Alternatives to Urbanism? Reconsidering Oppida and the Urban Question in Late Iron Age Europe

Large complexes, known as oppida, in Late Iron Age Europe (c. second century BC to first century AD) represent one of the most significant developments in prehistoric Europe. Spread from southern Britain to Bohemia (Fig. 1), their large size and extensive ramparts have led to discussion of oppida being dominated by debate over whether they can be regarded as the first manifestation of urbanism in Europe (e.g. Collis 1984; Woolf 1993; Sievers and Schönfelder 2012). Yet, oppida have been surprisingly absent from wider reappraisals of urbanism (e.g. Cowgill 2004; Marcus and Sabloff 2008). This paper examines why oppida diverge from ‘traditional’ forms of urbanism and, without drawing direct analogies, explores how they compare to a range of large social centres around the world, whether classified as ‘urban’ or not. I aim to move beyond socio-evolutionary paradigms that prioritise particular forms of urbanism and caution against allowing the terminology of urbanism to become limiting, focusing on labels and perceived social complexity at the expense of interpreting the roles of these sites. I suggest the morphology of oppida reflects the nature of Late Iron Age societies and, like many other ‘mega-sites’, they represent alternative solutions to managing increasingly large social entities.

The Latin term oppidum (plural oppida) derives from classical sources; most significantly, Julius Caesar used it to describe several locations in Gaul in the mid first century BC. By the early twentieth century, similarities in scale (by which I mean the area an oppidum covers) and apparent role as the apex of social hierarchy, meant oppida were argued to be a pan-European monument, with the term subsequently applied to a range of sites across Europe. Oppida as a phenomenon, however, vary significantly, encompassing both well-defined, fortified sites in central France and more polyfocal, sprawling complexes in southern Britain and Eastern Europe (Figs. 2, 3). For some, this has meant that the category oppida is too heterogeneous to be meaningful. Despite their variation, however, a number of general characteristics can be ascribed to them: (1) they are usually very large in size (over 25–50 ha), contrasting with immediately preceding settlement forms; (2) they are defined (if not always enclosed) by extensive ramparts; (3) they represent some form of socio-political apex; (4) they acted as centres of exchange; and (5) most had roles as ritual centres. Finally, many are characterised by a relatively rapid development and short period of occupation. As discussed below, such characteristics are shared with a range of alternative low-density settlement forms around the world (cf. Fletcher 2009).

The recognition that some unenclosed agglomerations could be comparable in size and role to oppida (Fichtl 2013) increasingly suggests that defining oppida using tight criteria is problematic. Equally, the discovery that earlier fortified sites (Fürstensitze) (sixth–fifth centuries BC) were much larger than originally thought (Brun and Chaume 2013) indicates a more complex trajectory in the emergence of large centres in the Iron Age. These developments suggest a looser framework is required to explore oppida, allowing examination of processes of change, rather than stagnant analytical constructs based largely on morphology. This might mean that many sites traditionally excluded from oppida debates (‘royal sites’ in Ireland, developed hillforts in Britain, and unenclosed agglomerations) should be included in discussion of how Late Iron Age societies used places to articulate power. To varying degrees, all these societies faced similar pressures, including increasing population; greater long-distance exchange and interaction; the direct or indirect impact of colonial expansion; and changing social structures. Oppida might be a heterogeneous category, but this very diversity may illustrate how societies developed different ways of managing social complexity. Many Iron Age societies in Europe were on trajectories that did not culminate in towns, as they are traditionally understood, which have too often been regarded as the ultimate goal of all complex societies. Whilst the imposition of oppida may often have been planned (Collis 2000), their variety potentially also indicates their experimental nature, which need not have been successful or developed in ways that were intended.

The translation of oppidum as ‘town’ has often been used to affirm the urban character of oppida. The criteria used to define urbanism elsewhere might not be relevant to Iron Age Europe, however. Even Collis’ (1984) influential volume, ‘Oppida: First towns north of the Alps’, recognised that many oppida lack the density of occupation or socio-economic roles required to pass the ‘urban test’. Urbanism studies have since moved beyond checklist approaches, as used by Childe (1950) and Weber (1966), recognising that many aspects they deemed important related to their particular research spheres (Osborne 2005, p. 7). This has led to broader definitions of towns avoiding rigid criteria, such as population size or nucleation, and focusing instead on their relationship (as places of economic, religious or administration significance) with a hinterland (Smith 2007, p. 5).

In recent years, there have been attempts to reintegrate oppida into urbanism debates (e.g. Fernández-Götz et al. 2014), although most of these continue to define urbanism using traditional criteria: large (and permanent) population; economic diversity; centre for long-distance exchange; and role as a central place (e.g. Wendling 2013). Many of these approaches have been useful in re-emphasising the significance of these monuments and recognising that some are comparable, in size at least, to examples of urbanism elsewhere in the world. There remains a danger however, of continuing to regard urbanism as ‘an accolade to be awarded or withheld, not a problem to be investigated’ (Osborne 2005, p. 7). In so doing we may overlook the specific nature of Iron Age ‘urbanism’, whilst continuing to situate oppida in a classical context (implying analogies with Greek poleis, for example) and missing other, more pertinent, comparanda.

Comparing Iron Age mega-sites (whether oppida, Fürstensitze or hillforts) to classical towns reflects a deep-seated problem in Iron Age studies, that of defining a society’s complexity using concepts, such as ‘states’, that are ultimately derived from the colonial context in which debates were forged (McIntosh 1995a). As in urbanization debates elsewhere (Gaydarska 2016; Gaydarska 2017, in her introduction to this special issue; Yoffee 2005), the appearance of oppida has often been connected to arguments over state formation (Collis 2000). However, correlating particular forms of urbanism and scale with social complexity can be misleading: societies can be complex, in terms of social organisation, without many of the attributes we traditionally ascribe to states (Kohring 2012); whilst states might manifest urban centres in different forms (Smith 2003, p. 13; Jennings and Earle 2016).

Irrespective of our definition of urbanism, therefore, we need to know how application of the term ‘town’ to Iron Age Europe improves our understanding of these places or the communities that inhabited them. A more important task is to address why these complexes appear so different from the cities with which we are more familiar. Rather than entirely abandon debates over whether oppida were urban (Woolf 1993, p. 231), exploring oppida within a comparative framework may address the important issues of whether Late Iron Age complexes were distinctive or whether equivalent forms of monument existed elsewhere. Such comparison should include not only traditional ‘urban’ forms but also other forms of large social centres from around the world. Rather than focusing on whether or not oppida can be called ‘towns’, we need to explore their significance in changing social dynamics and how they related to existing social organisation. Whether comparative examples are described as urban is perhaps less relevant than the extent to which they indicate similar social trajectories and roles for monumental places to the oppida encountered in the European Iron Age.

One starting point for broadening analysis is to explore how oppida compare with other ‘mega-sites’ that are difficult to place within traditional urban definitions but also challenging to adequately describe using other terms. One such group is ‘low-density settlements’, some of which are regarded as manifestations of ‘low-density urbanism’ (Fletcher 2009). These are characterised as large settlements (often in the hundreds of hectares) with low-population densities (from as many as fifty to as few as four persons per hectare). Claimed examples of pre-industrial low-density settlements that are potentially ‘urban’ come from various times and from spaces including Mesoamerica, Eastern Europe and Africa (e.g. Kusimba et al. 2006; Chapman et al. 2014; Isendahl and Smith 2013). Fletcher (2009, p. 9) has suggested that some oppida, with their huge size yet dispersed nature, might represent examples of the phenomenon.

One of the few oppida examined in detail, Bibracte (France), certainly appears to show some characteristics of low-density urbanism. Constituting an enclosed area of c. 200 ha, the range of buildings now recognised at Bibracte suggests a population of c. 5000 on the hill of Mont Beuvray. As is the case with research into low-density urbanism elsewhere, however, by shifting the focus from (seemingly) well-defined centres to hinterlands, the dispersed nature of the complex becomes apparent. A contemporary unenclosed agglomeration at Sources de l’Yonne, covering c.120 ha, just 3 km from Mont Beuvray, with similar activities and occupation density, indicates that the ‘Bibracte complex’ was far larger (Fig. 4; Moore et al. 2013). Fieldwalking has also revealed a range of contemporary settlements scattered between Mont Beuvray and Sources de l’Yonne (Barral and Nouvel 2012), which may be elements of a ‘sprawl’ between these agglomerations rather than independent farmsteads. Did all these settlements effectively represent a single centre?

Bibracte is not alone in displaying such features, with a comparable phenomenon in the Auvergne. Three oppida (Corent, Gergovie and Gondole) previously claimed to have been occupied successively, have now been shown to have been (at least partly) contemporaneous, forming a complex encompassing c. 2500 ha (Fig. 4; Poux 2014, p. 162). These may represent isolated multipolar ‘centres’ (Poux 2014, p. 163) or, as suggested for Bibracte, elements of a larger, low-density settlement.

Some earlier Iron Age complexes also consisted of multiple agglomerations. At the Late Hallstatt site of Bourges, unenclosed agglomerations were contemporary with an enclosed centre, all part of the same complex (Brun and Chaume 2013, p. 323). This may imply such multi-polar centres were more widespread and had antecedents extending further back in time. It also emphasises that recognising such arrangements is largely a result of investigation strategies, which explore beyond enclosed elements to examine the wider landscape.

The true nature of these settlements remains somewhat enigmatic, but in all cases the limits of the complexes expanded well beyond (and were not defined by) the ramparts. This increases their scale considerably, to hundreds rather than tens of hectares. As such, they are more akin to Mesoamerican low-density centres, the areas of activity on Mont Beuvray and at Corent representing denser occupation foci within a more dispersed complex. Such arrangements are also somewhat comparable to low-density African centres (Fig. 5), such as Jenné-jeno, Mali (McIntosh and McIntosh 2003) and Afikpo, Nigeria (McIntosh 1995a, p. 11). These consisted of a collection of agglomerations (‘villages’) spread across a number of square kilometres, representing discrete social entities and/or craft activity areas. That the complexes at Bibracte and Gergovie–Corent represent the only examples of this phenomenon seems unlikely and further fieldwork in oppida hinterlands may reveal more sites with multi-centric arrangements.

Recognition that enclosure may not always have defined the extent of the complex begins to blur distinctions between what we regard as elements of an individual oppidum whilst also challenging the supposed distinctiveness of different types, for instance those in Britain versus those on the continent. What typifies many oppida is a lack of dense occupation and/or presence of large open spaces, sometimes associated with incomplete boundaries.

Such characteristics are most clearly seen at the so-called ‘polyfocal’ sites in Britain, which consist of earthworks, often stretching for many kilometres, encompassing huge areas of landscape (Fig. 3), from two to three hundred (Bagendon and Stanwick) to over two thousand hectares (Camulodunum). Within these complexes are ‘elite’ enclosures, sanctuary sites and areas of denser occupation alongside larger, seemingly open, areas. At some sites, areas of relatively intensive occupation can be defined (e.g. Bagendon: c.16 ha) but these are a small fraction of the broader complex. Defining the limits of these complexes is also problematic: Verlamion (St Albans), for instance, consists of elements spread over 7 km² (Bryant 2007). Some open areas may have been for farming, but this does not explain why farmland needed to be defined by such monumental earthworks. While these dispersed arrangements are often regarded as a peculiarly British phenomenon, some continental oppida also display them. Reims, for example, has been claimed to be part of a larger complex (Haselgrove 2007), whilst sites such as Heidetränk or Zavist (Fig. 3) also incorporate both upland and valley areas within complex dyke systems.

The size and form of such complexes means they are as much landscapes as they are ‘sites’. Indeed, the nature of these monuments appears focused on defining landscapes and dictating how people moved around them. At Bagendon, for example, the earthworks acted as a funnel, channelling movement towards the elite enclosures, and the earthworks at Verlamion also formed ‘processional routeways’ (Bryant 2007). The disproportional labour expanded on the ramparts implies that they were used to emphasise the power and significance of ‘place’. These elements appear to have been designed to create a sense of theatre: at Bagendon the main trackway led visitors through the lines of earthworks, past a hive of industrial activity (perhaps allowing glimpses of camps of people or herds of livestock on the plateaux above) and finally climbed up to the main enclosures (Fig. 6).

The prime role of such oppida, then, was in choreographing movement of people. In this they resemble Irish ‘royal sites’, which also involved polyfocal activity, associated with large-scale earthworks, creating ‘arenas’ focused on the sacral role of kings (Moore 2012). Arrangements of large complexes in such a way is not unique to Iron Age Europe. Comparison might even be drawn with complexes like fourteenth century AD Great Zimbabwe in Africa. Spread over c. 700 ha, this complex also used monumentalised pathways to direct movement toward ritual and power centres. Akin to the ways in which Great Zimbabwe has been described (Pikirayi 2016), such oppida are better regarded as ‘powerscapes’ than a form of urbanism: places where topography, architecture and activities (industry, exchange) were manipulated to communicate the status of the community and the power of the place itself.

The provision of large open spaces within such complexes is likely to have been integral to such functions. There are indications from classical sources that one of the key roles of oppida was as foci for group meetings at times of crisis, ceremonies and decision making (Fernández-Götz 2014, p. 390). Such places reflect the nature of social organisation in the Late Iron Age, based on forms of clientage (Collis 2000, p. 233). Textual and coin evidence indicates such relationships extended over hundreds of square kilometres, encompassing many communities. Open areas would have allowed for periodic assemblies combining political, ritual and economic functions, enabling leaders to administer a dispersed populace without direct control or permanent population centres. This echoes the role of assembly places in Early Medieval northern Europe and even sites like Great Zimbabwe, all of which exhibit evidence for sacral kingship and negotiated forms of power where emerging elites used assembly to maintain their status (Fig. 7). Early Medieval assembly sites, although often with little in the way of structural elements, echo the nature of (some) oppida, with their focus on manipulating landscape settings as major meeting places, whilst at the same time being morphologically incredibly diverse (Semple and Sanmark 2013).

Their role as assembly places means that some oppida also show affinities with other large but ephemeral social centres, such as the so-called mobile capitals of Ethiopia. Dating from the fifteenth–nineteenth centuries AD, these consisted of small, permanently occupied nuclei, focused on a royal compound and occupied by just a few hundred people. At certain times of year they were augmented by temporary dwellings, with a concomitant increase in population to many thousands, and became the focus of power (Fletcher 2009, p. 8), while other activities, such as temporary markets, also sprang up (Horvarth 1969). Such centres, for example at Addis Ababa, could spread over many kilometres. Despite the obvious economic and environmental differences, it is possible to imagine that some of the dispersed oppida fulfilled analogous roles, with small permanent populations that occasionally (for tribute, negotiation, war) amassed hundreds or thousands of people within their bounds. This reflects Strabo’s (Geography IV.5) description of the ‘cities’ of Late Iron Age Britain as settlements in woodland for the corralling of men and cattle, which were not occupied for long, representing essentially temporary assembly sites rather than permanent centres of population. Whilst the veracity of classical depictions can (like the term oppidum itself) be questioned, it appears Greek and Roman authors were attempting to describe places for which they had no easy analogy. It is surely pertinent that such description come close to definitions of other assembly places found in Iron Age and Early Medieval Europe. As Horvarth (1969, p. 219) emphasises for the Ethiopian examples, these centres had the roles and scales of ‘cities’ but they were not permanent and were not urban as traditionally defined.

Even at more densely settled sites, spaces for the assembly of significant numbers of people appear to have been important. Bibracte, for example, contains substantial plazas, whilst oppida such as the Titelberg contained open spaces combining public assembly with ritual buildings (Metzler et al. 2006). Other structures also imply a combination of gathering and ritual, for example the theatre-like structure at Corent (Poux 2012), whilst at this multicentre location the seemingly empty plain (Fig. 4) might even have formed a deliberate open area for temporary congregations. That sanctuaries appear to have been central to many oppida reinforces the impression that ritual and social authority were intimately combined and may have acted as one of the draws for periodic assemblies (Fichtl et al. 2000). Such roles reflect the limited evidence for oppida being placed to exploit suitable agricultural land and indications that, whilst they were engaged in production and exchange, this was not their primary role (Fernández-Götz 2014). This lack of focus on an economic role is shared with some other sites we struggle to define as urban. These include, for example, the relatively large Hawaiian royal centres, dating to the eighteenth century AD, that had populations dispersed over many hectares, which included agricultural areas and foci of wooden temples and royal centres (Smith 2012). Meanwhile, ritual centres such as Chaco Canyon, in North America (Yoffee 2005, p. 168), also retained small populations of specialists that were only augmented by larger congregations at certain times of year, or had roles that were primarily religious. In both cases the extent to which they can be described as urban has been debated, Smith (2012, p. 337), for example, arguing for such status at the Hawaiian sites based on a broad definition which stresses the presence of institutions within these settlements that affected a larger hinterland.

Whether we should similarly extend such definitions of urbanism to the polyfocal oppida in Europe is perhaps less important than recognising that, despite their divergent form, a unifying thread emerges in the role of oppida: that their main function was for assembly rather than as population centres. The origins of oppida might explain this role. Many, despite their short duration, were located in pre-existing socially significant places in the landscape. On the continent, a number of oppida appear to have emerged from pre-existing sanctuaries that may have combined assembly and ritual (Fernández-Götz 2014). Elsewhere, the apparent emptiness of oppida locations prior to their construction (Hill 2007, p. 32), in what by the Late Iron Age were intensively occupied landscapes, suggests some locations were special places, perhaps assembly places that left little archaeological trace (Haselgrove 2007, p. 509).

The presence of pre-existing assembly places at oppida, like assembly locations in other contexts (cf. Semple and Sanmark 2013), is often hard to prove because they frequently lack structural evidence. Such locations, however, may explain the choice of oppida sites, on dominant mountain tops and in marshy valleys, seemingly ill-suited to traditional forms of urbanism. Such places were also often situated on agricultural interfaces marking liminal places in the landscape and allowing for access from different regions. On the continent at least, a high proportion of such locations contained earlier structures (Fig. 8) and, although usually unoccupied prior to the establishment of the oppidum, may have already been significant in local consciousness. It seems likely that such places were important in local identity, connecting communities to landscape features and ancestrally significant places. Such processes exploited cultural memory, transforming places with existing social resonance but transmuting what that locale meant in the present (cf. Holtorf 1996). Such utilisation of earlier monuments is well recognised for Early Medieval assembly, where it represented elites exploiting sites of ancestral power (Semple and Sanmark 2013, p. 531). Conveying a link to the communities’ past, whilst physically transforming these places, may similarly reflect the changing nature of power in the Late Iron Age, with locations previously focused on negotiated power now dominated by smaller sectors of society.

This raises the question of why in certain areas—such as East Anglia in Britain, and the Netherlands—although such places existed, and Late Iron Age societies with comparable levels of complexity emerged, locations were not monumentalised. Does this suggest the presence of somewhat different forms of social structure, perhaps controlling the emergence of hierarchy (cf. Haas 2001)? Or was power expressed in ways that did not require physical centres, perhaps peripatetic or through portable symbols? The juxtaposition of equally complex societies with and without ‘oppida’ suggests that the reasons why some complex social forms required monumental centres and others did not were themselves complicated (cf. Jennings and Earle 2016).

It is clear that the form of oppida reflects the social context in which they emerged and the nature of social systems that inhabited them. Many regions of the European Iron Age show scant evidence for hierarchical social systems before the first century BC, with little sign of elite representation, such as differentiated burial or material culture. Even in areas where hierarchies are claimed (eastern France and Germany), social stratification appears less marked—both within and between groups—than in preceding periods (Diepeveen-Jansen 2007, p. 385). Instead, these societies appear to have been heterarchical (Crumley 2003; Hill 2011), with a range of levelling mechanisms, such as labour potlatch, to maintain social equilibrium and minimise the power of community leaders (Hill 2011). Combined with the appearance of coinage and more differentiated burial rites, as noted above oppida have tended to be regarded as illustrative of state formation and (re)emergence of a hierarchical society dominated by kings who either resided within the oppida or used them as their administrative centres (e.g. Metzler 1995).

The morphology of oppida suggests, however, that transformations in social structure were more complex than the emergence of a simple hierarchy. The spatial layout of some oppida (e.g. Villeneuve-St-Germain; Conde-sur-Suippe; Manching) included enclosed settlement units, with their own houses, storage facilities and courtyards, reminiscent of contemporary rural settlements (Fig. 9). Such similarity has been argued to signify the transfer of rural elites into oppida (e.g. Wendling 2013, p. 473), or the physical manifestation of aristocratic authority (Fernández-Götz 2014, p. 384). It may, instead, demonstrate the very opposite: by retaining ‘rural’ settlement forms, the social unit it represented (probably extended households) continued as the social locus, downplaying status distinctions. Such an arrangement is not unique to Late Iron Age oppida: the outer settlement at the Late Hallstatt (sixth–fifth century BC) site of the Heuneburg also contains farmstead-type enclosures in distinct blocks below a more densely occupied hilltop (Kurz 2012). Such social forms are more redolent of negotiated power, with households retaining power, as in the heterarchies that preceded these centres.

Such layouts seem likely to mark the tension in transforming what were rural, heterarchical societies into more centralised social forms. This is characteristic of some quasi-urban centres where the social building block (the household) remained fundamental. In the large pre-colonial west African centres, often described as mega-sites or proto-urban, such as Ile Ife, the household remained the social and economic basis, despite an overarching hierarchical social structure (Ogundiran 2012). Elsewhere, the ‘village’ clusters recognised in other parts of Africa, such as the agglomerations of the Igbo in Nigeria and Jenné-jeno in Mali (McIntosh 1995a; McIntosh and McIntosh 2003) show little sign of social hierarchy. Instead, these were organised on a heterarchical basis, comprising village groups based on compounds for extended households, similar to those suggested for Conde-sur-Suippe (Fig. 9; Forde 1964, p. 50), with discrete areas appearing to reflect group autonomy rather than social or economic hierarchy (McIntosh 1995b, pp. 75–76; McIntosh and McIntosh 2003). In a similar way to the African examples, it seems likely that a core function of such arrangements was to organise work parties (perhaps on a neighbourhood basis) with the existence of farmstead-like compounds and evidence from many oppida that they were largely agriculturally self-sufficient, suggesting the household remained the locus of economic and social reproduction. Such forms of clustering in African centres marked the desire of heterogeneous groups to be ‘part of an urban entity without being subsumed by it’ (McIntosh 1995b, p. 76). As with the situation at the Trypillia mega-sites, where communal houses allowed for a set of nested social units to maintain the larger social entity (Chapman et al. 2014, p. 396), we may be seeing a situation where the heterarchical nature of pre-existing societies underpinned oppida spatial arrangements, even if society was already transforming into something more centralised and hierarchical.

Even for oppida where a more hierarchical social structure might be envisaged, such as Bibracte, the classical literature and archaeological evidence continues to suggest that leaders held power in trust, and that oligarchies of influential families vied for power (Collis 2000). It may be pertinent, therefore, that at most oppida the common ‘monumental’ aspect was the ramparts, rather than a central temple or palace. Some ramparts saw frequent rebuilding (four times at Bibracte between c.100 BC and 30 BC, representing a rebuild every generation). This was matched at Bibracte by frequent reorganisations of the interior. Considering the suggested nature of Late Iron Age society, this may emphasise leaders’ need to mobilise the community to make repeated statements of control over the physical space, the act of construction materialising the bond between leader and community.

Oppida as phenomena were not static and, like all forms of urbanism (Smith 2003, p. 17), were socially transformative places, where habitus is likely to have reconfigured social relationships. Thus the mega-centre nature of Bibracte and Gergovie–Corent in the mid first century BC is unlikely to reflect the role of those sites, or the social relationships they represented, in the late second century BC. We should not be surprised by this: the appearance of oppida was created by social transformation, to which their development in turn contributed. By drawing groups together and fixing assembly places more permanently it is likely their presence provided a new stage on which individuals and communities could seek and express power. In some cases, this may have led to their abandonment, as seen at relatively short-lived sites like Conde-sur-Suippe, with the social experiment of maintaining heterarchical social forms in an agglomeration perhaps found to be unsustainable. At others, such as Bibracte, they morphed into different architectural forms or changed their emphasis, increasingly focusing on production and trade, getting closer to common concepts of urbanism. The question of whether such a trajectory was common, however, is hard to gauge, having been obscured by the subsuming of these societies into the Roman Empire. Incorporating the rise and fall of the large centres of the early Iron Age in central Europe, an impression of cycles of boom and bust for mega-sites over the Iron Age has been suggested for parts of Europe where Rome’s influence was less direct (Salač 2012). This may indicate that a similar process of decline would also have taken place elsewhere, irrespective of the intervention of colonial conquest. The fact that the polyfocal ‘royal sites’ in Ireland never developed into urban-like centres might also imply that sites elsewhere in Europe were not on a trajectory to nucleated urbanism.

Comparing oppida to mega-sites and assembly places elsewhere in the world emphasises that oppida might indeed ‘represent an indigenous and separate urban tradition’ (Alexander 1972, p. 847). More importantly, however, it also reveals shared attributes with ‘alternative’ forms of urban and central places. Some share similarities with African and Mesoamerican low-density urbanism, others display greater affinities with Early Medieval assembly places. The very diversity of oppida (often important in including them in, or excluding them from, debates) thus reflects the heterarchical nature of the societies in which they emerged, which were nevertheless all transitioning to social forms dominated by smaller sectors of society. In some instances, this led to complexes more comparable to low-density urbanism; at others they retained looser ‘assembly’ like structures. The commonalities in all these societies, of negotiated power and household mode of production, meant that seldom, if ever, did they develop forms of urbanism similar to those in the classical world. The form of oppida instead appears to reflect the tensions within Late Iron Age societies between pre-existing heterarchical social organisations, which down-played status differentiation, and the emergence of larger entities that required social cohesion. As seen in the emergence of Early Medieval assembly sites, the diversity of oppida represents ways to articulate these transforming power relations. The implication is, perhaps, that these societies did not develop nucleated urbanism because they could not, but that they deliberately chose alternative forms of places to articulate society.

The dangers in cross-comparison should not, of course, be underestimated (see Gaydarska 2016; Gaydarska 2017, in her introduction to this special issue). The economic, social and environmental contexts of many alternative urban forms alluded to here are significantly different from oppida, making direct analogy impossible. Comparison does, however, indicate that oppida are not an entirely unique settlement form but are part of a much larger suite of agglomerations that do not sit easily within traditional definitions of urbanism. Recognising similarities between some oppida and assembly sites also emphasises that we should be wary of restricting debate purely to a question of ‘urban versus non-urban’ central places. Some oppida may even represent an important transition from un-monumentalised assembly to permanent urbanism. Through exploring comparisons with a range of significant centres around the world we may be able to move beyond social complexity paradigms that privilege high-density nucleation as the ultimate aim of all societies, enabling us to contextualise Iron Age societies in the diverse social forms of the pre-colonial world. Whether we can ascribe the label ‘urban’ to many of these centres, or to oppida specifically, has often overshadowed the more crucial question of why such societies developed complex mega-sites which contrast with the urbanism of the Classical and Near Eastern world. That relatively similar trajectories appear to have been followed in different geographical and temporal locations might imply that a host of complex societies developed alternative mechanisms and places which allowed these societies to function without an urban network in the traditional sense. Rather than expand our definition of urbanism to incorporate such sites, there is tantalising evidence that we should develop a terminology for these complexes, emphasising their different social and morphological context. Despite their problems, terms such as ‘mega-sites’ at least allow us to move beyond definitions such as ‘proto-urban’ which are in danger of imposing an anachronistic and hierarchical perspective on the complexity of many societies and social centres around the globe, including those of the European Iron Age.