Fertile Imaginations: Pastoralist production and a new interpretation of a Roman period relief sculpture from Bath

D. Martin Goldberg

Previous studies of ‘Celtic’ deities and religious concepts have been largely dependent on evidence from the Roman period when theonyms appear in Latin inscriptions for the first time and there is an increase in figural representations of divine entities. Deconstruction of the term ‘Celtic’ is well rehearsed, but particular criticism has been aimed at constructing a pan- European religion from the diverse but fragmentary sources available, and especially backprojection into the Iron Age from Roman period evidence. Due to the heavy focus on iconography and because of the artistic medium, interpretations have often been made from an art historical perspective, sometimes using comparative data from all over Western Europe. This type of interpretation tends to remove the deities from their local context. Many of the attributes of deities have been explained in broad basic terms as symbolising prosperity, protection or more frequently fertility, without going into any greater detail. The question considered here is: beyond vague notions of fertility and well-being, can deities be shown to reflect more specific concerns of the local population? Belief systems were a potent and influential factor in ancient societies and would have had both subtle and definite impacts on the way people conducted their everyday existence. Ritual as a distinct category for study has been questioned (e.g. Bruck 1999), and therefore, ritual considerations should not be completely divorced from other aspects of studying the past. The following discussion will explain the inter-relationship between four iconographic images that appear on a relief carving from Bath and its relevance to pastoralist production in Roman Britain.

A relief sculpture from Bath, (CSIR 1.2 39) During the excavations of 1878–1890 at the Roman religious complex of Bath, a small relief carving, Fig.1, was discovered (CSIR 1.2 39). Four image types are depicted; two main figures of a male and a female deity, three smaller cloaked figures sometimes referred to collectively as genii cucullati, and an animal identified as a ram (Cunliffe and Fulford 1982: 11; Ross 1974: 428). Ross favoured an interpretation of the male and female deities as Loucetius Mars and Nemetona because of an altar (RIB 140), also from Bath, dedicated by Peregrinus who identifies himself as a Treveran (Collingwood and Wright 1995: 44). However, there is no contextual information linking the sculpted relief to the altar. The altar is also the only British dedication to Loucetius Mars and Nemetona, who are more commonly found on the west bank of the Rhine, precisely where Peregrinus claims to hail from. The images on the Bath relief can be shown to have many parallels with other Romano-British stone monuments and must therefore be considered in their local context rather than as an imported phenomenon. As a caveat, Ross herself conceded that the symbolism of the Bath sculpted relief was essentially British, markedly regional and pointed to the three small hooded figures as an especially insular feature (Ross 1974: 271).

Figure 1: The Bath relief depicting a horned ‘Mercury’, a Goddess with a vessel, three genii cucullati and a ram (Bath Museum)

Toynbee suggested that the male god was Mercury (1964: 158), which is supported by his usual caduceus attribute, but this in turn has led to a speculative designation of the couple as British equivalents of Mercury and Rosmerta, another divine couple with a very widespread continental distribution, but with no epigraphic evidence for their worship from Britain. A cluster of similar divine couple sculptures from the Cotswold region frequently come under the heading of Mercury and ‘Rosmerta’ (e.g. CSIR vol.1.7 78–82; Henig 1993: 26–28). The Cotswold sculptures depicting a divine couple are never accompanied by epigraphy, and this seems to have largely hindered interpretation, hence the search for continental equations. The lack of epigraphy has the potential to be just as significant as any continental parallels and can be interpreted as characteristic of a local cult that resisted the Romanised epigraphic habit (c.f. Webster 2003: 49).

A fascination with fertility Within the concept of the divine couple, as it occurs in iconography throughout Northern Europe, is a general acknowledgement that the partnership promoted fertility. In particular, the fertile imagery of the goddess is revealed through a variety of symbols such as cornucopiae, vessels, and fruits (Green 1992: 66–67). Rosmerta’s name means ‘the Great Provider’ and Mercury and Rosmerta as they appear on the continent epitomise this type of iconographic image, to which the British examples undoubtedly bear some resemblance.

Moving away briefly from the divine couple, it might be instructive to look in more detail at the smaller hooded figures frequently called the genii cucullati. Their name originates from two altars found in a Roman period shrine at Walbelsdorf in Austria. The dedications on the altars were inscribed ‘genio cucullato’– to the hooded genius (Egger 1932: 311–323 cited in Webster 1986: 67). Although the altars were without any associated iconography the term has since been employed to describe divine images from all across Europe where the main characteristic is a hooded cape or cloak (Green 1992: 104). In Britain they usually occur as a triad of figures, often perceived as being of diminutive status, as they appear in relation to the divine couple on the Bath relief. Studies of ‘Celtic’ art have seen triplication as a common method for increasing the potency of the divine image and this characteristic is thought to separate the British examples from their continental counterparts (Green 1991: 100–109). The purpose of this paper is not to completely deconstruct the labels that have been applied to these images, but merely to point out the continental origin of many of these epigraphic labels and how these can influence interpretation. In order to avoid the negative impact of deconstruction, the term genii cucullati will still be employed in the following pages, as it seems a fair, descriptive label.

The fertility interpretation for these hooded cult images is often emphasised without specific elaboration. On three of four examples from Cirencester (CSIR vol. 1.7 101, 103, 104; Henig 1993: 34–35), the cucullati are in the company of a female divinity frequently interpreted as a ‘Mother-Goddess’ (Mater) representing the land or territory and thus establishing their association with ‘abundance and fecundity’ (Green 1992: 104). Even when two of the triad accompanying the Mater are possibly armed, fertility remains paramount in the interpretation that their role is to protect her from barrenness, disease and famine (Green 1992: 104). On the other Cirencester examples, the genii and the goddess are often holding rounded or disc-like objects, sometimes identified as cakes, fruit or eggs.

Figure 2: Mater and genii cucullati from Stratton, Glos.(Robert Wilkins FSA reproduced courtesy of Institute of Archaeology, Oxford)

A very similar relief comes from Daglingworth Roman villa, near Gloucester, (CSIR vol.1.7 102), which uncharacteristically shows a dedicatory inscription that reads ‘cudae’ (RIB 129). Green states that this potential theonym implies prosperity and well-being but without detailing any supporting etymological evidence (Green 1992: 74). The mater is again depicted with an ambiguous rounded object in her lap, and appears to be receiving a similar offering. One of the clearest examples is from close by at Stratton, Fig. 2, (Henig 1998: 186–189) where the mater has a large round object in her lap. Henig interprets this as a fruit and by analogy an apple, but notes that if it were to scale, as the rest of the image is, it would have to be a large melon. Away from southwest Britain, six examples of genii cucullati without the accompanying mater have been found on the northern frontier (CSIR vol. 1.6 152–156, 485; Coulston and Phillips 1988: 61–63, 162). A relief from Netherby again shows each of them offering a rounded object (Webster 1986: 68–69).

So, the main interpretation of the function of the mater and her attendant genii cucullati has been a general association with fertility. The hood has been seen as having phallic symbolism and the shape has even been related to breast imagery (Green 1992: 104). Webster provides a neat summary of previous scholarly opinion on the symbolism and attributes of the genii cucullati showing how generalised and compounded the fertility interpretation can become:

‘It is possible, as M.W. Deonna in his major study has suggested, that the phallic shape of the cucullus and its all-enveloping quality symbolised mystery and darkness as well as the death-regeneration cycle. He concludes that, in Professor Toynbee’s words, “they were deities of death and after-life, of life through death, of healing and of fertility, as well as acting as protection from all kinds of evil”. Like many other deities, they obviously represent the whole cycle of birth, death and regeneration.’ (Webster 1986, 67)

This would seem quite a wide sphere of influence for what Webster then explains are minor spirits due to their classification as genii; a strange contradiction, given this is a modern designation based on an altar from Germany and none of the hooded figures from Britain are ever named or classified as such. These previous interpretations of the symbolism of the genii cucullati may well be valid and possibly intended by those who created the images. However, in many cases a concern with fertility would have had a specific focus. For example, in the wine producing regions of Gaul the iconography of certain deities show associations with viticulture. This was, and still is, an important part of the local economy. The cult images testify that this major concern of the population came under the influence of supernatural forces. This has been realised and commented on in studies of Gallic religion (Green 1992: 226). As an example that religious expression can be intimately connected with local economy, this may seem rather obvious, but it is one of the few where this sort of specific connection can be easily made. A more prosaic interpretation can be shown to link the iconography of the Bath relief through a proposed relationship with sheep husbandry as an important form of economic production in Roman Britain.

The Divine Couple in Britain: Mercury and Rosmerta? There are several stone sculptures of a divine couple from the south west of Britain that closely resemble the Bath relief. The best preserved of three potential examples (CSIR 1.7 78–80; Henig 1993: 26–7) is from Gloucester, Fig. 3, but a scatter of others are known from Cirencester (CSIR 1.7 81), Wellow in Somerset and Nettleton, Wiltshire (CSIR 1.2 116–117). Although unnamed, these sculptures have been linked to the continental depictions of Mercury and a native consort, Rosmerta, ‘the Great Provider’. Several readings of this divine marriage using gender and post-colonial perspectives have recently been proposed (Webster 1997: 325– 327; Aldhouse-Green 2003a: 97). Rosmerta has sometimes been seen as simply a suitable feminine accompaniment to the cult of Mercury, but she is depicted independently and with consistent attributes in Burgundy. The identification of the British goddess as Rosmerta is tenuous although she may be a related type of deity. The attributes that she consistently possesses in Britain suggest that, like Rosmerta, she was a divine entity in her own right, independent of her consort (Green 1992: 180).

Figure 3: ‘Mercury and Rosmerta’ from Gloucester (Gloucester museum)

The goddess appears on numerous examples with what Webster has described as ‘a magic tub with its bands of binding’ (Webster 1986: note for Plate 10) and what Green has perhaps more objectively described as a wooden iron-bound bucket, often with stirring implement (Green 1992: 180). More recently, Aldhouse-Green has suggested they resemble the wooden buckets found in Late Iron Age burial contexts (2003b: 41–42). Building on previous work etymologically linking the cudae inscription to the Cotswolds (Yeates 2004: 2–8), and considering the same group of sculpture as here, a recent paper by Yeates at the 2005 Roman Archaeology Conference used the ‘burial buckets’ to establish a material connection to the Iron Age as a precedent for the cult of a tribal goddess of the Dobunni. There is a recognisable regional grouping of related imagery in the cucullati and mater, the goddess with the vessel, and the divine couple, from the Cotswold region. However, the images of both the cucullati and the goddess with the vessel, from beyond the Cotswold region, and especially along the northern frontier, suggest that they relate to wider practices and beliefs.

Hilda Ellis Davidson (1999: 93) has provided the most convincing explanation for the bucket and stirring implement in a study of the relationship between goddesses and dairy production in Northern Europe. She has suggested that most examples of the ‘magic tub’ closely resemble a very functional and mundane item: the old-fashioned upright churn variously known as a dash-, stand-,or plunger churn that was historically used for dairying throughout Britain and on isolated farms in the Pennines up until the nineteenth century. Unaccompanied images of this goddess with the upright churn have also been found at Lemington, Gloucestershire (CSIR 1.7 94; Henig 1993: 32), Stitchcombe in Wiltshire (CSIR 1.2 119) Corbridge, Fig. 4, and Newcastle (CSIR 1.1 115, 183; Phillips 1977: 42, 59–60), and her appearance without her consort suggests that she is a native goddess whose attributes indicate a sphere of influence encompassing dairy production.

Figure 4: Goddess stirring a churn from Corbridge (English Heritage)

How might this link to dairy production reflect on the divine marriage? The Mercury on the Bath relief is ambiguous because of the wear on the detail of the carving, although there also seems to have been a deliberate ambiguity in some of the British depictions of Mercury. Many British images of Mercury show the wings of the petasus directly attached to his head as a zoomorphic attribute rather than the classical standard of the winged helmet. The depiction of zoomorphic divine images has more in common with traditional pre-conquest imagery in Northern Europe than with the classical artistic canon (Webster 2003: 40). The common symbolism of rams, horns and snakes in various combinations, establish the connections between the northern European horned god and the Classical Mercury with examples from both Gaul and Britain (Ross 1974: 172–221). A relief of a horned god from Cirencester shows the common attributes of a pair of ram-horned serpents, and on either side of the horns what has been interpreted as the open tops of purses filled with coins (Ross 1974: 185). A crude carving of a horned figure from Great Chesters on the northern frontier also holds Mercury’s caduceus attribute (Ross 1974: plate 41). Ross suggested that the conflation of Mercury with ‘the horned god of the North’ was through his role as protector of the flocks, which fell under Mercury’s jurisdiction in highland areas of the Mediterranean (1961: 77). In the iconography of the North West provinces a ram or sheep usually replaced Mercury’s familiar cult animal the goat, as exemplified by the Bath relief. In Britain, the ram would be a more appropriate accompaniment for Mercury, representing pastoralism as an important economic practice. Surprisingly, the ram on the Bath monument is the only one of the images whose fertile symbolism is occasionally superseded by focusing on its aggressive characteristics (cf. Ross 1974: 243–244 ‘emblem of war’ and the ram as a cult animal linked with native warrior gods and genii cucullati, especially demonstrated by altars from Lower Slaughter, Gloucestershire). However, for the following interpretation of the Bath relief it is necessary to emphasise the pastoral role of native ‘Mercury’ and the fertility aspects of the ram as a primary factor in sheep husbandry.

The structuring of the Bath relief displays a conscious composition and a deliberate diagonal balancing of imagery. Mercury is positioned diagonally opposite to the ram with which he would normally be associated and the goddess with the churn diagonally opposite the genii cucullati. The Mater who frequently appears with the cucullati elsewhere is not depicted with the dairying apparatus, but the Bath relief appears to reference the relationship between goddess and cucullati, using it as an artistic balancing principle, emphasising the interconnection of the various elements and the harmony of the piece as a whole. Rather than a vague or generalised notion of fertility, this relief provides a direct connection to specific socio-economic practices associated with sheep husbandry. The link between the ram, pastoral horned ‘Mercury’ and the other two image-types is established through the attributes of the goddess with the churn and the genii cucullati who each represent the main secondary products of sheep husbandry: dairy and wool.

Sheep husbandry and dairy production In the early fourth century A.D., British sheep were described as ‘distended with milk and loaded down with fleece’ (Panegyricus Constantino Augusto VI, 9). In general, studies have shown that meat production was not the focus for sheep husbandry in the Iron Age and Roman periods. Sheep can be utilised for all of the potential resources they provide which include dairy, blood, wool and manure whilst alive, and meat, fat, bone, skin, gut and horn after slaughter (Ryder 1983: 713). Most ancient faunal assemblages indicate a balanced management strategy that sought to maximise the availability of all those resources according to their relative importance. If nutritional concerns are assumed to be paramount then milk production is by far the most efficient way of utilising the energy potential of sheep husbandry (Ryder 1983: 720). Where specific management strategies for sheep can be determined from the faunal assemblages of the first millennia B.C. and A.D. they seem to have been geared towards the secondary products of milk and wool (Wild 1982: 113–114; Grant 1989: 136).

The milking of sheep was a common practice in the pre-modern era (Ryder 1983: 720– 725). Challinor (2004) has shown that dairy products from sheep continued to be used throughout the Medieval period and into the nineteenth century in remoter parts of the British Isles and has begun research on pottery assemblages from Shetland in order to show the importance of dairy produce in later prehistory. The analysis of residues from pottery vessels indicates that dairying was an important part of Iron Age farming practices, supplementing evidence for husbandry patterns that had previously relied on faunal analysis (Evershed et al. 2005). Serjeantson (forthcoming) has shown that the intensification of sheep husbandry began as early as the Middle Bronze Age, in tandem with other forms of agricultural intensification, and faunal assemblages indicate that some sheep flocks were closely managed for milk production. Agricultural intensification would have led to some pressure on communities to maximise output and dairy production has been shown to be ten times more efficient in realising energy potential than meat production. The importance of sheep husbandry to arable farming systems should never be underestimated and the rotation of fields left to pasture while the sheep graze and re-fertilise the soil would have been a vital strategy during periods of intensification and is perhaps part explanation for the increase in sheep numbers in relation to pigs and cattle during the Iron Age (Cunliffe 1991: 173). Successful sheep husbandry embodied ‘fertility’ on many levels and may have been a strategy for the display of wealth, social prestige, and reciprocal exchange. Changes in animal husbandry occurred during the Roman period, with an increase in the proportion of cattle bones in faunal assemblages along with butchery patterns characteristic of intensive meat production (King 1984). Cows obviously form an important part of dairy production, but are considered much more valuable for their meat than sheep are. In general, there was an increase in specialised economic strategies (Hambleton 1999: 60) and these variations suggest that the growth of cattle-raising was linked to Roman urban and military areas, while many rural sites displayed continuity with Iron Age patterns and the predominance of sheep (Grant 1989: 136). Classical writers such as Caesar, Strabo and Pomponius Mela perpetuate the stereotype that Britain was rich in flocks of sheep (Wild 1982: 116). We can envisage that the milking of sheep formed a vital part of dairy production in the past and that the symbolism of the divine marriage as expressed on the Bath relief could be meant to facilitate the productivity of this economic practice.

Roman writers considered cow’s milk inferior to the milk of sheep and goats (Columella Rust. VII.2.1) and regarded ewe’s milk as the most nourishing (Ryder 1983: 721). Sheep’s milk contains double the fat and protein content compared with cow’s milk and is particularly suitable for cheese making. Cheeses are high-energy foodstuffs that require no cooking, are easily stored through winter, and are the most efficient way of utilising dairy products throughout the year. Butter production requires circa 2.3 times more milk per unit than cheese, which is also less prone to contamination and ruin, making it a more viable and stable option than other dairy products (Challinor 2004: 165–166). The upper roof spaces of roundhouses could be ideal storage spaces for foodstuffs because of the drying and antiseptic qualities of the circulating smoke (Pope forthcoming) and smoked or aged cheese would seem particularly plausible. Regional cheese varieties were already recognised during the Roman period with the most famous example being Roquefort, praised by Pliny the Elder in A.D. 79 (Ryder 1983: 722). The round, oval or cake-like objects frequently displayed by the genii cucullati and Mater, Figs. 2, could just as easily be interpreted as cheeses in their many and varied traditional forms, Fig. 5. The genii cucullati might also symbolically represent the other main by-product of sheep husbandry, wool.

Figure 5: A small example of the many forms of traditional and atisan cheeses (A. Wilson)

Genii Cucullati and wool production Green comments that ‘a striking feature of the iconography of the cucullati is the homogeneity of their dress’ (Green 1992: 104). There is actually significant variation in the lengths and styles of garments but she is essentially correct in that the hooded cloak is a primary symbol intended for communication by these sculpted images. The popularity of the types of hooded garments and woollen cloaks worn by the cucullati is attested by numerous other images and various literary references (Webster 1986: 66), indicating how suitable and common this garb was for outdoor wear and travelling. A heavy woollen cloak would have afforded protection from the elements for the shepherds who were responsible for the care and protection of the flocks. The shepherds were not only an integral part of the process that produced the hooded cloaks, but would have been one of the primary users. Their appearance could have become intimately associated with their role and its ultimate product. The cucullati may be the supernatural equivalent, guardian spirits of the flocks, which would account for their hooded cloaks and the protective aspect of their iconography. Davidson suggested a similar function for the cucullati as antecedents to the later widespread folklore of pucks, brownies and other supernatural helpers (1989: 111–115) and curiously this folklore incorporates motifs of clothing and especially cloaks, as well as a frequent association with dairy production.

When considering the native economy of northern Britain, Piggott was the first to suggest a connection between a wool industry in the region, the heavy cloaks of the genii cucullati and the byrrus britannicus, the British hooded cloak listed in Diocletian’s Edict of Prices (Piggott 1958: 27). The only two British items to appear in the Emperor’s price-list are both woollen products. The tapete was a heavy wool rug at the top of its section and therefore considered the best in the Empire. Of greater relevance for Piggott was the byrrus, the hooded wool cape that was also a high quality product, judging by its price (Wild 1982: 120). These two items demonstrate the importance of the British wool industry and Piggott thought that the appearance of the cucullati could symbolically represent one of those products.

Piggott commented on how British cloaks held their price into the eighth century, and the products of the wool industry continued to be a valuable export into the Middle Ages (Piggott 1958: 27). Unfortunately, Piggott’s study was heavily reliant on historical analogy and his model for northern Britain of a purely pastoral society has been refuted by recent studies, showing that the north was a mosaic of upland and lowland resources with mixed farming strategies (Miles 1989: 119). Sheep husbandry may not have been the marginal activity it became in the modern era and need not be mutually exclusive from other agricultural systems. Unfortunately faunal assemblages are unavailable for many areas of northern Britain because the mostly acidic soils hamper preservation (Hambleton 1999: 16). Although not a typical settlement site, indications from Stanwick, the large lowland enclosure in the Tees valley, suggest that proportions of sheep were higher in the earliest phases with a gradual increase in the proportions of cattle bone during later phases (Hambleton 1999: 111).

Spindle whorls, loom weights, bone weaving combs and other tools associated with spinning and weaving are common in finds assemblages throughout Britain and show evidence of sheep management and wool production in the archaeological record for both the Iron Age and Roman periods (Grant 1989: 136). The processing of raw wool and weaving are tasks that are not seasonally determined, but can be carried out at any time of the year. Woollen textile production was most likely to have been organised on a small-scale during the Late Iron Age and Roman period, judging from the widespread distribution of items associated with weaving (Wild 1982: 119). For the Roman period, Vindolanda near Hadrian’s Wall has produced numerous examples of woollen fabrics that help support the suggestion of woollen manufacturing traditions in this region (Wild 2002). However, a better association between the genii cucullati and sheep husbandry can be found from examining the local context in southwest Britain, particularly rural sites where the genii cucullati have been found and where faunal assemblages from the Roman period are more informative.

Genii Cucullati in a local context Gloucestershire and the Cotswolds are historically major sheep-rearing and wool producing areas of the country, as are the upland areas of the Cheviots to the north of Hadrian’s wall, and the Pennines to the south. As Piggott suggested, the importance of these local economies may be a genuine connection between the two areas where the genii cucullati are found, but more than historical analogy is needed to support this. As with the north, generalisation glosses over what would have been a myriad of regional systems, even between regions on either side of the Cotswolds. However, the general pattern for central and western southern Britain shows that sheep predominated in animal husbandry regimes during the Iron Age (Grant 1989: 136). By the Roman period, samples show an emphasis on older animals indicating that wool production may have become more important and there is some evidence for selective breeding (Maltby 1998: 426–7). Sheep husbandry and a focus on secondary products seems to have retained its importance in areas that were best suited to this economic practice, where it was the most viable and perhaps traditional option. However, the distinction between steak-eating Romans and mutton (or cheese) fed natives is a simplification and polarisation of what would have been much more complex interactions and adaptive strategies. The site at Wilcote, Oxfordshire, might be taken as a typical ‘native’ site in central southern Britain, with greater proportions of sheep bones in the faunal assemblage, and yet its origins are as a Roman military staging post. Even in this distinct initial stage with a more variable assemblage, sheep bones predominated, probably due to the availability of local resources. This suggests that sheep were the focus of local husbandry regimes, and consequently during the later de-militarised phases, Wilcote reverts to type. Several unusual deposits of articulated sheep and dogs from the later phases have even been interpreted as ritualised acts marking the revitalisation of traditional practices (Hamshaw-Thomas 1993: 176–177).

The two sites of Wycomb and Lower Slaughter in Gloucestershire have produced sculpted images of genii cucullati and stand out as rural religious centres. Wycomb can be compared with other rural temples from the Cotswolds, such as Uley. There were many problems in recovery, preservation and consequently interpretation of the faunal assemblages from these sites, but sheep/goat bones were generally better represented especially from the Syreford Mill excavations, which have been interpreted as typical of traditional or Roman period native rural settlement (Timby 1998: 350; Maltby 1998: 428). The ditched compounds and enclosures at Syreford Mill, Wycomb, Lower Slaughter and nearby Bourton-on-the-Water, indicate stock management, and these sites exhibit other vernacular building traditions (Ibid.: 350, 388–389). Cloth manufacture is indicated by common finds of loom weights, spindle whorls and needles at all the sites, and iron shears have also been found at Springhill, Bourton and Lower Slaughter (Ibid.: 359, 382, 388–389). Strong evidence for the cult of Minerva has been noted for the Cotswold region and two examples of ‘native’ depictions of this goddess were deposited in Well 5 at Lower Slaughter along with other sculptures depicting two genii cucullati, a warrior triad and an altar depicting a local god and a ram. Minerva was, amongst other things, the patron of weaving and spinning and combined with the other images from the well, this dovetails nicely with the imagery from the Bath relief. Mortaria and colanders are well represented at Lower Slaughter, which may be significant because these are the type of vessels that would be particularly suitable for cheese making. H.E.M. Cool has recently discussed the unusual popularity of mortaria in Roman Britain. Not only do mortaria occur on remote sites, but there was a disproportionate interest in the acquisition of them for rural sites, where they form a higher percentage of assemblages than in urban areas. This calls into question the assumption that they represent the adoption of ‘Romanised’ foodways and revitalises the original suggestion made by previous experts in mortaria studies, that the main purpose was for cheese making (Cool 2004). Examining the local context of Wycomb and Lower Slaughter shows the importance of sheep husbandry in the archaeological record, with the focus on the secondary products of wool and dairy reflected symbolically in local religious sculpture and highlighting the relationships inferred from the Bath relief.

Discussion Woolf has recently suggested that one of the characteristics of the growth of ancient empires was a revolution where religion became disembedded from traditional societies and a plurality of belief became possible, separate from cultural/ethnic identity (2005). This may be apparent within the Roman Empire from a pan-European perspective and especially when looking at the highest social levels but the possibility of discrepant experiences and identities should also be expected (Mattingly 2004). In the context of Roman Britain these should be most apparent in rural settlements like Lower Slaughter, away from urban and military areas. However, negotiation between traditional embedded beliefs and classical Mediterranean culture need not be polarised and we should be aware of levels of variation and adaptation expressed through material culture (as with mortaria), but especially visually exemplified by religious art (Webster 2003). The British depictions of Mercury, the goddess with the churn and the cucullati show these levels of variation perfectly. These images represent the religious reflex of traditional, local economic practices expressed through the new medium of stone sculpture, based on classical models, but usually without epigraphy. Webster has shown how classical artistic proto-types could be appropriated by local artists through mimesis or they may be ambiguously treated (like Mercury’s horns/wings) with the intention of depicting more locally appropriate imagery, as has been suggested for the cheeses held by the goddess and cucullati (Webster 2003: 40).

Stone-carved representations of supernatural beings are seen as an influence of the Mediterranean world and a prominent factor in the distribution of these sculptures from the Cotswolds and along the northern frontier may have been the availability of suitable stone for carving. The small chalk relief from Rushall Down, Wiltshire may be instructive because it represents the cucullati in a medium that was less likely to survive the test of time, but was very common on the chalk downs of Wiltshire (CSIR 1.2 103). Stone sculpture may not have been the most common way of representing or propitiating the supernatural forces that influenced dairying and the wool industry. The recognition of ritualised deposits of animals for both the Iron Age and Roman period hints at broadly related practices and beliefs, albeit with localised variations (Grant 1991; Scott 1991). The widespread distribution of mortaria, spindle whorls, loom weights and the other accoutrements of pastoralist production from all over Britain indicate that associated beliefs may be equally widespread, and the depositional contexts of those items await further interpretation. Engagement with these beliefs may also be through practices that are not generally archaeologically visible, perhaps through the products themselves. Before the imagination creates images of little woollen dollies or recalls adverts featuring Douglas, the Lurpak butter man, there may be rare organic artefactual survivals in the finds of ‘bog butter’ from Ireland and Scotland. These had been suggested as votive offerings of dairy produce or surplus contained in a variety of vessels and deposited in bogs and marshes (Cunliffe 1993: 16). Recent analysis has shown that two thirds of the examples from Scotland derived from dairy fats (Evershed et al 2004: 270–275). Hunter has considered them in relation to other vessel deposits in Scotland, as they seem to fit into a distributional gap (Hunter 1995: 119), but they could also be another manifestation of the ritualisation of dairy production. Other equally important social and economic practices would have also received ritualised expression. Further examples of traditional religious beliefs adapting and being expressed by a rapidly changing and radically altered material culture are to be expected within the new social order introduced to Britain by the Romans.

Conclusion The disparate and widespread nature of the evidence for ‘Celtic’ religion encourages broad analogies and comparative interpretation. Many studies range across the width of Europe and the span of a millennium or more, emphasising fertility, without any specific explanation. The aim of this paper has been to show that informative conclusions could be made about certain sculpted images from the Roman period, by examining them in a more local context. A definite message was intended by the person/s who commissioned these examples of religious sculpture, which may have had much to do with fertility and its attendants of well-being, protection and prosperity, but could also be much more specific and intimately related to their own lives. Previously the over-riding interpretation linking Mercury, his consort goddess, and the genii cucullati has been the notion of fertility. A more explicit link between these three representations of supernatural beings has been suggested through a new interpretation of the relief from Bath. This piece of iconography contains elements of religious expression that can all be related to sheep-husbandry and its by-products of dairy and wool. All three appear together, horned Mercury, the native goddess with her churn, the three genii cucullati and below the goddess the primary element in the fertility of sheep flocks, the ram. Surprisingly the only figure that has not been regularly associated with fertility is the ram and yet its role in husbandry relates directly to the fertility of the flock. The Bath relief sculpture is particularly informative as it shows the relationship between sheep husbandry and its by-products of dairy and wool, expressed through a variety of religious imagery that appear in combination elsewhere. These cult images are a result of human beings interacting with the unknown, the supernatural forces that affected their lives on a daily basis and could influence the success or failure of their economic ventures and even their basic means of subsistence. I imagine that the fertility aspects of native ‘Mercury’, the cucullati and the associated goddess reflect the concerns of those involved in sheep-husbandry and its by-products; the shepherds tending the flocks, who would have worn the heavy woollen cloaks woven in the settlements during the quiet periods of the year, when the butter-churning and cheese-making were completed.

Department of Archaeology, University of Glasgow

Acknowledgements Permission to use Figure 3 was kindly granted by Gloucester museum, Figure 4 by English Heritage and Figure 1 by Bath Museum. Martin Henig graciously provided Figure 2. The photographer Andrew Wilson helped with the presentation of all images, especially Figure 5 taken by him courtesy of I.J. Mellis, cheesemonger. I owe many thanks to Bill Hanson, and Katherine Forsyth for reading and commenting on several early drafts of this paper and to Martin Carruthers, Amanda Brend, Gordon Noble, and particularly Tessa Poller for much advice, cups of tea, and good humour. Thanks also to my family, Shelly, and all my friends, who have given me great support over recent years.