Les Gaulois sont-ils dans la plaine ?
Reflections on settlement patterns in Gaul in the 1st century BC

In the thirty years since Olivier published his ground-breaking study Structures d’habitat et fortifications de l’âge du Fer en France septentrionale¹, our knowledge of Iron Age settlement has changed out of all recognition, especially for rural sites. Fuelled by the growth of rescue archaeology, the evidence from France is among the most plentiful and best studied in Europe. Olivier himself has been at the forefront of this development in understanding, through his own excavations at Levroux, Mont Beuvray and Bourges; through his many publications and syntheses²; and by stimulating a new generation of researchers to follow in his footsteps with agenda-setting studies of their own.

Crucially, Olivier was also among the first to promote the need for standardised terminology to permit effective comparison between different regions and the power of computer databases for harnessing and analysing the rising mountain of evidence generated by rescue archaeology, of which we are just starting to reap the benefits³. Always too, Olivier has taken a broad view of Iron Age settlement and economy, setting the French evidence in a wider European context, thus allowing us to see which phenomena are confined to particular areas and which are widespread⁴. We are therefore pleased to be able offer this short study of one aspect of Iron Age settlement evolution, which seems to recur in several places.

Developments in the 1st century BC

Across central and northern France and in many other areas of Europe, the period from the 3rd century BC (La Tène C1) onward is marked by a steady rise in rural settlement numbers (and by implication in total population). This increase carried on almost everywhere until the 2nd or 3rd centuries AD, after which numbers fell again during the late Empire. This demographic explosion was closely linked to the opening up to intensive cultivation of heavier but more fertile soils in areas of the landscape such as plateau tops that had hitherto been marginal for settlement. The ready availability by c. 300 BC of mature iron technology for ploughshares, axes and other tools and of crops suited to cultivation on heavier soils was evidently a key driver in this process of agricultural intensification and settlement expansion; the warmer climate⁵ must also have contributed. Larger settlements became widespread, first unfortified agglomerations like Levroux, then the fortified oppida.

During the later Iron Age, enclosed farmsteads, linear ditches and field systems all became widespread in many different areas of Europe. Their form and layout indicate that this rise in enclosure was strongly influenced by agricultural concerns: to separate arable from pasture, and for stock management; to drain land otherwise too wet for use; and to mark property boundaries in a landscape that was fast filling up. The cooperation and labour involved in digging and maintaining these ditch systems may also have served to integrate groups more closely. Pollen analysis provides another indicator of the extent to which land use was now more specialised and communities were exploiting a more diverse range of landscapes such as wetlands⁶. On the alluvial terraces of major rivers, land closest to the valley bottom farms was often mostly pasture, whilst the arable fields were situated further away on the lower slopes and plateau tops. Surviving woodland was presumably heavily managed for fuel and building materials. The nature of crop husbandry also changed significantly⁷, as did grain storage methods, with a reduction in storage pits and storage facilities at producer sites⁸. In the Auvergne, the wet plains of the Limagnes were subjected to systematic drainage⁹.

As the number of excavations and the precision of our chronologies have increased¹⁰, it has become clear that the growth in Iron Age settlements was not unchecked from La Tène finale to the high Empire. Across much of France, from the Auvergne¹¹ to lower Normandy¹² and from the Pays de la Loire¹³ to Picardy¹⁴, an apparent drop in the number of occupied rural sites occurs in the 1st century BC, especially in La Tène D2, after which numbers rise again in the Gallo-Roman period. This reduction is also apparent from proxy indicators, with the incidence of coins and other finds at rural settlements also falling in the 1st century BC, whilst certain types of artefact and activities are virtually confined to oppida and other larger settlements¹⁵. The open nucleated settlements which developed during the previous century are also strongly affected, losing many of the activities carried out there and in some cases possibly disappearing altogether; obvious examples include Acy-Romance, Aulnat / La Grande Borne, Basel, Feurs, Levroux, and Roanne. A range of explanations has been proposed, according to whether the decrease in rural settlements is seen as genuine or due to problems with the evidence. Possible solutions in the latter category include a shift from enclosed settlements to less visible open sites and/or the difficulty in dating rural sites and lack of diagnostic material in La Tène D2 in particular. Explanations in the first category include depopulation due to Julius Caesar’s brutal campaigns, and – perhaps most popular – nucleation of a sizeable part of the rural population into the new oppida or other agglomerations¹⁶.

In the most comprehensive analysis of its kind so far –and one which shows the value of inter-regional databases– Nouvel et al.¹⁷ have compared the number of occupied rural sites by period from La Tène B1 (c. 400 BC) to Gallo-romain précoce across the entire East of France, from Alsace to the eastern border of Île-de-France. As the authors are the first to admit, the data are patently uneven. Very limited evidence is available from certain regions (notably Champagne-Ardennes) and environments (e.g. plateaux), whilst the major river valleys are heavily over-represented thanks to the focus of mineral extraction (and thus rescue archaeology) on the alluvial terraces. Disparities between regional chronologies and in the methods of data collection also contribute to the over-representation of some phases at the expense of others. Even so, their results are of great interest.

First of all, their data not only underline the ubiquity of this reduction in rural site numbers in the 1st century BC following the increase in La Tène C2–D1, but also indicate that this phenomenon of a peak in settlement followed by a reversal took place earlier (by at least a generation) in the East (Alsace), than further west (Lorraine, La Bassée). This would seem to eliminate the Gallic War as a primary cause, but this sequence does match another well-known pattern, whereby late La Tène oppida seem by and large to rise –and fall– earlier in central Europe than in the West¹⁸. Second, La Tène D2a appears everywhere as a particularly unstable period, marked by a multiplication of both settlement creation and abandonment. Third, the density of occupation in the valley bottoms, intensely exploited since the mid 1st millennium BC, goes into permanent decline after the 2nd century BC, whereas on the slopes and plateaux the reduction in occupied sites is much less marked; some areas even experience a modest increase, although here the influence of different methods of data collection poses a problem. The bulk of the data from these other environments comes from field survey¹⁹, which tends to emphasise continuity, whereas excavation often highlights apparent gaps and breaks in sequences, which may or may not be real.

This paper will seek to expand on the conclusions reached by Nouvel and his colleagues in two ways. First, we will attempt to put this temporary “reversal” in settlement evolution in France in context by examining 1st century BC developments in three adjacent regions: southern Germany; the southern Low Countries; and Britain. We will then turn to Picardy, an area not examined by Nouvel et al. and finally try to nuance some of their conclusions by revisiting two areas of Burgundy.

Southern Germany

Whether the apparent dearth of rural settlement of 1st century BC date in southern Germany is real and, if so, how to explain this was a matter of debate well before the earlier dating of La Tène D1–D2²⁰ opened up a clear gap between the end of the oppida there and the Roman annexation under Augustus. It is now generally agreed that the oppidum of Manching in Bavaria went into sharp decline after 80 BC²¹, being finally abandoned in the mid 1st century BC. A similar pattern applies across the rest of southern and central Germany, with the Dünsberg in the Lahn valley –where occupation continued to the end of the 1st century BC– the only major oppidum east of the Rhine known to have outlasted the La Tène D1b/D2a transition.

The recent suggestion that the well-known Viereckschanzen of southern Germany were primarily enclosed farmsteads, and not cult sites after all, does little to fill this gap. A new study of eighteen examples implies that their main period of occupation was in La Tène C2–D1, and they were all abandoned by the mid 1st century BC, mostly within La Tène D1²². Indeed, whilst some differences in form, size and architecture are evident between the Viereckschanzen and their French counterparts, in general terms, the German sites adhere to the same general pattern of an increase in enclosed settlements in the later Iron Age, followed by a decline. Moreover, but perhaps not unsurprisingly, the chronological rhythm most closely mirrors neighbouring Alsace Lorraine, with numbers peaking in La Tène D1a.

Explanations of the perceived void in southern Germany range from a shift of the remaining population after a period of economic decline into small, largely invisible open farmsteads, to progressive emigration of the existing La Tène inhabitants and their replacement by groups of Germanic migrants from the north. Rieckhoff²³ has suggested that this process might have been set off by an epidemic and/or an environmental crisis. On the other hand, this notion of an “empty landscape” is increasingly open to challenge²⁴. Pollen cores show no sign of a break in agricultural activity in the period between the end of the oppida and the arrival of the Romans, and gradually a number of sites that date to this period are coming to light. Some like Harting²⁵ would not have been assigned to this period if they had not had some diagnostic brooches, but others like Leonberg and Stöffling have yielded substantial coin and metalwork assemblages, whilst Seebruck has a series of tree-ring dates suggesting continuous occupation from the late Iron Age into the Roman period²⁶.

In short, it seems likely that the break in the German sequence has been somewhat exaggerated. More likely, what we are seeing is a decline in site numbers in La Tène D2 following a highpoint in La Tène C2/D1, just like the reversal that is apparent in France. This in turn suggests that migrations has been overemphasised as a cause of the changes in southern Germany in the 1st century BC, since a similar reversal in the settlement pattern is apparent at this period as far away as the Auvergne or Vendée, surely beyond the reach even of the secondary disturbances provoked by population displacement east of the Rhine. Nor is there much sign of cultural replacement in Gaul during the relevant period, despite the widespread evidence of settlement instability.

Roman intervention is even less likely to be a relevant factor in southern Germany than in Gaul. The area was not overrun until the reign of Augustus, yet the changes –which affected the oppida as well as rural sites– were underway before the end of La Tène D1. Equally, whilst the proposed role of the Romans as a catalyst in setting off a wave of migrations seems questionable, disease and/or environmental crises should not be overlooked as possible explanations for a decline in population and/or settlement numbers over large areas.

Belgium, southern Netherlands and the Rhineland

Our second case study comes from the zone extending from the Flanders coast through the south Netherlands to the Rhine. Until recently, large-scale excavations were a comparative rarity over most of this area and the chronology in use is still fairly coarse. Consequently, the sub-divisions used in France within La Tène finale are difficult to apply here, obscuring many of the finer details of settlement evolution. Nevertheless, drawing on data about sites excavated since 1998 collected as a part of a Leverhulme Trust-funded research project by the Universities of Leicester and Reading, some observations about rural settlement in this zone can be offered.

The first point is that in contrast both to France and to other neighbouring areas like Westphalia, there is a chronological bias to the earlier or middle Iron Age across this entire zone, with relatively few sites dating to the later Iron Age. In the Rhineland, only 22 of 110 Iron Age sites investigated since 1998 clearly post-date c. 250 BC. The same pattern is found in the cover sand area in the south of the Netherlands, where two-aisled longhouses were the predominant house form during the late Iron Age; elsewhere in the Netherlands, where three-aisled longhouses predominated, site densities conform to a more familiar pattern, with numbers gradually increasing to a peak in the later Iron Age. Belgium follows the same pattern as the Rhineland and the south of the Netherlands, with site numbers falling after c. 250 BC having peaked between c. 500-250 BC, although it should be noted that due to the different planning regimes in Flanders and Wallonia, the majority of excavated Iron Age rural settlements are in the sandy areas of the former and very few have been excavated in loam areas, which are mainly in the latter region.

There is nothing however to suggest that this reduction in occupied sites after c. 250 BC indicates a demographic downturn. Rather it would seem that an earlier-to-middle Iron Age landscape of dispersed settlements comprising three to five single-phase farmsteads whose locations shifted periodically, gave way in the late Iron Age to a more stable arrangement²⁷. Farms now remained in one place for two or three building phases and clustered together in small nucleated and even sometimes enclosed settlements. In the Netherlands at least, most of these sites continued to be inhabited throughout the later 1st century BC and into the Roman period. At a broader landscape scale, a second process of nucleation occurred in the later Iron Age, especially in the sandy areas. In contrast to the earlier Iron Age, when a great variety of landscape zones were exploited, later Iron Age occupation is largely confined to soils with relatively high loam contents (ibid.). This focus on a particular soil type contrasts with the more usual pattern for the later Iron Age not only in France but also in Britain, where the range of landscape types exploited is generally more diverse than before.

In the sandy areas of Belgium, later Iron Age settlements also appear to be occupied for longer than previously, but on the loam, the admittedly limited evidence from here points more to the regular creation and use of settlement enclosures on the model of northern France. In the Rhineland – despite the decline in later Iron Age settlement numbers – the older suggestion of a complete break in the mid 1st century BC, analogous to southern Germany, can be similarly discounted, given the mounting evidence for sites like Vilich-Müldorf, Elsdorf-Etzweiler and Jüchen-Neuholz with a continuous sequence across the late Iron Age to Roman transition.

In sum, this zone on the fringes of the north European longhouse tradition seems to follow a different trajectory, particularly in the sandy areas. The fall in site numbers begins much earlier, at a period when settlement was expanding strongly in northern France and elsewhere, but ushers in a greater degree of stability in recourse to particular locations. Due to a lack of fine chronological precision, it is not yet clear whether there were further changes in the 1st century BC, but there is nothing to suggest a sharp break in the settlement record at this time.

Southern Britain

As in the Low Countries, later Iron Age chronology in Britain is not susceptible to the same level of sub-division as in the La Tène zone, even in the ceramic-rich areas in the south and east. Where we have to rely on radiocarbon, chronologies are coarser still, making the quantification of site numbers occupied at successive periods all but impossible. Even so, it is clear that in many areas of Britain, the 1st century BC was a period of particular instability in the settlement record compared to the preceding centuries, although a fall in site numbers seems to be the exception to the rule.

The changes are especially marked in a zone extending from Wessex to East Yorkshire, which encompasses many of the most densely populated and socially developed areas of middle Iron Age Britain²⁸. In the 1st century BC, a major restructuring of the settlement pattern is apparent almost everywhere throughout this zone. In Wessex, the large developed hillforts like Danebury were abandoned and their population instead dispersed into smaller settlements, several of which had been occupied at an earlier era. The vast majority of these late Iron Age sites were also inhabited in the Roman period²⁹, although whether there was direct continuity is not always clear.

In the areas surrounding the Wessex chalkland, which include significant expanses of hitherto marginal clays, heaths and wetlands, new sites and settlement types appear in large numbers in the late Iron Age³⁰, and again there seems to be a strong degree of continuity into the Roman period. Further north, in the English Midlands, long-established agglomerated settlements like Crick, Fengate and Humberstone declined in size through the 1st century BC and their inhabitants presumably moved to other sites, whilst on the East Yorkshire chalk, settlement territories defined by linear earthworks gave way to enclosed “ladder” complexes of fields integrated with settlements³¹.

Many parts of South-East England see a significant increase in settlement numbers in the late Iron Age. Often this expansion is not fully apparent until the end of the 1^st century BC, but this is probably mostly a function of the changes in material culture in the Augustan period, which makes such sites both more visible and more readily dateable. Not all parts of the South-East follow this trajectory, however. Sealey³² has drawn attention to a marked decline in the number of roundhouses on sites in Essex from the end of the middle Iron Age and many existing settlements such as Stansted Airport were abandoned altogether in the 1st century BC. He suggests that this decline reflects a real reduction in population, but if so the cause is unclear; there is no evidence of adverse agricultural or environmental changes at this date, but other factors, such as epidemics, would not be as readily visible.

Over much of western and northern England and southern Scotland, the trajectory of rural settlement from the later 3rd century BC closely resembles France, with a proliferation of rectilinear enclosures, often on virgin sites and clearly associated with a process of infilling hitherto thinly occupied areas of landscape³³. Whilst these enclosures are archaeologically more visible than the amorphous open settlements that preceded them, the rapidity and scale of woodland clearance attested by pollen cores, particularly from c. 200 BC, shows beyond doubt that this represents a genuine process of settlement growth and agricultural intensification. Many of these farms were occupied for generations, with little change apart from some rebuilding, until the 1st century BC, when their character altered significantly. Across North-East England and southern Scotland, for example, the ditches around settlements were now allowed to fill up, but the farms themselves continued to be occupied and took on looser forms of organisation, often expanding over the old boundary ditches. Bayesian modelling of the radiocarbon evidence suggests that this shift to open settlement took place over a short period around c. 50 BC³⁴. Unlike Essex, these changes may be linked to population growth, since the average number of roundhouses on sites increases at this time and several extensive open settlements can also trace their origins to this horizon.

Thus, whilst no single pattern dominates, the settlement record in Britain displays some interesting parallels with France, with the 1st century BC emerging as a period of rapid change in some areas and dislocation nearly everywhere. This included the demise of many long-established larger settlements, whose inhabitants relocated to other sites, where they then remained in the Roman period. In contrast to France, the late 1st century BC and early 1st century AD seems to have been a period of particularly rapid growth in many areas. A wave of emigration precipitated by the Gallic war might be a contributory factor, particularly in some coastal areas, where refugees may already have had ties, but this can hardly explain developments in areas as far away as northern England or southern Scotland.

Northern and Eastern France

Having examined these other regions, we may briefly return to France, starting with Picardy. The Aisne valley area is one of very few where oppida as well as rural sites been extensively examined and evidence from surface prospection is also available to offset the bias to the alluvial terraces³⁵. As Figure 1 shows, there is a clear rupture in the rural settlement record in the 1st century BC, which closely coincides with the occupation of the successive – and (mostly) short-lived – oppida, strongly suggesting that much of the rural population abandoned their farms for a generation or two to live instead in these fortified centres. And when farms reappear on the terraces in La Tène D2b or Gallo-romain précoce, they were often much larger than before, consisting of multiple households.

A growing number of excavations in the side valleys and on the plateaux are starting to bear out the earlier surface prospection in suggesting that the picture is different away from the valley bottoms. In particular, recent work on the Soissonais plateau has identified a rural settlement at Ploisy, which seems to have been continuously inhabited from La Tène D2a (fig. 1, Ploisy 1b), thus overlapping with both Villeneuve-Saint-Germain and Pommiers. This suggests that some of the many plateau sites identified by surface prospection as occupied in La Tène D2/Gallo-romain précoce might have originated at the same period³⁶. On the other hand, prospection provides little indication of direct continuity on the plateau from La Tène C2/D1 (as indeed with the break between Ploisy 1a and Ploisy 1b).

This suggests that the type of settlement pattern associated with developed hillforts on the Wessex chalk in southern Britain³⁷ may also apply in northern France. Most of the people living closest to the oppida moved into these new fortified sites, whilst further afield, farms and other sites continued to be inhabited, but under the political and economic domination of the larger centres. We would emphasise, however, that oppida were more than just collections of farms, since they invariably yield evidence of numerous activities of a character and intensity not seen at other contemporary sites³⁸.

When the results of excavations in and near the Oise valley³⁹ are added to the Aisne valley data (fig. 2a), it appears that across the entire region, settlement in the secondary valleys and on the plateaux underwent a reversal of similar magnitude to that of the valley bottoms between La Tène D1 to La Tène D2. Both zones recover rapidly in the Augustan period, although when the results of surface prospection are included, the plateaux emerge as the area of strongest growth.

Turning briefly to the Auvergne and more precisely the Clermont-Ferrand region, a very similar break is obvious in the settlement pattern in the 1st century BC. As in the Aisne valley, it is correlated with a succession of three short-lived oppida close to each other⁴⁰. An increase in population density during the 3rd and 2nd centuries BC in the plains of the Limagnes is apparent from both excavation and survey evidence, with a few major nucleated settlements –such as Aulnat / La Grande Borne⁴¹ and also Aigueperse, some 30 km to the north– and a high density of rural settlements, mostly short-lived⁴². Some show a longer occupation, nucleating to several domestic units by the mid-2nd century, like Le Pâtural⁴³. Use of the uplands is still impossible to quantify but their exploitation is shown by enclosed settlements revealed by rescue excavations along the A89 (e.g. Le Bru at Saint-Ours-les-Roches).

This network shows a major break around 100 BC: not a single one of the numerous pre-existing sites shows clear evidence of continuity during the 1st century BC. The correlation with the foundation of a central oppidum is obvious but the clustering of a few thousand people on a hilltop of c. 70 ha is by no means an adequate explanation for a change of this significance and a major demographic downturn cannot be ruled out. Nevertheless, continuing agricultural use of plain in the 1st century BC is shown indirectly by the persistence of the field system into the Roman period (as shown locally⁴⁴). One supplementary explanation for the break would be that the reorganization of rural settlements around 100 BC was followed by a great stability of the newly founded sites until the end of the Roman period. This would explain the frequency of late La Tène period artefacts (especially easily identifiable sherds of Dressel 1 amphorae) collected during fieldwalking on Gallo-Roman sites. Excavation of Roman sites has unfortunately so far provided little decisive evidence for or against this hypothesis, except for a few positive cases (Les Chazoux at Gannat, Chaniat at Malintrat).

Turning to the Seine-Yonne confluence (La Bassée) and the lower Yonne valley on the borders between Ile-de-France and lower Burgundy, a somewhat different picture emerges (fig. 2b)⁴⁵. The number of valley bottom sites in La Tène D2 is lower than in La Tène D1, but here the principal fall occurs in the Augustan period, a time when numbers were rising again in Picardy. In La Bassée at least, site numbers then climb again to a new peak during the 1st century AD⁴⁶, after which another decline ensues. This suggests that different processes were at work in these two sectors, unless of course there is a problem in recognising Augustan sites in the Seine-Yonne area. This is conceivable, since La Bassée is further from the major centres of Augustan pottery production in central France than the Picardy sites are from the Gallo-Belgic industries of the Vesle valley. Another reason would be if many of the Augustan sites in the region were small, short-lived open settlements like the La Tène D2b site at Étigny on the Yonne⁴⁷, since these are not easy to find. The shorter length of the Augustan period may also be a distorting factor.

Although more sites were occupied in La Tène D2 in the Seine-Yonne sector than in Picardy, the numbers in figure 2b conceal a similarly dynamic picture on the ground. None of the La Bassée sites and only three on the Yonne terraces show continuity through from La Tène D1 to Gallo-romain précoce. Once again, it seems likely that aggregation of the population into larger settlements may be a factor in this high level of discontinuity. The nucleated settlement at Varennes-sur-Seine (and its elite satellite at La Justice), the oppida at Villeneuve-sur-Yonne and Sens were all occupied for at least part of La Tène D2. In the secondary valleys and on the plateau of lower Burgundy, however, the still-tiny sample of excavated rural sites suggests a quite different pattern, with occupied site numbers actually rising from La Tène D1 to La Tène D2 and then remaining constant into the Augustan period. This accords well with the larger body of evidence from surface prospection⁴⁸, which suggests less instability and more continuity away from the main river valleys.

The number of excavated sites in eastern Burgundy and Franche-Comté is too small to rely heavily on, but there is even less evidence of a break here (fig. 2c). In the valleys of the Saône and the Doubs, there are admittedly hints of a slight decline during the 1st century BC, but on the plains beyond, the reverse is true. Overall, this sector shows greater stability, with a number of agglomerations like Mâlain, Mirebeau and Verdun-sur-le-Doubs, with origins in La Tène C2/D1, apparently developing continuously through the 1st century BC.

This brief review of Picardy and two of the zones examined by Nouvel et al.⁴⁹ broadly repeats the patterns documented in their more detailed study. Although the settlement trajectories all show some signs of a reversal during the 1st century BC, there were clearly differences between regions in intensity and in timing, and between topographic zones within a single region. The prominent peak in La Tène D1 and subsequent fall in site numbers is primarily a feature of the major river valleys of the Paris Basin, from which Picardy emerged much faster. In the Saône-Doubs and on the plateau of lower Burgundy, any decline in La Tène D2 was negligible.

Conclusion

This short survey has touched on later Iron Age settlement trajectories over a sizeable part of Western Europe, from Bavaria to southern Scotland. Even a superficial examination like this leads to some interesting conclusions. The 1st century BC emerges nearly everywhere as a period of significant changes in the settlement record, which in many areas coincided with a net decline in the number of rural sites, putting an end to the period of sustained increase that started in the 3rd century BC. The one major exception is the area between the Flanders coast and the Rhine, where a reduction in occupied sites began much earlier, but was allied to the development of an increased attachment to place. This region generally has closer affinities with the zone dominated by the north European three-aisled longhouse tradition and evidently shared in the social and economic patterns that characterised agricultural communities there.

In our view, there is little doubt that the observed settlement “hiatus” during the 1st century BC is genuine, rather than somehow being a function of the archaeological evidence. Nor is it restricted to France, with a sizeable part of Germany showing a similar deficiency in sites and possibly also parts of southern Britain. The downturn is most striking in the densely populated major river valleys, but clearly did not happen everywhere at the same time or to the same degree, even in regions that were not very far apart, as the comparison of the Aisne-Oise, Seine-Yonne and Saône-Doubs sectors shows. The extent to which areas of the landscape outside the major river valleys were affected is more difficult to assess, owing to the limited number of excavations, but so far it seems that any interruption was less pronounced in the secondary valleys and on the plateau tops.

The nature of the underlying causes remains difficult to pin down, but it is likely that a multiplicity of factors were involved. The Gallic war may well have had some impact on its later stages, but the initial manifestations of the decline are too early for Roman expansion to have been a significant factor. Movement of rural population into oppida and other agglomerations many well be relevant in some areas and in these cases there is no reason to suppose any fall in population. But aggregation offers no more than a partial explanation, since in Germany the decline in rural settlements and in oppida went hand in hand. At the same time, the geographical extent of the region ultimately affected implies some common links. If nothing else, the high level of connectivity between the different peoples of late La Tène Europe might have played a part, ensuring that initially local crises –due for example to repeated crop failure or an epidemic, or enforced migration, or conflict for control of resources or territory– eventually reverberated over a far wider zone. The fact that major river valleys were also major routes may be relevant here.

This connectivity might well have contributed to some of the changes that we see in the 1st century BC in Britain, particularly the decline of some larger settlements, but over much of the island, the process of settlement expansion that commenced around the same time as in France seems to continue unabated or even to intensify. This is significant, as it would seem to rule out more global factors such as climate change from a major role in developments on the continent. In Britain, too, many of the sites founded in the late Iron Age were occupied without break into the Roman period, contrasting with France, where continuity across this transition is more unusual. This brings us to a final point, which may not be without relevance: whilst the average life span of rural sites in Roman Gaul was longer than in the Iron Age, even in the high Empire, sites were regularly abandoned for no reason that is obvious to us today and fluidity of rural settlement remained the norm right up until the full Medieval period.

References

• Audouze, F. and O. Buchsenschutz (1989): Villes, villages et campagnes de l’Europe Celtique, Paris.
• Bayard, D. and J.-L. Collart, dir. (1996) : De la ferme indigène à la villa romaine, Actes du II^e Colloque d’AGER tenu à Amiens en 1993, RAPic Numéro spécial 11, Amiens.
• Bertrand, I., A. Duval, J. Gomez de Soto and P. Maguer, dir. (2009): Habitats et paysages ruraux en Gaule et regards sur d’autres régions du monde celtique, Actes du XXXI^e Colloque de l’AFEAF, Mémoire 35, Chauvigny.
• Birkhan, H., dir. (2007) : Kelten-einfälle an der Donau, Österreichische Akademie der Wissenschaften Philosophisch-Historische Klasse Denkschriften 345, Vienne.
• Blancquaert, G., T. Lorho, F. Malrain and Y. Menez (2009): “Bilan et perspective de recherches sur les sites ruraux au second âge du Fer”, in: Bertrand et al. 2009, 5-23.
• Brun, P. and P. Ruby (2007): L’âge du Fer en France, Paris.
• Buchsenschutz, O. (1984): Structures d’habitats et fortifications de l’âge du Fer en France septentrionale, Mémoire de la Société Préhistorique Française, 18, Paris.
• Buchsenschutz, O. (2004): Les Celtes de l’âge du Fer dans la moitié nord de la France, Paris.
• Buchsenschutz, O. (2006): “Le monde rural et ses productions, II^e-I^er s. av. J.-C.”, in: Haselgrove, dir. 2006, 55-65.
• Buchsenschutz, O. (2007): Les Celtes, Paris.
• Buchsenschutz, O. and P. Méniel, dir. (1994): Les installations agricoles de l’âge du Fer en Ile-de-France, Études d’histoire et d’archéologie 4, Paris.
• Dobiat, C., dir. (2002) : Dürrnberg und Manching. Wirtschaftsarchäologie im ostkeltischen Raum, Römisch-Germanische Kommission Koll. Vor-und Frühgeschichte 7, Francfort.
• Collis, J. (1984): Oppida: Earliest Towns North of the Alps, Sheffield.
• Corsi, C. and F. Vermeulen, dir. (2010): Changing Landscapes – The impact of Roman Towns in the Western Mediterranean, Proceedings of the International Colloquium, Castelo de Vide – Marvão 15th-17th May 2008, Bologne.
• Courbot-Dewerdt, C. (2004) : L’évolution des établissements ruraux entre la fin de l’âge du Fer et la mise en place du système des villae dans le quart nord-ouest de la France, I^er siecle avant-II^e siècle après J.-C, Thèse de doctorat, Université de Paris I, Paris.
• Cunliffe, B.W. (2008): The Danebury Environs Roman Programme. A Wessex Landscape during the Roman Era. Volume 1, Overview, Oxford University School of Archaeology Monograph 70, Oxford.
• Deberge, Y. (2007) : “Les établissements ruraux fossoyés en Basse Auvergne du III^e au II^e s. av. n.è.”, in : Mennessier-Jouannet & Deberge 2007, 221-241.
• Deberge, Y., C. Vermeulen and J. Collis (2007). “Le complexe de Gandaillat / La Grande Borne: un état de la question”, in: Mennessier-Jouannet & Deberge 2007, 14-19.
• Deberge, Y., J. Collis and J. Dunkley, ed. (2007): Le Pâtural, Clermont-Ferrand, Puy-de-Dôme: un établissement agricole gaulois en Limagne d’Auvergne, DARA 30, Lyon.
• Fichtl, S. (2005): La ville celtique. Les oppida de 150 av. J.-C. à 15 ap. J.-C., (2^nd édition), Paris.
• Gerritsen, F. and N. Roymans (2006): “Central Places and the Construction of Tribal Identities. The Case of the Late Iron Age lower Rhine Region”, in: Haselgrove 2006, 251-266.
• Gransar, F. (2000): “Le stockage alimentaire sur les établissements ruraux de l’âge du Fer en France septentrionale: complémentarité des structures et tendances évolutives”, in: Marion & Blancquaert, dir. 2000, 277-298.
• Guichard, V., C. Ballut, D. Delhoume and Y. Deberge (2007) : “La place de l’âge du Fer dans l’histoire de la mise en valeur de la Grande Limagne d’Auvergne: l’apport des recherches d’archéologie préventive sur le tracé de l’autoroute A710”, in: Mennessier-Jouannet & Deberge, dir. 2007, 205-220.
• Hamilton, D. (2010): The use of radiocarbon and Bayesian modelling to (re)write later Iron Age settlement histories in east-central Britain, Thèse de doctorat en archéologie, University of Leicester, Leicester.
• Haselgrove, C. (1996): “La romanisation de l’habitat rural dans la vallée de l’Aisne d’après les prospections de surface et les fouilles récentes”, in: Bayard & Collart, dir. 1996, 109-120.
• Haselgrove, C. (2004): “Society and polity in late Iron Age Britain”, in: Todd, dir. 2004, 12-29.
• Haselgrove, C. (2005): “A new approach to analysing the circulation of Iron Age coinage”, Numismatic Chronicle, 165, 129-174.
• Haselgrove, C. (2009): The Traprain Law Environs Project: Excavations and fieldwork 2000-2004, Edinburgh.
• Haselgrove, C. (2011): “Reflections on the Iron Age Background to the Emergence of Villa Landscapes in northern France”, in: Roymans & Derks, dir. 2011, 45-60.
• Haselgrove, C., dir. (2006): Celtes et Gaulois. L’Archéologie face à l’Histoire. Les mutations de la fin de l’âge du Fer, Bibracte 12/4 Glux-en-Glenne.
• Haselgrove, C. and T. Moore, dir., The Later Iron Age in Britain and Beyond, Oxford.
• Hill, J. D. (2007): “The dynamics of social change in Later Iron Age eastern and south-eastern England c. 300 BC-AD 43”, in: Haselgrove & Moore, dir. 2007, 16-40.
• Kaenel, G. (2006): “Agglomérations et oppida de la fin de l’âge du Fer: une vision synthétique”, in: Haselgrove, dir. 2006, 17-39.
• Leroyer, C., M. Boulen, D. Marguerie et al. (2009): “Base de données et S.I.G. palynologiques sur l’Âge du Fer en France: autre approche du paysage végétal et de son anthropisation”, in: Bertrand et al., dir. 2009, 447-468.
• Maguer, P. and D. Lusson (2009): “Fermes, hameaux et résidences aristocratiques entre Loire et Dordogne”, in: Bertrand et al., dir. 2009, 423-459.
• Malrain, F., S. Gaudefroy, F. Gransar, G. Auxiette and P. Méniel (2005): “La protohistoire récente”, RAPic, 3/4, 127-176.
• Malrain, F. and E. Pinard (2006): Les sites laténiens de la moyenne vallée de l’Oise du V^e au I^er siècle avant notre ère, RAPic Numéro spécial 23, Amiens.
• Marion, S. and G. Blancquaert, dir. (2000): Les installations agricoles de l’âge du Fer en France septentrionale, Études d’Histoire et d’Archéologie 6, Paris.
• Matterne, V. (2001): Agriculture et alimentation végétale durant l’âge du Fer et l’époque gallo-romaine en France septentrionale, Archéologie des Plantes et des Animaux 1, Montagnac.
• Mennessier-Jouannet, C. and Y. Deberge, dir. (2007): L’archéologie de l’âge du Fer en Auvergne, Actes du XXVII^e Colloque de l’AFEAF, Clermont-Ferrand, 29 mai-1er juin 2003, Monographies d’archéologie méditerranéenne, Lattes.
• Nillesse, O. (2009): “Activités, métiers, vie quotidienne dans les établissements ruraux de l’ouest de la France à travers l’instrumentum (Hallstatt D / début du Haut Empire)”, in: Bertrand et al., dir. 2009, 45-83.
• Nouvel, P. (2004): Des terroirs et des hommes, dynamiques des organisations spatiales dans le bassin de l’Yonne moyenne et leur évolution de la fin de l’âge du Bronze au Haut Moyen Âge, Thèse de doctorat en archéologie, Université de Bourgogne, Dijon.
• Nouvel, P., P. Barral, S. Deffressignes, V. Riquier, J.-M. Séguier, N. Tikonoff and M. Zehner (2009): “Rythmes de création, fonctionnement et abandon des établissements ruraux de la fin de l’âge du Fer dans l’Est de la France”, in: Bertrand et al., dir. 2009, 109-151.
• Petit, C., dir. (2005) : Occupation et gestion des plaines alluviales dans le nord de la France de l’âge du Fer et à l’époque gallo-romaine, Annales Littéraires 786; Série Environnement, sociétés et archéologie 8, Besançon
• Rieckhoff, S. (1995): Süddeutschland im Spannungsfeld von Kelten, Germanen und Römern, Trierer Zeitschrift, Beiheft, 19, Trèves.
• Rieckhoff, S. (2002): “Der Untergang der Städte. Der Zusammenbruch des keltischen Wirtschafts- und Gesellschaftssystems”, in: Dobiat, dir. 2002, 359-380.
• Rieckhoff, S. (2007): “Wo sind sie geblieben? – Zur archäologischen Evidenz der Kelten in Süddeutschland im 1. Jahrhundert v. Chr.”, in: Birkhan, dir. 2007, 409-440.
• Roymans, N. and T. Derks, dir. (2011): Villa Landscapes in the Roman North. Economy, Cultures and Lifestyles, Amsterdam Archaeological Studies 17, Amsterdam.
• Séguier, J.-M. (2005): “Dynamiques d’occupation du secteur de confluence Seine-Yonne à la fin de l’âge du fer et à l’époque romaine”, in : Petit, dir. 2005, 107-120.
• Séguier, J.-M. and G. Auxiette (2008): “Le Brassot à Étigny (Yonne): un établissement rural de La Tène finale de la vallée de l’Yonne”, RAE, 57, 185–204.
• Sharples, N. (2010): Social Relations in later prehistory. Wessex in the first millennium BC, Oxford.
• Simons, A. (1989) : Bronze- und eisenzeitliche Besiedlung in den Rheinischen Lössbörden, BAR Int. Series 467, Oxford.
• Todd, M., dir.(2004) : The Blackwell Companion to Roman Britain, Oxford.
• Trément, F., C. Mennessier-Jouannet and J. Argant, dir. (2007): “Le bassin de Sarliève : occupation du sol et paléo-environnement à l’âge du Fer”, in: Mennessier-Jouannet & Deberge, dir. 2007, 385-400.
• Trément, F. (2010): “Romanisation et dynamiques territoriales en Gaule centrale. Le cas de la cité des Arvernes (II^e s. av. J.-C.–II^e s. ap. J.-C.)”, in: Corsi & Vermeulen, dir. 2010, 85-104.
• Von Nicolai, C. (2009): “La question des Viereckschanzen d’Allemagne du Sud revisitée”, in: Bertrand et al., dir. 2009, 245-280.
• Wells, P. S. (2005): “Creating an Imperial frontier: Archaeology of the formation of Rome’s Danube borderland”, Journal of Archaeological Research, 13, 49-88.

Footnotes

1. Buchsenschutz 1984.
2. Buchsenschutz 2004; 2006; Buchsenschutz & Méniel 1994.
3. Blancquaert et al. 2009.
4. Audouze & Buchsenschutz 1978; Buchsenschutz 2007.
5. Brun & Ruby 2007.
6. Leroyer et al. 2009.
7. Matterne 2001.
8. Gransar 2000.
9. Guichard et al. 2007.
10. Kaenel 2006.
11. Trément et al. 2007.
12. Courbot-Dewerdt 2004.
13. Maguer &Lusson 2009.
14. Malrain et al. 2005.
15. Haselgrove 2005; Nillesse 2009.
16. Haselgrove 1996; Maguer & Lusson 2009.
17. Nouvel et al. 2009.
18. Collis 1984.
19. Nouvel 2004.
20. Rieckhoff 1995.
21. Fichtl 2005; Kaenel 2006; Rieckhoff 2007.
22. von Nicolai 2009.
23. Rieckhoff 2002.
24. Wells 2005.
25. Rieckhoff 1995.
26. Wells 2005, 67-71.
27. Simons 1989; Gerritsen & Roymans 2006.
28. Hill 2007.
29. Cunliffe 2008.
30. Sharples 2010.
31. Haselgrove 2004.
32. Paul Sealey: “Where have all the People gone? A puzzle from late Iron Age Essex”, lecture to the Antiquarian Society Annual Spring Conference, 19 march 2011 (From Camolodunum to Durobrivae: aspects of Roman life in the eastern region).
33. Haselgrove 2009.
34. Hamilton 2010.
35. Haselgrove 1995.
36. Haselgrove 2011.
37. Sharples 2010.
38. Fichtl 2005; Haselgrove 2005; Nillesse 2009.
39. Malrain & Pinard 2006.
40. Trément 2010.
41. Deberge et al. 2007.
42. Deberge 2007.
43. Deberge et al. 2007.
44. Guichard et al. 2007.
45. Nouvel et al. 2009.
46. Séguier 2005.
47. Séguier & Auxiette 2008.
48. Nouvel 2004; Nouvel et al. 2009.
49. Nouvel et al. 2009.