A Local Barrow for Local People? The Ferry Fryston Cattle in Context

David Orton

Introduction It is an oft-noted point that animal bones from archaeological sites are traditionally used primarily to address questions of subsistence. Recent work on both the Roman period and the Pre-Roman Iron Age has stressed that even apparently mundane deposits of animal bones and other waste should not in fact be seen as the passive results of economic processes, but rather as the meaningful products of structured social action (e.g. Grant 1991; Hill 1995; Lauwerier 2004; Richardson 1997). Nonetheless, there remains a relative reluctance to interpret even large, articulated or otherwise unusual deposits of animal bones from Roman rather than prehistoric contexts in non-functional terms (see Morris, in prep.). The mass deposit of cattle remains at Ferry Fryston presents an interesting case in which any non-ritual explanation would be almost impossible to uphold, but where the nature of the practices involved is open to considerable debate.

The site of Ferry Fryston, initially referred to as Ferrybridge, was widely publicised in both the specialist and non-specialist media as a possible case of Iron Age feasting following its discovery in 2003. Since then, post-excavation work and a series of radiocarbon dates have altered this picture considerably. This paper provides a brief overview of Ferry Fryston before presenting a re-interpretation of the cattle deposit based on the latest evidence. Finally, consideration is given to the possible implications of the site, and how it might be interpreted with regard to wider questions of local and regional identity in Roman Britain. Ferry Fryston was excavated by Oxford Archaeology North on behalf of the Highways Agency. The bones were studied primarily by Andrew Bates, while Gillian Jones and the author carried out age-at-death studies on the main cattle assemblage. In the latter case, the study formed the basis of a master’s dissertation at the University of York. As such, it should be made clear at the outset that while this paper draws heavily on both Bates’s and Jones’s work, the views expressed do not necessarily reflect theirs, or those of Oxford Archaeology.

Ferry Fryston The Ferry Fryston square barrow was uncovered during the M1/A1 motorway extension in West Yorkshire. The excavation revealed a Middle Iron Age chariot burial with a single inhumation dated to the second to fourth centuries B.C. (352–291 cal B.C. plus 227–165 cal B.C.; 2168±20 B.P., 2 σ calibrated to 95%). This is an interesting discovery in itself since the site lies well outside the usual distribution of Arras Culture chariot burials in East Yorkshire, although an example was excavated even further afield near Edinburgh in 2001 (Carter and Hunter 2003).

More interesting in the current context, however, is the enormous assemblage of bones found deposited in the upper fill of the ditch around the burial, designated as SG 32. A total of 12,779 fragments were recovered from this deposit, of which Bates identified 2816 to species (Table 1). Of these, 2807 definitely derive from cattle, with only 9 fragments positively identified as other taxa. The identified bones were predominantly mandibles, although most parts of the body were represented to some extent. The absolute minimum number of individuals represented was calculated as 162.

Table 1: Species representation in SG 32 (courtesy of Andy Bates) Species NISP Horse 1 Cattle 2807 Pig 1 Sheep/Goat 5 Sheep 1 Red Deer 1 Cattle/Horse 1 Cattle/Red Deer 602 Medium Mammal 2 Large Mammal 3266 Unidentified Mammal 6092 Total 12779

The cattle remains were initially assumed to have been deposited within living memory of the inhumation, possibly in some kind of funerary feast or other commemorative event. On the basis of the stratigraphy, this was suggested to have taken place ‘perhaps over the course of a few days’ (Murray et al. 2004: 9). The first radiocarbon date obtained on one of the bones from the upper fill, however, placed it firmly within the Roman period, probably within the late first or early second century (70–176 cal A.D. plus 190–212 cal A.D.; 1890±21 B.P., 2 σ calibrated to 95%), two to five centuries later than the central inhumation. Obviously a Roman date does not preclude a feast or short-term event. However, several more recent lines of evidence suggest a rather different interpretation.

1) Age-at-death analysis Age-at-death studies were undertaken by Jones and the author with the aim of testing a ‘singlekill’ hypothesis. While cattle can technically calve all year-round, it is usually in the farmers’ interests strongly to encourage a natural tendency towards spring births. While we have no direct textual sources for Romano-British animal husbandry, both Classical (e.g. Columella De Re Rus. VI. 24. 1–2) and Early Modern British authors (e.g. Fitzherbert 1882 [1534]: 57) advocate exclusive spring birthing, and the reasoning behind this would have applied equally in Roman period Yorkshire. Assuming that cattle births are concentrated strongly in the spring, and given a suitably precise technique for age estimation, a single mass slaughter, even spread over days or weeks, should result in clear age cohorts within the assemblage. An event in autumn, for example, should result in cattle aged around six months, eighteen months, and thirty months and so on. Of course, a repeated cull over a number of years would produce the same result if very tightly seasonal. Nonetheless, the analysis of age-at-death has the potential to determine whether an assemblage could or could not have derived from a single slaughter, provided that sufficiently high resolution can be achieved.

Figure 1: Estimated mortality profile for the Ferry Fryston cattle based on radiographic assessment of tooth development (for full explanation see Orton 2004). 0 5 10 15 20 25 30 35 40 45 A (H 0-½mo) Bb-g (H 1, 2) Bt (S 5-6; H 6) Cb (H 6, 8, 8) Cd (H 9, 10) Ce (H 9, 9) Cf (H 12, 13) Cg (H 12) Ct (S 15-18; H 13-18) Db (H 18) Dc Dd (H 18) De Df (H 20,21,22) Dg (H 22) Dt (S 24-30; H 23) Eb (H 28-32) Ec Ed (H 31,32,36) Ee Fe (H 3y2m) Ff (H 4y) G (H 3y5m – 7y) M2xyz (H 5-10y ) M3x (H 9-13y) M3y M3z (H 12y) M3zm+ (H 16,19y) Mandible wear stage Frequency M3 b M2 b M1 b

Figure 2: Mortality profile for the Ferry Fryston cattle based on tooth wear (courtesy of Gillian Jones. Numbers in brackets are suggested ages. For definitions of stages, see Bates et al. forthcoming; Jones and Sadler in prep.). 0 1 2 3 4 5 Age-at-death (years) Estimated relative frequency (arbitrary units)

This is not the appropriate forum to detail the methodologies employed, but one study used a refined version of a conventional eruption and wear stage scheme (Bates et al. forthcoming; Jones and Sadler in prep.), while the other relied on measurements taken from digitised radiographs of the mandibles (Bates et al. forthcoming; Orton 2004). The studies produced consistent results, with both demonstrating a strong clustering of the cattle into age groups of a little under 1.5 and 2.5 years (Figs. 1 and 2). In both cases, however, a considerable number of individuals fall between these groups. The results are therefore more supportive of a phenomenon with a fairly strong seasonal, probably summer/autumn focus than of a single one-off event.

Atmospheric data from Reimer et al (2004);OxCal v3.10 Bronk Ramsey (2005); cub r:5 sd:12 prob usp[chron] Ferrybridge 1500BC 1000BC 500BC BC/AD 500AD Calendar date Sequence {A= 99.7%(A’c= 60.0%)} Boundary _Bound Phase Burial Cattle in lower fill 1 96.6% Cattle in lower fill 2 98.6% Pig humerus in grave 104.2% Pig skull in grave 97.9% R_Combine Inhumation 97.2% Phase SG 32 Cattle bone 1 (R-28487/1) 99.3% Cattle bone 2 (R-28802/3) 100.1% Cattle bone 3 (R-22251) 99.5% Cattle bone 4 (R-22252) 102.8% Cattle bone 5 (R-22250) 103.2% Boundary _Bound

Figure 3: Calibrated radiocarbon dates for Ferry Fryston.

2) Radiometric dating Figure 3 shows the full suite of dates currently published for Ferry Fryston. The original determination from the cattle in SG 32 is clearly inconsistent with those obtained more recently, the latter suggesting a third or fourth century date for the deposit. Six further determinations obtained very recently corroborate this picture, with one roughly equivalent to R-28487/1, four from the third to fourth centuries, and one intermediate (J. Montgomery pers. comm.).

3) Body part representation Andrew Bates’s work on body part representation shows a very distinct pattern within the Ferry Fryston assemblage (Fig. 4). Heads are most abundant, and a reasonable number of the first two neck vertebrae are also represented, presumably having been included with the head following decapitation in some cases. Other vertebrae are extremely rare. Forelimb elements are common, especially the scapula, while hind limbs are all but absent. The most striking point, however, is that the forelimbs are overwhelmingly right-sided, with very few left-sided post-cranial elements making it into the assemblage at all. This pattern clearly indicates the deliberate selection of elements, and is reminiscent of that seen for sheep at the Romano-British shrine of Great Chesterford in southern England. In this case, mandibles dominated the assemblage, followed by metapodials and then by right forelimbs (Legge et al. 2000: 155; Baxter forthcoming). The apparent selection by side, albeit less clear-cut, is also seen at the Harlow Temple, where right mandibles are consistently at least twice as abundant as lefts in all phases (Legge and Dorrington 1985: 124).

0 20 40 60 80 100 120 140 160 180 Skull (Maxilla) Mandible Atlas Axis Cervical Vertebra Thoracic Vetebra Lumbar Vertebra Scapula Humerus Radius Radius/Ulna Ulna Metacarpal Carpal Pelvis Femur Tibia Metatarsal Calcaneus Astragalus Phalanx 1 Phalanx 2 Phalanx 3 Element Minimum Number of Elements Left MNE Right MNE MNE

Figure 4: Body part representation among the Ferry Fryston cattle (courtesy of Andrew Bates).

Ritual? The deliberate selection of body parts need not relate only to ‘ritual’ activities, and indeed an assemblage with an excess of heads and forelimbs at third/fourth century Wigford, Lincoln, has been interpreted as secondarily deposited butchery waste (Dobney et al. 1996: 24). Such an interpretation would be extremely difficult to maintain for Ferry Fryston, however. While overrepresentation, even to the point of exclusivity, of fore- versus hind-limbs might be explicable in terms of butchery and processing practices, it is hard to see how practical concerns could extend to the selection of right-sided elements over left. Given the extreme right:left ratios seen at Ferry Fryston – rising to 129:2 for scapulae (Table 2) – even the most sceptical observer is forced to accept that the selection of elements for deposition did not follow purely functional principles.

Table 2: Right versus left cattle forelimb elements. Element Right MNE Left MNE Total Right/Left Scapula 129 2 131 64.5 Humerus 54 2 56 27.0 Radius 50 2 52 25.0 Radius/Ulna 3 0 3 ~ Ulna 24 4 28 6.0 Metacarpal 40 3 43 13.3 Carpal 7 1 8 7.0

The location of the deposit also militates strongly against a functional explanation. Quite apart from the ‘coincidence’ of its situation on a pre-existing monument, it would be hard to account for the bones’ origin in these terms. If the assemblage did represent butchery waste, then its sheer size and the almost exclusive presence of a single species would suggest the byproducts of large-scale production, and there is simply no plausible source for such waste. While ‘Romanised’ farmsteads are known at Ferrybridge (Roberts and Richardson 2005: 216– 7), and possibly Ferry Fryston itself (Faull and Moorhouse 1981: 151–2), the nearest appreciable settlement is the fort and vicus 2km away at Castleford, occupied in one form or another for much of the period between 70 and 400 A.D. (Fossick and Abramson 1999: 19). There is in any case no sign of the kind of butchery which one might expect from such a deposit. Meat production waste from urban and military sites in Roman Britain is typically characterised by chopping (e.g. Maltby 1989: 89–91); only one trace of which was recorded in the Ferry Fryston assemblage.

Discussion Having excluded the possibility that the Ferry Fryston cattle represent everyday butchery waste, we can move on to consider the possible form of ritual activity underlying their deposition at the barrow. The discrepancy in radiocarbon dates on the bones from SG 32 requires an explanation, and four possible scenarios can be outlined. Firstly, the deposit may have formed over a substantial period of time, spanning much of the second, third and possibly fourth centuries, with the periodic addition of fresh cattle remains, probably on a more-or-less seasonal basis considering the age-at-death results. This is currently the excavators’ preferred interpretation (Bates et al. forthcoming). Secondly, the early dates may be misleading. Carbon dating is of course a probabilistic method, and in theory the true date might lie in the right-hand tail of the early dates and the far left hand side of the others. Given the number of later dates, however, and the fact that only some overlap at all with the earliest at 95%, the odds against all these determinations representing the same underlying event must be astronomical. Thirdly, one cannot rule out the possibility that the early dates came from residual bone fragments. There was some evidence for residuality in the assemblage, with a few beaker pottery fragments and some lithics found alongside the bones, and the nine definitely non-cattle bone fragments may well reflect the same phenomenon. It is possible that a few of the cattle bones entered the deposit in the same way, but given the very limited evidence for residuality and the general apparent integrity of the deposit, it would seem exceptionally unfortunate for two of these to have been selected for dating. Finally, some of the bones may have been curated elsewhere for a considerable length of time before their eventual deposition at the barrow in a single episode.

The fourth explanation is advocated here for a number of reasons. Most strikingly, the original assessment report by Oxford Archaeology North argued very strongly for the shortterm formation of the deposit on stratigraphic grounds. The deposit was remarkably dense and homogeneous, containing no finer silt lenses as might be expected from gradual accumulation (Murray et al. 2004: 9). Terry O’Connor (pers. comm.) drew the same conclusions following a site visit, noting that should the cattle prove not to have come from a single kill, then their deposition cannot have been primary. Since the radiocarbon dates and the age-at-death results do seem to militate against a single kill, the possibility that at least some of the bones represent a secondary deposition in the ditch must be considered. Put simply, either the excavators’ original assessment of the deposit was off the mark, or the dates are misleading, or secondary deposition took place. Given the calibre of both the excavators and the radiocarbon lab, the latter seems the most plausible explanation.

Support for this explanation is provided by the lack of contemporary artefacts in the deposit, and the very small number of non-cattle bones: these are not suggestive of a deposit left open for a period of centuries or even of decades. In addition, there is practically no evidence for gnawing on the cattle bones. At the temple site of Wanborough, by contrast, considerable evidence for canid gnawing on bones from a foundation deposit suggests that they were exposed on the surface for some time prior to eventual deposition (Nicolaysen 1994). If the Ferry Fryston deposit did accumulate in situ over a long period, then it must have been well protected from carnivores and from the accumulation of general debris.

One might expect to see differences in bone colour and preservation in an assemblage deriving from multiple episodes of activity. While there is substantial gradation in the surface condition at Ferry Fryston, no discrete groupings are apparent (G. Jones, pers. comm.). Since the bones clearly underwent a considerable deterioration after the final deposition, however, any initial differences in patination could easily have been obscured.

The lack of articulation noted in the assemblage also strongly suggests that the remains had been kept elsewhere for at least a short period following use, especially since there is very little evidence for the disarticulation of limbs on the bones themselves. Of the ten cut marks recorded, six relate to decapitation, one probably for the removal of the forelimb, and three to the defleshing of the scapula and humerus (see Bates et al. forthcoming). No marks consistent with disarticulation within the forelimb were noted. The original arrangement of body parts might alternatively have been disturbed by exposure to the elements and occasional scavengers, but one would still expect to see some articulated units in a primary deposit of this size. If one accepts that cattle remains were in any case deposited secondarily well after slaughter, the idea of long term curation followed by an eventual single episode of deposition becomes less of a jump.

Finally, the curation of some of the cattle bones simply is not as hard to swallow archaeologically speaking as it might initially sound. Examples of apparent curation and redeposition of animal bones may be found in numerous periods and regions, from Beaker Period Northamptonshire (Davis and Payne 1993), to Late Iron Age France (Meniel 1985: 144–6). There is no reason to assume that similar practices could not have taken place during the Roman period in Britain, and potential comparisons with Romano-British temples become relevant here. King (2005) has recently provided a comprehensive review of animal bone assemblages from Romano-British sites interpreted as shrines or temples. While there is no suggestion of curation at any of these sites (although note the evidence for the redeposition of bones at Wanborough, mentioned above), many of the assemblages are suggestive of consistent patterns of sacrifice and deposition over a considerable period of time, with at least some of the resulting remains retained within the religious compounds.

The similarity in body part representation between Ferry Fryston and Great Chesterford is clearly very interesting in this light, and further parallels with southern Romano-British temples are seen in the age profiles. At Harlow, all the sheep appeared to have been killed in the autumn (Legge and Dorrington 1985: 131–2), while at Great Chesterford there were apparently two seasons of slaughter, one in the autumn and one in the spring (Legge et al. 2000: 154). On the basis of both epigraphic and environmental evidence, Isserlin (1994: 49) has suggested that spring and autumn may have been peak periods for various forms of ritual activity in the Roman period, including monument dedication and sacrifices. Autumn killing of caprines has also been noted at Uley (Levitan 1993), Chelmsford (Luff 1992), and more tentatively at Hayling Island (King 2005: 338). As noted above, the Ferry Fryston cattle appear mostly to have been killed in the late summer or autumn. In the light of these comparisons it is worth noting that an undated square post-hole structure 29m from the barrow has tentatively been suggested as a possible Romano-British shrine by the excavators (Fig. 5).

In their analysis of a similar but much older assemblage of cattle from Irthlingborough, Davis and Payne (1993) cite a paucity of premolars as evidence for the curation of cattle skulls over time. No such pattern is seen at Ferry Fryston. Despite the poor surface condition, the mandibular tooth rows, at least, were surprisingly complete. While this could be taken as evidence against their curation, it might alternatively suggest a relatively careful storage of the bones, perhaps in some form of ossuary, rather than the simple disturbance and redeposition of a gradually accumulated deposit. The lack of artefacts or other species does not, after all, suggest secondary middening. On the other hand, if only a small fraction of the bones were curated or redeposited, then this would not in any case affect the overall appearance of mandibular completeness in the assemblage.

The interpretation of the nearby structure must remain speculative due to a lack of artefacts or dateable material, but is it plausible that it might represent a shrine or ossuary associated with the barrow? The similarities with Great Chesterford in terms of body part representation are persuasive but certainly not conclusive. King’s (2005) review of faunal assemblages at temples and shrines failed to identify any cases in northern England, but this seems just as likely to relate to preservation, excavation and publication biases as to any actual archaeological pattern.

Figure 5: Plan of the Ferry Fryston site, showing the square barrow (2230) and undated structure (2242) (courtesy of Oxford Archaeology).

The view is taken here that the question ‘is it a shrine?’ is itself somewhat misleading. After all, the categories of shrine and temple in Roman Britain are far from unproblematic. Forcey (1998: 87) has argued that the religious and mortuary domains cannot truly be separated, and that Romano-Celtic temples in particular are fundamentally associated with the dead, although one might argue that if this was the case, then we should expect to see human bones more frequently associated with them. The temples, in his view, are foci for the commemoration of events and of people who may belong to the mythical rather than historical past, and he cites the location of the Harlow and Lancing temples close to Bronze Age barrows as examples of this (Forcey 1998: 92). Following this line of thinking, Ferry Fryston would appear to have many of the salient features of sites such as Harlow even if no associated temple or shrine building was ever actually present.

To summarise, a range of depositional models for the Ferry Fryston assemblage may be put forward, the best supported of which involves the accumulation and curation of selected cattle body parts for a considerable period of time, possibly in some kind of ossuary associated with the barrow. Eventually, at some time in the third or fourth centuries, these were deposited in the ditch in a single event, probably alongside a considerable number of relatively freshly killed animals. It should be noted that none of the data provided here implies a continuous process of accumulation, still less an even distribution of additions over time. On the contrary, it seems likely from the few dates available, that a large proportion of the cattle were killed towards, or at, the end of the sequence. Continuity is in any case almost impossible to demonstrate from this kind of assemblage regardless of the dating budget (Bayliss and Orton 1994). This is not a problem, however, since the interpretations offered below do not require that it be demonstrated; the very act of curation implies a form of continuity regardless of the regularity with which new remains were added.

Implications With regard to the wider implications of the Ferry Fryston cattle, the key feature would appear to be the change in the nature of activity from curation to deposition; an apparent practice of seasonally sacrificing cattle and retaining certain body parts, possibly at a shrine of some sort, gives way to their mass deposition at a prehistoric barrow. Williams (1998: 77) has suggested that ‘ancient structures may have held an important place in the construction of local and regional identities’ in Roman Britain, and that they were probably linked to specific deities or mythical ancestors. In many cases these might be expected to have a particular local or regional significance, and indeed Williams’ map of known cases of monument re-use shows distinct regional patterning.

The large scale deposition of animal remains – some long curated, others perhaps recently sacrificed – at a prehistoric barrow that has stood in the landscape since time immemorial is thus suggestive of a need to re-assert local identity, to renew ties with the landscape and the ancestral past. Since this change in ritual practice at Ferry Fryston appears to have taken place at some point in the third or fourth centuries, it brings to mind Scott’s theory that a revitalization, or ‘nativism’, movement took place in later Roman Britain; Scott describes a model from the anthropological literature in which:

A people who have been subjugated by an occupying force and administration start to bring back and re-work old myths, incorporating new material culture, in particular at times of economic stress and occasionally after hundreds of years. (1991: 119).

She goes on to suggest that certain changes in Romano-British ritual practice, particularly animal and infant burials at villas, can best be understood in terms of just such a phenomenon. Such revitalization movements apparently encourage the re-adoption of old customs, and may try to revive certain desirable elements of the perceived ancestral past. One might therefore expect them to be reflected, amongst other things, by the intensification of activity at ancient sites, with their likely ancestral connotations. Just such a trend was noted for later Roman Britain by Dark (1993: 136) in a review of monument re-use.

While there is much to be said for this, one would do well to avoid the idea of a provincewide movement. Scott’s model could be taken to carry implications of a kind of ‘pan- Britishism’, a widespread reassertion of a uniform perceived ‘native’ Iron Age identity as contrasted with the reality of Roman occupied Britain in the third and fourth centuries. Williams (1998: 77–78) has criticised this view on the basis that it places too much emphasis on links with the Iron Age, leading to unwarranted generalisations regarding the motivations for monument re-use across Britain. It seems far more likely that any intensification in ritual activity linked to a reassertion of identity would be a localised process, with communities feeling the need to stress their ancestral claims to the local landscape, and to the myths, deities and ancestors associated with it, through a wide variety of practices which may or may not have involved any explicit or even implicit reference to the pre-conquest past.

To the extent that any revitalization phenomenon was province-wide, this is much more likely to reflect broadly similar reactions to broadly similar conditions of social insecurity than any coherent movement celebrating native British identity. This is not to say that the trend identified by Scott is not real or important, only that it can be all too easy to slip into thinking in terms of conflict, overt or otherwise, between ‘Roman’ and ‘native’ identity as monolithic blocks, a point that is relevant in all areas of Roman period archaeology, not just those concerned with animal sacrifice or monument re-use. To the present author’s mind, intensified activity at prehistoric monuments such as Ferry Fryston highlights an altogether more specific, local aspect of identity, in which contrasts between the rather dubious categories of ‘Roman’ and ‘native’ could potentially have been far less important than those between, for example, ‘local’ and ‘stranger’.

Of course, this model does not require a distinct local tradition of animal sacrifice, and indeed the apparent pattern at Ferry Fryston does have parallels with sites in other parts of the country in terms both of seasonality and of the selection of body parts. The affinities with Harlow and Great Chesterford, for example, have already been noted above. What is important is the increased focus of ritual activity on the barrow itself, as represented by the sudden decision to switch from curating the cattle remains to depositing them en masse in the ditch. One possible challenge to the ‘local’ nature of activities at Ferry Fryston comes from strontium isotope analysis (Montgomery et al. forthcoming). Samples taken from six of the cattle produced very varied results, suggesting that they did not derive from a single herd. None of the measurements were consistent with origins in the immediate vicinity of the site: four of the animals could potentially have been raised on nearby geological zones, but two had values higher than any yet known from Britain. Since a similar measurement was obtained from the human burial in the barrow, and such values are rare even outside Britain, it is currently hard to know how to interpret this data. The substantial variation in tooth measurements also suggests dispersed origins for the cattle (G. Jones, pers. comm.), but this dispersal could be in time as much as in space.

The Ferry Fryston cattle also have very few, if any, parallels in the pre-Roman Iron Age of Yorkshire, but again actual continuity in ritual practice is not a necessary corollary of an attempt to reassert ties to the past and to the land (Evans 1985: 89). As Scott stresses, it is people’s perceptions of the past than are recreated or manipulated, rather than actual past practices (Scott 1991, 119).

With all of this having been said, it is worth mentioning one interesting, tantalising even, feature of the Ferry Fryston site. While large cattle bone deposits associated with funerary monuments in the Iron Age of Yorkshire are rare, remains from at least five cattle were found in the lower fill of the barrow ditch. Two direct dates on these bones show them to be roughly contemporaneous with the inhumation (Fig. 5). The juxtaposition of two unusual cattle assemblages at the same site is intriguing, but it is hard to see how they might be connected considering the length of time likely to have elapsed between them. On the other hand, since we have no firm date for the start of accumulation of the bones in SG 32, it is just possible that some form of very specific tradition associating the Ferry Fryston barrow with cattle persisted for centuries. In this respect it is worth noting that at both Harlow and Uley the pattern of animal sacrifice began in the Late Iron Age and continued through to the fourth century (Legge and Dorrington 1985; Levitan 1993). While Williams (1998: 76) rightly notes that Roman period monument re-use most often followed long periods of abandonment, this does not preclude the possibility that in some specific cases, a degree of continuity may have existed. Continuity in ritual activity at prehistoric monuments is in any case extremely hard to demonstrate (Meade 2004: 82).

Of course, this suggestion is extremely speculative, and even if there was some degree of continuity in the mythic associations of the Ferry Fryston barrow, this need not imply any continuity in the way that these associations were recognised or used. Whether or not the Roman period activity referenced any previous tradition at the site, I believe the sudden switch from curation to deposition in the third or fourth century can best be interpreted in terms of the negotiation of identity at an explicitly local level.

This is the interpretation of the Ferry Fryston cattle preferred here; but for the sake of balance, it is worth outlining an alternative narrative. Rather than expressing a need to reassert links with the perceived past, the deposition could conversely represent the end of ritual activity at Ferry Fryston as people moved on to new practices; the putative ossuary was cleaned out, and the contents deposited at the barrow in a fitting gesture of closure. Instead of seeing the deposit as an attempt to establish continuity with the past, it might rather represent a clean break with the past. This interpretation holds a certain attraction, and indeed the Ferry Fryston cattle deposit does come at the end of a long sequence of ritual activity in the area (see Roberts and Richardson 2005; Brown et al. forthcoming).

Conclusion The best supported explanation for the Roman-period cattle from Ferry Fryston involves their gradual slaughter over a significant time period, with the curated remains eventually deposited in the barrow ditch alongside those of a number of more recently killed animals. While there is insufficient evidence to interpret a nearby structure as a shrine associated with the barrow, one can nonetheless draw several parallels with certain Romano-British temples, especially with those classified by King (2005: 357–9) as ‘group A’.

The eventual placement of the cattle in the barrow ditch represents a dramatic shift from curation to deposition, the interpretation of which is open to question. While the line taken here is that the deposit reflects a reassertion of local identity though emphasis on ties to the land and the past, it could equally be seen as a radical break with that past. Whichever of these interpretations one prefers, the crucial point is that the negotiation of identity in Roman Britain will often have had a strong regional and/or local aspect that may be obscured by generalist models. Ritual activity at ancient monuments is particularly suggestive of this, and the Ferry Fryston deposit can best be seen in these terms.

Department of Archaeology, University of Cambridge

Acknowledgements Thanks are due to Oxford Archaeology, and especially to Andy Bates and Angela Boyle for involving me in the Ferry Fryston project. The MSc dissertation on which much of the above is based was supervised ably by James Barrett of the University of York, and invaluable practical support was provided by Steve Dobson, Roland Müller, Sonia O’Connor, Terry O’Connor and Joachim Wussow. Useful comments and suggestions on the interpretation were offered by J.D. Hill, Naoise Mac Sweeney, Mark Maltby, Martin Millett and Clive Orton. Ian Baxter, Mandy Jay, Gillian Jones and Janet Montgomery all provided access to unpublished data. The text was revised following comments by the TRAC editorial team and an anonymous reviewer. Finally, the alternative interpretation of the deposit as a break with the past was suggested by Fraser Brown.