domestic fowl on romano-british sites: inter-site comparisons of abundance

13
Domestic Fowl on Romano-British Sites: Inter-site Comparisons of Abundance MARK MALTBY { School of Conservation Sciences, Bournemouth University,Talbot Campus, Fern Barrow, Poole, Dorset BH12 5BB, UK ABSTRACT The paper compares the abundance of domestic fowl bones with those of sheep/goat and pig from published and unpublished reports on Romano-British animal bones assemblages. The samples are drawn from 68 urban, military, nucleated, villa and unromanized rural settlements. Variations in abundance of domestic fowl within and between these settlement types are noted. Some methodological problems encountered in this survey are discussed and suggestions are made about how more reliable results could be achieved. # 1997 by John Wiley & Sons, Ltd. Int. J. Osteoarchaeol., 7: 402–414 (1997) No. of Figures: 2. No. of Tables: 5. No. of References: 12. Key words: archaeozoology; domestic fowl Gallus gallus; pigs; sheep; Romano-British; Britain; methodology. Introduction Several studies have compared the faunal remains from different types of Romano-British settlements. These have attempted to compare aspects of the exploitation and consumption of animals and their products and to trace changes in the domestic stock used in the period. For example, the relative abundance of the major domestic mammals has been shown to vary on different types of Romano-British sites. 1 Cattle and pigs tend to be better represented on major urban and military sites, whereas sheep/goat tend to be relatively more common on rural ‘native’ settlements. Comparisons of assemblages from urban, suburban and rural sites centred on Dorchester (Dorset) and Winchester (Hampshire) 2 have shown significant differences in the relative abundance, mortality patterns, morphology and butchery of domestic mammals. Such contrasts lead us to consider whether there were also intersettlement variations in the abundance of domestic fowl exploited and/or consumed on British sites during the first to fourth centuries AD. Domestic fowl appear only to have been introduced into Britain in the later Iron Age, but are usually the most abundant avian species represented on Romano-British sites. 3 However, although there has been an extensive review of wild birds represented on Romano-British sites, 4 there has been no survey of the abundance of domestic fowl apart from restricted comparisons of species abundance from a small number of sites. 3 This paper will attempt a preliminary investigation of domestic fowl abundance on a wider range of Romano- British sites and discuss the problems of such intersite surveys. Methods and problems The initial research involved the collection of data from 123 assemblages from 68 settlements International Journal of Osteoarchaeology, Vol. 7: 402–414 (1997) CCC 1047–482X/97/040402–13$17.50 Received 1 November 1995 # 1997 by John Wiley & Sons, Ltd. Accepted 25 November 1996 { Correspondence to: M. Maltby, School of Conservation Sciences, Bournemouth University, Talbot Campus, Fern Barrow, Poole, Dorset BH12 5BB, UK.

Upload: bournemouth

Post on 10-Jan-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

Domestic Fowl on Romano-BritishSites: Inter-site Comparisons ofAbundance

MARK MALTBY{

School of Conservation Sciences, BournemouthUniversity,Talbot Campus,FernBarrow, Poole, Dorset BH12 5BB, UK

ABSTRACT The paper compares the abundance of domestic fowl bones with those of sheep/goat and pigfrom published and unpublished reports on Romano-British animal bones assemblages. Thesamples are drawn from 68 urban, military, nucleated, villa and unromanized ruralsettlements. Variations in abundance of domestic fowl within and between these settlementtypes are noted. Some methodological problems encountered in this survey are discussedand suggestions are made about how more reliable results could be achieved. # 1997 byJohn Wiley & Sons, Ltd.

Int. J. Osteoarchaeol., 7: 402±414 (1997)No. of Figures: 2. No. of Tables: 5. No. of References: 12.

Keywords: archaeozoology; domestic fowl Gallus gallus; pigs; sheep; Romano-British; Britain;methodology.

Introduction

Several studies have compared the faunalremains from different types of Romano-Britishsettlements. These have attempted to compareaspects of the exploitation and consumption ofanimals and their products and to trace changesin the domestic stock used in the period. Forexample, the relative abundance of the majordomestic mammals has been shown to vary ondifferent types of Romano-British sites.1 Cattleand pigs tend to be better represented on majorurban and military sites, whereas sheep/goat tendto be relatively more common on rural `native'settlements. Comparisons of assemblages fromurban, suburban and rural sites centred onDorchester (Dorset) and Winchester(Hampshire)2 have shown signi®cant differencesin the relative abundance, mortality patterns,morphology and butchery of domestic mammals.

Such contrasts lead us to consider whetherthere were also intersettlement variations in theabundance of domestic fowl exploited and/orconsumed on British sites during the ®rst tofourth centuries AD. Domestic fowl appear onlyto have been introduced into Britain in the laterIron Age, but are usually the most abundantavian species represented on Romano-Britishsites.3 However, although there has been anextensive review of wild birds represented onRomano-British sites,4 there has been no surveyof the abundance of domestic fowl apart fromrestricted comparisons of species abundancefrom a small number of sites.3 This paper willattempt a preliminary investigation of domesticfowl abundance on a wider range of Romano-British sites and discuss the problems of suchintersite surveys.

Methods and problems

The initial research involved the collection ofdata from 123 assemblages from 68 settlements

International Journal of Osteoarchaeology, Vol. 7: 402±414 (1997)

CCC 1047±482X/97/040402±13$17.50 Received 1 November 1995# 1997 by John Wiley & Sons, Ltd. Accepted 25 November 1996

{Correspondence to: M. Maltby, School of ConservationSciences, Bournemouth University, Talbot Campus, FernBarrow, Poole, Dorset BH12 5BB, UK.

in Britain (Figure 1) on the numbers of domesticfowl bones recorded in published bone reportsor in site archives made available to me (Table1). Many of the samples are the same as used inprevious surveys of Romano-British sites,1,4±6 butalso include data from more recent analyses.Counts were compared with those of otherdomestic species. This involved making anumber of decisions about how to quantify thematerial and to address a number of problemsencountered in its collation.

Identi®cation

Most reports do not state explicitly the criteriaused in the identi®cation of domestic fowl bones.Other galliformes theoretically could be presentbut have not been distinguished from domesticfowl, perhaps due to the lack of an extensivereference collection.7 This is probably not amajor problem because, where authors havediscussed their identi®cations, they have usuallyonly been able to positively identify domestic

Domestic Fowl on Romano-British Sites 403

# 1997 by John Wiley & Sons, Ltd. INT. J. OSTEOARCHAEOL., Vol. 7: 402±414 (1997)

Figure 1. Map of Roman Britain showing settlements and sites used in the survey. See Table 1 for key to numbers.

404 M. Maltby

INT. J. OSTEOARCHAEOL., Vol. 7: 402±414 (1997) # 1997 by John Wiley & Sons, Ltd.

Table 1. Settlements and sites in the survey and source of the data (see Figure 1).

Map settlement/site Data sourcesa

1 Abbotstone Down M. Maltby, 1986, AML 58/962 Abingdon B. Wilson, 1991, Oxon. 56: 1±15.3 Alcester Maltby, unpublished report4 Appleford Wilson/D. Bramwell, 1980, Oxon. 45: 84±905 Bancroft 1973±78 B. Westley, 1987, in D. Mynard, Roman Milton Keynes (Bucks Arch.

Soc. Monograph 1)5 Bancroft 1983±86 B. Levitan, 1990, AML 58/906 Barton Court Farm Wilson, 1986, in D. Miles, CBA 507 Bishops Cleeve Maltby, unpublished report8 Brancaster R. Jones et al., 1985, in J. Hinchcliffe, East Anglian Archaeology 239 Braughing R. Ashdown/C. Evans, 1981, in C. Partridge, Britannia Monograph 2

10 Brighton Hill S. Maltby, 1995, in P. Fasham et al., Wessex 711 Butter®eld Down Maltby, unpublished report12 Caerleon Baths T. O'Connor, 1986, in J. Zienkiewicz, The Legionary Fortress Baths at

Caerleon12 Caerleon Museum S. Hamilton-Dyer, 1993, Britannia 24: 132±613 Caerwent B. Noddle/Bramwell, 1983, Arch. Camb. 132: 63±7014 Carlisle, OGL S. Stallibrass, 1993, AML 93/9314 Carlisle, S. Lanes Stallibrass, 1993, AML 96/9315 Charlton Kings Noddle, 1991, TBGAS 109: 85±8616 Chester, Northgate B. West, archive report17 Cirencester, Mil. C. Thawley, 1982, in J. Wacher/A. McWhirr, Early Roman Occupation

at Cirencester17 Cirencester, Querns Maltby, unpublished report17 Cirencester, Sheep Maltby, unpublished report18 Coggeshall O. Bedwin, 1988, Essex Arch & History 19: 7919 Colchester, Balk. R. Luff, 1985, Colchester Archaeology Report 1219 Colchester, LWS Luff, 1985, Colchester Archaeology Report 1219 Colchester, Midbgh Luff, 1985, Colchester Archaeology Report 1219 Colchester, Sheepen Luff, 1985, Colchester Archaeology Report 12

20 Coombe Down P. Smith, unpublished report21 Derby, Racecourse M. Harman/Bramwell, 1985, DAJ 105: 148±53219±2122 Dorchester, Aling. Maltby, 1988, AML 182/8822 Dorchester, GY Maltby, 1993, in P. Woodward et al., DNHASM 1223 Durrington Walls Westley, 1971, in WAM 65: 122±524 Earith D. Phillipson/F. Gilmore, 1967, PCAS 60: 9±1025 Easton Lane Maltby, 1989, in P. Fasham et al., HFCM 626 Elsted A. Saunders, 1980, SAC 118: 221, 227±827 Exeter Maltby, 1979, Exeter Archaeological Reports 228 Farmoor Wilson, 1979, in G. Lambrick/M. Robinson, CBA 3229 Figheldean J. Egerton et al., 1993, WAM 86: 38±4030 Frocester Court Noddle/Bramwell, 1979, TBGAS 97: 51±6231 Gloucester, E. Gate Maltby, 1983, in C. Heighway, The East and North Gates of Gloucester32 Godmanchester Luff, 1993, AML33 Gorhambury A. Lockyer, 1990, in D. Neal et al., EH 1434 Great Chesterford D. Serjeantson, 1986, PCAS 75: 37±935 Great Dunmow Luff, 1988, in N. Wickenden, EAA 4136 Ilchester Levitan, 1994, in P. Leach, Ilchester Volume 237 Lambeth Locker, 1988, in P. Hinton, Museum of London 338 Latimer R. Hamilton, 1971, in K. Branigan, Latimer39 Little Chester Harman/Bramwell, 1985, DAJ 105: 148±53219±2140 Little Somborne Maltby, 1981, in P. Catherall et al., The Southern Feeder41 Little Woolstone B. Westley, 1987, in D. Mynard, Roman Milton Keynes (Bucks Arch.

Soc. Monograph 1)42 Longthorpe B. Marples, 1974, Britannia 5: 122±843 Mantles Green G. Jones, 1994, Records of Bucks. 34: 173±444 Magniovinium Locker, 1987, Records of Bucks. 29: 108±1545 Micheldever Wood J. Coy, 1987, in P. Fasham, HFCM 546 Middleton Stoney Levitan, 1984, in S. Rahtz/T. Rowley, Middleton Stoney47 Ower Coy, 1987, in P. Woodward, DNHASM 648 Owslebury Maltby, 1987, AMP 6/87

(continued )

fowl. Signi®cantly, perhaps, other galliformeshave been recorded very rarely on Romano-British sites.4

A greater problem concerns the identi®cationof immature bones. Some authors do not recordthem to species; others sometimes place them in`probable domestic fowl' categories; othersrecord them to species. These decisions couldhave a signi®cant bearing on the counts of fowlbones recorded.8 Unfortunately, most authors donot discuss this problem.

Retrieval

Any comparisons of bird bones with largermammals must take into account retrieval biases.

These have been clearly demonstrated insamples of all periods, including York4,9 andSouthampton.10 Sieving experiments haveshown that bird bones are consistently under-collected. Ideally, only sieved material should beused in comparisons but this would at presentgreatly restrict intersite comparisons.

The alternative is to rely solely on unsievedmaterial. This has the advantage of increasingthe number of sites but has many problems.Domestic fowl bones will be underrepresented incomparison with larger mammalian species.There is no easy way of determining whetherthis bias is consistent between sites, although itis possible to devise indices to provide anindication about retrieval ef®ciency.11 It was,however, decided to base the initial analysis onthe assumption that the retrieval rates were

Domestic Fowl on Romano-British Sites 405

# 1997 by John Wiley & Sons, Ltd. INT. J. OSTEOARCHAEOL., Vol. 7: 402±414 (1997)

Table 1. Continued

49 Portchester Castle A. Grant, 1975, in B. Cunliffe, Soc. Ant. 3250 Puckeridge A. Croft/Ashdown, 1979, in C. Partridge, HAM 751 Raynes M. Smoothy, 1989, Essex Arch & History 20: 1±2952 Roughground Farm G. Jones/Levitan, 1993, in T. Allen et al., Excavations at Roughground

Farm, Lechdale53 Scole G. Jones, 1977, in A. Rogerson, EAA 554 Silchester, S. Gate Maltby, 1984, in M. Fulford, Britannia Monograph 555 Slonk Hill S. Sheppard/Coy, 1977, SAC 116: 133±956 Stantonbury B. Westley, 1987, in D. Mynard, Roman Milton Keynes (Bucks Arch.

Soc. Monograph 1)57 Wantage Maltby, unpublished report58 Watercrook P. Fi®eld, 1979, in T. Potter, Romans in NW England59 West Stow P. Crabtree, 1990, in S. West EAA 4860 Westhampnett Smith, unpublished report61 Whitcombe J. Buckland-Wright, 1990, DNHAS 112: 86±9062 Wilcote J. Hamshaw-Thomas/N. Bermingham, 1993, BAR 23263 Winchester, SG Maltby, 1986, AML 490863 Winchester, VRX-XVI Maltby, unpublished report63 Winchester, VRIV-VI Maltby, unpublished report63 Winchester, Hyde A. Maltby, unpublished report63 Winchester, Def. Maltby, unpublished report25 Winnall Down Maltby, 1985, in P. Fasham, JFCM 264 Worcester S. Scott, 1992, TWAS 13: 88±9265 Woughton B. Westley, 1987, in D. Mynard, Roman Milton Keynes (Bucks Arch.

Soc. Monograph 1)66 Wroxeter B. Meddens, unpublished report67 Wycomb Maltby, unpublished report68 York O'Connor, 1988, The Archaeology of York 15/2

aAML, Ancient Monuments Laboratory Report; Arch Camb, Archaeologia Cambrensis; BAR, British Archaeological Report(British Series); CBA, Council for British Archaeology Research Report; DAJ, Derbyshire Archaeological Journal; DNHASM,Dorset Natural History and Archaeological Society Monograph; EAA, East Anglian Archaeology; EH, English HeritageArchaeology Report; HAM, Hertfordshire Archaeological Monograph; Museum of London, Museum of London Monograph;Oxon, Oxoniensia; PCAS, Proceedings of the Cambridgeshire Archaeological Society; SAC, Sussex ArchaeologicalCollections; Soc.Ant., Society of Antiquaries Research Report; TBGAS, Transactions of the Bristol and GloucestershireArchaeological Society; TWAS, Transactions of the Worcestershire Archaeological Society; WAM, Wiltshire ArchaeologicalMagazine; Wessex, Wessex Archaeology Report.

roughly comparable. Sites where there was apossibility that not all bones had been retainedfor examination were not included in the survey.

Preservation

Domestic fowl bones are fragile. Although theysuffer less butchery damage than larger species,destruction by scavengers and other taphonomicprocesses is great. In general reviews of this type,one has to assume that such biases are consistentbetween sites, although this is unlikely.However, samples where enough information isavailable to assess such biases are rare.

Intrasite variation

Making overall counts of bones recovered fromexcavations of a site assumes that the sample isrepresentative of the bones deposited and thespecies exploited in the whole settlement. Thiscan be challenged where small areas only of thesettlement have been investigated or in caseswhere there is good evidence for intrasitevariability in the faunal remains. For example, anumber of Roman towns have produced evi-dence for the large-scale disposal of parts ofcattle carcasses, which have biased bone countstowards cattle.12 In this survey, samples fromdifferent excavations within the same settlementwere kept separate and samples subdivided intoclear chronological groups within or betweensites were also considered as discrete groups. Anumber of settlements are therefore representedby more than one sample. This allows investiga-tion of possible chronological variations and alsoprovides some indication of variability withinthe assemblages.

Comparisons with other species

The main species exploited for meat at thisperiod were cattle, sheep and pig.1 Assessment ofthe importance of domestic fowl in the dietrelative to these requires counts of domestic fowlbones to be compared with those of these

domestic mammals. Some of the problems ofcomparisons using simple counts of bones orfragments (number of identi®ed specimens;NISP) have been outlined above. O'Connor,9

however, has demonstrated that interestingresults can be achieved using such simplemethods despite their acknowledged problems.He compared the relative percentages of (i)cattle and sheep; (ii) pig; and (iii) domestic fowland goose from a number of Anglo-Saxonsamples and showed both similarities and varia-tions in assemblages from sites of different types.

In Roman samples, however, numbers of cattlebones are very dif®cult to compare directly withsheep/goat and pig and even more so with birdbones. Their greater size, better preservation andparticularly the tendency for carcasses to bebulk-processed on some sites creates majorproblems for realistic comparisons. In addition,cattle bones are often heavily fragmented onRoman sites, which leads both to further biasesin bone counts and to greater variability betweendifferent specialists in what is counted.Accordingly, cattle are excluded in this analysis.

Comparisons between domestic fowl, sheep/goat and pig bones may be more reliable becausethere are smaller (albeit still signi®cant) differ-ences in the sizes and densities of their bones.More of the sheep and pig meat is likely to havebeen cooked on the bone, which would increasethe likelihood that bones would be discarded askitchen waste along with bones of domesticfowl. This should decrease the problems ofintrasite variability. Domestic fowl are still goingto be underrepresented but the biases will be lesssevere than they would be if cattle bones areincluded in the calculations.

Sample sizes

Many excavations produced samples too small tobe included in this survey. Assemblages thatcontained fewer than 100 fragments of sheep/goat, pig and domestic fowl were usuallyexcluded. In a number of cases, assemblagesfrom different phases have been combined toproduce a sample of suf®cient size. This will

406 M. Maltby

INT. J. OSTEOARCHAEOL., Vol. 7: 402±414 (1997) # 1997 by John Wiley & Sons, Ltd.

preclude any investigation of diachronic changeon such sites.

Quanti®cation

This proved to be the most problematic area ofthe initial survey. One of the major problems ofusing counts of individual specimens (NISP) isthat what is counted depends upon the methodsof different workers and these can vary sig-ni®cantly. There are some variations of whatbones of domestic fowl are included in thecounts, but the major problems arise in thevariations in how mammal bones are counted.Some people include ribs and vertebrae in theircounts; others either do not or count just aselection of them. Some authors include limbbones that survive just as shaft fragments; othersonly count fragments which include articularsurfaces. Others count only a selection of thebones; others publish only minimum number ofindividuals represented by each bone. There arearguments for and against all of these ap-proaches. There is the added problem that insome published reports, no information is givenabout how counts were obtained nor any dataabout which bones were present.

I have where possible included assemblageswhere counts of all limb bone fragments appearto have been made. It is, however, possible thatsome of these totals relied on more restrictedcounts, but the information about this is notexplicit. Ribs have been excluded in cases wherethey have been assigned to a `sheep-sized'category rather than being speci®cally identi®edas sheep/goat or pig. Bones from complete orpartial skeletons were excluded where they wereclearly indicated. There still may be signi®cantvariations in the counts that are derived fromdifferent workers, however. Ways of improvingthe consistency of intersite comparison arediscussed at the end of the paper.

Settlement types

The main purpose of this paper is to investigatewhether there are variations in the frequencies inwhich domestic fowl bones are recovered from

settlements of different types. These settlementsrequire de®nition. Seven categories were used.

Major urban sites

These included samples from the major coloniaand civitas capitals. They include assemblagesboth from intramural sites and from extramuralsettlement. Possible contrasts between the dietsof inhabitants in the centre of towns and those inthe suburbs were therefore not investigated indetail.

Nucleated sites

This is a broad grouping that includes townssuch as Alcester, which arguably could havebeen included in the major urban category. Bycontrast, it also includes relatively small settle-ments, which may have had few functions otherthan those concerned with agriculture. Sitesdescribed as small towns, roadside settlementsand large villages are included in this category. Italso includes some larger civilian settlementsassociated with military sites, such as Carlisle. Itcould be argued that these would be betterplaced with the military sites. However, as it hasbeen suggested that diet may have variedaccording to the different geographical originsof inhabitants of Roman Britain,1 it has beendecided to keep these samples separate from themilitary sites.

Villas

This category was not as easy to de®ne as wouldbe perhaps thought. Settlements that clearly hadstructures built of materials usually associatedwith villas could be safely included in thiscategory. Sometimes, however, excavators havebeen very cautious in attributing villa status incases where others would have had less hesita-tion. There is also the problem of satellitesettlements, which may have formed parts ofvilla estates but lacked the buildings usuallyassociated with villas. These have been excluded.

Domestic Fowl on Romano-British Sites 407

# 1997 by John Wiley & Sons, Ltd. INT. J. OSTEOARCHAEOL., Vol. 7: 402±414 (1997)

Some villas, such as Frocester, developed fromearlier forms of settlement, and so were includedin this category only in their later forms. Thiscategory is designed to represent higher statusrural sites, although it is recognized that statusneed not be re¯ected merely in buildingtypologies.

Industrial

This category includes sites at which the mainfunction appears to have had an industrialpurpose; Sheepen ironworking complex nearColchester is an example of such a site.

Other rural

These include all sites not de®ned as villas andwhich are believed to be smaller than thenucleated settlements included above. They areassumed to belong to farming families orcommunities that were arguably of a lower statusthan those occupying villas.

Military

This group includes all types of military estab-lishment. Again, it would have been interestingto subdivide this group further but there werenot enough samples available to justify this.

Oppida

Samples from two late Iron Age oppida of the®rst century AD are also included.

Assemblages from temple sites were notincluded in the survey.

Regional subdivisions

The sites have been assigned to six regions(Tables 2 and 3). It is possible that there wassigni®cant regional variation in the abundance ofdomestic fowl. However, because of the uneven

representation of settlement types, analysis ofregional variation awaits the accumulation ofadditional samples, and is not considered furtherhere.

The data

Data from the 123 samples are summarized inthe form of percentages of domestic fowl, sheep/goat and pig bones (Table 2). In Table 3 thesamples have been split by region and settlementtype. Figure 2 gives the average percentages ofdomestic fowl, pig and sheep/goat by settlementtype, and Table 4 shows domestic fowl as apercentage of domestic fowl and sheep/goat (FS)and of fowl and pig (P) respectively, calculatedby totalling NISP from all sites. Table 5 showsthe variability in numbers of domestic fowlbones recovered, expressed as a percentage ofthe total domestic fowl and sheep/goat bones(TNISP), in samples from the same settlementtypes.

Results

The samples (Table 3) vary considerably in size.The largest was 8362 bones; 29 samplescontained more than 1000 but 24 had fewerthan 200 fragments. The largest number are frommajor urban sites and these included more largesamples than other types of settlement. Thosefrom nucleated sites tended to be smaller. Nonucleated sites in the South and South Westregions are represented. The samples from villaswere only from the Midlands and the South East;unfortunately, none were found in the Southregion to compare with the large number ofother rural settlements from that region. Mostother rural assemblages were from the South, butmost of these produced small samples of fewerthan 500 fragments. Only 13 military samplescould be used, of which only one had over 1000fragments. Very few samples were from indus-trial sites and oppida. No region produced aneven representation of samples from differenttypes of settlement.

Domestic fowl usually represent a minorcomponent of the assemblages but averages

408 M. Maltby

INT. J. OSTEOARCHAEOL., Vol. 7: 402±414 (1997) # 1997 by John Wiley & Sons, Ltd.

Domestic Fowl on Romano-British Sites 409

# 1997 by John Wiley & Sons, Ltd. INT. J. OSTEOARCHAEOL., Vol. 7: 402±414 (1997)

Table 2. Percentages of domestic fowl, sheep/goat and pig on Romano-British sites. Rg, region (see Table 3); D, assemblagedate (century AD); S, settlement type (r, rural; n, nucleated; v, villa; m, military; u, urban; i, industrial; o, oppidum), %F,percentage fowl of total number of fowl, sheep/goat and pig bones; %S, percentage sheep/goat; %P, percentage pig; FS,percentage fowl of total number of fowl and sheep/goat bones; FP, percentage fowl of total number of fowl and pig bones; NISP,number of fragments of fowl, sheep/goat and pig.

Settlement, Site Rg D S %F %S %P FS FP NISP

Abbotstone Down, Hants S 1 r 51 80 20 51 3 323Abingdon, Oxon WM 1±2 n 2 88 11 2 13 130Alcester, Warks WM 1±2 n 5 82 13 6 28 542

WM 3±4 n 7 62 31 10 17 3768Appleford, Oxon WM 2±4 r 2 80 18 2 8 66Bancroft, Bucks 1973±78 EM 1±4 v 10 55 36 15 22 2011Bancroft, Bucks 1983±86 EM 4 v 4 69 26 6 13 1817Barton Court Farm, Oxon WM 1±2 r 2 69 29 3 7 142

WM 3±4 v 7 70 23 9 22 1390Bishops Cleeve, Glos WM 2±4 v 3 72 25 5 12 144Brancaster, Norfolk EM 2±4 m 2 85 13 2 14 836Braughing, Skeleton Green SE 1 o 7 25 68 21 9 1772Brighton Hill South, Hants S 1 r 0 84 16 0 0 855Butter®eld Down, Wilts S 3±4 r 51 91 8 51 5 617Caerleon, Legionary Baths SW 1±3 m 32 33 35 48 47 500

SW 4 m 5 45 50 10 9 665Caerleon, Museum Gardens SW 1±2 m 23 27 51 46 31 420Caerwent SW 3±4 u 15 53 32 22 32 194Carlisle, Old Grape Lane N 1±2 n 2 62 36 3 4 237

N 2±3 n 2 52 46 4 5 461Carlisle, South Lanes N 1±2 n 2 60 39 3 4 576

N 2±3 n 51 48 52 2 2 378Charlton Kings, Glos WM 1 r 2 78 20 2 8 226Chester, Northgat WM 1±3 m 14 30 56 32 20 348Cirencester, Military sites WM 1 m 4 62 34 5 10 416Cirencester, Querns Road WM 2 u 12 60 28 17 30 280

WM 3±4 u 12 62 25 16 33 333Cirencester, Sheep St WM 1±2 u 4 65 31 6 12 534

WM 2±3 u 9 65 26 12 26 144WM 4 u 13 51 36 21 27 210

Coggeshall, Essex SE 1±4 r 1 79 20 1 5 103Colchester, Balkerne St SE 1 u 3 56 41 6 7 1282

SE 1±2 u 4 62 34 7 11 1972SE 2±4 u 14 49 37 22 28 2776

Colchester, Long Wyre St SE 1±4 u 24 40 37 37 39 1253Colchester, Middlesborough SE 1±4 u 18 41 41 30 30 210Colchester, Sheepen SE 1 i 5 39 56 11 8 2994Coombe Down, Wilts S 1±4 r 0 84 16 0 0 81Derby, Racecourse EM 2±3 i 51 86 13 1 6 476Dorchester, Alington Avenue S 1±4 r 51 82 18 51 3 352Dorchester, Greyhound Yard S 1 u 12 60 28 17 30 405

S 1±2 u 10 55 35 16 23 4534S 2±3 u 11 53 36 18 24 2631S 3±4 u 10 51 40 16 20 831S 4±5 u 14 49 37 22 27 2226S 5 u 16 48 36 25 31 798

Durrington Walls, Wilts S 3±4 r 0 90 10 0 0 126Earith, Hunts EM 3±4 r 0 88 12 0 0 100Easton Lane, Hants S 1 r 0 83 17 0 0 82Elsted, Sussex S 1±3 r 0 89 11 0 0 343Exeter, 1971±75 sites SW 1 m 8 54 38 14 18 725

SW 1 u 7 57 36 11 17 801SW 2 u 6 55 40 10 13 1128SW 3 u 12 40 48 23 20 377SW 4 u 13 44 43 23 23 1244

(continued )

410 M. Maltby

INT. J. OSTEOARCHAEOL., Vol. 7: 402±414 (1997) # 1997 by John Wiley & Sons, Ltd.

Table 2. Continued

Settlement, Site Rg D S %F %S %P FS FP NISP

Farmoor, Oxon WM 3±4 r 0 87 13 0 0 122Figheldean, Wilts S 1±4 r 0 81 19 0 0 209Frocester Court, Glos WM 1 r 51 69 30 51 51 769

WM 2±3 r 2 77 22 2 7 803WM 4 v 6 66 28 8 18 1080WM 4±5 v 0 42 58 0 0 284

Gloucester, East Gate WM 1±4 u 6 62 32 9 16 253Godmanchester, Hunts EM 3±4 v 5 73 22 7 20 296Gorhambury, Herts SE 1 v 4 41 55 9 7 606

SE 2 v 4 47 48 8 8 749SE 3±4 v 3 56 41 6 7 334

Great Chesterford, Essex SE 3±4 n 2 76 23 2 6 264Great Dunmow, Essex SE 2±4 n 2 83 15 2 10 231Ilchester, Almhouse Cutting A SW 1±2 u 17 66 17 20 49 220Lambeth, Borough High St SE 1±3 u 6 45 48 12 11 1658Latimer, Bucks EM 2±4 v 11 38 51 23 18 179Little Chester, Derby EM 1±3 m 10 68 22 13 31 672

EM 2±3 m 6 75 19 7 22 755EM 3±4 m 3 79 17 4 16 783

Little Somborne, Hants S 1 r 51 90 9 1 9 215S 4 r 0 86 14 0 0 118

Little Woolstone, Bucks EM 1±4 r 51 85 14 51 5 258Longthorpe, Cambs EM 1 m 3 66 31 5 10 902Mantles Green, Bucks EM 2±4 v 51 82 18 51 3 1299Magiovinium, Bucks EM 2±3 n 4 79 17 5 18 1179Micheldever Wood, Hants S 1±4 r 0 70 30 0 0 510Middleton Stoney, Oxon WM 1±4 r 2 81 17 3 13 208Ower, Dorset S 1 r 51 36 64 51 51 372

S 4 r 6 65 29 8 17 230Owslebury, Hants S 1 r 51 69 31 51 1 8362

S 1±2 r 51 72 28 1 3 3457S 3±4 r 51 83 16 1 5 6635

Portchester Castle, Hants S 3±4 m 4 53 44 7 8 6096Puckeridge, Herts SE 1 o 3 48 48 7 7 923Raynes, Essex SE 1±4 r 2 81 17 2 11 539Roughground Farm 1957±82 WM 3±4 v 0 79 21 0 0 74Roughground Farm 1990 WM 3±4 v 6 64 31 8 15 127Scole, Norfolk EM 1±2 n 5 81 13 6 29 292

EM 3±4 n 2 78 20 2 9 562Silchester, South Gate S 1 u 51 77 23 51 3 142

S 1±2 u 2 55 42 4 5 165S 4 u 7 41 51 15 13 177

Slonk Hill, Sussex S 1±4 r 51 55 45 1 2 543Stantonbury, Bucks EM 1±4 r 0 92 8 0 0 730Wantage, Oxon WM 1±4 v 3 84 14 3 16 197Watercrook, Cumbria N 1±3 m 51 68 32 51 2 964West Stow, Suffolk EM 1±2 i 51 74 26 51 2 378Westhampnett, Sussex S 1±2 r 5 71 25 6 16 106Whitcombe, Dorset S 3±4 r 0 91 9 0 0 209Wilcote, Oxon WM 1±3 n 4 83 13 4 22 3303Winchester, Staple Gardens S 1±2 u 3 57 41 5 6 148

S 3±4 u 2 61 36 4 6 610Winchester, Victoria Rd X±XVI S 2±4 u 4 72 24 5 14 4149Winchester, Victoria Rd IV±VI S 2±4 u 2 64 34 3 6 1661Winchester, Hyde Abbey S 2±4 u 2 71 26 3 8 1073Winchester, City Defences S 2±4 u 68 30 3 7 1514Winnall Down, Hants S 1 r 51 86 13 51 2 962Worcester, Sidbury WM 1±2 n 2 85 13 2 12 532

WM 3 n 2 83 16 2 9 453WM 3±4 n 51 75 25 51 2 665

(continued )

between settlement types do vary (Figure 2).Overall, domestic fowl and pigs were bestrepresented on urban sites and pigs are alsowell represented on military and villa sites.Domestic fowl were generally moderately re-presented on military and villa sites. Both pigsand domestic fowl were less well represented incomparison with sheep/goat on nucleated sitesand also on other rural sites. Domestic fowl werecommonly absent at these sites. The percentageof domestic fowl and sheep/goat combinedappears to give the greatest discriminationbetween settlement types (Table 4).

There was remarkably little variability in thedomestic fowl:sheep/goat percentages from ruralsites (Table 5), where 71 per cent of the samplescontained less than 1 per cent domestic fowl,and 89 per cent had less than 2 per centdomestic fowl. Only one (6 per cent) of thesamples from nucleated settlements had less than

1 per cent domestic fowl; however, none hadmore than 10 per cent. Counts from the villasites were very variable, but the number of sites(15) is small. Domestic fowl were consistentlybetter represented on villa sites than on otherrural sites, with 86 per cent of the sampleshaving between 3 per cent and 10 per cent fowl.In the 13 military samples, 61 per cent of thesites had over 6 per cent domestic fowl, but thesamples are few and rather variable.

This was also the case with the samples fromthe major urban sites, although there were manymore of these. Many produced large percentagesof domestic fowl: 79 per cent produced over 6per cent domestic fowl and in 53 per centdomestic fowl contributed over 16 per cent ofthe total of domestic fowl and sheep/goat bones.Of the eight urban samples with less than 6 percent domestic fowl, six were from Winchester; asmost of these samples were from suburban sites

Domestic Fowl on Romano-British Sites 411

# 1997 by John Wiley & Sons, Ltd. INT. J. OSTEOARCHAEOL., Vol. 7: 402±414 (1997)

Table 2. Continued

Settlement, Site Rg D S %F %S %P FS FP NISP

Woughton, Bucks EM 1±4 r 0 82 18 0 0 116Wroxeter WM 1±2 u 12 51 37 19 25 429

WM 2 u 9 46 45 16 16 488WM 3 u 4 51 45 7 9 1564

Wycomb, A40 site, Glos WM 1±4 n 4 53 43 7 8 150York, Tanner Row N 1±2 u 20 36 44 36 32 220

N 2±3 u 15 47 38 24 28 2459N 4 u 15 43 42 26 27 105

Table 3. Samples by settlement type and region. Columns 9±12 shows samples in four size classes. S, South (Dorset,Hampshire, Wiltshire, West Sussex, Berkshire); SE, South East (Kent, East Sussex, Surrey, Middlesex, Essex, Hertfordshire);SW, Wales and South West (Devon, Cornwall, Somerset, Avon); EM, East Midlands (Norfolk, Suffolk, Cambridgeshire;Huntingdonshire, Buckinghamshire, Leicestershire, Northamptonshire, Lincolnshire, Bedfordshire, Derbyshire,Nottinghamshire); WM, West Midlands (Oxfordshire, Gloucestershire, Warwickshire, Hereford and Worcestershire,Staffordshire, Shropshire, Cheshire); N, North (Lancashire, Yorkshire, Cumbria, Durham).

Region NISP

Type ofsettlement S SW SE EM WM N Total 5200 201±500 501±1000 51000

Major urban 15 6 6 9 3 39 7 11 6 15Nucleated 2 3 8 4 17 2 7 5 3Villa 3 5 7 15 5 3 2 5Industrial 1 2 3 2 1Other rural 21 2 4 7 34 10 12 9 3Military 1 4 5 2 1 13 4 8 1Oppidum 2 2 1 1Total 37 10 16 19 33 8 123 24 39 31 29

rather than from the centre of the city, this maybe a factor in the low representation of domesticfowl. The other two urban samples with less than6 per cent domestic fowl were small early Romanassemblages from near the South Gate atSilchester and are probably atypical.

Conclusions

All conclusions from this analysis must remaintentative, but some trends have been identi®ed.These deserve further investigation. Does thefact that domestic fowl are consistently morecommonly recovered from urban sites than fromrural sites re¯ect a genuine difference in theirabundance and contribution to the diet?Undoubtedly, urban assemblages are often betterpreserved, but some of the rural samples, such asOwslebury, also had quite good preservation andfowls were few there. It also seems unlikely that

variable retrieval rates can explain these con-sistent differences.

Domestic fowl tend to be higher on siteswhere pigs are also well represented. This mayre¯ect dietary preferences of different groups asKing1 has argued. It may also indicate that bothspecies were commonly kept in towns.

Further analysis will probably also demon-strate more clearly differences in the exploitationof domestic fowl between other types ofsettlement. Here the preliminary evidence sug-gests that Romanized settlements, that is majortowns, some military sites and some villas, tendto produce high percentages of domestic fowl.Samples from unromanized sites, that is rural andmost nucleated settlements, tend to produce lowpercentages. Further studies are needed toexamine these patterns further. However, thereare indications that domestic fowl were morepopular on Romanized sites but less readilyaccepted as a food resource on native settle-ments. The spread in the popularity of domesticfowl in Roman Britan over time also meritsfurther investigation. However, the variability inthe results from different types of site and therelatively small number of samples availablemeans that comparisons between regions orthrough time are premature.

Methods must be devised to improve theseintersite comparisons. Large numbers of suitablylarge sieved samples are not yet available. Inorder to try and minimize biases of retrieval,fragmentation and preservation, a method maybe to select only certain bones of sheep/goat andpig for comparison. Small bones such as thecalcaneum could be compared with domesticfowl bones, because they are rarely recovered ina fragmentary state and are not particularlyrobust. However, using just one or two bonetypes in such comparisons will rarely producesuitably large samples. A selection of smaller andless robust bones that are more comparable withfowl bones could be used, such as the ulna,hyoid, calcaneus and astragalus and ®rst pha-langes of sheep, the hyoid, ®bula, calcaneus,astragalus, lateral metabopodials and ®rst (axial)phalanges of pig. However, not every workerrecords all these bones. Alternatively, domesticfowl bones could be compared with some of theless robust articulator surfaces of limb bones

412 M. Maltby

INT. J. OSTEOARCHAEOL., Vol. 7: 402±414 (1997) # 1997 by John Wiley & Sons, Ltd.

Figure 2. Percentages of domestic fowl, pig and sheep/goatfrom different Romano-British settlement types, calculated bytotalling NISP from all sites.

Table 4. Percentage of bones of domestic fowl plus sheep/goat (FS) and of fowl plus pig (P): average by settlement type.

FS FP

Major urban 14 20Military 9 14Villa 8 14Nucleated 5 17Other rural 0.9 3

(distal scapula; proximal humerus; proximal anddistal femur; proximal tibia). Exclusion of shaftfragments would remove the problem of differ-ential fragmentation. However, variablecarnivore destruction of articular surfaces maybe a factor that cannot easily be controlled.Again the number of comparable samples isdecreased because not everybody recordswhether the articular surfaces are present.

Some initial analysis using a selection of suchbones is currently taking place. It is hoped thatthe results of that study will expand thispreliminary survey and will form a separatepublication. It should provide a more accuratecomparison of the relative numbers of sheep,pigs and domestic fowl consumed. There areindications that on some sites the view thatchicken was a `useful supplement to the diet' maygreatly underestimate the frequency of itsconsumption.

Acknowledgements

I am grateful to Pippa Smith, Sue Stallibrass andBeverley Meddens for the use of unpublished data.

References

1. King, A. Animal bones and the dietary identity ofmilitary and civilian groups in Roman Britain,Germany and Gaul. In: Military and Civilian inRoman Britain (edited by T. Blagg and A. King).Oxford: British Archaeological Reports (BritishSeries) 136, 1984: 187±217.

2. Maltby, M. The meat supply in RomanDorchester and Winchester. In: Urban±RuralConnexions: Perspectives from EnvironmentalArchaeology (edited by A. Hall and H.Kenward). Oxford: Oxbow Monograph 47,1994: 85±102.

3. Maltby, M. Iron Age, Romano-British and Anglo-Saxon animal husbandry Ð a review of the faunalevidence. In: The Environment of Man: the Iron Age tothe Anglo-Saxon Period (edited by M. Jones and G.Dimbleby). Oxford: British ArchaeologicalReports (British Series) 87, 1981: 155±203.

4. Parker, A. The birds of Roman Britain. OxfordJournal of Archaeology, 1988; 7: 197±226.

5. King, A. A comparative survey of bone assem-blages from Roman sites in Britain. Institute ofArchaeology Bulletin, 1978; 15: 207±232.

6. Luff, R. A Zooarchaeological Study of the RomanNorth-western Provinces. Oxford: BritishArchaeological Reports (International Series)137, 1982.

7. Coy, J. The provision of fowls and ®sh for towns.In: Diets and Crafts in Towns (edited by D.Serjeantson and T. Waldron). Oxford: BritishArchaeological Reports (British Series) 199,1989: 25±40.

8. Coy, J. [this volume]

9. O'Connor, T. Bones from 46±54 Fishergate. TheArchaeology of York, Vol. 15, Fascicule 4.London: Council for British Archaeology, 1991.

10. Bourdillon, J. and Coy, J. The animal bones. In:Excavations at Melbourne Street, Southampton 1971±76(P. Holdsworth). London: Council for BritishArchaeology Research Report 33, 1980: 79±121.

11. Maltby, M. Patterns in faunal assemblage varia-bility. In: Beyond Domestication (edited by G. Barkerand C. Gamble). London: Academic Press, 1985:33±74.

Domestic Fowl on Romano-British Sites 413

# 1997 by John Wiley & Sons, Ltd. INT. J. OSTEOARCHAEOL., Vol. 7: 402±414 (1997)

Table 5. Variations in the percentages of domestic fowl compared with sheep/goat: %fowl, percentage domestic fowl of totalnumber of domestic fowl and sheep/goat bones; (%) percentage of samples of relevant settlement type falling within eachfrequency range.

% fowlUrban

(%)Military

(%)Villa(%)

Nucleated(%)

Rural(%)

51 3 8 20 6 711±2 0 8 0 41 183±5 18 23 13 29 96±10 15 23 53 24 311±15 10 15 7 0 016±20 23 0 0 0 021±25 21 0 7 0 0426 10 23 0 0 0Number of samples 39 13 15 17 34