Pragmatic colonialism, pragmatic agency

Pragmatism is concerned with the historical emergence and practical consequences of ideas, meaning and practice, considered in its full diverse expression in social life (Preucel Reference Preucel2006; Preucel & Mrozowski Reference Preucel, Mrozowski, Preucel and Mrozowski2010; Rorty Reference Rorty1991). Pragmatism’s concern with practical consequences gives it an ethical and normative core as in implications for what someone will, should and ought to do (Peirce Reference Peirce1998, 346). Pragmatism asks what impact a set of ideas, practices and social structures has on the world and on the people living within it.

This discussion of pragmatism is related to two key ideas: pragmatic colonialism and pragmatic agency. Pragmatic colonialism argues that colonialism is not a uniform process, but that the key strategies of colonialism (extraction of wealth and control over land) had to be adapted to the local conditions of the territory that was being annexed. The historic conditions structured the agentic capabilities of the imperial administration in establishing the colonial administration (Emirbayer & Mische Reference Emirbayer and Mische1998, 973), thereby producing a unique form of colonial rule specific to southern Roman Britain.

Pragmatic agency follows naturally from this, arguing that local communities had to adapt themselves to the new realities of living within a new colonial structure, and that patterns of activities and behaviour are the consequence of adaption to colonial rule to maintain social cohesion (Cipolla Reference Cipolla2013, 19–24). Pragmatic agency has been used in historical archaeology to circumvent the resistance/domination and rebellion/acculturation dichotomy by recognizing that local communities in a colonial context would have sought to maintain local community relations in the face of dramatic upheaval, and that as a consequence of this, adaption to external circumstances would have been short-term and reactive (Mitchell Reference Mitchell2009, 22; Preucel Reference Preucel2006, 50). Whereas traditional accounts of agency treated it abstractly, pragmatic agency properly appreciates that a community’s ideas, beliefs and worldviews were not only historically situated but were a direct consequence of the historical conditions within which they were situated (Gardner Reference Gardner2012; Rorty Reference Rorty1991, 23).

The concern with the historicity of imperial rule is a key one of postcolonial Roman archaeology, with theories such as discrepant identities, creolization and globalization all exploring the historicity of imperial rule and imperial identity-making (Lambert Reference Lambert2024). Whereas globalization theory has developed the idea of the Roman Empire as an increasingly interconnected and interdependent world of localities, people and goods (Pitts & Versluys Reference Pitts, Versluys, Pitts and Versluys2015) which produces a homogenized and homogenizing imperial material culture (Pitts Reference Pitts2019; Pitts & Versluys, Reference Pitts and Versluys2021), ideas of creolization, performativity and discrepant identities have sought to reconceptualize the role of subaltern agents under imperial rule, aiming to give them agentic capacities and not just making them dupes of imperial ideologies (Mattingly Reference Mattingly2004; Revell Reference Revell2016; Webster Reference Webster2001).

Globalization theory allows us to understand the imperial structure as an integrated whole, but at the cost of the consideration of the insidious nature of imperial power (Fernández-Götz et al. Reference Fernández-Götz, Maschek and Roymans2020), the active reorganization of the colonial landscape as part of imperial conquest and rule and an explicit disavowal of the contextualization of object relations within interpersonal social systems (Pitts Reference Pitts2019, 87–92). To restate the classic point: material relations are expressions of underlying social relations.

Creolization, discrepant identities and other postcolonial-influenced theories of identities allowed us to place subaltern peoples back into the process of history, but at the cost of situating agency within broader social structures (Millett Reference Millett2025, xxxiii). What this does is present agency and structure as independent concepts that are externally related (this is the standard image of the duality of structure and agency), instead of properly situating agentic capabilities within the structure; the ability to act only makes sense in the structure it is embedded within.

Pragmatic agency allows us to consider the historicity of action and identity-formation under imperial rule, but also, and against modern theoretical strands that would seek to jettison agency in favour of other theoretical concepts, it breathes new life into the concept (Gardner Reference Gardner2021). Pragmatic colonialism helps theorize the emergence of the imperial political economy that allows for a materialist analysis of ritual and cosmological activities to be undertaken. The key thing to note is that these processes are embedded within the imperial political economy of Roman Britain, as both process and structure: it emerges out of the web of social relations that make it up, while also structuring them at the same time (Hamilton & Shin Reference Hamilton and Shin2015).

The Later Iron Age and Roman political economies of southern Roman Britain

I explore the LIA and Roman political economies and historical contexts of southeastern and south-central Britain, corresponding to the modern county regions of Kent, East and West Sussex, Surrey, Greater London and Hampshire (Fig. 1), roughly, as more in-depth studies of the archaeology of these regions exist in the literature, alongside an expansive archaeological record (Champion Reference Champion, Moore, Revell and Millet2016; Fig. 2). By focusing on the LIA–Roman transition, we can investigate the specificities of imperial rule that caused the pattern of relations observed in the archaeological record.

Figure 1.

England with the boundaries of the modern counties. (Image retrieved via Wikimedia Creative Commons License ‘English ceremonial counties 1998 (named)’ by English_ceremonial_counties_2010.svg. Nilfanion derivative work: Dr Gred is licensed under CC BY-SA. Image non-adapted.)

Figure 2.

Selected sites in Roman England and Wales as recorded by RRSP. The fertile central and lowland regions are, unsurprisingly, the most populated, c. 43–410 ce. (Data: derived from Ordnance Survey Data © Crown Copyright and Database Right 2014.)

Before beginning this analysis, a short description of the concept ‘political economy’, because there are a number of definitions and uses. I will use the definition set out by Marx, and I will provide the full quotation because it sums up my own approach to analysing and understanding political economies:

Material stuff is produced by the social relations that organize labour, such that in analysing material relations, we are also analysing social relations. How people invariably organize themselves into social structures is conditioned by their productive and reproductive forces: the material stuff that people make, consume, discard and remake, consume and discard (McGuire Reference McGuire1992). This not only presents us with a historically specific analysis of social and material relations; it also gives us a materialist foundation for analysing cosmological belief systems.

The data that form my detailed empirical analyses for Roman Britain are taken from the Roman Rural Settlement Project (henceforth RRSP; Brindle et al. Reference Brindle, Smith, Allen and Fulford2016). The RRSP is a database for rural settlements in Roman England and Wales which collects information on sites and their associated publications, including monographs, articles and reports, with the aim of exploring the value of commercial archaeology (Fulford & Holbrook Reference Fulford and Holbrook2011).

Settlement patterns in the MIA-LIA were, despite the emergence of the oppidum as the site of exchange and consumption of trans-continental goods and other large-scale activities associated with emergent political kingdoms, predominately roundhouses and other small settlements usually housing small kinship groups. There have been debates about whether LIA societies were hierarchical, egalitarian or heterarchies, but the important thing to note is that there was short- to long-term exchange network interdependence not integrated within any formal or informal structure, and this was in sharp contrast to the later emergence of the imperial political economy (Hill Reference Hill, Moore and Armada2011), as well as having the capacity to shift between these different political modes, a choice no longer available to them once the territory had been annexed (Wengrow & Graeber Reference Wengrow and Graeber2015).

The MIA–LIA period (roughly from bce 400 to 43 ce, ending with the conquest of Britain by Claudius; the LIA itself ranges across bce 100–43 ce) was one of substantial reorganization of the landscape. Emergent settlement patterns, changes in pottery production and advances in agricultural activities generated the conditions that would make Roman colonization of the period possible (Cunliffe Reference Cunliffe2005). As will be explored, there are differential regional patterns during this period, and this would have had a later impact on Roman colonialism in the region (Fig. 3).

Figure 3.

Map of MIA–LIA settlement in relation to Roman urban centres. From right to left: Thanet, Bourne Park, Bigberry, Homestall Wood, Quarry Wood Camp, Springhead, Hayling Island, St Catherine’s Hill, Oram’s Arbour, Sudden Farm, Hengistbury Head. (Data: derived from Ordnance Survey Data © Crown Copyright and Database Right 2014.)

Unlike other regions in Britain, southeast Britain had no significant settlement patterns dated to the MIA, and pre-Roman activity in the region relates principally to the LIA period. The evidence suggests a large expansion in settlement and industrial activity, indicating substantial reorganization of social relations in the region during the LIA period (Champion Reference Champion and Williams2007, 119–20; Hill Reference Hill, Haselgrove and Moore2007a, 24). The emergence of hillforts such as Bigberry, a potential precursor to the civilian capital of Canterbury in the later first–second centuries ce (Williams Reference Williams2003, 222), where a 6 km trackway acted to link up disparate settlements, as well as the existence of ritual metalworking, demonstrated increased activity during the LIA period (Ashbee Reference Ashbee2005, 160–62; Millett Reference Millett and Williams2007, 158). Other sites such as Homestall Wood and another potential hillfort at Bourne Park suggest that the Canterbury hinterland was a socially vibrant region (Wallace & Mullen Reference Wallace and Mullen2019, 79–80). There was a Roman road at the oppidum at Quarry Wood Camp, which formed an important part of the iron trade to the Weald and continued to have some uses into the Roman period (Howell Reference Howell2014, 63–4). The LIA ritual site at Springhead later became a major roadside settlement (Andrews et al. Reference Andrews, Biddulph, Hardy and Brown2011, 190–92) and villas such as Keston, Minster-in-Thanet and Fishbourne all have evidence suggesting pre-Roman settlement activity (Booth Reference Booth and Bird2017).

Ritual activity in southeast Britain was varied and included the ‘Aylesford-Swarling’ tradition, wherein cremated remains were interred in decorated wooden pots with imported amphorae and pottery which were likely influenced by Gallic burial traditions, and dating at some burials places these types of burial to the first century bce, with the possibility of some occurring earlier in the second century bce (Champion Reference Champion, Moore, Revell and Millet2016, 160–62). There were other local traditions of inhumation in East Kent and Thanet (Parfitt Reference Parfitt1998) and Brisley Farm (Stevenson Reference Stevenson2013, 157–62), as well as the LIA ritual complex at Springhead. Many of these different ritual activities are evidence of the complexity of the sacred/profane dynamic (Champion Reference Champion, Moore, Revell and Millet2016).

The LIA pattern in the southern-central region is different compared to the southeast, with the major difference being a longer-term settlement pattern from the MIA and earlier (Sharples Reference Sharples2010). Changes in much of southern Britain reflected long-term trends, and patterns observed during the Roman period were dependent on activity in the region stretching back to the MIA–LIA (Sharples Reference Sharples2010, 8–14). Also important in this context are the regional trade connections with Gaul during this time, and the rise of major sites of exchange and consumption such as at Hengistbury Head (Cunliffe Reference Cunliffe2005, 170–80). Hillforts such as St Catherine’s Hill and Oram’s Arbour show settlement activity in the MIA, although the former appears to have been abandoned by the first century bce (Cunliffe Reference Cunliffe2005, 104). The Sudden Farm enclosure appears to have been settled consistently from the seventh century bce to the fourth century ce with a few periods of abandonment (Stevens et al. Reference Stevens, Lightfoot, Hamilton, Cunliffe and Hedges2013, 257). Like Canterbury, the urban town at Silchester was the site of a major LIA oppidum and likely served as the seat of the Atrebates Kingdom (Creighton Reference Creighton2006, 20–24).

Ritual activity in the LIA southern-central region is dominated by the Hayling Island complex. The LIA shrine had multiple phases of construction and depositional evidence included sacrificed pigs and sheep, human bones, horse harnesses, currency bars, coinage from Britain, Gaul and Republican Rome, which were deposited in specific zones within the shrine (King & Soffe Reference King, Soffe and Rudling2008, 139–40).

Changes in agricultural activities in the LIA included increased centralization of crops in storage pits (Van der Veen & Jones Reference Van der Veen and Jones2006, 226) and new techniques such as field drainage, free-threshing wheat and mono-cropping continued into the Roman period (Van der Veen & O’Connor Reference Van der Veen, O’Connor and Bayley1998) alongside increases in the scale of production and the introduction of new crops such as fruits, herbs, vegetables, malted beverages and hay (Van der Veen Reference Van der Veen, Millett, Revell and Moore2016). Alongside this were changes in animal rearing and dairy production (Allen & Lodwick Reference Allen, Lodwick, Allen, Lodwick, Brindle, Fulford and Smith2017, 144), the use of corn-dryers and grain mills, innovations in wheel-thrown pottery (Shaffrey Reference Shaffrey2015). Many of the road networks established during the Roman period also had LIA antecedents (Allen 2016), and LIA cross-Channel exchange introduced Roman-style material goods to the territory (Champion Reference Champion, Moore, Revell and Millet2016, 158–9), a process turbocharged by the annexation, increased influx of goods as well as increase in local production in Roman Britain (Pitts Reference Pitts2019, 207–9). Britain was now plugged into this global network of goods circulating through a multitude of trade networks of production and consumption (Versluys Reference Versluys2014). The increases in technological production allowed for the conditions that would make colonization of Britain a project that could be pursued, alongside the existence of cross-Channel exchange networks and increasing settlement density (Marx Reference Marx1976).

With this in mind, we can now explore the development of the imperial administration in southern Roman Britain. The construction of the colonial administration had some key features. This involved the institution of the road network, the creation of military forts, the development of urban centres, the establishment of a legal system, which created a colonial administration tasked principally with wealth extraction through land redistribution and regressive taxation (Birley Reference Birley2005, 9–14; Mattingly Reference Mattingly2011, 132–9; Fig. 4). The establishment of a road network not only partitioned the region, but also served to facilitate trade relations across the region and connect up major urban and military sites with nucleated rural settlements, thereby creating secondary markets at nucleated rural settlements in the region (Brindle Reference Brindle, Allen, Lodwick, Brindle, Fulford and Smith2017, 277–80), and allowed for administrators, tax officials and military officials attached to the administration to access large swathes of the rural landscape more easily to extract taxes and resources from the regions. The development of urban centres recreated the means by which imperial administrators could create the politico-legal bureaucracy for administering the region, as well as replicating the social conditions they were most comfortable within. London developed as a major port town and military-administrative centre, which, alongside Canterbury, acted as a major centre of consumption and distribution in the wider region (Orengo & Livarda Reference Orengo and Livarda2016). The urban centres at Silchester and Winchester in Hampshire were settled in the political territory of the Atrebates Kingdom, whose seat of power was likely the oppidum that preceded the construction of Silchester (Fulford Reference Fulford, Augenti and Christie2012).

Figure 4.

Major urban sites discussed in text and other selected sites from Roman England and Wales, c. 43–410 ce. (Data: derived from Ordnance Survey Data © Crown Copyright and Database Right 2014.)

There are several similarities between the southeast and southern-central regions, including the nature of agricultural activities, the presence of the road network, the number of settlements with LIA antecedents (Allen Reference Allen, Smith, Allen, Brindle and Fulford2016, 82) and the development of the imperial administrative bureaucracy, but there are also major differences. As noted already, the southern-central region had a longer-lived settlement chronology than the southeast (Creighton Reference Creighton2006, 20–24). By the middle of the third century ce, settlement patterns and activity in the southeast had contracted significantly, with the majority of farmsteads seeing either a decline in material evidence or the wholesale abandonment of sites (Allen Reference Allen, Smith, Allen, Brindle and Fulford2016, 82; Table 1a, 1b). There are several reasons for this, but the important thing to note is while this pattern occurred in the southeast, there was a very different pattern occurring in the southern-central region, which had a longer-lived settlement pattern, including a major reorganization of both the urban centre at Silchester (Fulford & Timby Reference Fulford and Timby2000) and the wider rural landscape (Allen Reference Allen, Smith, Allen, Brindle and Fulford2016, 136). It is likely that the role of the imperial administration in both regions contributed significantly to the different trajectories that exist, and this was impacted by the LIA histories. The likeliest explanation is that the longer-lived settlement patterns in the central-southern region equipped those communities with a greater resilience to adapt to colonial annexation and extraction compared to the southeast.

Table 1a.

Total number of settlement types by time period in Roman Kent. Where settlements overlap time periods, they will be included in both time periods. The division is based on the RRSP classification for major site types; cf. Table 4. Where sites are in multiple classifications, they will be added to all classifications that are applied to them, except where villas are classified as villas/farms, then these sites will be counted as villas. Where sites are farms with other associated features such as field systems and industrial centres (counted under ‘other’), they shall be classed as farms. Sites solely identified as ritual, industrial or rural landscape will be placed in ‘other’. This cross-classification explains the high numbers of farmsteads in the fourth century ce, because many of them were also villas, and villa construction was on the rise in the third–fourth centuries ce.

Table 1b.

Numbers of settlements in the Hampshire region by settlement category, following RRSP classifications. The increase noted in the fourth century ce reflects a real pattern in the data, in stark contrast to Roman Kent.

Political economy to cosmology to political economy

The previous section provided a materialist analysis of the political economies of LIA and Southern Roman Britain. The question now is to situate ritual activities within this materialist analysis. What is the relationship between rituality and the imperial political economy? Furthermore, how do cosmologies arise from ritual activities?

With that in mind, let us define what we mean by cosmology. As a science, it can be understood narrowly as the study of the universe as an ordered system, the laws that govern it and that deal with the structure and evolution of the universe (Hawley & Holcombe Reference Hawley and Holcomb2005, 4–5). The philosophy of cosmology seeks to understand the nature of the cosmos as a physical entity related to geometry, matter, fields, dark matter and energy, physical laws and probability (Ellis Reference Ellis2014, 5–17) and metaphysical questions concerning determinism, causality, intelligent life and complexity (Ellis Reference Ellis2014, 18–22).

As an anthropological theory, it relates to the idea that an understanding of the universe is an integral part of human cultures and human societies, the notion that a community’s world view and belief system is based on a conceptual understanding of the structure and ordering of the universe (Darvill Reference Darvill2008). This is especially prevalent when thinking about the cosmologies of indigenous communities in colonial metropoles and peripheries, who had and continue to have original and unique understanding of how their world functions, and the political consequences of taking these cosmologies seriously in relation to the continued fight for the freedoms of indigenous people to live life on their terms, free from the continual colonial legacies of the modern world (Sundstrom & DeBoer Reference Sundstrom and DeBoer2012). Cosmology encapsulates the empirical, the normative and the abstract in terms not only of how historic communities constructed and understood their cosmologies, but also how we as archaeologists understand them (Bhaskar Reference Bhaskar2008).

One key idea to grasp is that cosmologies provide a narrative of a community’s role in the universe, in order to guide actions (Bennett & Royle Reference Bennett and Royle2004, 54). Cosmologies naturalize and provide reasons and justifications for the way the world is ordered and how it should be ordered. Cosmologies thus infuse the beliefs, institutions and relations within a community by providing meaning that anchors individuals and their actions within a worldview (Allan Reference Allan2018, 12). This also means that cosmologies are value-laden, insofar as the values, ethics and material activities are all interconnected and they give reasons and motivations to act in certain ways and to maintain social relations that are conducive to the overall cosmology, thereby tying these relations, and other such ‘mundane’ activities, within the wider cosmology (Bell Reference Bell1997, 122). This thereby allows the communities in question to comprehend the logic of the structure as a conceptual whole, regardless of the empirical accuracy of such comprehension.

As already noted, cosmologies emerge from the political economy to give unity to social and material relations with the aim of making sense of these relations. In other words, cosmologies are built up from the production and reproduction of social life, while also giving them meaning and structure (Boivin Reference Boivin2009, 273–7). Cosmologies structure political economies; the whole point of the cosmological structures is to give meaning to social relations. What is at stake during ritual activities is what cosmological structure is generated and what narrative is told, in order that it can act as a systematic belief system to maintain the integrity of the community, and where consent for the cosmological system is tacitly provided through participation in ritual activities and the political economy at large (Gramsci Reference Gramsci1971, 428–38). Cosmologies are adaptive systems that allow for a sense of collective cohesion in the face of dramatic changes, such as those brought about by the Romano-British imperial annexation. Cosmological belief systems are therefore abstractions of these concrete processes, acting as a narration that provides a complete explanation of the totality of social relations (Marx Reference Marx1976, 163–4). It is helpful to think of cosmologies as structures of rationalization (Chibber Reference Chibber2022, 111–14) that seek ultimately to make sense of the world, that seek to explain the world and to offer prescriptions on how to navigate the world (Peirce Reference Peirce1998, 346).

Rituality, pragmatic agency and cosmologies in southern Roman Britain

This section will explore the types of ritual activity present during Romano-British imperial rule. Table 2 details name and locational data for major sites, site categorization based on the RRSP’s own model and key publications. Figure 5 gives the location of major sites in southern Roman Britain.

Table 2.

Tabulated information of selected key sites discussed in text, their regional location, their morphological categories as defined by the RRSP and key publications associated with each site. As noted by Millett (Reference Millett2025, xxxvi), such morphological categories can tend towards homogeneity, but situating them within the imperial political economy of southern Roman Britain can inculcate us from this tendency. Note: RC = Romano-Celtic.

Figure 5.

Discussed sites collated in Table 2 as related to the road network and major urban centres in the region. From right to left: Monkton, Ickham, Westhawk Farm, Ospringe, Springhead, Dartford, Lullingstone, Keston, Odiham, Hayling Island, Thruxton, Alton, c. first–fourth centuries ce. (Data: derived from Ordnance Survey Data © Crown Copyright and Database Right 2014.)

The reason for focusing on ritual activities is that ritual spaces are the arena wherein the logic of a community is made most apparent, where the relationship between individuals and community, between individuals and the cosmos and between communities and the cosmos is made most explicit (Durkheim Reference Durkheim1995, 33–5). Most of the population lived in rural settlements away from the urban and military sections (some estimates put it at around 90 per cent: Mattingly Reference Mattingly2006, 356) and research programmes have long emphasized urban, military and villa architecture in imperial contexts in isolation from the wider imperial political economy (Mattingly Reference Mattingly2011). The focus on rural communities is not only a redress to this imbalance but also a recognition that one cannot talk about ‘Roman Britain’ in any meaningful sense without investigating the other 90 per cent.

The empirical data will refer to domestic ritual activities, or ritual activities associated with domestic dwellings, in part because there is a marked increase in this type of ritual activity in Roman Britain, with around 50 per cent of recorded rural settlements having some evidence of internal domestic shrines or temples (Smith Reference Smith, Smith, Allen, Brindle, Fulford, Lodwick and Rohnbogner2018, 136, 142–61), and specialized constructed sacred spaces only became a common feature after the conquest (Smith Reference Smith2001, 67), suggesting that the development of these spaces was a causal consequence of the emergence of the imperial political economy, and that constructed sacred spaces within domestic sites was a common response across the region, suggesting that this was a key pragmatic response to imperial annexation.

The RRSP defines ritual activities in three broad categories: constructed sacred spaces, religious objects (Whitehouse Reference Whitehouse and Wilkins1996) and structured depositions (Garrow Reference Garrow2012). Our usual concern with ritual activities concerns whether depositions are deliberate, or ‘rubbish’, and how to infer meaning from them (Hill Reference Hill2007b; Whitehouse Reference Whitehouse and Wilkins1996). Our case studies pertain to deliberately constructed sacred spaces within domestic sites, and this deliberate construction is evidence that the communities were marking out such spaces for the purposes of conducting ritual activities, as a direct consequence of imperial annexation. Such communities would not have understood their ritual activities through the sacred/profane dichotomy, but as adaptive responses to the imperial political economy.

Rituality, pragmatic agency and cosmologies in Roman Kent

We will principally consider two types of settlements, villas and roadside settlements, because much of the ritual activity we will investigate is associated with these site types (Booth Reference Booth, Booth, Champion and Foreman2011, 246; Smith Reference Smith, Smith, Allen, Brindle, Fulford, Lodwick and Rohnbogner2018, 147, 182; Table 3). For clarification purposes, Table 4 details the RRSP’s classification system. As noted, the most common ritual activity in the region is demarcated sacred spaces within domestic dwellings, and while funerary sites were located in the wider landscape based on small finds and enclosures, there was not enough evidence to draw meaningful interpretations; as such we will restrict ourselves to those sites with substantial evidence (Smith Reference Smith, Smith, Allen, Brindle, Fulford, Lodwick and Rohnbogner2018, 168–9).

Table 3.

Tabulated data showing split between different religious/ritual sites at domestic dwellings. The ‘others’ category refers to a range of ritual activities, not all of which are associated with domestic dwellings, and where material evidence is scant. In Hampshire 20 per cent of all recorded sites were funerary sites are associated with domestic dwellings (with 0 per cent being sacred sites); and in Kent, 26 per cent of all ritual sites were sacred spaces associated with domestic dwellings.

Table 4.

A classification of site types in the RRSP database (Fulford & Brindle Reference Fulford, Brindle, Smith, Allen, Brindle and Fulford2016, 10). These classifications are not mutually exclusive, and sites regularly cross multiple major and minor site types.

For the most part, there is a key split in the types of ritual activity occurring within villas and roadside settlements, although there is of course a fuzziness in our categories, and we should not draw hard and fast boundaries between such differences (Brück Reference Brück1999, 314). One interesting fact that does emerge from all these sites is that the ritual spaces appear to have been constructed alongside the domestic dwelling, indicating that the development of sacred spaces was not merely ad hoc, but was likely a planned feature. Furthermore, we can sub-divide ritual activities in the region between those sites in which built shrines were the key ritual architecture and other sites which depict more complex ritual activities.

The villa sites at both Darenth and Thurnham had shrines that were likely dated to the initial period of construction. Furthermore, both sites lacked evidence of structured depositions, and this was starker in the case of Darenth, where the shrine enclosed a well found with no depositions, and artefacts were regularly deposited within wells (Black Reference Black1987, 1–8). Pottery and brooches found within the shrine at Thurnham dated the construction of the shrine to the same period as the first period of the proto-villa, around 70 ce (Booth & Lawrence Reference Booth and Lawrence2006, 42–4). The pottery and brooches were detritus and not structured depositions (Garrow Reference Garrow2012), and as such we can group Thurnham along with those other sites for which ritual and sacred space was not accompanied by ritual depositions. There is also a range of roadside settlements including the sites at Westhawk Farm, Monkton, Ickham and Each End, wherein the principal ritual activity was the presence of either masonry or timber-built structures, all of which have very little to no evidence of ritual or structured depositions. The shrine at Westhawk Farm was most likely planned and built as a single entity based on the uniform shape of the structure, suggesting that the location and the decision to build it were deliberate (Booth Reference Booth2001, 12), and while the evidence is not clear for the other sites, it would not be a stretch to infer that the shrines were built in conjunction with the domestic site, likely suggesting that whatever activities were being conducted within these spaces, they relied not on the ritual deposition of artefacts or cultic activity (Booth et al. Reference Booth, Bingham and Lawrence2008, 379) but on the conviviality generated by the social interactions of the communities within these spaces and the repeated, creative and structured interactions that these spaces generated (Berggren & Stutz Reference Berggren and Stutz2010, 175–9). The memories, emotions and sensorial responses associated with the act being within these spaces were experienced collectively (Antczak Reference Antczak2017, 147–8), fully embedded in the social system the community was part of.

We can go further and argue that such social interactions within these spaces was the key ritual activity undertaken within the shrines; acts of collective and mutual memory-making and remembrance (Van Dyke Reference Van Dyke2019). At Westhawk Farm, the shrine did not contain evidence indicating ritual deposition (Booth Reference Booth2001, 17–18). As noted earlier, depositions in wells and watery contexts were key indicators of ritual activity, and such activities continued in some form into the early Roman period (Champion Reference Champion, Moore, Revell and Millet2016). At Westhawk Farm, a nearby well had deposits including waterlogged wood from an oak, pottery dated from the second century ce and 227 coins, again dated to the second century ce, were found in wells (Booth Reference Booth2001, 11–17; Garrow Reference Garrow2006). The most important thing here is that these depositions were found away from the shrine, and not within it. This demonstrates two things: first that, and as noted above, the nature of ritual activity within the shrine was oriented towards ends not associated with ritual depositions; and whatever cosmological belief system was constructed within the shrine was flexible enough to encompass a variety of different behaviours beyond the formal ritualized space, allowing for individuals to act and interpret such belief in personal terms (Dornan Reference Dornan2004). As for claims that the shrine may not be ritual in nature, the fact that it was built at the same time as the domestic settlement, and that the shrine was oriented south-east similar to IA roundhouses, indicates otherwise (Oswald Reference Oswald, Gwilt and Haselgrove1997, 92), and also points to the role of social memory among the community at Westhawk (Van Dyke Reference Van Dyke2019).

The exception to the rule is Springhead, a major LIA ritual site that later became a major roadside settlement in the early Roman period and was the only roadside settlement with multiple Romano-Celtic temples which contained structured depositions. LIA activities were oriented around the nearby river Ebbsfleet, and included both cremation burials and deposition of coins and metalwork (Andrews et al. Reference Andrews, Biddulph, Hardy and Brown2011, 190–92). Depositions continued into the early Roman period with the construction of five temples (three were Romano-Celtic) in the early second century ce. Votive offerings of jewellery including brooches and hairpins were carefully chosen for their shape, colour, motifs and materials, and were just as important as the type of offering being given (Andrews et al. Reference Andrews, Biddulph, Hardy and Brown2011, 94–8; Puttock Reference Puttock2002, 115). Other structured depositions included animal bones (sometimes only a skull, which was likely borrowed from Romano-Gallic burial traditions: Woodward Reference Woodward1992), and infant burials interred with a headless bird (Alaimo Reference Alaimo2016, 160–62). It is more than likely that Springhead’s LIA ritual activities continued into the Roman period, and the emphasis on ritual depositions in the LIA explains why they continued into the early Roman period, in contrast to the other roadside settlements, which do not have this same history.

There are also a range of sites with ritual activities related to burials/cemeteries. At the villa site of Lullingstone, the Romano-Celtic temple/mausoleum contained inhumations for two adults (probable male and female in their mid 20s) buried under the temple in a chamber. The burials were dated to ce 300 based on grave goods such as silver spoons, copper alloy, ceramic flagons, glass vessels and whetstones (Meates Reference Meates1979, 125), and a signet ring was found belonging to the governor of Britain and later emperor Publius Helvius Pertinax; and villas were expansive architectural structures for the cultivation of local elites (Taylor Reference Taylor2013). An infant burial associated with the fourth-century ce church was interred with wheat grains, a small ceramic vessel, a decorated copper-alloy vessel and five coins dated to the later fourth century ce. There is also evidence that a horse’s head was laid above ground in front of the temple structure alongside cattle skulls from the late third century ce. Further animal skeletons included cats and dogs (Meates Reference Meates1979, 125–8). The roadside settlement at Ospringe has a cemetery containing mostly cremation burials, but there were some inhumations; many of these burials do not survive in any meaningful way, making inferences about social relations difficult (Whiting et al. Reference Whiting, Hawley and May1931). Excavations at another roadside settlement at Dartford uncovered two stone coffins alongside a number of cremation burials. Pottery and iron deposits were found interred with the bones, which allowed for the cemetery to be dated to the later Roman period, in contrast to many other cemeteries, which are of an earlier date (Herbert Reference Herbert2011, 105–8).

The ritual activities described above were all attached to domestic settlements, and this should be seen as a pragmatic response to the annexation of the land and the inability to establish relationships across the region, because the establishment of the colonial bureaucracy and the associated institutions reshaped how these communities interacted with the wider landscape which acted as a control mechanism. (Deleuze Reference Deleuze1992). This would have had profound consequences on their cosmological belief systems having to rationalize their retreat from the landscape and back into clearly demarcated domestic sites. The establishment of the imperial political economy necessitated new ways of living, with the structure mediating and organizing social relations. The differences noted between the different types of ritual activities should be considered as localized instantiation of this general pattern of ritual activities as embedded within the overall settlements and their community relations. The diversity in ritual activities is matched by the diversity of cosmological belief systems that each community would have constructed, expressed and acted within. Differences between roadside settlements and villas should be approached with regard to the principal roles they played in the region, the former as rural markets of exchange and consumption and the latter as estates for managing land, as well as sites for the production of particular types of elite Romano-British identities related to either local elites or attachés of the imperial administration (Smith Reference Smith, Smith, Allen, Brindle and Fulford2016). Further differences should be seen as unique localizations of ritual activities and pragmatic agencies.

Ontogenesis and the making of the cosmological landscape

As noted in the introduction, it is only after the empirical analysis has been done that we can fully detail the process of ontogenesis. To do this, we can explore Bhaskar’s distinction between the empirical, the actual and the real in his account of causal mechanisms (Bhaskar Reference Bhaskar2008, 47). For Bhaskar, the empirical constitutes the domain of experiences, the actual constitutes the domain of events and the real constitutes the domain of mechanisms. The empirical refers to the observation of actions in the world as individual experiences, the actual refers to the integration of these instances together into a coherent picture and a recognition that they form part of a collected whole. The real refers to accounting for the collective whole as a mechanism, in exploring how the collective whole acts as a causal network which produces the phenomena under observation (Illari & Russo Reference Illari and Russo2014, 123). Mechanisms help us understand how a given result is brought about by the precise configuration of specific relations in the system (Glennan Reference Glennan2002, 344), as well as accounting for change systematically (Crellin Reference Crellin2020, 12-17).

I am going to modify this image to explain the process by which cosmological belief systems emerge, beginning with the empirical, which includes the daily interactions with the institutions and networks of the colonial administration through exchange relations, the manufacturing of resources and goods and the material relations through which communities maintain their daily lives. The actual refers to the recognition that these interactions form part of a complex system of relations that are interconnected to the wider colonial administration, and that they constitute a totality (Lukács Reference Lukács1971). The real is the process by which these material and social relations become naturalized, as cosmological belief systems developed as a consequence of the pragmatism displayed by local communities are embedded into the fabric of how they understand and interpret the world. This creates a process whereby cosmological structures emerge from the political economy while ordering these relations into a comprehensible whole. It is this process by which historical processes become naturalized, and the way in which different actors seek to naturalize historically contingent relations that I call ontogenesis. Ontogenesis is a genetic process that creates new forms of understanding and interacting with the world within and through the conceptualization of social relations (Lévi-Strauss Reference Lévi-Strauss1966; Reed Reference Reed2019). Normatively constructed relations about how communities should function become naturalized into a descriptive understanding about how the world actually works, and it is this naturalization that serves as a narrative structure that allows for communities to make sense of the dramatic changes brought about by Romano-British imperialism. This can be usefully contrasted with Ingold’s notion of ontogenesis as the generation of being, as opposed to its essence, one in which being is constantly renewed and generated differentially within what he calls a ‘one world anthropology’ (Ingold Reference Ingold2018, 166–8). Ingold takes his own idea of ontogenesis from the French thinker Simondon, for whom ontogenesis was akin to individuation, of the immanent differential process by which life is renewed within itself, constantly marking itself through this internal act of differential renewal (Deleuze Reference Deleuze2014, 260; Simondon Reference Simondon, Crary and Kwinter1993, 300–305). Put more prosaically, it is about the differential and individual process of becoming, as opposed to the static and essentializing ‘being’ (Harris & Cipolla Reference Harris and Cipolla2017, 138–46). Despite opposing his idea of ontogenesis to ontologies (and multiple ontologies in particular), Ingold’s theory is still metaphysical in its attempt to account for the nature of reality. On the other hand, my own use of ontogenesis as a process is non-metaphysical. It is not aimed at developing any idea of how reality in itself works, but names the process by which constructed social relations are naturalized and how cosmological structures emerge that condition and determine the range of ritual activities and belief systems that communities living within these relations can, and do, express.

Conclusion: making communities

This paper has sought to justify a single premise: that understanding and exploring the cosmologies and cosmological structures of a given community is a necessary part of archaeological research. To accomplish this, however, we need to understand that how communities conceptualize the world is conditioned by how they experience it. This paper has sought to establish an inductive process between local communities, the imperial political economy and ritual activities in order to demonstrate not only how they are connected, but to demonstrate that a rigorous understanding of the empirical evidence is a prerequisite for careful analysis and speculation of the cosmological structures of a community. We may not be able to provide an individualized account, but we can certainly delineate a narrow range of cosmologies that communities could take up, in part because the social relations they are embedded within would restrict the range of belief systems they could imagine. Inferring the cosmologies of ancient communities is believed by most archaeologists to be a fool’s errand, available only to those working either with ancient texts or with descendant communities, usually citing Hawkes’ (Reference Hawkes1954) ladder of inference, in which religion is the hardest aspect of society that one can infer. We can flip this by arguing that it is only once a fully materialist analysis of communities has been given that we can provide a route through to analysing cosmologies (Engels Reference Engels1970). This paper has been one attempt at providing this materialist analysis, demonstrating the viability, if not the finality, of this project.