An alleged Far West? The Romanisation of the countryside in western Gaul

Cécilia Courbot-Dewerdt

Introduction Even after more than a decade of fierce debates on its significance and on the concepts it conveys, using the ‘R’ word always seems to necessitate some degree of explanation and justification. Even if it is a useful means of encompassing all the different processes of social, cultural, and economic evolution undertaken by people inhabiting territories that have been incorporated as a province of the Roman Empire, this ‘R’ word is so laden with various theories, that these long circumlocutions are usually preferred in academic works. Most of the time, discarding the word does not change our way of addressing the issue. This paper is precisely about how, sometimes, we still consider the archaeological data of the Late Iron Age/Early Gallo-Roman period from a narrow Romanisation point of view, despite our best efforts, whether using the ‘R’ word or forsaking it. Romanisation will be used in this paper in an historiographic sense, that is, the way archaeologists have looked, and still look, at the data at hand in relation to the Roman domination of the provinces. As it has already been said for Roman Britain, there was probably more to countryside life in Roman Gaul than just Romanisation (Hill 2001: 12), but it looks as if we still have difficulties determining what that ‘more’ is. When contrasting the archaeological dataset of two regional surveys, we thus unsurprisingly tend to analyse differences during the Gallo-Roman period in term of ‘resistance’ or ‘integration’ to the Roman Empire. Yet, regional surveys can contribute to a better understanding of the rural settlement system and its evolution, as they usually tend to consider the phenomenon from a long-term point of view. Too much focus on the Late Iron Age/Gallo-Roman transition has resulted in a bias that leans firmly towards a Romanisation perspective. However, trying to include the Iron Age data in the analysis of Gallo-Roman settlements can shed new light on old questions. This paper will contrast the evolution of rural settlement during the late La Tène and Early Empire in well-studied Picardy and in western Gaul to try to understand the discrepancies between the two territories without falling back on the usual Romanisation answers.

Romanisation from Picardy to Northern Gaul: the re-construction of a story The choice of the comparison areas, Picardy and western Gaul, may appear quite arbitrary at first but is partly determined by the historiography of Gallo-Roman archaeology. A PhD thesis recently completed on the evolution from late La Tène farmsteads to Roman villas in western Gaul has been the starting point of this enquiry (Courbot-Dewerdt 2004). During this research, it emerged quite early that the area chosen at first, a large north-western zone encompassing Brittany, Normandy, Pays de la Loire and Picardy was too heterogeneous for the survey purposes. In fact, the Picardy data stand out in many different ways, from issues relating to rescue archaeology organisation and methods, to the layout of the Iron Age and Gallo-Roman settlements. Thus, the research has been limited to a western zone that at first excluded Picardy (Fig. 1). The archaeological dataset collected is quite unusual since it is based mostly on rescue archaeology reports from one hundred and sixteen sites, dating from the late La Tène to the third century A.D. Detailed analysis of more than two hundred excavation reports has thus shed some new light on rural settlements in western Gaul, farmsteads and villas alike.

Figure 1: The area surveyed in western Gaul. Rescue archaeology operations are indicated by solid circles.

It is therefore not surprising that, since the beginning of this survey, several facts have appeared to be quite at odds with traditional views of the Romanisation of the countryside in northern France. Looking back consequently to the data that were used to build what can be called an ‘official story’ of the evolution from ‘native farm’ to Gallo-Roman villa, it appears that our understanding was derived mostly from sites and surveys in Picardy. There are some very good reasons for this state of affairs. First of all, Agache and his famous aerial surveys in Picardy in the 1970s and 1980s have been the necessary starting point of all the later research on native farms and Roman villas (Agache 1978). His extensive work is known not only in academic circles, but also to the general public, creating, for instance, a lasting picture of what a Gallo-Roman villa looks like. It is generally described as a very large settlement surrounded by a rectangular walled enclosure with numerous buildings built against the inner sides and the impressive main house at one end with its porticos overlooking the courtyard. This imagined typical layout of the Gallo-Roman villa has been reinforced by numerous scale models presented in museums and reconstructions in books (Fig. 2). This large courtyard villa of Picardy has thus become both the imagined norm and seen as the pinnacle of Romanisation in the Gaulish countryside.

However, since the first aerial reconnaissance of the courtyard villa by Agache (1973), numerous academic works have expanded our knowledge of Gallo-Roman villas in northern France. Since the 1980s, important archaeological surveys of the Oise and Aisne Valleys (associated with several academic research projects that have developed Agache’s work) have added a great deal to our impression of the typical villa layout. The excavation of large quarries in different Picardy valleys has also led to numerous rescue archaeology operations, particularly in the last ten years or so. This dynamic research explains why the first important seminar about the Romanisation of the Gaulish countryside was held in Picardy. A large part of the volume that was then published in the Revue Archéologique de Picardie (Bayard and Collart 1996) was dedicated to the presentation of the first results of various Picardy surveys. From Agache’s first publications (1978), series of papers and publications have thus drawn on the Picardy data to present the Romanisation of countryside settlements and the organisation of Gallo-Roman villas in northern France. Important handbooks about the Gallo-Roman countryside, like the two volumes of Les Campagnes en Gaule romaine (Ferdière 1988) published in 1988, even now very much in use for first-year students, still rely heavily on the Picardy surveys to explore themes like the evolution from native farm to Gallo-Roman villa or the layout of the latter. Consequently the history of the Romanisation of the countryside in northern France still appears to be dominated by the results of the first Picardy surveys. In other parts of Gaul, the first surveys failed to find such impressive courtyard villas. This is particularly the case for western Gaul where few villas were identified in the 1980s, compared to Picardy. At the beginning of a paper about Roman villas of Brittany, it was said that despite some people trying to present Roman Brittany as a human desert, there was a quite dense network of middle-range settlements discovered by aerial surveys (Galliou 1982: 95). But ten years later, the same author entitled his paper about the Romanisation in western provinces, L’Armorique Romaine: Mutations et Résistances, specifying in the introduction that new works show that Brittany was not a complete backwater during the Roman Empire but that it participated in the phenomenon of Romanisation, even if only ‘sur un mode mineur’ (Galliou 1992: 29). This view no doubt stems from the fact that the first villas to be excavated in Brittany were less impressive when compared with those in Picardy.

A large excavation at the end of the 1980s provided a better understanding of the layout of an important villa near Rennes (Provost 1990). As late as the mid-second century A.D. the dwelling walls were still built in timber and wattle and daub, despite the use of stone foundations and tiled roofs. During the first century A.D., the main dwelling was a quite simple affair with three or four rooms lined with a portico. It developed into an important three-aisled building during the third century A.D. Only a few other buildings have been discovered inside the ditched, or possibly fenced, enclosure in which it stood. The settlement layout lacks the Roman symmetry found in Picardy courtyard villas, even in its latest developments at the beginning of the Late Empire. Most of the Gallo-Roman villa excavations conducted in western Gaul since then have revealed middle-range settlements.

Figure 2. A model of a villa reconstructed from the archaeological excavations at Ferin, North France (Archaeological museum of Douai).

This notion of discrepancy between western villas layout and size and Picardy courtyard villas has been quite reinforced by the results of several aerial surveys in these regions. Regional surveys in western France have revealed numerous Iron Age enclosures but only a few villas. As a result academic works have therefore focused on Iron Age settlements (see for example Leroux et al. 1999). The discovery of enclosures dating back to the early Gallo- Roman period lead to the study and excavation of several Gallo-Roman farmsteads, very similar in their layout and building techniques to Late Iron Age ones. These Gallo-Roman enclosed farmsteads, coupled with the small number of villas discovered, tend to incite explanations that focus on Romanisation and resistance to integration in the Empire or underdevelopment. Contrasting the archaeological data of western Gaul with that of Picardy during the 1980s, some authors argued that Gallo-Roman farmsteads and middle-sized villas were linked to delayed or superficial Romanisation. They tended to explain what they saw as a lesser degree of Romanisation due to the so-called remoteness of western Gaul, and specifically Brittany, from major Roman roads and trade traffic (see for example Langouët 1991). Other territories have seen the development of such an history of archaeological theories, like the Fenland in East Anglia, where poverty has been used to explain what was seen as a ‘failure to Romanize’ (Fincham 1999: 47). The deconstruction of this notion has led a recent work to propose a new, post-colonial model of evolution for the Fenland (Fincham 1999: 47–51). This idea of western Gaul as a ‘far west’ underdeveloped and under-Romanised during the Early Roman Empire has also been seriously contested recently in France but there is still a lingering thought that western Gaul may not have achieved a high level of integration into the Roman Empire.

The contribution of regional surveys: from late La Tène settlement layouts to villa designs in western Gaul and Picardy The difficulty in archaeology is that ‘[l]’archéologue est d’abord l’inventeur de l’objet sur lequel il va ensuite travailler’ (Burnouf 1996), i.e. the archaeologist creates the very data he will be working on; therefore, the way data are acquired can deeply influence the story archaeologists are trying to reconstruct and early misconceptions can have enduring effects. The respective histories of Iron Age and Gallo-Roman archaeology in Picardy and western Gaul have tended to accentuate apparent discrepancies. A closer look at the primary archaeological data may thus lead to a better understanding of the phenomena observed. It is interesting to note, for example, that the Picardy countryside is not solely made up of extensive courtyard villas. A well-documented Picardy villa like Dury (Somme) is quite similar in size to middle-range settlements in western Gaul (Quérel and Feugère 2000). Courtyard villas are so impressive they tend to attract academic interest, but it has to be kept in mind that most of them only achieved this development quite late during the Early Empire. Interestingly, looking back in time to late La Tène farmsteads layouts in western Gaul and in Picardy can also shed some new light on Gallo-Roman villa designs.

During the Late Iron Age in western Gaul, rural settlements were scattered in the countryside, each farmstead standing alone, surrounded by pastures and a little farther away cornfields. The features that survive best in the archaeological record are the ditched enclosures that surrounded the farmstead buildings. Three different layouts have been identified from the dataset of western Gaul enclosures: a single isolated enclosure, two joint enclosures and an enclosure that was part of a small field system. Enclosures themselves are mostly rectangular with a more symmetrical and regular shape toward the Augustan period. Some enclosures were even trapezoidal with the main entrance situated on the broader side. This specific layout would have thus created an enhanced visual impression for anyone entering the enclosure. The farmstead was clearly divided in two areas: a housing space, delimitated by a specific entrance or a simple fence, and a working area. There was perhaps a kind of progress from outer space to domestic and then private areas. The organisation of the entrances suggests a main passageway leading to a working area with granaries, kilns and the like (Courbot-Dewerdt forthcoming). From this space, access was possible to the house. It is important to note that only a few buildings have been discovered in the enclosures, two to five being the usual case and most of them what are thought to be four-post granaries. More interestingly, in western Gaul, extensive excavations have revealed only single houses, or very rarely two houses that might have been occupied at the same time. The latter case only occurred on important farmsteads, with ten or more buildings being identified inside their enclosures. The analysis of archaeological remains retrieved on these sites points to high-status sites: many kinds of artefacts like manufactured goods, brooches, coins and even, sometimes, weapon elements (Fig. 3). Only a handful of the rural settlements excavated in western Gaul were of this twohouse kind. Interestingly, in these cases, both houses were situated in the same enclosure and seemed to have no distinctive features that would suggest any difference in their status. Furthermore, they also share the same work area. Late Iron Age farmsteads in western Gaul seem to have been occupied and worked in most instances by a single family living in a thirtymeter- square timber-built house.

Figure 3: Two-house settlements in Western Gaul.

Late La Tène rural settlements in Picardy appear to have had a quite different layout. Enclosures and fences were still the main structures that organized the farmsteads, but a higher number of buildings can generally be observed. A recent academic work about the Iron Age settlements of the Oise Valley thus points out that, in most cases, three to four of these buildings are houses (Malrain et al. 2002: 137–158). In his PhD, Malrain (2000) has proposed a four-stage site hierarchy, from small open settlements to farmsteads with two joint enclosures. All three enclosed stages present several houses used simultaneously. More importantly, one of the houses is clearly singled out, not by its architecture, but because it was established in a central place in the courtyard and even, for the higher status settlements, separated from other domestic units by a sub-division within the larger main enclosure. Bearing in mind these differences in the social and architectural organisation of rural settlements in Picardy and western Gaul, it is interesting to look at what occurred to them during early Gallo-Roman times, that is, during their so-called ‘Romanisation’. The development of rural settlements in western Gaul during the Gallo-Roman period can be described as a non-stop process with peaks of important changes. The first occurred during the second half of the first century A.D., when the first farmsteads that can be named ‘villas’ were built. The layout of these elementary villas is quite similar to the Late Iron Age farmsteads: a rectangular enclosure, one house built against the fence, and a few buildings (Fig. 4). Only the house, and sometimes a granary, was built with stone foundations and the walls were still made of wattle and daub. The residences already had something of a Roman air with tiled roofs, porticos and whitewashed walls. In the meantime, traditional farmsteads ceased to be built and are progressively abandoned, some to be replaced by a villa built some distance away, others to be farmed as fields. The second major changes date from the middle of the second century A.D. Some of these elementary villas were abandoned, while others began a new stage of expansion. In the latter case, the residence was rebuilt on a larger scale, with several more rooms being added to the original layout. Most of the time, the house walls were rebuilt in stone at this point. New economic activities also took place in the villa courtyards, like largescale tile production. More importantly, new domestic units were built, probably intended for a steward. All of these data thus point to an important change in the scale of economic activities in villas with the resulting need for a larger staff to run the farm. It is important to note that, from the end of the second century A.D., villas in western France were very similar in scale to courtyard villas in Picardy, even if their layouts still differed slightly.

Figure 4: Elementary villas in western France.

With courtyard villas initially being considered as an obviously Roman phenomenon, it took some time before academic research became interested in the early phases of their development. But in investigating the first stages of courtyard villa development, the layout and social organisation appeared to be very similar to those of the preceding settlement (Haselgrove 1995). In the Late Iron Age and early Gallo-Roman farmsteads, one house was singled out at one end of the enclosure, facing a large courtyard where domestic units and farm buildings were ranged on opposite sides. The ‘Les Grèves’ site at Beaulieu in the Aisne Valley was thus organised from the first century B.C. to the end of the first century A.D. (Haselgrove 1995), despite several phases of rebuilding. The courtyard villa of Verneuil-en-Halatte in the Oise Valley was organised in a similar fashion in its first stages (Haselgrove 1995). Toward the mid-first century A.D., the main residential building was rebuilt in a more Roman architectural style and the fence was replaced by a stone wall. Later, the other domestic buildings were progressively rebuilt on stone foundations. This social organisation with one prominent house dominating a courtyard where discrete domestic units were symmetrically ranged closely resembles the layout of Late Iron Age farmsteads in Picardy.

With these two regional studies, it clearly appears that, until at least the end of the first century A.D., changes occurred mostly in the household architecture and building techniques. Discrepancies between western Gaul and the first villas in Picardy appear thus to have nothing to do with a phenomenon of ‘Romanisation’ as such, but are linked to previous differences in social organisation: one household running a farmstead in western France versus a higher status household working with several dependant families in Picardy. In the same historical and political context, rural settlements evolved in quite parallel ways but from distinct starting points. When the research focus is not only on issues to do with Romanisation, alternative explanations can be found for phenomena observed in specific areas. Social organisations dating back to the Late Iron Age continued to be dominant in the early Roman period and thus still affected the layout of settlements during the Early Empire, even as other social and economic changes took place. In turn, these explanations can shed new light on rural settlement evolution in relation to Roman rule. Despite the way early villas apparently retained the layout of late La Tène farmsteads and, thus, some aspects of their social organisation, important changes occurred in the rural settlement system during the early Roman period. Hence, even if a Gallo-Roman farmstead were discovered under each and every villa excavated, many Late Iron Age and early Gallo-Roman farmsteads would have disappeared, having been replaced by a field system. The relatively high site density observed in western Gaul for the late La Tène period decreased progressively until, at the end of the second century A.D., a developed villa had replaced less than half of the farmsteads (Fig. 5). This pattern of events suggests that a major concentration of land ownership occurred during the early Gallo-Roman period with a resulting change in settlement type and number. The figures for Picardy appear to be quite similar, despite the fact that a process of settlement aggregation was already underway during the first century A.D. This evolution can therefore be traced back to what has been called Romanisation but that is mostly driven by the political and economic integration of Gaul as a province of the Roman Empire. Thus, comparing detailed regional surveys appears to be one way to better understand processes of evolution because it allows archaeologists to separate large-scale phenomena that may be linked with the Roman Empire from regional Late Iron Age specificities.

Direction de l’archéologie du Douaisis, Douai (France) 0 5 10 15 20 25 30 35 40 45 50 125 B.C. to 75 B.C. 75 B.C. to 25B.C. 25 B.C. to A.D. 25 A.D. 25 to A.D. 75 A.D. 75 to A.D.125 A.D. 125 to A.D. 175 A.D. 175 to A.D. 225] Number of settlements Number of foundations Number of desertions

Figure 5: The evolution of countryside settlements from late La Tène to Early Empire in western Gaul.

Acknowledgements I would like to thank the TRAC 2005 team for offering me a bursary that allowed me to present this paper and J.D Hill for his advice and support.