The Lost Domain: the rural archaeology of Africa in the Roman Empire - M DE VOS RAAIJMAKERS, and B. MAURINA, eds. 2019. RUS AFRICUM 4. La fattoria bizantina di Aïn Wassel, Africa Proconsularis (Alto Tell, Tunisia). Lo scavo stratigrafico e i materiali. Roman Archaeology 58. Oxford: Archaeopress. Pp. xiv and 438. ISBN 978-1-78969-115-3.

I first read Henri Alain-Fournier's masterpiece Le Grand Meaulnes (sometimes titled in English “The Lost Domain” or “The Lost Estate”) in the early 1980s, while working on the UNESCO Libyan Valleys Survey in Libya.¹ The archaeology of North Africa's ancient landscapes has been central to much of my subsequent career, but, like the obsessive and passion-driven quest of Augustin Meaulnes, the rediscovery of Africa's lost estates has been far from straightforward.² In part this relates to the fact that this has been a less-frequented domain of study in traditional approaches to “Roman Africa.”³ To be sure, archaeological field survey has made important progress over the last decades and Roman-era rural archaeology has grown in prominence within the discipline.⁴ Of particular importance is the fact that Maghribian archaeologists have become heavily involved in rural field survey.⁵ They are increasingly at the forefront of such work.⁶ However, the amount of excavation work carried out at rural sites remains minimal and little of it is fully published.⁷ This is a major hindrance to knowledge and understanding of the rural communities, so the volume under review here is a most welcome addition to a small corpus of published excavations on rural sites.

Mariette de Vos Raaijmakers and her team are to be congratulated on their work in excavating to a high professional standard and impressively documenting the site of Aïn Wassel, albeit that (as so often in archaeological endeavors) there has been a considerable gap in time between excavation (1994–1996) and final publication (2019).⁸ This is a model archaeological report covering stratigraphic aspects, plus detailed presentation of various categories of finds, including pottery, glass, metal, coins, stone artifacts, and palaeo-economic markers (plants and faunal remains). The volume is a further milestone in an important series of final reports emerging from the project that provide a mass of new information about the rural landscape around Thugga (Dougga). Previous volumes have included a detailed site gazetteer from the rural field survey (vol. 1), focused studies on the aqueduct system for Thugga (vol. 2), and the road network and milestones (vol. 3).⁹ The interpretation of the dossiers of information presented in these volumes is complicated by the fact that there is not much synthetic discussion provided there, though the publication series is still evolving and some impressive overview chapters have been published elsewhere.¹⁰ There are only a small number of comparably well-published studies of rural landscapes in North Africa that approach this level of comprehensive reporting.¹¹

The site of Aïn Wassel, situated on a low hill above a spring, is especially interesting because a Severan copy of the Hadrianic Law pertaining to imperial estates in the Medjerda valley was found there.¹² This is one of a group of seven large inscriptions detailing elements of the administration of multiple imperial estates in the area of the Wadi Medjerda (ancient Bagrada). The excavation thus offers us a stratigraphic window into a world hitherto known only from the epigraphic hearsay of that group of inscriptions. Indeed, in many conventional accounts of the story of Roman Africa, descriptions of rural activity make surprisingly little reference to archaeological evidence at all, relying to a large extent on the epigraphic testimony – a significant difference in comparison to the situation in other provinces.¹³ The Thugga survey has significantly advanced knowledge of the landscape context of the imperial estates in the Medjerda valley area and on their formation from earlier large estates, some of which related to close associates of Augustus, such as T. Statilius Taurus (6–12).¹⁴

As is clear from the English abstract at the head of Chapter 1, Aïn Wassel was specifically targeted for excavation “in order to investigate the work and living conditions of the sharecroppers who had asked Septimius Severus [for] the application of the law” (1). However, as is clear from the title of the volume, what was actually encountered when excavation descended into the rectilinear building range visible at the surface proved of much later date than the expected High Empire settlement. The Late Antique building range incorporated a lot of earlier stonework in secondary reuse in its walls but did not shed light on the layout of earlier features at the site. The volume under review is thus a richly documented account of the Byzantine phase of occupation at the site, but the equivalent structural evidence for the imperial estate of the 1st–5th c. CE and the Vandal period of the 5th–6th c. CE remains largely lost from view, though to some extent attested in the finds record through residual material. This does not detract from the fundamental importance of what has been achieved here, but it is nonetheless frustrating that a prime expected outcome of the project could not be realized because the activity in the area of the site selected for excavation turned out to be exclusively Late Antique in date.¹⁵ A further limitation is the flimsiness of the excavation record from rural sites in North Africa as a whole. As we shall see, the story of Late Antique Aïn Wassel is an important one – though it is ironic also that most of the other published rural excavations to date have also related to Late Antiquity rather than the earlier Imperial period.¹⁶ This has revealed the Late Antique landscapes of North Africa to be much more thriving than sometimes imagined by a colonialist-inspired discourse of decline.¹⁷ But it does not provide a good viewpoint from which to assess the story of the earlier Imperial period.

The volume reviewed here commences with a wide-ranging essay by de Vos Raaijmakers (1–55), which presents the excavation at Aïn Wassel in the context of the landscape and the work of the overall project. The themes addressed are somewhat miscellaneous and the running order of topics a bit idiosyncratic, as she attempts to both introduce and offer a synthesis of some of the results. She provides excellent insights into the landscape setting of the site and its relation to the reconstructed boundaries of the huge imperial estate, the saltus Neronianus, that it lay within. The reconstruction of what is known of Byzantine and Christian sites across a broader area (17–26) is also an important section. There is a long discussion on regional economic activity (including some overview on olive oil production in towns and countryside in Late Antiquity (26–55).

The bulk of the volume is taken up with a fairly standard account of the excavation and the finds. The stratigraphic report by Barbara Maurina is descriptively thorough, proceeding room by room and phase by phase (56–142, with plentiful use of Harris matrices, sections, plans, and photographs), but with only a comparatively brief interpretative overview at the end (142–48). From the surface survey traces of the settlement, it was clear that there were multiple buildings arranged around a large open area. Construction was in the opus africanum tradition, with vertical orthostats at intervals on main walls and infilling in smaller blockwork (57 for a plan locating the excavated area within the overall settlement traces visible at the surface). The construction of the excavated building was dated to the late 6th or early 7th c. CE and its abandonment occurred by the early 8th c. CE. The existence of earlier structures and occupation on the hill can be postulated from a range of other evidence – architectural fragments reused in walls, a number of epigraphic finds including earlier tombstones and the Severan copy of the Hadrianic Edict, and a large volume of residual pottery (primarily of the 4th–5th c. CE). The fact that a building that was not constructed until the Byzantine era was associated with so much Late Roman pottery certainly suggests intensive earlier occupation close by, but it shows the dangers and difficulties of interpretation, when there is such scant comparative excavated data from the region. For this reason, we cannot assume that the stratigraphic sequence encountered applied equally to all parts of the Aïn Wassel hill. Had the trench been differently located within the large settlement area visible at the surface, more evidence of earlier phases of building might well have been encountered, not just of the Late Roman era as suggested by the excavated pottery, but from at least the 1st c. BCE onwards. This point is supported by the survey dating evidence.¹⁸ Indeed Carthaginian coin finds suggest that the ultimate origins of settlement could go back even earlier (see below).

Chapter 3 (149–244) by Martina Andreoli and Silvia Polla presents a detailed report on an assemblage of more than 8,000 sherds of the domestic pottery recovered in the excavations, with 1,250 forms identified. This is the most important rural pottery assemblage published to date from North Africa, though >70 percent of the material was deemed residual in its excavated context. The assemblage is thus of high interest for our knowledge of regional pottery consumption between the 4th and 8th c. CE, although the occupation of the Byzantine building fell in the last 100 years or so of this timespan. This will be a very important assemblage, especially as our knowledge of rural sites and their material culture improves. The vessel glass (302–12, report by Martina Andreoli) constitutes a similarly mixed assemblage, with 132 fragments relating to a minimum of 56 different vessels. Once again much of the fragmented material appears to have been residual within the site stratigraphic sequence, but there are quite a few stemmed goblets of Byzantine type, which provide an indication of flourishing Late Antique convivial practices.

The catalogue of amphora and jars by Barbara Maurina (245–94) is based on ca. 2,000 sherds, with 290 attributable to form. This is quite a large volume of amphora from a small excavated area of a rural production site and demonstrates regional contacts (also commented on in Chapter 1, 36–37) as well as contributing some new additions to the inland (though not immediately local) production of amphorae. The capacity of the newly identified Aïn Wassel types 1–3 ranges from 115–120 liters (36, 266–69), which is at the very limit of what a camel could carry as a pair of balanced vessels. That might suggest that the large amphorae at the site (extending also to large-capacity Sidi Jdidi types), were primarily used for storage purposes at the rural settlement – perhaps in relation to olive oil or wine production,¹⁹ rather than being intended for regular use in the carriage of the produce to the distant ports. Overland transport of estate olive oil was more likely to have made use of animal skin containers. Although dating is not certain for all the types present, and some degree of residuality is evident, the material seems to confirm that Late Antique African amphora production continued in response to still significant rural surplus production of oil and wine.

Of the other finds reports, the ceramic building materials (295–301, by Martina Andreoli) represents a pretty small assemblage of 73 fragments of what are referred to as “roof tiles” (tegulae, with a few fragments of imbrices) and a few bricks and vaulting tubes. This material appears mainly to have been reused as general construction material in walls of the Byzantine farm rather than to have been a Byzantine production for a specific use. In any event the low occurrence levels suggest that tiles were not generally used for constructing roofs in this region. The small collection of 42 metal small finds (313–29, by Martina Andreoli) includes tools, domestic equipment, and items of personal adornment, and the concentration of finds within two rooms of the building favors contemporaneity of loss/deposit with the occupation phases. This is a small assemblage, so wider conclusions are tentative, but again there are indications of a reasonably high standard of living being maintained at the farm well into the Byzantine era. The chronological peculiarities of the site are again evident in the small coin assemblage (330–38, Silvana Abram), where of a total of 14 coins, no fewer than three were Carthaginian bronzes of the 4th–3rd c. BCE, but with no Republican Roman issues and a sole 2nd-c. sestertius. The remaining 10 coins included seven fractions of folles of the 4th–5th c. CE, two Vandal nummi of the late 5th c. CE and a Byzantine quarter siliqua dating to 582–602 CE, this last providing the closest dating evidence of the construction of the Byzantine building.

Chapter 9 (339–62) by de Vos Raaijmakers catalogues stone finds under a number of heads and includes some really interesting material: a sundial, eight funerary inscriptions, a number of architectonic elements (some in situ, some reused), mortars and mills, and numerous olive press elements. Quite a few of these press components were reused as building stone in the Byzantine house, but some at least were part of an active olive processing installation. Again, while the Byzantine building was clearly part of a still thriving farming community, there is a strong presumption from the distribution of the reused stone elements that a large and well-appointed prior settlement had been heavily quarried for building materials (see 36 and 57 for plans showing the location of the reused stone elements).

A most welcome aspect of the report is the documentation of the faunal remains (363–92, by Jacopo de Grossi Mazzorin et al.) and botanical assemblage (393–402, by Daniela Moser et al.). The main assemblage of animal bone (ca. 1,034 fragments, with 334 identified to species) relates to the occupation phase of the Byzantine building and this was the focus of the report. Ovicaprines constituted 34% of the identified material, pigs 28%, cattle 10%, and chickens 14%, with a small number of dog, small mammal, camel, and donkey/equine bones. However, the comparanda sites cited (378–80) are urban and chronologically diverse, again illustrating the paucity of rural evidence.²⁰ The botanical remains (carpological and anthracological) relate primarily to the local trees, notably the olive, the stone pine, the arar, and the narrow leaved ash. It is a pity that there was not more varied material and a deeper chronological base from which to more fully evaluate the regional environment and any changes across time.

There is much of interest in the rich dossier of information made available in this report. As noted already, it opens an important window on rural communities in North Africa, but it needs to be recognized that this is a very narrow spyhole, potentially set with distorting glass. To tangle the metaphor even more, as far as the imperial estate of the early centuries CE is concerned, we are looking through the wrong end of the telescope. One of the most important consequences of the Late Antique focus of the pottery assemblage from the excavations is that it enabled the project team to acquire a detailed knowledge of Late Roman, Vandal, and Byzantine ceramics in the region. This led to identification of much more extensive continuing rural settlement in Late Antiquity at the surveyed sites around Thugga. Indeed, in terms of total site activity, the Late Roman, Vandal, and Byzantine periods have the largest numbers of identified sites based on the ceramic evidence from surface assemblages.²¹ The temptation to view this as evidence of a “gradual increase in the rural population” over time with a peak in Late Antiquity is understandable,²² but on the evidence of the excavation perhaps needs to be tempered. For one thing, it is well known that surface assemblages do not always bear a close resemblance to subsurface ones, with earlier phases often under-represented vis-à-vis late phases.²³ The high level of site continuity well into Late Antiquity in the Thugga region is certainly a very important phenomenon, but ceramics of this date may be over-represented in the surface material assemblages collected in survey, obscuring evidence for earlier phases of site occupation. A related point is that the project's excavated Late Antique assemblage has potentially amplified the range and volume of diagnostic material that might be identified as dating to this phase in comparison to earlier phases where phase recognition is more dependent on fewer fossil markers, because of less detailed knowledge of domestic pottery of the 1st–3rd c. CE. As a consequence, I am less certain that we have anything like the full story of the sites associated with the lost domains of the early to mid-Roman era represented in the survey ceramics.²⁴ The “pots equals people” style of demographic modelling needs to be resisted.

The great imperial estates of northern Tunisia had a major focus on cereal production, but it has long been known from the inscriptions relating to the lease arrangements applied to the sharecroppers on these estates that the olive and vine were also cultivated on some scale on land less suited to cereal culture. An interesting point (33–34) is that the omission of mention of vines in the Aïn Wassel and Lella Drebbia inscriptions that relate to the saltus Neronianus may indicate that this part of the landscape was not suited for vines, though they are attested on other estates in the area.²⁵ The Rus Africum survey has dramatically increased our knowledge of the rural buildings in this landscape and has identified at least 370 presses in the Thugga area (an average of a press every 1 sq. km). One of the great achievements of the careful survey of sites is that so many of the ancient farms can be recognized to have contained presses, as evident in the superb catalogue of plans in volume 1 of the Rus Africum series.²⁶ That certainly suggests a significant level of production that exceeded local subsistence needs.²⁷ A missing element in the final publications of the project is a definitive account of the abundant traces of olive presses and comparisons with other regions of North Africa, where there is now a very significant level of data available, primarily from survey.²⁸ A really important aspect of the demonstration of the continuing vitality of the Aïn Wassel settlement into the Byzantine era is that olives remained an important element of the rural economy. But what was the scale of Late Antique oil production? On the one hand, the unexpectedly late chronology of the press building excavated certainly demonstrates that some of the press elements recorded by the survey must go with the Late Antique phases, but the fact that many press elements were recorded in reuse contexts in late building walls also suggests a long history of oil production here. This and the potential chronological biases of the ceramic dating evidence currently work against fuller interpretation of changing production over time. We desperately need an excavation of an early or mid-Imperial press building in the Thugga region (and preferably one linked directly to the imperial estate) to help understanding of production development and change.²⁹

A couple of final points to reflect on concern the extent to which the world of the imperial estates was a literate one and how this impacted on the wider rural landscape in the hinterland of Thugga. Dossey has argued that Christianity had an important part to play in the rise of rural literacy in Late Antique North Africa.³⁰ This point seems to find some support in the prevalence of Late Antique written documentation (tablets and ostraca) from rural locations in other regions of North Africa.³¹ The Thugga regional survey has recorded 450 rural inscriptions (not including milestones), of which 178 were new discoveries during the work (341).³² This is an unusually high level of epigraphy in the context of Roman rural communities, though, to judge by the published material from the survey overall and Aïn Wassel in this volume (341–45), most of it relates to the early to mid-Imperial period rather than to Late Antiquity.³³ The presence of multiple large imperial estates in the landscape around Thugga, requiring abundant boundary markers and erection of monumental stones recording the legal frameworks of their functioning, no doubt contributed to the unusual engagement with Latin text in the countryside. Another factor may have been the embedding from an early Imperial date of a pagus of Roman citizens alongside the Numidian population in the town of Thugga. The implications of this perhaps need further reflection³⁴ – again we seem to have a ghost image of the imperial domain intruding into the site's Byzantine story here. It does not disprove Dossey's idea of Late Antique “textual communities” becoming ever more important, but it does raise the possibility that this area was an early outlier that anticipated that trend.

The reflections in this review are in no way meant to undermine the quality of the work presented by the Trento/INP team, but rather, I hope, they have served to highlight how valuable the assembled data are for exploring issues of interpretation, even if, for want of a larger corpus of excavated rural sites, our search for the lost domain will continue to produce more questions than conclusive answers.