Introduction

The interpretation of religious sites in Britannia is often undertaken in relative isolation from their secular surroundings. Roman archaeology as a discipline has focused predominantly on structural morphology and identification of deities, or ritualized deposition of material culture when interpreting Romano-Celtic temples/shrines and the practices undertaken there. Ritual practices such as sacrifice and structured deposition have been discussed, but almost always interpreted in relation to the sacred site rather than placed in wider contexts of practice; that is, how those using the sacred site lived their lives, including in relation to animals (Lewis 1966; Wilson 1975; Smith 2001; Hingley 2006). Archaeological categorization of ritual sites by form, deity or (occasionally) practice thereby separates them from the wider populations of people, animals, objects and places from which they originate. If wider activities in the landscape are mentioned, they are seen as being de facto shaped and controlled by the temple (Smith 2018: 173–174).

The widespread deployment of the concept of the ‘sacred landscape’ (e.g. Irvine Steinsapir 2005; Stek 2009; Newson 2019) has helped entrench this dichotomy; whilst early proponents of the idea such as Alcock (1993) were careful to ensure that their discussions acknowledged the interplay between different facets of practice in the landscape, this subtlety has bled away in light of recent enthusiasm for illuminating memory, aesthetics and belief above secular occupations and relations (e.g. Ferris 2021). More traditional approaches to religious space in Britannia also remain common, where even if their broader secular contexts are considered, the nature of interplay between religious practice and everyday life remains less explored (Robinson 2001; Jackson and Burleigh 2018). This article aims to show how a major aspect of ritual practice at temples and shrines in Roman Britain — interaction with animals — can be reconnected with the wider secular contexts from which animals and people derive.

Interaction with animals is not only a prism through which human society can be better understood (Poole 2015), but a fundamental constitutive part of human worldviews and belief systems, landscapes and timescapes (Chadwick 2016). In pre-industrial societies, the significant majority of the human population engaged in agriculture, in almost every case necessitating daily and lifelong interactions with animals (Fulford 2004). Archaeological interpretation separates ritual activity from this deep lived context, even where it involves the sacrifice of animal lives, thereby divorcing ritual from its secular context and relegating animals to the status of material objects. In this paper, through an examination of the evidence at the South Wiltshire Temple, we argue that taking account of animal lives and social associations beyond the immediate ritual context of their deposition provides an important new method of interpreting sacrificial activity, and the social context and meaning of ritual practice.

Approaching Animals

Much recent theoretical work has sought to situate cultural transformation and expression in Britannia in richer contexts through perspectives derived from theories of materiality and post-humanism (Chadwick 2016; Gardner and Wallace 2020; chapters in Sesvold and Webb 2020). Such archaeological accounts conceptualize places by interrogating the networks and interplay of human and non-human agents in the world, and the experiences, practices, temporalities and symbolisms by which places are mutually constituted (Garland 2016; Wallace and Mullen 2019). Non-human agency and network-theory have provided useful avenues to allow the exploration of the materiality of places, whilst avoiding environmental determinism, although some work has in our view over-emphasized the influence of non-human actors.¹ Considerable theoretical work also remains to be done to reconcile the tensions in such models between the scales and rhythms of lived human experience and those of other agents and processes in the landscape, whether the faster life-rhythms of animals, or slower processes such as soil degradation or climatic change (Walsh 2004; 2013). Additionally, these theories do not currently work well with the challenge of thinking through the scale of the Roman Empire and its influence, and the power of structural inequality across the Roman world at different scales (Fernández-Götz et al. 2020; Gardner 2021).

A practice-based perspective may avoid the theoretical difficulties of posthuman and materiality approaches outlined above, whilst retaining their rich facility for characterizing particular material encounters (Gardner 2021). Scholars such as Giles (2007) and Chadwick (2012; 2016) have emphasized the importance of practice, that is, the ways in which individuals and communities interact with the world — landscape, object, animal and beyond — in their contexts of ‘being in the world’. Landscape, animals and the other involved entities of the world have different forms of interaction with the human actors, but are all active, even if for some this is primarily through the materiality of their inherent properties provoking, affording or suggesting human action.² Where such arguments have been made for prehistory and the Romano-British period they usually emphasize the depth of engagement with place, landscape, the material world, and other living actors including animals (Chadwick 2016; Erskine 2021). Crucially, practice theory affords greater insight into social and longer-term aspects of human material experience, compared to posthuman and materiality theories outlined above. Long-term practices inscribe themselves on the landscape, blending human and non-human agencies and processes.³ Temples, boundaries, hollow-ways, quarries, fields and streetscapes are all constituted by shared practices, mediated by individual agency and non-human agents and processes; humans, animals, landscapes and material constraints create these places (Kalayci and Wainwright 2021). Archaeology, with allied disciplines such as Ancient History and Environmental Science, has the tools to unpick those practices and develop narratives of the people, animals and places with which they are entangled.

Glocalization theory, if allied to theories of practice, may allow an even fuller expression of lifeways, cultural transformation and expression in the Roman period. The essential argument of glocalization is that as the Roman world became increasingly connected, there developed a larger shared material and social assemblage mutually recognized by those living under Rome’s rule (Hitchner 2008; Pitts and Versluys 2014; van Alten 2017). Glocalization theory sees this increased connectivity as the key characteristic of the Roman world, rather than its ‘Roman-ness’ or otherwise. This perspective has been critiqued as anachronistic or entrenching models underlying Romanization (see discussion by van Alten 2017: 143). These critiques neglect that, whilst of course the institutions, impositions and material of Roman imperialism are a major part of what is transmitted by increased connectivity, the colonizing power cannot fully control what else is communicated, or any feedback affecting the nature of those ‘Roman’ practices and structures. Crucially, glocalization theory balances its emphasis on connectivity with recognition of the complex and diverse ways in which increasingly mutually recognized material and social assemblage could be refracted by individual, local, regional or provincial lifeways, and thereby feedback through the system as a whole (van Alten 2017; Díaz-Guardamino et al. 2020).

By considering practices in the wider context provided by glocalization we may therefore be able to begin to develop broader understandings of the ways in which people’s lived experiences changed during the Roman period. This paper attempts to provide a worked example of that theoretical potential, drawing on interaction with animals at the South Wiltshire temple (SWT) and comparable religious sites in the south-west of later Roman Britannia. We will first review the evidence for religion in this period, and the roles of animals within it, to provide appropriate context for the case study.

Religion in Late Roman Britannia

Religious belief in third and fourth century AD Britannia was extraordinarily diverse and changed significantly over this short time (Henig 1995; Watts 1998; Petts 2003; 2016). Roman paganism appears to have been concerned with orthopraxy — completing set rituals and practices in order to ensure an effect — rather more than orthodoxy, which implies uniformity of belief and conformity to certain doctrines (Scheid 2005; Rives 2007: 49–50; Smith 2018: 182). For much of this period, there was relative freedom to believe and worship based on individual inclination, but religious upheavals in the central Empire undoubtedly had significant results in Britannia (Mattingly 2006: 347–348). During the third century AD, paganism, including the Imperial cult, was the dominant mode of religious expression in the Roman Empire, but this label covers diverse religious and ritual practices. Many people in rural society in the Roman north-west provinces continued to worship deities with roots in the Iron Age such as Cernunnos or Toutatis, albeit sometimes syncretized with Mediterranean deities such as Apollo-Cunomaglos or Sulis-Minerva (Henig 1995; Derks 1998). Some deities tended to be venerated by particular professions, such as Mithras’ strong connection with the military (Walsh 2018: 25–30), but much religious or ritual practice was localized (Rives 2007: 54–87). This too fits harmoniously with the glocalization theory outlined in the previous section, which can convincingly explain both the mechanisms of religious change from the pre-Roman period, and the diversity of the resultant tapestry of belief (Peer 2023).

Ritual was, of course, a continuum, encompassing formal religious behaviour through to structured deposition of waste and many forms of practice between (Goodman 2011: 167; Chadwick 2012). Belief in genii loci is attested across the Western Empire, and prehistoric traditions of the deposition of metal and other objects in watery places and other significant landscape locations continued to develop and diversify through access to new media and traditions (Hingley 2006; Bland et al. 2019; Henry et al. 2019).⁴ Nevertheless, Roman deities, and to a lesser extent those syncretized with them, are the deities best attested to by durable material culture such as coinage, mosaics, statuettes, epigraphy and sculpture; the media — and message — of the elite classes. Even within Wiltshire, Apollo, Hercules, Jupiter, Mars, Mercury, Minerva, Silvanus, Venus and Vulcan are all convincingly represented in the archaeological evidence for religious activity (Robinson 2001; Durham 2012).

In the first half of the fourth century AD, cessation of official persecution of Christians and the ensuing rapid, if uneven, increase in Christian influence in Imperial power structures led to greater material expression of Christianity within Britain, including among elites (Petts 2003; 2016; Rees 2020).⁵ Despite this, investment in pagan temples and displays of belief also continued well into the second half of the fourth century AD in Britannia (Woodward and Leach 1993; Esmonde Cleary 2014; Smith 2018: 203–204; Henry et al. 2020). The short-lived pagan revival under Julian (AD 361–363) has been suggested by Moorhead (2001) to be key to understanding a floruit of late Roman pagan shrines and temples in Britannia in the years afterwards, especially in the south-west, but the challenges of linking specific historical events to archaeological chronologies are considerable.⁶ It appears that, unlike in the German provinces (Kousser 2010) or Gaul (Goodman 2011), in Britannia the ‘end of the Gods’ did not take place until the early fifth century (Esmonde Cleary 2014). Overall, it is difficult to align the archaeological evidence of the highly diverse religious and ritual behaviours in late Roman Britain with wider political and religious change in the Empire, though undoubtedly disruptions to wider religious networks and shared vocabularies of religious practice must have caused responses in practices across Britannia. One of the key questions provoked by the glocalization paradigm here must be to explain the slow pace of religious change in mid to late fourth century AD Britannia in comparison to the rest of the Roman West; perhaps this hints at an already fragmenting network of connections to Britannia.

Animals in Religion in Late Roman Britannia

Animals played a key role in pagan religious practices in the Roman world. The most important of such practices included sacrifice, divination, deposition of votive offerings, and feasting (Henig 1995; Rives 2007: 24–27; Gerrard 2009; Maltby 2012; Smith 2018: 182–183). The material remains of many such interactions are accessible through the archaeological record and can therefore be interrogated (Allen 2018: 192–199; Smith 2018: 182–191). It is important to note that the regularity of blood sacrifice reduced in the core of the Empire, especially the East, in the later third and fourth centuries AD (Bradbury 1995), although it is unclear whether there is any corresponding change in sacrificial activities in Britannia. Elsewhere in the Empire, a reduction in blood sacrifice is partly attributed to the particularly vehement rejection of this aspect of pagan religion by Christians (especially Christian Emperors), and partly to the lack of resources on the part of civic and religious institutions and individuals required to sustain earlier volumes of sacrifice (Bradbury 1995). In Britannia, the first of these factors may have been less significant than elsewhere in the Empire due to a (relatively) lower level of Christianity amongst the population (Smith 2018: 204), and the second must have to some extent been mitigated by the evident stability and wealth of lowland elites in the late Roman period alongside Britannia’s apparently smaller volumes of sacrifice in the earlier Roman period. Certainly, there is little evidence of any reduction in the volume of animal sacrifice until the later fourth century AD at the Uley temple complex, which provides what is probably Britannia’s largest thoroughly researched zooarchaeological assemblage from a religious site (Levitan 1993).

King’s (2005) review of animal remains at temples in Britannia demonstrated the selection of animals and considerable evidence for seasonality in slaughter patterns at several major sites (Uley, Harlow, Great Chesterford and Hayling Island; King 2005: 358). King (2005) established five categories of site based on prevalence of animal bone and features within the assemblage.⁷ While this is a useful way to describe various practices, any broad attempt at categorization necessarily focuses on commonalities between assemblages at sacred sites, rather than contextualizing ritual actions in relation to wider landscape practices. To move interpretation beyond both broad categorization and comparison to literary evidence, we suggest considering religious sites and their animals in the context of Roman society. Indeed, King’s discussion (2005: 357–363) demonstrates the diversity of practices represented within each group, and his article concludes that while the new material forms and ritual practices developing in Britannia in the Roman period owe much to Graeco-Roman influences, the specific forms of these interaction with animals and religion more broadly at each site are strongly mediated by pre-existing practices.

Allen’s (2018: 196–199) more recent brief synthesis of faunal remains from shrines and temples contrasts to King’s (2005) more variable findings, indicating that there is a distinct difference between assemblages from temples and those from shrines.⁸ Allen argues that shrines may show a very diverse range of assemblage compositions between species, whereas temple assemblages tend to demonstrate a distinct preference for ovicaprids and/or pigs (predominantly the former) rather than cattle, and chickens are found more commonly at temples than shrines. Greater standardization of sacrifice at temple sites may also be mirrored in the formality and standardization of the architecture, suggesting a more widely shared set of ideas and rituals than the more localized shrine sites.

It is widely recognized today that religion and ritual do not stop at the walls and ditches of Britannia’s temenos enclosures. Ritual practices involving animals in the wider rural landscape are endemic in pre-Roman and Roman Britannia, including the deposition of partial or whole animals (Morris 2011), feasting associated with burial (Orton 2007), deposition of objects in watery places (Fulford 2001; Rogers 2013: 196–199), or the raven ‘familiars’ associated with druids, attested in both ritual deposits and sculpture (Serjeantson and Morris 2011). This article uses a single case study of the South Wiltshire Temple (SWT), with recourse to key comparanda; the general principles are, however, potentially applicable to the interpretation of any ritual activity.

The South Wiltshire Temple

A significant late Roman temple has recently been excavated in southern Wiltshire; the location of this site is known as South Wiltshire to prevent illegal metal detecting (Henry et al. 2020). Here, geophysical survey following numerous and unusual finds by a metal detectorist — for example, mutilated coins pierced by iron studs and miniature iron votives — revealed an extensive Late Iron Age to Late Roman landscape of enclosures and settlement, including ironworking, quarrying and crop processing. Within this wider settled landscape was a large rectangular timber framed building, floored and roofed in local limestone (Figure 1). An ambulatory enclosed a small rectangular central room, with evidence for structures restricting internal views. The central room was initially built around a single post which formed the focus for repeated and dense deposition of artefacts and ecofacts including animal remains. The post was removed sometime in the fourth century AD and additional slabs laid down, with deposition continuing at a high intensity until the late fourth century.

Figure 1

South Wiltshire Temple under excavation; photo is looking west along the south ambulatory, central room and ritual pit to right of photo. Photo: PASt Landscapes Project.

The very large artefactual assemblage of coins, miniature objects, curse tablets and other objects deposited primarily in a central votive pit demonstrates that the building was a pagan temple, and indeed one tablet explicitly names it as such. The tablets also provide name of a god, Bregneus, who appears to have been associated with ironworking and hunting, based on the artefactual and epigraphic evidence from the temple.⁹ The artefactual assemblage and its religious implications, including links to wider landscape practices of ironworking, hunting, and farming, has been discussed at length by Henry et al. (2020). Votive deposits of coins, some mutilated, miniaturized objects, bells, items of personal adornment such as rings, hair pins, and brooches, and curse tablets, were sacrificed at the temple. Some appear to have been fixed to the temple’s superstructure or less permanent elements in order to be displayed, and there are clear chronological episodes of the mutilation of low denomination coinage during the temple’s use. Objects of particular relevance to this paper include votive iron miniatures of spears and a series of representations of animals; these will be discussed in detail later in this paper. There was evidently a public element to the religious practices carried out using artefacts within the temple, perhaps made more dramatic through the burning of mustard seeds (Henry et al. 2020: 18) and the relatively confined space for movement within the building (Henry et al. 2020: 7–9). In essence, Henry et al. (2020) argue that whilst the practices taking place using artefacts link to wider grammars of religious behaviour at sites in Britannia and beyond, the specific choice of artefacts can be related to their resonances in the immediate landscape of the temple.

Building on this, we note that the artefactual assemblage also contains several objects representing animals, including a lion-headed buckle, seven horse-and-rider brooches, and a zoomorphic brooch depicting a deer; as well as two figurines, both depicting large felines. One copper alloy figurine depicts a leopard (Worrell and Pearce 2014: 423–425); the other figurine, of shale, is only partially preserved and shows the front right paw of an unidentified feline on a decorated square base. The former was found by metal detecting in the vicinity of the temple; the latter was recovered from the temple itself.

Animals at the South Wiltshire Temple

The animal bone assemblage comprises c. 700 fragments overall. Although the assemblage is relatively small, all deposits were fully coarse-sieved and intensive flotation sampling was undertaken. The faunal material was recovered from the central temple structure, which was fully excavated, as were parts of the external open area within the temenos, and a section of the temenos ditch itself. As such, we can be confident that the sample is representative of the assemblage available for retrieval from the excavated areas, and that the amount of animal bone deposited in the immediate temple grounds was genuinely relatively limited, especially compared to (for example) the metalwork assemblage deposited at the site. The assemblage of identified bones is, however, larger than at least 4 of the 25 sites reviewed by King (2005); at seven of these sites the total number of bones remained unknown. Regular cleaning and maintenance of the temple, or differential deposition of bone into unexcavated areas of the wider temple site, are possibilities and therefore the scale of animal sacrifice is not directly estimated in this paper.

There are no large mammal remains from Roman phases, with the domestic mammal assemblage composed of sheep and pig, including neonatal suckling pig (Table 1). There is no evidence of differential selection of body parts for deposition from either sheep or pig. Age data is extremely limited from the sheep remains but indicates broadly animals being killed at a range of age stages from 1.5 years onwards. Chicken remains are present in small quantities and with no indication as to the sex of the birds. In addition, there is a distinct smaller component of non-domestic resources, including wild game birds and fish imported from coastal and riverine environments. Some of the considerable oyster assemblage appeared to have been deliberately deposited leaning against the internal superstructure of the building (Henry et al. 2020). Oyster deposition may have been succeeded by the use of blue mussel shells in the central pit. A high number of micromammal remains were present and likely represent a small commensal population of mice and voles, probably indicating the presence of botanical food sources within the temple building.

The most notable feature of the faunal assemblage is a single pair of fragmented antlers excavated within the central room of the temple. The antlers are from a mature adult stag, and are attached to a section of cranium, indicating that the animal died during the autumn or winter, when the antler is fully grown and mineralized, but prior to shedding after the rut (Fletcher 2014). The antlers were found with two curved nails close to each side of the joins to the skull. The nails and the position of the fragments within the collapse layers suggest it is possible that the antlers were mounted within the inner room.

Animals and Landscape Practice at the South Wiltshire Temple

The South Wiltshire Temple lies within a few miles of a major contemporary villa (Roberts 2018), and just east of the wide Blackmore Vale, which is likely to have been forested in the Roman period, perhaps suggesting a suitable landscape context for the hunting of deer. In the medieval period this was the Forest of Blackmore, running north into Gillingham Forest and the Forest of Selwood (Bond 1994: 120; Draper 2006). Landscape practice and settlement in the closer environs of the temple was dominated by extensive mixed agriculture on the chalk downs to the north, and in nearby valleys (Fitzpatrick and Crockett 1998). Sheep are the prevalent species in late Roman zooarchaeological assemblages in the locality (Hammon 2008; Allen 2017). Limestones from several quarries in the Nadder Valley were extracted and worked in this period (Fitzpatrick and Crockett 1998; McBain and Nelson 2003), with the river to the south, the Mendips lead road to the north, and the road between Poole Harbour and Bath to the west providing major transport routes through the region. The Great Ridge, along which the lead road runs, bears several large nucleated settlements of Roman date originating in the Late Iron Age, and there are several Roman religious sites slightly further south of the road (Crawford 1928; Cunnington 1930; Henry et al. 2020; Roberts 2021). Whilst these nucleated settlements and smaller farmsteads dominate the settlement pattern of the area, unusually, there are many more shrines recorded in the vicinity than villas. There has, however, been little modern archaeological investigation at almost all of these sites, and they receive little attention in regional syntheses (e.g. Corney 2001: 5; cf. Draper 2006).

This brief consideration of the site and its immediate and regional contexts makes clear that several strands of practice shared by significant elements of its communities are articulated in the temple’s faunal and artefactual assemblages; this discussion will focus on the former. Undoubtedly, the emphasis on sheep, which comprise c. 40% of identified hand collected bone, reflects the importance of pastoral farming in the area, influenced by the chalk downland landscape of the region which is more suited to grazing sheep than to pasturing cattle (Hammon 2008; Allen 2017). Unlike Uley, Harlow or Great Chesterford, the small assemblage prevents ageing data being used to identify seasonal slaughter of ovicaprids (King 2005). However, for the SWT, the small number of animals represented suggests that the sheep which ended up at the sites were drawn from local working flocks, indicating a degree of interdigitation between secular and ritual practices.

Ownership of sheep flocks is unlikely to have been restricted to any stratum of society, with classical sources referring to flocks owned by individual shepherds, large estates and others (Fronto, Ad Marcus Caesar 2.12; Varro De Re Rustica 2.10), albeit the applicability of these Mediterranean sources to Britannia is tenuous. Levitan (1993: 300) and King (2005: 334) have argued that the unusual nature of the goat-dominated assemblage at Uley, together with evidence for caprine coprolites and hay in the palaeo-environmental assemblage, suggests that a large temple flock was maintained on site. The same may be true for other large rural temple sites with extensive and highly specific sacrificial practices such as Harlow (King 2023). However, a number of other major shrine sites with large animal bone assemblages (Springhead, Tabard Square London, Chelmsford, Heybridge) show age and species distributions which indicate links to the local agricultural economy (King 2023: 232–233). While a temple flock cannot currently be ruled out for the SWT site, the apparent scale of sacrifice and the consistency with the surrounding agricultural economy would indicate that an origin in local flocks appears more likely.

These sacrifices were of the most common domestic animal. Within a rural agricultural context, these may have been understood as a sacrifice not only of the animal itself, but also of its future value within the flock, comparable to the Biblical tradition of dedicating the ‘first fruits’ of the harvest (Leviticus 23: 10–14). The extent to which the origin of the animal was known, or would have mattered, would have depended on the status and activities of the person offering the sacrifice. Since sheep were common and available, they may also have been acquired and offered as commodities — animals with value, but without full understanding of their agency or knowledge of their life history — by craftspeople or elite villa owners with little direct connection to the living flocks. The meaning of sheep in such sacrificial contexts will thus have been mediated by their relations of practice to the people involved. The scale of the practice at larger sites like Springhead, which are interdigitated with the local economy with sacrifices being made by those beyond the local population, would also have a significant impact on the local economy; those purchasing sheep to sacrifice would increase the flow of lower denomination coinage into rural communities, and may even have incentivized breeding for characteristics deemed suitable for sacrifice.

At the SWT, the absence of cattle from Roman levels (a single cow bone was found in topsoil) and significant percentage of pig bones, further demonstrates deliberate selection from the available major domesticates available in the local landscape, mirroring many other Romano-Celtic temples (Allen 2018: 196–199). Cattle being of relatively limited sacrificial importance at these, the most formal of rural religious sites, is a notable departure from wider Roman sacrificial practices, perhaps indicating a significant provincial refraction of the period’s increasingly connected religious traditions (Aldrete 2014: 28).

Another departure from supra-provincial practices is the importance of deer. The dramatic display of mounted red deer antlers at a focal point within the shrine building suggests that the visual impact and display of these animal remains was more important than the particular event of the stag’s death. Whilst the stag from which they came was certainly killed, whether this was a sacrifice at the shrine or the result of more conventional hunting is uncertain, although the latter appears more likely. The antlers form an important visual part of the shrine, indicating a link between deer, hunting and Bregneus, echoing the prevalence of miniaturized spears at the site. Horns and antlers have also been linked with various gods, including Cernunnos (Aldhouse-Green 2004) and Diana (Toynbee 1973: 143). However, their explicit ritual symbolism is also entangled in secular perceptions of deer based on landscape practice. Rives (2007: 80–81) highlights the importance of Silvanus and Diana in Roman Dacia, and implicitly suggests this may represent the interpretatio Romana of indigenous deities with strong connections to woodland. Their enduring importance may have rested at least partly on the continuing secular significance of woodland in landscape practice in Dacia, a quite heavily wooded province. Red deer and roe deer are both native to Britain, but there is little evidence of deer being hunted in the pre-Roman Iron Age (Rainsford and Roberts 2013). Deer remains become increasingly frequent in assemblages throughout the Roman period, but their consumption appears to have been socially stratified, with deer most common on villa sites and in high-status buildings on military sites but found rarely in rural communities (Rainsford and Roberts 2013). Red deer antlers with evidence of mounting have also been found at the shrines at Flexford, Surrey (Allen 2016) and Witham, Essex (Luff 1999), although in both of these cases the antler showed evidence of working and was not attached to cranial bone.

As in the later Medieval period, the control of deer and the ability to serve and consume venison in the Later Roman period can be viewed as an expression of landownership (Smith et al. 2018: 348). There is evidence of Roman landowners deliberately buying woodland or enclosing land in order to maintain deer (Allen 2014), including the importing of fallow deer from the Mediterranean and their maintenance in parks such as the Isle of Thanet, and at Fishbourne (Sykes et al. 2006; 2011). Hunting in the Roman world was a social performance, demonstrating elite social power through interaction with a powerful symbolical animal, reinforced by the employment of lower-status persons to manage a successful hunt (Ammianus Marcellinus 28.4.18; Sykes 2014: 66). Within the context of the temple, the display of deer antlers acts as an emblem of both the divine and the wild. However, their symbolism is also inextricable from the narratives of power and social inequality in which the concept of the deer is enmeshed. The association of deer antler with the divine may have acted to reinforce this narrative, placing the elite of society as having the ability to transgress these boundaries or possessing privileged media of communication with the gods.

Chickens form a relatively small component of the assemblage from the SWT, and unlike deer, they occur as disarticulated bones, predominantly associated with the central pit. As such, it appears very likely that chicken remains were also deposited in the pit as part of repeated ritual activity at the site, although the specific processes between ‘living chicken’ and ‘deposited bone’ unfortunately cannot be reconstructed from the current evidence. The use of chickens in ritual context is relatively common in Britannia. Chickens were arguably initially imported primarily for their symbolic and ritual value, and there is evidence for cockfighting in many Roman towns (Best et al. 2022: 879). The cockerel was associated with the Roman god Mercury, and at Uley, which is dedicated to the god, chickens and particularly cockerels comprised a substantial proportion of the assemblage (up to 8% in some phases; Levitan 1993; Doherty 2013). Chickens were also regularly used as sacrifices in Mithraic and other cultic rites in Britain and beyond (Lentacker 2003; King 2005). However, chickens were also used frequently for less specific ritual purposes, being present on at least 8 of the 25 religious sites surveyed by King (2005), and regularly included as offerings within Roman inhumation and cremation burials (Philpott 1991; Hill 2017).

The inclusion of chickens at the SWT can also be associated with socio-economic inequality in a rural context. By the Late Roman period, chickens were being exploited for meat and eggs. Chicken remains are most commonly found in major towns, roadside settlements and villas, and are relatively rare on rural sites (Allen 2017; Maltby et al. 2018: 1005–1009). Considering their infrequent presence on rural sites, there is likely to have been differential access to chickens in the Late Roman countryside, with the average chicken not venturing further into the rural landscape than the villa barnyard. The presence of chicken at the SWT further attests to high-status and cosmopolitan tastes in the use of animals at the site, also reflected in the deposition of suckling pig, wild game birds, and fish species such as red sea bream. There are notable similarities between the SWT assemblage and that from the nearby Deverill villa of (in addition to the major domesticates) suckling pig, red deer (including worked antler), and chicken (including immature bone hinting at rearing of chickens). Although red sea bream was not recovered at the Deverill villa, a quite wide range of marine fish was, suggesting links to coastal fisheries (Rainsford 2015).

Animals of the Imagination at the South Wiltshire Temple

The ritual deposition of animals at the temple also encompassed the artefactual representations of animals discussed above, including some creatures highly unlikely to have been seen in contemporary Britannia. Leopards, lions and other large felines are native to northern Africa in the far reaches of the Roman Empire. Leopards are depicted both in hunting scenes and in mythological contexts in Britannia, and are often associated with Bacchus (Toynbee 1973; Hobbs 2016). Some scholars suggest that leopards can be associated with Bacchus even in the absence of a depiction of him (Andrási 2008: 115), but this interpretation is far from certain at SWT.

Whilst leopards and lions were imported for arena shows in Rome, there is little evidence that they were exported to the far north-west provinces. The difficulty of transporting animals as far as Rome, both in terms of expense and in terms of the welfare and survival of the animals, makes it unlikely that they were exported much further (MacKinnon 2006). Furthermore by the second century AD, numbers of wild felines available for even the highest status Roman amphitheatres had been considerably reduced through over-hunting, capture for the arena and private households, as well as increased human settlement in the African provinces and consequent habitat reduction; more local animals such as bears, deer, and bulls were consequently more common in amphitheatres in this period (MacKinnon 2006).

It is therefore very unlikely that even a member of the Romano-British elite would have seen a leopard or lion in Britain.¹⁰ The association of the leopard figurine with the temple is thus an indication that some of those interacting with the place were linked with wider networks of practice, engaging in a locally or regionally contingent way with those networks, i.e. glocalizing. They had either seen such creatures in other parts of the Empire, or experienced imagery of such animals through media such as mosaics (Witts 2016), manuscripts, or iconography on portable material culture (Durham 2012). It is also relevant that leopards and lions were still perceived as dangerous wild animals, and that hunting of these animals was associated with elites in other parts of the Empire; this may tie into the elite hunting iconography represented by the deer antlers and miniature spears deposited at the temple, hinting at supra-provincial understandings of animals and landscape practice.

The other feline figurine from the site consists solely of the front right paw of an animal; identification to species is challenging, but Durham’s (2012) work suggests that this is unlikely to be a cat, as the only cat figurine she lists is posed wholly differently, resembling a toy rather than a miniature votive in its execution (Gurney 1990, fig 7.2). Identification as a lion is possible, given the appearance of a probable lion in shale on a table leg found in Dorchester (Timby et al. 2014: 289–290). However, the figurine is more likely to represent a second leopard, given the presence of the first find, or perhaps a panther, given both the greater numbers of panther figurines (9) than lions (6) and the poses for those for which Durham (2012, 3.35.7) references photographs more closely resembling this example. Any of these cases, though, provide further evidence of the ritual significance of imagined animals drawn from a wider geographical network to the people using the temple.

Animals, Ritual and Power

The structure and material deposition at the South Wiltshire Temple show that this is a site with widespread connections, used and influenced by a rural elite with access to material wealth. Similarly, many of the animals at the site are heavily enmeshed in power relations. Red deer and chickens are linked in secular contexts with elite consumption and wealth, and it is impossible to consider that these associations are not retained even within an explicitly religious context. Wild animals — including deer and exotic large felines — are present as symbols, referencing a hunting iconography associated with well-connected elites. The primary value of these animals in this context is in what they represent; the potential that their agency can be transgressed and dominated, and that these ‘wild’ animals can be caught, possessed and killed by people who then employ them as symbols.

This was not, however, the case for all animals sacrificed at the temple. The sacrifice and offering of sheep occurred widely across Roman Britain, in ritualized and secular contexts — indeed, as Chadwick (2012) argues, these are points on a continuum of ritual practice, rather than separate spheres. Being widely available and, individually, of less economic value than larger animals such as cattle, sheep made for versatile sacrifices available to both elites and those who did not have access to animals such as chickens or deer. This may be echoed in the metal finds assemblages from the temple; whilst curse tablets, bronze ritual leaves/plaques and amber indicate wider networks of practice, many of the miniature iron objects deposited are likely to have been created by craftspeople associated with the adjacent iron smelting site (Henry et al. 2020). Although miniature objects are fairly widespread finds on temple/shrine sites in Britannia (Smith 2018), iron miniature objects are highly unusual, and appear to speak to a particularly local refraction of this wider tradition (Kiernan 2009).

In essence, people sacrificed objects and animals — whether real or symbolic — that meant something to them within their networks of practice. In terms of the sacrifices’ materiality, the embodied landscape power through associations with hunting, the visual impact of the antlers, and the embedded craft knowledge of the miniature iron objects exemplify this point best at the South Wiltshire Temple. People undertook these religious acts in a temple context that drew on much wider vocabularies of religious practice in terms of its architecture and the act of sacrifice, yet refracted through local, regional and even supra-provincial understandings of landscape practice. Tellingly, this could take place in a temple singularly dedicated to a local god, Bregneus, strongly supporting arguments that orthopraxy, not orthodoxy, was at the heart of the shared assemblage of Roman pagan practices (Scheid 2005). The act of carrying out ritual practices may have been more important than the particular deity to whom they were dedicated.

As well as linking to supra-provincial grammars of ritual action such as sacrifice, at a local or regional level these practices were connected to broader, widely-understood animal cosmologies, which are rooted in secular practices and habitus. Cosmology incorporates the body of beliefs and associations held about particular animals, including their relative values and their associations with particular classes or types of people. It is from these broader cosmological understandings and entanglements that animals were selected as appropriate for ritual practices, and different selections were relevant to different communities and mediated through the shared practices of those communities. This could include the animals themselves; their value and relations are defined in significant part by their agency and histories, even up to the moment of sacrifice (Aldrete 2014). Indeed, it may be that their social agency in the human world was greatest during their final moments, as their actions took on meanings to the audience and participants in the ceremony which were contingent on historical and religious precedent. Nevertheless, this global facet of animal agency during sacrifice could simultaneously be understood with more local interpretations of their actions. The essence and meaning of ritual lies in its ability to speak to the world within which people find themselves — for example, harvest rituals would be highly relevant to farming communities, but less so to those who were less immediately connected to the land. Neither animals nor ritual practice can be understood in isolation from ‘secular’ practices, economies and power relations.

Through this comparative discussion of the animals entangled in the ritual assemblage of the South Wiltshire Temple and other temples in Britannia, we have demonstrated that whilst the meanings of animals at temples are deeply embedded in particular contexts of landscape, social, and economic practices at a range of scales, there are aspects of a much wider shared assemblage of religious practices active across Britannia and beyond. The concept of glocalization meshes with the significance of orthopraxy in paganism under the Roman Empire, emphasizing our key argument that while increased connectivity and imperial occupation engendered a wider shared material and social assemblage of religious acts, the nature and significance of those acts and materials was strongly refracted through local, regional or supra-regional practices. Interactions with animals at the SWT crystallize a number of these wider strands. They highlight not only connectedness — orthopraxy, architecture and the supra-provincial entanglements of exotic animals — and local/regional landscape practices such as sacrifices of sheep, deer hunting and miniature iron objects, but also departures from ‘Roman’ practice at Britannia’s most formal (sensu Smith 2001) rural religious sites, such as the apparent disregard for cattle at most Romano-Celtic temples, and the importance of deer. This discussion has also demonstrated the importance of engagement with theoretical perspectives when interpreting the complex interplay of human, animal and religion in Britannia.

In this article, we have argued that animals and other materials sacrificed in ritual activity in the past have tended to be separated in archaeological interpretation from the very real associations that they held in past societies. As such, our interpretations can become focused on religious aspects — the association of particular animals or suites of animals with particular deities — or occasionally the economic context of sacrifice, for example when considering the large-scale goat sacrifices at Uley. These neither acknowledge the animals as living beings with a history of human interactions, nor do they touch on the social aspects of ritual practice. We have proposed both practice theory, considering the meaning created by the full range of interactions between humans and animals, and glocalization, how aspects of cultural practice from across the Empire’s networks were adopted in new places at different scales, as middle-range theories to help develop a sense of past cosmologies surrounding particular animal species.¹¹ Applying this to the SWT case study highlights the ways in which some very obviously ‘ritual’ aspects, such as the displayed deer antler, also had social implications, connected via deer hunts to elites and elite power, and indicating a degree of social authority in those who established and maintained the temple. The materials used in sacrifice — oyster shells, chickens, uncommon fish, sheep, statues, miniaturized iron objects — show various levels of connectedness to the core of Empire, and various levels of social status.

Taking account of the secular materiality of ritual practices — to whom these animals and materials would have been available, the actions they afford, and the level of social prestige or expense they imply — may provide a new perspective on sites of ritual action. People conducted ritual activity with materials that had meaning to them and which were accessible to them, as well as those which were considered appropriate to the deity and the practice. Close consideration of these contexts reveals a level of social complexity around temples, with some able to participate and some excluded. Social structures and authority are also frequently reinforced by connection with the sacred, and considering the materiality of ritual allows us to view ways in which this may have been taking place. It is infrequently acknowledged that animals in the past were living beings, rather than objects. Acknowledging their lives as well as the symbolism of their ritual deaths is important perhaps ethically as much as theoretically.

Acknowledgements

Many thanks to Terry O’Connor and Denise Wilding for insightful comments on drafts of this paper. Research on this site was led by DR, Richard Henry and Steve Roskams, and Rebecca Venn managed the post-excavation programme. Funding for the SWT project and analysis was from a wide range of sources (see Henry et al. 2020), with the Society of Antiquaries of London being the lead funder, for which support we are very grateful. Many thanks to our reviewers for their comments on this paper.

Competing Interests

The authors have no competing interests to declare.