The Shifting Baselines of the British Hare Goddess

The rise of social zooarchaeology and the so-called ‘animal turn’ in the humanities both reflect a growing interest in the interactions of humans and non-human animals. This comparative archaeological study contributes to this interdisciplinary field by investigating the ways in which successive human cultures employed religion to conceptualise and interact with their ecological context across the longue durée. Specifically, we investigate how the Iron Age, Romano-British, early medieval English, medieval Welsh, and Information Age populations of Great Britain constructed and employed supranatural female figures – Andraste, Diana, Ēostre, St. Melangell, and the modern construct ‘Easter’ – with a common zoomorphic link: the hare. Applying theoretical concepts drawn from conservation ecology (‘shifting baselines’) and the study of religion (‘semantic centres’) to a combination of (zoo)archaeological and textual evidence, we argue that four distinct ‘hare goddesses’ were used to express their congregations’ concerns regarding the mediation of violence between the human in-group and other parties (human or animal) across two millennia.

. Julius Caesar, writing in the mid-first century BC, describes Britain as a region inhabited by a range of social groups who kept large numbers of cattle, and who "Leporem et gallinam et anserem gustare fas non putant" [regard it as taboo to eat hare, chicken, or goose] (Commentarii de Bello Gallico V.12;Edwards, 1919, pp. 250-251). Caesar's use of the term fas ('religiously permitted'), rather than ius ('judicially permitted'), suggests that the prohibition on the consumption of these animals was an explicitly religious matter (Ferguson, 1987, pp. 749-750). This may explain the rarity of hare bones in midden deposits at most British sites in this period (Allen & Sykes, 2011;Crummy, 2013). Those that are known are frequently identified as part of ritual assemblages in associated bone groups, as at Blackhorse Road (a mid-Iron Age site with multiple enclosures, some potentially the remains of round houses, near Letchworth, Hertfordshire; Legge et al., 1988), Winnall Down (one of a pair of linked Iron-Age settlement sites just east of Winchester, Hampshire; Maltby, 1985;cf. Davis, 2014), and Winklebury Camp (an Iron-Age plateau fort, Basingstoke, Hampshire; Jones, 1977) -although whether these animals were sacrificed, kept as pets, or served some other purpose remains open to interpretation (Crummy, 2013;Fowler, 2020). The hare's special status might be further witnessed in the use of periphrastic 'noa names' for hares in a range of Celtic languages (Forbes, 1905, pp. 168-170), potentially reflecting wider European hare taboos (Boyle, 1973;Krappe, 1942, pp. 198-201). That said, faunal evidence from contemporary sites on the continent suggest hunting and consumption of hares was more common beyond Britain (Green, 1992, pp. 50-52;Meniel, 1987, pp. 89-100).
Despite the hare's apparent special status in Iron-Age Britain, any link to a particular deity is less clear. British religions in this period are poorly understood, with textual accounts and inscriptions from the late Iron Age mostly reflecting Roman ideas about who the Britons and their gods were. We will likely never firmly establish the names and identities of a coherent pantheon of deities -if such a thing ever existed. Nonetheless, archaeological evidence does allow us to suggest some general trends, particularly when it comes to gods as mediators between humans, animals, and their shared ecological contexts. A number of anthropomorphic figures displaying zoomorphic features from across western Europe are traditionally understood to represent deities (Green, 1986(Green, , pp. 167-199, 1992, sometimes due to their appearance in Roman religious iconography and Latin inscriptions. These famously include a widespread motif of a cross-legged male figure with stag's horns, sometimes accompanied (as on the interior Panel A of the Gundestrup cauldron) by animals including deer, snakes, and bulls (Nielsen et al., 2005), and who is identified as "Cernunnos" on the Gallo-Roman Pillar of the Boatmen, a first-century AD monumental pillar with Latin inscriptions and a hybrid iconography of Roman and pre-Roman divinities (Bober, 1951). Female Gallo-Roman examples may include Artio, seemingly a goddess associated with bears (Delamarre, 2003, pp. 55-56;Deyts, 1992, p. 48;Zangemeister, 1905: 5160, p. 31), and the equine goddess Epona, absorbed into Roman religion as patroness of cavalry (Deyts, 1992, pp. 51-57;Euskirchen, 2006), a bronze statue of whom was recovered in England (Johns, 1971), for all she does not appear to have been the recipient of cult in Britain (Mackintosh, 1995, pp. 29-37).
Miranda Green argued that such deities "were associated with hunting and […] seemed to have a dual attitude to animals, as guardians and hunters, in an intimate and mutual bond" (Green, 1992, p. 134). While it is entirely possible that hare hunting was associated with a particular Iron Age British goddess, no iconography of a female figure with lagomorphic features has yet been found. We can, however, draw on Cassius Dio's second-century account of the mid-first century rebellion of the Iceni, an ethno-political group in eastern England, which records how their leader, Boudicca, released a hare from under her dress as a type of pre-battle "μαντείᾳ" [divination] (Historia Romana LXII.6b.1-2; Cary, 1925, pp. 90-93). The hare's motions apparently conveyed positive news, and Boudicca is reported to have thanked "Ἀνδράστη" [Andraste] -presumably the same figure as "Ἀνδάτης" [Andate] in whose sacred grove the Iceni later celebrated their victory (Historia Romana LXII.7b; Cary, 1925, pp. 94-95). Boudicca's reported address to Andraste suggests she was seen as a protective, tutelary deity by the Iceni, who could be entreated to provide "νίκην καί σωτηρίαν καί ἐλευθερίαν" [victory and survival and freedom] (Historia Romana LXII.6b.1-2; Cary, 1925, pp. 90-93). This presentation of Andraste's domains lines up with the general observation that female supranatural powers in Iron-Age Ireland, Britain, and Gaul frequently bear responsibility for a range of activities associated with both fertility and violence (Green, 1995, p. 29;Herbert, 1996). Iconographic representations of female figures with wargear, such as the first-century BC Dinéault statue with its goose-crested helmet and the Redones coin showing a mounted female figure with a shield (both from first-century BC Brittany; Green, 1995, pp. 34-35), have been interpreted as depictions of such figures. However, later in his Historia, Dio claims that Ἀνδάτης was the local name for Νίκην [Nike] (Historia Romana LXII.7b; Cary, 1925, pp. 94-95), the Greek-language name of the Roman goddess and personification of martial success Victoria (Scherf & Strothmann, 2006). This makes it difficult to tell whether the characteristics Dio attributes to Andraste -particularly her martial power -were genuinely associated with the Iron-Age British deity, or were instead projected onto her via interpretatio romana.
Notably, the only potentially pre-Roman image of a hare from Great Britain was found during field walking at Woodcock Hall, an unexcavated settlement site (and later Roman military outpost) in Norfolk (Brown, 1986, pp. 39-40), an area dominated by the Iceni during the first century (cf. Allen, 1970). The leaded-bronze hare and a similar bronze stag figure are small (c. 40mm and 30mm long respectively), and the remains of lugs suggest they may have been attached to straps or clothing ( fig. 1). Whether they held any religious or cultural significance is unclear, as it is quite possible that they were 'merely' decorative. That said, just as Iron-Age British cultures appear to have exhibited taboos around the consumption and naming of hares as noted above, the otherwise complete lack of lagomorphs in British iconography from this period might imply that there was a similar prohibition against its representation in pictorial or sculptural form. If so, the Woodcock Hall figure could well have been regarded as sacrally charged due to its contravention of such a restriction.
Thus if there was ever a particular deity linked to the hare in the British Iron Age (which does not seem unlikely given its apparent special status), Andraste is the best -indeed, the only -candidate we know anything of. Quite what her semantic centre may have been depends on our willingness to trust the details of Dio's account (McDougall, 1991;Millar, 1964). Yet even if Dio did draw on the Roman Victoria in his presentation of Andraste, such interpretatio romana would still have required some underlying similarity between the two goddesses. Considering the general roles played by goddesses elsewhere in societies that spoke a Celtic language, we therefore believe Andraste's semantic centre to have been based in warfare, and the power over life and death that followed from this.

Roman Britain: Diana Andraste
Animals were undoubtedly important in Imperial Roman religion, playing roles in mythology, iconography, and ritual (Johnston et al., 2016). Most Roman deities appear to have had connections with particular animals -Jupiter and the eagle (Toynbee, 1973, pp. 240-243;Graf & Ley, 2006), Diana and the deer (Reinken, 1999;Toynbee, 1973, pp. 143-147), and so on -although the hybridisation of anthropomorphic and zoomorphic features witnessed in British (and related) religions is not a widespread feature of Roman religion. Divine shapeshifting, where it occurs in narratives like those Jupiter inherited from his Greek predecessor Zeus (e.g. Lucian's Θεῶν Διάλογοι, 6.2;MacLeod, 1961, pp. 262-263), seems to have been an either/or matter; and iconography tends to present gods accompanied by their favoured animals rather than partially embodying them. It thus seems possible that a Roman goddess could have been associated with the hare -but lagomorphs do not appear in mainstream Roman religion. Nonetheless, provincial culture could differ markedly from Italic paradigms, with Romano-British reflexes of Roman religion exhibiting syncretic elements drawn from regional pre-Roman societies (Daubney, 2010;Goldberg, 2009;Haeussler & King, 2007Haeussler, 2008Haeussler, , 2012Henig, 1984;Smith et al., 2018, pp. 120-204;Watts, 1998;Webster, 1987). Here we will summarise an argument that a local Romano-British reflex of Diana exhibited just such an association (Murphy, 2020).
With the establishment of the province of Britannia and its subsequent cultural hybridisation in 43 AD, local societal attitudes to lagomorphs appear to have shifted. Romano-British culture engaged with the broader ecological context in significantly different ways from its Iron-Age predecessors (Allen & Sykes, 2011;Sykes, 2009;van der Veen, 2008), with a greater emphasis on hunting and fishing as sources of protein (Smith et al., 2018, pp. 78-119) and the deliberate introduction of new species such as the fallow deer (Sykes et al., 2006), chicken (Sykes, 2012), and the European rabbit (Sykes & Curl, 2010) at high-status sites. The hare appears to have gained greater prominence as a result of these shifts, appearing in leporaria -dedicated lagomorph enclosures -at sites like Fishbourne, an opulent elite residence with multiple animal enclosures near the settlement of Noviomagus Reginorum, today's Chichester (fig. 2; Allen & Sykes, 2011;Cunliffe, 1998). Evidence from elsewhere in the Roman cultural sphere suggests hare-hunting was a male pastime, sometimes conducted from horseback (e.g. Martial's Epigrams XII.14;Ker, 1920, pp. 326-327), although hares also appear in pet-like contexts and in xenia mosaics as part of an idealised rural diet ( fig. 3; Toynbee, 1973, pp. 200-203). The fourth-century hare mosaic from the Beeches Rd. site in Corinium Dobunnorum, a Romano-British town (now Cirencester, Gloucestershire), bears witness to British expressions of such motifs (Cosh & Neal, 2010, pp. 91-94).
Hare iconography thus becomes markedly more common in Roman Britain ( fig. 4), and also begins to appear in religious contexts (a partial catalogue of such motifs is under preparation as Henig 2018). Generally speaking, Romano-British deities exhibit a range of syncretic positions, including the accretion of pre-Roman epithets -perhaps most famously in the rededication of the Aquae Sulis springs at Bath to Sulis Minerva (Cunliffe, 1988(Cunliffe, , 1996Cunliffe & Davenport, 1985;Dark, 1993;Davenport, 2007). Romano-British reflexes of syncretic deities can also be traced in personal jewellery, including enamelled bronze or brass plate brooches (Johns, 1996a, pp. 170-182;Mackreth, 2011). Plate brooches are a subset of fibulae (so-called 'safety-pin' brooches), which were a common part of dress or accessory in the late Iron Age and Romano-British periods. While the more common crossbow fibulae are typically associated with male dress, the lack of solid archaeological context for most plate brooch finds -which are typically loose finds, likely as the result of casual loss during use -does not allow us to draw broad conclusions regarding their use (Mackreth, 2011, vol. 1, pp. 234-235.) The preponderance of plate brooches at Colchester, where a large civilian population was massacred during Boudicca's rebellion, might suggest that they were more generally worn by women than by men (Mackreth, 2011, vol. 1, pp. 234-235). That said, there is a great diversity of form, and plate brooches can typically gather less fabric than other fibulae, suggesting that they may have been more decorative than other types of brooch: it is thus likely that different plate brooches carried different symbolic values, and were used by different segments of the population.
Many plate brooches are geometric, but some are emblematic, depicting animals and objects understood to reflect the iconographies of particular Romano-British deities, perhaps serving as a form of pilgrim's badge or visual marker of the wearer's religious affiliation: horse-and-rider brooches to the Romano-Celtic rider deity (Ferris, 1986); horses to Epona (Johns, 1996b); dogs to Nodens or Asclepius (Simpson & Balance, 1998); cockerels, sandals, purses, and flies to Mercury (fig. 5;Crummy, 2007), and so on. Yet the largest grouping of zoomorphic brooches -hareshas yet to be convincingly assigned to any known deity ( fig. 6). It is possible that these brooches had no religious connotations: proverbially fertile hares were common gifts between romantic partners (Lazenby, 1949, p. 301), and the motif of the hound chasing the hare could represent romantic pursuit. Nonetheless, the widespread connection of other zoomorphic plate brooches to the cults of particular deities suggests, to us, that there was likely a religious element to this lagomorphic jewellery, at least for many of its wearers.    The explosion in hare iconography from the mid-Roman period suggests that any Romano-British hare deity must have been the product of some level of syncretism -which does not simply "Romanise" or "Celticise" one deity or the other, but rather produces new hybrid reflexes (Goldberg, 2009;Haeussler & King, 2007;Haeussler, 2008Haeussler, , 2012. We are therefore sceptical of suggestions that either Roman Venus (Boyle, 1973, p. 324) or Iron-Age Andraste (Lee, 2018) could have been the intended referent of the hare-shaped plate brooches, and believe the most likely candidate is a local Romano-British reflex of an imported Roman deity, adapted to British cultural and ecological contexts. While there are a number of Romano-British hunting gods, notably Apollo Cunomaglos and the Cotswold hunter-god (both relatively well attested in southern England; e.g. Roman Inscriptions of Britain (hereafter RIB): 3053; Corpus signorum imperii Romani (hereafter CSIR) 1.7: 110-115, pp.37-39; cf. Merrifield, 1996;Murphy, 2020), we believe the most likely candidate is a reflex of Diana-Artemis, the Classical goddess of land outside the civitas, and thus patron of wild animals and hunters (Scheid, 2006).
Inscriptions across Roman Britain witness Diana's popularity (RIB: 138, 316, 1126(RIB: 138, 316, , 1209(RIB: 138, 316, , 2174(RIB: 138, 316, , 2122, particularly in military contexts. It is important to note that Roman forts were accompanied by thriving civil populations, and both soldiers and civilians would have participated in the cultural hybridisation noted above (Birley, 1973). However, the two altars conventionally understood to portray Diana visually do so without reference to her traditional deer iconography ( fig. 7; RIB: 1121, 2343; cf. CSIR 1.1: 60, pp. 23-24; CSIR 1.10: 73, pp. 42-43). Deer are also absent from the plate-brooch corpus, which features just one unequivocal stag (Mackreth, 2011: 8097, vol. 2, p. 128). This may have been due to the strong association between deer and pre-Roman deities such as Cernunnos, Callirius (RIB: 194;Green, 1997, p. 183), andSilvanus Cocidius (RIB: 1207;Webster, 1986, pp. 75-76), suggesting Diana's cult would have needed to develop alternative iconography in a British context so as not to overlap with established symbolic systems. If this alternative drew on another animal that featured prominently in Romano-British hunting culture, like the hare, the change may not have been any more jarring than the replacement of Mars' classical woodpecker (e.g. Plutarch's Quaestiones Romanae 21;Babbitt, 1962, pp. 34-37) with the goose that accompanies the (potentially Romano-Germanic) Mars Thincsus at Housesteads, a military outpost on Hadrian's Wall, the northern border of the Roman province (RIB: 1593; CSIR 1.6: 159-161, pp. 64-65; Green, 1986, pp. 113-114). Diana appears elsewhere in syncretic form with local deities like Abnoba in the Black Forest of Germany, (Heinz, 1981(Heinz, , 1982Kotterba, 1996) and it is possible her conflation with a British deity -like Andraste -provided Diana with a ready-made iconography and cult structure. Indeed, while hare brooches are not uncommon in southern England (particularly east of the Fosse Way), there are particular concentrations in Suffolk and Norfolk (Mackreth, 2011, vol. 1, p. 183), areas dominated by the Iceni, further strengthening Andraste's candidacy as the pre-Roman deity hybridised to from a British reflex of Diana Artemis.
Given both Andraste and Diana Artemis were linked to violence (via warfare and hunting respectively), it seems likely that Diana Andraste's semantic centre also exhibited this association. However, while the hare is known for its violent spring boxing, most of the hare brooches give little sense of movement (Mackreth, 2011, vol. 2, plate 124-125), and some seem to show it hiding (Mackreth, 2011: 14434, 15411, vol. 2, p. 128), a motif also witnessed on a tombstone from Housesteads (RIB: 1618; CSIR 1.6: 198, p. 82), where the hare may have been understood to represent fecundity in the afterlife. Alternatively, this cowering may have been intended to draw on Italic traditions of Diana as the patron of the non-elite population (Beard et al., 1998, p. 3) and as a provider of sanctuary to the pursued, as witnessed in Pausanias' second-century BC account of ritual combat through which an escaped slave could become the priest of Diana Nemorensis (Helládos Periēghēsis II.27.4;Johns, 1918, p. 393). Thus if we are correct in identifying the Romano-British Diana Andraste reflex as the deity behind the hare-shaped plate brooches, we therefore believe her semantic centre lay in protection from violence and the granting of sanctuary.

Early Medieval England: Ēostre
The very existence of the goddess Ēostre in pre-Christian England is controversial. Bede, writing in 725 AD, claimed the 'English' name for the month when Easter occurred was Eosturmonath (De temporum ratione, 15; Jones, 1997, pp. 329-322), apparently a name derived from the pre-Christian spring festival dedicated to the eponymous deity. The existence -and subsequent Christianisation -of such a goddess might go some way to explaining why the English language went against the grain in developing a vernacular term for the Christian spring festival instead of borrowing the Latin pascha, as seen in most Germanic languages (Yorke, 2006). However, Bede's desire to conform to contemporary scholarly mores (Lendinara, 1991) provides motivation for him to have exaggerated or even invented Ēostre, and there has been some scepticism regarding her historicity (Knobloch, 1959, pp. 42-44;Page, 1995, p. 124;Udolph, 1999;Weinhold, 1869, p. 52). On the other hand, the heterogeneous religious cultures of early medieval England (Carver et al., 2010;Wilson, 1992), which were set in a landscape far less densely populated than during the Roman occupation, provide ample space for the development of unusual deities, and convincing arguments have been made that link Ēostre with a localised religious tradition in Eastern England.
Early scholarship focused on Ēostre's etymological links to Germanic terms for 'east' (Grimm 2007, p. 240-241; a connection still commonly accepted today; Orel, 2003, p. 30;Shaw, 2011, pp. 55-59), and attempted to link her to a group of Indo-European goddesses whose names designate 'dawn' in local languages as a Germanic reflex of an Indo-European dawn goddess *h a éusōs (Mallory & Adams, 2006, pp. 409, 432), alongside Ushas (उषस् ) in Sanskrit, Éōs (Ἠώς or Ἕως) in Greek, Aurora in Latin, and potentially Aušrinė in Lithuanian (Greimas, 1992, p. 77;Helm, 1950;Shaw, 2011, p. 55). However, Ēostre's name was not the local vernacular term for daybreak (Old English daegrǽd < OE daeg [day]), and there is no evidence that her cult was widespread in either England or on the continent. Thus it has been suggested that Ēostre was likely a highly localised goddess (Shaw, 2011, pp. 49-71), perhaps similar to the etymologically analogous continental matronae Austriahenae (Kolbe, 1960, p. 58;Neumann, 1987, pp. 109-111;Roymans, 1990, p. 50). In this model, Ēostre's name derives not from Old English ēast [east], but rather from *ēastor [from the east] (Shaw, 2011, pp. 55-61), suggesting she may have been a tutelary deity for a sub-tribal social unit, the *Ēastorwara, based in an area in eastern Kent that likely included the modern village of Eastry -one of three English place names that share an etymological root with her name (fig. 6). In this model, Bede presented a local Kentish name for what he likely knew as Aprīlis mensis or Aprēlis mōnaþ [the month of April] as part of his efforts to describe the pre-Christian insular calendar. Bede is known to have acquired source materials for another of his works, the Historia Ecclesiastica, from the church centre at Canterbury in eastern Kent (Brooks, 1989, p. 59), which might explain how he came to know of Ēostre in the first place (Shaw, 2011, pp. 64-67).
We therefore accept Ēostre as a historical deity, and as a tutelary goddess her semantic centre would have been focused not on a particular function or role, but on her congregation: she was a local goddess, for local people. However, despite the clear link between lagomorphs and contemporary British Easter traditions (Lauritsen et al., 2018), there is no evidence whatsoever to link Ēostre to hares, and popular suggestions to the contrary appear to be based on misunderstanding or projection. How long the Easter Hare may have existed before it was recorded in seventeenthcentury Germany is unclear (von Franckenau, 1682, p. 10); as are its relationships with the early twentieth-century Osterfuchs (Easter Fox) recorded in north-western German folklore (Schnitzler, 1959) and the seemingly Victorian Easter Bunny in Britain.

Medieval Wales: St. Melangell
Of the potential British 'hare goddesses' examined here, the longest serving appears central to the cult site of Pennant Melangell, Wales (fig. 8). This site is located at the head of a steep-sided glacial valley in the Berwyn Mountains in Montgomeryshire, and presently takes the form of a twelfth-century stone church dedicated to St. Melangell (figs. 9-10; Latin Monacella). Excavations under the church have identified a number of charcoal-filled pits from the Middle Bronze Age (Britnell et al., 1994, pp. 54, 90), charcoal samples from which were carbon dated to 1514-1119 BC (Britnell et al., 1994, p. 95). Fragments of human bone suggest these pits were cremation pyres, although no burials have yet been identified (Britnell et al., 1994, p. 91). On the basis of comparable sites elsewhere, the excavators suggest there may have been a central burial mound on the location now occupied by the church, but there is no evidence the site was settled during this period , indicating Pennant Melangell likely served as a high-status burial site. There is no evidence the site saw use thereafter until the early Middle Ages, from when a number of early Christian graves have been identified, the variant alignments of some graves indicating they predate the present church. This structure may have replaced an eleventh-century timber-framed church (Britnell et al., 1994, pp. 90-94), likely constructed in an attempt to Normanise the pre-existing Christian burial site. There is therefore evidence suggesting that Pennant Melangell was a long-standing sacral site, and that even if it did not see continuous use between the Bronze and Middle Ages, its previous role as a liminal point in the landscape -between lowlands and highlands, living and deadfacilitated its adoption as a Christian centre in the Middle Ages.
The medieval focus of this sacred site was St. Melangell, who, while not technically a deity in her own right, acted as a recipient of cult and an intercessor with the divine, making her functionally equivalent to the other supranatural figures examined in this study. A late-fifteenth or early-sixteenth century Latin account, the Historia divae Monacella, draws on earlier sources of unknown date to outline Melangell's alleged life. It claims she was a sixth-century Irish king's daughter who settled at Pennant as a hermit, where she protected a fleeing hare from a hunting prince. Struck by her piety, the prince granted her land to found a religious community on the site (Pryce, 1994). The Historia is primarily a hagiographic text, and the literal historicity of its account is doubtful (Pryce, 1994, pp. 33-34). Notably, there is no archaeological evidence of early medieval occupation at Pennant, meaning that unlike comparable sites elsewhere in Montgomeryshire (Williams, 1990, p. 6), the site does not appear to have hosted a clas -Welsh cenobitic institutions which were incorporated into the Cistercian Order en masse during the twelfth century (Aston, 2001;Davies, 2003;Golding, 1996, pp. 101-105;Henken, 1983Henken, , 2003. Any religious community must therefore have been small and intensely local. Nonetheless, the presence of pre-Norman Christian burials combined with the construction of an expensive stone   church in the twelfth century suggest Melangell's cult may have been long standing. Indeed, the complex includes Ffynnon Cwm Ewyn, a spring (inaccessible due to forestry work during the authors' visit in May 2018) purported to offer remedies to rheumatism, scrofula, and skin conditions during the Middle Ages (Evans, 1994, p. 12;Jones, 1992, p. 202;Cyffin, 1910, p. 241, Rhŷs et al., 1911. Healing wells are well known in the North Atlantic, and some -such as St. Winefride's cult at nearby Holywell (Scully, 2007;Webb, 2002) -became highly successful pilgrimage centres in the Middle Ages (Cormack, 2007;Healy, 2001;Jones, 1992;Logan, 1980;Morris & Morris, 1982;Ray, 2014). Melangell's connection to Ffynnon Cwm Ewyn is less emphatic than St. Winefride's to her spring, but the parallel does raise the prospect that a historical hermit inspired the medieval cult (Golding, 1996;Pearce, 2003;Henken, 2003). Alternatively, it is possible that the medieval Melangell figure is the result of widespread missionary efforts to Christianise sacred waters in Britain, which repackaged pre-Christian gods and spirits in acceptable Christian terms (Green, 1995, pp. 188-202;Lawrence, 1996;Pearce, 2003;Ross, 1962). Either way, her worship appears to have been well established and protracted, and deeply connected with the local landscape.
This connection is borne out in Melangell's embodiment of local social authority: her Historia's use of the hagiographic "hermit and hunter" motif -where the saint protects a fleeing animal from a pursuing hunter (Alexander, 2008, pp. 113-131;Golding, 1996;Salisbury, 1994) -reflects twelfth-century tensions between ecclesiastical and secular powers, which were clashing over rights to marginal landscapes (Golding, 1996, pp. 114-118). Hunting in such landscapes, like those surrounding Pennant Melangell, became a key activity for the Welsh aristocracy in the twelfth and thirteenth centuries (Jenkins, 1993). What is unusual about the Melangell tradition is how the Historia stresses local authority at the site, rejecting regional judicial and religious jurisdictions, and claiming Melangell's lay successors at Pennant had the right to grant "asylum, refugium et tutela" [perpetual asylum, refuge and protection] to those fleeing persecution (Pryce, 1994, p. 38). This local character and connection to sanctuary has persisted to modern times, today expressed in a church-run mental health retreat hosted in a nearby cottage (Keulemans & Burton, 2006).
Despite this interest in Pennant Melangell's local character, Melangell is portrayed as the patron saint of hares, and her cult has made significant use of lagomorph imagery since at least the construction of late fifteenth-century rood screen depicting events from the Historia (with which it might be contemporary; fig. 11; Ridgway, 1994, pp. 130-135). Similar imagery is found in modern emblems and souvenirs (figs. 12a-d; Ridgway, 1994;Parkinson & Watkinson, 1994). An eighteenth-century account of local folklore further connected Melangell to both hares and the offer of sanctuary, noting that hares were locally called Mwyn-Melangell [Melangell's lambs], and that [t]ill the last century so strong a superstition prevailed, that no person would kill a hare in the parish; and even later, when a hare was pursued by dogs, it was firmly believed that if any one cried 'God and St. Monacella be with thee', it was sure to escape (Pennant, 1883, vol. 3, p. 164). In this, Melangell is represented as an intermediary between humanity, animals, and other supranatural powers, perhaps reflecting the liminal location of Pennant Melangell itself. We therefore believe that despite Melangell's close connection with the hare, her semantic centre is clearly located in her role as a tutelary figure of a particular placethe cult site at Pennant -offering sanctuary and protection within a prescribed area. In this, the hare is the ideal cult symbol, reinforcing her links to both local wilderness and hunted prey.

Information-Age Britain: Easter, an Internet Meme Goddess?
Easter appears as a recently-constructed 'pagan goddess' in contemporary Britain -that is, an imagined figure conceived as a deity by a population, few of whom actively worship her (Cusack, 2007), but who employ her to fulfil a specific cultural role. For Easter, this is as the focal point of a widespread annual debate on the origins of the Christian spring festival and its putative links with early medieval Ēostre, the Babylonian goddess Ishtar, and their alleged lagomorph attendants (Lauritsen et al., 2018). This discourse is reflected in traditional media (e.g. Telegraph Reporters, 2018) but has gained particular traction on social media, where memes (in the general sense of images containing short texts) express a variety of views, some more academically sound than others (figs. 13a-c). Many of these texts engage directly with one another, and some are explicitly palimpsestic ( fig. 13d), making efforts to 'correct' perceived inaccuracies in the cultural consciousness. Likely due to the largely urbanised and increasingly irreligious nature of modern British society (Cotter, 2016;Cotter, Quadrio, & Tuckett, 2017), the figure of the 'British hare goddess' has thus been adopted and updated to reflect concerns more relevant to the twenty-first century, where she expresses anxieties regarding the reliability of information in the digital space and the origin of contemporary cultural traditions. Easter thus does not mediate between humans and other animals, but rather allows humans to mediate their connections to and memberships of particular social groups, and we would therefore argue her semantic centre is rooted in identityor the lack thereof.

Conclusion: Shifting Baselines of Violence Mediation
Having established the semantic centres of five potential 'British hare goddesses', we can now turn to the comparative longue durée. Four of the five figures we have examined exhibit contemporary links with hares. The only exception is Ēostre, whose association with lagomorphs is anachronistic. It is nonetheless possible to trace a shifting baseline between the remaining deities: in the Iron Age, Andraste's role was based in warfare, thus mediating between her worshipers and their armed foes (notably the Roman Empire). In Roman Britain, the nominal peace enforced by Imperial rule channelled violence into hunting, and Diana Andraste's cult appears to have looked to the goddess for protection, as depicted on some of the lagomorphic plate brooches. In medieval Wales St. Melangell's semantic centre was based on the sanctuary of her cult site, likely expressing her congregation's concerns regarding land usage and the distant religious and political authorities that claimed jurisdiction over Pennant Melangell. Finally, modern conceptions of Easter employ her to express anxieties relating to social identity in the Information Age, using her to mediate the variant more-or-less British, more-or-less Christian identities of their own selves and their wider society; and by implication the threats of violence that all too often accompany less-British and less-Christian identities. We would therefore conclude that these four distinct 'hare goddesses' all express their congregations' concerns regarding the mediation of violence between the human in-group on the one hand, and other parties (human or animal) on the other. This is not to say that these supranatural female figures are reflexes of a single continuous 'British hare goddess' worshipped unceasingly for over two millennia. Our study has shown that the primary characteristics and concerns of these deities are diverse and not grounded in the hare. Rather, the lagomorph serves as a reflection of their semantic centres: hares are highly fecund, and can be violently competitive, but all four cults regarded the hare as a prey animal, and thus a useful symbol on which to project social anxieties about violence. As such, our analysis reinforces two key tenets of social zooarchaeology and the humanistic animal turn: firstly that animals are more than sources of calories and raw materials (as they were viewed in traditional zooarchaeology), and need to be understood "in their broader social context" (Russell, 2011, p. 400). Secondly, that animals are more than passive bearers of human symbolism (as they have often been regarded in traditional humanistic disciplines; Petersen, 2016), and that humans and animals have complex bilateral relationships, where animal behaviours shape human actions and attitudes, which in turn shape the lived reality of animals. Our study has further explored how human societies, particularly those in a common geographic area, have employed religion to negotiate their direct relationships with animals, but also, through those relationships, their own broader ecological and ontological contexts. We therefore hope that this article will prove beneficial not only in understanding the specific case of the 'British hare goddess' and her concern with the mediation of violence, but also demonstrate the viability of comparative archaeological approaches to human interactions with the supranatural, animals, and their shared ecological context in the longue durée.
Acknowledgements: Parts of this research were funded as part of the Exploring the Easter E.g. -Shifting Baselines and Changing Perceptions of Cultural and Biological "Aliens" project, funded by the AHRC (project reference AH/N004558/1). Publication costs have been provided by the Religious 'Stuff': The Cognitive Science of Ritual Props in the Romanisation of Britain and the Christianisation of Iceland project, funded by the University of Iceland's Division of Science and Innovation. All persons who meet authorship criteria are listed as authors, and all authors certify that they have participated sufficiently in the work to take public responsibility for the content, including participation in the concept, design, analysis, writing, or revision of the manuscript. Murphy and Ameen share responsibility for the conception and design of study; Murphy takes responsibility for the acquisition and analysis of data, the drafting of the manuscript, and the revision of the manuscript for important intellectual content; Murphy and Ameen share responsibility for the approval of the version of the manuscript submitted. We wish to thank a number of colleagues whose input has been invaluable, especially, Naomi Sykes, Greger Larson, Phil Shaw, Sean Doherty, Tom Fowler, Joel Alves, and Adrian Bott, as well as two anonymous peer reviewers.