the hisloriques Carte archëologique), monumenls þeréafter, south (Figure 10.1). Such broad geographic coverage a large-scale initiative to inventory all archaeological makes it possible to explore variation in the archaeology remains in the country (BenBaaziz 1992).The Tunisian of several regions of the country. The wide distribution of govemment has sponsored the Carle archéologique in surveys throughout the country, therefore, supplies an order to create an inventory ofsites, as well as to protect additional motivation for a comparative analysis of their against their destruction in the face of rapid modern results. development, and to encourage greater public awareness Only the Dougga and Segerrnes Surveys have published local heritage. Table lists the l9 recent projects, of l0.l final book-length reports (De Vos 2000;Dietz et al' 1995; with references to their principal publications; I 0 ofthem Ørsted et al. 2000).It is thus important to stress that any were conducted in the context ofthe Carte archëologique. conclusions drawn from a comparison ofsurveys here are preliminary. On the other hand, most projects have com- pleted their fìeldwork and are currently compiling and David L. Slone

./r,.l Lower Oued Sejnane Tel/ Cañhage E I Dougga ( Sraa Ouertane

! [ vioioi Leptiminus s¡di ! nonia el-Hani Oued A geria \ Cherita

Sahe/

H¡gh Sfeppe

El Meknassi I ttFll L-. Gafsa

N Figurel0'l RecentfieldsurveyprojeclsinTunisia.(TheSahetPotteryProject[notshown]coversalargeareabetween

I h e Lept im in u s a nd Sufe tu I a- Mas c I ia n ae Su rveys. j

. preliminary analyzing their data for publioation. reports Oued Cherita, Rohia, and Sraa Ouertane Surveys). Ade- in journals or conference proceedings have varied from quate maps of individual sites are also lacking, again detailed examinations of individual sites (e.g. Kasserine particularly (although not exclusively) among the reports Survey), to overviews of regional trends (e.g. the Coast- of the Carte archëologique surveys. Despite signifìcant lines, Jerba, and Sufetula-Masclianae Surveys), and to differences in extent and quality of publication, all 19 inventories of types of sites discovered (e.g. the Sidi el projects have been included here in order to create as Hani and Sousse Surveys). Some of these reports have large a database as possible for interregional comparison. considered the importa The volume of the data already collected aid presented about cultural change by these surveys is more than sufficient to justiff com- (e.g, the Kasserine and parative analysis. especially those ofthe presented their findings in depth, nor come to grips with the data in any synthetic fashion (e.g. the Cafsa, iøi¿i¿i, Problems and Possibilities in Comparative Survey: A North African Perspective 135

Suruey Project Area (sq. km) Team Size TranspotT A¡lifacts lnterdisc. Off-site data Carte archéologique Projects Bir Mcherga 320 2-3 car El Meknassi 640 2-3 caf Gafsa 2560 2-3 car Mididi c. 100 2-3 car Oued Cherita 640 2-3 caÍ Oued Sejnane 610 2-3 car Rohia c.100 2-3 car Sidi El Hani 640 2-3 car Sousse 420 2-3 car Sraa Ouertane c.100 2-3 car Other Projects Carthage Survey ' 825 1 car ./ Coastlines Survey >1 300 4+ car Dougga Survey 150 4+ fwn Jerba Survey 580 4+ fw ./ Kasserine Archaeological Survey 75 4+ fw ./ Leptiminus Archaeological Survey 10 4+ fw ./ Segermes Survey c.150 4+ fw ./ Sahel Pottery Survey 20000> 4+ car ./ Sufetula-Masclianae Survey c. 400 2-3 car Table 10.2 Methodological comparison of recent surveys in Tunisia (Key: fw : fi.eldwalking; interdisc. = inter- disciplinary.)

METHODOLOGY people. They investigated new sites where they discovered them in the course of examining known ones, or where At first observation, the recent survey projects present local informants pointed them out. The differences in severe problems of comparability. They have been con- methodology employed by these two groups of Tunisian ducted by both indigenous andforeign scholars, working survey projects conform to the well-known case of'earlier' in several geographic zones, with variably-sized teams and 'recent' Greek surveys examined by John Cherry. His and budgets. But perhaps the most critical point to em- study demonstrated the impact of methodological differ- phasize is that their directors have brought different ences on site recovery (Cherry 1983: fig. l; see also methodologies to bear upon the landscapes they have Wilkinson et al., Ch. 14, this volume, Figure 14.1), studied. Table 10.2 displays the methods used by these providing empirical evidence that 'intensive', closely- surveys. The purpose of the summary in this section is not spaced fieldwalking procedures tended to find 70-100 to debate the relative merits of different techniques (such times the density of sites than more 'extensive' archaeo- a debate might be indeed be worthwhile for Tunisian logical reconnaissance investigations. In addition, it showed archaeologists to hold elsewhere, since there are real that the more intensive surveys found a greater proportion differences of opinion), but to provide an overview of of small-sized sites, and thus revealed a more complete their methods without which any comparative assessment settlement hierarchy. would be impossible. This overview is intended to sum- Table 10.3, a compilation of the basic results from the marize methodological differences without presenting an surveys, reveals a similar, though less pronounced, pattern exhaustive treatment of every single one. , in the Tunisian evidence. The survey projects which have employed rigorous, systematic methods in collecting their Survey 'Intensily' data have found many more sites than the others. Each of the more intensive projects has recorded among the highest Two groups of intensive' and 'extensive' survey projects totals of sites per sq. km, as shown in Figure 10.2. In all stand out in Table 10.2. Surface survey by freldwalkers at situations the more intensive the survey methodology a high level of intensity (i.e., at an interval of 25 m or less) applied, the more sites were found. Vy'here it was possible has been performed by only five projects in Tunisia to make a comparison to earlier surveys, the Atlas Archéo- (Dougga, Jerba, Kasserine, Leptiminus, and Segermes logique of the 1880s and 1890s in all cases found fewcr Surveys). On these surveys, several people (at least three, sites than the surveys of the 1980s and 1990s (Figure and usually five to seven) formed the fieldwalking teams. 10.3 The research design ofthe other surveys mainly involved ). re-documenting sites previously found by the 19th-century surveys (see Ben Baaziz1992 lor the Carte archéologique; Site Definilion and Function bibliography in Table 10. I for the other surveys). In most What individual survey projects call a 'site' a notorious cases, archaeologists on these survey projects explored the - definitional problem in regional archaeology often landscape by car and worked in teams of one, two, or three - 136 David L. Stone

Suruey Proiect Area (sq.km) Iofal sr'fes Prehist. Pre-Roman Roman lslam¡c Modern Atlas

Cafte archéolog¡gue Bir Mcherga 320 158 0- 4. 143 NR - 210 El Meknassi 640 85 5. 2. 50 NR NR NC Gafsa 2560 291 30 max. 30 max. 200 85 max. 85 max. 48

Mididi c.100 75 5 72 NR Oued Cherita 640 176 2 2 NR NR NR NC Oued Sejnane 610 165 2. 37 80 NR 85 25 Rohia c 100 22 3 19 NR Sidi El Hani 640 '125 NR NR 104 NR NR 81 Sousse 420 186 2. 4. 128 39499 Sraa Ouertane c 100 65 0. 42 26 NR

Other Projects Carthage Survey 825 136 57 121 26 40 104 Coastlines Survey >1 300 209 NR NR NR NR NR Dougga Survey 150 206 NR 63 132 2 6NR Jerba Survey 580 446 NR '183 216 276 273 NR Kasserine Survey 75 164 0 23 149 0 12 NR

Leptiminus Survey 10 30 0 3 17 4 6 6 Segermes Survey c.150 c.250 4 I c.220 NR NR 80 Sahel Pottery Survey 20000> 29 29 kilns

Suf.-Masc. Survey c. 400 125 't25 NR

Tablel0.3 BasicquantitativeresullsoflherecenlsurveysinTunisia,basedonnumberscompiledfromthepublications * listed in Table 10. I . (Keyr : datafrom the Atlas préhistorique de la Tunisie,' - : survey did not investigate this period; NR : not reported in publication; NA : not available; NC : not covered åy Atlas archéologique)

involves complex issues of interpretation and analysis ftom agroville to villa, opus africanum farm, courtyard (Cherry et al. I99l:31-54; Gallant 1986). In general, the farm, and small structure (Hitchner 1988). The Carthage more recent Tunisian surveys do not disagree over what Survey identified three levels of Roman rural sites: villages, they consider a site, but they have many more differences villas, and small-scale farmsteads (Greene forthcoming). in the use of intensive fieldwalking procedures, collection The rest ofthe surveys have additional categories ofsite methods, and interdisciplinary approaches. By contrast, and more or less elaborate site hierarchies. Some of these the surveys display a remarkable similarity in their use of differences may be due to real variations in the make-up of the term 'site', employing it to describe a distinct con¿en- settlements in the survey region, but they indicate one of tration ofartifacts related to recurrent patterns ofhuman the basic ways in which data from individual projects are activity in the i past.' not directly compatible with those from others. Such a definition distinguishe{a site from, for example, lithic artifacts or potsherds not found in association with other artifacts but likely to derive from 'off-site' activities. Artifact Colleclion and Sampling Methods Investigation of off-site data, which cân reveal specific Collection of materials from individual sites provides a agricultural activities, refuse disposal, and intensity of quantitative basis for the analysis oftheir dates ofoccu- landuse (see Alcock et al. 1994; Foley l98l; Wilkinson pation, their functions, and their relationships to other 1982; 1989), is a feature offour Tunisian survey projects sites. Most practitioners of Mediterranean survey would (the Dougga, Jerba, Leptiminus and Segermes Surveys). agree that a multi-stage design of increasingly rigorous Although there is basic agreement among the Tunisian collection procedures will produce the most reliable surveys regarding the definition of site and off-site ma- results. In the recent Tunisian surveys, only the Carthage, terials, there is little uniformity in the assignment of a site Coastlines, Dougga, Jerba, Kasserine, Leptiminus, Sahel hierarchy. The Kasserine Survey, for example, has de- Pottery, and Segermes Surveys have so far collected signated a hve-tiered typology of Roman rural sites ranging artifacts from sites (Table 10.2). These projects also Problems and Possibilities in Comparative Survey: A North Afrfuan Perspective t31

2.5

E I 151 ú o o i ct tn q, È U' rl

I

ou]

I olII ill tu ,€ oPuP *.d *t".-{.{3,""' "*'

ß)

300

200

(,to .ñ 3 tso ¡¡6' E z

100

Leptiminus Oued Sejnane Segermes Sidi El Hani Sousse Carthage Bir Mcherga Survey Project Fígure 10.3 Comparison ofsites found by Igth-century and recenl surveys in Tunisia. 138 David L. Stone

incorporated a multi-stage component into their research the employment of variant chronological schemes or dcsign in which sites werc rcvisitcd, and artifact collection terminologies. Alcock (1989: I l, n. 20; 1993) found the took place with more thorough sampling strategies on use of multiple chronological schemes to be a diffrcult, subsequent visits. The other projects (the Sufetula-Mas- but not irresolvable, problem in her comparative study of cfianae and all of the Carte archéologique surveys) did Greek surveys. In Tunisia, individual surveys constructed not gather artifacts at any stage. Sadok Ben Baaziz, the their own chronological systems without regard to whether director of the Carte archëologique, has argued (1992: they conformed with the systems used by other survey 13) that this approach is not relevant for the broad and projects, presenting problems of data compatibility. For rapid coverage of the country which his project aims to example, the Carthage and Segermes Surveys spent con- accomplish; therefore, ârchaeologists working on the Carle siderable time collecting and dating pottery from their archëologique simply estimate the date of a site based on sites and have broken them down by arbitrary 50-year the brief observation of su¡face matérials. On small sites, (Segermes) and 10O-year (Carthage) periods. Such cate- if the density and scatter of artifacts are restricted, a gories are more precise than those ernployed by the subjective analysis of date and function may be fairly Coastlines and Dougga Surveys which subdivide pottery easy to make; but on larger sites, where the surveyor may from broad periods (such as Roman) into multiple com- not have had the ability to investigate all remains closely, ponents (such as early, middle, and late Roman). By the subjectivity ofthe analysis inevitably increases. For a contrast, the Carte archëologique has not made artifact functional interpretation ofeach site, the Carte archëo- collection a priority: it assessed the date ofa site by brief logique relies upon surviving architectural elements, observation of surface remains and therefore did not divide whereas the collection ofartifacts enables other projects periods into multiple units. Recorded Roman sites date to draw on a wider array of material evidence. These between the first century BC and the seventh century AD, differences in collection methods mean that detailed but only occasionally did the author of a Carte archéo- information about sites is available from some surveys, logique publication specifo a smaller window within these while others have gathered basic data only. eight centuries.

S ingle-per iod E mphøs is Interdiscip linary Studies Virtually all of the recent survey projects have examined Contrary to many current Mediterranean investigations, sites of all periods from prehistoric to modern, but it is f,reld survey projects in Tunisia have concentrated nanowly clear that Roman archaeology has been their principal on the archaeological record alone and have avoided focus. Two have even concentrated exclusively, or almost interdisciplinary investigations. Two exceptions are the exclusively, on Roman data (the Sahel Pottery and Sufetula- Segermes and Coastlines Surveys. At Segermes, palyno- Masclianae Surveys). Finally, Roman archaeologists com- logical and geological studies have substantially improved. prised the principal staff even on survey projects which knowledge of vegetational history and the formatioq of evaluated sites from all periods in their regions (the Carte the natural landscape (Dietz et al. 1995). The union of archéologique, Dougga, and Leptiminus Surveys). In part , geomorphological and archaeological techniques on the the single-period emphasis is due to the fact that rural Coastlines Survey has provided a detailed landscape settlement was widespread in the Roman period, and reconstruction of the lower Medjerda valley over the past therefore the presence ofRoman archaeologists in Tunisian 6,000 years (Chelbi et al. 1995). Other surveys carrying regional surveys is essential. But this emphasis also derives out interdisciplinary analyses in Tunisia have yet to publish in part from the guidelines of some permits issued to foreign many of their studies, though that picture may change archaeologists, which stipulated that in-depth reJearch before long. could be carried out only on the Roman-period sites. The Segermes Survey, which includej specialists from a number of different periods (Prehistoric, Punic, Roman) and Summary disciplines (archaeology, ceramics, geology, palaeobotany) It is obvious that the quality of the data available from on its staff, appears different from the other surveys, Tunisiansurveys varies significantly. Archaeologists who although even this project systematically examined material consider comparative studies of survey databases else- from only the Prehistoric, Punic, and Roman periods. The where similarly face this critical problem. Are some Jerba Survey,co-directed by Roman and Medieval archaeo- surveys simply too extensive in approach to provide any logists, is the sole project that has a fully diachronic focus, worthwhile information? Should one omit the evidence and its results accordingly show a much greater balance of from these surveys and evaluate instead only the more site numbers between periods (Table 10.3). rigorous ones? Beyond this question, of course, lie further problems of comparability among the intensive surv€ys, such as the use ofdifferent collection methods or chrono- Site Dating and Artifact Chronologies logical systems. How big a role do these methodological One issue affecting use ofdata from different surveys is differences play? Problems and Possibilities in Comparative Survey: A North African Perspective 139

Suruey project Suruey area (sq. km)

Kasserine 75 149 40 0.13 o.27

Mididi 100 72 29 0.29 0.40

Segermes 150 220 27 0.18 o.12

Sufetula-Masc. 400 1% 44 0.11 0.35

Rohia 100 19 11 0.'t 1 0.58 '5 Sraa Ouertane 't 00 % 0.05 0.19

Oued Sejnane 6'10 80 22 0.04 o.28

Carthage 825 121 13 0.02 0.11

Sousse 420 128 4 0.01 0.03

Sidi El Hani 640 104 2 0.003 0.02 Table 10.4 Olive oil presses found by recent Tunisian surveys (ranked by presses/sq. km).

COMPARING THE DATA new agricultural installations. First, the hill-country in the Given the preliminary nature of many of the survey hinterland of Carthage, where the new 3rd- and 2nd- publications, as well as the methodological discrepancies, century BC sites are located, consists ofeasily exploitable there are limits to the questions that can be asked. It seems soils. Second, the overall site distribution pattem in the best to begin with an evaluation of site numbers and site Roman period, when these sites contain numerous agri- types for both the pre-Roman and Roman periods, In the cultural features (presses, millstones, inigation), is quite following section, a synthetic arnlysis of the l9 most recent similar. Finally, the presence of Punic fine wares and projects presents the long-term trends in patterns of amphoras also suggests occupation (Greene, forthcoming). settlement and landuse which the surveys have docu- The Jerba Survey has likewise documented the presence mented. of rural sites in the 3rd and 2nd centuries BC. On Jerba, rural villas and smaller farms make up a landscape that seems geared toward surplus production of wine and olive Pre-Roman Period oil, based on the evidence ofpottery from nearby kilrs. From approximately the 7th to 2nd centuries BC, the These products were likely exchanged for cereals, as emerging Numidian kingdom and the Carthaginian state subsistence crops fare poorly on the island (Fentress 200 I : controlled different parts of Tunisia. To simplify the 262-64). The survey has also documented the presence of terminology, this period will be defined as pre-Roman, isolated tombs and cemeteries in the rural landscape. rather than Protohistoric, Punic or Numidian; each of the Thus, the evidence from recent surveys argues for latter, though commonly used, refers to part but not all of significant developments in Tunisian landuse in the 3rd Tunisia during this time period. Survey data from the pre- and 2nd centuries BC, although rural exploitation varied Roman period indicate several activities taking place in by region. At Carthage and on Jerba, isolated rural settle- the countryside and significant regional variatidn in the ments as well as moderately-sized villages first fìlled the types of sites found. A small number of towns and tombs countryside at this time; such organization, in which have been identified in the areas of the Dougga, Jerba, communities relied on small-holders and peasants who and Leptiminus Surveys. The$raa Ouertane and Kasserine commuted between town and country, is typical of more Surveys, respectively, found42 and 23 megalithic tombs, developed societies in the eastern Mediterranean in the but no settlements. Cemeteries containing rock-cut tombs contemporary Hellenistic period. Elsewhere in Tunisia, known as haouanel are present in other survey regions where surveys made an effort to study the pre-Roman such as the Oued Sejnane (37) and Segermes (2)' period, high concentrations of funerary monuments were In only two survey areas is there much evidence for pr€sent in the rural landscape, These areas contained widespread rural activity beyond the construction oftombs' haouanet, or megalithic tombs, which seem to have The Carthage Survey data show minor occupation ofthe developed in the 4th, 3rd, and 2nd centuries BC in response city's hinterland during the 7th to 5th centuries BC (7 to the increasing expropriation of territory by the Numidian sites) and the 4th century BC (9 sites), but a very significant kings and Carthage. The overall picture - ofa landscape rise in settlement density marked the 3rd to 2nd centuries with rural farms emerging in two coastal zones, but BC (50 sites). Three factors suggest that the expansion of dominated in all other areas by small towns and cemeteries rural activity may well have been linked to the presence of - is clear. t40 David L. Stone

250

I Roman sites i 200 I tr Pre-Roman___-_l sites I

oth .Ë 150 U) o (,L .C¡ E 100 zf I

50

0 I :LIL d*""".,$ ...oo.opr-o. "'"

Survey Project Figure 10.4 Pre-Roman and Roman site numbers compared in several survey areas.

Roman Period may have equalled or nearly attained the levels produced During the Roman period the number of rural sites in- around Kasserine (Table 10.4). The Roman-period land- creased significantly in almost every region of Tunisia. scape in Tunisia contains additional features such as roads, Table 10.3 presents the total number of sites for all periods bridges, cemeteries, boundary stones, and kilns, although studied by the recent Tunisian surveys; Roman sites farms signifìcantly outnumber other site types (see the surpassed pre-Roman sites in all but one case, often by Dougga and Segermes Surveys). The chronology ofrural several times. The high density of rural sites occupied in development is not well documented, though the ,bäom, the Roman period appears clearly in Figure 10.4. The visible in the survey data appears to start earlier and finish character ofthe new sites, moreover, signifies an important later in coastal regions. change in the nature of landuse in the Roman period. All Beyond this general picture, individual survey areas ofthe surveys found evidence for rural farms, a form of exhibited some difference in settlement patterns during settlement which became the new ordering mechanism of the Roman period. Figure 10.5 displays the number of the landscape at this tin.re, in contrast to the presence of Roman sites per sq. km found by all ofthe surveys. These farms in just two of the survey areas from the p..-&.ornun numbers relate to actual densities of settlement in antiquity, period. Roman-period farms have courtyards with attached but are also shaped by other factors such as survey buildings, cisterns, and a rangp of ceramic wares, arguing intensity, the durability of local building materials, and for year-round occupation, rather than seasonal use at the post-abandonment conditions. Nonetheless, the numbers large majority of the sites (e.g. the Dougga, Kasserine, demonstrate regional variation, with many sites appearing and Segermes Surveys). The substantial number of sites in surveys in the high steppe and Sahel and comparatively with agricultural features, such as olive-presses, millstones, fewer sites per km in the low steppe, upper Tell, and and water-related facilities, points to the surplus pro- lower Tell. Within Tunisia, there were 'discrepant ex- duction and cxport ofgrain, and olive oil, a link especially periences' utder Rouran rule, as the decline in site numbers well attested by the Coastlines, Dougga, Kasserine, Lepti- in the Sraa Ouertane in particular suggests (Mattingly minus, Mididi, Sahel Pottery, and Sufetula-Masclianae 1997; Said 1993:3143). Some regions resþonded more Surveys. Estimates of total production capacities of olive rapidly or favorably to the imposition of empire, although oil for the Kasserine Survey area in years with average or the general trend points toward a signihcant expansion of high rainfall reach well beyond regional subsistence rural settlement everywhere. requirements (Mattingly 1988; 1994). As the densiry of Based on the evidence collected by the surveys, it is olive-presses in other survey areas indicates, some regions currently impossible to carry this analysis further into the Problems and Possibilities in Comparative Survey: A Norlh African Perspeclive r4t

2.50

2.00

xE ú 1.50 an o CL th 't.00 Ë0, th ',

0.50

0.00 rrr-lr-rrlil tl ---d-' -¿ - *uo rn*..o"d o" ""."i "."{ø'S ".t- adtù'

Figure r 0. 5 site r"":;å";"::i lo^on pè,ioa.

Arab, Ottoman, or French colonial periods. With the tion, since this evidence can be assessed in quantitative exception ofthe Jerba Surveyrirecent Tunisian projects terms. have placed very little emphasis on recording material Even though many projects have not yet completed from these periods; it is true that excavations have neg- their analysis and publication, this comparative study has lected to undertake detailed relevant site and ceramic been useful in taking stock of the present situation. It has studies as well. An inability to recognize ceramic evidence illustrated the methodological shortcomings of current from these periods may contribute to their present low site work and enumerated areas to be explored in future totals. investigations. Nor should archaeologists be hesitant tp undertake such an analysis before final publication of results occurs; a reluctance to undertake a compaiative analysis of survey dataatan early stage will only prolong CONCLUSION the time before survey data become useful and will only Although recent surveys in Tunisia ranged from extensive delay the development of a suitable agenda for future projects conceived for the purpose of recording the pre- research. sence of sites and monuments, to intensive projects aimed The results summarized here should offer encourage- at producing detailed multi-period reconstructions of ment that survey work in Tunisia has a key role to play in landscapes, such methodological variance has ñot pre- the future. At the moment, several concems must be raised. cluded comparative study. From the pre-Roman to the Field surveys in Tunisia must strive to include a wider Roman period, as the Tunisiau countryside came to partici- variety ofspecialties as regular foci ofinvestigation. Too pate in a wider sphere of economic and cultural relations, few surveys have considered geomorphological, palaeo- the survey data from many regions lead to a similar environmental, and other questions pertaining to landscape conclusion. There appeared a transformed landscape, evolution, in addition to acquiring archaeological data. characterized by frequent rural farming settlements and Reconstructing the history of fertile and marginal zones agricultural facilities, rather than scattered towns and will greatly improve the comprehension of landuse and cemeteries. Without evidence from field surveys, this settlement patterns. Detailed ceramic analysis must also transformation would hardly be so visible. Prior to the become a more prominent feature of survey reports since recent survey projects, scarce references in ancient texts, it is impossible to document the extent of occupation or limited surface reconnaissance, and inadequate - mostly the range of functional activities on a site without know- unscientific - excavations offered only a very imprecise ledge of its ceramic assemblage. Two potential issues to picture of the changes between the pre-Roman and Roman be studied through ceramic analysis include the chronology periods. There is no doubt that on its own the survey of the appearance of rural farms in the country and the evidence provides a more accurate historical reconstruc- arrival in quantity of imported consumer goods{signified, David L. Stone t42

amphoras)' A for example, by the presence of Dressel 2/4 ãìu"¡.or,ì" approach to regional history must also be adopted. The distribution is poorlY understood, th nazionale de the landscaPe maY signal dans la.plaine de Ben Baaziz, S. pation humaine the Arab lrt Colloque sur side during Period, Rohia et le 6ans I'antiquité' era may point to the arrival du Nord 3: 289-300' ioi ir.lguti-on in the óttoman l'Histoit'e et de I'Afríque three issues oiu n"i phase of olive oil exportation',These Paris, CTHS, attention antiques de la région de Sidi El i*J.*y could be illuminated by greater --FtenBaari",S. (1938) Lçs sites "tfters) The suitability de l'lnstitut National de I'art et I'archéologie ìo all material .uid"n.. in the countryside' fl""i. Butletìn Tunisian of it"fA."*"y as a technique for the study øf the in order to landscape no longer needs justifrcation, but - projects nrust ,nake nË* contributions - individual survey concerns begin to address some of these methodological in a serious fashion.

ACKNOWLEDGEMENTS I would like to thank Sue Alcock, John Cherry' Bruce advice Hir.hrr"., and David Mattingly for their excellent in unJrrrgg.*ions while I worked on the topic discussed for the this paper, f,trst as part of my Ph'D' dissertation Uniuotlay of Michigan, and second in the course of this data in conferenðe. Elizabeth Fentress kindly offered Bo the Jerba Survey publication' Several other advance of 'il,üì'il;i;de I'lnstitut offered comments on my paper members of the conference le 5' Janvier- for which I am also grateful' Juin A' (1914-32) d;;i, Reinach, S' and Merlin' Atla laTunisie' 2nd edition' Paris' E' Leroux' Chelbi, P' (1995) La baie d'Utique et REFERENCES son ,Trousset'antiquité: une réévaluation géoarchéo- A' and Fantar' M' (1973) "-iìorpui;on"Acquaro, 8,, Bartoloni' P', Ciasca' archeologica al Capo Bon I' Collez.ione di studi ch ¡eniciZ. Rome, Consiglio nazionale delle richerche' area: interim ''-i.liirr.¡,¿áv*un, P. (196i) Theãrchaeology ofthe Bri,nay'Exploration Group Annual Report: 60-?7 ' Ad;;;";, e., ani sirnpson' W' G' (1966) Archaeologv of the ' -õU.itf Exploration Group u u."a: ,e"ond interim report' Bralhay Annual RePort:153-70' in the Gieek landscape' Alcock, S. E. (1989) Roman imperialism Journal of Roman Archaeolog 2: 5-34' tandscapes o/Rom&n Alcock, S. E. (1993) Graecia cTpa: The P'"ttl Gre,ece. Cambridge, Cambridge University .. lntensive 'Alcock, S.8., Cherry, J' F', and Davis' J' L'.(1994) tde Classical of ,uru.y, agricultural practice' and .landscape C,"."". tn'I. Morris (ed') Classical Greece: Ancient Hislories Cambridge and Modern Archaeologies: 137-70' Cambridge' UniversitY Press. dans,la région de Annabi, M. li. (fSSg) Prospection archéologique ' el I'archéologie Sourr". Bulietin de I'lnstilut Nolional de I'art Comptes Rendus (Fascicule 2): 17 -3 l' A"".bi,ï. K. (l99la) Arché rlogie de paysage à oued Cherita' 'art 'archëologie Comptes AìulLt¡n ¿" t lrstitut National de I et I Rendus (Fascicule 4):'l-28 en Tunisie: le Annabi, tr¡. k. (rSStU) Prospections archéologiques sitedeHenchirBirEnnadhour'lnColloquesurl'Hisloireet Editions iÀrchéologfe de t'Afrique du Nord 4: 549-54' Paris' du CTHS. (I892-19I3)Itlas archéo- 13:403-41 8' Babelon, 8., Cagnat, R' and Reinach' S IogY (1992) Approches du paysage antique de la vallée de logique de Ia Tunisie' Paris, E Leroux' Cnatä T. Problems and Possibilities in Comparative Survey: A North African Perspective t43

I'Oued Sejnane (Feuille n. 5 au l/50,000). Bulle¡in de I'lnstitut perience in lhe Roman Empire: ll7-39, Journal of Roman t. National du Palrimoine Comptes Rendus (Fascicule 5. Janvier- Archaeology Supplementary Series 23. Portsmouth, RI, Journal Juin 1990):33-57. of Roman Archaeology. I Greene, J. A. ( I 983a) 'Atlas archéologique de la Tunisie' and recent Mattingly, D. J., Stone, D. L., Stirling, L. M. and Ben Lazreg, N. l archaeological reconnaissance near Carthage. ln D. Keller and (2000) Leptirninus (Tunisia): a producer c¡ty? ln D. Mattingly D. Rupp (eds.) Archaeological Survey in the Mediîerranean and J. Salmon (eds.) Economie.s beyond Agricullure ín the Area: 133-36. BAR Intetnational Series 155. Oxford, British Ancient World: 66-89. London, Routledge. Archaeological Reports. Maurin, L. and Peyras, J. (1991) Romanisation et traditions Greene, J. A. (1983b) Carthage survey. ln D. R. Keller and D. W. africaines dans la région de Bir Mcherga. Cahiers de Tunisie Rupp (eds.) lrchaeological Survey in the Medilerranean Area: 154-55:105-148. I I 97-99. BAR International Series I 55. Oxford, British Archaeo- Neuru, L. ( I 987) Red-slipped wares ofsouthwestern central Tunisia: e logical Reports. new evidence. Rei Crelariae Romanae Faulorum Acta 25--26: Greene, J. A. (1984) Canadian Carthage survey 1983: preliminary I 75-88. (eds.) 1 report. Echos du monde classique/Classical views: 214-18. Ørsted, P., Carlsen, J., Sebai, L. L, and Ben Hassen, H. 'I Greene, J. A. ( I 986) The Carthaginian Countryside: Archaeological (2000) Africa Proconsularis: Regional Studies in lhe Segermes Reconnaisance in the Hinterland of Ancient Carthage. Ph.D. Valley ol Norlhern Tunisia. Vol. tII: Hìslorical Conclusions. Thesis, University of Chicago. Aarhus, Aarhus University Press. Greene, J. A. (1992) Une reconnaissance archéologique dans Paskoft R., Slim, H. and Trousset, P. (1991) Le littoral de la I'arrière-pays de la Carthage antique. ln A. Ennabli (ed.) Pour Tunisie dans I'antiquité: cinq ans de recherches géoarchéo- sauver Carthage: Exploration el conservalion de la cilë pu- logiques. Comptes Rendus à I'Academie des Inscriptions et nique, romaine, et byzantine:195-97. Paris, UNESCO. B el les-Lel lres: 5 I 546. Greene, J. A. (forthcoming) Ager and 'Arosol: Rural Seltlemenl Paskofl R. and Trousset, P. ( I 99 t ) Les sites submergés de Tunisie. (ed.) Thracia Pontica II/, ac¡es du symposium ) and Agrarian History in îhe Carthaginian Countryside. In M. Lazarov Greene, J. A. and Kehoe, D. P. (1995) Mago the Carthaginian on internalional de Sozopol (octobre I 988): 367-84. Sofia. Centre s agriculture: archaeology and the ancient sourc€s. In M. Fantar d'archéologie subaquatique. I (ed.) Actes du III" congrès international des études phénicíennes Peacock, D. P. S., Bejaoui, F. and Ben Lazreg, N. (1989) Roman el puniques II: I l0-17. Tunis, Institut National du Patrimoine. amphora production in the Sahel region ofTunisi a. ln Amphores économique: dix ans de recherches'.179- K Gsell, S. (1911) Atlas archéologique de I'Algërie. Algiers, A' roma¡nes el histoire Jourdan. 222.Collecaionde I'Ecole Française de Rome il4. Rome, Ecole Hitchner, R. B. (1988) The Kasserine Archaeological Survey, 1982- Française de Rome. (1990) Roman a 86. Antiquités Africaínes 24: 74|. Peacock, D. P. S., Bejaoui, F. and Ben Lazreg, N. ,l Hitchner, R. B. (1989) The organisation of rural settlement in the pottery production in central Tunisia. Journal ofRoman Archae- Ci llium-Thelepte region (Kasserine, Central Tun isia) . L' Afr ica ologt 3: 59-84. romana 6'.387402. Peacock, D. P. S. and Tomber, R. (1989) Roman amphora kilnli in the Sahel ofTunisia: petrographic investigation ofkiln matçrial ) Hitchner, R. B. (1990) The Kasserine Archaeological Survey, 1987. and An tiquit és Afr ic ai nes 26: 23 I -60. lrom a sedimentary environmenl. In I. Freestone A. Mid- ; Hitchner, R. B. (1995) Irrigation, terraces, dams and aqueducts in dleton (eds.) Recent Developments in Ceramic Petrology the region of Cillium (mod. Kasserine). ln P. Trousset (ed') British Museum Occasional Paper 8 I : 289-304, London, Br¡tish Productions et Exportations Africaines: Actualités Archéo- Museum. t logiques en Afrique du nord antique el medievale. Colloque sur Renfrew, C. and Wagstaff, M. (eds.) (1982) An Islqnd Polity: The Cambridge, Cambridge n l'Histoire et I'Archéologie de I'Afrique du Nord 6.1: 143-58. Archaeologt of Exploitation in Melos. Paris, Cahiers des Travaux Historiques et Scientifiques. University Press. McDonald, W. A. and Rapp, G. R. (eds.) (1972) The Minnesota Said, E. W. (1993) Culture and Imperialism. New York, Knopf. Messenia Expedìtion: Reconstrucling a Bronze Age Regional Stone, D. L., Stirling, L. M., and Ben Lazreg, N. (1998) Suburban around Leptirninus (Tunisia): e Environment. Minneapolis, University of Minnesota Press. land-use and ceramic production I e Mattingly, D. J. (1988) Oil for export? A comparative study of interim report. Journal qf Roman Archaeologlt l: 304-317. olive-oil production in Libya, Spain, and Tunisia' Journal of Trousset, P. (1974) Recherches sur le limes Tripolilanus: du choll Roman Archaeology 1: 33-56. el-Djerid à lafronlière tuniso-libyenne, Paris, Centre National Mattingly, D. J. (1992) The field survey: strategy, methpdology, de la Recherche Scientifique. and preliminary results. [n N. Ben Lazreg and D' J. Mattingly T¡ousset, P.(1992) La vie littorale et les ports dans la Petite Syrte (eds.) Leplininus (Lamta): A.Roman Port Town in Tunisia. à l'époque romaine. ln Afrique du Nord anlique el médi¿vale: I'Histoire et 7 Reporl no. /.: 89-120. Journal of Roman Archaeologt Sup- speclacles, vie portuaire, religions. Colloqtre sur I plementary Series 4. Ann Arbor¡ Journal ofRoman Archaeology. I'Archéologie de I'Afrique du Nord 5:317-32. Paris. Editions Mattingly, D. J. (1994) Regional variation in Roman oleoculture: du CTHS I some problems of comparability. In J. Carlsen, P' Ø¡sted and J.- Wilkinson, T. ( 1982) The definition of ancient manuring zones by E. Skydsgaard (eds.) Landuse in lhe Roman Empíre' Analecta means ofextensive sherd samplìng techni ques. Journal of Field Romana Supplement 22: 9l-106. Rome, Danish Institute at Archaeologt 9:323-33. Rome. Wilkinson, T. (1989) Extensive sherd scatters and land-use in- Mattingly, D. J. (1997) Imperialism and territory: Africa, a land- tensity: some recent ¡esults. Journal ol Field Archaeologt l6: scape of opportunity? In D. J. Mattingly (ed.) Dialogues in 3t-46. Roman Imperiolísm: Power, Díscourse, and Discrepanl Ex-