URBANlSATlON IN ATLANTlC EUROPE IN THE lRON AGE
by John COLLlS Professor, Departament of Archaeology and Prehistory. University, Sheff ield, England
In this paper I shall primarily be considering central and northern France and Britain in the two centuries before the Roman conquest. In France this is essentially the second and beginning of the first century BC, ending with Caesar's conquest of Gaul in 58-51 BC; in Britain it will concern the first century BC and the first half of the first century AD -the south of England was invaded in 42-43 AD, but the north was only conquered about a generation later. I shall also use the term <.urbanisation)>in a loose way. Though many would accept that some of the oppida mentioned by Caesar had many characteristics we associate with urban sities, -centres of trade, industry and administration- in the late lron Age there are many other sites and settlement systerns which were almost as complex as urbanised societies, and perhaps on the threshold of urbanisation. Though terms like <
CENTRAL AND NORTHERN FRANCE
These are areas which ostensibly follow the major trends of central Europe, culminating in the appearance of the oppida during La Tene DI. However, generally we know little about settlement patterns in the preceding periods, And much of this area is devoid of the cernetery evidence we have for some areas of central Europe, e.g. the La Tene A burials of the Hunsrück Eifel or the La Tene B-C flat inhumation cemeteries of Hungary and Czecholovakia. Champagne and the Aisne valley are the exceptions, as here there is an almost continuous tradition for burial. However, archeological obsession with these cemeteries has elevated the Champagne to an importance wich it probably did not possess in the lron Age, and only now is settlement evidence emerging from sites such as Bourges which will provide a more balanced view. Settlement archaeology in France is still in its infancy, so I shall concentrate on three regional studies which have been under way since the 1970s- 1980s, in the Auvergne, in the Berry, and in the Aisne valley.
THE AUVERGNE
Great importance was accorded to the Arverni, with their kings Luernios (((the richest man in al1 Gaub), his son Bituitos, or the first French national hero, Vercingetorix, by classical sources such as Posidonius, Caesar and Livy. This has neves been matched by the archaeology. Until recently burials were virtually unknown, and none fall into the rich class (Loison et al. 1992). Details of the chronology too are only just being clarified. However, we perhaps now know more about the settlement pattern and its development in the late lron Age than anywhere else in France. Two major settlement excavations have taken place, Aulnat and Gerzat, wiht many other chance finds or small scale excavations, and a continuing tradition for intensive field walking and survey. The systematic surveys by Mills (1986), while producing important evidence form most periods, revealed little La Tene material. In part this may be due to sampling problems -the sites rnany not be evenly distributed across the landscape- but also because many of the sites lie in the plains, the <
Settlements belonging to La Tene A (fifth century BC) are at present poorly represented, but are generally in low-lying situations. From La Tene B a period of intensive occupation begins on the clays and gravels of the river terraces, and especially on the terre noire of the limagnes. The settlement pattern gradually intensifies with more and more settlements appearing in the plain up to La Tene DI at the end of the second century BC. Generally the sites seem to be small in size, around 300 m in diameter. The sites are especially dense in the area to the east of Clermont Ferrand around the modern villages of Aulnat and Gerzat, and are often only a few hundred metres apart. Many of these settlements have produced evidence for industrial activity. The excavation of Aulnat-La Grande Borne has produced traces of a wide range of industries from the end of La Tene B until La Tene DI, in fact virtually al1 the industries later found on the oppida though not al1 were present at one time (Collis 1980). Bone buttons and bone dice were manufactured especially in the later phases, glas was probably worked, and some bone tools and haematite crayons may be from pottery manufacture. lron slag is especially common in the earlier phases, and is mainly from smithing. lron was imported in the form of paddle shaped currency bars, and the site has produced a wide range of iron objects -spears, scabbard fragments, a cauldron hook- but the only items certainly manufactured were brooches. Several hundred fragments of crucibles were found, mainly containing traces of copper and tin, and in a later phases some contained gold and silver (Smith and Collins i982), but in none of these cases is it clear what was being made except for one pit which produced moulds for making flans for gold and silver coins (Tournaire et al. 1982). Textiles were also produced, and awide range of bone points are perhaps for net making or a sort of one needle knitting. The settlement at Gerzat presents a considerable contrast, though the evidence relates mainly La Tene DI. Glass was worked -there is a fragment of a cake of raw purple glass, and one or two bone points are present, but are rare in comparison to Aulnat. The only other industry present is blacksmithing, in the form of bun-shaped lumps os slag, and is much more common than in thecontemporary phase at Aulnat. Several other sites have produced evidence for bone working or iron working, and at three sites, Lezux (Mennesier-Jouannet1991 ), Chamalieresand Randan (Miallier 1984), pottery kilns have been excavated. The range and variety of local pottery developed enormously during La Tene C, such as the highly original painted wares. The development of contacts with the Mediterranean world can also be documented at Aulnat. In the earliest phase, La Tene 82, only coral was definitely being imported. In La Tene C fine black slipped table wares form ltaly appeared along with painted white wares form the Marseilles region and mortaria of unknown origin. In thefinal phases in addition to these imports large quantities of Campanian amphorae (Dressel la) are found on al1 sites, while Aulanat produced a carnelian intaglio (Henig and Collins 1987). The pattern that is emerging is of a number of dispersed sites which are engaged in trade and industrial production.Around Clermont Ferrand there is a cluster of sites which collectively seem to function as an urban settlement, but lacking the density of population found on the later oppida.
Some time during LaTene DI, most, if not al1 of these sites were abandoned. The latest brooches are typical Nauheim fibulae. The latest coins form both Gerzat and Aulnat are potin Nash 594 ((along cou>>(Malacher and Collins 1992). These are also found on the plateau site of Corent, and it also has some Nauheim brooches. But Corent has also produced some coin types which are not found on the lowland sites -the (cau renard>)type La Tour XII 3966-3669 of which nearly 200 exemples are recorded from Corent. A handful is known from another plateau site the C6tes de Clermont, but is otherwise confined to one or two other finds from other sites. Other coin types show a similar, though not so exclusive pattern- the inscribed MOTVDIACA and EPOS types and the ADCANAVNOS series for which dies have been found on Corent. Though some finds may have been wrongly assigned to Corent, the pattern which emerges is of a shift of the concentration of population to Corent during La Tene DI, with subsequent occupation, or a least coin using, confined to the site in the period around the Roman conquest. No large scale excavation has yet taken place on Corent. It is a naturally defended plateau of some 75ha, of which some 40 ha have a dense scatter of amphorae, mainly Dressel lb, and other finds. The site has been heavily pillaged by treasure hunters and collectors. Some 2000 coins are recorded, mainly in private collections. Our intensive surface collection is defining areas of densest occupation, but sondages suggest the lron Age occupation level is very thin, and has largely been destroyed in ploughing. Other than coin manufacture, there is at present only limited evidence for metal working among the finds. Corent survived as a major settlement only for about a generation, perhaps less. No traces of defences survive, but the slopes are generally so precipitous that they were perhaps unnecessary. Its probably successor is at Gondole, a low-lying site of about 35 ha on a grave1 terrace between the river Allier and a stream, the Oison. The spur is cut off by a massive bank and wide ditch typical of the Fécamp series, though these ofter cover a murus gallicus. Field walking has produced a small amount of Dressel lb amphora, a Jezerine brooch, al1 typical of La Tene D2, and a d~zencoins typical of the final phase at Corent. there is however, also a scatter of occupation, as yet unmapped, outside the defences. Sondages in the interior have produced Gallo-Romaine Précoce finds, just post-datin the Roman conquest. The next site in the sequence is the hill-top site of Gergovie, which started around 30-20 BC. Its coin list is dominated by the latest inscribed Gallic coins -late EPAD and VERCA coins- as well as the early Augustan coins from colonies such as Nimes. The defences are of Roman date. Though this site fits Caesar's description of Gergovia, there are no finds dating that early, and the name Gergovie was only given to the plateau in the 19th century. The excavations of the 1930s and 1940s revealed an industrial area with ¡ron working, and stone built houses. Except for a temple, this site too was deserted within a generation during the reign of Augustus, and a new town laid out in the valley, Augustonemetum, on the site of modern day Clermont Ferrand. The problem presented by these three settlements of Corent, Gondole, and Gergovie is the lack of contemorary settlements. Though the site os Aulnat -La Grande Borne and Corent could in part be contemporary- both nave Nauheim brooches and potin coins -it is more likely that the abandonment of Aulnat and the establishment of Corent happened at approximately the same time while the brooches and coins were current. Some pottery types typical of the latest phase of Aulnat, such as the Jatte dlAulnat, are so far absent from Corent. There is a possibility that occupation continued at or near Gerzat, and there is a recent find of a MOTVDIACA coin from the site, but the present lack of the relevant brooch and coin types, or of late Campanian or early Arretine imports, suggets a break. Later Arretine wares are present in the cemetery which overlies the La Tene settlement. From the csattered finds elsewhere in the Grande Limagne, it would seem the pattern presented by Aulnat and Gerzat -abandonment during La Tene DI, re- occupation in the late augustan period- is fairly typical, though it must be stressed, for instance, that potin coins are only known from Gerzat and Aulnat, the only two sites which have been extensively excavated. Future excavations, especiatly of villa sites, could change this picture. This model of total abandonment of openn sites and nucleation into defended sites does present problems. Firstly the 40 ha site of Corent hardly accounts for the massive population formerly located in the Grande Limagne. Though there may have been a drop in population with the defeat of Bituitos by the Romans in 123 BC, the Arverni were still a major power at the time of Caesar. The ravaging of the countryside by Brutus following Caesar's defeat at Gergovia in 52 BC it too late to account for the phenomenon. These problems are even greater in areas such as Riom and Issoire, where there are no defended sites at all, though we know little about what was happening in this perod on the highergrounds of the Massif Central. The Auvergne is discussed in detail because it has a fairly complete settlement sequence, and provides a model for other parts of France; firstly the abandonment of dispersed open settlements and the foundation of the defended oppida during La Tene DI ; and secondly in the fluid nature of the oppida -short-lived sites which were densely occupied but quickly abandoned, until finally in the Roman period permanent undefended administrative towns were established, either on the sites of existin oppida (Paris, Besan~on),or on new accessible sites (Clermont Ferrand, Autun).
THE BERRY
On the western side of the Massif Central, the most comparable work to that in the Auvergne is that carried out by Olivier Buchsenschutz and Nigel Mills in and around Levroux (Buchsenschutz et al. 1988). Little is known about the Early LaTene period in the Berry except old finds of burials with imported Mediterranean vessels, and more recent finds of settlements with imported pottery around Bouges (Rouffier et al. 1985)) supporting the importante placed on the Bituriges by Livy and their involvement in the Gallic migrations to northern ltaly in the early 4th century BC. In the Late LaTene,the pattern of oppida in the Berry is unusual, with a number of small or medium size sites, but no massive enclosures. The picture at Lezoux parallels, but is slightly different from, the Auvergne. Only one open settlement is known, Les Arenes at Levroux (Buchsenschutz et al. 1992), but this was already nucleated and could be described as a small urban settlement. It was certainly involved in industrial production, including coin manufacture (Tournaire et al. 1982). The settlement starts in La Tene C, and was abandoned in La Tene DI. The succeeding oppidum on an adjacent hill top has largely been destroyed by later medieval activity. It was about the same size as the open settlement, and was defended by a murus gallicus, later replaced by a Fecamp type rampart. It was abandoned in the Augustan period, when a Roman town was established at the foot of the hill, partly overlying the previous lron Age open settlement. Nigel Mill's field survey around Lezoux has proved largely negative for the lron Age, with Roman farms and villas first appearing in the first century AD.
THE AlSNE VALLEY
Early an Middle La Tene settlement patterns are known mainly form cemeteries such as Pernant, but some settlements have also been excavated in recent years (Royrnans 1990). Occupatión was largely confined to the grave1terraces of the river, and despite extensive survey, both field walking and aerial photography, the surrounding hills seem to have been largely devoid of settlement in the lron Age, except perhaps in the final stages when there are some coin finds (Demoule and lett 1985). Most of the work on the oppida has concentrated on the middle reaches of the river, especially the large scale excavations at Guignicourt and at Villeneuve St. Germain. Both are low-lying sites adjacent to the river, with bankand ditch defences, and in both cases a dense but highly organised pattern of streets, including palisade enclosures containing large timber buildings, as well as various other small houses and workshops. lndustries such as iron working, coin manufacture, etc, are represented. Both are relatively short-lived, as in the Auvergne, with occupation lasting perhaps a generation or so. In the area of Soissons, the sequence of events parallels that in the Auvergne, with an early phase of open settlements such as Berry-au-Bac, or the newly discovered site at Damaray (Haselgrove pers. comm.) which form initial sondages may well parallel Aulnat in its industrial activity. Settlernent was then concentrated on the low lying settlement of Villeneuve St. Germain (Audouze and Buchsenschutz 1989), but thas was abandoned for the hill-top oppidum of Pommiers which lies on the opposite side of the river, and is defended by a Fécamp rampart. The coin series at Villeneuve is dominated by uninscribed and potin coins, Pommiers by the later types, such as the inscribed CRlClRV series. One pecufiarity shared by both these defended sites is the division of the site into four sectors by ditches which run at right angles to each other across the site. In the case of Villeneuve, Debord et al. (1 988) have suggested this may be sorne sort of subterranean structure with a roof supported on massive timber posts. Pommiers was abandoned in the Augustan period, for the Roman town beneath modern Soissons.
From these three different areas of France we see a roughly parallel sequence from open settlements to nucleated oppida which often have a short life-span, even when densely occupied. This change happens over much of central and northern France, southern Germany and Switzerland at a time when the Nauheim fibula was in fashion. At Besan~onand Yverdon (Orcel et al. 1992) the earliest buldings are dated by dendrochronology to around 120 BC, and this would seem a reasonable datefor the shift elsewhere-several of thesesites were well established by the time Caesar arrived in 58 BC. In other parts of France generally only part of the sequence is known, for instance the shift from Bibracte / Mont Beuvray to Autun. However, ín certain areas the situation is much more stable, especially in the valley of the Saone, and it is these stable sites about which we know much less, largely because they underlie modern towns and so are inaccessible- Chalons-swr-Saone, Besancon, Reims, Bourges, París. For some we only have documentary evidence, but some, like Roanne, may well start relatively early, and were continuously occupied throughout the Late La Tene and on into the Rornan period (Guilhot et al. 1992). However, this pattern of uniformity may well be misleading. In parts of France there are few or no oppida, and we know virtually nothing of the settlement patterns. In Champagne the cemetery evidence suggests considerable continuity on the small farms and settlements thoughout the later lron Age and early Roman period, with no major shift to nucleatedsettlements. Brittany presents yet another case, with a longer tradition for hill-forts, but little extensive excavation has taken place until the last decade. Recent excavations of minor farming settlements, and the excavation of the multi-phase defended site of Paule will revolutioniseour knowledge (Arramond and le Potier 1990), but generally it would appear that Brittany shares more in common with southern England than it does with the rest of France, and it is to southern England that we now turn.
THE BRlTlSH lSLES
The pattern of hill-fort and related structures in Britain is extremely complex, both chronologically and geographically, the earliest defended hill top sites date to the Neolithic (e.g. Crickley Hill and Carn Brea - for a general recent summary of Brithis prehistory see Darvill1987). They appear in increasing numbers in the late Bronze Age from about 1000 BC. Some hill-fort sites, especially in western England were reoccupied in the Late Roman period, especially in the fourth and fifth centuries, and a very small number where turned into burhs in the tenth century as centres of defence against Viking attack. The vast majority of sites however belong to the lron Age, 600 BC to 100 AD (Cunliffe 1991). There is a general geographical trend from east to west. In the east hill-forts are rare if not unknown, and tend to be abandoned early in the Iron Age, perhaps around 500-400 BC. This process of gradual abandonment of a hill-fort style of life gradually shifts further to the west during the later phases of the lron Age. In a north-south strip through central England, parts of Wales, and into Scotland, hill-forts tend to be relatively large, generally around 5 to 20 ha. In the extreme west and north the landscape is dominated by small fortified homesteads of 0.5 -1 ha, with many regional variations such as the ((rounds>)of Cornwall, and the (
The term (~portof trade,) is a term used by economic historians for sites at which exchange took place (Polanyi 1963), usually on the boundaries between tribes or states, but where the control of the site was independent of the adjacent states - the free port of Danzig is a modern exarnple. Many punic or Greek sites such as Massilia or Naucratis fall into the category. Sites need not necesarily be ports, they can be inland on boundaries. They may be temporary, seasonal, or major urban centres engaged in industrial production. One site which may fall into this category on the south coast of Britain is Hengistbury Head (Cunliffe 1987). It controls river routes on the Stour and Avoh which run into the heartland of Wessex, and also lies at one end of one of the main short cross-channel routes, leading to the Channel Islands, Normandy and Brittany. It lies just behind the lsle of Wight, the ancient Vectis, which could be the lctis referred to in connection with the tin trade, and it is also likely to be 4he emporion), referred to by Strabo. This area is on a cultural boundary between the hill-fort zone of Dorset, and an area of Hampshire typified by open settlements in thefirst century BC. By the time of the Roman conquest, these are different tribal areas, in Dorset the Durotriges, in Hampshire the Atrebates. It is situated on a peninsula cutting off Christchurch Harbour form the sea, and is defended by an undated double bank and ditch. The site was occupied around 500-400 BC, so the defences could be earlier, but are more likely to belong to the end of the Middle lron Age around 100 BC. In this period part of the enclosed area was occupied by typical round houses. This period, up to around 50 BC was the 1 major period of trade contacts, especially with Brittany, and imported pottery from northern Armorica is not uncommon. Mediterranean goods include Dressel la amphorae and figs. The range of pottery from other parts of southern Britain is extensive, including material from Cornwall and Devon, the tin producing areas. Industrial production encompassed the working of iron, bronze, silver, gold, and glass, though some of this activity is poorly dated. Occupation continued into the Late lron Age, when palisade enclosures replaced the former free layout, but trade fell away, and later amphorae such as Dressel lb from ltaly and Pascual I from Spain are rare. The hinterland seems to extend from Cornwall to central Hampshire. Amphorae turn up sporadically on hill-forts such as Maiden Castle, but in areas like central Hampshire it is especially the small farming settlements which produce them. For instance, the extensively excavated farrn at Owslebury (Collis 1970) produced fragments of at least five or six amphorae, suggesting they were being imported for special occasions every 10-15 years (Williams 1981). The distri bution of silver Armorican coins in southern Britain may be another indicator of this trade. Two other port sites are known. One, Mount Batten at Plymouth, has largely been destroyed (Cunliffe 1988), but has produced an interesting range of bronze ornaments both from Britain and elsewhere along the Atlantic coast, and its main period of importante may pre-date Hengistbury. It has however produced one or two sherds of Breton pottery. It lay at the mouth of one of the main tin producing rivers, the Plym, and one piece of tin ore, cassiterite, is recorded. Given the rarity of Breton pottery, it may have played a subsidiary role to Hengistbury Head in the Middle lron Age. The final site is Poole Harbour, which seems to have replaced Hengistbury Head in the second haIf of the first century BC. Excavations have been lirnited (Woodward 1987), but several sites around the harbour, or on islands within the harbour itself, have produced fine wares from central Gaul and Pascual 1 amphorae. All these sites lie in areas which at the time of the Roman conquest, seem to have tacked any centralised political organisation other than a loose tribal federation. They are thus unlikely to have been centrally controlled.
BRlTlSH OPPIDA
The use of this term in England is different from the continental usage. It is used to denote large enclosures of several hundreds if not thousands of hectares, defined by massive linear dykes, and dating to the end of the first century BC and continuing into the first century AD. They are largely confined to southeast England, though one example, Stanwick, is known in the north. Sorne, Colchester, Silchester, and St. Albans, developed into major Roman towns, and Colchester (Camulodunum) was considered to be the British capital at the time of the conquest, and Claudius himself travelled to Britain to supervise its capture. In the century between Caesar's brief invasions of 55 and 54 BC and the Claudian invasion, British society underwent a major evolution, demonstrable from the written, numismatic, and archeologicalsources. At the time of Caesar we are told of tribal groupings and local kings, and a larger organisation of tribes which collaborated to select an overall leader, Cassivellaunus, to conduct the campaign against the Romans. The only <
ln the last thirty years there has been a general shift from migration models as a rneans of explaining the lron Age, to socio-economic models. One of the most favoured class of these new models is the World Economy or Centre (Core) - Periphery models (e-g. Brun 1987, Rowlands et al. 1987). Within this general geographical framework in the Late Iron Age, Spain, France, and Britain can be viewed in terms of threeconcentricarcs.The inner arc isformed by the Mediterranean iittoral, consisting of Phoenician, Greek and Roman colonies or their successors, and small native urban agglomerations with an urban character. Beyond this is a second zone in which large urban agglomerations appear, though their size may be more an indication of a primitve settlement system. In includes the oppida of France and England which form the subject of this article. Beyond this is a further zone of hillforts and defended homesteads, more typical of the highland areas of the Atlantic Iittoral. Attractive and useful though this Romano-centric model is as a framework for analysis, it provides only a partial picture, and it is perhaps time for us to recall a more Atlantic perspective, such as we already have for studying the Neolithic. The simple trajectory -Megalithic tornbs, Bronze Age hoards, lron Age hill-forts- is certainly an oversimplification, but it does emphasise a unity along the Atlantic coast. Had we continued our arc around the Mediterranean, and extended the outer circle into central Europe, we would have found the hill-fort absent, even in areas where it had existed in earlier times (e.g. eastern Germany, Poland), so there is sornething special about the Atlantic that deserves treating it as something of an entity. The to-ing and fro-ing along the coast, the trading and piracy, the intermarriage and local warfare that went on throughout Prehistory did not produce an absolute unity, but it did mean that ideas spread quickly, and parallel developments occurred. The problem is that especially in the lron Age, when in these areas there is only a limited material culture surviving in the archaeological record, these contacts are only rarely demostrable in truly tangible terms -the Spanish- southwestern French style brooches form Cornwall, the Armorican pottery and coins from southern England, rare British CO~~Sin western France (Boudet 1992), or the Navan ape. Less tangible are similarities between pottery styles or hill-fort rampart constructions. However, perhaps the most stimulating approach is to assume the similarities (e.g. in the appearance hill-fort sacieties), and instead look at the differences in these hill-fort and early urbanised societies. As I have tried to emphasise in this paper, even though many of the processes going on in these societies rnay be similar-the development of trade with the Mediterranean, the increasing complexity of industrial production, etc- the reactions of individual societies may be very different, and a comparative approach will help highlight these differences, and hopefully the reasons for them. Simply laberlling cites by typological definition as «hill-forts* or q