Chop and Change: Specialist Cattle Carcass Processing in Roman Britain

Mark Maltby

Introduction Full-time professional butchers are found in most complex societies and particularly in urban centres. It is likely that specialist butchers, who regularly acquire large numbers of animals, will develop processing methods that will be repetitive and possibly distinctive. In addition, it is possible that waste from their activities will sometimes be deposited in large quantities. Urban macelli, where butchers and other specialists plied their trade are known in many Roman towns, including examples in Britain such as Wroxeter (Ellis 2000). This paper explores the extent to which the presence of these specialist butchers can be traced through the analysis of cattle bone assemblages from towns and other civilian settlements. One of the features of animal bone assemblages from Roman sites is the distinctive nature of the processing marks found on cattle bones. Certain types of marks recur frequently in some assemblages, but are much rarer in others. Similarly, there have been numerous discoveries of substantial dumps of cattle remains that indicate the disposal of large amounts of processing waste, and the accumulation of specific types of bones for large-scale processing in some instances. Both of these traits indicate that specialist butchers operated in certain settlements and that they employed standardised methods of butchery using specialist equipment and facilities. In other settlements, however, these processing marks are absent, or restricted in their frequency and distribution. This paper will review the evidence for some of these processing marks from Romano-British civilian sites. Particular attention will be given to three types of marks: (1) superficial blade marks on upper limb bone shafts; (2) axially split upper limb bones; and (3) transversely broken metapodials. The first marks were made during the filleting of meat from the bones; the other two provided access to marrow. The reasons for the variations in the frequency of occurrence of these marks will be discussed, and the economic and cultural implications arising from the studies will be considered.

General trends in Roman carcass processing Previous discussions of changes in butchery practices in Roman Britain (Grant 1989; 2000; Maltby 1989; 1998a) have noted the increase in the incidence of cleaver marks on cattle bones on Romano-British sites compared with Iron Age assemblages. Wilson (1978) in his study of bones from Abingdon, Oxfordshire, first illustrated the types of incisions that have since commonly been described on Iron Age sites in southern Britain. However, quantification of butchery marks is fraught with difficulties because of variations in recording methods and differential preservation. Therefore, statements that chop marks tended to become more common in the Roman period have usually not been supported by quantified data. Table 1 partly addresses this by comparing the number of records of cleaver and (occasional) saw marks with the number of bones bearing knife cuts in several assemblages recently quantified by the author.

Table 1: Cattle bones with chop/saw marks and knife cuts from Iron Age and Romano-British assemblages Site County Period Type Chopped /Sawn Knife Cut % Chopped Source Battlesbury Bowl Wilts Iron Age hillfort 38 249 15 Hambleton /Maltby nd Wellingborough Northants Iron Age rural 12 33 27 Maltby 2003a Lechlade Glos Iron Age rural 3 11 21 Maltby 2003b Biddenham Loop Bedford Iron Age rural 0 13 0 Maltby nd1 Biddenham Loop Bedford Roman rural 16 12 57 Maltby nd1 Owslebury Hants Iron Age rural 45 159 22 Maltby 1987 Owslebury Hants ERB rural 121 486 20 Maltby 1987 Owslebury Hants LRB rural 53 107 33 Maltby 1987 Stanstead ACS Essex LIA rural 13 16 45 Mainland 2004 Stanstead ACS Essex ERB rural 17 35 37 Mainland 2004 Holwell Quarry Herts LIA-RB rural 8 21 28 Maltby nd4 Marsh Leys Bedford Roman rural 37 15 71 Maltby nd2 Shefford Bedford Roman rural 14 11 56 Maltby nd3 Wantage Oxon Roman roadside 37 14 73 Maltby 1996 Caerwent Gwent Roman town 1944 268 86 Hambleton/ Maltby in prep. Winchester Hants Roman town 1869 148 93 Maltby in press

These results largely support the impression gained from previous studies. Although cattle bones with chop marks have been recorded in Iron Age assemblages, they are largely outnumbered by those damaged by finer incisions. Chop marks become more common in Romano-British contexts and are particularly prevalent in urban assemblages such as those from Caerwent (Hambleton and Maltby in prep.) and Winchester (Maltby in press). Significantly, in two assemblages from rural settlements that have Iron Age and Romano- British phases (Owslebury and Biddenham Loop, Maltby 1987; nd1), a higher percentage of chopped bones were encountered in the later assemblages. This is a very crude measure that does not take into account variations in the types of mark occurring in different parts of the body. Nevertheless, it demonstrates that methods used to process cattle changed markedly in Roman towns, and to a lesser extent in rural settlements.

Table 2: Recorded observations of some specific butchery marks in major Romano-British towns Town Site Split Up Limb Blade Up Limb Chopped Ramus Trimmed Scapula Holed Scapula Chopped Fem Caput Source Caerwent Basilica 11% 18% x x x x Hambleton & Maltby in prep. Chichester Cattlemarket x x x x Levitan 1989 Cirencester Cemetery x Thawley 1982 Cirencester Chester St 39% 20% x x x Maltby 1998a Colchester Various x x Luff 1982;1993 Dorchester Colliton Park x x x Hamilton-Dyer 1993 Dorchester Greyhound Yard 17% 21% x x x x Maltby 1993 Exeter Various x x x x Maltby 1979a Gloucester East Gate x Maltby 1983 Gloucester Eastgate St 78% x Levine 1986 Leicester Bonners Lane x Baxter 2004 Leicester Clarence St x Baxter 2005 Lincoln Waterfront x x x x x Dobney et al. 1996 Lincoln Wigford x x x Dobney et al. 1996 London Mithraeum x Macready & Sidell 1998 Southwark NW Southwark Low % x Pipe 2003 Southwark 199 Borough High St x x Locker 1988 Southwark Borough High St x Ainsley 2002 St Albans Folly Lane x Locker 1999 Silchester South Gate x Maltby 1984 Winchester Crowder Terrace x Coy and Bradfield in press Winchester Defences 17% 21% x x x Maltby in press Winchester Staple Gardens 15% 28% x x x Maltby 1986 Winchester Hyde Abbey 16% 16% x x x Maltby in press Winchester Victoria Rd G2-4 x x x Pfeiffer in press Winchester Victoria Rd IVVI 18% 18% x x x Maltby in press Winchester Victoria Rd XXVI 29% 20% x x x Maltby in press Town Site Split Up Limb Blade Up Limb Chopped Ramus Trimmed Scapula Holed Scapula Chopped Fem Caput Source Wroxeter Basilica 21% x x x x x Hammon 2005 York Blake Street x O’Connor 1987 York Tanner Row x x x O’Connor 1988 Up Limb = humerus, radius, femur and tibia; Chopped Ramus = chop marks on or near mandibular condyle; Trimmed Scapula = blade marks on spine or body; Fem cap = femur caput chopped through x = present but not quantified

The distinctive nature of Roman butchery has been noted by a number of authors who have studied military and urban assemblages (e.g. Dobney 2001; Dobney et al. 1996; Maltby 1989; 1998a; in press; O’Connor 1988; Stallibrass 1999). Table 2 lists a selection of distinctive marks that have been explicitly noted in analyses of bones from major urban sites. More detailed descriptions of these and other marks can be found in Maltby (1989; in press). The traits listed in Table 2 are indicative of various processes including the segmentation of the carcass (chopped mandibular ramus; chopped femur caput), filleting (blade marks on upper limb bones and scapula), marrow extraction (axially split upper limb bones), and hanging up meat most probably for preserving the meat (scapulae with holes in blade).It should be emphasised that the list includes only those cases where the butchery trait is specifically mentioned. It is extremely likely that further investigations would reveal that all the types of marks would be encountered in assemblages from major towns. The consistency of the butchery marks and their widespread occurrence suggest that specialist butchers using new methods of carcass processing were operating in these towns. As indicated in Table 3, there is also abundant evidence from these towns for large accumulations of cattle bones that were discarded after various stages of carcass processing. These dumps have been found both near the centre of towns (for example, Dorchester, Maltby 1993 and Wroxeter, Hammon 2005), in peripheral areas (for example, Chichester, Levitan 1989 and Winchester, Maltby in press), or in both (for example, Colchester, Luff 1993; Exeter, Maltby 1979a; and Cirencester, Maltby 1998a). There are a few cases of waste derived from more than one phase of processing (for example, Chester Street, and Cirencester, Maltby 1998a). However, most consist of discrete accumulations derived from primary processing (heads and feet), joint preparation (femur caput, scapulae) marrow extraction (most of upper limb deposits), bone working (for example, some of upper limb deposits in Victoria Road, Winchester, Maltby in press); scapulae deposits at Crowder Terrace, Winchester (Coy and Bradfield in press), or horn working. This suggests that discrete stages of processing were often undertaken separately, perhaps in different locations by different specialists.

It is clear, therefore, that distinctive types of large-scale processing of cattle carcasses developed in the Roman period and that the evidence for such activities has been found commonly within bone assemblages from major Roman towns. How widespread were these activities? Do similar accumulations appear in other types of settlement in Roman Britain? The following sections will briefly examine three types of processing evidence to examine to what extent processing methods varied.

Table 3: Recorded observations of large accumulations of cattle bones in major Romano-British towns Town Site Upper Limbs Scapulae Femur Caput Head/Feet Feet Head Horn Cores Source Chichester Cattlemarket x Levitan 1989 Cirencester Cemetery x Thawley 1982 Cirencester Chester St x Maltby 1998a Cirencester St Michael’s x Levitan 1990 Colchester Balkerne Lane x x x Luff 1982;1993 Colchester Culver St x Luff 1993 Colchester Gilberd School x x Luff 1993 Colchester Long Wyre St x Luff 1993 Dorchester Colliton Park x Hamilton-Dyer 1993 Dorchester Greyhound Yard x Maltby 1993 Exeter Goldsmith St x Maltby 1979a Exeter Rack St x Maltby 1979a Gloucester East Gate x x Maltby 1983 Gloucester Eastgate St x Levine 1986 Gloucester Westgate St x Maltby 1979b Leicester Bonners Lane x Baxter 2004 Leicester Causeway Lane x x Gidney 2000 Leicester Clarence St x Baxter 2005 Leicester Newarke St x Baxter 1996 Lincoln Waterfront x x x Dobney et al.1996 Lincoln Wigford x Dobney et al. 1996 London Mithraeum x Macready/Sidell 1998 London Walbrook x C-Brock/Armitage 1977 Southwark Borough High St x x Ainsley 2002 St Albans Folly Lane x Locker 1999 Silchester Basilica x Grant 2000 Silchester South Gate x Maltby 1984 Silchester Manor Farm x Maltby 1984 Winchester Crowder Terrace x Coy/Bradfield in press Town Site Upper Limbs Scapulae Femur Caput Head/Feet Feet Head Horn Cores Source Winchester Hyde Abbey x Maltby in press Winchester Victoria Rd G2 x Pfeiffer in press Winchester Victoria Rd X-XVI x x x Maltby in press Wroxeter Basilica x x x Hammon 2005 York Tanner Row x x O’Connor 1988 x = present

Blade marks on cattle upper limb bones These distinctive filleting marks consist of shallow scoops of bone removed from ridges and other protuberances along the shafts of upper limb bones (Fig. 1). In addition to the humerus, radius, femur and tibia (Table 2; Fig. 2), they are also commonly encountered on the ulna, and rather less frequently on the astragalus, pelvis and other bones (Maltby 1989; in press). Experiments have suggested that they were created by the use of the tip of a cleaver during the stripping of meat from the bone (Seetah 2006).

It has already been established that such marks are frequently observed in assemblages from large towns (Table 2), but relatively few attempts at quantification have been published. There are problems in quantifying blade marks, as they do not appear on all areas of the bones even if they have been filleted by this method. Therefore estimating the percentages of fragments with such marks is a fairly crude measure. However, it is interesting to note that where frequencies of such marks have been noted, fairly consistent results have been obtained. In several assemblages around 20% of the total number of fragments of femur, tibia, radius and humerus have been damaged with blade marks. These include assemblages consisting almost entirely of cattle processing waste (Eastgate Street, Gloucester, Levine 1986; Chester Street, Cirencester Maltby 1998a), and those that include a much wider range of bones (Basilica, Caerwent, Hambleton and Maltby in prep; Greyhound Yard, Dorchester, Maltby 1993; and various Winchester sites, Maltby in press). There are some variations in blade mark frequencies on different bones and between sites in Winchester (Fig. 2). It is perhaps significant that there is a gradual decrease in the percentage of blade marks the further away the assemblage was from the centre of the Roman town. The highest percentage (28%) was found in the Staple Gardens assemblage (not illustrated) near the centre of the town; the lowest (16%) was obtained in the assemblage from the Hyde Abbey site, located some distance to the north of the town walls (Maltby in press). There is very little evidence for similar marks on Iron Age sites in Britain (Maltby 1989). They have been recorded on other types of civilian Romano-British site, but where quantified, never to the same extent as in the major urban sites (Table 4). They are completely or almost entirely absent from some rural settlements. They have been noted more commonly in villa and larger settlements, although unfortunately very few samples have been quantified.

Split Upper Limb Bones Axially split upper limb bones (Fig. 3) also occur commonly in many towns. Table 2 demonstrates that they have been recorded in a large number of urban assemblages. However, their frequency of occurrence appears less consistent than for blade marks on the same elements. In quantified samples, the percentage of split humeri, radii, femora and tibiae ranges between 11% and 78%. The highest percentage is derived from a discrete assemblage consisting almost entirely of these bones recovered from three pits in Eastgate Street, Gloucester (Levine 1986). This clearly demonstrates that such bones were sometimes accumulated in large numbers for secondary processing. Axial splitting allows easy access to the marrow (Seetah 2006). The under-representation of epiphyses in accumulations such as in Gloucester and York would suggest that the bones were gathered to extract marrow as a discrete product, rather than the bones being used in large-scale stock production (O’Connor 1988: 117; Stokes 2000; Dobney 2001: 40). Indeed, it should be emphasised that although the vast majority of upper limb bones in urban sites have been broken, it is only a certain proportion that were split axially, indicating that they were specially selected for processing marrow in bulk. Similarly, groups of accumulated mandibles have been observed in Lincoln that were also probably processed for marrow (Dobney et al. 1996: 25). In some cases split limb bones were reduced further when they were used to produce bone artefacts, for example in Winchester (Maltby in press; Pfeiffer in press). The lower percentages of split upper limb bones found in assemblages such as in the vicinity of the Basilica in Caerwent (Hambleton and Maltby in prep.) probably indicate that the waste from these specialist processes has been redeposited and had become mixed with other material. The same explanation would account for the variations in later Roman assemblages from sites in the Winchester Northern Suburb and Defences (Table 2; Fig. 4). The highest percentages of split bones were found in deposits of the Victoria Road site where bone working waste had been deposited (Maltby in press).

Figure 1: Cattle tibia from Winchester showing blade marks near distal end

Figure 2: Percentage of cattle upper limb bones with blade marks from later Roman sites in Winchester 0% 10% 20% 30% 40% VR X-XVI R HT F VR IV-VI R T H F HA R TH F Defences R T F H 0% 10% 20% 30% 40% VR = Victoria Road; HA = Hyde Abbey; Defences = Jewry St, Crown Hotel and Henly’s Garage % = percentage of total number of specimens of each type of upper limb bone bearing blade marks F = femur; T = tibia; H = humerus; R = radius

Table 4: Recorded observations of blade marks in assemblages from other Romano-British civilian sites Settlement Type Blade Up Limb Source Alcester, Warwickshire small town x Maltby 2001a Alington Avenue, Dorchester urban hinterland 3% Maltby 2002a Asthall, Oxfordshire small town x Powell et al. 1997 Biddenham Loop, Beds rural 0% Maltby nd1 Home Farm, Bishop’s Cleeve rural 3% Maltby 1998b Stoke Road, Bishop’s Cleeve rural 0% Maltby 2002b Brighton Hill S. Hampshire rural 0% Maltby 1995 Marsh Leys Farm, Beds rural 4% Maltby nd2 Neatham, Hampshire small town x Done 1986 Owslebury, Hampshire rural 0.2% Maltby 1987 Shefford, Beds rural 0% Maltby nd3 Shepton Mallet, Somerset small town x Pinter-Bellows 2001 Snodland, Kent villa x Hamilton-Dyer 1995 Wantage, Oxfordshire roadside settlement 11–15% Maltby 1996; 2001b Wilcote, Oxfordshire roadside settlement x Hamshaw-Thomas et al. 1993; 2004 Winnall Down, Hampshire rural 0% Maltby 1985a Wortley, Gloucestershire villa x Maltby/Hambleton in prep. Wycomb, Gloucestershire small town 0% Maltby 1998c Up Limb = humerus, radius, femur and tibia x = present but not quantified

Figure 3: Axially split radius from Roman Winchester

Figure 4: Percentage of cattle split upper limb bones later Roman sites in Winchester

Again, comparisons with other settlements (Table 5) show that split upper limb bones are much more likely to be encountered in major towns. Such bones are encountered very rarely in Iron Age assemblages (Maltby 1989), and are often absent from assemblages from Romano-British rural sites such as Owslebury (Maltby 1987) and Biddenham Loop (Maltby nd1). They have been found in some of the assemblages from nucleated settlements, none of which unfortunately have been quantified. One suspects that they are less common than in large urban complexes but further research is required to confirm that impression. They have, however, been recorded on a few villa sites including Wortley, Gloucestershire. Here, although found in some numbers, axially split bones appear to be limited to one or two deposits, suggesting that the practice was restricted both in extent and time.

0% 10% 20% 30% 40% 50% VR X-XVI F T H R VR IV-VI F T H R HA F T H R Defences F H T R 0% 10% 20% 30% 40% 50% VR = Victoria Road; HA = Hyde Abbey; Defences = Jewry St, Crown Hotel and Henly’s Garage % = percentage of total number of specimens of each type of upper limb bone recorded as split F = femur; T = tibia; H = humerus; R = radius

Table 5: Recorded observations of split upper limb bones in assemblages from other Romano-British civilian sites Settlement Type Split Up Limb Source Alcester, Warwickshire small town x Maltby 2001a Alington Avenue, Dorchester urban hinterland 0% Maltby 2002a Asthall, Oxfordshire small town x Powell et al.1997 Biddenham Loop, Beds rural 0% Maltby nd1 Castleford, Yorkshire vicus x Berg 1999 Home Farm, Bishop’s Cleeve rural 0% Maltby1998b Stoke Road, Bishop’s Cleeve rural 0% Maltby 2002b Elms Farm, Essex temple complex x Albarella 1998 Keston, Kent villa Low % Locker 1991 Marsh Leys Farm, Beds rural 0% Maltby nd2 Owslebury, Hampshire rural 0% Maltby 1987 Rocester, Staffordshire small town x Hammon 2000 Shefford, Beds rural 0% Maltby nd3 Shepton Mallet, Somerset small town x Pinter-Bellows 2001 Stonea, Cambridgeshire vicus x Stallibrass 1996 Tort Hill East roadside settlement x Albarella 1998 Wantage, Oxfordshire roadside settlement 0% Maltby 1996; 2001b Wilcote, Oxfordshire roadside settlement x Hamshaw-Thomas et al. 1993; 2004 Winnall Down, Hampshire rural 0% Maltby 1985a Worcester, Worcestershire small town x Nicholson/Scott 2004 Wortley, Gloucestershire villa x Maltby/Hambleton in prep. Wycomb, Gloucestershire small town 0% Maltby 1998c Up Limb = humerus, radius, femur and tibia x = present but not quantified Transversely broken metapodials

In contrast with the upper limbs, axially split cattle metapodials are uncommon in Romano- British assemblages. However, most of them have been broken transversely, presumably to extract the marrow. Occasionally, cleaver marks are located near the break (Fig. 5). The increased fragmentation of metapodials has been noted in Romano-British assemblages when compared with Iron Age samples, and preliminary research has demonstrated that such an increase was not the result of increased damage due to scavenging animals (Maltby 1985b). Table 6 shows that percentages of complete or largely complete metapodials are very low in urban assemblages when compared with those from rural sites. This indicates that marrow extraction from metapodials tended to be more intensive on urban sites. This is reflected even in cases where large numbers of foot bones were discarded after primary butchery, for example in the South Gate assemblage at Silchester (Maltby 1984). The percentages of complete metapodials from Winchester would have been even lower had the bones from two particular shafts been omitted (Maltby in press). These bones were found in association with a number of partial and complete skeletons of several species. Faunal assemblages in these features are atypical of the rest of the Winchester sites and they may have included a significant amount of structured deposition, which may have incorporated the complete metapodials.

Table 6: Percentage of cattle metapodials consisting of >75% of the bone from Romano-British settlements Settlement Type NISP Metatarsals % >75% Complete Source Caerwent urban 224 9 Hambleton/Maltby nd Silchester South Gate urban 28 4 Maltby 1984; 1985b Winchester VR X-XVI urban 277 11 Maltby in press Winchester VR IV-VI urban 49 14 Maltby in press Winchester Hyde Abbey urban 44 0 Maltby in press Winchester Defences urban 42 5 Maltby in press Brighton Hill South rural 34 24 Maltby 1995 Holwell Quarry rural 32 28 Maltby nd4 Marsh Leys Farm rural 56 18 Maltby nd2 Winnall Down rural 28 64 Maltby 1985a; 1985b Settlement Type NISP Metacarpals % >75% Complete Source Caerwent urban 198 7 Hambleton/Maltby nd Silchester South Gate urban 22 5 Maltby 1984; 1985b Winchester VR X-XVI urban 157 15 Maltby in press Winchester VR IV-VI urban 50 4 Maltby in press Winchester Hyde Abbey urban 18 11 Maltby in press Winchester Defences urban 37 8 Maltby in press Brighton Hill South rural 17 41 Maltby 1995 Holwell Quarry rural 20 30 Maltby nd4 Marsh Leys Farm rural 37 32 Maltby nd2 Winnall Down rural 21 52 Maltby 1985a; 1985b Biddenham Loop rural 20 45 Maltby nd1 VR = Victoria Road

Figure 5: Cattle metatarsal from Late Roman Winchester with transverse cleaver marks on shaft

Discussion The examples of processing evidence outlined above confirm that distinctive butchery methods became employed during the Romano-British period. There is a general acceptance that the methods of specialised butchery of cattle carcasses evident, particularly in Roman towns, were derived from military practices (Grant 1989; Maltby 1989; Stallibrass 1999; Berg 1999; Dobney 2001; Seetah 2006). Although these butchery practices have been recognised for a long time, there has been only a limited discussion about how widespread or consistent these practices became. Comparative studies are still severely handicapped by the paucity of quantified data and imprecise descriptions that are characteristic of many faunal reports, particularly with regard to explicitly noting the absence of specific butchery traits. However, it is becoming increasingly evident that such processing was not practised consistently on all Romano-British settlements. By recording the presence, absence and, ideally, the frequency of distinctive butchery traits, it should be possible to trace where specialist processors were operating and begin to consider to what extent they controlled the provision of beef products. In contrast, on settlements where traditional methods of butchery continued to be practised, it could suggest that the residents were less integrated into the new economic systems that prevailed in the major towns. The surprising consistency in the frequency of blade marks on cattle upper limb bones in urban assemblages suggests that perhaps most of the filleted beef consumed in the major towns was obtained from cattle processed by specialist butchers. If that was the case, this has implications on how a major proportion of the meat supply was controlled and redistributed, as noted by Grant (1989) and Maltby (1989). Traditional means of procurement of beef must have been superseded by the emergence of professional traders who acquired large numbers of animals for processing. The proportion of stock slaughtered and butchered by non-specialists must have decreased markedly. The slaughter of large valuable animals and the redistribution of their meat (fresh or preserved by salting and/or smoking) must have been significant events for their owners and their families. The process may often have been associated with significant social or ceremonial occasions (for example, commemorations, marking rites of passage, celebrating seasonal events, and gatherings of kin or trading partners etc.). The emergence of urban centres with the increased demands for food must have led not only to a change of economic emphasis towards wholesale procurement, slaughter and redistribution by professional traders, but also probably severely disrupted social and ritual practices associated with traditional means of meat redistribution.

The presence of specialist butchers filleting meat is also attested at a number of small towns. Sadly the quantification of their occurrence is generally lacking and it would be interesting to establish whether their frequencies reached the levels encountered in larger towns, thus implying the dominance of wholesale processing of beef in those settlements too. Specialist butchers were certainly present in roadside settlements such as Wantage (Maltby 1996; 2001b), where filleted limb bones have been found in two small excavations in fairly high quantities (Table 4). They were also present in villas such as Wortley (Maltby and Hambleton in prep.), and Snodland (Hamilton-Dyer 1995), but whether that was a widespread phenomenon has yet to be established. The lack of comment regarding such occurrences in a number of faunal reports from villas perhaps suggests that it was uncommon. At Wortley, most of the blade-marks were found on split bones from one major accumulation (see below), and it is possible that this was associated with specific specialist processing that was intensive but perhaps not of long duration. This raises the possibility that some specialist butchers were itinerant.

Blade-marked upper limb bones have been found very infrequently in other rural settlements. Either specialists only rarely operated in such settlements or, more likely perhaps, the new methods (and associated technology) were not adopted wholeheartedly by the residents. This is not necessarily to say that they were disengaged from the new urban procurement system. It has been suggested, based on ageing and sexing evidence, that some of the cattle from rural settlements such as Owslebury could have been acquired by professional butchers in settlements such as Winchester (Maltby 1994; in press). Filleted beef acquired from urban markets or shops would also leave no trace in the archaeological deposits of the rural settlements.

With regard to the possible acquisition of beef from towns on rural settlements, it is perhaps worth noting that the only limb bone that has types of butchery marks commonly found on both rural and urban assemblages is the scapula. At both Winchester and Owslebury, the most common butchery marks on the cattle scapula were found on the edge of the spine, particularly where it rises from neck of the glenoid cavity. Other rural settlements have also produced scapulae chopped in a similar manner, for example Winnall Down (Maltby 1989; in press), Biddenham Loop (Maltby nd1), and Marsh Leys Farm (Maltby nd2). It has been suggested that similar marks found on specimens from Lincoln represent the preparation of scapulae and the attached shoulder meat for salting (Dobney et al. 1996: 26–27; Dobney 2001: 40–41). Seetah (2006) has demonstrated that the marks were most likely to have been made with the tip of a cleaver blade; a method that urban butchers were also employing on the upper limb bones. This raises the possibility that settlements such as Owslebury were importing cured shoulders of beef on the bone. Longitudinal knife cuts were found on the blades of a few of such scapulae at Owslebury and Biddenham Loop, suggesting that traditional methods of butchery were subsequently carried out to fillet the meat from the bone. Similar marks were found much more rarely in Winchester (Maltby 1989; in press) and Cirencester (Maltby 1998a). An alternative explanation to account for the presence of these distinctively butchered scapulae on rural settlements was that they were processed there in the same manner as in towns. Perhaps it was the methods of curing beef that became more widespread, rather than all methods of carcass processing.

Marrow became more important in the Romano-British period in general, as indicated by the increased fragmentation of metapodials. Again, however, the intensity of the exploitation seems to have been greater in the urban settlements than elsewhere. In addition to this general increase, it appears to have been a common practice to accumulate upper limb bones to obtain larger quantities of marrow. This would suggest that most axial splitting was associated with specialist, rather than routine, processing. Such processing appears to have taken place more frequently in large towns, where, of course, suitable bones would be available in abundance. There is, however, sometimes evidence for similar processing in other types of settlement, although noticeably not on any of the non-villa rural settlements in this survey. Marrow was therefore processed in bulk, commonly in major towns, and perhaps fairly frequently in small towns and some roadside settlements. The bones were processed either by specialist butchers themselves, or by people who obtained filleted bones from them and processed them using cleavers. It has not been clearly established whether the marrow obtained in this way was gathered for a specific purpose or for more general use. Dobney (2001: 40) has suggested some possible uses for marrow, including lamp oil, cosmetics, soaps, and medicines, in addition to its value in cooking. Others have suggested that some of the accumulations represent the preparation of quantities of glue. Further research is still required into this topic. The occurrence of the discrete concentrations of split limb bones at Wortley indicates that specialist production took place at the villa at some time during its occupation. However, the quantity of bones required may have been greater than could be provided by the number of cattle routinely butchered there. It is possible that the limb bones were brought in from elsewhere for the processing of large quantities of fresh marrow for a particular (industrial?) project. It is clear, therefore, that variations in carcass processing are apparent on Romano- British sites. The evidence for changes in cattle carcass processing has implications both in economic and social terms. Beef production became more important in the Roman period (King 1999), and the need to provision towns in particular led to significant changes in how animals were acquired, processed, and their products distributed. It is possible to trace the activities of specialist processors and to recognise where they operated. It is also possible to observe on which settlements traditional butchery practices, and perhaps traditional forms of meat redistribution, were maintained, at least in relation to the cattle slaughtered at those settlements. Wholesale beef processing resulted in the loss of identity of the animals involved. People who ate beef processed by urban butchers are much less likely to have known who originally owned the animal, in contrast to traditional forms of processing. The social dimensions of the relationship between the cattle owner and the recipients of the meat and other products therefore changed fundamentally.

It is hoped that this research will encourage further explicit statements to be made about the presence, absence and frequency of various butchery traits encountered on different types of Romano-British sites. Such studies have the potential to shed further light upon far-reaching changes in the economic and social spheres in Britain subsequent to the Roman invasion.

School of Conservation Sciences, Bournemouth University

Acknowledgements I would like to thank Albion Archaeology; Andy Hammon; The National Museum of Wales; Wessex Archaeology; The Winchester Museum Service and David Wilson for the use of unpublished material. Helen Rees designed Figures 2 and 4.