Animals In Roman Times In the Dutch Eastern River Area

RCGM LAUWERIER^IVJIIVIALS IN ROMAN TIMES IN THE DUTCH EASTERN RIVER AREA Lauwerier, R. 1988.pdf m(im NEDERLANDSE OUDHEDEN 12 [ PROJECT OOSTELIJK RIVIERENGEBIED i NEDERLANDSE OUDHEDEN Editorial Board: W.A. van Es, J.F. van Regieren Altena and G. H. Scheepstra RIJKSDIENST VOOR HET OUDHEIDKUNDIG BODEMONDERZOEK, Kleine Haag 2, 3811 HE Amersfoort, The Netherlands 1 J.E. BOGAERS De Gallo-Romeinse tempels te Eist in de Over-Betuwe. 1955 (out of print). 2 W. GROENMAN-VAN WAATERINGE Romeins lederwerk uit Valkenburg Z.H. 1967 3 P.J.R. MODDERMAN Linearbandkeramik aus Elsloo und Stein. 1970, 3 vols. (out of print). 4 J.A. BRONGERS 1833: Reuvens in Drenthe. 1973. 5 T. CAPELLE Die frühgeschichtlichen Metallfunde von Domburg auf Walcheren. 1976, 2 vols. 6 J.A. BRONGERS Air Photography and Celtic Field Research in the Netherlands, i^ld, 2 vols. 7 T. CAPELLE Die karolingischen Funde von Schouwen. 1978, 2 vols. 8 J.H.F. BLOEMERS Rijswijk (Z.H.), 'De Bult', eine Siedlung der Cananefaten. 1978, 3 vols. 9 W.A. VAN ES/W.J.H. VERWERS Excavations at Dorestad i, The Harbour : Hoogstraat I. 1980. 10 G.F. IJZEREEF Bronze Age Animal Bones from Bovenkarspel. 1981. 11 W. PRUMMEL Excavations at Dorestad 2, Early medieval Dorestad, archaeozoological study. 1983. } ROEL C.G.M. LAUWERIER ANIMALS IN ROMAN TIMES IN THE DUTCH EASTERN RIVER AREA NEDERLANDSE OUDHEDEN 12 | OOSTELIJK RIVIERENGEBIED I SDU uitgeverij, 's-Gravenhage Rijksdienst voor het Oudheidkundig Bodemonderzoek, Amersfoort ISBN 9012058244 NEDERLANDSE OUDHEDEN 12 PROJECT OOSTELIJK RIVIERENGEBIED I ANIMALS IN ROMAN TIMES IN THE DUTCH EASTERN RIVER AREA ROEL C.G.M. LAUWERIER 3 Amersfoort 1988 This study was carried out at the Biologisch-Archaeologisch Instituut of the University of Groningen, thanks to a grant from the Foundation for Archeological Research, subsidized by the Netherlands Organisation for the Advancement of Pure Research (zwo) (1980-84). In 1985 this research was financially supported by the Rijksdienst voor het Oudheidkundig Bodemonderzoek. The Computing Centre of the University of Groningen offered facilities in the last phase of this study. Contents Acknowledgements 9 1 Introduction n 1.1 ARCHEOLOGICAL RESEARCH IN THE EASTERN RIVER AREA ii 1.2 THE ARCHEOZOOLOGICAL RESEARCH i3 1.3 THE SITES i6 2 Methods i8 2.1 THE EXCAVATION METHODS i8 2.1.1 THE NATURE OF THE EXCAVATIONS l8 2.1.2 2.1.3 ARCHEOZOOLOGICAL CONSEQUENCES COLLECTING BY HAND AND SIEVING: AN EXPERIMENT l8 21 2.2 THE ARCHEOZOOLOGICAL METHODS 27 2.2.1 RECORDING AND PROCESSING OF THE DATA 27 2.2.2 THE SPECIES DETERMINATION 27 2.2.2.1 General points 27 2.2.2.2 Aurochs and domestic cattle 28 2.2.2.3 Wild boar and pig 30 2.2.2.4 Sheep andgoat 31 2.2.3 THE QUANTITATIVE ANALYSIS 31 2.2.4 MEASUREMENTS 33 2.2.5 2.2.6 SEX WITHERS HEIGHT 33 34 2.2.7 2.2.8 AGE DETERMINATION OF THE SEASON OF SLAUGHTER 35 36 2.2.9 2.2.10 BUTCHERY MARKS PATHOLOGICAL CONDITIONS, TRACES OF GNAWING, BONE ARTEFACTS, 40 ETC. 42 3 The bone material in the settlements 43 3.1 NIJMEGEN IN THE EARLY ROMAN PERIOD 43 3.1.1 3.1.2 INTRODUCTION THE BONES THAT MAY POSSIBLY BE ASSOCIATED WITH THE CASTELLUM 43 (NIJMEGEN la) THE BONES FROM THE SETTLEMENT ON THE VALKHOF AND 47 3.1.3 SURROUNDINGS (NIJMEGEN Ib-c) 50 3.2 3.3 NIJMEGEN IN THE MID-ROMAN PERIOD 52 3.2.1 INTRODUCTION 3.2.2 THE BONES FROM THE CASTRA 3.2.3 THE BONES FROM THE CANABAE LEGIONIS 3.2.4 THE RELATION BETWEEN THE CASTRA AND THE CANABAE LEGIONIS 52 53 57 59 NIJMEGEN IN THE LATE ROMAN PERIOD 64 3.3.1 INTRODUCTION 3.3.2 THE 4TH-CENTURY SETTLEMENT ON THE VALKHOF (NIJMEGEN IV) 3.3.2.1 3.3.2.2 3.3.2.3 3.3.2.4 Introduction The hand-collected material The sieved samples The density of finds 64 64 64 3.3.3 THE 4TH-CENTURY CEMETERY 'MARGRIET' 3.3.3.1 Introduction 3.3.3.2 Dishes and bones 3.3.4 3.4 3.5 3.6 THE RELATION BETWEEN THE SETTLEMENT AND THE CEMETERY THE CASTELLUM AT MEINERSWIJK 86 3.4.1 3.4.2 86 86 INTRODUCTION THE BONES THE MILITARY VICUS AT KESTEREN 88 3.5.1 3.5.2 88 88 INTRODUCTION THE BONES NATIVE FARMSTEADS 90 3.6.1 3.6.2.1 Introduction 3.6.2.2 The bones 90 90 90 92 92 93 THE VILLA IN DRUTEN 95 THE NATIVE SETTLEMENT IN HETEREN 3.6.1.1 Introduction 3.6.1.2 The bones 3.6.2 3.7 THE NATIVE SETTLEMENT IN EWIJK 7.1 7.2 INTRODUCTION THE BONES 7.3 THE RELATIONSHIP BETWEEN BONES, BUILDINGS AND OCCUPATION 7.3.1 Druten II 7.3.2 Druten III 7.4 FOUNDATION DEPOSITS CONSISTING OF ANIMAL SACRIFICES 7.4.1 The circumstances of the finds 7.4.2 The horse burials: whole or half skeletons? 7.4.3 Additional data 3.8 67 72 74 76 76 76 82 THE GALLO-ROMAN TEMPLES AT ELST 8.1 INTRODUCTION 8.1.1 8.1.2 8.2 8.2.1 8.2.2 8.2.3 8.2.4 The period before the temples were built The temple period THE BONES The period before the temples were built The temple period Suovetaurilia Discussion and conclusions 95 97 97 98 99 104 104 105 no III III III III 112 113 "5 118 119 4 Production and consumption of animal products 122 4.1 THE DOMESTICATED (MEAT-YIELDING) MAMMALS 122 4.1.1 4.1.2 THE GENERAL PICTURE THE OCCURRENCE OF PIG AND THE FACTORS OF THE ENVIRONMENT AND 122 THE NATURE OF THE SETTLEMENTS THE OCCURRENCE OF SHEEP/OOAT AND THE FACTORS OF THE I26 4.1.3 ENVIRONMENT AND THE NATURE OF THE SETTLEMENTS I28 4.1.4 THE OCCURRENCE OF CATTLE AND THE FACTORS OF THE ENVIRONMENT AND THE NATURE OF THE SETTLEMENTS 4.2 128 THE PRODUCTIVE VALUE OF PIG, SHEEP/GOAT AND CATTLE 129 4.2.1 4.2.2 129 I3I PRODUCTION PRODUCTION OF WOOL, DUNG, MEAT AND MILK PIG: MEAT SHEEP: 4.2.3 CATTLE: 4.2.4 AGRICULTURE AND STOCK-BREEDING PRODUCTION OF TRACTION POWER, MANURE, MILK AND MEAT 133 14O 4.3 THE POULTRY YARD 142 4.4 HUNTING i43 4.5 FISHING 147 4.6 MOLLUSCA 150 4.7 THE SLAUGHTERING OF LIVESTOCK AND THE PROCESSING OF MEAT 150 4.7.1 DOG 153 4.7.2 HORSE 153 4.7.3 4.7.4 CATTLE SHEEP/GOAT 155 157 4.7.5 4.7.6 PIG WILD MAMMALS 159 159 4.8 THE MENU i6o 5 The animals not used for human consumption 162 5.1 HORSE 162 5.2 DOG 164 6 Withers height of the farm animals 166 6.1 INCREASE IN THE SIZE OF CATTLE: ROMANIZATION IN STOCK-BREEDING PRACTICES 166 THE WITHERS HEIGHTS OF THE OTHER FARM ANIMALS 169 6.2 6.2.1 SHEEP/GOAT 169 6.2.2 PIG 170 6.2.3 HORSE 170 Summary 174 Samenvatting i77 Appendix: Butchery mark code igi Abbreviations 213 References 214 Photographs and figures 224 Tables m1-m35 on microfiches 225 Ackno\A/ledgements I wish to express my sincere gratitude to: Dr. A.T. Clason, head of the Archeozoological Department of the Biologisch-Archaeologisch Instituut (BAI), for initiating this investigation, for giving me the opportunity to carry it out, for constant guidance, encouragement, expert support and the freedom she gave me throughout this study. Prof. Dr. G.J. Boekschoten (University of Groningen) for being my promotor, for advice, criticism and encouragement. Prof. Dr. J.H.F. Bloemers (Rijksdienst voor het Oudheidkundig Bodemonderzoek (ROB)/University of Amsterdam), who headed the Eastern River Area project and who, together with Dr. A.T. Clason, apphed for the grant to finance this research, for valuable discussions and critical reading of preliminary versions of the manuscript. Mr. R.S. Hulst, Mr. P.A.M. Zoetbrood and Dr. W.J.H. Willems discussions and practical help. (ROB) for Prof. Dr. W.A. van Es (ROB) for reading the preliminary version of the manuscript and for the opportunity to work within the framework of the Eastern River Area project (ERA) and to publish this investigation as a volume of the series 'Nederlandse Oudheden'. Prof. Dr. H.T. Waterbolk (BAI) for advice. My colleagues at the Archeozoological Department of the BAi, Mr. D.C. Brinkhuizen, Mr. H. Buitenhuis, Ms. A.M.P. Kersten, Dr. W. Prummel and Mr. J.T. Zeiler for stimulating discussions, and Mr. T.P. Jacobs and Mr. R.J. Kosters for technical assistance. I gladly remember the good times we had together. Mr. R. de Bruin, Mr. O. van Schalkwijk and Mr. L.Th. van der Weele (Computing Centre, RUG) for their interest. I thank Mr. L.Th. van der Weele for advice and help with the cluster analysis. Ms. M. Bierma the references. (BAI) for hbrarian help and Mr. W.C. Mank (ROB) Mr. S.P. Cordes and Mr. R.J. van Ewyck (Centrale Fotodienst, the photographs. for correcting RUG) for making Mr. H.R. Roelink (BAI), Mr. H.M.C, de Kort, Mr. J.C.A. Hulst and Mr. J.H. van Vlierden (ROB) for preparing the drawings. Mr. A. Buisman and Mr. K. Greving (ROB) for their help in taking sieve-samples and their hospitality at the excavations. Ms. S.M. and Gelder-Ottway M.Sc. for translating the text into English. Mr. G.H. Scheepstra (ROB) for editing this publication. All other members of the Biologisch-Archaeologisch Instituut. My wife, Jacqueline E.M. Lauwerier-Verdenius, for criticism and help in many ways, in spite of her own busy work. 10 1 Introduction 1.1 ARCHEOLOGICAL RESEARCH IN THE EASTERN RIVER AREA The Eastern River Area is the district around the town of Nijmegen, and is characterized by the presence of the three rivers, the Rhine, the Waal and the Meuse. The region is covered approximately by the sheets of the Topografische Kaart van Nederland (Topographical Map of the Netherlands) scale i: 25,000 39F-H, 40 A-B-C-D-E-G, 45 F, 46 A-B and includes the adjacent region in Germany (fig. i) (Bloemers et al. 1980b). In the Roman period, approximately from the beginning of the Christian era up to AD 400, the Eastern River Area was a densely populated district for that time (Willems 1983). It was the centre of the home territory of the native tribe of the Batavi. Other inhabitants of the region were Roman or Gallo-Roman immigrants. During the entire Roman period it formed part of the border region in the northwestern part of the Roman Empire, that from the middle of the i st century until the second half of the 3rd century, and probably also thereafter, was protected by a chain of fortifications (Bogaers and Rüger 1974: map 2). The Roman influence made itself strongly felt, and as a result the original inhabitants underwent a more or less drastic acculturation process (Van Es 1981; Bloemers 1983). In the 3rd and 4th centuries the region was subject to invasions launched by Germanic tribes. The number of settlements in the Eastern River Area is very large. In a study on the region made by Willems (1983) almost 400 sites are enumerated. The types of settlement vary considerably. Native farmsteads have been found in Heteren and Ewijk (Hulst 1971a; Hulst and Noordam 1974), native or Gallo-Roman farmsteads of a more elaborate type in Druten (Hulst 1978). Small military fortifications, castella, were situated in Nijmegen and along the Rhine, for example at Meinerswijk and most probably also at Kesteren (fig. 2) (Noviomagus 1979; Willems 1980a; Bogaers and Rüger 1974). The economic, military and administrative centre of the district was Nijmegen. It was the chief town of the civitas Batavorum and from the time of the reign of the emperor Trajan (98-117) it was called (Ulpia) Noviomagus. From a military point of view Nijmegen is characterized by the presence of legionary camps, of which notably the mid-Roman castra with the surrounding canabae legionis have left their traces (Noviomagus 1979). To the north of Nijmegen, in Eist, there was a Batavian sanctuary, presumably the national temple of the Batavi (Bogaers 1955)The geological situation in the region is determined by the presence of the great rivers: the Rhine, the Waal and the Meuse. The Holocene fluviatile deposits in the centre of the region are flanked by much higher Pleistocene soils, such as the II Fig. I Sites in the Netherlands dating from the Roman period that have been investigated archeozoologically: • castella, • other settlements. The region outlined is the Eastern River Area. ice-pushed ridges to the north of the Rhine, to the south of Nijmegen, and the cover-sands of North Brabant in the southwest (fig. 2) (Willems 1981a). Since c. 1974, studies that have been made in the Eastern River Area have been brought together and integrated within the framework of the Eastern River Area project (ERA) of the Rijksdienst voor het Oudheidkundig Bodemonderzoek (ROB) (Van Es 1977). The chief aim of the research within the ERA project is the study of the socio-economic developments in the Eastern River Area based on the analysis of historical, archeological, geographical and ecological data. The research is concentrated on the Roman period, but the whole time-span covered begins in the late Iron Age and ends in the early Middle Ages (Bloemers et al. 1980b). The archeozoological research, as based on the analysis of the finds of faunal material, attempts to meet the above-described objectives, concerning specifically hunting, stock-breeding and the natural environment. 12 iiiiiiiiiiiiiiiiiiiiii:i^^^^^^^ Fig. 2 The Eastern River Area. A reconstruction of the geological situation during the Roman period showing the situation of the settlements. I Pleistocene deposits, 2 fiood-basin deposits and peat, 3 pre-Roman channel zone deposits, 4 Roman channel zone deposits, 5 present-day river-channels, 6 boundaries of deposits, 7 reconstructed boundaries of deposits, 8 castra, 9 castellum, 10 possible castellum, 11 town, 12 agricultural settlement, 13 temple. Scale i : 250,000 (after Willems i98i,fig. 13). AND^VAN"euY'KIS/^ïI S, •8 D10 m 11 • 12 o 13 1.2 THE ARCHEOZOOLOGICAL RESEARCH In the Eastern River Area, little research has been carried out in the past on the archeozoological aspects of the inhabitation during Roman times. The only extensive report that has been published in this field concerns the faunal material that was found during the excavation of the Gallo-Roman temples in Eist (Kortenbout van der Sluijs and Audretsch 1955). Other publications mainly give a list of species found during incidental excavations such as those of the castra of 13 Nijmegen (Clason 1977a: table 15*1; Thijssen 1976; 1977; 1980) and of the native settlement of Heteren (Clason 1977a: table 15). Some aspects of the bone research in the area have been published by Lauwerier (1986a; 1986b). In the other areas of the Roman Netherlands research on hunting and stock-breeding has been incidental and so far no comprehensive study on this subject has been published (Van Es 1981). Publications have appeared concerning the faunal material from the castella at Velsen (Clason 1967; Vons 1977; Gordijn-Vons 1977), Zwammerdam (Van Wijngaarden-Bakker 1970) and Valkenburg (Clason i960; 1967; Prummel 1975) (fig. i). In addition data have been published concerning a few native settlements: Vlaardingen (Clason 1967), Rijswijk (Clason 1978) and Leiderdorp (Van Mensch 1975). The archeozoological material from the native settlements of Houten and Rijswijk has recently been the subject of two undergraduate research projects carried out at the Biologisch-Archaeologisch Instituut in Groningen (Taayke 1984; Ket 1987). Data on the fish and birds from the castella of Valkenburg and Velsen have been included in three articles of a more general nature (Brinkhuizen 1979a; Clason and Prummel 1979; Clason et al. 1979). An article on hunting in the Netherlands in Roman times has been published by Van Mensch (1977). Butchery and processing methods for cattle from the forts of Zwammerdam and Valkenburg are discussed by Van Mensch (1974) and Van Mensch and IJzereef (1977). Part of the bone material that was found in the graves of a 4th-century cemetery in Nijmegen is dealt with in an article on bone finds in graves from the Roman period (Lauwerier 1983b). Luff (1982) gives a summary of the archeozoological research of the northwestern provinces of the Roman Empire, and King (1984) a summary of the finds of cattle, sheep/goat and pig within the borders of the Roman Empire. From the unconquered Netherlands to the north of the limes, bone material has been published from the settlements of Egmond, Wij ster, Sneek, Paddepoel and Kimswerd (Clason 1984; 1965; 1962; Knol 1983; Milojkovic and Brinkhuizen 1984). The archeozoological research in the Eastern River Area is intended to be a comprehensive study of hunting, stock-breeding and related aspects of the natural environment in the district around Nijmegen. The research is based on the study of subfossil faunal material collected from archeological excavations of various kinds of settlements dating from Roman times in the Eastern River Area. The aim of this research is to provide answer to the question which role the animals played in Roman times in the Eastern River Area. From pedological research we known that in Roman times the Eastern River Area was an ecologically diverse region, including on the one hand the low-lying parts close to the rivers where water was abundant, with slightly higher ridges of land that were suitable for stock-breeding and arable farming, and on the other hand the considerably higher soils of the Veluwe, the Rijk van Nijmegen, the Reichswald and Het Land van Cuyk. The lower-lying area would have been rich in wildfowl and fish. The higher soils were presumably still largely covered with forest and suitable for the hunting of larger game and for pasturing swine. One of the aims of this study is to ascertain to what extent this ecological situation is reflected in the species composition of livestock and game animals in the different *i The data are erroneously mentioned in the table under 'Noviomagus; Roman town'. This should read: 'Nijmegen, castra'. As a consequence of this error, in the publications by Luff (1982: table 5: 2, 7, ix) and by King (1984: table 4) data for the castra are presented as being data for the civilian settlements. 14 kinds of settlements situated in different environments. Similar studies, demonstrating this relationship, have already been carried out for other localities (Nobis 1955; Prummel 1979a; Reichstein 1975). The pattern of consumption for the various settlements has been investigated per settlement. On the basis of the data obtained the settlements are compared with one another. For each settlement an attempt has been made to determine the composition of livestock and game animals as far as possible, with respect to species, age and relative size. Moreover, the relationship to the excavated soil traces is established as far as possible. The questions that arise here are, for example: where was butchering carried out, and where did people eat? Did the Roman soldiers eat meat that differed from that consumed by the native inhabitants? Which butchery techniques were used? Were animals used for sacrifices? What was the economic function of animals, other than for meat consumption? What kinds of grave-goods of animal origin were provided at the burial of the dead? The comparison of these data for the various settlements can indicate the extent to which the pattern of consumption of the settlements differ (for example military versus civilian, Roman versus native), what the relative proportions were of livestock and game animals (for example legionary fort versus native farmstead) and what kind of mutual relationship may have existed (production of local inhabitants for the army; import by the army from elsewhere?). The Romans are known to have been good stock-breeders (White 1970). It is therefore also possible that the Roman presence in the Eastern River Area had some influence on stock-breeding. For example, they may have introduced better exploitation techniques, a better planned and more purposeful breeding programme, and the import of better breeding animals. The presence in Nijmegen of a large Roman military unit, notably in the period AD 70-104, and the concentration of a civilian population that developed in its immediate neighbourhood could have been a stimulus towards the introduction of these improved breeding methods, that would have resulted in a greater supply of meat being available to feed the Roman army and especially the civilians of the town. In other words, did the Romanized farmer start producing for a market? Differences in the composition of the livestock that cannot be associated with environmental factors could be the result of a change in exploitation and management. In Southern Germany, for example, it is known that in Roman times cattle were kept that were larger than those of the immediately preceding period. This is explained by the introduction of better breeding animals from Italy (Zeuner 1967; Boessneck et al. 1971). Not all of the find-complexes have been analyzed to the same extent. On the one hand the amount of time available plays a role here, on the other hand the nature and quantity of the material determine the value of the data obtainable for further analysis. Thus the material from Ewijk has been identified only as to species and skeletal element. In the case of other complexes the material has also been studied to obtain data on other variables such as sex, age at time of slaughter, measurements, pathological condition, and the presence of any traces of burning, gnawing or butchery. The extent to which these data were analyzed varies from site to site. In chapter 3 an outline description of each site is given, together with details of the total amount of bone material found per site. The information provided includes the percentage distribution, in terms of both numerical frequency and weight of the bones of the various species, the degree of identifiability of the material, and a re-estimation of the frequency- and weight-percentages with the unidentifiable material included (see 2.2.3). A few complexes are analyzed further in chapter 3. These are the castra and the 15 canabae legionis of Nijmegen, the 4th-century settlement and the cemetery of Nijmegen, the villa in Druten and the sanctuary of Eist. These complexes were selected primarily because they provide sufficient material for further analysis. In addition, in the case of a few sites, such as the 4th-century cemetery presently called 'Margriet' and the sanctuary in Eist, despite the small quantity of bones present the material was further analyzed in connection with the special nature of these sites. The more detailed data on the remaining complexes, often combined with the data on other complexes, are presented in chapter 4, 5 and 6 in the discussion of the Eastern River Area as a whole. For reasons of efficiency, in chapter 2, in the section 'methods', under the heading 'the species determination' (2.2.2), for aurochs/domestic cattle and wild boar/pig not only the methods are described but also the results of these species determinations. 1.3 THE SITES The faunal remains from the sites described immediately below constitute the material studied in this research. A more detailed description of the sites is given in chapter 3. The dates mentioned refer to the period from which the bone material has been derived. For mid-Roman Nijmegen data ar lacking for the period after AD 120, the time when the city of Noviomagus flourished. In the sandy soil of the sites of this period not a single bone has been preserved (personal communication J.K. Haalebos, KUN). The name given in brackets after the brief description of each site is the name used from here on in this publication to refer to the site concerned. Nijmegen in the early Roman period (15/10 BC-AD 70) Finds from the period up to AD 25, possibly connected with a small castellum found on the Trajanusplein and its surroundings. (Nijmegen la) Finds from the period after AD 25, belonging to the early Roman settlement on the Valkhof and its surroundings. (Nijmegen Ib-c) Nijmegen in the mid-Roman period (AD 70-120) The legionary fort on the Hunerberg. (Nijmegen castra) The surroundings of the army barracks, the canabae legionis. (Nijmegen canabae) Nijmegen in the late Roman period (4th century) A defensive ditch of the late Roman settlement on the Valkhof. (Nijmegen IV) The cemetery to the east of the settlement on the Valkhof. ('Margriet' cemetery) Meinerswijk: a castellum that was in use from the beginning of the ist century until the beginning of the 5th century. (Meinerswijk) AD Kesteren: A military vicus that was most probably situated next to a castellum, the location of which is not (yet) known. Inhabited from about AD 70 until some time in the 3rd century. (Kesteren) Heteren: a native settlement, inhabited from the middle of the ist century AD until the middle of the 2nd century. There may also have been occupation at an earlier date. (Heteren I and II) 16 Ewijk: Native settlement that was inhabited from the pre-Roman Iron Age until the end of the 2nd century AD. (Ewijk I and II) Drutcn: A native settlement that under the influence of the military presence in Nijmegen developed into a villa. Inhabited from the pre-Roman Iron Age (?) until the end of the 2nd century AD. (Druten I, II and III) Eist: Gallo-Roman temple complex. Two successive temples from AD 50-70 and from AD 70 until some time in the 3rd century. In the period before the first temple was built this spot may have been a place of cultic importance already in pre-Roman times. (Eist pre-temple and temple) All the sites were excavated by staff of the Rijksdienst voor het Oudheidkundig Bodemonderzoek (ROB) in Amersfoort. Only the excavations of the casta of Nijmegen in the years 1951 and 1957-1967 were carried out in cooperation with the Rijksmuseum van Oudheden in Leiden. 17 2 Methods 2.1 THE EXCAVATION METHODS 2.1.1 THE NATURE OF THE EXCAVATIONS Almost all of the complexes discussed here were excavated because excavation at a later date would not have been possible due to the construction of houses and roads, the installation of sewers, etc. On account of this imminent development, the excavations had to be carried out within a limited period of time, and this time-limit largely determined the method of excavation that was chosen. Under these conditions it was impossible to carry out excavations with scrupulous care: soil-traces and material finds could not be given one hundred percent attention. These rescue-excavations thus concentrated on providing as full a picture as possible of the archeological features, while the material finds were only collected in part at a few excavations. The collecting was done by hand. From 1975 on sieving was also carried out. The 4th-century defensive ditch on the Kelfkensbos is one site where sieve-samples (i.e. samples obtained by sieving) were also taken incidentally. The excavation technique that was used can be broadly described as follows: The layer above the Roman cultural level and the soil that has been disturbed by post-Roman activities are removed by means of a mechanical digger. Then, if time permits, the Roman cultural level is dug through at different levels 10-25 cm apart, to facilitate the collection of finds by hand; subsequently the mechanical digger is used to get down to the next level. The surface of each layer is tidied up using a spade, and then the archeological features are recorded on drawings. In all cases the finds visible in the surface are collected. The archeological features are dug through entirely or partly and any additional finds thus coming to light are collected too. Then the mechanical digger is used again: 10-25 cm of soil is dug away and the following surface is investigated. In some exceptional cases this working method was not followed. For example, during the older excavations of the temples in Eist and the oldest excavations of the castra in Nijmegen all digging activities were carried out using a spade. Similarly, the graves of the 'Margriet' cemetery were very carefully exposed with the aid of a trowel and a brush. 2.1.2 ARCHEOZOOLOGICAL CONSEQUENCES The excavation method outlined above has various consequences. The first consequence of excavating at artificial levels and collecting whatever is visible in the surface is that large fragments are more likely to intersect a surface, or to be almost in contact with it. Theoretically, in digging down 10 cm, when the surface is inspected meticulously, a spherical object with a diameter of 10 cm has a 100% 18 chance of being noticed and collected, while a similar object with a diameter of 5 cm only has a 50% chance. In the excavations in the Eastern River Area this is not a very serious problem because most of the bone material is found in the archeological features, refuse pits, ditches, etc. that are dug through anyhow, so that also the material between two levels is reached. Moreover, large bones of the larger meat-producing animals like cattle and red deer are cut into more pieces than the smaller bones of the smaller animals. As a result of this the difference in fragment size becomes levelled out to some extent. A second effect of the mechanical excavation down to the next level is that it is especially disadvantageous for the recovery of whole skeletons. As the soft parts of a cadaver rot away, the weight of the overlying soil may break some of the ribs, for example, and consequently a skeleton is often found in a relatively thin layer. If the uppermost parts of a skeleton are not spotted in the surface, or if they are indeed spotted but not recognized as part of a complete skeleton, then by excavating with the mechanical digger it is possible that a large part of the skeleton will be lost without anyone noticing. This is precisely what happened in the case of the skeletons of horses at the villa in Druten (see 3.7.4). Collecting mainly by hand has the negative effect that here again large fragments stand a greater chance of being collected than small ones, because a large fragment is more conspicuous. Thus it can be predicted that bones of horses, which in the Eastern River Area were not broken into pieces, will be more readily collected than those of cattle, and bones of cattle more readily than those of sheep and pigs, for example. To put it briefly, the numbers and weights of the bones of the various animal species are not representative for the occurrence of these animal species as faunal remains in the soil. In this way very small animals like birds and fish will be severely under-represented (Clason and Prummel 1977). Only by extensive sieving is it possible to arrive at a more balanced picture. The above-described negative effect of collecting by hand is aggravated by the pressure of the time-limit imposed on the excavations under discussion here. The positive selective advantage conferred on larger fragments increases with the volume of soil examined per unit of time. In the excavation of the 4th-century defensive ditch in Nijmegen (Nijmegen IV), for example, the tempo was particularly high: in a space of 43 working days a team of 5-10 people together with a mechanical digger investigated ditch-fill measuring about 1,600,000 m^ in volume. The collection of finds could not have been done optimally. Another point is that for the different excavations the time-limit imposed varied considerably, due to all kinds of external circumstances. In comparing the various complexes, this means that the degree of selectiveness of the data obtained varies from one complex to another. The problem with interpreting archeozoological data is not so much that these have been obtained by collecting selectively, but that one does not know the degree of selectiveness. This problem can be avoided if samples are taken not only by means of hand-collecting but also by carrying out a thorough, well-planned program of sieving. With sieving it is possible to have two objectives: a qualitative and a quantitative one. The qualitative objective, wanting to know which species occurred, is the least exacting. After a few samples have been sieved the species list grows quickly, especially as far as birds, fish and small mammals are concerned. Table I gives an overview of the increase in the number of species in early medieval Dorestad (Prummel 1983) when in addition to collecting by hand also sieve-samples are taken (mesh width 10 and 4.5 mm). For a qualitative insight into the occurrence of domesticated and wild mammals, domesticated birds and larger molluscs it is possible to rely on the data of hand-collected material. The number of wild mammal species increases with the use of the 4.5-mm sieve. The 19 TABLE I Dorestad: numbers of species found by means of collecting by hand and by means of combining hand-collecting and sieving (source: Prummel 1983). hand domestic mammals domestic birds wild mammals wild birds fish moUusca handele mm sieve hand+iomm +4.5 mm sieve 7 7 7 2 2 2 8 8 10 I 12 22 13 25 14 12 20 12 extra species that were collected in this way are all small rodents. For wild birds and especially for fish, sieving is indispensable. Using both the lo-mm sieve and the 4.5-mm sieve the number of species increases considerably. The strategy that can be applied to arrive quickly at as complete a qualitative picture as possible is as follows. Sieve-samples are taken from the different types of archeological features of the settlement to be excavated. Subsequently the archeozoologist analyzes series of samples originating from the different types of archeological features. This analysis of series is repeated until no new species are found. This procedure is the very minimum standard that should be met for an excavation if one is to arrive at well-founded conclusions. In this way insight is gained into the natural environment, and information is acquired about hunting, fishing and dietary habits. The quantitative proportions of the various sources of food remain unknown, however. Unfortunately, the complexes investigated in this publication do not come up to this minimum standard. Of the complexes that have been studied, only during the excavation of the 4th-century defensive ditch in Nijmegen (Nijmegen IV) were a few sieve-samples taken. This means that in my research I had to concentrate on the larger domesticated and wild mammals (sheep, goat, pig and larger animals). The information on birds, fish and other small animals is extremely scanty. In order to gain some idea of the role played by these groups of small animals, I examined sieve-samples not only from the 4th-century ditch but also from excavations carried out in 1982, of which the hand-collected material does not fall within the scope of this study. The second, quantitative objective involved in sieving, to form a picture offer example the importance of stock-breeding, fishing and fowling for supplying meat, is more exacting. The strategy to be applied will cost the excavators as well as the archeozoologist much more time and effort than if attention were devoted exclusively to qualitative data. It will be necessary to take an extensive series of random samples from the object to be excavated. The more heterogeneous the separate sieve-samples are, the greater the number of sieve-samples should be. Seeing that there is often not enough time to apply such a sieving strategy, due to various external causes, and even if time is available it often cannot be optimally made use of for economic reasons, a less time-consuming strategy will usually have to be settled for. This strategy is based primarily not on the question 'What is present in the soil?' but rather on the question 'What is the excavator leaving behind in the soil?'. If this last factor is known then for material excavated manually it is possible to make a correction for this loss. To get a good idea of what the excavator leaves in the soil it is necessary to directly combine the information provided by the sieve-samples and by the hand-collected material. This can be achieved by collecting bones from for example a number of refuse pits and parts of ditches with the aid of the spade, the eye and the hand according to the excavation method followed and with the usual 20 attention and speed. Subsequently the earth from the same refuse pits is sieved to collect the remaining bone material. Then it is possible to directly compare the hand-collected material with the total amount of bone material present in the pits, namely the combination of the hand-collected and the bones obtained by sieving. For those species that occur both in the hand-collected and the sieved material, it can now be estimated what percentage of the bone material of a certain species present in the soil can be retrieved by collecting by hand. On the basis of these data it will then be possible to make a correction to account for the material collected by hand elsewhere on the site. The correction factors obtained will also be applicable for other excavations where material is collected by the same team working at the same speed and using the same method, provided that the fragmentation per animal species does not differ in the different excavations. The advantage of directly combining the techniques of hand-collecting and sieving lies in the fact that in using both techniques exactly the same group of bones present in the soil is investigated completely. If these two techniques were to be carried out independently, then to obtain the same information more sieve-samples would have to be taken, since the representativeness of both kinds of sampling would be involved. Both must be representative for the bones present in the soil in their entirety, while by combining the techniques the data provided by both the hand-collected material and the sieved material are a priori representative of the population of bones under investigation. The working method outlined above can be applied notably when one wants to acquire quantitative data on domesticated farm animals and large wild mammals. For species that occur only very sporadically or not at all in the hand-collected material, the only possible way of acquiring such quantitative data is to carry out a random sampling procedure on a broad scale. In the case of the complexes investigated in this publication, sieving was not carried out in order to obtain quantitative data. Wherever percentage ratios between animal species are mentioned, these are necessarily based on the data provided by the hand-collected material, without any corrections being made for the selectiveness involved in collecting. Seeing that this study is mainly concerned with the comparison of find complexes this absence of any kind of correction factor is somewhat less dramatic, because the topics discussed are mainly about similarities and differences in species composition, and not so much about the absolute proportions that the different animal species had, for example, in the local human diet. In this connection it was indeed taken as a basic assumption that the degree of non-representativeness of the material from the various sites is more or less equal. It is impossible to say whether this assumption is justifiable. 2.1.3 COLLECTING BY HAND AND SIEVING: AN EXPERIMENT As mentioned previously, in the case of the bone complexes that form the subject of this study data obtained by sieving are absent almost everywhere. Nevertheless, in order to gain some kind of insight into what has been lost as a result of hand-collecting, in Nijmegen a hand-collecting/sieving experiment has been carried out as outlined in the above section. The material comes from five refuse pits dating from the early ist century on the terrain of the Sint Jozefschool in Nijmegen (find nos.; Nijmegen 218/18, 21, 27, 62, 66) (fig. 7). These pits belong to the early Roman occupation of the Valkhof and its surroundings, of which the hand-collected bone finds obtained during previous excavations are discussed in section 3.1.3 (Nijmegen Ib-c). The pits were dug through by one of the members of the excavation team, with the bone material being collected by 21 TABLE 2 Hand-collecting/sieving experiment. The hand-collected and the total hand-collected and sieved material from five pits in Nijmegen. Frequencies and weights (g). Mesh width: 5 mm. Nijmegen 218/ i8 cattle sheep or goat pig horse cattle-horse size sheep-pig size pig-cattle size small rodent small rodent size domestic fowl mallard woodcock Nijmegen 218/ 21 hand hand+ hand n weight sieve n weight n weight 4 - 49-0 - - 4 I - Nijmegen 218/ 27 hand-|sieve n weight 49.0 24.4 - 32 3 6 - 513.6 16.0 30.5 - 43 hand n weight hand-(sieve n weight 17 - 659.1 45.8 126.8 - 7 3 - 269.1 22.4 - 18 4 5 - 434-3 7-1 43.8 - 12 I 5.2 6 22.5 17 106.4 6 42.6 6 I 4 5-3 6.7 I I 9-7 1-4 53 5-3 65.2 5 9 5-5 11.3 18 30 24 116.2 - - - - - - - - - - - - - - - - I o.i - - - - — - _ - _ - _ - _ - _ - _ — _ - 2 2.2 - _ - - - I 0-3 - I 0.8 4 2.3 10 4-7 _ - _ - 30.5 21.8 bird unidentified - - - - amphibian unidentified - - - - - - - - - - - pike tench allis shad or twaite shad salmon Cyprinidae - - - - - - — — — — — _ _ _ _ _ I 0.9 - - - - - - I 0.5 - - - - I 0.1 I 0.1 mussel - - - - - I 0.6 - - - - total identified total unidentified Total 4 49-0 49-0 25 35 291.5 61.8 31 83 114 488.6 34-5 594.6 832.3 73 177-8 73 146 lOIO.I 10 II 4 73-4 16.8 90.2 560.1 0.0 5 9 14 41 0 fish unidentified - 52 353-3 173-3 661.9 * % of total * * % of identified material hand in the usual way. Subsequently all the soil from the pits was sieved using a sieve with a mesh width of 5 mm. The soil from the pits was dark but not very sticky, so hand-collecting and sieving were relatively easy. It should be noted that the digging and the hand-collecting were done with scrupulous care, and that a considerable amount of time was devoted to this, in comparison with the usual practice in similar excavations where this is generally not possible. In such an incidental case as this it is inevitable that whoever is doing the hand-collecting sees the sieving experiment to some extent as a means of checking his own personal capability and expertise. The results of the experiment are presented in table 2 (fig. 3). As a result of sieving the qualitative data are increased considerably, even with such a small number of samples. In the hand-collected material only three species are present, 22 hand n weight 12 679.2 _ — I 6.2 tiand+ sieve n weight I 2 3-4 16 31 iand + sieve n weight 25 3 6 882.2 3 3 827.4 32.5 113.0 - - - I 2-3 25-5 17.1 23.8 3 4 18.8 3.0 70.5 - - 15 28 38 20 25.5 - n weight 29 1031-3 II.2 6 33-6 3 2 hand 32.5 151.8 19.2 29.1 hand+ hand hand+ sieve hand sieve n %* %** 75 53.6 79.8 6 4-3 6.4 13 9-3 13.8 - - - 17 12.1 II 7-9 13 9-3 - 5 0.3 - - - - - - - I 0.1 - - - - - - _ _ Total weights Total numbers Nijmegen 218/66 Nijmegen 218/62 n 0/ * /o %** 119 22.8 56.4 25 4.8 11.9 32 6.1 15.2 0.2 0.5 I 53 10.2 81 15.6 147 28.2 - 2 0.4 4 0.8 I 0.2 I 0.2 2.4 0/ * n/ ** /o /o 2338.3 86.4 91.4 48.5 1.8 1-9 172.1 6.4 6.7 - - 109.4 17.2 18.0 4.0 0.6 0.7 - - - - - - - - 0/ * 0/ ** /o /o 3055-9 74.4 86.1 96.6 2.4 2.7 380.4 9-3 10.7 2.3 0.1 0.1 3238 81.8 141-3 7-9 0.3 0.0 0.2 0.0 4.0 0.1 0.1 0.9 0.3 0.0 0.0 0.0 0.0 2.0 3-4 0.0 2 1.8 - - I 0.9 - — — - - — - - - - - 7 1-3 - - I 1.6 4 2.9 19 3-7 2.3 0.1 8-4 0.2 - I 0.1 - - - - - - I 0.2 - - 0.1 0.0 14 3-7 _ - _ - 2.7 0.8 - 14 I I 0.2 0.2 0.0 0.0 0.2 5 1.0 3-7 0.8 0.5 0.9 1-5 0.1 0.0 I - 0.1 0.0 I - 0.0 0.0 0.0 0.0 7 1-3 0.1 0.0 3-9 0.1 I 0.2 - - 0.6 0.0 5 1-5 _ — - - 20 2.8 - - 5 0.5 - - - - - - - 64 1083.4 70.7 78 142 1154.1 26 972.9 7 33 21.8 13 3 16 _ _ _ — 685.4 28.9 714.3 — _ _ — — 994-7 I 0-5 - - I 0.7 - - 38 1071.0 120.9 87 125 1191-9 94 67.1 46 32.9 140 - - 40.5 310 59-5 521 211 1-9 0.5 0.5 6.6 0.5 0-5 0.5 2.4 0.5 2558.9 94.6 147.0 5-4 2705.9 — weight _ - _ _ _ _ - - 5 I.O weight - - 0.0 3548.7 86.4 559-5 13.6 4108.2 while in the total material twelve different species are represented. Seeing that four of the five pits have been shown, by sieving, to contain species that do not occur in the other pits, it is very likely that by sieving the filling of a few more pits the number of species will be increased even further. The expected increase in the number of species following the investigation of more pits using the spade and hand-collecting is considerably lower. The previously excavated part of the settlement, of which the hand-collected material is discussed in section 3.1.3, yielded a total number of only seven species (see table 9). The extent to which the quantitative data are influenced by sieving can be deduced from the percentage data shown in table 2. Comparing the number of fragments of cattle bones as collected by hand and the total number of bones actually present, inasfar as they were bigger than 5 mm, it is evident that these 23 10cm I Fig. 3 Collecting by hand and by means of sieving, p. 24: hand-collected material from refuse pit Nijmegen 218/62; p. 25: extra material obtained by sieving the residue of the pit with a 5-mm sieve (see also table 9). 24 values differ by as much as 30.8 %. Since cattle bones are easy to collect by hand, this species is extremely over-represented. On the other hand, small animal species are extremely under-represented. The fish, for example, form only 0.7 % of the hand-collected material, while after sieving the percentage of fish turns out to be 5.6%. Even if we disregard the unidentifiable fragments, that strongly influence the percentages notably with the sieved material, the maximal positive difference, that for cattle, is still 23.4%. The differences between hand-collected bone material and the total amount of bone material really present in terms of percentages by weight are smaller but still considerable. The greatest difference can again be observed with the bones of cattle, over-represented in the hand-collected material. This difference by weight amounts to 12%, in the case of the percentages of the identified bones. The differences in terms of weight percentages of the small animal species appear to be very slight, however. The weight percentage for fish, for example, is increased after sieving from 0.0 "0 to 0.2 "„. The effectiveness of collecting by hand during the experiment is shown in table 3. The effectiveness is expressed as the percentage of hand-collected material with respect to the total amount of hand-collected and sieved material. These percentages have been calculated both for individual species and for groups. The percentages have only been calculated when of any one species or group at least ID fragments were present in the total amount of hand-collected and sieved material. '^ In»«»» ^n n I c ^h 10cm TABLE 3 EfRctiveness of hand-collecting for the various species and groups. The effectiveness is expressed in the percentage of hand-collected material with respect to the total of the hand-collected and the sieved material. effectiveness (%) cattle sheep or goat pig with respect to numbers with respect to weights 63 24 77 41 45 53 50 total cattle-horse size total sheep-pig size total mammals 22 29 72 43 66 total birds 16 20 3 I 45 72 26 66 total fish total identified total unidentified total 15 27 The effectiveness is greatest for cattle. Of the identifiable fragments present 63 %, or 77 % by weight, were noticed and picked up during the hand-collecting. For 25 the other domesticated farm animals, sheep/goat and pig, this proportion is considerably less: 24 %, or 50 % by weight, and 41 %, or 45 % by weight, respectively. The effectiveness was low for birds (16%, or 20% by weight) and very low for fish (3 %, or i % by weight). The above-mentioned percentages can only be regarded as an indication of the effectiveness of the method of collection during the experiment. The number of bones available for calculating the percentage per species or per group is in most cases too small to provide any hard data. If only to justify the use of the concept of 'percentage', the number of fragments of a species for which the effectiveness is calculated has to be at least 100. Percentages calculated for 25 fragments, as in the case of sheep/goat and birds, are not representative. Nevertheless the relation between fragment size and effectiveness is evident. There is a distinct ranking order, from large to small, in the series 'total cattle-horse size', 'total sheep-pig size', birds and fish. The effectiveness of the hand-collecting for this group is 53 %, 22 %, 16 % and 3 %, respectively; in terms of weight percentages: 72 %, 43%,20%andi%. Of the total number of bones from the five pits 27 % (66 % by weight) were collected by hand. The extent to which the effectiveness of hand-collecting can vary is evident if we compare these data with data from early medieval Dorestad (Prummel 1983: table 16). For the four samples from Dorestad, each consisting of more than 600 fragments, the effectiveness of the total lay between 1.2 and 1.8% (by weight, between 11.2 and 30.7%). This is very much lower than in Nijmegen, certainly if one realizes that a lo-mm sieve was used in Dorestad. One of the factors that would have played a role here is the great deal of attention paid to hand-collecting in Nijmegen, in contrast to the collection of bones in Dorestad (personal communication W. Prummel). This illustrates the fact that the effectiveness data are specific for a particular excavation. Thus the data can only be used for those excavations where the conditions, the excavation technique, the amount of attention paid, the nature of the soil and the size of fragment per species are the same as at the place where the hand-collecting/sieving experiment was carried out. For the bone complexes from the Eastern River Area that will be discussed in the following chapters the above-mentioned data therefore carmot be used. Table 4 gives an example of the use of data concerning effectiveness based on hypothetical data from an imaginary excavation where the circumstances are the same as with the hand-collecting/sieving experiment. The frequencies obtained, with the accompanying frequency percentages, are shown in the first two columns. By multiplying the frequencies obtained by the reciprocal of the effectiveness value one arrives at the actual number of fragments found. The last column gives the frequency distribution of the real values. For weights the same method can be applied. TABLE 4 Frequencies of the hand-collected material from a hypothetical excavation involved in the hand-collecting/sieving experiment and the 'real' frequencies as estimated with the aid of the effectiveness data. 26 hand-collected 0/ n /o cattle sheep or goat pig birds fish unidentified effectiveness 0/ /o 1000 50 400 300 150 50 20 63 24 15 8 41 16 3 5 3 15 100 'real' values n % 1587 1667 732 938 1667 667 22 23 10 13 23 9 Fig. 4 Skeleton of a cow stylopodium humérus femur Pelvis Scapula radius tibia Cranium zygopodium ulna fibula carpus tarsus autopodium metacarpus metatarsus phalanges } Metatarsus Carpus Metacarpus I Phalanges Phalanges 2.2 THE ARCHEOZOOLOGICAL METHODS 2.2.1 RECORDING AND PROCESSING OF THE DATA For every bone or bone fragment the following data were recorded as far as this was possible: species; skeletal element; part of the skeleton; side (left or right); sex; special features such as the presence of traces of gnawing, of burning, of chopping, cutting or sawing, and the occurrence of pathological conditions; weight; rough estimate of the age of the animal at the time of slaughter; age data based on the dentition and the state of fusion of the epiphyses; measurements; find numbers and dating. All these data were computerized. A slightly modified version of the numerical code 'Knocod', as developed by Uerpmann (1978), was used for this purpose. Uerpmann's system was extended with a code for recording more precisely the presence of traces of chopping, cutting and sowing (see section 2.2.9). The data were processed with the aid of the computer of the Computing Centre of the University of Groningen (RUG). The (statistical) programming apparatus available there was used for most data. For the processing of the metrical data a somewhat modified version of a programme developed by Uerpmarm was used (for details of this programme see Frummel 1983: 53, 54). An overview of the various skeletal elements and their nomenclature is shown in fig. 4. 2.2.2 THE SPECIES DETERMINATION 2.2.2.1 General points The bone fragments found were identified as far as possible to species level. This was done with the aid of the comparative collection of recent mammals, birds, amphibians, fish and molluscs of the Biologisch-Archaeologisch Instituut of the University of Groningen (BAI). In addition use was made of the literature in the field of identification (i.a. Schmid 1972). 27 As a general rule complete bones can be easily identified. However, most of the bones from archeological sites are extremely fragmented as a result of such activities a slaughtering, the cutting up of bone with meat into portions and the smashing of bones for the purpose of extracting the marrow. Especially in the case of closely related species like sheep and goat it is not possible to identify the exact species, and the bones concerned will have to be classified as 'sheep or goat'. Also problems arise when one tries to distinguish between wild and domesticated forms of one species, e.g. between aurochs and domesticated cattle, and between wild boar and pig. Since these are different forms within one single species there are usually no morphological differences, and one has to make use of differences in size. The identification of aurochs and domesticated cattle, of wild boar and pig, and of sheep and goat will be discussed in more detail later on. In the case of small fragments without any distinct morphological characteristics it is impossible to make an identification to species level. This unidentifiable part of the material is divided into major categories such as 'birds, unidentifiable' and 'fish, unidentifiable'. The unidentifiable mammal bones are divided into size classes, as proposed by Uerpmann (1978). The classes used are: 'no size assigment', 'rabbit size to medium size dog', 'medium size dog to wild boar size' and 'European red deer size to domestic cattle size'. In this publication, for the last three classes abbreviated terms are used: 'rodent size', 'sheep-pig size' and 'cattle-horse size'. 2.2.2.2 Aurochs and domestic cattle The problem with distinguishing between aurochs. Bos primigenius, and domestic cattle. Bos taurus, is that both of these are representatives of the same species. Domestic cattle are the domesticated form of the aurochs. Consequently there are no distinct morphological differences. The only features of a bone that would suggest that it came from an aurochs are a greater compactness and a more pronounced profile at the points of attachment of the muscles (Boessneck et al. 1971:97). The most important criterion for ascribing a bone to domestic cattle or aurochs is the size of the bone. Various authors have established criteria for sizes and indices for the purpose of distinguishing between the wild and domesticated form (i.a. Requate 1957; Bökönyi 1962, 1972; Stampfli 1963; Bohlken 1964J Degerbol and Fredskild 1970; Von den Driesch and Boessneck 1976). The biggest problem that arises here is that young aurochs bulls and adult aurochs cows are often difficult to distinguish metrically from large domestic cattle, notably the adult bulls. Thus Stampfli (1963: 161) notes that the lower range of breadth variation for the aurochs measurements can overlap with the measurements for domesticated oxen and bulls. As a result of this overlap different authors place the caesura in the measurements for aurochs and domestic cattle in a different position, which can have a considerable influence on the ratio between the wild and domesticated animals within a settlement. Clason (1972) has pointed out the implicit dangers when one uses these data without giving due consideration to the criteria used. Another problem involved in distinguishing between domestic cattle and aurochs is that the animals of aurochs populations may be larger or smaller in size in different biotopes (Bökönyi 1962; Von den Driesch and Boessneck 1976), while at the same time aurochs and domestic cattle are able to interbreed. An extra point that makes it difficult to distinguish between the two species for animals in the Roman period is the development of the size range of aurochs and domestic cattle. In the course of this period the aurochs show a decrease in size (Degerbol and Fredskild 1970; Von den Driesch and Boessneck 1976), while notably in Roman times domestic cattle can be very large (Boessneck et al. 1971). 28 Consequently the size ranges of the two groups of animals are closer to each other in the Roman period than in earlier times. Three bone fragments could immediately be identified as aurochs. These are two skull fragments, each with part of the horn-core, and a separate point of a horn-core, that come from 4th-century Nijmegen (find number: Nijmegen 182/6). The circumference at the base of the two horn-cores (306 and 317 mm) fall well within the size range for aurochs bulls (Degerbel and Fredskild 1970). This still applies if we take the much higher values given by Bökönyi (1962) as a criterion for aurochs bulls. The measurements for the other cattle in the Eastern River Area are considerably smaller (89-127 mm, n = 4o). The separate point, that probably belongs to one of the two skull fragments, is also ascribed to aurochs on account of its size. To find out whether there are any other remains of aurochs in the material apart from the horn-cores mentioned, all the measurements for cattle have been compared with those for aurochs as given by Degerbol and Fredskild (1970). In this way it was possible to compare 653 measurements of 13 different kinds of bones from 4th-century Nijmegen. Of these 653 measurements, 175 fall within the range of the aurochs cows and 12 within the range of the bulls. The fact that 175 measurements fall within the range of the aurochs cows is not very meaningful, since almost all measurements of domestic bulls fall within a range that overlaps with that of the aurochs cows. Exceptions to this rule are the measurements of the length of the metapodials and of the cheektooth row of the mandibula. Although for any one particular species there are in principle no absolute limits for the measurements of different forms of that species, the measurements of metapodials of domestic cattle and aurochs given in the literature show no overlap. The length of the cheektooth row shows only a slight overlap between the measurements of aurochs cows and domestic bulls. Of the 34 metapodials and the 13 mandibulae from 4th-century Nijmegen that permit these measurements to be taken, none of the measurements falls within the range for aurochs. In the following, all the appropriate measurements of bones of cattle from the Eastern River Area will be tested according to the measurement criteria of Degerbol and Fredskild (1970). These measurement criteria are not absolute, however. Because some of the material studied by Degerbel and Fredskild dates from after 3000 BC, their classification too is partly based on interpretation. On the basis of the length of the cheektooth row and the length of the metapodials, it should be possible to distinguish between bulls and cows of domestic cattle and aurochs. On the basis of other measurements, only the aurochs bulls can be identified with certainty. The length of the premolar row of seven mandibulae, varying from 57.9 to 59.7 mm, fall within the range for aurochs bulls (Nijmegen 183/13 (3 x), 29/32, 34/36; Heteren 1969/37, 70; Druten 65/4). The measurements given by Degerbol and Fredskild for the length of the premolar row of the mandibula show a complete overlap for aurochs cows and bulls, and a large overlap between aurochs and domestic cattle. This measurement is thus a poor criterion for distinguishing between aurochs and domestic cattle. The largest measurement (59.7 mm) is that of a mandibula of which also the length of the cheektooth row (144.6 mm) could be measured. As stated above this length measurement is one of the few measurements that permit a reasonably good distinction to be made between aurochs and domestic cattle. As this mandibula clearly comes from a domesticated animal, I consider the seven mandibulae mentioned all to be of domestic cattle. It should also be noted that of the 45 mandibulae from the Eastern River Area for which the length of the 29 cheektooth row could be measured, none fell within the size range for aurochs cows or bulls. In the case of a scapula from Druten (19/4), all of the four measurements that could be used for distinguishing between the two species fell within the size range for aurochs bulls. A scapula from 4th-century Nijmegen (182/6) with a minimum length of the collum of 74.6 mm just falls within the lower limit of the measurements for aurochs bulls. Since the three other measurements of this scapula fall outside the range of measurements for aurochs bulls, I consider this scapula to be one of a domestic animal. For five metacarpal bones the thickness of the proximal end, varying from 47.0 to 48.6 mm, falls within the range of the measurements for aurochs bulls (Nijmegen 183/13, 14, 185/12; Druten 19/4 (2x)). In the case of two of these bones, with proximal thicknesses of 47.3 and 48.6 mm, it was also possible to measure the length. These length measurements, 233.6 and 234.6 mm respectively, fall within the range for aurochs cows. As stated above the length of the metacarpus is one of the few measurements that can be used for distinguishing between aurochs cows and domestic cattle. Consequently I consider all five metacarpal bones as coming from aurochs. A metacarpus for which only measurements of the distal end are available (Nijmegen 106/21) I also consider to be of aurochs, since these measurements correspond to those of the two metacarpals that are ascribed to aurochs on the basis of their length measurements. A pelvis fragment from Druten (63/7) I consider to be of aurochs because the measurements are much larger than those of the rest of the material. The height and breadth of the collum are 48.8 and 33.0 mm respectively. It is impossible to compare these measurements with data in the literature since no such data for aurochs are available in the literature. An astragalus from Heteren (1969-70/37) with a maximum lateral length of 87.0 mm and a maximum distal breadth of 57.2 mm falls within the range for aurochs bulls. For four calcanei the maximum breadth falls within the range of the aurochs bulls (Nijmegen 183/15; Druten 5/4, 42/2; Meinerswijk 14/8). In view of the fact that the total length measurements for these bones even fall outside the range for aurochs cows, I consider these calcanei to be of domestic cattle. Three metatarsal bones have a measurement that is indicative of aurochs: the minimum breadth of the diaphysis, 37.0 and 38.9 mm, of two bones from Nijmegen (183/13, 15) and the thickness of the proximal end, 60.0 mm, of a bone from Meinerswijk (14/55). For these bones, however, the measured or estimated lengths fall well outside the range of the sizes of aurochs cows, so I consider these bones too to be of domestic cattle. 2.2.2.s Wild hoar and pig With mature animals it is usually possible to distinguish reasonably well between wild boar. Sus scrofa, and domesticated pig. Sus domesticus, on the basis of metrical data (Becker 1980: 47; Teichert 1969: 239). In addition, the presence of more pronounced articular surfaces and points of attachment for muscles indicates that a suid bone can be ascribed to wild boar. Problems arise, however, with bone material of immature animals and with small bone fragments on which the above-mentioned features cannot be observed. Moreover, in favourable nutritional circumstances the size of domestic pigs rapidly increases, as a result of which the measurements may fall within the size range for wild boar (Teichert 1969: 240). Another point that must be taken into account is the possibility of the occurrence of crossbreds in the archeological material. In this investigation the measurements of the bones of Sus are compared as far as 30 possible with the measurements given by various authors for pre- and protohistoric wild boar and domestic pigs (Boessneck et al. 1963: 60-68; Boessneck et al. 1971: table 158, 179; Luhmann 1965: table 34; Becker 1980: table 30, 33; Kratochvil 1981). Only ten suid bones from the Eastern River Area have been identified as wild boar. Four fragments of the upper jaw (two matching pairs) are of wild boar (Nijmegen 183/13, 14). The length of the molar row varies from 78.0 to 83.1 mm. Since few measurements of wild boar are available in the literature, the maximum measurements mentioned for domestic pig will have to serve for the purposes of comparison: Haithabu 66.4 mm (Becker, 1980), Manching 71.0 mm (Boessneck et al. 1971), Magadalensberg 67.5 mm (Luhmann 1965), Mikulcici 71.0 mm (Kratochvil 1981). A very large canine from the lower jaw (Nijmegen 182/6) and a canine from the upper jaw (Nijmegen 183/13) undoubtedly come from a male wild boar. Also a scapula with a minimum length of the collum of 30.1 mm I consider to be of wild boar (Druten 1/33). The following maximum values for domesticated pig may be mentioned: 21.5 mm (Boessneck 1963), 25.5 mm (Kratochvil 1981), 26.6 mm (Becker 1980), 28.5 mm (Boessneck et al. 1971). The minimum values for wild boar are 26 mm, 27.8 mm, 28.7 mm and 30.3 mm, repectively (Boessneck et al. 1963; Becker 1980; Luhmann 1965; Becker 1980). Three tibiae with a maximum breadth of the distal end of 36.1 mm, 39.5 mm and 37.2 mm are clearly of wild boar (Nijmegen 183/13 (2x); Druten 19/4). The minimum values for wild boar are 32.9 mm, 33 mm, 33.9 mm, 36.3 mm respectively (Becker 1980; Boessneck et al. 1963; Becker 1980; Luhmann 1965). 2.2.2.4 Sheep and goat With many bones it is difficult, or even impossible, to distinguish between sheep, Ovis aries, and goat, Capra Mr eus. In 1964 Boessneck, Müller and Teichert published an extensive study on osteological differences between domesticated sheep and goats based on material from a comparative collection. In 1969 Boessneck published a summary of the most important differences. For the archeological study of the Eastern River Area I have taken as a basis the criteria given by Boessneck in his summary (1969), since many of the other differences described by Boessneck et al. (1964) were found to be insufficiently distinct or even misleading when checked against the comparative collection of the BAI. An explanation for this phenomenon could be that the extra differences between sheep and goat that are described by Boessneck et al. (1964) are not actual differences between the species sheep and goat, but between the breeds of sheep and goat that the authors studied. It is also possible that my own ability to distinguish between these bones is somewhat limited. For the identification of the species on the basis of the horn-cores the data of Schmid (1972: 90-91) and Boessneck et al. (1964: 21-23) have been used. In these publications, however, it is pointed out that with different breeds of sheep and goat, and with young animals and females in general, the criteria described often do not permit a definitive species identification. 2.2.3 THE QUANTITATIVE ANALYSIS In archeozoology three groups of methods are used for the quantitative analysis of bone material recovered from the soil. These are the number of faunal remains methods (NR), the minimum numbers of individuals methods (MNI), and the weight method. A fourth group, consisting of probabilistic methods, is now being developed (Lie 1980; Wild and Nichol 1983). I have made use of the NR method and the weight method. 31 The widely used MNI methods involve the determination of the minimum number of individuals of a species that is represented by the bone material found. Subsequently the MNI values of the different species are compared to determine the relative importance of those species. This method reduces deviations that arise when the bones of different species are fragmented in a different way or when the skeletons of the different species consist of a different number of skeletal elements. Proportional results indicate, however, that the corrections are negligible or non-existent (Gautier 1984). A disadvantage of the method is that animal species of which only a few bone fragments are found become greatly over-represented (Payne 1972; Uerpmann 1973). As a result, the MNI values of species with a widely different number of identified bones are not comparable (Grayson 1978). In addition, the MNI is largely dependent on the size of the archeological units such as layers, trenches or groups of pits, within which the MNI values are calculated (Grayson 1973). Another problem is that the MNI methods are not uniform (see e.g. Kranz 1968; Bökönyi 1970; Chaplin 1971; Perkins 1973; Gasteel 1977). The different MNI methods give different results (Bobrowski 1981). In publications it is often not indicated which method has been used, and consequently it is not possible to compare data from different publications. If the MNI method used is indeed indicated, it is nevertheless usually impossible to compare the bone complexes for which different MNI methods were used. In comparing such complexes the data will have to be reassembled to permit the application of a uniform MNI method. For this purpose it is necessary to have access to the primary data or even to the bone material itself, which in practice is almost possible. Another objection to the MNI methods is that different samples from a population give widely varying MNI data. This is in contrast to the NR methods, with which these data are fairly constant (Gilbert et al. 1981). In view of all these drawbacks it is evident that the MNI methods are less suitable for the analysis of bone material, certainly if different bone complexes of varying size and composition are to be compared. Since such comparisons form a part of this study, the MNI method has not been used here. With the number of faunal remains method (NR), the basic data consist of the number of identified bone fragments per species. The disadvantage of this method is that species for which the bones are more fragmented than those of other species are over-represented. Also, in comparing bone complexes it is possible for the material of one species to be more fragmented in one complex than in another, and consequently the comparison of the bone frequencies from different complexes becomes dubious (Lasota-Moskalewska and Sulgostowska 1976-1977). These differences in fragmentation can arise as a result of different butchering methods and differences in the resistance of bones to the detrimental action of soil. A second disadvantage is that the skeletons of different species do not consist of exactly the same number of bones (Poplin 1976). For this reason Grant (1975; 1976) applied the 'epiphysis only' method that provides a correction for differences in fragmentation and differences in the numbers of skeletal elements with different species. As it turned out, however, the results of the 'epiphysis only' method and the NR method largely corresponded in a relative sense. In view of the fact that in this study of bone complexes in the Eastern River Area the bone complexes are only compared in a relative sense, on account of the way in which the bones were collected, it is the NR method that has been used. In applying this method, fragments that clearly come from a skeletal element, such as loose teeth that fit into a jawbone, are counted together with that element as one fragment. In addition to the NR method the weight method, as described by Uerpmann (1973) has been used. This method involved the comparison of the total weights 32 of bone of the different species. The reasoning behind this method, that is still a matter of discussion, is that with all mammals the ratio of bone weight to flesh weight is more or less constant. By comparing the bone weights of different species it is therefore possible to gain some insight into the relative importance of these species for the production of meat. An advantage of this method is that the degree of fragmentation of the bone has little influence on the results. The use of the weight method together with the NR method makes it possible to assess the extent to which the above mentioned fragmentation problems play a role in the use of the NR method. Clason (1972) and Uerpmann (1973) have pointed out the importance of including also the unidentified bone material in the quantitative analysis. The percentages of unidentified material give an idea of the representativeness of the identified material. As mentioned in section 2.2.2, the unidentified material is divided into size classes such as 'cattle-horse size' and 'sheep-pig size'. By means of allocation the unidentified material is attributed to the species within the groups (Prummel 1983). The way in which this allocation is done is illustrated with reference to the material from Nijmegen la (see tables 6, 7 and 8). The percentage division of the identified bones within a size class is ascertained. For example, for the 'cattle-horse size' group from Nijmegen la the species concerned are cattle and horse (table 7). The number of unidentified fragments from the same size range are divided according to the previously established percentages of the species of the size range concerned (this process being referred to in this publication as 'allocation'). Subsequently these numbers of bones per species, as obtained by means of allocation, are added to the numbers of identified bones per species. The data thus obtained are presented in the tables, in which the total amounts of identified and allocated unidentified material are listed (table 8). In the same way also the mass of the unidentified bone material is allocated to the identified material. 2.2.4 MEASUREMENTS Archeozoologists record measurements of their material for various reasons. In the first place measurements are necessary for distinction between wild and domesticated forms of a species (see 2.2.2). In the second place measurements are necessary for obtaining information about the body size of the animals. In this study it is interesting to find out whether changes in body size occurred during the Roman period, and if so, whether these changes can be associated with Roman influences on native stock-rearing practice (see chapter 6). In this publication the withers height is also used, this being a function of the body size (see 2.2.6). The measurements recorded are all standard measurements. The measurements that were recorded are those proposed by Uerpmann (1978) in his 'Knocod' system. The way in which the measurements were taken is according to Von den Driesch (1976). The measurements of mammals and birds are presented, on microfiche, in the tables mi2-m25 (see 'Tables mi-m35 on microfiches'). 2.2.5 SEX Only a very few skeletal elements allow direct identification of the sex of the animal concerned on the basis of the so-called primary sexual characteristics. Thus with deer, except for reindeer, antlers and the bony pedicles of the frontal 33 bone only occur in males. Some male carnivores, rodents and insectivores have a penis bone, the os penis or baculum. The canine teeth of pigs are different in males and females, and consequently also the structure of the upper and lower jaws. With horses stallions usually have canine teeth in the upper jaw, while in mares these teeth are generally absent or very small. This difference is also evident in deer. A less conspicuous difference is the presence of spurs on the tarsometatarsus of cocks. In hens these are usually absent, or if they are present they are generally smaller. Capons, i.e. castrated cocks, sometimes have spurs. A difference between the sexes that is generally present is the different shape of the pelvis. In an archeological context pelvic bones are usually found in a broken state, so this feature can only be made use of to a limited extent. With other skeletal elements the measurements of males and females may differ if there is a difference in size or weight of the whole animal between the sexes, this being a secondary sexual characteristic. The problem with this criterion for distinguishing between the sexes is the overlap that exists between the measurements of bones of animals of different sexes. The presence of castrated animals among the domesticated farm animals makes it even more difficult to distinguish between the sexes correctly. A second problem with the use of measurements as criteria for distinguishing between the sexes is than an observed difference in measurements is not necessarily caused by a difference in sex, but can also be explained by a difference in breed and by the difference between the wild and the domesticated forms of a species, as described in section 2.2.2. Thirdly, different amounts of available food can result in a variable size of adult animals, irrespective of their sex (Herre and Rohrs 1973). To get round these problems to some extent it is possible to incude more measurements in the analysis, where appropriate combined in the form of indices. The advantage of the use of indices is that it is not so much the differences in size of the bones that are expressed but rather the differences in measurement ratios, i.e. the shape of the bones. 2.2.6 WITHERS HEIGHT The estimation of withers height, i.e. the height of the animal measured from the ground to the shoulder, is based on the fact that there is relation between the length measurements of the individual skeletal elements and the withers height. Von den Driesch and Boessneck (1974) have summarized the various studies that have been carried out on this relation, and have added their own comments. The relations, except those for dogs and horses, are all expressed by means of a simple linear comparison: withers height = c x bone length, where c is a constant. In the research on the Eastern River Area the estimation of withers height of cattle, pig, sheep en goat is carried out using the values recommended by Von den Driesch and Boessneck (1974) for the constant c. For horse the class division according to Vitt (see: Von den Driesch and Boessneck 1974) is used. Seeing that within the classes one can expect a more or less linear relation between the withers height and bone length, in addition to making a class division also the individual withers heights were estimated (cf. Prummel 1983). For the estimation of the withers heights of dogs linear regression comparisons are used: withers height = a+b x bone length. The constants a and b have been worked out by Harcourt (1974). Estimation of withers height with the aid of a regression comparison is preferable to a normal linear comparison because with the regression comparison it is not assumed that a = 0. 34 TABLE 5 Reference to the data for age determination as given by Habermehl (1975) used in this publication. species cattle sheep/goat pig horse dog tooth eruption (pages) epiphyseal fusion (pages) 95- 96 122-123 150-151** 50- 53 159-161 104-105 121* 150 48 166 * data according to Duerst (1926). ** data for late-maturing breeds. With all estimated withers heights it must be taken into account that these are nothing more than indications of the real withers heights of animals in the past. Von den Driesch and Boessneck (1974) mention the following factors that can influence the relation between bone length and withers height: age, sex, nutrition and genetically determined variation in size and proportions. In addition, it must be remembered that the multiplication factors, the class divisions and the regression comparisons used are average values that have been obtained on the basis of a limited number of skeletons of a few recent breeds. An example of the relative value of estimated withers heights is given by Frummel (1984: 171-173). She showed that the metapodials and other long bones from the same population of cattle can give different values for the withers heights. This was the case with the medieval settlements of Dorestad, Hamwih and Haithabu. As an explanation she mentions the lesser reliability of the factors for the long bones on account of the small number of data on which the factors are based. Another explanation mentioned by Prummel is a difference in the ratio between the length of the autopodium and the stylopodium or zygopodium. 2.2.7 AGE In the analysis of bone material from an archeological context various methods are available for determining the age that an animal had reached when it died or was killed. Chaplin (1971) mentions the following criteria on the basis of which age can be determined: eruption and replacement of the teeth; wear of the teeth; growth structures; fusion of the epiphyses; the closure of the sutures of the cranium; development of antlers; size and shape of the bones; qualitative phenomena such as the ossification of the tendons, where these are attached to bones. Grigson (1982) gives an overview of the publications that have appeared concerning age determination in cattle. She points out that none of the methods for determining age is 100% accurate. For this study the methods used for determining age are those that are based on the degree of fusion of the epiphyses, the eruption and replacement of the teeth and the wear of the teeth. In a few cases other data will be made use of, notably those concerning the closure of the sutures of the skull, the development of antlers and the size and shape of bones. The age of domesticated animals on the basis of data on teeth and epiphyseal fusion has been established according to Habermehl (1975), who gives detailed information on these age-related features in the form of tables (see table 5). Inasfar as Habermehl distinguishes between early maturing, normally maturing 35 and late maturing breeds, the data on late maturing breeds has been used. This is because it is assumed that the domesticated farm animals in the Roman period bear the closest resemblance to the least developed breeds of the present day. For the age determination of wild animals the data of Habermehl (1961) were used. The age data on the basis of epiphyseal fusion are analyzed according to the method proposed by Chaplin (1971: 128-130). With age determination on the basis of eruption, replacement and wear of the teeth the jawbones are first of all divided into classes according to the stage of tooth eruption and replacement (direct data); e.g. Ml erupting; Ml erupted but M2 not yet through; M2 erupting. The other jawbones, from which these clear data cannot be deduced but which do have some teeth present, are added to the classes just mentioned on the basis of the wear of the teeth (indirect data). This addition is made after the comparison of the wear of teeth from the second group (indirect data) with that of the teeth from the first group (direct data) from the same site (see e.g. table 58). In making such a direct comparison of jawbones of cattle from an archeological population it is possible to get round the problem mentioned by Grant (1978) that animals of the same age from different populations can show different stages of tooth wear on account of local factors such as the kind of food eaten and the nature of the soil where the animals grazed. To acquire information on the age after the eruption of the M3, fully developed M3 teeth, both those occuring loose and in the jawbone, can be divided according to the stages described by Grant (1982). Naturally the above-mentioned environmental factors impose limits on the value than can be attached to the data on wear of the M3 when the ages of animals from different sites are compared. In interpreting the data it must be borne in mind that various other factors may affect the age distribution pattern, such as the tendency of excavators to collect mostly large bones (i.e. often bones of older animals), the differential destruction of bones of different ages and the fact that the limits of the classes for age determination are not precise points in time but rather time ranges (Watson 1978). Finally it should be noted that the ages that are mentioned are in fact the ages attained by animals in modern comparative collections. These may differ to a greater or lesser degree from the real ages attained in the past (Meitinger 1983; Bull and Payne 1982; Bullock and Rackham 1982; Noddle 1984). 2.2.8 DETERMINATION OF THE SEASON OF SLAUGHTER In this study no attempts have been made to ascertain the season during which slaughter of farm animals may have taken place, although other authors have done this for example for pigs (Clason 1967; Mohl 1978; Frummel 1983; Van Wijngaarden 1980; IJzereef 1981). I have refrained from attempting to ascertain the season of slaughter because the basic data that are necessary for determining this, i.e. the frequency and time of slaughter as deduced from the bone material, are too uncertain. With the aid of historical data, Meitinger (1983) has shown that the data generally used by archeozoologists for determining age at the time of slaughter on the basis of dentition, as given by Silver (1969) and Habermehl (1961; 1975) for cattle in the past, can give results that deviate considerably from the real historical situation. Seeing that in the case of cattle, for which the age at slaughter can be calculated on the basis of dental and epiphyseal data, the size and variation of these deviations are not known, it is not possible to determine the season of slaughter for cattle. For pigs in former times the data on the season of slaughter deviate less or hardly 36 Fig. 5 Number of births a month registered by the Great-Yorkshire herdbook in 1949 (after Dommerhold 1951). number x 100 6 • at all (Bull and Payne 1982). In the case of pig, the frequency and time of birth are the uncertain factors that make a determination of the season of slaughter dubious (Lauwerier 1983a). A number of authors have estimated the season of slaughtering of pigs on the basis of the age of the animal at the time of death (see above). This procedure is based on the assumption that piglets are born once a year, in March, April or May. On the other hand Boessneck and Von den Driesch (1979) and Hatting (1981) take into account the possibility of two farrowings a year, in February-March and August-September. However, there is biological and historical evidence to suggest that the frequency and time of birth are not definitely fixed but can vary depending on several factors such as food supply, climate, economic factors and the farmer's know-how. The modern sow is on heat every three to four weeks, this not being related to any particular season (Reinders 1879^ Dommerhold 1951; Pond 1983). Therefore the sow could actually be served throughout the year. However, in professional breeding as well as in backyard farming the practice is to have two farrowings a year (Williams 1977). In the Netherlands most litters are born in March and September (fig. 5) (Dommerhold 1951). The sow is pregnant for about four months, according to the count of pig-breeders 3 months, 3 weeks and 3 days. With a regime of two farrowings a year, the sow can feed her litter for two months. With modern indoor farming the maximum is 2.7 cycles of reproduction and lactation a year (Pond 1983). Under normal conditions wild boar mate from the end of October until the end of November and their litters are born in March and April, just like the domesticated pig after a gestation period of four months (Mohr i960; Frädrich 1972). In the Netherlands the mating season for wild boar is in December and January, while the litters are born in April and May (Eygenraam 1973). In years when food is scarce the mating season may be delayed until spring, in which case the litters are not born until June or later. Yet when acorns and beechnuts are plentiful the wild boar can have two litters a year (Oloff 1951; Mohr i960; 37 Eygenraam 1973). From the above the inference is drawn that the possibility of the domestic pig being served throughout the year and the possibihty of having more than one farrowing a year are not characteristics of domestication but are natural characteristics of the species Sus scrofa. The earliest historical sources relating to pig-breeding are treatises on agriculture from Roman times. These are the Rerum Rusticarum of Varro dating from the ist century BC and Columella's De Re Rustica and Pliny's Naturalis Historia, both from the ist century AD. Varro writes: 'The best time for the service is from the beginning of the west wind to the spring equinox (from the middle of February until the 21st of March), as in this case the litter is produced in summer. For the sow is pregnant for four months and will thus bear her young when the land is rich in food' (II.4.7). 'Pigs born in winter are apt to grow thin an account of the cold and because mothers drive them off on account of the scantiness of milk...' (11.4.13). And contrary to what White (1970: 318) maintains about this, Varro goes on to say: 'Her (the sow's) year is naturally divided into two parts, as she bears twice a year, being with young for four months, and giving suck for two' (II.4.14). So Varro describes a regime of two farrowings a year, one in June-July, the other in December-January, adding that the latter is of poorer quality. Pliny mentions the same regime and the same time for the service (VIII.250). As already described by White (1970), Columella mentions two different regimes. The first one, 'in out-of-the-way regions where raising stock is the only thing which pays' (VII.9.4), produces only one litter a year. The piglets are born in June, 'when the grass is already of stronger growth, so that the porkers may find the milk at the perfection of its full strength and also, when they cease to be suckled at the udder, they may feed on stubble and the fruits also which fall from leguminous plants' (VII.9.3). The second regime is for 'districts near towns (where) the sucking pig must be turned into money, for then its mother is saved trouble by not having to rear it and will more quickly conceive and produce another offspring, and so bear twice in the same year' (VII.9.4). In early medieval sources, like Charlemagne's Capitulare de villis, I have not been able to find any information on the subject (Franz 1974). Walter of Henley's Le Dite De Hosebondrie and the anonymous Hosebonderie, two treatises written in early French about husbandry, give information on pig-breeding on 13th-century manors in England (Lamond 1890: 29, 75). The instruction given by Walter of Henley is: 'three times a year ought your sows to farrow, unless it be for bad keeping'. Trow-Smith (1957: 126) calls this a quite impossible frequency because of the brief period of only six or seven days between farrowing and the following service. Yet this is not impossible because after the litter is taken away the sow is on heat within a few days (Dommerhold 1951: 260; Pond 1983). Perhaps Walter of Henley aims at the breeding regime of five litters in two years, with which system there can be three litters in the same year. The anonymous Hosebonderie mentions breeding twice a year: 'And you must know that a sow ought to farrow twice a year...'. Neither of these two treatises mentions the times of farrowing. The accounts of the demesne farm at Wellingborough (Page 1965: 40, 133) mention the time of farrowing twice, in both cases indicating a breeding system of two farrowings a year: 'And (he renders account) for 24 young pigs (received) from the issue of the month of April. And for 18 from the issue of the month of August. And for 20 from the issue of the month of September' (account of the year 1283-1284). And for the year 1322-1323 is noted down: 'Item he renders account for 13 young pigs received from the issue of the month of November and for 5 young pigs from the month of March...'. Also The Book of Husbandry of Master Fitzherbert, dating from 1534, (Skeat 1882: 75) gives an indication of more than one season for farrowing: 'And if thy 38 sowe hauwe moo pygges than thou wik rere, sel them, or eate them, & rere those pygges that come about lententime specyally the begynnynge of somer, for they can-not be rered in winter, for cold, without great coste'. A breeding regime with three farrowings a year like that recommended by Walter of Henley is mentioned by Markham in 1614 in his Cheape and Good Hvsbandry (Markham 1614; 89): 'A sow will bring forth Pigs three times in a yeere, namely at the end of euery tenne weekes...'. The period of 10 weeks mentioned is absolutely impossible and it must be a scribal error or it suggests ignorance as far as pig-breeding is concerned. Handbooks on husbandry and veterinary surgery of the last 200 years all mention two farrowings a year (Leopold 1798; Hentze 1857; Anon. 1874; Carrington et al. 1894; Mangold and Reicherter 1900; Reinders 1904; Kroon 1924; Dommerhold 1951; Williams 1977). Mangold and Reicherter (1900: 301) also mention, like Reinders (1904: 98), a system of five farrowings in two years but do not recommend it because this regime is too great a strain on the sow, while the piglets can be suckled only for a short time and are born partly in an unfavourable season. Dommerhold (1951: 262) also mentions the possibility of three farrowings in two years. In modern indoor pig-farming, as mentioned before, more than two farrowings a year are normal (Pond 1983). Kroon (1924: 2) stipulates that with a regime of two farrowings the sows have to be foddered well, and Hentze (1857: 10) mentions that this system is very common in regions where trade in pigs is important. One could assume that this implies that in cases where these conditions are not met, only one farrowing is usual. From the foregoing it appears that the frequency and time of birth of the wild boar and the domestic pig vary considerably. The frequency is one, two or three farrowings a year or five farrowings in two years. The months in which the litters are born almost cover the whole year. Only for the wild boar can it be said that under normal circumstances the litters are born once a year in March, April or May. To summarize, the frequency and the time of birth of pigs are determined by the following factors. (i) The biological possibilities. In principle the sow, domesticated or wild, can be fertile at any time and can have more than one litter a year. (2) Food supply and climate. From historical data and the wild boar's mating pattern it appears that food supply determines to a great extent how the biological possibilities are realized. The amount of woodland and pasture available and the associated potential food supply, that is to some extent climatologically determined, will increase or decrease the number of litters a year. The food supply available also determines in which months the farmer wants to have young pigs. The cold weather in the winter limits the possibihty of litters in this season. (3) Economic factors. Are pigs raised for the market or for home consumption? Must the animal be fat or lean? Is there a demand for sucking pigs, piglets or full grown animals? It is evident that these last economic factors are partly dependent on taste, for instance sucking pig was preferred by the Romans. (4) Know-how. Finally a farmer's know-how plays a role in employing a certain breeding regime. The procedure for determining the season of slaughtering of pigs, wild or domestic, cannot be applied without due consideration. At every site where one wants to use this procedure one should first take fully into account all the factors mentioned with specific reference to the site concerned. The likelihood that the Romanized farmers in the neighbourhood of Nijmegen had the Roman know-how of employing a breeding regime means that it is possible that their pigs had two litters a year. The climate in the Netherlands provides the opportunity of producing more than one litter a year, and only limits farrowing in the middle of 39 winter. If food supply was not a limiting factor and if breeding piglets was economically attractive then pigs could have been born twice a year. If the pigs were born twice a year this could possibly have been in February-April and in August-September. All these possibilities form too many uncertain factors to permit the determination of the season of slaughter of pigs with any degree of accuracy. 2.2.9 BUTCHERY MARKS Butchery marks, the degree and manner of fragmentation of bones and the spatial distribution of the fragments of different parts of the skeleton can inform us about what happened to the animal after death. Thus the very frequent occurrence of chop and cut marks on the various skeletal elements show how the carcass was divided into pieces for meat distribution and consumption. Specific traces like a smashed forehead can show that an animal did not die a natural death but rather was killed intentionally (IJzereef 1981: 67, 68, 73). Wells containing a very high percentage of horn-cores of goats and cattle in medieval 's-Hertogenbosch indicated the presence of tanneries (Prummel 1978). Distinctive traces of damage on shoulder blades of cattle from the Roman town of Augusta Raurica and from the Roman castella in Valkenburg (Zuid-Holland) and Zwammerdam show that beef was conserved by smoking (Schmid 1972; Van Mensch and IJzereef 1977). The characteristic way in which the extremities of long bones were chopped off in the Roman castellum in Zwammerdam indicates that the bones served as a source of marrow (Van Mensch 1974). In Augusta Raurica severe fragmentation of bone was found to be connected with glue production, as has been shown by chemical analysis (Schmid 1968). Also the way in which butchering was done can be deduced from the occurrence of butchery marks and the fragmentation of the bones (see e.g. Maltby 1979: 38-40; IJzereef 1981: 73-76; Von den Driesch and Boessneck 1975). If one wishes to acquire such information as to what happened to the carcass of an animal after it died, then in the course of analysis of the bone material it is necessary to systematically collect data concerning butchery marks, fragmentation and the spatial distribution of the skeletal elements within the settlement. For the last two points, fragmentation and spatial distribution, this is no problem. The systems that have been developed for recording bone data are intended to include this information (see e.g. Armitage 1978; Uerpmann 1978). With regard to the information on butchery marks the possibilities for recording these are very limited with the code systems for bone data that have been published to date. These systems only allow for the inclusion of scanty information in the data recorded, like the mere presence of a 'cut mark' or a 'chop mark'. If any conclusions arc to be drawn from butchery marks about for example the method of butchering, it will be necessary to establish not only whether a chop mark or cut mark is present, but also where such traces are located on a bone. With large complexes this is impracticable, and in such cases the data on butchery marks will have to be established in a different way. In the past this was done, at least in the archeozoological department of the BAI, using a card system. The butchery marks were sketched individually on cards. Subsequently, on the basis of these sketches some idea of the butchery method used could be gained. The problem with this method is that it depends on rather subjective analysis, that is very time-consuming, notably at the stage of recording the data. In establishing an effective system for recording butchery marks the following conditions must be met: (i) it must be possible to record more or less exactly the kind of butchery trace present and its location on a bone; (2) the way in which the 40 data are recorded must be objective and therefore repeatable; (3) for rapid analysis it must be possible to process the data in a computer; (4) it must be possible to analyze the data statistically; (5) the system must be efficient. Biddick and Tomenchuk (1975) have developed a system for recording butchery traces and fracture surfaces on long bones that fulfils the first four conditions. With this system the data are recorded in the form of a vertical and polar coordinate system. Although in theory this system is appealing, in practice it is complicated and, moreover, it is limited to the analysis of long bones. As far as I known this system has not been applied by other archeozoologists. In order to fulfil the fifth condition, that of eflSciency, without neglecting the other four conditions too much, I have reverted to the old card system with sketches. This has resulted in a code-book with numbered sketches of butchery marks per skeletal element (see appendix). Included in the code are all butchery marks that occur on the material from the Eastern River Area plus a few other butchery marks from other excavations: Bouqras (Clason 1977b); Rijswijk (Zuid Holland) (Clason 1978); Paddepoel (Groningen) (Knol 1983). The code-book consists of the following elements per skeletal part. In the left-hand column is given the code number of the butchery mark. The numerical code consists of two ciphers, so for any one skeletal element a maximum number of 99 different butchery marks can be described. With the complexes investigated it turned out that the number of different butchery marks per skeletal element did not exceed 40. Butchery marks that do not occur in the code-book can be categorized for example under code 99, which means that the fragment concerned is kept apart. If there are many bones with the same butchery marks that fall into this remainder category then it is always possible to add new codes to the code-book. The advantages of a simple code with two ciphers are that the recording of data costs little effort, that the storage space for the information is small, and that the data can be processed statistically in a simple way. The codes per skeletal element form a continuous series of codes beginning with 01, and so on. It was also possible, for example, to reserve the codes 01 to 19 for butchery traces at the proximal end of a bone, the codes 20 to 39 for the diaphyses, etc., so there is a logical connection between the code and the location of the butchery mark on the bone concerned. As a result of this, however, there are many unused codes between the used ones, so the code-book becomes steadily bigger. By making use of a continuous series all the butchery-mark codes for a skeletal element now fit into at most two opposite pages, which is very convenient for anyone using the code-book. In the second column the bone with the butchery mark on it is represented schematically. The butchery mark is indicated by means of a thicker line. If the line extends beyond the outline of the skeletal element, then the bone has been completely broken through at that point. The bones of cattle have been taken as standard bones for the code-book. This is because in the Roman period ruminants, notably cattle, were consistently the most important group of animals. For a few drawings of butchery traces, however, on account of the atypical form of the bone the skeletal elements of another animal species have been used (see e.g. mandibula 29). The choice of the left bone in the case of paired bones is for the sake of conformity with the 'Atlas of animal bones' of Schmid (1972). The third column indicates the position of the butchery mark. An overview of most of the indications of such positions is given in the overall drawing that is recorded for each skeletal element. The nomenclature used is as far as possible the same as that used in 'A guide to the measurements of animal bones from archaeological sites' (Von den Driesch 1976; figs. 3 and 4). 41 In the fourth column, in the places where a drawing is used for two butchery marks, the codes for those butchery marks are repeated. The fifth column indicates whether the code refers to a butchery mark that is a trace of cutting, chopping or sawing. Cut marks occur only on the surface of the bone. They cause only slight damage in the form of a very narrow groove. Traces of chopping occur on the surface of a bone or they form part of a fracture surface. The chop marks on the surface differ from the cut marks inasmuch as they are much wider and deeper. Moreover, they are always absolutely straight. The chop marks that form part of a fracture surface can be recognized as such because of the fact that this part of the fracture surface is absolutely straight and smooth. Also, on account of the rounded shape of most bones the chop mark extends further slightly at one end or at both ends into the bone surface. Such chop marks are often accompanied by one or more parallel superficial chop marks that were probably inflicted when the bone was not broken through at the first blow. In addition to the chop marks mentioned there are also traces of splintering (see e.g. tibia, mark 22). These originate when a joint of meat is struck in a longitudinal direction by means of a butcher's cleaver or chopper. When a bone is struck in this way, then a small splinter of bone flies off. Usually a number of such traces occur together on a bone. Traces of sawing have a characteristic appearance of a flat surface on which groups of more or less parallel scratches are present. The best way of learning to recognize and distinguish between the above-mentioned butchery marks is to take a bone and actually do some 'butchery' on it using a knife, a chopper and a saw. The last column gives a verbal description of the butchery marks. The data that have been recorded with the aid of the code-book can be processed by computer so as to provide frequency tables and diagrams or, in combination with e.g. fragmentation data, matrices. Statistical analyses can be carried out for the material. 2.2.10 PATHOLOGICAL CONDITIONS, TRACES OF GNAWING, BONE ARTEFACTS, ETC. While the bone material was being studied, any traces of pathological deformities, gnawing or burning were also noted. Any bone artefacts that were present in the complexes studied were recorded as well. The first three of these particular categories were recorded together with the rest of the data according to Uerpmann's 'Knocod' system (1978). In the analysis of the data no special attention has been devoted to the seldom occurring pathological conditions, to the traces of burning, to the frequently occurring traces of gnawing (caused mainly by dogs) or to the few bone artefacts that were found. In the text of chapters 3, 4 and 5 some incidental remarks will be made concerning these phenomena. 42 3 The bone material in the settlements 3.1 NIJMEGEN IN THE EARLY ROMAN PERIOD 3.1.1 INTRODUCTION The oldest traces of the occupation of Nijmegen in the Roman period date from the time of the rule of the emperor Augustus (27 BC-AD 14). They may be connected with the military expeditions of Drusus in the years 12-9 BC, the aim of which was to conquer the region between the Rhine, the Danube and the Elbe (Bloemers 1979a). On the edge of the ice-pushed ridge to the south of the Waal valley three occupation terrains dating from the early Roman period can be distinguished (figs. 6 and 7). In the east there is the Kops plateau (figs. 6:D and 7:C). According to current views this must be regarded as a military settlement that came into use around 10 BC. The terrain was abandoned by the Roman army, probably around AD 10, and was taken over by the Batavi, who used it mainly as pasture (Bogaers and Haalebos 1979a). Willems (1984), however, stated that there is no convincing evidence that it did not remain a military camp until around AD 70. On the second terrain was situated the large Augustan army camp with a surface area big enough to accommodate two legions (figs. 6:C and 7:B). The castra must be dated to somewhere between 12 BC and AD 16. The presence of only a few traces of occupation within the army camp suggests that the camp was in use for only a very short time and only periodically (Bloemers 1979b). To the west of the army camp there was a cemetery with a maximal extent of 20,000 m^ where an estimated number of 4000 graves were present (Bloemers I979d). It is very likely that most of the people who were buried in this cemetery are to be associated with the third complex, the occupation of the Valkhof and its surroundings. The bone material dating from the early Roman period that forms the subject of this study comes from this westernmost terrain. On this terrain, called 'Valkhof and surroundings', there was situated a small castellum (fig. 6:B) and a large Gallo-Roman civilian settlement (fig. 6:A and 7:A). The castellum was probably built between AD 10 and AD 20, and after AD 30-40 it was no longer in use. It was surrounded by a V-shaped defensive ditch and an earthen wall, and was large enough to accommodate ca 500 men. Within the fortification no traces of buildings have been found; it is therefore supposed that this was a fortified camp where the soldiers lived in tents. The function of the camp was probably to maintain and control a position of authority that had already been established here (Bloemers 1979c). The large civilian settlement that from the middle of the ist century on also extended over the terrain of the small military fortification was probably the central place of the civitas of the Batavi, that is referred to in classical sources as Batavodurum and Oppidum Batavorum (Bogaers 1979a). The settlement dates 43 ï>Ql3 [5^4 ^2 p^^ ^ 10 Fig. 6 Nijmegen: sites from I2 BC - AD 25. A civilian settlement on the Valkhof and surroundings, B castellum, C large Augustan legionary camp, D settlement on the Kops Plateau. I contour lines, 2 railway, 3 excavated area, 4 topographical coordinates, 5 outlines of buildings (not on this map), 6 double ditch around the Augustan legionary camp, 7 inhabited area; investigated and/or many finds, 8 inhabited area; not investigated and/or few finds, 9 ditch, certain and hypothethical trajectory (not on this map), 10 cemetery; investigated and/or many finds, 11 cemetery; not investigated and/or few finds, 12 road. Scale 1:10,000 (after Willems 1984, fig. 66). from the time of Augustus until about AD 70, and extended over an area of at least 22 ha. The material that has been studied comes from the eastern part of the terrain 'Valkhof and surroundings', the place where the Trajanusplein is now situated 44 11 - 12 -«»,.-, (fig. 8). This is the place where the small fortification was situated, over which the occupation of the Valkhof later extended. The bone research is concerned with excavation trenches 28-36, 58, 73-75 and 78-86. The material mainly comes from pits and ditches. The oldest bones, dated la, may be associated with the castellum. This cannot be established with any certainty on account of the absence of any distinct structures within the fortification. This bone complex is dealt with in section 3.1.2, and is designated 'Nijmegen la'. The younger bone finds, dated Ib-c, certainly belong to the settlement on the Valkhof. They are dealt with in section 3.1.3 and are designated 'Nijmegen Ib-c'. 45 1^. v^ s]^ Eö: .a«° 4 • [^^ ^;;;]^ ^- Fig. 7 Nijmegen: sites from AD 25-70. A civilian settlement on the Valkhof and surroundings, B large Augustan legionary camp, C settlement on the Kops Plateau. I contour lines, 2 railway, 3 excavated area, 4 topographical coordinates, 5 outlines of buildings (not on this map), 6 double ditch around the Augustan legionary camp, 7 inhabited area; investigated and/or many finds, 8 inhabited area; not investigated and/or few finds, 9 ditch, certain and hypothethical trajectory, 10 cemetery; investigated and/or many finds, 11 cemetery; not investigated and/or few finds, 12 road, 13 excavation trench 218: location of the refuse pits of the hand-collecting/sieving experiment (see 2.1.3). Scale 1:10,000 (after Willems 1984, fig. 66). 46 •j- --- L__ 19 13 • 3.1.2 THE BONES THAT MAY POSSIBLY BE ASSOCIATED WITH THE CASTELLUM (NIJMEGEN la) A survey of the hand-collected material, dated to 10 BC-AD 25, is given in tables 6 and mi. The measurement data are given in table mi2. Approximately half of the bones or three-quarters of the mass could be identified to species level. Cattle is the most important species, accounting for 69.6% of the bones (80.6% by weight). Pig comes in second place with 21% (11.8% by weight), followed by sheep/goat with 7.3% (4-4% by weight). Within the sheep/goat group the presence of both sheep and goat could be demonstrated. In addition, a few bone fragments of horse were found, and one of dog. A high percentage of bones could not be identified: 50.5% (24.3% by weight). 47 TABLE 6 Nijmegen Ia: the hand-collected material. Frequencies and weights (g). species domestic mammals cattle sheep goat sheep or goat pig horse dog total identified number Jo weight 192 69.6 7071 80.6 I 0.4 21 0.2 I 0.4 46 0.5 18 6.5 319 3.6 58 5 21.0 1032 II.8 1.8 265 3.0 I 0.4 21 0.2 276 49-5 8775 75-7 18 0/ /o unidentified mammals no size assignment sheep-pig size cattle-horse size 16 6.4 5-7 53 58 1-9 2.1 248 87.9 2704 96.1 total unidentified 282 50.5 2815 24.3 Total 558 11590 TABLE 7 Nijmegen la: allocation of the unidentified mammal bones to the mammal species according to the proportions of the mammal species per size class among the identified bones (weight in g). species sheep-pig size sheep and sheep/goat pig dog cattle-horse size cattle horse number % weight % 4 25-3 16 26.8 12 73.4 42 71.7 0 1-3 I 1-5 242 97-5 2607 6 2.5 97 96.4 3.6 number /o weight 434 80.4 9678 24 4-4 402 70 13.0 1074 II 2.0 362 31 I 0.2 22 0.2 TABLE 8 Nijmegen la: summary of the identified and the allocated unidentified hand-collected bone material (weight in g). species domestic mammals cattle sheep, goat and sheep/goat pig horse dog 0/ /o 83.9 3-5 9-3 These unidentifiable bones consist to a great extent of vertebrae and ribs: 34% (48% by weight). Also skull fragments have a high score: 19% (14% by weight). The other fragments are mostly fragments of bone that could not be identified as part of any particular skeletal element (table mi). The underrepresentation of the small animals with respect to the large animals among the unidentified material is very significant (7^ = 6.978, df= i, p<o.oi). In table 7 the fragments that could not be identified to species level but that could be assigned to a size class are distributed over the species found belonging to the 48 Fig. 8 Early Roman Nijmegen: excavation trenches from which the bone material has been studied. I boundary of excavation trench and trench number, 2 ditch of castellum, 3 road, 4 Nijmegen 28/70: concentration of bones. size classes concerned. The distribution was done on the basis of the percentage distribution of the identified species within a size class. Table 8 gives a summary of the identified and the allocated unidentified bone material. The differences between the values shown in table 8 and those in table 6, for the identified material, are considerable. For the number of fragments of cattle the difference in percentage value is 10.8, for pig 8.0 and for sheep/goat 2.9. Only for horse and dog are the percentage differences minimal: 0.2 and o.2'/ó. In terms of weight percentages the corresponding differences are 3.3, 2.5, 0.9, O.i and 0.0, respectively. Seeing that in the class of large animals in addition to cattle only horse is present, and that horse occurs only rarely with respect to cattle, it is most likely that most of the unidentified bones in the size class of large animals come from cattle. If we assume that the underrepresentation of the small mammals is not or is only to a slight extent due to the method of collection, then to gain an impression of the species composition we can best consider table 8, the summary of the identified and the allocated unidentified material. Cattle then score highest with 80.4%, followed by pig with 13% and sheep/goat with 4.4%. 49 In percentages by weight the corresponding values are 83.9, 9.3 and 3.5. For horse and dog the percentages are 2.0 and o.2"o (by weight 3.1 and o.2"„). No bone material dated to la was found in the excavation trenches 30, 32, 36, 73, 74; 75J 78-84 and 86. In the trenches 28, 29 and 35, in the vicinity of the western exit of the castellum, most of the bone material was found: 78.1% (81.9% by weight). If the bone material is indeed associated with the castellum, then it seems likely that the bone refuse was dumped outside the fortification in the vicinity of the exit. More than a fifth of the material of Nijmegen la, in terms of both frequency and weight, comes from the only refuse pit that contained more than 100 fragments (128) (find, no.: Nijmegen 28/70) (Fig. 8). This trench contains significantly more pig bones than the rest of the find numbers (/^= 14.142, df = I, p<o.ooi). After allocation the percentage of pig in the trench is 24"/o (19% by weight) as compared to 10% (7% by weight) in the rest of the material. However, this bone material is not suggestive of any particular activity at the.spot concerned. 3.1.3 THE BONES FROM THE SETTLEMENT ON THE VALKHOF AND SURROUNDINGS (NIJMEGEN ib-c) Tables 9 and m2 give a survey of the hand-collected material from the period AD 25-70. The measurement data are presented in table mi3. More than half of the number of bones, altogether more than three-quarters of the total weight, could be identified to species level. Among the domesticated animals cattle is the most important species, followed by pig, sheep/goat, horse and dog with values of 69.3%, 14.8%, 11.2%, 3.3% and 1.2% respectively. Also in terms of weight percentages cattle is the most important species with 79.3%, followed by horse with 8.7%, then pig (6.8%), TABLE 9 Nijmegen Ib-c: the hand-collected material. Frequencies and weights (g). species domestic mammals cattle sheep goat sheep or goat pig horse dog O/ /O weight 592 2 I 69-3 0.2 0.1 10.9 14.8 24138 ()/ 79-3 0.4 o.i 1.2 134 19 1019 2085 2643 183 2 0.2 235 0,8 total identified 854 54.6 30456 78.4 unidentified mammals no size assignment sheep-pig size cattle-horse size 25 49 637 3-5 6.9 89.6 68 179 8145 0.8 2.1 97.1 total unidentified 711 45-4 8392 21.6 wild mammals Cervus elaphus Total 50 number 93 126 28 10 - red deer 1565 3-3 38848 3-3 6.8 8.7 0.6 TABLE 10 Nijmegen Ib-c: summary of the identified and the allocated unidentified hand-collected bone material (weight in g). species domestic mammals cattle sheep, goat and sheep/goat pig horse dog wild mammals red deer number 0/ /o weight /o 1198 116 153 57 12 77.8 7-5 9-9 3-7 0.8 31411 1233 2193 3441 192 81.0 3-2 5-7 8.9 0.5 4 0.3 308 0.8 sheep/goat (3.8%) and dog (0.6%). Within the sheep/goat group the presence of both sheep and goat could be ascertained. The wild animals are represented by only two fragments of red deer. The percentage of unidentified bones is high: 45.4% (21.6% by weight). They mainly consist of fragments of ribs, vertebrae and unrecognizable skeletal elements (table m2). The underrepresentation of the small animals with respect to the large animals in the unidentified material is very significant {j^ = 60.680, df= I, p<o.ooi). In table mzó the fragments that were imidentifiable to species level but that could be assigned to a size class are distributed over the species found belonging to the size classes concerned. The distribution is based on the percentage distribution of the identified species within the size classes. Table 10 gives a summary of the identified and allocated unidentified material. The percentages for cattle rise considerably: with 8.5% to 77.8% (by weight with 1.7% to 81%). The values for sheep/goat and pig fall to 7.5% and 9.9% respectively (by weight 3.2"„ and 5.7%). The values for horse, dog and red deer change only slightly: the percentages are 3.7%, o.V% and 0.3%, respectively (by weight 8.9%, 0.5% and 0.8%). For the same reason as mentioned in section 3.1.2, the values given in table 10 can be considered as giving the most reliable picture of the species composition of the settlement. No bones with a dating of Ib-c were found in trenches 30, 36, 58, 79, 80, 81, 83, 84 and 86. In trenches 32, 29 and 35 the largest quantities of bone were found: 40.3%, 18.4% and 10.4% of the total quantity (by weight 37.3%, 16.1% and 12.4%). Three pits are conspicuous on account of the large quantities of bones that they contained. These are the pits with the find numbers Nijmegen 32/25, 32/29 and 35/5, that contained 510, 94 and 154 bone fragments, respectively. The composition of the bone content of these pits did not differ from that of the other find numbers. The sieving experiment that was carried out with the contents of five pits in excavation trench 218 (fig. 7) (see 2.1.3) yielded a few species that do not occur in the above-mentioned hand-collected material. These data show that in the settlement on the Valkhof and surroundings also birds, fish and molluscs were eaten. Bone fragments were found of domestic fowl, duck, woodcock, pike, tench, shad and salmon, as well as shells of the common mussel. Also found were bone fragments of a small rodent, and unidentifiable small bone fragments of birds, fish and amphibians. 51 >• / • J » / A ft 1 _ ^ ft — \ L i o'^ K \ I I I V •^^ 1 - ~\ :: 1 5; •ti '/'ƒ,/ /(' ' -.(^i -?^i \ 5v 52 / S:: , '(f Fig. 9 Nijmegen in the mid-Roman period. A civilian settlement (Ulpia) Noviomagus (Batavorum), B (commercial?) settlement along the Waal, C legionary camp (castra)j D canabae legionis. Hatched: settlement, cross-hatched: cemetery. Scale i : 25,000 (after Willems 1984, fig. 76). '; A JJU^ ' ' ' ', rf 3.2 NIJMEGEN IN THE MID-ROMAN PERIOD 3.2.1 INTRODUCTION In the mid-Roman period, after the rebellion of the Batavi in AD 70 had been put down, there were four main terrains in Nijmegen that were of importance (fig. 9). The civilian settlement of Batavodurum disappeared from the Valkhof and came to lie further west (fig. 9:A). The settlement rapidly grew into a town de facto, and from the time of the reign of the emperor Trajan (98-117) it was called (Ulpia) Noviomagus. Later, probably in the second half of the 2nd century, the settlement acquired the privileges due to a town and the status of municipium (Bogaers 1979b). A small settlement arose after 70 along the Waal, at the foot of the Valkhof (fig. 9:B, io:A). This settlement, that remained in use until the end of the 4th century, was presumably a trading quarter that was most probably associated with canabae belonging to the legionary camp (Bogaers 1979b). The bone material from the mid-Roman period that forms the object of this study comes from two terrains situated to the east: the castra (fig. 9:C, io:B) and the surrounding canabae legionis (fig. 9:D, io:C). On the Hunerberg a military occupation was present in the period AD 70-175, of which the most important was the Legio X Gemina. Probably eight out of the ten cohorts of the legion, about 5000 men altogether, lived in the castra during the period 71 vintil about 104. Before the Legio X Gemina the Legio II Adiutrix was stationed for almost a year in Nijmegen. After the departure of the Legio X Gemina the camp was occupied by the Vexillatio Britannica (ca 104 until 121), the Legio IX Hispana (121 until 130?) and a detachment of the Legio XXX Ulpia Victrix (ca 130 until 175) (Bogaers and Haalebos 1979b). If we include the large army camp dating from the time of the emperor Augustus, then according to current views there must have been five successive army camps present on the Hunerberg. The last of these, built in ca 89 by the Legio X Gemina, was entirely a stone-built construction, that served until the end of the occupation, until AD 175. The camp extends over a surface area of 16.5 ha, and was enclosed by a V-shaped defensive ditch, except on the north side, where a steep slope is present. On the west, north and east sides remains have been found of a defensive wall fortified with towers and buttresses. Around the army camp remains have been found of the canabae legionis, the camp village (Bloemers 1979e). A camp village fell under the jurisdiction of he camp commander but was inhabited by civilians such as craftsmen, traders, inn-keepers, women of easy virtue, and the womenfolk and illegitimate children of the legionary soldiers, who were only allowed to marry after they had completed their term of service. In the excavated parts of the canabae gravel roads have been found as well as traces of houses or workplaces made of stone or wood. In the eastern part there was a large building that possibly served as a market hall or warehouse. Also in this part of the canabae two ovens have been found, that could not have been used at very high tempertures seeing that the inner surfaces are blackened with soot but are not reduced to a cinderized state. It is possible that these ovens were used for smoking meat. To the southwest of the army camp there was an amphitheatre that was built at the time of the occupation of the Legio X Gemina and that remained in use until some time during the 3rd century. The period of maximum activity in both the castra and the canabae legionis lasted from during the 70's until ca AD 120. All of the bone material investigated dates from the period AD 70 to 120 (dating Id-IIa) (fig. 10). Concerning the castra, all the bone material has been studied that came from the excavations carried out by the Rijksmuseum van Oudheden and the ROB in the years 1951-1967 and from the campaigns of the ROB from 1972 until 1978. This entire bone complex is dealt with in section 3.2.2 and is designated 'Nijmegen castra'. The bone material from the canabae legionis comes from the trenches 1-24, 49, 5i> 53-545 61-66, 68-69, 98-99, 101-103, 106, 109-110, 112,117-122, 124-125, 129, 143, 149, 167-168, 177 (eastern canabae); 48, 56, 130, 132, 136-139, 150 (southern canabae); 28-36, 58, 73-75, 78-86 (western canabae). The entire bone complex is designated 'Nijmegen canabae' and is dealt with in section 3.2.3. 3.2.2 THE BONES FROM THE CASTRA An overview of the hand-collected material from the castra is given in tables 11 and m3. The measurement data are presented in table m 14. The percentage of unidentified bones is fairly high: 44.4%, while the weight percentage of 19.1% deviates less from the values found elsewhere in the Eastern River Area. This can be explained partly by the fact that the bones from the castra are relatively poorly preserved and consequently fragile. The average weight of the unidentified bones is therefore very low: 6.6 g. Cattle score highest with 50.3%, followed by pig and sheep/goat with 25.6 and 10.2% (in terms of weight 66.2%, 14.3",, and 5.0%, respectively). In the sheep/goat group only the presence of sheep could be demonstrated. 53 W\^ ŒO]^ zag ^^5== I ^^ mn^ t^^ Fig. 10 Nijmegen: sites from AD 70-early-2nd century. A (commercial) settlement along the Waal, B legionary camp (castra), C canabae legionis. I contour lines, 2 railway, 3 excavated area, 4 topographical coordinates, 5 outlines of buildings (not on this map), 6 ditch, 7 inhabited area; investigated and/or many finds, 8 inhabited area; not investigated and/or few finds, 9 ditch, certain and hypothethical trajectory (not on this map), 10 cemetery; investigated and/or many finds, 11 cemetery; not investigated and/or few finds, 12 road, 13 excavation trenches from which the bone material has been studied, 14 concentration of shoulder blades of cattle, 15 concentrations of fragments of mandibulae and skulls of cattle. Scale i : 10,000 (after Willems 1984, fig. 77). 54 ---._ ^- _ 12 13 1 14 • 15 Horse accounts for 2.9% of the material (9.7% by weight), and dog for 1.2% (0.4% by weight). The wild mammals, red deer, roe deer and hare, together constitute 2.0% (3.5% by weight) of the material. The bird bones found (7.6%; 1.0% by weight) are mostly from domestic fowl. The other bird species are greylag goose (wild or domesticated), duck and dove or pigeon. On the basis of the measurements (table 12) the dove species represented could be the stock dove {Columba oenas), the rock dove (Columba livia) or the domestic pigeon that is descended from it (Fick 1974). In addition two fragments of oyster shells were found. The only clearly recognizable fish bone is the lowermost os pharyngeum of a chub, Leuciscus cephalus. Unfortunately this bone is not numbered, so it cannot be ascertained whether the bone did indeed come from the castra. 55 TABLE II Nijmegen castra: the hand-collected material. Frequencies and weights (g). species number domestic mammals cattle sheep sheep or goat pig horse dog domestic birds domestic fowl 0/ /o weight % 8i6 50.3 24071 66.2 I 164 O.I 10.1 23 0.1 416 25.6 1794 5186 14-3 47 2.9 1.2 3521 20 139 9-7 0.4 88 5-4 274 0.8 4-9 wild or domestic birds Anser anser Anas platyrhynchos Columba sp. greylag goose mallard dove or pigeon 20 1.2 76 12 0.7 0.2 26 0.2 0.1 4 0.0 wild mammals Cervus elaphus Capreolus capreolus Lepus capensis red deer roe deer brown hare 27 1-7 0.3 1135 31 5 106 I 0.1 4 0-3 0.0 2 o.r 2 0.0 1623 55.6 36361 80.9 6.3 11.6 82.1 moUusca Ostrea edulis total identified unidentified mammals no size assignment sheep-pig size cattle-horse size unidentified bird remains unidentified fish remains - oyster 4 229 17.8 543 388 30.2 668 52.0 994 7060 9 12 I I total unidentified 1295 Total 2918 44-4 8610 19.1 44971 The unidentified mammal bones consist for 34.1% (71.4% by weight) of vertebrae and ribs, and for the rest mainly of fragments of unrecognizable skeletal parts (table ms). Comparing the mammal bones found for the two size classes, we see that the percentages for the unidentified bones from the sheep-pig size class (36.7%3 12.3% by weight) are lower than for the identified bones from this class (39-7%5 20.1% by weight). Testing the numbers of the unidentified and the identified bones from the two size classes gives a very significant difference between the identified and the unidentified bones (/^ = 8.223, df= i, p< 0.005). In table m27 the fragments that could not be identified to species level but that could be assigned to a size class are distributed over the species in those classes. In table 13 an overview is given of the identified and the allocated unidentified bones for each species. Comparing this table with that for the identified bones (table 11), we can see that there is a shift of 2.8% with the number of fragments of cattle. In the case of the other species the value shifts not at all or only a few tenths of a percent. With the weight percentages the most considerable shifts are 56 TABLE 12 Measurements of the pigeon (or dove) bones. The measurements* are those according to Fick (1974), which differ in a few cases from the standard measurements according to von den Driesch (1976) that are used elsewhere in this pubhcation. skeletal element humérus humérus tibiotarsus tibiotarsus measurements (mm) find no. 107/33 107/33 107/33 96/15 GL DP BP KC BD TC 43.6 43.6 56.5 17.0 17.8 II.9 II.8 6.4 6.5 5.0 5.0 2.9 3-0 10.2 ID. I 6.4 3.6 3-6 - 91 8.8 - — GL: greatest length; DP: diagonal of the proximal end; BP: breadth of the proximal end; KC: smallest breadth of the corpus; BD: breadth of the distal end; TC: depth just above the condyles. TABLE 13 Nijmegen castra: summary of the identified and the allocated unidentified hand-collected bone material (weight in g). species domestic mammals cattle sheep and sheep/goat number 0/ weight /o 29987 2066 5898 4389 158 67.5 4.6 1429 271 682 82 53-1 10.1 25.4 3.0 33 1.2 47 8 1-7 0.3 1417 121 3-2 0.3 I 0.0 4 0.0 133 4-9 392 0.9 fish I 0.0 I 0.0 mollusca 2 0.1 2 0.0 pig horse dog wild mammals red deer roe deer brown hare birds 13-3 9-9 0.4 evident for cattle and pig, 1.3 and 1.0% respectively. The shifts in value for the other species are less than i %. In addition to mammal bones the unidentifiable material includes nine bird bones of domestic fowl/duck size and a fragment of a fish bone. 3.2.3 THE BONES FROM THE CANABAE LEGIONIS Tables 14 and m4 give an overview of the hand-collected material from the area around the army camp. The measurement data are presented in table mi 5. Of the bones 63%, or 84.2% by weight, could be identified to species level. Once again cattle is the most abundant species with 83.4% (86.5% by weight), followed by pig and sheep/goat with 7.0 and 3.6% (3.4 and 1.9% by weight). Within the sheep/goat group only the presence of sheep could be demonstrated with certainty. The domesticated mammals are represented in addition by horse and dog, with values of 3.5 and 0.8%, respectively (7.2 and 0.2% by weight). Among the birds, that account for 1.0% (0.1% by weight) of the identified material, the wild or domesticated greylag goose is the most abundant species, followed by the domestic fowl and the wild or domestic duck. The wild mammals, aurochs and red deer, together form 0.8";, (0.7% by weight) of the material. Within this group red deer is the more important species. Only one fragment of a fish has been found: a tench (Tinea tinea). 57 TABLE 14 Nijmegen canabae legionis: the hand-collected material. Frequencies and weights (g). species domestic mammals cattle sheep sheep or goat pig horse dog number 0' '0 weight /o 1643 6 83.4 56430 0.3 1035 65 138 3-3 7.0 86.5 1.6 0.3 69 3.5 0.8 3-4 157 0.2 0.3 14 0.0 12 0.6 0.1 47 3 0.1 2 15 domestic birds domestic fowl wild or domestic birds Anser anser Anas platyrhynchos greylag goose mallard wild mammals Bos primigenius Cervus elaphus aurochs red deer fresh water fish Tinea tinea total identified unidentified mammals no size assignment sheep-pig size cattle-horse size unidentified bird remains - tench 207 2247 4674 7-2 0.0 I 0.1 0.7 69 361 0.1 14 I 0.1 I 0.0 I97I 63.0 65245 84.2 35 155 952 31 13.6 71 581 11554 0.6 4.8 94-7 83-4 r; total unidentified 1159 Total 3130 0.6 22 37.0 12228 15.8 77473 A total of 37";, (15.8% by weight) of the bones have not been identified to species level. The unidentified bones consist for 3,1% (45% by weight) of ribs and vertebrae, and for the rest mainly of unrecognizable fragments (table m4). If we compare the identified and the unidentified mammal bones from the two size classes (table 14), then we see once again that the bones of the small mammals are underrepresented within the unidentified group (;^^ = 3.903, df= i, p<o.05). The percentage differences between the identified and the unidentified material are, however, small compared to other sites. For the small animals the percentage for the identified material is 11.5%, and for the unidentified material 14.0%. In terms of weight the differences are smaller: 5.6"',, and 4.8% respectively. Table m28 gives the allocation data for the bones from the two size classes, and in table 15 an overview is given of the identified and allocated unidentified material. Comparing table 15 with the data in table 14, we can see that the percentage shifts per species that are a result of the unidentifiable state of a proportion of the material are very small. The maximum shift for the number of bones is i.o"„ in the case of cattle, while the other shifts are 0.5",, or less. In terms of weight there is only a shift of o.i"o in the case of cattle. 58 TABLE 15 Nijmegen canabae legionis: summary of the identified and the allocated unidentified hand-collected bone material (weight in g). species domestic mammals cattle sheep and sheep/goat pig horse dog number 0/ 2548 82.4 120 233 107 3-9 7-5 3-5 25 0.8 weight % 67025 1440 2605 86.6 5552 182 7-2 0.2 1-9 3-4 wild mammals aurochs red deer 2 0.1 81 O.I 22 0.7 430 0.6 birds 36 1.2 86 0.0 I 0.0 I 0.0 fish In addition to the above-described mammal bones 17 fragments of bird bones were present among the unidentifiable material. 3.2.4 THE RELATION BETWEEN THE CASTRA AND THE CANABAE LEGIONIS If we compare the bone material from the castra with that from the cabanae legionis then we see a great difference in the relative proportions of the domesticated meat-producing mammals, i.e. cattle, sheep/goat and pig. The percentages for sheep/goat and pig are considerably higher in the castra. On the basis of the data obtained after allocation, the percentage value for sheep/goat in the castra is 11.4% as compared to 4.1% in the canabae (in terms of weight 5.4% as compared to 2.0%), while the percentage value for pig is 28.6% in the castra as compared to 8.0% in the canabae (in terms of weight 15.5% as compared to 3-7%)For a difference such as this in the bone material there are three possible explanations, (i) In the castra relatively many small farm animals were eaten, in particular pigs. (2) The pattern of meat consumption was actually the same in the castra and the canabae, but a proportion of the cattle were slaughtered in the canabae and consumed in the castra. (3) The pattern of meat consumption was the same, but a proportion of the - larger - cattle bones were removed from the castra after slaughtering had been done. We shall first of all consider this third point. Obviously refuse and in particular the malodorous refuse of slaughtered animals would have been taken away from the castra. The large quantities of uninvestigated bone material from the rubble layers just outside the eastern entrance to the army camp may be indicative of this (trenches 147, 152, 155-157, 159-160). If in particular slaughter refuse of cattle had been removed from the castra, then one would expect the percentage of cattle bones in these rubble layers to be very high, and the percentage of sheep/goat and pig bones to be very low. This information is not yet available, however. Yet if we look at the distribution of the skeletal elements of cattle in the castra (fig. 11), we have to reject the hypothesis of this selective rubbish disposal. If we use as reference material the data of all investigated complexes from Roman Nijmegen then it is conspicuous that precisely the bones that generally produce the biggest fragments and that are therefore the easiest to dispose of, i.e. the metapodials, are found in abundance in the army camp. Also the second hypothesis, the slaughtering of cattle in the canabae and consumption of the meat thus provided in the castra, is refutable on account of 59 Fig. 11 Cattle: distribution of skeletal elements. A total for Roman Nijmegen (n = 3208), B castra (n = 529), C canabae legionis (n=i350). Sk: skull, Ma: mandibula, Sc: scapula, Hu: humérus, U-R: ulna and radius, Pe: pelvis, Fe: femur, Ti: tibia. Me: metapodia. % A 30 20 1 10 - \ 0 B weight-% 30 Roman Nijmegen 30n 1 \ J Castra '"- 20100- C 10- M to 1.0-1 Canabae legionis - 30 30 20 20H - 10 10 1 0 Fig. 12 Cattle: distribution of skeletal elements. A eastern part of canabae legionis (n = 678), B western part of canabae legionis (n = 623). Sk: skull. Ma: mandibula, Sc: scapula, Hu: humérus, U-R: ulna and radius, Pe: pelvis, Fe: femur, Ti: tibia. Me: metapodia. Sk Ma Sc Hu 1 0-R 1 Pe 1 Fe Ti He 7o A 1— 1 0- 0 Sk Ma Sc Ha Sc Hu 0-R Pe Fe li Me Ti Me weight-7o 50 50-1 Canabae-east 103020100- B 1,0 30 tOn Canabae-west - 30- 20 10 70- - ID- —r-r-^^- 0 Sk Ma Sc Hu U-R Pe Fe Ti Me n-J 1 Sk Hu 1 U-R 1—\ Pe Fe the percentages of metapodials in the castra. Metapodials are typical items of slaughter refuse on which there is no meat present. Consequently, in modern slaughterhouses as soon as the animal has been killed and its neck cut open the metapodials are first of all cut off and disposed of. The percentages of metapodials in the castra are higher than in the canabae and also higher than in the entire material of Nijmegen. There is no evidence to suggest that metapodials were specially brought to the castra for bone working. Nor is it likely that metapodials were specially brought there for the extraction of the marrow, since 60 TABLE I 6 Nijmegen canabae legionis west. Butchery marks on the shoulder blades of cattle (code: see appendix 'Butchery mark code'). category A B C code number 2 20 3 6 7 8 14 II 12 13 13 5 2 2 4 2 2 I? 19 16 D 20 6 E 32 6 33 34 36 jg 40 3 F 4 5 a proportion of the bones are complete. It is therefore most probable that the cattle were brought into the army camp on the hoof, and were slaughtered on the spot. We are thus left with the first hypothesis that more meat of sheep/goat and particularly pig was eaten in the army camp than in the canabae. This will be discussed in further detail in the section 'production and consumption of animal products' (see 4.1). Although the comparison of the bone spectra for cattle from the army camp and from the canabae gives us no reason to assume that there were certain places where slaughtering was done and where meat was further processed or consumed, the bone spectra for parts of the canabae and notably for some pits are indeed suggestive of this (fig. 12). In the eastern canabae the very large quantities of skull fragments and especially mandibulae are striking, while shoulder blades are conspicuously almost absent. In the western canabae precisely a large quantity of shoulder blades were found, that are moreover less fragmented than usual. On account of the slight degree of fragmentation this quantity finds expression particularly in the weight percentages. Data from the southern canabae are disregarded because of the small number of bones present there. Among the 892 bone fragments identified as cattle from the eastern canabae there was only one scapula fragment, which is extremely little in view of the normal picture. We must conclude from this that the shoulder blades of animals slaughtered in the eastern canabae were transported elsewhere. One possibility is that they were taken to the western canabae where larger quantities of shoulder blades were found than is usual. Of those excavation features of the western canabae that contained more than 100 bone fragments, the pit Nijmegen 30/1-3, 7-9, the well Nijmegen 74/17, 21 and the pit Nijmegen'74/18 (fig. 10), the last two of these are responsible for this overrepresentation of shoulder blades in the western canabae (fig. 13). The butchery marks that have been found on the shoulder blades from the western canabae are presented in table 16 (the codes refer to the butchery mark code in the Appendix). These butchery marks can be grouped in six categories, A-F, of which the first four are associated with removal of the shoulder blade from the thoracic girdle. The butchery marks of categories E and F are indicative of the smoking of shoulders of meat (i.a. Schmid 1972; Van Mensch and IJzereef 1977). Schmid presumes that the hole in the scapulae at the spot where the bone is thinnest may have been caused by a hook used for hanging up the shoulder of meat for the purpose of smoking it. In three scapulae from the western canabae a hole is clearly present, while in many more scapulae there could have been such a hole present. In many cases the proximal part of these shoulder blades has become broken off and has disappeared, so it is not possible to ascertain whether a hole was formerly present. The second indication that the meat may have been smoked is the presence of traces in category E. These traces, that often occur in combination, could have been caused by the scraping of meat from the bone; indeed one of the eflfects of smoking is that the meat becomes more firmaly attached to the bone (Van Mensch and IJzereef 1977). Similar indications of the smoking of shoulders of meat are frequently found, not only in settlements of the Roman period but also in earlier and later times (see i.a. Schmid 1972; Van Mensch 1975; Tiessen i960; Van Mensch and IJzereef 1975, 1977; Kokabi 1982). All in all, the evidence suggests that shoulders of meat were taken from the eastern to the western canabae; also cattle slaughtered elsewhere may have been taken there. These shoulders may have been smoked in the neighbourhood of the above-mentioned well and pit in trench 74, after which the meat could have been scraped off the bones. The place of consumption of this meat could have been 61 Fig. 13 Cattle: distribution of skeletal elements. A Nijmegen 74/17 (n = 47), B Nijmegen 74/18 (n= 114), C Nijmegen 30/1-3,7-9 (n = 259). Sk: skull, Ma: mandibula, Sc: scapula, Hu: humérus, U-R: ulna and radius, Pe: pelvis, Fe: femur, Ti: tibia, Me: metapodia. % A B weight-'/o 50- Nijmegen Bo- 7i. /17,21 U). ton 30- 30 20- 20 10- 10- 0 0 50- Nijmegen 50- 7t»/16 W- iO- 30- 30- 20- 20- 10- 10-. 0 C 0 30n Nijmegen 30/1-3,7-9 20 10 0 Sk Ma Sc Hu U-R Pe Fe Ti Me Sk Ma Sc Hu ' U-R Pe Fe Ti Me either the canabae or the castra. Naturally it is also possible that no transport took place from east to west and that also in the eastern canabae there was a spot where shoulders of meat were collected and processed. If such a spot existed then it must lie in a part of the eastern canabae that has not been investigated archeologically. The third place where more than 100 fragments were found in the western canabae, the pit Nijmegen 30/1-3, 7-9, is not suggestive of any particular butchery activities. The distribution of the skeletal elements more or less corresponds to the general pattern in Nijmegen (compare fig. 13 C and 11 A). Ahhough shoulder blades of cattle are absent from the eastern canabae, fragments of skulls and mandibulae are present there in large quantities (fig. 12 A). Of the excavation features that yielded more than 100 fragments, the section of the defensive ditch Nijmegen 1/135, the pit 106/21 and the rubble layers 129/3-4, 6, lo-ii, the defensive ditch and the pit account for this excessive amount of fragments of mandibulae and skulls (fig. 14). Also the finds from what may have been a small oven mainly consist of skull and jawbone fragments (Nijmegen 143/5J 8). The rubble layers, on the other hand, contain not only a minimal quantity of shoulder blades but also a minimal quantity of mandibulae. The metapodials, butchery refuse of the first order, are overrepresented, however, even if we take into consideration the low percentages for mandibula and shoulder blade. Evidently mainly butchery refuse was disposed of here, in addition to refuse from the kitchen and dining table. The missing shoulder blades have been discussed already above. It would appear that also the mandibulae and to a lesser extent the skulls were kept apart and processed separately after the 62 Fig. 14 Cattle: distribution of skeletal elements. A Nijmegen 1/135 (n = 359), B Nijmegen 106/21 (n= 119), C Nijmegen 143/5.8 (n= 56), D Nijmegen 129/3,4,6,10,11 (n = 34). Sk: skull. Ma: mandibula, Sc: scapula, Hu: humérus, U-R: ulna and radius, Pe: pelvis, Fe: femur, Ti: tibia. Me: metapodia. % A B C weight-% 70^ Nljmegen 70n 1/135 60- 60- 50- 50- (.0- 1,0- 30- 30- 20- 20- 10- 10- n_ n 60 n Nijmegen SO- 40- LO- 30- 30^ 20- 20- 10- 10- Nijmegen 1.0 J 30- 30- 20- 20- 10- 10- fl- D 1 50^ K3/5,8 10- , 50-| , Nijmegen 1 r- n 129/3A,6,10,11 1,0- 30- 30- 20- 20- 10- 10Sk Ma 1 Sc 1 1 Hu U-R Pe Fe Ti Me — 1 50- 40- , 1 60n 106/21 50- 50-1 1 Sk Ma Sc Hu U-R Pe 1— Fe Ti Me animals had been slaughtered. The few visible butchery marks on the extremely weathered material provide little firm evidence of butchery techniques. The diastema was cut through immediately in front of the tooth row (butchery mark 4, 7 and 27; see appendix). In addition mark 19 is present, a chop mark immediately below the processi, caused by the detachment of the jawbone from the skull. A possible explanation is that the skulls and jawbones in the neighbourhood of the places where they occur in high concentrations were 63 processed there for the production of brawn. To make this meat product the head of a pig or ox is cooked together with other meat for several hours. Subsequently the bones are removed, the meat is finely minced and herbs are added, after which the mass is left to cool and solidify in a mould. To summarize, on the basis of the evidence available so far the following can be said about the relation between the castra and the canabae. In the castra considerably more meat of sheep/goat and especially of pig was eaten than in the canabae, although most of the meat consumed was provided by cattle. Since we have only relative data at our disposal it is possible that in the castra just as much meat from sheep/goat and pig was eaten as elsewhere but that less beef was consumed there. This would imply that the soldiers ate less meat than the people living in the canabae. As a general rule cattle that were eaten in the castra would have been slaughtered there; the bone material does not provide any clear evidence to suggest that animals were mostly slaughtered in the canabae and consumed in the castra. In the canabae some sites can be pointed out, however, where a specialization in the processing of meat took place. In the western canabae this probably involved the smoking of shoulders of beef. In the eastern canabae ox-heads were processed, probably into brawn. It is possible that these meat products from the canabae were partly or entirely intended as a source of food for the inhabitants of the army camp. 3.3 NIJMEGEN IN THE LATE ROMAN PERIOD 3.3.1 INTRODUCTION It appears that after the middle of the 3rd century the occupation of Nijmegen was concentrated around the Valkhof and its surroundings, situated on the ridge of land to the east of (Ulpia) Noviomagus and to the west of the terrain of the former castra (Wynia 1979) (fig. i5:A). This concentration can be explained by the steadily increasing disturbances along the border of the empire caused by the invading Germanic tribes from the other side of the Rhine, and by the ultimate collapse of the Roman lines of defence. In addition the increase in water level and the accompanying frequent floods may have caused the low-lying bank of the Waal in the west to have become unsuitable for occupation (Bogaers 1979b). Nevertheless, it is evident for a few sparse finds that the old town was not abandoned by its inhabitants altogether. Here below an account will be given of the bone material that was found in one of the defensive ditches that surrounded the settlement on the Valkhof in its successive phases (fig. 16). Subsequently a description will be given of the faunal material from the cemetery on the terrain of the nursing home 'Margriet', one of the two cemeteries that most probably belonged to the fortification on the Valkhof (fig. I5:C). The bone finds from the settlements and cemeteries will then be compared. 3.3.2 THE 4TH-CENTURY SETTLEMENT ON THE VALKHOF (NIJMEGEN IV) 3.3.2.1 Introduction During the first half of the 20th century various excavations were carried out and observations were made that were indicative of a 4th-century occupation of the Valkhof (Thijssen 1980). The most important data concerning a late Roman settlement on the Valkhof and its surroundings were provided by the six 64 TABLE 17 Nijmegen VI: the hand-collected material. Frequencies and weights (g). species domestic mammals cattle sheep goat sheep or goat pig horse dog number /o 971 62.0 5 I 8 0.3 61.9 I.O 0.4 25 0.0 0.4 2.6 0.5 4447 1927 2963 490 1-7 0.3 1565 80.1 171128 94.0 I 159 349 16 total identified aurochs elk (moose) red deer wild boar /o 105953 184 118 263 5128 49201 429 domestic birds domestic fowl wild mammals Bos primigenius Alces alces Cervus elaphus Sus scrofa weight 6 2 33 8 0.1 0.5 10.2 22.3 0.1 2.1 O.I O.I 0.2 3.0 28.8 0.3 I.I unidentified mammals no size assignment sheep-pig size cattle-horse size 0.3 4-9 94-9 42 0.4 19 369 134 10666 98.4 total unidentified 389 19.9 10842 6.0 Total 1954 1.2 181970 defensive ditches that have been excavated since 1969 (Bloemers 1986). These ditches probably represent different phases of the settlement. The largest defensive ditch, that contained the bone material to be discussed below, was found in 1969 on the building site of the cultural centre 'De Lindenberg' (Bogacrs 1969). In 1979 and 1980 it became possible to trace the further course of the ditch on 'De Lindenberg' and on the 'Kelfkensbos' (fig. 15:6) (Bloemers et al. 1979; 1980a). Although the entire course of the ditch could not be traced, on account of digging activities undertaken in the i6th and 20th centuries, it is clear that the ditch enclosed the whole area of the present-day Valkhof, ca 3 ha (Bloemers 1986). The moat is V-shaped in cross-section, and is 14-15 m wide and 5-6 m deep. The 31 coins that were recovered during the excavation by Bogaers (1969) all date from between 330 and 350 (Haalebos 1976: 204-205). The coins and the pottery found during the later excavations confirm a dating for the ditch in the first half of the 4th century. Apart from this large defensive ditch, to the south of it a system of two parallel ditches has been found (fig. 15:9). This double ditch also dates from the 4th century and enclosed an estimated area of at least 8 ha. In addition traces have been found of three ditches that ran to the north of the large ditch and parallel to it. These three ditches cannot be dated with certainty. Concerning the chronology of the six ditches and of the different phases of the settlement on the Valkhof so far little can be said with any certainty (Bloemers 1986). 65 is^ Eö: D m^^ WM' urn S? 9 Fig. 15 Nijmegen: late Roman sites. A settlement on the Valkhof, B cemetery around Marienburg, C cemetery around Hugo de Grootstraat ('Margriet'). I contour lines, 2 railway, 3 excavated area, 4 topographical coordinates, 5 outlines of buildings (not on this map), 6 ditch around the Valkhof, 7 inhabited area; investigated and/or many finds, 8 inhabited area; not investigated and/or few finds, 9 double ditch around investigated and uninhabited area, 10 cemetery; investigated and/or many finds, II cemetery; not investigated and/or few finds, 12 road. Scale i : 10,000 (after Willems 1984, fig. 86). 66 f- ~ ~ L- 12 Also the nature of the settlement on the Valkhof is still unclear. Data provided by the cemeteries that most probably belonged to the settlement, in which remains of men, women and children have been found, indicate that we are concerned here with a civilian occupation that may have included a military component (Bloemers 1986). 3.3-2.2 The hand-collected material An overview of the hand-collected material is given in tables 17 and ms. The measurement data are presented in table mi6. The percentage distribution for both the number and weight of bones is given separately for the identified and the unidentified bones. The most abundant species is cattle, with ca 62% of bones in terms of both 67 Fig. 16 Excavation trenches on the Valkhof and the position of the 4th-century ditch. I boundary of trench and number of trench, 2 ditch. Scale i : 2,000. numbers and weight. Of the meat-providing domesticated animals pig comes in second place with 10.2",, (3.0% by weight). Sheep and goat, both of which are definitely present, account for less than i % of the material. Among the bones from the ditch horse is relatively abundant: about a quarter of the bones in terms of both numbers and weight. Dog accounts for 1% (0.3% by weight). Only one bird species has been found, namely the domestic fowl. The seven fragments form 0.4% of the material (in terms of weight <o.05%). The wild mammals, aurochs, elk, red deer and wild boar, together constitute 3.1% of the number of bones (5.7"o by weight). Within the group of wild mammals red deer scores highest as far as the number of fragments is concerned. The bone fragments of aurochs and elk that were found are rather heavy, on average i kg; consequently the percentages of these species in terms of weight are relatively high. Of the bones 2o"o (6% by weight) could not be identified. Almost half of the number of these unidentified fragments (6o"o by weight) are vertebrae (table 68 TABLE I 8 Nijmegen IV: summary of the identified and the allocated unidentified hand-collected bone material (weight in g). species domestic mammals cattle sheep, goat and sheep/goat pig horse dog wild mammals aurochs elk red deer wild boar birds 0/ number 7ü weight 1234 63.2 0.8 62.0 22.7 112822 576 5232 52390 0.9 438 0.2 7 0.4 0.1 4735 2055 2.6 2 42 2.1 3155 9 0.4 500 1-7 0.3 7 0.4 25 0.0 15 174 443 18 8.9 0.3 2.9 28.8 I.I m5). If we compare the bones belonging to the two size classes, we see a very significant difference between the identified and the unidentified bones (j^= 18.035, df = I, p<o.ooi). Among the unidentified bones the small animals are underrepresented. In table m29 those fragments that could not be identified to species level but that could be assigned to a size class arc distributed over the species belonging to the classes. The distribution is based on the percentage distribution of the identified species within a size class. Table 18 presents an overview of the identified and allocated bone material. Comparing the data in this table with those for the identified bones in table 17, we can see that the error that is due to part of the material being unidentifiable is very small, in terms of both numbers and weight. The differences are of the order of tenths of a percent, at most 1.2% in the case of the number of fragments of cattle. On account of the low weight of the unidentified fragments the percentage distribution of the weights of the species is hardly affected. The maximum difference here is 0.1%. About a quarter of the bone material from the defensive ditch is of horse. During the excavation it was observed that in any case in trench 183 some of the horse bones lay in the soil in an articulated state (personal communication J.R.A.M. Thijssen). In view of this fact and of the relatively high average bone weight, and seeing that no butchery marks are to be found on any of the bones, it may be concluded that the animals were not slaughtered for consumption. Therefore we are probably concerned here with horses that were used for riding, or as pack animals or draught animals, that died from natural causes, or that were killed in action, or that had been put down on account of their advanced age, or because they were diseased or considered to be no longer fit to do their job. It is remarkable that in contrast to other sites (see 4.7.2) there are no indications of the animals having been skinned. Naturally, this does not mean that there is any evidence to suggest that the horses from the ditch were not skinned. When the cattle bones were studied to ascertain the possible presence of aurochs (see section 2.2.2.2), it was found that six fragments dating from the 4th century belonged to wild cattle. These consist of two skull fragments with fragments of horn-core and the isolated point of a horn-core with find no. 182/6, and three metacarpals with the find nos. 183/13, 183/14 and 185/12. The skull fragments, a left and a right one, overlap slightly and therefore must come from two different animals. The isolated point and the left skull fragment, both showing recent fractures, possibly belong together. The three metacarpals, 69 Fig. 17 Nijmegen, 4th-century ditch: antler fragments of red deer, a-b skull fragments from which the antler has been sawn off below (a) or above (b) the pedicle, c thin piece of antler that has been sawn off the beam, d beam fragment from which thin pieces have been sawn off on two sides, e-f tine, g fragment of beam or tine. of which only the proximal parts have been found, are in any case from two different individuals. When the pig bones were studied to see whether wild boar was present (see section 2.2.2.3), it was found that eight fragments came from wild pigs. The bones concerned are two right tibiae (find no. 183/13), a canine tooth from the upper jaw (183/13) and one from the lower jaw (182/6), and four upper jaw fragments (183/13, 14). The four upper jaw fragments, two left and two right, most probably come from two individuals. The age of these individuals, on the basis of the eruption of the teeth, is older than 24 months. The two canines, both from male animals, come from animals older than 10 months in age. The two tibiae, that come from two individuals, indicate an age of more than 2 years. In the case of one of the tibiae the processus of the distal epiphysis has been cut off (butchery mark 27; see appendix). Red deer is the most frequently occurring species of game animal in the bone material from the defensive ditch. The 33 fragments that were found come from various parts of the skeleton (table ms). There are no antler fragments present that had obviously been naturally shed. With the five fragments including the base of an antler, the antler base is fused with the skull. These five fragments must therefore come from hunted animals. There are no indications that collecting shed antlers for the purpose of making artefacts was of any importance. The presence of saw marks and shaving marks on antlers suggests that they were used as raw material for making certain objects. The antlers were sawn off the skull above or below the pedicle (fig. 17:3, b). Evidently the pedicle was not always valued as raw material. Fig. I7:c shows semi-finished artefacts sawn from the dense cortex of the beam, the central axis of the antler. The irregular outer 70 Fig. 18 Nijmegen, 4th-century ditch: skull fragment of an elk showing many traces of sawing, chopping and cutting. surface has been partly smoothened, probably with a knife or plane in view of the longitudinal parallel grooves present. Fig. lyid shows a piece of a beam from which thin pieces of'antler have been sawn off in this way on two sides. The antler tine with the find no. 186/6 has been cut off at the wide end in the shape of a wedge, while the upper and lower surfaces have been smoothened somewhat (fig. I7:f). In view of the irregular structure of the wedge-shaped extremity and the smoothened surfaces, a chopper or axe must have been used on this antler. The tine 183/13 has been sawn oflf at the base, while at the tip it has been worked on four sides (fig. lyre). The piece of antler that is illustrated in fig. 17:9 is a sawn-off fragment of a tine or of a thin part of the beam. Also these last three fragments are probably semi-finished artefacts. Two bones of elk have been found, a metatarsus and a fragment of a skull (find, no. 183/14 and 182/6). Both bones come from adult animals. The skull fragment comes from a male animal and shows many traces of sawing, chopping and cutting. The occipital bone, that part of the skull that is connected to the first cervical vertebra, has been chopped through from left to right. Clearly this was done to separate the head from the trunk. The left pedicle has been broken off, while the right one has been sawn off entirely. Evidently the antler was used as raw material for making particular objects. There are many cut marks and chop marks running diagonally over the frontal bone. It is possible that this very tough part of the skull was used for a while as a cutting or chopping block (fig. 18). 71 Fig. 19 Nijmegen, 4th-century ditch: western north-south profile of excavation trench i6i showing the position of the sieve samples. I dirty-black filling with much charcoal, 2 clean yellow sand, 3 lower limit of soil disturbed after the Roman period, 4 sieve sample. EI3-3.3.2.3 The sieved samples Samples for sieving were taken from the 4th-century defensive ditch in five places (table 19). The three samples with trench no. 161 were taken from the western north-south profile of trench 161 (fig. 19). The samples 161/37 and 161/38 come from two dirty-black layers that contain much charcoal. Sample 161/39 comes from the deepest part of the ditch, that was filled with clean yellow sand. Of the two other samples one (182/6) comes from the dirty-black fill of the ditch in trench 182, while the other (183/20) comes from a greyish-brown fill in the ditch, from trench 183. In view of the spatial distribution of the samples within the ditch and the very small number of samples from such a large object - 5 samples from a total volume of 415 litres from a ditch containing ca 1674,600 litres - it is clear that these samples cannot tell us anything about the ditch as a whole. They only give local information about the profile in trench 161 and the sampling locations in trenches 182 and 183. In addition these five samples can provide further information about the occurrence of species that were overlooked when hand-collecting was being done, although such information can only be very limited in view of the small number of samples taken. All samples were first sieved using first a sieve with a mesh width of 5 mm and then one with a mesh width of 2.5 mm. Tables 20 and 21 give an overview of the faunal material found by means of sieving. TABLE 19 Nijmegen IV: sieve-samples; trench/find nos, x- and ycoordinatcs, height levels (-|-NAP) and sample volumes. 72 trench/find no. 161/37 161/38 161/39 182/6 182/20 x-coordinate y-coordinate height (-I-NAP) volume (1) 188.1544 188.1543 188.1615 188.2705 188.2817 428.8362 428.8359 428.8325 428.7978 428.7941 29.80 29.10 27.40 30.67 29.90 TOO 75 100 125 15 Nijmegen IV: sieved material from the 5 mm fraction (5 mm sieve). Frequencies (n) and weights (g). TABLE 20 161/38 161/37 sample no. sample size (1) lOO 75 n g domestic mammals cattle sheep or goat n g 182/6 125 n 5 1 13 - red deer birds, unidentified 183/20 15 g n g 33-6 0.7 127.9 10 - 43.7 - I 715.2 - 2 6.1 - I O.I I 0-3 0-3 fish Perca fluviatilis Salmo salar cf. trutta Sphyraena sp. unidentified perch salmon or sea trout barracuda moUusca Ostrea edulis unidentified oyster unidentified mammals no size assignment sheep-pig size cattle-horse size 5.6 3 2-9 I 7-4 7 3 5 16 18.0 17 10 Total TABLE 21 — I 0.2 2 0.2 — _ — I I 0.4 0.3 _ — 125 26 17 47-8 344 55-7 13 2 0.7 196 334.7 26 761.7 3-0 51 _ - 14.0 0.0 I.I 2.1 - Nijmegen IV: sieved material from the 2.5-5 tnm fraction (residue of 5 mm sieve sieved with 2,5 mm sieve). Frequencies (n) and weights (g). sample no. sample size (1) 161/37 161/38 161/39 100 75 100 n fish Esox lucius Cyprinidae unidentified g 3.0 0.8 1.8 pig horse wild mammals Cervus elaphus 161/39 100 n g n g n 0.1 - - - 182/6 25* g - pike I unidentified birds 183/20 15 n g I 0.0 2 0.0 4 0.1 n g - - 0.3 unidentified amphibians 0.0 unidentified mammals no size assignment 73 4-9 16 1.2 Total 76 5-3 16 1.2 0.0 37 0.2 0.5 45 0.3 0.5 only one-fifth of the residue of the 5 mm sample was sieved with the 2.5 mm sieve. 73 In addition to a few fragments offish bones that were unidentifiable to species level in both the 5 mm and the 2.5 mm fraction, a fragment was found of the salmon or sea trout (Salmo salar cf. trutta) as well as one of perch (Perca fluviatilis), both in the 5 mm fraction. In the 2.5 mm fraction also pike {Esox lucius) was found to be present, as well as two fragments of fish of the family Cyprinidae. The 5 mm fraction of sample 182/6 contained an exotic element: the vertebra of a fish belonging to the genus Sphyraena (see 4.5). In the 5 mm fraction of sample 182/6 a fragment of the shell of an oyster (Ostrea edulis) was found, as well as a small piece of another indeterminate mollusc. In the 2.5 mm fraction of this sample there was a fragment of an amphibian, that was not identified any more closely. Unidentifiable fragments of birds were present in the two fractions of sample 161/37. For the reasons mentioned above we cannot use the data provided by these sieve samples for drawing conclusions about the content of the ditch as whole. Consequently the data also cannot be used for testing the reliability of the hand-collected material as far as domestic farm animals and wild animals are concerned. The fact that in the hand-collected material small animals like birds, fish and amphibians are greatly underrepresented or are totally absent is generally well known (see i.a. Clason and Prummel 1977; Payne 1975; Watson 1972), and is emphasized by the data from the hand-collected and the sieved material from the ditch. Here too it is once again confirmed that the small fraction of 2.5 mm makes no meaningful contribution to the data on the larger mammals (sheep to cattle size range), since the fragments of bones of these species are too small to permit their identification. Concerning the samples from the western north-south profile the following observations can be made. The two samples taken in the uppermost layers, 161/37 and 161/38, both come from dirty-black fill containing a lot of charcoal (fig. 19). On the basis of the occurrence of so much charcoal in the layers it could be assumed that the rubbish, after it was thrown into the ditch, was burned there. This could have been done to reduce the quantity of rubbish or to combat offensive smells. However, since there are no traces of burning on either the sieved or the hand-collected bone material, there is no evidence to suggest that animal refuse was burned in the ditch. The sample from the deepest part of the ditch, 161/39, contains no bone material at all. The fact that the fill of this deepest part of the ditch consists of thin layers of clean yellow sand suggests that this sand became washed into the ditch when it was still being used as a defensive moat and not as a rubbish dump. In that case one cannot expect to make many finds there. The sample from the deepest part provides no evidence to confirm or refute this view since there is almost no bone material at all preserved in the clean yellow sand (cf. 3.3.3.2 (cemetery)). Even if bone material had been thrown into the deepest part of the ditch, it is unlikely that we would have been able to find any. 3.3.2.4 The density of finds The density of finds from the ditch is expressed in numbers of finds per m^ (n/m^) and the weight of bone per m^ (g/m^). The units of volume for which the find densities have been calculated are blocks of no standard size. These are arbitrary units that were determined by the particular circumstances connected with practical aspects of the excavation. The blocks are listed together with the associated find numbers plus the find densities per block in table 22. In view of the arbitrary subdivision of the ditch and the varying intensity of collecting during the excavation, the possibilities of making a spatial comparison of the find densities in the ditch are limited. Nevertheless, there are indications that in terms of both frequency and weight a high concentration is present in 74 TABLE 22 Nijmegen IV: density of finds per excavation unit (block). For each block the block number is given, together with the associated trench/find numbers, the block volume and the density of finds in terms of numbers and grams per m^. block find nos. volume* (m') n/m' elm' 13 I 161/35,36 lOO.O 0.190 2 182/19,20 182/5,22 180.6 0.050 5 0.000 0 182/6,7,16-18,21 70.4 197-6 1.027 189 183/12 100.8 0.000 0 98.9 II.891 790 183/14-20 104.2 2.418 323 185/no finds 107.9 0.000 0 185/2,10 254.3 48.8 0.244 30 1.312 120 99-4 63-3 175-7 35-8 0.895 107 3 4 5 6 7 8 9 183/5.13 10 185/6-8 II 185/9,11,12 12 i86/no finds 13 14 15 186/1 186/12 186/13 186/14,15 r6 total 0.000 0 0.296 0.279 20 28.2 0.248 27 8.7 0.690 38 1674.6 1.167 109 38 data provided by P.A.M. Zoetbrood (ROB). blocks 7 and 6 ofthat part of the ditch that is situated in trench 183 (fig. 20 and 21). The adjacent blocks, insofar as they include the lowermost part of the ditch (4, 10, 11), also have a somewhat higher density of finds than the rest of the ditch. This suggests that in that part of the ditch where trench 183 is situated much bone material was thrown away, and that this is detectable in adjacent parts of the ditch too. For the potsherd material the find densities have been determined in g/m3 per trench (table 23). Although the find densities for pottery were determined more roughly, with larger volume units, the pottery densities present the same general picture as the bone material: a concentration in that part of the ditch where trench 183 is situated. From this it can be concluded that this part of the ditch was used as a rubbish dump. The reason for the concentration precisely here can be explained if we assume that at this spot there was an exit from the settlement. An argument for this is the fact that trench 183 is probably situated halfway along the whole length of the ditch. However, if such an exit existed then there must have been a bridge present, and one would expect to be able to find some traces of a transition from the ditch to such a bridge. Yet no such traces are in evidence; at this point the ditch shows no special features, while the exit on the west side of the settlement is a passageway through the ditch, i.e. a place where the ditch is interrupted (Bogaers 1969: 2). Another possibility is that there was a road that came to a dead TABLE 23 Nijmegen IV: density of finds of pottery per trench. For each trench the weight of pottery found, the volume of the ditch and the density of finds are given (data provided by P.A.M. Zoetbrood, ROB). trench weight (kg) volume (m^) density of finds (g/m^) 0 161 2.8 182 24.0 183 46.2 ±100 448.6 303.9 185 22.9 510.4 6 152 45 186 8-9 311.7 29 75 Trench 161 -^ Fig. 20 Nijmegen, 4th-century ditch: schematic longitudinal cross-section. The density of finds per block is indicated in terms of fragments per m'. 1 present-day ground surface, 2 lower limit of soil disturbed after the Roman period, 3 block no. end next to the ditch, precisely for the purpose of conveniently dumping rubbish. This is all mere speculation, seeing that nothing is known of the structure within the settlement. 3.3.3 THE 4TH-CENTURY CEMETERY MARGRIET 3.3.3.1 Introduction On both the cast and the west side of the 4th-century fortification on the Valkhof cemeteries have been found that most probably belong to the settlement. The eastern burial ground, the cemetery now named 'Margriet', was partly excavated in the years 1980-1983 (Bloemers et al. 1980a, 1981; Willems et al. 1982, 1983) (fig. I5:C). It is estimated that this cemetery contained some 850 graves, of which 326 have been excavated; these were almost exclusively inhumation graves. In some graves dishes were found containing animal bones as gifts for the dead (partly discussed previously: Lauwerier 1983b). As material for archeozoological investigation usually only the remains of meals are available: garbage and refuse of meals in the past. The interesting thing about the bone material from the cemetery is that we are not dealing with the garbage and refuse but with the meal itself: a sort of plate-service for the dead. 3.3.3.2 Dishes and bones During the excavation 326 graves were found of which 72 contained items of pottery on which one could expect to find bones: plates, dishes and casseroles (personal communication P.A.M. Zoetbrood, ROB). Although during the excavation of the graves bones were often observed on the dishes, the contents of only a few dishes have remained preserved, on account of the very unfavourable conditions for preservation in the sandy soil of the cemetery. Even a slight touch caused the bones to disintegrate into dust. The human skeletal material gives a 76 Trench 161 0 151-200 » 1-50 201-400 ^M 101-150 600 - 800 ^ Fig. 21 Nijmegen, 4th-century ditch: schematic longitudinal cross-section. The density of finds per block is indicated in terms of bone weight per m'. 1 present-day ground surface, 2 lower limit of soil disturbed after the Roman period, 3 block no. clear indication of how poorly bone has been preserved. In ca 35% of the graves no bone material is present any longer. In most of the other graves only a fragile bony pulp of the skull remains. Less than 10% of the graves contain fragments of both the skull and of the bones of the legs (personal communication P.A.M. Zoetbrood, ROB). The contents of a few dishes were treated with a preservative during the excavation, in order to preserve them as far as possible prior to archeozoological analysis. The contents of the dishes will be discussed further below. A summary of the data is given in table 24. The percentages mentioned in the table under contents recovered indicate approximately the proportion of the contents of the dishes, i.e. of the undisturbed soil containing the bones, that were conserved during the excavation. The data concerning the pottery and the human skeletons were obtained from P.A.M. Zoetbrood (personal communication, ROB). No. I. Pig: distal half of a right humérus. The distal epiphysis is fused, but because the bone is very small the age of the animal at the time of slaughter would not have been much more than one year. The dish also contains a few unidentifiable fragments of mammal bones. No. 2. A few unidentifiable bones. No. 3A few unidentifiable bones of an infantile mammal. No. 4. Pig: diaphysis of a right humérus. The distal epiphysis is not fused: age younger than one year. 77 Nijmegen IV: overview of the pottery from the 'Margriet' cemetery that has been studied (x: presence of animal species estabHshed). TABLE 24 grave find no. pottery place contents recovered (%) I 2 3 4 5 6 7 8 9 10 II 12 13 14 15 16 17 18 19 20 21 61 61 61 85 105 105 126 153 197 213 238' 251 252 257 260 271 271 294 307 307 308 192/61 192/62 192/63 195/27 201/6 201/9 202/54 206/6 211/197-4 214/213-1 210/I90-I 215/251-1 215/252-1 216/257-17 216/260-19 219/271-7 219/271-9 219/294-8 219/307-30 219/307-35 219/308-8 TS-dish (Chenet 304) coarse ware (Pirling 127/128) TS-dish (Chenet 324) TS-dish (Chenet 304) TS-dish (Chenet 313) coarse ware (Pirling 120/122) coarse ware (Pirling 127/128) coarse ware (Pirling 127/128) coarse ware (Pirling 127/128) coarse ware (Pirling 127/128) TS-dish (Dragendorf 31) coarse ware (Pirling 126) coarse ware (Pirling 121/122) coarse ware (Pirling 122) coarse ware (Pirling 121/122) TS-dish (Pirling 26) coarse ware (Pirling 122) TS-dish (Pirling 34) TS-dish (Chenet 304) TS-dish (Chenet 304) TS-dish (Pirling 34) in in in in in in in in in in in in in in in in in in in in in niche niche niche niche niche niche niche niche niche niche niche niche middle of grave grave outside coffin grave outside coffin niche niche grave outside coffin grave outside coffin grave outside coffin niche 10 0' 0» 10 25 40 75 10 100 75 75 100 75 100 100 100 100 100 100 100 100 No. 5. Cattle: processus olecrani of the right ulna of an adult animal. No. 6. Pig: fragment of the head of a sucking pig. In both the maxilla and mandibula the milk incisors, the Pd3's and the Pd4's are present. The milk premolars are not worn, the Pdz's are not yet present: age between 7 and 10 weeks. Unidentifiable small fragments of bones. No. 7. (fig. 22) Pig: head of a sucking pig. The cranium is fragmented. The Pd2's are just breaking through: age about 10 weeks. Domestic fowl (?): articulated parts of ulna, radius and humérus, probably from a domestic fowl. Three large unidentifiable fragments could be parts of the hind legs. Also a lot of bird rib fragments are present. The position of the articulated bones of the wing, the unidentifiable long bones and the fragments of ribs strongly suggest that we are concerned here with the bones of one complete bird that was interred in an intact state. No. 8. Unidentifiable fragments of a young mammal. Some fragments are from long bones. No. 9. Domestic fowl: articulated skeleton. Clearly identifiable are the two humeri, the 78 human skeleton animal species cattle _ - pig dom. fowl unident. parts still present X X - _ - X - X X - - X X - - X X - X - X no bones no bones no bones no bones no bones no bones no bones no bones no bones no bones X X skull skull skull skull+arm skull+femur skull+femur no bones skull skull complete skull no bones no bones no bones no bones ± complete ± complete no bones no bones no bones no bones sex age f f f f m m ? f f f ? ? ? ? ? 5^65 56-65 56-65 ? ? ? ? 31-40 41-50 41-50 child^ 31-40 16-20 26-35 •> ? ? adult^ child^ child^ adult^ adult^ child^ ' bone fragments were also collected ^ on the basis of the dimensions of the coffin. ' previously disignated as grave 190-1. right ulna, a phalanx, fragments of the vertebral column, ribs, the sternum, parts of the pelvis and the two femora. Also unidentifiable fragments were found, including parts of long bones. Cattle: on top of the skeleton of the domestic fowl a fragment of a rib was found. No. 10. A few unidentifiable mammal bones. No. II. Domestic fowl: almost complete articulated skeleton. Missing parts are the head, the right femur, the right foot and the left leg. If this leg was present, it most probably lay on the broken and cleaned fragment of the dish. It is remarkable that the dish was made in the 2nd or 3rd century AD and was placed in the grave only much later (personal communication J.H.F. Bloemers). No 12-21. No bones found. On the basis of bone finds in the cemetery in Courroux, Bechert (1982: 284) proposes that the dead were supplied with second-rate meat while the funeral guests kept the choicest pieces for themselves. The finds from the 'Margriet' cemetery give a different impression. That we are dealing here with meals and not with the refuse of meals is best illustrated by the articulated skeletons of domestic fowl. One can only find articulated skeletons if complete animals have been buried from which eatable parts have not been cut off. For the two best conserved 79 Fig. 22 Nijmegen, cemetery 'Margriet': dish of coarse ware containing the remains of the head of a sucking-pig and fragments of a domestic fowl (?). skeletons of domestic fowl from the graves nos. 197 and 238 it is striking that there are no traces of the head and the uneatable parts of the hind legs. This brings to mind the well known picture of a fried chicken, from which the head and feet have been cut off. From Roman times too there are examples of such a way of preparing fowl. The ornamented bronze lid, probably of a vessel for food, from Mundelsheim from the 2nd or 3rd century AD, is one such example (fig. 23) (Paret 1938; personal communication Ph. Filtzinger, Württembergisches Landesmuseum Stuttgart). The fowl is presented on the border of the lid: the head and the ends of the hind legs have been cut off. Also represented are a ring of piglets and, flanked by grapes, a hare. Fowls prepared in this way are also known from the cemeteries of Leuna, Wessling, Neuburg and Kempten (Gandert 1953; Keller 1971, 1979; Mackensen 1978). In Nijmegen a humérus of a sucking pig was found on two occasions, this being indicative of good quality meat (graves nos. 61 and 85). Also heads of sucking-pigs have been found (graves nos. 105 and 126). Nowadays we tend to consider the meat of this part of the animal to be of poorer quality, but there are indications in written sources that this kind of meat was in fact a delicacy in Roman times. Macrobius (3, 13, 12) tells about a banquet that Lentullus offered his guests in about 70 BC on the occasion of his ordination as a priest. Besides oysters, thrushes with asparagus, fried hare and fowl also fried heads of pigs were served. But pig's head is also known to have been a delicacy in later times, as in medieval and Tudor England (Wilson 1973: 82). And in the folk tale of 80 Uilenspiegel, as placed by De Coster (i867)in 16th-century Flanders, Lamme complains that he has to live on bread and water while the nobility feast on head of wild boar, among other things (III, 29). All in all, the dishes from Nijmegen on which bone remains were found give the impression that food was placed in the graves that was not second-rate or the refuse of a meal of, for example, the funeral guests, having only a symbolical meaning, but rather a good meal for the dead. In view of the small number of finds it is not possible to discern any relation between the ages of the deceased persons and the meals they were given. All that can be said is that the dead of all ages, children as well as adults, were provided with meat as a grave gift. Cattle and pig bones were found both in graves of women and in the only grave known to be that of a man. Domestic fowl was found only once in the grave of a person of known sex, in this case a woman. In Courroux birds were found exclusively in the graves of women. From this it has been concluded that birds were specifically grave gifts for women (Martin-Kilcher 1976). The pottery that was studied (12-21) shows no traces of bones at all. From this it could possibly be concluded that boned meat was given in these cases. Yet if we consider the human skeletal material then we see that human remains were found Fig. 23 The bronze lid from Mundelsheim (West Germany) showing rings of fowl and piglets and in the centre a hare (photograph, Württenbergishes Landesmuseum Stuttgart). 81 TABLE 25 Overview of the occurrence of cattle, sheep/goat, pig and domestic fowl in graves within the northwestern part of the Roman Empire. species identified cemetry early-Roman Kempten Weisenau Brugg mid-Roman Hörafing Eining Stephanskirchen Regensburg Courroux cattle sheep/goat pig dom. fowl II 7 128 I 2 - I 37 - I 7 I 4 I I - 47 4 2 I late-Roman Nijmegen Krefeld Neuburg Neuss Augsburg Göggingen Burgheim Valley Potzham Altenstadt Wessling Oudenburg 3 5 4 I I — 25 3 19 Total Percentage 23 6 18 4 277 69 21 2 I 4 6 - 3 I 2 I I I 2 7 84 in only one grave. Evidently the conditions for preservation in the graves were so poor that no conclusions at all can be drawn from the finds of empty items of pottery in the 'Margriet' cemetery. 3.3.4 THE RELATION BETWEEN THE SETTLEMENT AND THE CEMETERY It is obvious that we should compare the bone material from the 'Margriet' cemetery with that from the stronghold on the Valkhof (3.3.2). Comparing these two sets of bone material, we can see that the values for frequency of occurrence of bones of cattle, sheep/goat, pig and domestic fowl are completely different at the two sites. Unfortunately, however, the number of identifiable finds from the cemetery is so small (9!) that no conclusion whatsoever can be drawn from this comparison. In order to make a more meaningful comparison between animal bones from settlements and those from cemeteries, archeozoological data have been collected relating to 27 settlements and 20 cemeteries or individual graves within the Roman Empire. For the cemeteries a random selection has been made from the available literature in which faunal material is mentioned, from sites in the Netherlands, Belgium, Germany, Austria and Switzerland (table 25). These include early Roman cemeteries in Kempten (Mackensen 1978), Weisenau (Kessler 1927) and Brugg 82 TABLE 26 General overview of the occurrence of cattle, sheep/goat, pig and domestic fowl in settlement refuse within the Roman Empire. species cattle sheep/goat pig domestic fowl mean percentage distribution of the number of fragments 60 26 2 (Tomasevic and Hartmann 1972); cemeteries dating from mid-Roman times in Horafing (Gerhardt and Maier 1964), Eining (Kellner 1965), Stephanskirchen (mentioned in Kellner 1965), Regensburg (Mackensen 1973) and Courroux (Martin-Kilcher 1976; Kaufmann 1976); late Roman cemeteries in Krefeld (Pirling 1974), Neuburg (Keller 1979), Neuss (Harke 1980; Reichstein 1980), Augsburg, Goggingen, Burgheim, Valley, Potzham, Altenstadt, Wessling (Keller 1971), Oudenburg (Mertens and Van Impe 1971) and the previously described cemetery in Nijmegen. In the table no distinction has been made between animals or parts of animals that were found on dishes or plates and those that were simply placed in the grave. Possibly a few bones are present that accidentally came into the grave together with the earth used to cover the remains of the deceased. As in the great majority of cases the faunal material found was described as clearly belonging to the grave, this will hardly influence the overall picture of the occurrence of the various kinds of animals interred. The numbers shown in the table indicate the frequency with which the animal species occur in the graves. In the case of one grave containing faunal remains of one species on different plates, separate counts have been made for each plate. Pig is the most abundant species, with a frequency of 69%, followed by domestic fowl at 21%. Cattle and sheep/goat account for only 6 and 4% respectively. In addition to the species mentioned in the table other species occur incidentally: horse in Oudenburg, dog in Neuss and Courroux, hare or rabbit in Horafing, goose in Courroux and Oudenburg, eider duck (Somateria mollissima) in Neuss, fish in Weisenau and oyster in Oudenburg. The few fragments of horse and dog that have been found in graves probably cannot be considered as representing food for the dead as these species were not normally eaten in Roman times (Luff 1982). Martin-Kilcher (1976) indicates the possibility that dog may have been provided as a companion for the journey to the hereafter. As an indication of this he mentions the presence of a ceramic figure of a dog in one of the three graves in Courroux that contained dog remains. Table 26 gives an overview of the settlement refuse of the most frequently consumed mammals and the domestic fowl in Roman times. The table includes data from 27 sites where a total number of more than 100 bone fragments of cattle, sheep or goat, pig and domestic fowl have been found. The sites concerned are of both civilian and military settlements in the Netherlands, Germany, Austria, France and Switzerland (Clason 1977a: tables 15 and 18). The table gives the mean percentage distribution of the number of bone fragments. This overall picture of the species composition in all the settlements considered jointly corresponds more or less to the frequency spectra within the individual settlements (Clason 1977a: 126). The data from the settlements and the cemeteries cannot be directly compared because they express different quantities, namely numbers of fragments and numbers of individual animals respectively. This problem is a consequence of the different ways of presentation of the bone material in the literature consulted, that is partly accounted for by the different nature of the objects excavated. In a self-contained structure like a grave it is more sensible and more meaningful to indicate individuals than in the refuse pits of a settlement. However, this difference in presentation does not prevent us from gaining insight on a broad scale into the differences between settlements and cemeteries. The most conspicuous differences are as follows. Cattle, that are represented in the settlements by 60% of the fragments, are found in very small quantities in the cemeteries. Pig, on the other hand, that constitutes 26% of the number of fragments in the settlements, is the most important animal in the cemeteries, certainly if one compares the mammal species with one another. Among the 83 Fig. 24 Frequency distributions of cattle, sheep/goat, pig and domestic fowl in settlements, cemeteries and the cookery book of Apicius/Caelius. The frequency percentages are based on numbers of fragments for the settlements, the frequency of occurrence for the cemeteries and the number of times a species is mentioned in the cookery book. settlements p;:::> Cemeteries 60- III cookery book W- 20- cattle sheep/goat pig domestic fowl mammals, sheep/goat comes in third place in both the settlements and in the cemeteries, although the percentage for the settlements is higher than that for the cemeteries. The domestic fowl, that rates 2% in the settlements, is the second most abundant animal in the cemeteries. The above-mentioned differences can partly be explained by a different method of excavating cemeteries and settlements. If the cemeteries have been excavated more meticulously than the settlements or if sieving has been carried out, something that is not mentioned in most of the publications concerning the cemeteries, then it is possible that notably the quantitative difference in domestic fowl remains may be explained as a result of this (Clason and Prummel 1977; Clason et al. 1979). That the differences for mammals can also be attributed to different excavation techniques seems unlikely. If this were indeed the case, then one would expect that in addition to pig also sheep/goat, that fall into the same size class, would score higher, yet this is not so. The low numbers of bird remains in the settlements may also be due to gnawing by dogs. Yet this does not explain the difference in occurrence of mammals between the settlements and the cemeteries. An explanation for the scarce occurrence of cattle in the cemeteries could be that beef, in contrast to pork, was completely removed from the bone before being placed in the grave, so that in the cemetery no traces of cattle are to be found. Also Mackensen (1978) mentions this possibility. The considerable differences for domestic fowl are difficult to reconcile with such an explanation, however. Finally, it is well possible that the dead were provided with a special meal that was different to what the living were accustomed to eating every day. Concerning these two last possible explanations for the differences between the settlements and the cemeteries, it cannot be said which is the more likely. However, if the latter explanation is correct, then one may suggest what the reason is for the frequency distribution of the bone material. In fig. 24 the data of the tables 25 and 26 are presented alongside the frequencies with which the different animal species are mentioned in the cookery book of Apicius/Caelius. This cookery book dates from the ist century AD. It was written by the well-to-do gastronome Apicius, and, in the form of the book that is known to us, may have been revised by a certain Caelius at the end of the 4th century or beginning of the 5th century (Forbes 1965). For the quantitative data concerning the occurrence of the various animal species in the cookery book the English translation of Flower and Rosenbaum (1958) has been used. In addition to the 84 species listed in the figure, namely cattle, sheep/goat, pig and domestic fowl, in the cookery book there is occasional mention of a great number of other species: wild sheep, wild goat, red deer, fallow deer, boar, rabbit, hare, dormouse, goose, duck, hazel hen, partridge, pheasant, peacock, crane, ostrich, parrot, pigeon, wood pigeon, turtle dove, thrush, fig-pecker, flamingo, electric ray, murena, eel, conger eel, anchovy, scorpion-fish, perch, sea-perch, sea-bream, gold-bream, dentex, red mullet, gray mullet, horse mackerel, tunny fish, bonito, sole, sheat-fish, 'cornuta', prawn, sea crayfish, squill, lobster, squid, cuttlefish, octopus, mussel, oyster, snail, sea urchin and jellyfish. Again it must be pointed out that the frequency percentages in fig. 24 cannot be directly compared with those for the cemeteries and settlements. In the figure different quantitative categories stand alongside one another: numbers of fragments for the settlements, numbers of individuals for the cemeteries and the number of times a species is mentioned in the cookery book. However, the figure does appear to serve a purpose in demonstrating general tendencies. If we assume that beef was not cut from the bone and that the frequency with which normally available kinds of meat of cattle, sheep/goat, pig and domestic fowl were mentioned in an exclusive cookery book is a measure of the extent to which the diff'erent kinds of meat were appreciated, then a broad comparison of the data from the three sources concerned indicates that the dead were not provided with ordinary, everyday food but with something more festive. Naturally the argument for this is dependent on the validity of the two presuppositions. In the above, an attempt has been made to explain the apparent discrepancy in the frequency of faunal remains from cemeteries and settlements. No evidence is available to support any of the explanations. The two explanations, that the difference in the species spectra is caused by different excavation methods or by gnawing by dogs, seem to be the least acceptable, notably for the mammals. The lack of clear information as to the method of excavation in the literature consulted limit the possibilities for testing this explanation, however. The two other explanations are diametrically opposed to each other and are dependent on the question as to whether or not the beef that was placed in the graves was removed from the bone. If the answer is yes, then this provides an explanation for the different bone spectra, at least as far as the mammals are concerned. If the answer is no, then the explanation lies in the fact that the dead were provided with a different kind of meal, possibly one regarded as being of higher quality, from that normally eaten by the living. This investigation is merely an initial attempt to compare the meals for the dead with those for the living, and as such it has many limitations. The relation between a settlement and the associated cemetery has not been considered, with the exception of the cemetery and settlement in Nijmegen. To gain further insight into this material an analysis will have to be made, according to a rigidly applied research method, of the different factors that play a role in the interpretation of the data (Jones 1977). Research will have to be focussed on among other things the relations between the settlements and the associated cemeteries, the geographical differences, the nature of the settlements, the degree of Romanization, the relation between the faunal material found in the graves and the other archeological finds from the graves, and the way in which bones are found in the graves (whether on dishes or not). 85 3.4 THE CASTELLUM ATMEINERSWIJK 3.4.1 INTRODUCTION In 1979, in the Meinerswijk polder (municipality of Arnhem), to the south of the course of the Rhine in Roman times, the remains were found of a stone building and a system of ditches around a series of Roman castella that had been built at the same spot (Willems 1984). The castella may possibly be identified as Castra Herculis, although this is a matter of some discussion (Bogaers 1968; 1981a; 1981b; Willems 1980b; 1981b; 1984). The Roman occupation of Meinerswijk can be divided into six periods, of which the first falls in the 2nd decade AD while the last one ends at the beginning of the 5th century. Faunal material was found for all periods with the exception of the last one; thus the material dates from a period lasting from the 2nd decade AD until some time in the first half of the 3rd century. In view of the small numbers of bones found faunal material will be discussed here only in its entirety. 3.4.2 THE BONES An overview of the material found in Meinerswijk is given in tables 27 and m6. The measurement data are given in table mi7. Of the 149 animal bones found 87.9% (97-1% by weight) could be identified to species level. Among the domesticated mammals, cattle score highest with 67.9%, followed by pig and sheep/goat with 16.8% and 6.1% (in terms of weight percentages: 90.4%, 4.7% and 3.1%, respectively). In addition two bones of a dog were found, namely a mandibula and a phalanx I. Domesticated birds are represented by two bones of domestic fowl. Two bones of red deer were found. One bone has been identified as greylag goose, wild or domesticated, and one bone comes from a duck of indeterminate species. Four fragments of pike were found. The only marine species found is oyster. The mammalian fragments that could not be identified to species level mainly consist of fragments of ribs and other, indeterminate skeletal parts belonging to the group of the large mammals (table m6). The small animals are underrepresented compared to the material identified to species level. Significance cannot be tested using a /^ test because of the too low expected frequency of unidentified material belonging to the group of the small mammals (Schefler 1979). In table m30 the fragments that could not be identified to species level but that could be assigned to a size class are distributed over the species belonging to those classes. The distribution is based on the percentage distribution of the identified species within the size classes. An overview of the identified and the allocated unidentified material is shown in table 28. The most important shift can be seen for the frequency percentage of cattle, that increases by 3.0",, to a value of 70.9%. Pig decreases by 1.6% to reach 15.2%. The other shifts are less than 1.0%. Among the percentages by weight the maximum shift is only o.2"o, namely for cattle. One animal species that belongs to the context of Meinerswijk but that is absent from the list is horse. It is possible that horse does not occur in Meinerswijk. It is more likely that horse remains have simply been missed quite by chance. In this connection the combination of two factors may play a role: the small size of the excavation area (222 m^; incompletely excavated on a vertical scale), and the 86 TABLE 27 Meinerswijk: the hand-collected material. Frequencies and weights (g). weight number species domestic mammals cattle sheep or goat pig dog domestic birds domestic fowl 89 67.9 8 22 6.1 16.8 I 6140 212 /o 90.4 0.8 321 22 31 4-7 0.3 2 1-5 3 0.0 wild or domestic birds Anser anser - greylag goose I 0.8 18 0.3 wild mammals Cervus elaphus - red deer 2 1-5 43 0.6 I 0.8 2 0.0 wild birds Anatidae fish Esox lucius - pike 4 3.1 9 0.1 moUusca Ostrea edulis - oyster I 0.8 19 0.3 131 87.9 6789 97.1 total identified unidentified mammals sheep-pig size cattle-horse size I 5.6 17 94-4 4 199 98.0 total unidentified 18 12.1 203 2.9 Total TABLE 28 species domestic mammals cattle sheep/goat pig dog 2.0 6992 149 Meinerswijk: summary of the identified and the allocated unidentified hand-collected bone material (weight in g). number /o 106 70.9 8 5-5 15.2 weight 6338 214 90.6 31 4.6 0.7 323 22 2 1.6 44 0.6 2.7 23 0.3 fish 4 4 2.7 9 0.1 moUusca I 0.7 19 0.3 wild mammals red deer birds 23 I 0.3 87 fact that bones of horses, unlike those of other animals, are often not spread over an occupation area but are rather clustered. It is known that at other sites cadavers of horses were dumped in certain places or were buried (see e.g. Nijmegen IV, Kesteren and Ewijk). 3.5 THE MILITARY VICUS AT KESTEREN 3.5.1 INTRODUCTION Roman Kesteren is represented by two cemeteries and by traces of an extensive occupation lasting from the last decades of the ist century AD until some time in the 3rd century (unless stated otherwise, the information given in this introduction is derived from R.S. Hulst, ROB, personal communication). On the basis of excavations carried out in 1968, 1977 and 1984 (Hulst 1968; Hulst and Noordam 1977; Hulst et al. 1984) and a few observations (Hulst 1969a; 1969b; 1971b) it is clear that we are concerned here with a long, narrow, continuous settlement measuring about 400 x 100 m to the south of the Nedereindsestraat in present-day Kesteren. The settlement is situated on the high south bank of an old river channel, that in Roman times already was largely silted up, though it may have become inundated periodically. This old channel is situated on the edge of a young meander loop within which the Rhine flowed in Roman times. The archeological features within the settlement consist of ditches and rubbish pits. The finds indicate that occupation here was most extensive from about AD 70 until the second half of the 2nd century, after which it diminished considerably. It is very likely that the settlement was a vicus, a village situated on the main road, that was inhabited by traders, craftsmen and innkeepers; thus it would have had no agrarian function. The finds from the settlement and from the larger of the two cemeteries are suggestive of the presence of a military unit; consequently it is thought that the settlement may well have been a military vicus belonging to an (as yet) undiscovered castellum (Bogaers and Rüger 1974: 70-71; Van Es 1981: 103-104). This could possibly have been the Carvo indicated on the Tabula Peutingeriana. Here below the animal bones that were found during the excavations of 1977 will be discussed. The finds from the 18 graves of horses that were found in the cemetery on 'De Prinsenhof (Hulst 1975) have been partly discussed by Prummel (1979b). 3.5.2 THE BONES An overview of the hand-collected material from the settlement at Kesteren is given in tables 29 and m7. The measurement data are given in table mi8. Of the 387 animal bones found 70.3% (by weight 88.8%) could be identified to species level. The fragments that could not be identified as to species are mostly of ribs, vertebrae, and indeterminate skeletal parts (table m7). If we compare the bones of the two size classes represented then we see a very significant difference between the identified and the unidentified bones (/^ = 46.142, df= i, p<o.ooi). The small animals are underrepresented among the unidentified bones. In table m3i the fragments that could not be identified as to species but that could be assigned to a size class are distributed over the species belonging to those classes. The distribution is based on the percentage distribution of the identified species within a size class. 88 TABLE 29 Resteren: the hand-collected material. Frequencies and weights (g). species domestic mammals cattle sheep or goat pig horse dog mollusca Ostrea edulis Buccinum undatum total identified unidentified mammals no size assignment sheep-pig size cattle-horse size oyster common whelk number /o weight /o 198 72.8 10433 77.6 43 6 15.8 2.2 630 4-7 1.0 21 7.7 2186 I 0.4 18 16.3 0.1 2 0.7 0.4 21 0.2 I 13 0.1 272 70.3 13440 88.8 139 I I.O 13.9 6 72 0.4 16 98 85.2 1614 95-4 total unidentified "5 29.7 1692 II.2 Total 387 4-3 15132 The data for the identified and the allocated unidentified material are summarized in table 30. The shifts that result from the allocation are very slight. The maximum shift in frequency percentage is for cattle: 1.6%. For sheep/goat the shift is 1.0%. The other percentage shifts, both in terms of frequency and weight, are at most 0.2%. Among the meat-providing animals cattle is the most important species, followed by sheep/goat and pig with values of 74.4%, 14.8% and 2.1%, respectively (in terms of weight: 77.8%, 4.5% and 1.0%). Horse scores 7.8% (16.3% by weight). One fragment of dog was found. The marine component is represented by two fragments of oyster and one of whelk. TABLE 30 species domestic mammals cattle sheep/goat pig horse dog mollusca Resteren: summary of the identified and the allocated unidentified hand-collected bone material (weight in g). weight number 287 57 8 30 I II768 77.8 2.1 688 152 4-5 1.0 16.3 0.1 74-4 14.8 7.8 2465 0.3 20 0.8 34 89 Heteren I and II: the hand-collected material. Frequencies and weights (g). TABLE 31 Heteren I species domestic mammals cattle sheep sheep or goat Heteren II number 0/ /o weight % number 0/ /o weight 0/ /o 175 634 9324 61.7 131 62.1 4507 54-2 I 0.4 100 0.7 - - - 27 9.8 230 1-5 31 14-7 327 3-9 pig II 4.0 303 2.0 4 1-9 25 0.3 horse dog 35 12.7 3488 23.1 42 19.9 3191 38.4 2 0.7 39 0.3 I 0.5 46 0.6 wild mammals Bos primigenius Cervus elaphus total identified - aurochs - red deer I 0.4 121 0.8 - - - - 24 8.7 1496 9-9 2 0.9 221 2.7 276 84.1 15101 954 211 77.6 8317 91.9 unidentified mammals sheep-pig size cattle-horse size 10 19.2 43 5-9 12 19.7 45 6.1 42 80.8 692 94.1 49 80.3 689 93-9 total unidentified 52 15-9 735 4.6 61 22.4 734 8.1 Total 328 15836 272 9051 3.6 NATIVE FARMSTEADS 3.6.1 THE NATIVE SETTLEMENT IN HETEREN 3.6.1.1 Introduction In Heteren, in 1968, 1969 and 1970, parts of a native settlement were excavated (Hulst 1969a; 1971a). The settlement was situated on a natural elevation on a stretch of fossil levee soil. The excavation revealed a complex of ditches in a rectangular pattern (ca 100 m wide), within which traces of occupation were evident. These traces consist of an enormous number of post-holes, among which it is possible to discern the presence of a farmhouse and a number of small outhouses (Van Es 1981: 169). The outhouses could indicate that the settlement mainly consisted of storage units for agricultural products. In addition, immediately beyond the western end of the settlement ditches were present that presumably served as parcel boundaries of fields or pastures. The bone material has been divided into two groups on the basis of the pottery finds. The first group comes from a context including both native pottery and imported Roman ware. The Roman ware indicates a dating from the middle of the ist century until the middle of the 2nd century. In the text this group is designated 'Heteren 11'. The bones belonging to the second group were found only in places where native pottery may possibly be associated with an older occupation. This group is designated 'Heteren I'. 3.6.1.2 The bones Tables 31 and mS present an overview of the hand-collected material of Heteren I and II. The measurement data are presented in table mi9 and m20. About 81% 90 TABLE 32 Heteren I and II: summary of the identified and the allocated unidentified hand-collected bone material (weight in g). Heteren II Heteren I species domestic mammals cattle sheep and sheep/goat pig horse dog wild mammals aurochs red deer 206 35 14 41 2 I 28 number weight number 62.0 10.7 9771 351 4-3 12.5 0.6 3655 41 0.3 8.6 322 127 1568 61.7 2.2 2.0 23.1 0.3 168 41 5 54 I 61.8 15.1 1.8 19.9 0.4 weight /o 4899 364 28 3469 51 54-1 4.0 0.3 38.3 0.6 240 2.7 0.8 9-9 of the total number of bones (94% by weight) could be identified to species level. The unidentifiable fragments largely consist of pieces of ribs, vertebrae and indeterminate skeletal parts (table m8). There are no significant differences between the ratio large: small animals in the identified material and the unidentified material at the p = 0.05 level (Heteren I: f- = 0.346, df = i, Heteren II: /^ = 0.080, df= i). The only reason whey the unidentified bones of the two size classes have been allocated (table m32) is because this same procedure has been applied to the other find complexes. The data for the identified and the allocated unidentified material are summarized in table 32. As one would expect, the percentage shifts with respect to the identified material of table 31 are minimal. On the basis of table 31 a very significant difference can be seen between the species composition of Heteren I and II (;(:^ = 27.692, df = 5, p<o.ooi). The wild species are combined in one group for the purpose of testing. The differences consist mainly in the larger proportion of horse in Heteren II and the much greater proportion of wild animals in the material of Heteren I. Among the meat-yielding species, cattle, sheep/goat, pig, aurochs and red deer, the proportion of wild animals as compared to domesticated animals is very significantly greater in Heteren I than in Heteren II [f- = i4-593> df = i, p<o.ooi). Whether wild animals were actually more important as a source of meat in Heteren I than in Heteren II is doubtful. Of the 24 fragments identified as red deer, 23 were parts of antlers. Of the two fragments from Heteren II, one is a piece of antler. Moreover there are indications that the antlers were collected separately and that they did not come from animals that had been hunted. The only piece of antler from Heteren I on which the base is present was certainly a shed antler. In addition, on an other fragment of antler chisel-shaped gnaw marks are present. Such gnaw marks are indicative of a red deer having gnawed a shed antler. Antlers were evidently used as a raw material for making particular objects. Traces of chopping are present on three fragments from Heteren I (mark 7 - see appendix). On the only fragment from Heteren II chop marks are present (mark 7) as well as a mark of careful sawing (mark 4). There is no significant difference between Heteren I and Heteren II with regard to the proportions of the domesticated meat-producing animals, cattle, sheep/goat and pig (/^ = 5.053, df = 2 p > 0.05). If we consider Heteren I and Heteren II jointly, then 62.4% of the material comes from cattle (58.9% by weight). Horse 91 TABLE 33 Ewijk I and II: the hand-collected material. Frequencies and weights (g). Ewijk I species domestic mammals cattle sheep sheep or goat pig horse dog wild mammals Bos primigenius Cervus elaphus 0/ number Ewijk II* /o weight % number % weight % 837 76.0 777 66.5 29203 65.3 O.I 0.1 I 0.1 9-5 3-7 9.3 1-3 2.0 58 40 274 5.0 3-4 23.4 15 1-3 95 662 785 13692 136 0.2 105 48555 40 1178 791 8838 80.7 I 1-5 r.8 30.6 0.3 2 0.2 100 0.2 2 0.2 43 0.1 41 102 14 334 1-3 14-7 0.6 - - 2 0.2 405 0.7 1102 71.6 60141 90.6 1169 63.1 44716 87.2 unidentified mammals sheep-pig size cattle-horse size 31 407 7-1 92.9 200 3-2 96.8 22 6036 663 3-2 96.8 76 6507 98.8 total unidentified 438 28.4 6236 9-4 685 36.9 6583 12.8 total identified Total - aurochs - red deer 1540 66377 1854 1.2 51299 * The bones from the two horse graves are not included. scores 15.9% (28.6% by weight), followed by the domesticated animals sheep/goat with 12.7% (3.7% by weight), pig with 3.2% (1.4% by weight) and dog with 0.5% (0.4% by weight). Of the wild animal species red deer is the most important: 5.2% (7.3% by weight). Only one fragment of aurochs was found, in Heteren I. A bone of the sheep/goat group from Heteren I could be identified positively as sheep. 3.6.2 THE NATIVE SETTLEMENT IN EWIJK 3.6.2.1 Introduction In 1973, on a natural elevation within the terrain known as 'De Woerdjes' in Ewijk, a settlement was found dating from the pre-Roman Iron Age (Hulst et al. 1973). In 1974, on the east flank of the elevation also traces of occupation were found, in this case dating from the Roman period (Hulst and Noordam 1974). The oldest of these traces of occupation may be contemporaneous with the settlement on De Woerdjes. The youngest occupation is represented by a ground-plan of a native farmhouse dating from the middle until the last quarter of the 2nd century AD. A second house ground-plan cannot be dated any more precisely than in the Roman period, but it probably also dates from the 2nd century. Outside the inhabited area the terrain is divided by means of a system of ditches, probably dating from the Roman period. Like Heteren, Ewijk could have been a kind of storage depot for agricultural products, possibly associated with a rural estate (van Es 1981: 230). The bone material from the excavation of 1974 has been investigated. This 92 TABLE 34 Ewijk I and II: summary of the identified and the allocated unidentified hand-collected bone material (weight in g). Ewijk I species domestic mammals cattle sheep and sheep/goat pig horse dog wild mammals aurochs red deer Ewijk II* number /o weight % number % weight % 1199 77-9 53625 80.8 1265 68.2 33621 65.5 126 8.2 3-2 9-5 1322 2.0 859 1-3 3.8 2.6 791 821 1-5 1.6 9762 14.7 24.1 15761 30.7 I.I 363 0.5 70 48 446 18 I.O 142 0.3 _ _ _ _ 0.2 3 447 0.7 113 50 0.2 0.2 49 146 17 3 3 0.2 0.1 * The bones from the two horse graves are not included. material has been divided into two groups. Bones belonging to the first group, that were found in places where only native pottery was present, is designated 'Ewijk r. Bones belonging to the second group were found in places were both native and Roman pottery were present; this material is designated 'Ewijk 11'. Francisca Zeiler, a student of archeology, assisted in the identification of the bone material from Ewijk. 3.6.2.2 The bones Tables 33 and mg give an overview of the hand-collected material from Ewijk. The percentage of identifiable material from Ewijk I (71.6%; 90.6% by weight) is somewhat higher than that from Ewijk II (63.1%; 87.2% by weight). The unidentifiable fragments are mostly parts of ribs, vertebrae and indeterminate skeletal parts. In Ewijk II there is also a considerable proportion of skull fragments in the size range of the larger animals (table mg). Both in Ewijk I en Ewijk II there is a very significant difference in the ratio of large to small animals for the identified and the unidentified material (Ewijk I: x^— i5-6i4, df = i, p<o.ooi; Ewijk II: /^ = 26.229, df= i, p<o.ooi). In table m33 the fragments that could not be identified to species level but that could be assigned to a size class are distributed over the species included in those classes. The distribution is based on the percentage distribution of the identified species within a size class. In table 34 the data for the identified and the allocated unidentified material are summarized. For the frequency percentages the percentage shift with respect to the identified material is at most 1.9%, namely for cattle in Ewijk I. For the weight percentages the maximum shift is only 0.2%. There is a very significant difference in the species composition of Ewijk I and Ewijk II (x^= 143.163, df = 5, p<o.ooi). Here the wild animals are considered jointly in one class. The same applies if we look at the domesticated meat-providing animals, cattle, sheep/goat and pig (x^ = 17.749, df = 2, p < o.ooi). The most important difference between Ewijk I and II with respect to the domesticated meat-providing animals is the much greater proportion of sheep/goat at the expense of cattle in Ewijk I. With the other animals a large difference is also evident especially in the case of horse. In Ewijk II the proportion of horse is considerably greater than in Ewijk I. Wild animals are of 93 Fig. 25 Ewijk: extremities, thoracic and pelvic girdle of horse from the horse grave Ewijk 10.10. Parts that were retrieved are shown hatched; the dotted lines indicate the fractures. A forelegs, front view; B hind legs, rear view. B little importance in both periods. Red deer occurs in both phases, aurochs only in Ewijk II. In addition to the material described above, two graves of horses were found in Ewijk II: Ewijk 10.8 and 10.10 (table 35). The skeletons lay in the soil in an articulated state. The two skeletons indicate how poor the conditions for preservation were for a large proportion of the bones from Ewijk. Although in the case of articulated skeletons we can be sure that they were once present in a complete state in the soil, those at Ewijk could be recovered only partially. 94 TABLE 35 Ewijk, horse graves: frequency and weights (g) per skeletal element. The frequency indicates the number of different specimens of a skeletal element of which at least one fragment was found. horse no. number cranium mandibula scapula humérus radius ulna carpalia/tarsalia metacarpus III pelvis femur tibia astragalus metatarsus III phalanges indeterminate vertebra unknown fragments Ewijk 10 10 Ewijk 10.8 — - weight - number I weight 1033 2 100 2 2 359 5 530 55 48 13 770 561 104 124 2 170 2 211 - 468 465 58 2 539 2 1258 893 136 I 2 I 2 2 2 2 6 I* 8* 117 263 198 23 40 2 2 2 2 - - 6 59* 190 581 20* 134 * number of fragments. Undoubtedly some fragments became lost during the hand-collecting procedure, despite the fact that collecting was carried out rather carefully (some of the bones were recovered with the surrounding soil attached). Fig. 25 shows to what extent the fragile bone material became fragmented in the course of time. Some of the fractures evidently occurred long ago in the soil, while some have occurred recently when the bones were being - rather carefully - recovered and sorted. The problems involving finds of skeletons of buried animals are discussed in more detail in section 3.7.3. 3.7 THE VILLA IN DRUTEN 3.7.1 INTRODUCTION In the years 1975-1978 a Roman villa was excavated in Druten on the terrain known as Klepperhei (Hulst 1978; 1980). The settlement was situated on the southern levee of the river Waal, 18 km west of Roman Nijmegen. The settlement consists of different buildings that are grouped around a large rectangular courtyard (fig. 26). Together the buildings form a large farmstead. On the west side was the residential part of the settlement, the pars urbana. To the east of this, separated by a fence or colonnade, was the pars rustica with the various farm buildings. Although the villa shows distinct Romanized features, it is essentially of native origin. The building of the villa must have started just after AD 70. The occupation continues until into the 3rd century. The bone material discussed here below has been divided into three parts (personal communication R.S. Hulst, ROB). Material originating from the first phase of development of the villa at the end of the ist century and the beginning of the 2nd century (dating Id-IIa) is designated 'Druten II'. The bone material from approximately the middle until the end of the 2nd century (about Ilb-IId) is called 'Druten III'. A group of bones found in a context where only native pottery is present may date from the pre-Roman Iron Age but may also be partly 95 Fig. 26 Druten: excavation plan showing the places where bone material has been found. The buildings are indicated with the numbers 1-22. The features containing bones are indicated by a find number consisting of two parts (see text). Black: pestholes, shaded: stone foundations, hatched: foundation ditches, circles: refuse pits, circular shape with hatching: well. Scale 1:1000 (after Hulst 1978, fig. 4). 96 contemporaneous with the beginning of the Druten II period (personal communication R.S. Hulst, ROB, 1985). This group is designated 'Druten I' Hulst (1978) gives as a reason for the origin of the settlement as a villa, in addition to its favourable situation for agrarian production, the presence of a surrounding region where there was a demand for agricultural products. The founding of the villa in the Druten II period may have been prompted by the stationing of Legio X Gemina in Nijmegen. In the Druten III period the villa would have had a market for its products especially in the central place of the civitas Batavorum, Ulpia Noviomagus. 3.7-2 THE BONES An overview of the faunal material that was found in Druten is given in tables 36 and mio. Not included in table 36 are the bones from the four horse graves from Druten II that will be discussed separately (see 3.7.4). The measurement data, including those for the horses, are given in table m2i, m22 and m23. The percentage of identifiable fragments varies from 78.4% in Druten II to 84.0 in Druten I (in terms of weight from 94.5"/, in Druten II to 96.6% in Druten I). The unidentifiable fragments are mostly parts of the skull, the ribs, vertebrae and indeterminate skeletal parts (table mio). In Druten I and III there are no significant differences at the p = 0.05 level in the ratio large:small animals between the identified and the unidentified material (Druten I: /^ = 0.416, df= i; Druten III: x^ = 0.092, df = i). For the material from Druten II there is a very slightly significant difference (/^ = 4.418, df = i, p < 0.05). The small animals are overrepresented among the unidentified material, this being a deviation from the pattern found elsewhere in the region. In table m34 the fragments that could not be identified to species level but that could be assigned to a size class are distributed over the species included in that class. The distribution is based on the percentage distribution of the identified species within the size classes. Table 37 presents a summary of the identified material and the allocated unidentified material. Distinct differences are evident between the three groups of bones from Druten. If we consider only the domesticated meat-producing animals, cattle, sheep/goat and pig, then there is a slightly significant difference between Druten I and II 0^^ = 7.141, df=2, p<o.o5) and there are very significant differences between Druten I and III (j^ = 25.842, df = 2, p < o.ooi) and between Druten II and III (/^ = 42.085, df=2, p<o.ooi). The proportion of cattle among the domesticated meat-providing animals increases after Druten II (table 38). In Druten I and II the percentage for cattle was found to be 66.4% and 65.1% respectively (by weight 81.2% and 82.0%), whereas in Druten III this value increases to 80.6% (92.7% by weight). The percentage for sheep/goat decreases from 24.870 in Druten I to 11.9% in Druten III (a decrease from 10.1% to 3% by weight). The highest percentage for pig (15.2%; by weight 9.5%) occurs in Druten II. In all three Druten phases considerable quantities of horse bones were found, as well as a few bones of dog. In Druten too wild animals are of little importance. In Druten I red deer is found, in Druten II red deer and wild boar, and in DrutenUI aurochs and roe deer. The proportion of wild animals appears to increase slightly in the course of time. However, the numbers of bones are too small to permit any firm conclusions about this. In any case the diversity of wild animals increases with the passage of time in the excavated material. The presence of domestic fowl has been ascertained only in Druten I, while the wild or domesticated forms of the greylag goose and wild or domestic duck occur in Druten III and Druten II respectively. In Druten II and III three different bird species are represented, each by one bone: in Druten II the crane and the long-eared owl, and in Druten III the cormorant. Remains of oysters were found in Druten II and III. 3.7.3 THE RELATIONSHIP BETWEEN BONES, BUILDINGS AND OCCUPATION Immediately below the bone data are analyzed in relation to the archeological features and in particular to the various buildings. Only identified bones from 97 TABLE 36 Druten I, II and III: the hand-collected material. Frequencies and weights (g). Druten I species number domestic mammals cattle sheep sheep or goat pig horse dog 0/ Druten II* /o weight % number 134 55-4 7230 53-9 290 - - - - I 48 17 36 4 19.8 7.0 14.9 883 764 4400 61 6.6 5-7 32.8 0.5 78 61 15.6 12.2 54 3 domestic birds domestic fowl 1-7 0.8 weight % 57-9 13796 63.7 0.2 19 1372 0.1 10.8 0.6 1539 4517 29 7-1 20.9 4 0.8 213 I.O I 0.2 22 0.1 0/ 6.3 0.1 0.1 wild or domestic birds Anser anser Anas platyrhynchos greylag goose mallard wild mammals Bos primigenius Cervus elaphus Capreolus capreolus Sus scrofa aurochs red deer roe deer wild boar wild birds Grus grus Asia otus Phalacrocorax carbo crane ling-eared owl cormorant I 0.2 5 0.0 I 0.2 2 0.0 moUusca Ostrea edulis oyster 6 1.2 147 0.7 total identified I 0.4 71 0.5 242 84.0 13416 96.5 501 78.4 21663 94-5 unidentified mammals no size assignment sheep-pig size cattle-horse size 16 30 34-8 65.2 87 395 18.0 82.0 54 84 39-1 60.9 268 21.2 999 78.8 total unidentified 46 16.0 482 3-5 138 21.6 1267 5-5 Total 288 13898 639 22930 The bones from the four horse graves are not included. pits, ditches and wells are considered. Faunal material from post-holes and foundation trenches, accounting for 14% in Druten II and 9% in Druten III, is disregarded. This has been done because this material may have been dug up from layers of soil dating from earlier periods. In the case of associated find numbers, one of the numbers is mentioned consistently, namely the lowest. 3.7.3.1 Druten II Druten II is represented by the buildings i, 6, 8, 9, 10, 11 and 20, and by the pit or sunken hut 16 (fig. 26). The animal remains that have been found in or around these objects are presented in the tables 39-43. Next to building i, in two pits, large quantities of horse bones have been found 98 Dru ten III number 522 10 68 49 335 10 0, /o weight % 50.1 I.O 57370 444 1414 2672 38590 55-1 0.4 190 0.2 6.5 4-7 32.1 1.0 1-4 2.6 37-1 0.0 4 7 0.4 1.6 1402 1-3 1405 I 0.1 1-3 0.0 I 0.1 II 106 0.1 0.1 23 2.2 516 0.5 1042 81.3 104I34 96.3 2 0.8 15 30 207 12.6 142 86.6 3829 0.4 3-6 96.1 239 18.7 3986 3-7 (table 39). These mainly come from two horse graves, possibly foundation deposits, that will be discussed in section 3.7.4. The other finds mostly come from two ditches west of the building. Building i shows the greatest diversity of meat-providing species. It is the only place where large game species, red deer and wild boar, are found. Also oysters and wild or tame duck occur only here. This diversity of species, including wild mammals, birds and oysters as well as the horses in the graves, fits in well with the picture of the special status of the occupants of the main building of the settlement. The relatively high percentage of pig may be regarded as a Roman or military influence (see 4.1.2). Also building 11 has a high percentage of horse (table 40). As with building i, this is accounted for by two horse graves (see 3.7.4)- All other finds come from a pit next to the southwestern corner of the building. Building 11 comes in second place as regards the diversity of species that were possibly a source of food. In addition to cattle, sheep/goat and pig, the crane and the long-eared owl have been found. These species were probably eaten (see 4.4). On the basis of the presence of the horse graves and the bird species, one could deduce from this bone material that building 11 was the second most important building as regards status. The bone finds in and around the buildings 8, 9 and 10 mainly come from a few pits and a well to the north of building 10, and from a pit to the south of building 9 (table 41). Pit 44.2 contained a large quantity of cattle bones of which 87% were fragments of vertebrae and ribs. With building 20 (table 42) the high percentage of sheep/goat is conspicuous. This high percentage of sheep/goat is mainly accounted for by pit 66.1 to the south of the building. A second pit containing many finds was situated north of the western part of the farmstead. In this pit also several skeletal parts of a dog were present. In the ditch of the southern foundation trench (25.14), not included in the table, a piece of worked antler of red deer was found. -pjjg f^^ ßj^^g fj.Qj^ j]^g pjj Q^ sunken hut 16 present the same picture as the material of building 20 (table 43). With building 6 only a few fragments were found in a post-hole and in a foundation trench. 3.7.3.2 Druten III Druten III is represented by the buildings i, 2, 3, 4, 12, 13, 14, 15, 17, 18, 19 and 22 and probably also by the buildings 5 and 7 (fig. 26). The animal remains that were found in or around these buildings are presented in tables 44-49. In contrast to the situation in Druten II, in Druten III only a few bones were found that are possibly associated with building i (table 44). The bones of cattle, sheep/goat, pig, red deer and horse were all found in the ditch to the south of the building. The species diversity of the animals that were a source of food is considerably lower than in Druten II, where also oyster, wild boar and duck were found. The finds from the vicinity of the bath house, no. 2, come from the well (table 45). In addition to cattle, sheep/goat and red deer the relatively large number of oyster shells is striking, finds that fit in well into a context of luxury that one associates with a bath house. If we assume that the bath house belongs to building i, then we can conclude that the species diversity for the meat-providing animals in this complex differs less from Druten II than is indicated by the data of building i only. However, in view of the small number of finds from Druten III, any comparison of Druten II and III is of limited value. The bone material from the vicinity of building 3 mostly comes from the northern ditch (table 46). In this ditch large quantities of bones of cattle and horse were found. Cattle bones in pit 15.2 included the various parts of a right 99 TABLE 37 Druten I, II and III: summary of the identified and the allocated unidentified hand-collected bone material (weight g). Druten I species number domestic mammals cattle sheep and sheep/goat pig horse dog Druten II* % weight % number % weight % 158 54-9 7474 53-8 360 63.4 20.5 928 6.7 109 7-3 803 5.8 42 14.6 4549 32.7 10.5 4761 20.8 5 84 67 4 1518 1679 6.6 21 56.3 17.0 13.1 14540 59 0.6 32 O.I 0.8 224 I.o I 0.2 24 0.1 3 0.5 9 0.0 6 0.9 147 0.6 1-7 64 0.5 wild mammals aurochs red deer roe deer wild boar 0.3 73 0.5 birds 0.7 mollusca 7-3 * The bones from the four horse graves are not included. TABLE 38 Druten I, II and III: percentage distribution of the domesticated meat-providing animals per period. frequency percentages period: cattle sheep/goat pig TABLE 39 feature ditch pit pit pit pit well ditch pit? total % without horse * percentage of total 100 weight percentages II III 66.4 24.8 65.1 19.7 80.6 11.9 8.8 15.2 7-5 II III 81.2 82.0 92.7 10.1 8.6 3.0 8.7 9-5 4-3 Animal remains found in and near building i (Druten II) (numbers). find no. 1-9 1.13 cattle sheep/ goat 33 I 8 - pig 13 I 2 — - 1.15 2 4 2 1.17 - - - 1.18 red deer wild boar mallard _ — - _ _ _ _ — 1.21 5 2 I 1.33 23 4 10 2 I I 1-35 - 2 I - - - 64 20 29 I 16 23 3 2 I 52 I I oyster horse 27 152 I 2 186 (60)* Druten III 0/ /o weight % 50.5 7-4 4-7 32.4 0.9 59595 1913 2750 40087 196 55-1 1.8 21 0.3 1.6 1456 1459 1-3 1-3 I 0.1 II 0.0 I 0.1 109 0.1 2 0.2 14 0.0 23 2.2 516 0.5 number 645 95 60 414 12 5 2.5 37-1 0.2 rear leg, that evidently belonged together. The finds from the well also include some bones that belong together, for example among the cattle bones a more or less complete skull with the associated cervical vertebrae. The bone finds in the vicinity of building 12 come from the well on the southwestern side and the ditch on the north side (table 47). A conspicuous feature is the species diversity of game animals in the well: red deer, aurochs and wild boar. This is the only place in Druten III where three species of large game animals occur. The finds in and around building 14 that has been dated in the last period of Druten III are shown in table 48. The cattle bones from well 10.4 probably all come from one animal. The bones found include parts of both forelegs, hind legs, both halves of the pelvis, the shoulder blades, and a few vertebrae and ribs. On the basis of the state of fusion of the epiphyses, the animal was 3.5-4 years old. In view of the presence of chop marks and cut marks on the bones, this animal must have been butchered for human consumption. The butchery marks are: scapula 5, 30 (x2); humérus 12, 13, 24, 31; pelvis i, 2, 3, 4, 5; femur 35 (see appendix Butchery mark code). In the eastern part of the settlement, the pars rustica, a large proportion of the faunal material cannot be associated with any degree of certainty with particular buildings. Some of the finds were also located in the central yard. For this reason these bone finds are considered jointly for the most part. TABLE 40 Animal remains found in and near building 11 (Druten II) (numbers). feature pit pit pit find no. cattle sheep/ goat pig 29 3 6 29 73 3 8 6 15 find no. cattle sheep/ goat pig 39.10 2 3913 44.2 I - - 2 I I I I I I - 5 5 - - - I 16 6 2 2 I - 5 6 7 4 2 3 98 80 16 13 8 7 (II)* 12.2 12.4 13.2 total % without horse owl horse 136 31 I I 168 (81)* * percentage of total. TABLE 41 Animal remains found in and near the buildings 8, 9 and 10 (Druten II) (numbers). feature pit pit pit pit pit pit ditch pit pit well 54-14 56.7 60.3 60.6 61.6 61.12 61.13 total % without horse 55 3 2 2 horse - 15 percentage of total. lOI TABLE 42 Animal remains found in and near building 20 (Druten II) (numbers). feature find no. pit pit pit ditch pit field field pit 25-5 25.8 25.9 27.8 32.4 65.2 65.3 66.1 total % without horse and dog cattle sheep/ goat 2 I - - 2 18 4 2 4 17 pig horse — - I 3 I 6 3 - I I - - 17 6 II - . 21 8 27 ID 19 (19)* cattle sheep/ goat pig horse 63 26 49 63 dog 3 (3)* * percentage of total. TABLE 43 Animal remains from pit/sunken hut 16 (Druten II) (numbers). feature find no. pit 28.2 % without horse (17)" percentage of total. TABLE 44 Animal remains found in the vicinity of building i (Druten III) (numbers). feature find no. cattle sheep/ goat pig red deer ditch I.I9 15 2 I I 79 II 5 5 (39)* horse % without horse horse 12 percentage of total. TABLE 45 Animal remains found in the vicinity of building 2 (Druten III) (numbers). feature find no. cattle sheep/ goat red deer oyster well 24.2 19 7 3 13 17 45 17 7 31 (29)* % without horse percentage of total. 102 TABLE 46 Animals remains found in and near building 3 (Druten III) (numbers). find no. feature pit ditch pit 3-4 3-5 3.6 3.8 5-4 7-3 7-4 7.6 15.1 15.2 15-3 pit well pit pit ditch ditch pit pit total % without horse and dog cattle _ sheep/ goat pig I - — - I I 3 3 I 62 — — ~ " ~ 5 5 5 5 I 80 I (40)* wild boar aurochs horse I I 3 3 7 I 2 3 6 (16)* I I 2 I 2 64 7 3 TOO 90 dog - 14 horse — - - 6 red deer - I I 7 — 2 6 2 — — — — 5 I 7 (4)* * percentage of total. An imai remanis louna in trle vicinity 31 Duiiaing vi l^i^ruic TABLE 47 feature ditch well total % without horse find no. 10.2 19.4 cattle sheep/ goat pig red deer I 6 30 3 6 36 69 4 8 6 2 12 4 I 10 percentage of total. In addition to the domesticated farm animals cattle, sheep/goat, pig, horse and dog, in the eastern part also large game animals, birds and oysters have been found (table 49). Once again horse is found everywhere in considerable quantities, 14%, but especially in the ditches to the northeast and southwest of the buildings 18 and 19. The filling of this last-mentioned ditch probably derives from building 19. Almost half of the material comes from this ditch. Among the domesticated meat-providing mammals sheep/goat clearly dominates over pig. A conspicuous feature is the occurrence of oyster shells in a pit in the entrance to the central yard (45.14) and in two pits to the north and south of building 22 (54.6 and 39.2). The pit in the entrance contains, in addition to bones of domesticated farm animals and oyster shells, a fragment of a humérus of red deer and a piece of antler of a roe deer. The presence of this single remnant of roe deer in Druten does not necessarily mean that roe deer were a source of meat in Druten, seeing that this fragment may have come from a shed antler. The two bird bones, one of a cormorant and the other of a greylag goose, come from the pit in the central yard (42.4) and the ditch northeast of the buildings 18 and 19. 103 TABLE 48 Animal remains found in and near building 14 (Druten III) (numbers). feature find no. pit well well 6.2 8.3 10.4 total % without horse cattle sheep/ goat pig _ _ _ 4 30 I I _ _ I 3 I _ - 34 83 2 4 8 I 2 (II)* 5 red deer horse I 5 percentage of total. The only aurochs bone comes from the ditch southwest of building 19. Besides the general differences between Druten II and III that are discussed in section 3.7.2, there are also differences in terms of the distribution of the species. In Druten II there is a distinct difference in the nature of the bones found between the pars urbana and the pars rustica. Large game animals, birds and oysters were found only in the pars urbana, mainly around building i. In the pars rustica only the meat-providing animals cattle, sheep/goat and pig were found. In the pars urbana pig predominates over the probably less relished sheep/goat (for the pars urbana as a whole 60% as against 40%). In the pars rustica sheep/goat is dominant over pig everywhere (for the pars rustica as a whole 70% as against 30<;ó). Horse graves occur only within or very close to the two buildings in the pars urbana. All three of these differences can be associated with a higher status of the occupants of the pars urbana. These clear differences are not in evidence in Druten III. In this phase game animals and oysters are found in both parts of the settlement. In both parts sheep/goat predominates over pig, although the percentage of pig within the sheep/goat-pig group in the pars urbana is somewhat higher than in the pars rustica (44% and 37% respectively). Special features such as horse graves are absent from Druten III. On the basis of the bone material from Druten II and III it can be concluded that the distinct difference in status in Druten II between the two parts of the settlement is no longer discernible in Druten III. If we presume that game animals, birds, oysters and high percentages of pig are indicative of high social status, then we can say that in Druten Ilia certain levelling out has taken place of the status of the occupants of the settlement. 3.7.4 FOUNDATION DEPOSITS CONSISTING OF ANIMAL SACRIFICES 3.7.4.1 The circumstances, oj the finds In two places within or just outside ground-plans of houses, pits containing horse bones were found (Hulst 1978: 136,141). In view of their situation these could possibly be interpreted as foundation deposits. Two rectangular pits, both lying along the same longitudinal axis, were situated at right angles to the north wall of building 11 (fig. 27B). The southern pit, with find numbers 12.4 and 12.9, designated 'pit 12.4', was situated in the porticus right next to the entrance. The pit is immediately adjacent to the foundation trench of building 11 and certainly must be associated with this building (personal communication R.S. Hulst,ROB). The northern pit, with find numbers 12.2, 12.3 and 12.8, designated 'pit 12.2', was situated just outside the porticus. 104 Animal remains found in the pars urbana (Druten III) (numbers). TABLE 49 feature pit ditch well well pit pit pit pit pit ditch pit ditch pit pit pit pit pit ditch ditch pit pit pit pit ditch pit pit find no. 25.10 25.11 26.14 31.6 32.16 3314 34-8 35-1 37-4 37-5 37-6 37.7 39.2 39-3 19.17 42.2 42.4 42.17 44.8 44.19 45-3 45.14 45-17 46.9 46.10 54.6 total 'V,, without horse and dog cattle I 103 9 2 I 3 I sheep/ goat 2 I 12 4 I I I - roe deer aurochs goose cormorant oyster horse 3 2 I 41 2 I 7 I 6 2 2 47 5 2 2 I - I I 261 44 74 12 dog 88 2 - I red deer _ 18 6 - 39 3 3 5 16 pig 21 17 4 26 7 10 3 200 (36)* 3 (I)* * percentage of total. This pit 12.2 probably also belongs to building 11, although it is not impossible that it belongs to building 12 (personal communication R.S. Hulst, ROB). In both cases the situation with respect to the buildings is the same. On the inner side of the east wall of building i there was also a pit with horse remains (fig. 27A, pit 1.17). This pit is more or less parallel to the wall in the middle of the main building. A second pit with horse remains, not mentioned in the literature referred to above, is situated at the same level with the same orientation in the porticus of building i (fig. 27A, pit 1.18). As the house ground-plan shows no indications of the site of the entrance, it is assumed, by analogy with the situation with building 11, that the entrance was situated somewhere near pit 1.17 (Hulst 1978: 136). On the basis of the archeological context the pits are dated to the end of the ist century AD (Druten II). If pit 12.2 does indeed belong to building 12 then this pit must date from the 2nd century (Druten III). 3.J.4.2 The horse burials: whole or half skeletons? This section is concerned with the following questions: Do the four pits contain the remains of four different horses? Were whole animals buried or only parts of 105 Fig. 27 Druten: situation of the foundation deposits consisting of horse sacrifices. A building i, B building 11. i. 17, 1.18, 12.2 and 12.4: pits containing horse skelettons. t • - I • % 10m I 10m them, and what conclusions can be drawn from this? What may have got lost during the excavations and what consequences does this have for the conclusions mentioned in the second question? The point of departure for answering these questions is that although no complete skeletons were exposed during the excavation, when the bones were removed from the soil it was observed that they were in their normal anatomical juxtaposition (personal communication R.S. Hulst, ROB). The first question can be answered in the affirmative on the basis of the skeletal elements that were found in the four separate pits. With the bone material found, only one skeleton can be reconstructed for each pit. The possibility that in two pits the remains of one and the same horse are present can be excluded because the different pits contain identical skeletal elements. To find an answer to the question whether complete horses were buried or only parts thereof, I have made use of schematic drawings of a horse flattened out, with splayed limbs, as shown in fig. 28. Seeing that the bones were observed in their normal anatomical position, I interpret the find of a metacarpal, for example, as 'forefoot', and that of, say, a tibia as 'hind leg'. If we now assume that: a) while the excavation was in progress, when for example a foot was present at the time of the excavation, of this foot at least one bone fragment was collected; b) there was never any possibility for dogs to have interfered with the material by dragging bits away; 106 Fig. 28 Druten: schematic drawings of the horses found in the pits 1.17, 1.18, 12.2, 12.4. In the drawings the horses are shown flattened out with splayed hmbs. I skull and mandibulae, 2 shoulder and cervical vertebrae, 3 thoracic vertebrae and ribs, 4 other vertebrae and pelvis, 5 left foreleg, 6 left forefoot, 7 left hind leg, 8 left hind foot, 9 right foreleg, 10 right forefoot, II right hind leg, 12 right hind foot. Light: no fragments found, dark: fragments found. 1.17 1 1 n 2 1.18 1 1 1 1 f 1 1 1 1^^ \i __ CI 12.4 1 1 1 1 1 1 t 1 1 6 5 2 9 10 3 |8|7 4 11 12 c) the pits in which the horses were buried remained more or less undisturbed, then we may conclude the following: Pit 1.18 contains a complete skeleton. Therefore here a complete animal was buried. With skeleton 1.17 the forefeet and the left hindfoot are absent. These missing parts could have been detached from the carcass before the animal was buried, with the aim of using the radii and the metatarsus for the manufacture of bone artefacts. These skeletal elements are ideally suited for working on account of their straight shape and relative thickness. With skeleton 12.4 the front half is missing. For some reason or other the horse was cut in two before burial, after which only the back halfwas buried in pit 12.4. With skeleton 12.2 the forefoot and hind feet are missing. The forefeet and hind feet may have been detached with the aim of using the metapodial bones for making artefacts. Naturally, these conclusions are completely dependent on the correctness of the above-mentioned assumptions and the point of departure taken that the skeleton or parts of it were present in the soil in their normal anatomical juxtaposition. Concerning this latter point there is no doubt: the excavator observed articulated skeletons. The same applies to the state of the pits: during the excavation it was observed that these had been more or less undisturbed. It is unlikely that dogs gained access to the buried horses and dragged bits of them away. No traces of gnawing have been found on the bones; if dogs had indeed been responsible for taking away the bones that are now missing, then it is most unlikely that in doing so they would have left no traces on the bones that we have found. With an excavation where no special attention is paid to retrieving skeletons, for example as with the very time-consuming cemetery research, the question remains whether it is correct to assume that from every skeletal part at least one fragment has been retrieved. To get some idea of what can become lost in the course of an excavation, the weight of the excavated horse material has been compared with the weight of the bones of a subfossil horse from one of the terpen (i.e. dwelling mounds in the Northern Netherlands) from the comparative collection of the Biologisch-Archaeologisch Instituut (BAI 226). Table 50A gives the weight of the bones of the horse from the collection and that of the complete bones of the excavated animals. In the calculations below it is estimated how heavy the skeletons of the excavated animals would be if they had been complete (expected bone weight). The expected bone weight is estimated by multiplying the total bone weight of the horse from the collection (14,193 g) by the ratio of the weight of the complete bones of the excavated skeleton with respect to the same bones of the horse in the reference collection (M. find/M. BAI). The number n indicates how many complete bones were available, and is used for the estimation. By subtracting the actual bone weight, one arrives at the weight of the bones not found during the excavation. The small bones that are not mentioned in table 50, for example the sesamoid bones, have also been disregarded in the calculations. 107 In table 50B it is indicated what percentage of the expected weight per skeletal element has been recovered. 1.18 M. find/M. BAi= 1.26 n=i5 expected weight weight of excavated bones 1.26x14-193 kg = missing 1.17 17.9 kg 11.3 kg 6.6 kg M. find/M. BAI = 1.19 n=2 expected weight weight of excavated bones 1.19x14-193 kg = missing — 37 /o 16.9 kg 2-3 kg 14.6 kg =86% If indeed no forefeet were present then the expected weight without forefeet is: 1.19X II.851 kg = 12.4 weight of excavated bones 14-1 kg 2-3 kg missing 11.8 kg M. find/M. BAr= i.ii expected weight weight of excavated bones n=5 i.iix 12.193 kg = missing = 84% 15-8 kg 2-9 kg 12.9 kg = 82% If indeed only half the carcass was buried, then the expected weight without the front half is: 1.11x7-504 kg = 12.2 weight of excavated bones 8-3 kg 2-9 kg missing 5-4 kg M. find/M. BAI= 1.60 expected weight weight of excavated bone missing n=3 1.6OX 14.193 kg = = 65% 22.7 kg 10.4 kg 12.3 kg = 54% If forefeet and hind feet were indeed missing, then the expected weight without forefeet and hind feet is: 1.60x12.443 kg= 19.9 kg weight of excavated bones 10.4 kg missing 9.5 kg =48% With horse i. 18 almost all the skeletal elements are completely or partly present. Of the larger bones only the right femur is missing. In total 37% of the bone weight is missing. This is largely accounted for by the missing femur, the ribs, most of which are missing, and the skull, scapulae and pelvis, of which large parts are missing. Since many recent fracture surfaces are present on the bones, while the fragments that would fit on to these surfaces are missing, it is evident that a much larger proportion of the horse must have been present in the soil. The conclusion that a complete animal was buried at this spot thus appears to be correct. 108 Druten: A. Weights (g) per skeletal element of a horse from the comparative collection of theBAi (no. BAI 226) and of horses 1.18, 1.17, 12.4 and 12.2. B: Weigth percentages of the horses from Druten with respect to BAI 226. TABLE 50 B. weight percentage with respect A. weight of complete bones to BAI 226 BAI Horse no. cranium mandibula scapula humérus radius ulna metacarpus 3 per element 1680 1.18 620 820 360 720 75 150 190 380 32 66 910 150 calcaneus 70 140 260 520 48 1.18 1.17 12.4 12.2 47 75 37 86 44 93 99 14 0 0 20 0 0 0 35 57 67 0 0 0 0 0 0 0 0 38 49 47 20 23 86 45 54 0 68 75 39 633 634 90 84 92 81 320 275 19 III I 70 84 92 52 46 0 99 0 48 0 91 0 47 0 0 0 56 50 680 480 1000 55 500 92 76 274 20 54 645 51 31 0 38 0 0 0 0 0 0 0 127 127 47 86 125 125 0 0 0 68 630 79 59 0 0 0 0 51 50 710 180 410 180 1750 indeterminate vertebra (g) unknown fragments (g) total weight (g) 226 247 8 8 990 75 metatarsus 2+4 phalanges atlas axis cervical vertebra thoracal vertebra lumbar vertebra sacrum rib 480 1150 astragalus metatarsus 3 12.4 1680 1405 310 495 575 33 455 1.17 per element total 410 metacarpus 2+4 pelvis femur patella tibia 226 225 29 0 0 99 56 4 0 14 0 100 15 202 I4I93 II288 75 87 45 2306 2963 10395 Of the well preserved material of skeleton 1.17 a great deal became broken and lost during the process of excavation. If the whole skeleton was present in the soil, then after recovery of the excavated bones 86% of the mass is missing. If the forefeet were not buried with the carcass, then 84% is missing. Almost all of the fragments found show a recent fracture at their extremities, even after parts that fit have been glued together. This implies that there must have been much more material present in the soil. In any case it is very likely that the fragments that were found come from those bones that were still in a complete state in the soil. If so much of the excavated skeletal elements was lost during the excavation then 109 it is very well possible that also the forefeet were present but were not recovered. Also with skeleton 12.4 the first impression suggested by the material is very doubtful. Among the 31 fragments there are 6 complete bones: three tarsals, a patella, and a splint-bone. All the other bones show recent fractures, so it may be assumed that all these bones were complete at the time of the excavation. Now 65% of the expected weight of the back half of the animal, as represented by bone finds, is missing. With the bone material being recovered in such an inaccurate way, it is not impossible that also the front half of the animal was present at the time of the excavation. For skeleton 12.2 the same applies as for 12.4. All the incomplete bones were broken recently, and therefore must have been still complete at the time of the excavation. Only a few ribs show a couple of old fractures. Of the bone weight that should be present 54% is missing, and if we assume that forefeet and hind feet were indeed absent this figure is 48%. All in all, it is clear that here no conclusions can be drawn regarding the presence or absence of certain skeletal elements. From the above it can be seen how much material can become lost with relatively large-scale excavations (cf. Nijmegen IV, 3.3.2.2). Moreover, it must be said that this loss of material and this inaccuracy in excavating are usually unavoidable. Thus all the excavations dealt with in this research were rescue excavations, during which there is no time for the application of meticulous excavation techniques. Much material can become lost, for example, during the removal of soil with the use of a mechanical digger, which may explain the absence of whole parts of the skeletons. But also when hand-collecting is carried out relatively carefully, it is still possible for a lot to be overlooked (see hand-collecting/sieving experiment, 2.1.3). For archeozoological research complete skeletons are of more value and provide more data than the sum of the skeletal elements of which they consist. Moreover, whole skeletons are of inestimable value for research into the development of breeds and for the improvement and development of archeozoological techniques. Consequently, if articulated skeletons are observed during an excavation, it is to be recommended that these are recovered in as complete a state as possible. 3.7.4.S Additional data The more or less complete horse 1.18 has well developed canines in both the upper and lower jaw. As these teeth are always present in the stallion and are usually absent in the mare, it is probable that 1.18 is a stallion. The values for withers height of the animal, estimated on the basis of the maximum length of seven long bones and four metapodials (table m22), all fall within the range 144-152 cm, according to the criteria of Vitt (see Von den Driesch and Boessneck, 1974). On the basis of the degree of epiphyseal fusion and the state of the teeth, the animal must have been older than 5 years. The withers height of horse 1.17 could only be estimated on the basis of the right femur. According to the criteria of Vitt, this animal falls into the size range of 144-152 cm. In view of the degree of fusion of the epiphyses, the animal was older than 3.5 years at the time of death. On the pelvis a pathological deformity is present on the flat part of the ilium. The bone may have suffered a fracture and subsequently healed. On the basis of the epiphyseal data, horse 12.4 was older than 4-5 years. On the basis of the tibia the withers height is 128-136 cm, and on the basis of the metatarsus it is 136-144 cm. All lumbar vertebrae and one of two thoracic vertebrae show complete fusion and/or exostosis: an indication that the animal was of advanced age. The withers height of horse 12.8 falls into the size class 144-152 cm. This could no be estimated from the lengths of the left femur and two tibiae. The epiphyseal suture of the proximal epiphysis of the femur is not completely closed. The other epiphyseal sutures that close around the age of 3.5 years are closed already. Therefore this horse could not have been much older than 3.5 years. The unfused epiphyses of the vertebrae confirm this. 3.8 THE GALLO-ROMAN TEMPLES AT ELST 3.8.1 INTRODUCTION The village of Eist is situated at a point where several stream-ridges come together. In Roman times Eist lay on a branch of the Rhine that was then navigable, at an important intersection of roads in the heartland of the tribe of the Batavi. In 1947 excavations were carried out by the ROB within and next to the Dutch Reformed church of Eist, that had been badly damaged in the Second World War. Traces were found of two early medieval churches and two Gallo-Roman temples (Glazema 1952). Bogaers (1955) demonstrated that in addition a yet older culture layer was present. The results of the excavation of the two Gallo-Roman temples plus the older culture layer are described by Bogaers (1955). Unless stated otherwise, the archeological information I have made use of is taken from Bogaers' excavation report of 1955. In addition to the two temples, there are traces of at least three other stone buildings dating from the Roman period (Bogaers 1970a; 1970b). On the basis of these finds, van Es (1981: 197) suggests that the sanctuaries may have formed part of a large temple complex. Whether there was also an occupied village at this spot is uncertain (Van Es 1981: 229). Willems (1984: no), on the other hand, considers Eist to be a vicus. Hulst and Greving (1981) determined the position of the possible western limit of the settlement. The bone material that will be dealt with in this study comes from the excavation of 1947. The material can be divided into two groups. The first group of bones comes from the culture layer predating the building of the temples, while the second group is associated with the temples themselves. 3.8.1.1 The period before the temples were built The oldest indications of human activity, predating the building of the temples, consist of a thin layer, sealed in by brown clay, containing scattered animal bones, fragments of charcoal and a few shells. This layer extends over the whole excavation area (Bogaers 1955: 42) and even beyond it (Bogaers 1955: 59). The layer is a distinctly self-contained unit, and probably dates from the early Roman period. Yet there are no indications that this was part of an occupation surface (Bogaers 1955: 42-43). Bogaers states that there is a possibility, 'though nothing more than that', that the finds belong to a purely native, Batavian cult site that existed before the stone temples. He regards the abundant animal bones (find nos. 230, 231, 236, 320, 322, 323) as being indicative of this. 3.8.1.2 The temple period The two temples found in Eist are both of the Gallo-Roman type, and are built of stone. The earlier temple (Temple I) was built in about AD 50 at the spot where up until then there had been the possibly native Batavian cult site mentioned above. It was a simple rectangular building, measuring ca 8.70 x 11.60 m, oriented approximately north-south. After Temple I had been destroyed by fire, very probably during the Batavian III Fig. 29 Eist: reconstruction of the second Gallo-Roman temple (after Bogaers 1955). 10 I 20 m I revolt of AD 69-70, immediately afterwards a new temple (Temple II) was built at the same spot, with the same orientation. This porticus temple, that was built on a much bigger scale, stood on a podium measuring ca 23 x 31 m, and consisted of a rectangular cella with a surrounding gallery (fig. 29). Temple II remained in use until into the 3rd century. In view of the Gallo-Roman architectural style and the materials used, it is evident that the temples were built by the army by order of the Roman military authorities. Nevertheless they are specifically native Batavian sanctuaries. On account of the exceptional size of the buildings, the temples may have been of more than regional importance. The animal bones that are considered to derive from the temple period mostly have no distinct stratigraphical context. According to Bogaers (1955: 140), most of the bones undoubtedly belong to the temples, although as a result of later digging activities some bones may also be present dating from the older culture layer predating the building of the temples. Bogaers proposes that the bones are the remains of animals that were sacrificed. In his view, the bones were buried in refuse pits and subsequently, as a result of later digging activities, they became spread out over the whole temple terrain. One such refuse pit is thought to have been found (Bogaers 1955: 43+fig. 48). The pit was certainly dug after Temple I had been built, and is probably associated with Temple II. Besides pieces of limestone and debris of roof-tiles, the pit contained many animal bones. The fact that below these bones the skulls of a pig, a sheep and an ox are present suggests to Bogaers (1955: 141-142) that these are the remains of a suovetaurilia sacrifice, that was possibly made on the occasion of the lustratio at the spot where the new temple was to be built after the destruction of Temple I (fig. 30). 3.8.2 THE BONES The faunal remains that were collected during the excavation of 1947 were previously studied by G. Kortenbout van der Sluis and A.P. Audretsch. The results of their study are presented in the form of a report (Bogaers 1955: 137-140). 112 Fig. 30 Eist: skulls of a pig, a sheep and an ox, the remains of a suovetaurilia sacrifice. TABLE 51 Eist: find numbers in which also human skeletal material is present. No. 236: pre-temple period; the rest: temple period. Find no. 40 181 185 188 191 194 199 235 236 302 304 336 number of fragments of human bones I 3 3 10 I 2 2 3 2 2 5 I For the sake of uniformity, the bones have been studied again as part of my own research. This has resuhed in a number of corrections with regard to species, skeletal elements and ages. Thus dog and goat have been removed from the species list, while the possibility of the presence of a few bones of Bos primigenius, the aurochs, has been ruled out. A few bones that were originally identified as coming from domesticated farm animals have now been identified as human bones (table 51). The domestic fowl has been added to the species list. As far as the skeletal elements are concerned, the most important difference with respect to the investigation of 1955 is that a considerable quantity of bones formerly identified as metatarsals, on which the minimum number of individuals was based, have now been identified as metacarpals. 3.8.2.1 The period before the temples were built A total of 45 bones have been preserved from the early period, before the temples were built. Most of these are cattle bones, while a few are of sheep/goat (tables 52 and mi i). The measurement data are given in table m24. Of the 42 remains of cattle, 24 come from the head, 16 from the forefeet and hind feet, and 2 are fragments of thigh-bones. It is remarkable that apart from the two thigh-bone fragments, no remains were found of the trunk, the stylopodium or the zygopodium. Nearly all of the bones give the impression of being immature. Of the 14 metapodials of which the distal parts of the diaphyses are preserved, none of the distal epiphyses are fused. This means that the animals were not older than 2-2.5 years. The only phalanx I that could be used for age determination came from a cow or bull older than 20-24 months. Three mandibulae are available for age determination. One comes from an animal aged ca 24-28 months, the others from cattle that were slaughtered at an age between 15-18 and 24-28 months. Summarizing, we can say that the cattle were slaughtered before or around the age of 2 years (table 53 and 54; fig. 31 A). Cut marks are present on four metatarsals. They are small incisions on the dorsal surface of the proximal end (butchery mark i - see appendix). They were very probably inflicted during the process of skinning, when the skin was being cut loose from the feet. The bones of sheep/goat that were found, two milk premolars and the diaphysis "3 Fig. 31 Ages of cattle at the time of slaughter on the basis of the epiphyseal fusion. A Eist, the period before the temples were built, B Eist, the temple period, C Eastern River Area, total without Eist. A 100-,% Elst.pre-lemple period B 100 C 100 Eist,temple period n=K ° n=119 Eastern River Area, total n = 121B 80- 80 so- Age at time of slaughter (months) 40' A B C D E F > 0-- 7/15 7/15--15/24 15/24--24/30 24/30--36/42 36/42--42/48 42/48 A/B/C D/E/F age A B C D E F age A B C DE F age of a femur, all come from immature animals. Of course the loose milk premolars may have ended up in the soil after they had been replaced by the permanent teeth of an animal, and consequently they do not necessarily give an indication of the age of the animals at the time of slaughter. It is evident that dogs were also present locally during the early period in Eist, in view of the gnaw-marks that are present on three metatarsal bones of cattle. TABLE 52 Eist, pre-temple period: the hand-collected material. Frequencies and weights (g). species 0/ niunber /() weight 0/ /o domestic mammals cattle sheep or goat 42 3 93.3 6.7 1836 16 99.1 0.9 total identified 45 lOO.O 1852 100.0 Total 45 1852 TABLE 53 Eist, pre-temple period: age of cattle at the time of slaughter, on the basis of the degree of fusion of the epiphyses (p: proximal; d: distal). TABLE 54 Eist, pre-temple period: ages of cattle at the time of slaughter, on the basis of tooth eruption (direct data). age (month) bone and epiphysis 20-24 phalanx I p 24-30 metacarpus d metatarsus d state of tooth* -1- ± M2 P P * +: erupted; ±: erupting; — : not yet erupted 114 number fused; killed after given age number not fused; killed before given age age (months) number of mandibulae of given age 15/18-24/28 ±24-28 2 I TABLE 55 Eist, m-temple period: the hand-collected material. Frequencies and weights (g)* number % weight 744 4 28 26 92.5 24500 96.5 0.5 65 0.3 3-5 3.2 312 1.2 I domestic birds domestic fowl species domestic mammals cattle sheep sheep or goat pig horse 0/ /o 462 1.8 0.1 35 O.I I 0.1 3 0.0 total identified 804 88.4 25377 97.6 unidentified mammals sheep-pig size cattle-horse size 2 1-9 5 0.8 104 98.1 618 99.2 total unidentified 106 11.6 623 2-4 Total 910 26000 * The skulls of the suovetaurilia are not included. TABLE 56 Eist, temple period: summary of the identified and the allocated unidentified hand-collected bone material (weight in g)*. species domestic mammals cattle sheep and sheep/goat pig horse birds /o weight % 848 33 27 93-2 3.6 3.0 25117 96.6 1-5 1.8 I 0.1 379 465 36 0.1 I 0.1 3 0.0 number 0/ * The skulls of the suovetaurilia are not included. 3.8.2.2 The temple period Of the 910 bones that were mostly associated with the temples, it was possible to identify 88.4% (97.6% by weight). These include bones of cattle, sheep, sheep/goat, pig, horse and domestic fowl (tables 55 and mil). The measurement data are presented in table m25. Of the identified bones, 96.5% by weight are of cattle. Pig scores 1.8% by weight, sheep and sheep/goat 1.5% by weight. In addition the material included a bone of a domestic fowl, the proximal half of a femur, and a phalanx I of horse. Gnaw marks on 16 bones bear witness to the presence of dogs on the temple precincts. Of the 106 skeletal elements that could not be identified to species level, two belong to the size range of sheep-pig; both of these are bones of the facial part of the skull. The rest of the unidentifiable bones fall into the size range of cattle-horse. These bones consist for 24% by weight of vertebrae. The other fragments are small pieces of indeterminate skeletal elements with an average weight of 5 g. Seeing that the bones within this size class that are identifiable to species level are all cattle bones, with the exception of one horse bone, it is most "5 TABLE 57 Eist, temple period: age of cattle at the time of slaughter, on the basis of the degree of fusion of the epiphyses. It is indicated what percentage of the cattle were slaughtered before and after a given age and during the interval between the given age and the preceding one (f: number fused; nf: number not fused; p: proximal; d: distal). killed after age (months) 7-15 bone and epiphysis 7-10 12-15 scapula d radius p f 15-20 20-24 humérus d phalanx I p 6 0 0 24-30 tibia d metacarpus d metatarsus d 36 42 calcaneus p femur p 42-48 42-48 42-48 42-48 ulna p radius d femur d tibia p 24-30 24-30 8 10 10 3 18 I 5 34 0 29 6 68 0 3 0 2 0 42-48 0 0 4 36-42 100 2 2 24-30 nf 10 l6 15-24 (%) 0 5 0 I 0 2 0 I 0 2 0 0 6 killed before killed between (%) (%) 0 0 90 90 94 4 100 6 100 0 likely that all or nearly all of the unidentified bones in the class of cattle-horse also come from cattle. If we make use of the data for the unidentified material by allocation according to the percentage distribution of species (table m35) then the resulting shift is negligible: only a few tenths of a percent in favour of domestic cattle at the expense of the small domesticated farm animals (tables 55 and 56). Cattle The cattle were slaughtered at an early age. On the basis of the epiphyseal data (n= 119), it is evident that all the animals were killed before they reached maturity (table 57 and fig. 31B). Of the animals 90% were slaughtered during the second year of life, and the remaining 10% during the third. The difference in the ratio between the fused and unfused epiphyses between 7-15 months and 15-24 months is strongly significant with a one-sided test at the p = 0.05 level (;C^ = 25, n = 312, df = i). The difference in the condition of the epiphyses between the other ages considered cannot be tested by means of a ;^^-test, on account of the low expected frequencies (< 5). The data provided by the mandibulae are indicative of a shift towards the beginning of the third year of age (table 58): 66% of the jaws come from animals older than 15-18 months and younger than 24-28 months, 25% come from animals that died between the ages of 24 and 28 months. Three mandibulae come from cattle aged 5-6 months, 15-18 months and more than 24-28 months, respectively. If we combine the data for the epiphyses and those for the mandibulae, then we can conclude that almost no animals were slaughtered that were younger than 15 116 TABLE 58 Eist, temple period: ages of cattle at the time of slaughter, on the basis of the eruption (d: direct data) and the wear (i: indirect data) of the teeth. state of the tooth* + ± Ml MI M2 MI M2 M3,P M2 M3,P M3,P age (months) number of mandibulae of given age 1 d <5/6 ±5/6 5/6-15/18 ±15/18 15/18-24/28 ±24/28 24-28 < 0 I 0 I II 0 killed in given age phase % 0 0 3 0 0 0 ID 3 66 7 I 25 I 0 3 * + : erupted; ± : erupting; — : not yet erupted months or older than 36 months. Most of the cattle would have been slaughtered during the second half of the second year of life, and some in the first half of the third year. As the animals were slaughtered at such a young age, hardly any measurements were taken, since only measurements of adult bones are of any use, as a general rule. As a result of this it is not possible to make any statements about the ratio of the sexes among the cattle. For the determination of withers height, only two full-grown metatarsals were available, with maximum lengths of 203.0 and 209.0 mm respectively. If we multiply these values by 5.45, the factor given by Von den Driesch and Boessneck (1974) for metatarsals of cattle of unknown sex, then we arrive at withers heights of no.6 cm and 113.9 cm. A number of bones display chop marks or cut marks. The only hyoid bone that has been found shows cut marks at the aboral end (mark i; see appendix), that were probably inflicted when the tongue was being cut away. On scapulae the following butchery marks are present: 2 (twice); 3 (three times); 13 (once); 36 (ones); 28 (once). In the case of a pelvis fragment the pubis has been chopped through (26). The cut marks running transversely over the middle of the dorsal surface of an astragalus (6), the small cuts on the dorsal surface of the proximal end of the three metatarsals (2) and the chop mark on the lateral side of the proximal end of a metatarsus (6) were probably all inflicted during the process of skinning, when the skin was being cut loose from the feet. Sheep and sheepjgoat Of the 32 bones of sheep or goat, 4 could be positively identified as sheep. All four of these are horn-cores or skull fragments with horn-cores. Not a single fragment indicates the definite presence of goat in Eist. AU fragments come from the head and the feet (table mi i). No remains of the trunk have been found. Only 3 mandibulae and 8 diaphysis-epiphysis extremities are available for age determination. In the three jawbones the M3's have erupted and are slightly worn. In any case they come from animals older than 18 months. The epiphyseal data do not provide any evidence that the animals were slaughtered before the age of 15-24 months. For one bone an age of less than 42 months could be ascertained (table 59). If we combine the age data for the mandibulae and for the older bones then we can say, with due reservation, that no sheep/goats were slaughtered before they were ±2 years old. A metacarpus of 127.8 mm and a metatarsus of 120.8 mm are available for the estimation of withers height. According to Teichert (see Von den Driesch and 117 TABLE 59 Eist, temple period: age of sheep/goat at the time of slaughter, on the basis of the degree of fusion of the epiphyses (p: proximal; d: distal). number fused; killed after given age number not fused; killed before given age humérus d radius p I I o o 15-20 20-24 tibia d metapodia d 2 3 o o 42 radius d number fused; killed after given age number not fused; killed before given age bone and age (months) epiphysis 3- 4 TABLE 60 Eist, temple period: ages of pig at the time of slaughter on the basis of the degree of fusion of the epiphyses (p: proximal; d: distal). bone and age (months) epiphysis 12 radius p I 24 tibia d 2 36 ulna p o 42 femur p femur d o I Boessneck 1974), the metacarpus and the metatarsus are indicative of heights of 62-64 cm and 54-56 cm, respectively. Cut marks are discernible on a humérus fragment on the medial side of the distal epiphysis (22), and on a metacarpus on the volar side of the proximal end (3); all of these marks could have been made in the process of skinning the animals. Pig The 26 fragments that have been identified as pig all come from the head and feet (table mil). In determining the age at the time of slaughter, we have to rely on scanty data (n = 10) provided by the degree of epiphyseal fusion (table 60). No data concerning the teeth are available. To give a very rough estimate, we may conclude that most of the animals were slaughtered before they were 3-3.5 years old, and some certainly before they reached their second year. A butchery mark is present on only one bone: a chop mark halfway along the diaphysis of a humérus (2). 3.8.2.3 Suovetaurilia As mentioned in section 3-8.1, three skulls, of a pig, a sheep and an ox, can be regarded as the remains of a suovetaurilia sacrifice (fig. 30). Since these skulls are of such special significance, they are discussed separately here. The ' measurements of these skulls are included in table m25, together with the other measurements for the temple period. The skull of the pig is for the main part complete. It includes left and right parts of the frontal, lachrymal, jugal and palatine bones and the maxillae. The brain-case is broken, and was possibly smashed intentionally. This could have happened when the animal was killed, or after slaughter to permit the removal of the brain. Pig brains were regarded as a delicacy in Roman times (see for example 118 the many recipes in the cookery book of Apicius CaeHus in which pig brains are an ingredient). The dentition is fully developed, so we may conclude that the animal was older than 20 months (Habermehl I975)- The sex of the animal can be ascertained from the maximum width of the alveolus of the right canine (±22.5 mm), and from its shape. The animal concerned was certainly a boar (i.e. a male). The second skull could positively be identified as sheep. This identification was made on the basis of the horn-cores (Boessneck et al., 1964; Schmid 1972), and of the pattern of the sutures between the frontal, parietal and occipital bones (Boessneck 1969). The base of the skull and the neurocranium are both present in their entirety; the tips of the horn-cores are missing. It is very likely that this animal was a ram, in view of the great thickness of the horn-cores; these had a maximum diameter of more than 6 cm, which is very large compared with measurements of other sheep horn-cores from the Eastern River Area. The basisphenoid and the presphenoid are fused, so the animal must have been older than 4-5 years (Habermehl 1975: 121). In view of the intact state of the brain-case, it is clear that this sheep could not have been killed by means of a blow on the head; nor was the brain removed. As for the bovine skull, the neurocranium and the right half of the base of the skull have remained preserved. The neurocranium of this skull has not been smashed by means of a heavy blow, either. The absence of the left half of the base of the skull can probably be accounted for by the head having been chopped off the trunk. The straight fracture running transversely across the left half of the occipital bone is indicative of this. Both of the horn-cores were sawn or hacked off, as is testified by the very distinct saw marks or chop marks on the parietal bone. The age of the animal at the time of slaughter cannot be determined accurately, but since the sutures of the skull have not closed up and the bone structure has a young appearance, the animal could not have reached maturity. The sex of the animal cannot be ascertained. 3.8.2.4 Discussion and conclusions Before we discuss the faunal material, we shall first consider the human bones that were found, that occur both in the early period and the temple period (table 51). If these bones have been correctly ascribed to these periods, then this means that people were buried on the terrain of the cultic site and the temples. Yet during the excavation no traces were found, apart from these isolated bones, that were indicative of burials in the Roman period. Bogaers (1955) points out that as a result of later digging activities the animal bones became spread out over the terrain and that also bones from the older cultural layer could have become mixed with those dating from the temple period. It is therefore possible that the human bones come from the churchyards of the later Christian churches, and that they ended up in the Roman material as a result of digging activities. It seems unlikely that any animal material dating from medieval times and later could have ended up among the Roman bones, since animal refuse obviously would not have been dumped in a churchyard. The most remarkable feature of the bone material from both the temple period and earlier is the pattern of ages of cattle at the time of slaughter. On the basis of the epiphyseal data, it is evident that in both priods the cattle were slaughtered for the most part during their second year of life. This is in marked contrast to the data for other places in the Eastern River Area, where most of the cattle were slaughtered only after the age of two years (fig. 31C). Also elsewhere within the Roman Empire mainly animals of more advanced age were slaughtered (Luff 1982: 261). This deviating pattern of ages at the time of slaughter is indicative of an intentional selection of young animals. In view of the fact that the corresponding data for the other settlements, both civilian and military, and both 119 native and Roman, are completely different, the choice of young cattle cannot be associated with the kind of people who used the terrain in Eist. For the temple period it is therefore clear that the choice of young animals must be associated with the special function of Eist as compared to other sites, namely as a sanctuary. Bogaers' view that the bones are remains of sacrificed animals can therefore be given further support by the fact that the pattern of ages of cattle at the time of slaughter for the temple period deviates conspicuously from that of normal meat-providing animals. Bogaers (i955) suggested the possibility that the terrain in Eist was also a place of cultic significance before the temples were built. His first argument for this is the occurrence of faunal material in the absence of any indications of an occupation layer. His second argument is based on continuity: after the early period the terrain was certainly hallowed ground. A third argument may now be added, namely the pattern of ages of the cattle at the time of slaughter. As described above, it is probable that the young slaughtered cattle of the temple period can be identified as sacrificed animals. By analogy, the cattle of the period before the temples were built, that shows the same pattern of ages at slaughter, must also be regarded as sacrificed animals. With this last argument the possibility that Eist was a place of cultic significance in the period predating the temples now becomes a probability. In the period before the temples, besides cattle also sheep/goat has been shown to be present, while in the temple period sheep, sheep/goat, pig, horse and domestic fowl are in evidence. The question that arises here is whether we should consider also the bones of sheep, sheep/goat and pig, apart from the suovetaurilia, as remains of sacrificed animals. The bones of sheep/goat from the period before the temples were built are so few in number that no definite conclusions can be drawn concerning this question. What is remarkable about both pig and sheep/goat bones from the temple period is that remains of the trunk, i.e. vertebrae, ribs, thoracic and pelvic girdle, are completely absent. These skeletal elements are also absent from the unidentifiable bones from the size class of sheep-pig (table mil). This situation could be explained by supposing that sheep/goat and pig were slaughtered on the spot, and that the head and feet were used for human consumption or for other purposes, while the trunk was taken elsewhere. An alternative explanation on the basis of the skeletal elements found is that the animals were slaughtered elsewhere and that only the head and feet were brought to the temple terrain. The second explanation seems unlikely, for two reasons that are related: parts of the autopodium were found and there are indications that the sheep or goats were skinned. If the animals had been slaughtered and skinned elsewhere, then it would be logical that the typical butchery refuse, the bones of the autopodium, were also removed at the same spot. This explanation would only hold good if we assume that butchery refuse was offered to the deity, which would indeed have been economical, but would hardly have been fitting as a token of veneration. We are thus left with the first explanation. The first part of this, that sheep/goat and pig were slaughtered on the spot, is very probable in view of the material found, and may be an argument that sheep/goat and pig were used as sacrificial animals. Whether this was actually the case depends first of all on whether the temple terrain was also occupied. If occupation was present, then the animals could simply have been slaughtered for human consumption. If no occupation was present, then a second question arises, namely if it is reasonable to assume that the visitors to or users of the temple terrain took live animals with them from home to Eist merely for the purpose of slaughtering them there for consumption. On the basis of 20th-century efficiency I would answer this second question in 120 the negative, but it is debatable whether rational arguments are valid in the sphere of religious practices. Because the first question cannot yet be answered, we cannot make any positive statements about the use of sheep/goat or pig as sacrificial animals. Gnaw marks bear witness to the presence of dog, either as an inhabitant of or as a visitor to the temple terrain. Seeing that no bone fragments of dog have been found, it is unlikely that dogs were either eaten or sacrificed in Eist. Concerning the femur of a domestic fowl and the horse phalanx nothing more can be said than that these two bones of these species were found on the temple terrain. The arguments put forward by Bogaers (1955) for regarding the three skulls of pig, sheep and ox as a suovetaurilia offering are corroborated by the archeozoological evidence. The first argument is the improbability that precisely at the site of the temple in Eist three skulls of precisely these animals ended up in a pit merely by chance. Such large fragments of skulls of the three animal species concerned are rare in any case in the bone material of the Eastern River Area, but nowhere else have such fragments been found in this combination. The second argument is that the sheep/goat skull can be positively identified as sheep. A third point that is worthy of mention is that the two animals for which the sex could be determined were both males. With regard to the horn-cores of the ram, it should be pointed out that these were very large compared with other sheep horn-cores found in the region, which suggests that the ram must have had a striking external appearance. 121 4 Production and consumption of animal products 4.1 THE DOMESTICATED (MEAT-YIELDING) MAMMALS 4.1.1 THE GENERAL PICTURE Among the domesticated mammals only cattle, sheep, goats and pigs were eaten. Neither horses nor dogs were a source of meat for human consumption (see 5.1 and 5.2). In the following we shall first of all consider the general picture presented by the percentage distribution of these meat-yielding farm animals. Subsequently (4.1.2-4.1.4) an attempt will be made to give explanations in more detail for the similarities and differences in the percentage distribution among the various complexes that have been studied. The criteria that are made use of in this connection are the function or nature of the settlements and their geological situation, which to a large extent will have determined the local environment. The information concerning the geology and the associated environment has been derived mainly from Willems (1981a; 1984). A broad overview of the geology of the region is presented in fig. 2. For further details see the text and the numerous maps (in particular the appendices) in the publications of Willems. The settlements that have been investigated differ too greatly in terms of function/nature and situation to permit an analysis of the data simply on the basis of chronology. For such a purpose it would be necessary for data to be available for more settlements with the same function/nature and situation, dating from different periods of time. The chronological factor is considered only in the case of those settlements where more than one phase can be discerned in the bone material. The data for the cultic site at Eist and the 4th-century cemetery in Nijmegen have usually been left out of the analyses because the bones from these complexes cannot be regarded as 'normal' refuse of slaughtering and consumption. In section 4.2 attention is devoted not only to the meat yields but also to the other products that could have been supplied by cattle, sheep/goat and pig, such as milk, wool, hides, bone, dung and traction power. The proportions of the different meat-yielding domesticated mammals have been compared on the basis of weights of bones and the numbers of bones. This is because bone weight gives a better indication of meat weight, while the numbers of bones give a better indication of the proportions of the numbers of slaughtered animals. It has to be pointed out that the data presented are not as definitive as they appear to be. If we compare the ranking order of the frequency percentages then we see that in many cases these values do not run parallel, on account of differences in the degree of fragmentation. These differences in fragmentation can be caused by different butchering techniques, but also by differences in the spectrum of the bones that have been preserved of the species in the various find 122 Domesticated meat-providing mammals: bone weights (g), frequencies and percentages per find complex. TABLE 6 I bone- weight (g) cattle sheep/ goat Nijmegen la Nijmegen Ib-c 9678 31411 Nijmegen canabae Nijmegen castra weight-% number of fragments sheep/ goat pig 86.8 90.2 3-9 9.6 434 3-5 6.3 1198 2605 5898 94-3 79.0 2.0 3-7 233 8.0 271 682 87.8 60.0 41 15-5 2548 1429 120 5-4 11.4 28.6 669 5232 95.0 0.6 4.4 1234 15 174 86.7 I.I 12.2 214 323 152 92.2 3-1 8 57 77-4 81.5 16.8 287 23 8 5.8 5-5 4.7 1.2 106 93-3 16.2 2-3 3-4 6.9 31 206 35 14 0.5 168 41 5 80.8 78.5 13-7 19.2 2.3 1-5 2.3 1199 1383 126 49 48 87.3 91-5 9.2 3.6 51 3-5 21 66.4 65.1 80.6 24.8 19.7 11.9 15.2 pig cattle 1233 1074 2193 76025 29987 1440 2066 Nijmegen IV 112822 Meiners wijk Kesteren 6338 11768 402 688 cattle sheep/ goat pig cattle 24 70 116 153 82.2 81.7 9771 4899 351 364 322 28 93-6 92.6 Ewijk I Ewijk II 53625 33621 1322 859 96.1 2.4 791 821 95-4 2.2 Druten I Druten II Druten III 7474 14540 928 803 8.7 158 1679 2750 81.2 82.0 92.7 10.1 1510 1913 8.6 9-5 360 59 109 3.0 4-3 645 95 Heteren I Heteren II 59595 70 84 60 sheep/ goat 4.6 7-9 pig 13-3 10.4 5-5 8.8 7-5 complexes. Moreover, if we consider the fact that during the excavation the bones were collected by hand, and that the intensity of bone collecting varies considerably from one excavation to another, it is clear that notably where there is a lot of fragmentation and/or the amount of time available for bone collection is limited, the values obtained will be less representative. We are thus concerned with find complexes that are not entirely comparable in the first place. Strictly speaking the bone data do not meet the requirements necessary for the purposes of comparison. The values presented for frequency percentages and weight percentages, and the conclusions drawn from these, must therefore be regarded merely as rough indications of a real situation in the past, that can no longer be accurately assessed. Everywhere cattle were the most important meat-providing species. The proportion of cattle within the group cattle, sheep/goat and pig varies from 60.0 to 91.5% (by weight 79.0 to 96.1%). Sheep/goat and pig come in either second or third place. An overview of the distribution of cattle, sheep/goat and pig at the different sites is shown in table 61. To permit comparison of the various settlements on the basis of these percentage distributions, first of all the points of correspondence and difference in these distributions have been considered. For this purpose cluster analysis has been made use of, which involves the grouping of units at increasingly higher levels, on the basis of points of correspondence and difference (compare, for example, the taxonomie working method employed in Linnaeus' Genera Plantarum (1737)). The hierarchical technique according to Ward ('Ward's method') has been used, which involves the ordering of the units, the settlements, by means of a series of consecutive fusions into groups, which in turn are ordered into larger groups (Everitt 1974). The procedure begins with the computation of a similarity or (Euclidean) distance matrix between the entities. At any particular stage the 123 Fig. 32 Dendogramof the clustering according to Ward's method of the find groups based on the weight percentages of cattle, sheep/goat and pig per find group. Nijmegen la -cl Nijmegen I b-c Nijmegen ^ , canabaeL Heteren I Meinerswijk Druten III Nijmegen IV Ewijk I Ewijk II Kesteren Heteren II Druten I Druten II Nijmegen castra" 10 20 -V 30 *^ 90 ESS (coefficient X1000) method fuses individuals or groups of individuals which are closest or most similar. It is proposed that at any stage of an analysis the loss of information which results from the grouping of individuals into clusters can be measured by the total sum of squared deviations of every point from the mean of the cluster to which it belongs. At each step in the analysis, union of every possible pair of clusters is considered and the two clusters whose fusion results in the minimum increase in the error sum of squares are combined. The error sum of squares (ESS) is given by: ESS = YJ ^i \/n = 1 where Xi is the score of the i-th individual The result is presented in a dendrogram. For the cluster analyses presented here the programs of CLUSTAN2 (Wishart 1982) have been used. The cluster analysis has been carried out on the basis of the percentage data from table 61. The results are presented in the dendrograms of fig. 32 for the weight percentages and of fig. 33 for the frequency percentages. A glance at the two dendrograms shows that the points of fusion for the weight percentages give lower error sums of squares than those for the frequency percentages. If we assume that weight percentages are more an approximation of meat percentages while the frequency percentages are more an approximation of the proportions of the numbers of individuals of the species, then we see that the differences between the settlements as regards the percentages of slaughtered animals of the various species are much greater than the differences in the proportions of consumption of meat of those species. In other words: although there are often distinct differences in the pattern of consumption if we count the number of slaughtered animals, this difference is far less evident in the values for meat weight. This is caused by the great differences in weight between the different species and the fact that the heaviest species, cattle, is by far the most frequently slaughtered species in all the settlements, as a result of which the cattle 124 Fig. 33 Dendogram of the clustering according to Ward's method of the find groups based on the frequency percentages of cattle, sheep/goat and pig per find group. Nijmegen la -ch Nijmegen Ib-c Meinerswijk — Nijmegen , _, canabae |_ Nijmegen IV —T Ewijk I Ewijk II 1 Kesteren Heteren II Heteren I Druten ill Druten I Druten II Nijmegen , ^ castra Z} 10 20 •V 30 ^ 90 ESS (coefficient «1000 ) percentage is predominant everywhere, with these percentage values consequently lying close together. To give an idea of the differences in weight: Slicher van Bath (i960, table IV), on the basis of historical data, gives figures for live weight of animals in the i6th, 17th and i8th century of 88-400 kg for cattle, 20-30 kg for sheep and 42-100 kg for pig. The most important caesura lies between the group consisting of the Nijmegen castra, Druten 1 and II, and the other settlements. This difference is caused by the relatively low percentages for cattle and the consequently high percentages for sheep/goat and/or pig in the first group of settlements. Within the group, and thus within the Eastern River Area, the castra is the most widely deviating site, with the lowest percentage of cattle (60%, or 79% by weight) and the highest percentage of pig (28.6%, or 15.5% by weight). In section 3.2.4 it is mentioned that the inhabitants of the castra may have obtained part of their supply of beef from the canabae legionis, which would have a slight negative influence on the percentage of cattle bones in the castra. Druten I and II have the highest sheep/goat percentages in the region (24.8 and 19.7%, or 10.i and 8.6% by weight, respectively). The proportion of sheep/goat in the castra is considerably lower (11.4% or 5.4% by weight). The fact that the castra and Druten II lie together within this most widely deviating cluster is difficult to explain only on the basis of environmental factors. We would then expect a clustering of Druten with the other settlements situated in the Holocene channel zone. It is more likely that the clustering can be explained by the binding factor for these two settlements: the Legio X Gemina. According to Hulst (1978), Druten II was mainly geared to the presence of this legion in the castra, and it showed unmistakeable traces of Romanization. Yet it is striking that also Druten I occurs in this widely deviating cluster. The bone material of Druten I comes from places with only native pottery that could originate from a native settlement of the pre-Roman Iron Age. A second possibility, however, is that this pottery should be dated to the beginning of Druten II (Hulst, ROB, personal communication). The fact that the bone material of Druten I shows a relatively strong resemblance to that of Druten II (there is 125 a weakly significant difference (/^ = 7.143, df = 2, p < 0.05)) and not to the material from other native settlements is an argument for the second possibility suggested by Hulst, namely that the material of Druten I originates for the most part from the beginning of Druten II. The archeological features that belong to Druten I, mainly situated in the southeastern part of the settlement, are probably indicative of a functionally different part of the villa rather than of a villa belonging to the previous occupation phase. Among the other settlements a cluster is formed by the frequency percentages for Kesteren, Heteren I, Heteren II and Druten III. These are distinct from the rest on account of the relatively high numbers of sheep/goat and the low numbers of pigs that were slaughtered. The common factor for the settlements in this cluster is their situation in the Holocene area that will have been suitable for grazing sheep. For keeping pigs, however, the environment on the Holocene soils was marginal. It is also remarkable that the native farmsteads of Ewijk, that had the same geological situation, have lower percentages for sheep/goat and fall outside the cluster. If we consider the weight percentages then this clustering is no longer evident. This is because the bones of sheep/goat and pig have low weights compared to those of cattle, so the weight percentages of these smaller animals are a slightly discriminating factor as far as clustering is concerned. We therefore do not take these data into account. The cluster we have not yet discussed, consisting of the sites of Meinerswijk, Ewijk and Nijmegen (with the exception of the castra), forms a less consistent whole than the previous one. The error sum of squares is thus much greater. This group is characterized particularly by the low percentage of sheep. With the settlements Ewijk, Heteren and Druten a distinction has been made between bone material that was found in archeological features with only native pottery, the possibly earlier phases I of these settlements, and material that was found in archeological features with Roman pottery, the phases II. From the dendrogram it is clear that the differences between the phases I and II of these settlements are relatively small. For Druten the above described situation can be explained by the fact that phase I is probably contemporaneous with the beginning of phase 11. This could also apply to Ewijk and Heteren. If not, then we can be certain that in these settlements in the course of time relatively few changes took place as regards the relative proportions of the numbers of cattle, sheep/goat and pig that were slaughtered. 4.1.2 THE OCCURRENCE OF PIG AND THE FACTORS OF THE ENVIRONMENT AND THE NATURE OF THE SETTLEMENTS The most conspicuous and most anomalous complex is the Nijmegen castra, with by far the highest percentages for pig in the Eastern River Area. Clason (1977a: 126-128; 1978: 431) associated a relatively high percentage of pig bones in settlements with Roman influences. This interpretation was based on a higher frequency percentage of pig with respect to sheep/goat in Roman castella and towns with a dominating Roman influence, as compared to native villages and farmsteads where the ranking orders were the other way round. This is supported by fact that the Belgac, who reared sheep and pigs on a large scale, exported wool products and salted pork to Rome and other parts of Italy (Strabo IV. 197). The fact that the highest percentage of pig was recorded for the most predominantly Roman complex in the Eastern River Area, namely the Nijmegen castra, is in agreement with this interpretation of Clason. Also Luff (1982) partially agrees with this view on the basis of her archeozoological study of the Roman Northwestern Provinces, though she correctly points out that also the 126 environment can be responsible to a large extent for the values for pig percentages. This was previously suggested by Nobis (1955) and has been demonstrated by Todd (1975: 120) and Prummel (1979a), among others. We shall therefore test Clason's interpretation on the material from the Eastern River Area, taking into consideration the environmental factors involved. Concerning these environmental factors the following point should be emphasized. A relatively high percentage of a species can be explained by a favourable environment. But this favourableness is itself relative: the environment as such can be favourable or unfavourable to a greater or lesser extent, for a number of animal species. Let us first of all consider the ranking order for the occurrence of pig-sheep/goat (table 61). On the basis of the frequency percentages pig dominates over sheep/goat in Nijmegen la, Nijmegen Ib-c, Nijmegen canabae, Nijmegen castra, Nijmegen IV and Meinerswijk. The weight percentage figures for these complexes shows the same distinct predominance of pig. A slight predominance of pig is evident in the weight percentages for Ewijk II, and Druten II and III. To what extent do the environmental factor and the 'Roman' factor play a role in this predominance? In this context the environmental factor means the possibility for putting out pigs in the deciduous forests to graze on beech nuts, acorns, etc., a practice which was very important in the past for pig rearing on a large scale (Ten Gate 1972). If we assume that these forests must have been mostly situated on the higher, Pleistocene soils - the ice-pushed ridge and the fluvioglacial deposits hardly offer any possibilities for arable land or pasture (see Willems 1984: 50) - then we can rephrase the question thus: to what extent did the factor of situation of the settlements on or near the Pleistocene area and the 'Roman' factor play a role in the predominance of pig? The settlements Heteren, Kesteren and Ewijk, that were situated in the Holocene area, have the lowest pig percentages, while pig is less abundant than sheep/goat. This was also to be expected in connection with the absence of an environment suitable for pigs. The find complexes from Nijmegen, on the other hand, have relatively high pig percentages. Here pig predominates over sheep/goat, which can be explained by the presence of forests on the Pleistocene soils in the immediate surroundings. The environmental factor thus certainly plays a role in the sense that with the settlements Heteren, Ewijk and Kesteren, that were situated on Holocene soils, the local surroundings limited the possibilities for pig rearing. The inhabitants of Nijmegen, on the other hand, had the choice: at their disposal they had not only forests for grazing pigs on the Pleistocene ice-pushed ridge and the fluvioglacial deposits, but also Holocene and possibly Pleistocene (the cover-sands) areas of pasture for keeping sheep. Within the Nijmegen settlements we see that the two 100% military complexes, the castra and the material of Nijmegen la, that probably came from the castellum there, contain the highest percentages of pig. From this we are able to conclude that a relatively high percentage of pig is associated with the 'military' factor. This is supported by finds from the third purely military settlement, the castellum at Meinerswijk. Although this site lies in the Holocene area, where one would expect sheep/goat to predominate, it is in fact pig that is predominant here. Evidently the military factor takes precedence over the environmental factor. It is most likely that the pigs of Meinerswijk were brought there from settlements on the other side of the Rhine on the Veluwe, just as heather sods must have been transported there from the same region (Willems 1984: 355). Also Druten I and II, and to a lesser extent Druten III, have relatively high pig percentages, even though sheep/goat is predominant, at least in terms of frequency percentages. According to Hulst (1978) Druten II was probably geared 127 to the presence of the loth legion in the Nijmegen castra and had a distinctly Roman character. The high percentage of pig in spite of the less favourable environment for pig rearing around the settlement thus fits in very well with the view that high percentages of pig go together with the predominance of a Roman cultural influence. In phase III of Druten, dated to after the departure of the loth legion, Druten became more oriented towards the town, according to Hulst. The fall in the percentage of pig in Druten III can therefore also be regarded as a consequence of the disappearance of the Roman military market of the villa. 4.1.3 THE OCCURRENCE OF SHEEP/GOAT AND THE FACTORS OF THE ENVIRONMENT AND THE NATURE OF THE SETTLEMENTS Sheep require an open, relatively dry environment with no forestation, like high pastures and heathland. According to Bekedam and Herweyer (1978: 181-184), the reason why sheep do not thrive on wet land is because they are very susceptible to infestation by liver-fluke {Fasciola hepatica), which can be fatal. Prummel (1979a) has shown for a number of settlements in the Netherlands that the percentage of sheep/goat is dependent on the possible occurrence of liver-fluke. The environment suitable for sheep in the Eastern River Area is to be found notably on the channel zone deposits in the centre of the region and on the cover-sands of the Pleistocene area. The keeping of sheep was concentrated in the Holocene area: the highest percentages of sheep are found in the settlements Kesteren, Heteren and Druten, that all lie on Holocene soils. The lowest percentage, on the other hand, is to be found in Nijmegen IV, which is not surprising in view of the local environment. The environment around the late Roman Nijmegen IV differs from that of the other, early and mid-Roman sites of Nijmegen inasmuch as it became much wetter (see 3.3.1). If we place the possibility for keeping sheep in Nijmegen in the immediately adjacent Holocene area - Nijmegen being situated on the border between Holocene and Pleistocene soils - then the low-lying area will have become wetter, and therefore less suitable for keeping sheep, because of the high water levels and the frequent floods in late Roman times. Among the settlements in the Holocene area Ewijk is anomalous inasmuch as it has relatively low percentages of sheep. This could indicate that sheep-keeping was less important in this settlement, compared with the other agrarian settlements Heteren and Druten. It is remarkable that there is no corresponding relative abundance of sheep in the military settlements Nijmegen la, Nijmegen castra and Meinerswijk. One would expect that Meinerswijk would score highest in view of the favourable environment for sheep in the neighbourhood. This is not the case, however. The castra, on the other hand, does have high percentages of sheep. This could be explained as follows. If there is a choice between pork and mutton, then in military settlements the preference will be for pork. However, if the neighbourhood of a densely populated settlement like the castra is not able to meet the relatively great demand for pork then the deficit is made up with mutton. 4.1.4 THE OCCURRENCE OF CATTLE AND THE FACTORS OF THE ENVIRONMENT AND THE NATURE OF THE SETTLEMENTS All over the Eastern River Area the most frequently eaten animals were cattle. It is unlikely that the environment played a direct role in the availability of cattle. 128 All settlements lie in or near the Holocene soils, areas where pastures can be expected. The problem of the wetness of the soil, that as we have seen was a critical factor for keeping sheep, hardly applies for rearing cattle; it is possible to keep cattle successfully on both dry and wet pastures. Also if we consider the nature of the settlements there is no discernible pattern of regularity as regards the proportion of cattle. The values for the cattle percentages in the settlements appear to be mainly dependent on the extent to which sheep/goat and pig were consumed, for which animals it was indeed possible to discern a relation between the environment available and the nature of the settlement. 4.2 THE PRODUCTIVE VALUE OF PIG, SHEEP/GOAT AND CATTLE 4.2.1 PIG: MEAT PRODUCTION An important indication of the purpose for which pigs were kept is the age at which the animals were slaughtered. The age of animals at the time of slaughter was determined on the basis of the data on the degree of epiphyseal fusion of the various skeletal elements (table 62); the jaw fragments that were found did not provide enough data to permit an analysis concerning slaughtering age, and consequently they are disregarded. The epiphyses are divided into three groups on the basis of the time at which they fuse: around 12 months; 24-30 months; and 36-42 months. In the table presenting the total data it is indicated what percentage of pigs was killed before and after the ages given and what percentage was slaughtered in the different age phases. In fig. 34A these data are presented graphically. Only 8% of the pigs survived the 36-42 months. The rest were slaughtered in more or less equal numbers in the first 3-4 years of life: 30% in the first 12 months, 28% in the period 12-24/30 months and 34% between 24/30 and 36/42 months. There is no sense in discussing the age at the time of slaughter in the different settlements and comparing the results, seeing that only a limited amount of data is available per site. Only the Nijmegen castra provides a quantity of data per age class that I would dare to call percentages. Of the pigs found in the castra 34% were killed in the first year of life, 32% between 12 and 24-30 months, 32% between 24-30 and 36-42 months, and 5% at an age of more than 36-42 months (fig. 34B). The fact that only a few pigs reached a mature age makes it clear that they were kept primarily for the production of meat. This is what one would expect, seeing that pigs, in contrast to cattle, horses, sheep and goats, are economically valuable primarily as a source of meat. Pork was highly favoured by the Romans. This is Fig. 34 Age distribution of slaughtered pig on the basis of the epiphyseal fusion. Nijmegen, castra Eastern River Area. A 40-,% n=36l. B 40i% n=180 Age at time of slaughter (months) A 0-12 B 12-24/30 C 24/30-36/42 D > 36/42 A B C D A B C D 129 TABLE 62 Ages of pig at the time of slaughter, on the basis of the degree of fusion of the epiphyses. In the totals (data for all the sites combined) it is indicated what percentage of the pigs were slaughtered before and after a given age and during the interval between the given age and the preceding one (f: fused; nf: not fused; p: proximal; d: distal). Nijmegen Nijmegen Nijmegen Nijmegen Nijmegen la Ibcanabae castra IV age (months) 12 bone and epiphysis 12 12 12 24-30 36-42 24 24 24 24-30 24-30 36 42 42 42 42 42 42 scapula d humérus d radius p total f nf f nf f nf f nf f nf II 4 ID 0 10 8 3 2 0 21 2 I I 2 0 - - I 0 I 0 2 0 2 2 I I 6 2 3 4 2 0 0 2 I 5 I 3 6 18 8 6 6 8 37 5 19 tibia d metapodia d phalanx I p fibula d calcaneus p total - - 6 9 3 2 4 I I 7 5 9 18 I 0 I - - I 0 0 5 3 3 3 6 - - 2 - 5 - 9 12 7 10 3 17 5 33 4 I ulna p humérus p radius d ulna d femur p femur d tibia p total - - 0 I - - I 12 0 4 0 I 0 I I I 0 6 0 I - - - - 0 2 I 2 2 I 0 - - 0 2 0 I II I 0 0 I 0 5 5 2 2 I 3 2 13 13 70 0 0 0 3 9 16 3 4 4 9 6 3 0 0 2 I 3 16 4 0 I 2 18 testified by the great number of recipes for pork in the epicurean cookery book of Apicius, far greater than for other kinds of meat (see 3.3.4). In written sources it is mentioned that in order to meet the demand for pork in Rome and the rest of Italy, salted pork was imported on a considerable scale from other parts of the empire, in the form of sides of bacon, hams and shoulder joints (White 1970: 320-321). Apart from bacon and pork in the narrow sense, also offal, such as the uterus and the brains of pigs, were eaten and considered a delicacy. If we assume that pork was as highly esteemed in the Eastern River Area of the Netherlands as it was in Rome, then the local farmers, and certainly those who were producing foodstuffs for the military market, would have aimed at establishing a breeding regime with two litters a year (see 2.2.8). All the more so seeing that sucking pigs could evidently be sold for a good price (Columella VII.9.4; see 2.2.8). It is clear that sucking pigs, or at least pigs less than a year old, were also relished in the Eastern River Area, in view of the fact that 30% of the slaughtered pigs were younger than one year of age. We may assume that this percentage is a minimum value, as young animals are possibly underrepresented among the faunal remains found (see 2.2.7). Most of the boars (i.e. the male pigs) were probably castrated with a view to obtaining better quality meat. Columella (VII.9.4) mentions that this has to be done at the age of 6 months, or, in the case of animals used for breeding, in the fourth or fifth year of life. Varro (II.4.21) writes that the best age for this is one year. Varro (II.4.7) recommends that sows be allowed to breed only after the age of 20 months, and that they should no longer be used for breeding after the seventh year of age. According to Columella (VII.9.1), with pig breeding attention should be paid to the quality of the boar, 'because the offspring is often more like its father than its mother', which of course is nonsense. The fully mature animals that were found in the Eastern River Area (8%) were probably all animals that owed their relatively advanced age to the fact that they were used for breeding. 130 Meinerswijk Heteren I Heteren II Druten I Druten II Druten III killed after f f f f f f f (%) nf (%) 27 39 7 23 6 36 21 86 79 63 77 70 19 16 4 3 5 47 44 41 80 50 26 42 24 23 56 59 I 20 3 14 65 50 74 58 28 I 23 13 14 96 93 82 91 96 93 90 92 34 nf nf nf nf nf nf ---- 10 10 --10 1020 103020 --10-105020 -01 10 ____ ____ ____ __oi 30 --I0 20 ______ __ 11 ____ ____ ____oi __ii5i __ __ ____ 02-- 01 04 01 ----01 TOTALS 20 killed before killed between 01 ____ ____oi 2 4 7 18 9 4 7 3 10 01 02-- 020703 12 8 I 3 I I 10 23 27 28 138 37 23 30 (%) 30 Such a low percentage of mature animals is no problem with pigs; since they are sexually mature at a young age, they can produce several litters per year and each litter produced consists of a number of piglets. In contrast to pork, the other products of pig will have had little economic value. The hairs could have been used for making brushes and the skin for making leather. The bones were probably used hardly or not at all for making particular objects. The bone of pigs slaughtered at such a young age is not yet mature and is still rather weak, and is therefore not very suitable for bone-working. Where possible the dung of pigs will have been used as manure, although the classical sources give differing opinions on the value of pig dung for this purpose (White 1970: 128, 321). 4.2.2 SHEEP: PRODUCTION OF WOOL, DUNG, MEAT AND MILK For the sake of convenience only sheep will be referred to here, seeing that where it was possible to distinguish between sheep and goats it turned out that mainly sheep are concerned in the Eastern River Area. Of the small number of bones that could be identified to species level with any certainty only 8% are of goat. An important indication of the use of sheep is the age at which the animals were slaughtered. For the same reason as with pig (4.1.2.1) the age at the time of slaughter was determined only on the basis of the data on epiphyseal fusion (table 63). The epiphyses are divided into three groups, depending on the age at which fusion occurs: from 3 to 5 months, from 15 to 24 months, and from 36 to 42 months. In the table the total data indicate the percentage of sheep that were killed before and after the ages given and what percentage were slaughtered in the different age phases. This age distribution of slaughtered animals is presented graphically in fig. 35A. About half (49%) of the sheep were slaughtered after the skeletal elements were 131 TABLE 63 Ages of sheep/goat at the time of slaughter, on the basis of the degree of fusion of the epiphyses. In the totals (data for all the sites combined) it is indicated what percentage of the animals were slaughtered before and after a given age and during the interval between the given age and the preceding one (f: fused; nf: not fused; p: proximal; d: distal). Nijmegen Nijmegen Nijmegen Nijmegen Nijmegen la Ib-c canabae castra IV age (months) 3- 5 15-24 bone and epiphysis 3- 4 3- 4 5 15-20 20-24 36-42 36 36-42 36-42 42 42 42 42 humérus d radius p scapula d total tibia d metapodia d total calcaneus p ulna p femur p humérus p radius d femur d tibia p total f nf 2 I 3 nf f nf f I I 2 0 0 0 I 16 3 3 6 I 0 6 6 4 0 I 3 I I 6 I 0 I 5 5 3 4 3 4 IG 8 - - 16 9 0 I - I 0 0 0 I - - I 0 3 3 6 _ - _ — - 0 I 0 I f nf 0 5 0 0 2 I - I 0 I 0 8 I 3 I 0 2 0 I 0 2 0 4 6 - - - f - - I 0 I I I 0 - - - - - - I 0 I 0 0 2 0 I 0 I - - 0 2 r I 0 4 0 4 3 6 nf 0 I 3 5 19 fused (after 36-42 months). In the first 3-5 months 13% were slaughtered; 26% were killed between 3-5 and 15-24 months and 12% between 15-24 and 36-42 months. Also in the case of sheep there is no sense in considering the different complexes separately, on account of the small quantity of data per settlement. We can probably make an exception for the castra: 27% in the first 3-5 months, 9% between 3-5 and 15-24 months, 40% between 15-24 and 36-42 months, and 24% after 36-42 months of age (fig. 35B). These data on age at the time of slaughter cannot be regarded as representative, however, in view of the small number of bones providing data available on epiphyseal fusion (72). The fact that half of the sheep reached maturity indicates that sheep, unlike pig, had an economic value in addition to the provision of meat. Their additional value lay in the production of wool, milk (cheese) and dung. The 39% of the sheep that died before the age of 15-24 months were kept mainly for the supply of meat. The youngest of these, 3-5 months old at the time of death, would not have provided any wool at all and must have served exclusively as a source of meat. The skins of these young animals may have been used as a kind of 'Astrakhan fur'. Moreover, by slaughtering these young lambs it would have been possible to milk the mother sheep, and thus to produce sheep's-milk cheese. The young animals were eaten in the castra in particular (27%). The sheep that were slaughtered between 3-5 and 15-24 months of age will have produced at most one fleece (at around 18 months). Wool production will have been one of the most important reasons for keeping older sheep. The production of wool is emphasized by various classical authors; especially white wool was preferred (White 1970). Until linen was produced on a wide scale at the end of the ist century, wool was by far the most important material for making clothing. Columella (VII.3.13) mentions two regimes for keeping sheep. In remote districts as many lambs are kept as possible, provided sufficient pasture is available. In the neighbourhood of towns, where there is a market for meat and milk, the lambs are sold before they begin to graze: the lambs provide meat and the ewes can be milked. With the latter regime it would have been necessary to keep only one lamb in five to maintain the flock. The rams may have been castrated. In classical sources from Roman times 132 Meinerswijk f nf Kesteren f nf Heteren I f nf Druten I Heteren II f f nf nf Druten II Druten III f f nf TOTALS killed after killed before killed between nf f (%) nf (%) 94 84 79 87 2 4 3 9 6 16 -- 10 ____ 201190 29 10 -10 -10 20 10 10 ____ 10 10 201040 __3o 10 40 51 14 o 21 01 30 31 10 -10 --10 --01 00 II 144032 105222 249254 20 __ --10 --- __ 02 --10 -- ____ --10 _--01 -____ -10 __io-21---____3i 01 01 40 __io-25 -10 10 12 0120 4721 II 61 29 49 I I 6 7 7 6 5 33 81 67 58 61 50 50 60 64 44 67 28 49 10 21 31 I I 4 4 9 3 13 35 (%) 21 13 13 33 42 39 26 50 50 40 36 56 33 72 51 12 mention is made of castration in the fifth month (Haugher 1921). Castration prevents problems that arise when too many rams are present in the flock; moreover, castrated rams tend to be heavier animals and yield better quality wool (Bottema and Clason 1979). Apart from the products mentioned above the skins will have been used as well. Also the production of manure may have been important. This will have been available in the form of dung mixed with litter, from the sheep-cot, and in the form of dung deposited on the fields when sheep were allowed to graze there after the harvest. There are no indications that bones of sheep and goats were used for making particular objects. 4.2.3 CATTLE: PRODUCTION OF TRACTION POWER, MANURE, MILK AND MEAT As with sheep and pig, the data on the age at which animals were slaughtered can inform us about the purposes for which cattle were used and the primary and secondary reasons for keeping cattle. The data on the age at the time of slaughter as based on the degree of epiphyseal fusion are shown in table 64. The epiphyses are divided into five groups, according to the age at which fusion occurs: from 7 to 15 months, from 15 to 24 Fig. 35 Age distribution of slaughtered sheep/goat on the basis of the epiphyseal fusion. A 60- % Eastern River Area, total B 60 1% Nijmegen, castra n = 72 n=217 kO- Age at time of slaughter (months) A B C 0- 3/5 3/5 -15/24 15/24-36/42 D > 36/42 - 20- A B C D A B C D 133 TABLE 64 Ages of cattle at the time of slaughter, on the basis of the degree of fusion of the epiphyses. In the totals (data for all the sites combined) it is indicated what percentage of cattle were slaughtered before and after a given age and during the interval between the given age and the preceding one (f: fused; nf: not fused; p: proximal; d: distal). Nijmegen Nijmegen Nijmegen Nijmegen Nijmegen la Ib-c canabae castra IV age bone and (months) epiphysis f nf scapula d radius p total 2 0 4 6 0 7-10 7-15 12-15 15-24 15-18 phalanx lip humurus d phalanx I p total 15-20 20-24 24-30 nf f nf f nf f nf 21 0 5 0 0 I 0 0 31 0 0 6 0 31 25 56 2 0 52 18 70 0 10 4 4 7 15 0 4 13 15 32 0 3 14 25 42 0 I 2 0 I 29 3 0 7 16 28 2 IG 0 2 51 4 41 3 - - 8 2 16 2 16 4 I 5 I 0 II 2 2 25 3 49 31 4 34 37 87 2 12 5 7 14 I I 17 16 3 17 I 3 3 6 I 0 8 I I I 0 2 8 3 14 4 4 8 5 29 34 6 I 3 4 6 8 2 9 5 14 3 0 I I 2 3 8 I II - - I - - 5 0 2 3 0 2 I 2 5 0 I 10 I I I 3 8 3 20 3 6 0 I I 0 4 6 5 13 4 18 24-30 tibia d 24-30 metacarpus d 24-30 metatarsus d total 36-42 42-48 Fig. 36 Age distribution of slaughtered cattle on the basis of the epiphyseal fusion. A Eastern River Area, total (excluding Eist and Ewijk), B Nijmegen, early Roman period (Nijmegen la and b-c), C Nijmegen, mid-Roman period (Nijmegen castra and canabae), D Nijmegen 4th century (Nijmegen IV), E agricultural settlements (Druten I, II and III, Heteren I and II), F Eist, temple-period. calcaneus p femur p total 36 42 42-48 42-48 42-48 42-48 42-48 A 80-|% humérus p ulna p radius d femur d tibia p total B Eastern River Area, total 80- % n = 2l,36 0 0 2 I 3 Nijmegen, early Roman period 2 I 4 3 17 C 80- % I 3 3 9 I 7 2 12 8 7 8 10 53 30 Nijmegen, mid-Roman period n = 371 60- iO- 40 20- A B C D E A F E I00-|7o D B E F Agricultural settlements A F 100-|% B C BO- % 0-- Nijmegen, Uli century D E F Eist, temple-period n = 3U D 7/15 7/15--15/24 15/24--24/30 24/30--36/42 36/42--42/48 42/48 0 n = 153 Age at time of slaughter (months) A B C D E F,> f n = 119 80- 80- n = 322 40- 40- 20- A 134 8 C D E F A B C D E F A B C D E F Meinerswijk Heteren I Heteren II Kesteren Druten II Druten I Druten III TOTALS killed after killed before killed between nf f nf f nf f nf f (%) nf (%) (%) I I o o 0 13 i6 0 143 90 0 I 29 2 233 96 96 96 6 4 3 4 3 7 2 2 3 4 7 I 4 5 4 4 4 4 - — - - I 0 2 0 0 I 2 I 2 10 8 24 97 I 2 0 - - 7 4 o 9 0 IIO I 2 I 2 I 10 4 21 8 231 78 81 21 22 21 83 81 20 62 19 17 19 9 f nf f nf f nf 6 0 4 0 I 2 0 I 6 0 I 3 o o I 5 f I 0 _ - I 0 3 9 _ _ _ _ 2 0 2 I 0 3 o 5 7 o o o I 0 6 3 2 0 9 4 0 I 3 I 10 3 3 0 I I - - I 2 I 3 2 3 3 2 2 o 2 2 2 I 0 I I I I II 6 0 9 6 4 o o II 6 2 5 6 5 24 3 8 74 87 99 260 I I — - 0 0 I O I I 4 7 0 _ - I I - - 2 0 I 0 I 2 10 2 I o 0 I 0 2 I I I 2 6 17 3 5 8 39 58 97 70 72 71 17 23 40 30 28 29 10 2 I 6 4 _ _ - — o I - - 0 I 2 0 2 0 5 9 9 31 3 9 7 23 33 6 26 69 43 53 51 50 54 15 8 23 36 37 119 31 57 47 49 50 46 17 0 I 2 I I o 0 I o I 2 2 0 2 - - o - - 0 I 2 2 - - 3 I 2 0 I 2 8 o I I 0 3 3 3 I 5 2 9 0 4 9 3 8 37 37 139 96 82 96 90 10 I 22 4 27 4 18 4 10 6 months, from 24 to 30 months, from 36 to 42 months and from 42 to 48 months. It has to be pointed out that the ages referred to here, that are based on the epiphyseal fusion of recent breeds, represent minimal values for cattle in the past; in Roman times the epiphyses probably fused at a much later age (see 2.2.8). In the table the total values indicate the percentage of cattle that were killed before and after the ages given, and the percentage that were slaughtered in the different age phases. This age distribution of slaughtered animals is presented graphically in fig. 36A. The data for individual settlements or combinations of settlements are shown in fig. 36B-F. In the following we shall be concerned mainly with the total data and the data for combinations of settlements. Most of the individual settlements provided too few data to permit any meaningful conclusions. Here it should be noted that the reconstructed patterns of age at the time of slaughter are based on 15 different parts of the skeleton. Moreover, for each skeletal element both a right and a left specimen can be found, so that theoretically for one individual there are 30 data that are involved in determining the pattern of ages at the time of slaughter. During the excavations no complete skeletons were found, so that in practice the chances that one individual influences the pattern in 30 places are not very great. Yet it is reasonable to assume that bone material from a refuse pit, for example, very probably contains skeletal elements that come from one individual. Just how probable this is cannot be estimated, however. In any case, we know we are concerned with one individual in the case of the few complete bones found for which both of the epiphyseal extremities provide age data. All in all, the number of individuals that determines the pattern of age at the time of slaughter is smaller or much smaller than the numbers indicated in the table and in the figures. The data from Nijmegen for the different periods and the combination of 135 agrarian settlements from the Holocene area show a relatively great similarity (fig. 36B,C,D and E). In the early (Nijmegen la and Ib-c), middle (castra and canabae) and late (Nijmegen IV) Roman period in Nijmegen the percentages of cattle that were slaughtered at a fully mature age, i.e. 42-48 months or older (phase F), were 63, 52 and 63%, respectively. In the agrarian settlements outside Nijmegen (Druten I, II and III, Heteren I and II) the corresponding figure is 51%. In Nijmegen, in the third year of life (phase D and E) 24, 30 and 28% of the cattle were slaughtered, respectively; in the second year (phase B and C), 14, 18 and 4%, and in the first year (phase A), o, o and 5% of the cattle. For the combined agrarian settlements the corresponding values are 14% (D and E), 29% (B and C) and 6% (A). Milk production The low percentages of cattle that were slaughtered in the first years of life indicate that the cattle that were eaten were not kept primarily for the production of dairy products, i.e. milk or cheese. If a cattle farmer wishes to concentrate on the production of fresh milk or cheese then the cows must be allowed to calve frequently. This is because lactation only occurs when the cow is with calf and after the calf has been born. The calf has to be taken away from the mother as soon as possible after its birth so that the farmer can use the milk. If the cattle farmer is aiming at milk production then the cow will have to be served again by a bull 10 weeks after the birth of the calf, thus ensuring continual lactation and the production of one calf per year. Dairy farming thus results in a large supply of calves. The annual calving yields a number of newborn calves that greatly exceeds the number required for maintaining the herd. Consequently there is a large surplus of calves available for slaughter. To give an illustration: at the present time (1968) about one-half of all beef cattle in the Netherlands are calves (Ministerie van Landbouw en Visserij 1970: table 16). Nowadays calves are taken away from milk-cows a few days after birth. It is doubtful whether calves were taken away so soon from their mothers also in Roman times. In contrast to modern highly bred milk-cows, more primitive breeds require stronger stimuli for the milk let-down, i.e. the discharge of milk from the mammary gland into the milk cisterns from which it can be drawn off through the teat (Amoroso and Jewell 1963). Under natural conditions the let-down of milk is a reflex in response to the sucking action of the calf. Amoroso and Jewell state that many cows, even today, and certainly those kept under primitive conditions, will not let down their milk at all unless the calf is present. With these animals the cow has to be able to see and smell her calf. In this case seeing and smelling her calf are the necessary stimuli in response to which the milk let-down occurs as a conditioned reflex. If we assume that the cattle from the Eastern River Area were kept for the production of milk, then this should be discernible in the pattern of ages at the time of slaughter. If the cattle in the Eastern River Area needed, like modern milk-cows, only low-level stimuli for milk production (for example the rattling of buckets or the voice of the milker or milk-maid) then the calves will have been taken away from their mothers soon after birth. If this was the case then we would expect a high percentage of slaughtered animals in the age phase A or, if they were kept for the production of meat, in the phases up to and including D (see 'meat production'). If the presence of the calf was necessary for stimulating milk production then the calves would have been available for slaughter only after I year of age, in the age phase A and B. In this case too the animals could have been kept for the production of meat and slaughtered in the phases up to and including D. Yet we do not find any high percentages of animals slaughtered in the phases 136 A-D, that one would expect in the case of cattle farming aimed at milk production. Precisely the population concentration in the mid-Roman castra and canabae legionis would have been an ideal market for a surplus of calves and young cattle. But we do not find them here either. Even if we take into consideration an underrepresentation of calves among the cattle bones (see 2.2.7), the percentages are still low compared to the bones of immature sheep and pigs from the settlements. The bones of calves found in 4th-century Nijmegen and the agrarian settlements probably come from animals that died as a result of an accident or disease. For the sake of comparison: nowadays the loss of calves aged 0-3 months is 6% (Van der Kerk 1976). Only in Kesteren is there evidence of a high percentage of slaughtered animals in phase A. However, in view of the very small quantity of data available (5 fused, 3 unfused) this observation is of no significance whatsoever. Also the classical written sources make it clear that, at least in Italy, the production of milk from cows was of no importance (White 1970: 277). Whenever milk production is discussed the milk concerned is that of sheep and goats. Cato does not mention fresh milk anywhere, and for cheese he only mentions the use of sheep's milk (76-82). Varro (II.ii.i) refers to milk as 'of all the hquids we consume as food the most nourishing', to which he adds 'first sheep's milk, and then goat's milk', implying that cow's milk does not belong to the category of milk for human consumption. This is also suggested by the fact that Pliny (XXV.53; XXVIII.33) mentions the consumption of cow's milk only for medicinal purposes. Another indication of this is that Varro (II.5.7-9), when mentioning the points to look out for with breeding animals, does not make any reference to the udder. Naturally, with the descriptions made by Roman agronomists we have to take into consideration the fact that they were describing the agriculture of a region that diflfered from the Eastern River Area with respect to geology, climate, and economic possibilities. The possibilities for dairy farming will have been less favourable in Italy than in Northwestern Europe, as is still the case today. Illustrative in this respect are the present-day (i960) differences in the average milk yield per cow: Netherlands 4220 kg, Belgium 3810 kg and West Germany 3430 kg; in Southern Europe the corresponding yields are much lower: Portugal 2270 kg, Spain 1380 kg and Italy 1940 kg (French et al. 1966). To summarize, we may conclude that the cattle in the Eastern River Area were not kept primarily for the production of milk. At most the cows could have been milked on a small scale for the production of milk or cheese for human consumption. Meat production The cattle whose remains were found in Nijmegen had served as a source of meat for human consumption, as is evident from the butchery marks and fragmentation. However, the cattle will not have been kept primarily for the production of meat (and fat), seeing that a high percentage of the cattle were adult animals. With cattle-farming aimed at meat production one would expect a high proportion of calves and young cattle. If the production of veal is not the main objective of a cattle farmer (or if there is no surplus of calves as a result of milk production), then the age at which beef cattle are slaughtered nowadays is between 1V2 and 2V2 years for cows (usually animals that were intended for milk production but that did not become pregnant), i V2 years for bulls, and 2-2V2 years for oxen (Ministerie van Landbouw en Visserij 1970). In terms of our slaughter-age phases, if meat production were the chief aim of the cattle farmer then we would expect to find high percentages in the slaughter-age phases A-D, which in fact is not the case. In Exeter in England, for which the importance of 137 Fig. 37 Age distribution of slaughtered cattle on the basis of the epiphyseal fusion. Age at time of slaughter (months) A B C D E F > o- 7/15 7/15-15/24 15/24-24/30 24/30-36/42 36/42-42/48 42/48 A 80 % B Nijmegen, castra 60 % Nijmegen, canabae legionis n = 226 n = H5 60- 20 A B C D E F A B C D E F meat production in the post-medieval period is testified by historical records, it is evident from the epiphyseal data that in the i6th and lyth-iSth century only 30 and 33% of the cattle respectively reached the age of maturity, while 56 and 57% were slaughtered in the phases A-D (Maltby 1979). In the Roman period the percentage of adult animals was much higher (74%), just as it was in the Eastern River Area. (In working out these percentages I have made use of only those epiphyseal data that were also used for the Eastern River Area.) As mentioned previously, notably the concentration of occupation of the castra and the canabae legionis will have constituted a good market for the beef. The Roman army certainly bought cattle from the local population, as is evident from the deed of purchase for a cow found in Tolsum (province of Friesland) (Boeles 1917; Van Es 1981, note 585). With the production of beef for the market we would expect to find precisely in the castra and canabae cattle of the slaughter-age phases A-D. There is a conspicuous difference between the castra and the canabae legionis (fig. 37A,B). Of the cattle eaten in the castra 25% were slaughtered in phase D, a very small percentage in phase E, and 63% in phase F. In the canabae legionis, on the other hand, we find no cattle slaughtered in phase D, 36% in phase E, and 42% in phase F. This indicates that among the cattle available for slaughter a selection was made according to age that differed in the castra and the canabae. Here it must be pointed out that these differences are not caused by the over-representation of shoulder blades and parts of the head in the canabae legionis (see 3.2.4): the shoulder blades influence only the percentage in phase A, that is 0% for both the castra and the canabae legionis; the bone fragments of the head are not included in the age estimations. The best quality meat, i.e. that of animals slaughtered in the age phases B, C and D, was eaten mainly in the castra; beef of less high quality (E) was eaten mainly in the canabae legionis, while the poorest quality beef (F) was consumed relatively more in the castra. It is possible that these categories correspond to particular groups of consumers, namely army officers, civilian inhabitants of the canabae and ordinary soldiers, respectively. To investigate this possibility it would be necessary to correlate the bone material that provided slaughter-age data with exact locations within the sites, and with pottery and other finds found at the same spot, from which the status of the occupants could be deduced. In the farmsteads the percentage of cattle slaughtered in phases B and C is 29%, which is relatively high compared with the Nijmegen sites. This higher percentage will partly have been the result of death due to natural causes. It is probable that more calves were kept than the average number required to maintain the herd, in order to ensure that enough mature animals would be available. If there were ultimately no setbacks in the form of abnormally high 138 mortalities among the cattle then these surplus animals could be slaughtered. These animals on the farmsteads that died or were slaughtered in phases B and C evidently served as a source of meat for the occupants of the farmsteads. Within the group of farmsteads it appears that in Heteren more cattle were slaughtered at an early age than on the farmsteads in Druten. Also in Kesteren it appears that the cattle were slaughtered at an early age. This could be an indication that these animals were kept for their meat. In view of the small quantity of data available it is not possible to draw any further conclusions. Previously I concluded that Heteren specialized in rearing cattle (see Willems 1984: 216); however, in view of the lack of sufficient data there is no justification for this conclusion. The age data for the cattle from Eist classified as sacrificial animals (see 3.8.2.4) show that these animals, that were killed notably in slaughter-age phase B, must have yielded good quality meat (fig. 36). In this case, which is exceptional for the Eastern River Area, we are probably concerned with calves that were fattened specially for sacrificial purposes. It is unlikely that these calves came from the herds of dairy farmers, as in that case we would expect to find animals slaughtered in age phase A (see above). Also the classical written sources indicate that cattle were not kept primarily for their meat (White 1970: 276-277). In Italy the production of working animals and animals for sacrifice were of primary importance. An argument for this is the scanty attention devoted to beef by Apicius in his cookery book. Another, much more weighty argument is the absence of information on meat production in the writings of Roman agronomists. This lack of information sharply contrasts with the attention given by these writers to the importance of the meat production of sheep and pigs and the importance they attach to cattle as working animals. Traction power and manure The mature age at which most of the cattle were slaughtered indicates that the animals were kept mainly as a source of traction power and manure. When only the supply of traction power and manure are the main objectives of a farmer a high rate of reproduction of the herd is not important and may even be disadvantageous. Columella (VI.24.4) recommends that working cows should only be allowed to calve once every two years, so that the cow does not have to cope with both work and pregnancy. Varro recommends that cows should not be served by bulls before they are two years old, but 'it will be all the better if they are four years old before they bear a calf (II.5.13). These recommendations are given with a view to obtaining good working animals for agricultural purposes. The importance that was attached to good working animals is also evident from the detailed conditions of guarantee that were stipulated when a working animal was purchased. This is in contrast to the trade in slaughtered animals or animals for sacrifice, for which no guarantee of quality was required or given (Hauger 1921). From the writings of the Roman agronomists it is evident that in Italy cattle were bred mainly for the traction power they could provide for agriculture (White 1970). It appears that the same situation prevailed in the Eastern River Area. Obviously cattle were important as working animals since horses could hardly be used for traction. Before the invention of the horse-collar and shafts in the 9th or loth century horses could pull only light loads of 200-300 kg, and thus only cattle were suitable for the heavier work (Slicher van Bath i960). Until the 9th-ioth century notably agriculture was completely dependent on the traction power provided by cows and oxen. Only in the loth century was it possible to use draught horses instead of oxen for the heavy ploughing work, after the introduction of improved harnessing for horses and of the three-course rotation 139 system, which resulted in the availabihty of more fodder for horses. Also the non-agrarian sector, the civilian population and the army, was largely dependent on cattle for traction power. Assuming that cattle were used mainly for agricultural purposes, as described above, we may also assume that in addition to the traction power they supplied the production of manure was of great importance. In the words of Varro (II.praef.5): 'since manure is admirably adapted to the fruits of the earth, and cattle are specially filled to produce it,...'. Cattle, being the most frequently occurring species in the archeozoological material, will have provided the bulk of the manure. If we take into consideration the relative proportions of manure production by farmyard animals: i cow = 2/3 horse = 4 pigs = 10 sheep (Sucher van Bath i960: 321-322), we can state that agriculture was almost completely dependent on the manure production of cattle. In this connection the question arises as to what extent the manure was lost on the pastures. On the basis of historical data the manure production per head of cattle per year for the 18th century has been estimated to be 3000-4000 kg for cattle kept and fed in the byre in winter for ±20 days. For cattle kept in the potstal, i.e. cattle brought into the byre and foddered there not only for the whole winter but also every night during the summer, an annual production of 10,000 kg per head was normal (Slicher van Bath i960: 322). Traces of cutting and sawing on the bones show that the animals were skinned for making leather and that horn was used (see 4.7.3). Bones of cattle were normally not used for making objects. 4.2.4 AGRICULTURE AND STOCK-BREEDING Broadly speaking, the pattern of ages at the time of slaughter of cattle, sheep/goat and pig for the settlements investigated present an overall picture of an agrarian economy based on mixed farming. Cattle were kept neither primarily for milk production nor primarily as a source of meat, even though beef was the kind of meat that was eaten most. The large numbers of cattle, that greatly exceeded the numbers of other farm animals, will have been kept mainly for the supply of traction power and manure that could be used for agricultural purposes. In addition to cattle, smaller numbers of pigs and sheep were kept for the production of meat, while sheep were also kept for their wool and milk. The manure of these smaller farm animals will also have been of economic value for fertilizing the fields. The numerical ratio between the 'agricultural animals', cattle, and the predominantly 'stock-breeding animals', sheep and pigs, shows that with the mixed farming economy there was a strong emphasis on agriculture. Additional support for this is evident from table 61, in which the ratios given are an indication of the ratio of the numbers of animals or of the quantities of meat that were eaten. The ratio between the number of cattle and the number of smaller farm animals that were present at the same time will be even more in favour of cattle. This is because, in view of the pattern of ages at the time of slaughter, the population of smaller farm animals reproduced at a more rapid rate than the cattle herd. In other words: because of the shorter life span from birth to slaughter the number of pigs and sheep that were present at any particular moment produced more individuals, and thus more potential archeozoological material than the same number of cattle. The same reasoning applies to the ratio between sheep/goat and pig. Because the pig population had a higher reproduction rate than that of sheep/goat, the number of pigs that were present at a certain moment will have been smaller than 140 the number of sheep and goats than is indicated in the ratio shown in table 6i. Theoretically, on the basis of the frequency percentages given in table 6i and the data on age at the time of slaughter it should be possible to estimate the relative proportions in which cattle, sheep/goat and pig were present in the settlements investigated. However, seeing that these data are only approximations of the real situation in the past, for which broad margins of error should be reckoned with, such an estimation can be nothing more than an accumulation of uncertainties giving the illusion of exactness. Consequently there is every good reason to refrain from making such an estimation. Naturally this picture of an agrarian economy concentrating on agriculture is applicable especially for the settlements investigated, and then only for those settlements for which relatively many data on the age range of cattle are available. It is not possible to draw any general conclusions for the entire Eastern River Area on the basis of these particular data. The conclusion arrived at by Willems (1984, section 11.3.2) that in the first half of the first century the emphasis in agriculture in the rural areas lay on stock breeding can be neither confirmed nor refuted on account of the lack of sufficient zoological data. At most we can say that the archeozoological data from Nijmegen (Nijmegen Ia and Ib-c) do not point in this direction; this could be explained by the anomalous nature of these settlements in Nijmegen with respect to those in the rural areas. For the mid-Roman sites, the castra and canabae legionis of Nijmegen, and the villa in Druten, the archeozoological data are in agreement with the conclusions drawn by Willems (1984, section 11.4.2) that particularly in the villae farming activities were concentrated on agriculture. This applies to Druten but also to other agrarian settlements, especially villae(?), on which the military settlements in Nijmegen were dependent for their food supply. The archeozoological data from the castra and canabae legionis will reflect not so much farming activities occurring on the spot but rather farming practices of those agrarian settlements in the neighbourhood that produced foodstuffs for the army or at least supplied the army with their surplus farm animals. The fact that after the arrival of the loth Legion the hills in Nijmegen were deforested and used primarily as grazing land, as is evident from palynological data (Teunissen and Teunissen-van Oorschot 1980: 272), is an extra argument for the presence of farms concentrating on agriculture in the surroundings of Nijmegen. In the immediate vicinity of the castra there will have been a great demand for pasture land on which the horses and draught oxen, and possibly also draught cows, of the army could be put out to graze. It seems unlikely that the army used these pastures also for grazing animals for the production of meat, since in that case we would expect much earlier ages at the time of slaughter for cattle in the castra and the canabae legionis. The theoretical possibility that the young animals from the castra and canabae legionis represent animals slaughtered for feeding the army while the older animals were used as draught animals on farms in the neighbourhood concentrating on agriculture, does not seem to be a very plausible explanation. A second theoretical possibility is that the army only consumed beef when its own cattle, used as pack animals and draught animals, died as a result of old age, disease or accidents. As we know that cattle account for by far the greatest part of the meat eaten, this would mean that the population of the castra and the canabae legionis will almost never have eaten (mammalian) meat. Seeing that by modern standards a healthy human being requires about 65 g of animal protein per day (Den Hartog 1972: 154), a quantity that is probably on the high side for a healthy person of the ist century, this second possibility of the army consuming only its own pack and draught cattle must be refuted. As we have seen in section 4.2.2, it is possible that the cattle in Heteren and Kesteren were primarily animals reared for slaughter. To permit definite 141 conclusions about this, however, it will be necessary to obtain more information on the age range of the cattle concerned. Only then will it be possible to shed light on the question as to whether the anomalous settlement form of Heteren, the so-called 'post-hole swarm' type, can indeed be associated with specializiaton in cattle rearing (Willems 1984: 215-216). 4.3 THE POULTRY YARD It is impossible to say to what extent domesticated birds contributed to the meat diet. When bone material is collected by hand the small bird bones will generally be very much underrepresented with respect to the larger, mammalian bones, or they may even be overlooked altogether. A survey of the domesticated and domesticated or wild bird species found per site is given in table 65. The only bird that was definitely domesticated was the domestic fowl, found in the Nijmegen castra and canabae, Nijmegen IV, the castellum in Meinerswijk and in Druten I, as well as in the temples in Eist and on the plates from the 4th-century cemetery in Nijmegen. The domestic fowl was introduced relatively late in the Netherlands. The oldest remains were found in a settlement dating from the pre-Roman Iron Age (± 300 Bc) at Zandvoort (Clason 19675 1977a). The practice of keeping hens in Northwestern and Central Europe only became general during the Roman occupation (Thesing 1977, Zeuner 1967). This spreading habit of keeping hens is thus regarded as a distinctly Roman influence originating from Italy, where various breeds of domestic fowl were kept by professional breeders (see i.a. Varro III.9; Columella VIII.2). The Roman character of the domestic fowl is also emphasized in the Eastern River Area. The highest proportion of the domestic fowl remains was found in the Nijmegen castra: 5.4% or 0.8% by weight. The castellum at Meinerswijk comes in second place with only 0.8% or 0.1% by weight. The oldest remains of domestic fowl may well be the two bones found in the settlement Druten I that is possibly of pre-Roman Iron age date (see 3.7.1 and The occurrence of domesticated, domesticated or wild and wild bird species per settlement (x: presence established). TABLE 65 Nijmegen Nijmegen Nijmegen Meinerscastra canabae IV .wijk domestic domestic fowl wild or domestic Anser anser Anas platyrhynchos Columba sp. wild Anatidae Grus grus Phalacrocorax carbo Asia otus 142 greylag goose mallard dove or pigeon X X X X X " crane cormorant long-eared owl — — - - Druten I Druten II Druten III Eist temple period 4.I.I). On the other hand, the find of domestic fowl in Druten I could be regarded as an extra argument for classifying Druten I together with Druten II (dating Id-IIa). Domestic fowls will have been kept for their meat and eggs. Recipes for both products are included in the cookery book of Apicius. The production of eggs will only have taken place outside the winter season (Columella VI11.5.1; Pliny X.74). This does not alter the fact that eggs could be eaten throughout the year, thanks to methods or conservation (Columella VIII.6). Roman written sources also mention the use of domestic fowl for cock-fighting, as oracular animals and as objects of sacrifice (Zeuner 1967). In sections 3.3.3 and 3.3.4 it has been described how chicken played an important role in the funerary ritual as a meal for the deceased. The remains of goose, duck and pigeon could come from both domesticated and wild birds. Goose was found in the Nijmegen castra and canabae, in Meinerswijk and in Druten III. The Romans ate both the flesh and the eggs of goose. To ensure the development of large goose livers, which were highly esteemed by the Romans, the birds were fattened with a mixture of flour, milk and honey (Pliny X.27). Furthermore goose grease was used for medicinal purposes (Pliny X.28) and goose feathers for stuffing cushions (Pliny X.27). It is also mentioned that the Germanic goose is smaller in size than the Roman goose, and that geese were driven on foot from the region of the Morini, in present-day Belgium, to Rome. Duck was found in the castra, the canabae and in Druten II. Whether the domesticated form of duck was present in Northwestern Europe at the beginning of the Christian Era is not certain (Zeuner 1967). The fenced enclosures used by the Romans for keeping wild duck that had been captured may also have been used for keeping duck that had been bred (see Columella VIII.15). A few bones of pigeons were found in the Nijmegen castra. In view of the size of these pigeon bones they could come from domesticated pigeons (see 3.2.2). Pigeon was an esteemed dish among the Romans. The pigeons were kept in 'columbaria' and were specially fattened. High prices were paid for good breeding stock. Pigeons were not only eaten but were also used as homing pigeons for carrying messages. 4.4 HUNTING In the Eastern River Area both mammals and birds were hunted. Just how important wildfowling was as a contribution to the meat diet of the local people is impossible to say. A meaningful quantitative comparison between the small bird bones and the much larger mammal bones can only be made on the basis of large quantities of sieved material. For the Eastern River Area, however, no such large quantities are available. What we can do is to see to what extent the mammal remains consist of game, i.e. wild animals that were hunted. If we compare the hunted mammals with the domesticated meat-providing mammals cattle, sheep/goat and pig, then we see that generally speaking hunting was of minor importance as a means of providing meat (table 66). To get a clear idea of the importance of game the finds of isolated fragments of antler have to be disregarded. Such isolated fragments could come from shed antlers that had been collected, as has been demonstrated for Heteren I, for example (see 3.6.1). The data for Heteren I show how important it is to take this into consideration: the percentages for game including and excluding isolated antler fragments differ considerably. They are 10.5% and 0.9%, or in terms of weight 14.0% and 2.1%., respectively! 143 TABLE 66 Overview of the bone material found of wild mammals and the percentage of wild species among the meat-providing mammals*. The values between brackets indicate the data without isolated antler fragments. Ilumbei site Nijmegen la Nijmegen Ib-c Nijmegen castra Nijmegen canabae Nijmegen IV Meinerswijk Kesteren Heteren I Heteren II Ewijk I Ewijk II Druten I Druten II Druten III Eist pre-temple Eist temple red deer roe deer 2 (I) 27 (i8) 14 (12) 33 (25) 2 (2) 24 (I) 2 (I) 2 (2) 2 (I) I (0) 4 (2) 17 (i6) of frag•ments percentage wild elk aurochs wild boar hare - - - - - - I I - 2 6 8 - - - _ - I _ - 0/ /o o.2 (O.l) 5 (5) — I (o) 2 4 I I _ - 2.3(1.7) 0.8 (0.7) 4-1 (3-5) 1.7(1.7) 10.5 (0.9) 1.2(0.6) 0.2 (0.2) 0.5 (0.3) 0.5 (0.0) i.i (0.7) 3-4(3.1) weight-% 0.9 3.9 0.7 8.1 0.6 (0.6) (3.0) (0.5) (7.9) (0.6) 14.0(2.1) 4-4 (3.2) 0.8 (0.8) 0,5 (0.4) 0.8 (0.0) 1.4(0.8) 4-5 (4.4) On the basis of hand-collected material with no allocation. Only a few complexes show percentages for game exceeding i % in terms of both numbers and weight. The proportion of game in the legionary fort of Nijmegen is 1.7% (3.0% by weight), higher than in the neighbouring canabae. The difference in the ratio wild: domesticated in terms of numbers is weakly significant (/^ = 6.266, df = i, p < 0.02). Consequently not too much importance should be attached to the difference between the legionary fort and the canabae. The greatest proportion of game in Nijmegen occurs in the 4th century. The quantity of game is significantly greater than for early or mid-Roman Nijmegen (for example with respect to the castra: /^ = 7.616, df= i, p<o.oi). The reason for this increase could be that game became more popular as a source of meat. It is also possible, however, that game animals became more readily available around Nijmegen in the late Roman period. The population in the Eastern River Area had declined sharply by that time (Willems 1984: 142). The town of Ulpia Noviomagus had been abandoned for the most part; only on the Valkhof was a settlement still present (see 3.3.1). Also palynological investigations indicate that the population in the surroundings of Nijmegen declined in the late Roman period (Teunissen and Teunissen-van Oorschot 1980: 272). As a result of this, larger areas of land will have remained unmanaged, permitting an increase in the population of wild animals. The third place where game account for more than i % of the faunal remains is Druten III (3.1%, or 4.4% by weight). A slightly significant increase in the proportion of game with respect to Druten III can be discerned (/^ = 6.260, df= I, p<0.02). No game animals (either mammals or birds) were found in the earliest period in Nijmegen of 10 BC-AD 25, or in the vicus at Kesteren or at the cultic site of Eist. In Druten I only an antler fragment was found. The absence of any evidence of 144 hunting activities at Eist can be associated with the function of Eist as a sacred place. Most of the bone material found at Eist can be regarded as the remains of sacrificed animals (see 3.8.2.3). Evidently game animals were neither used for sacrificial purposes nor were they taken to the sacred place for ordinary consumption. The absence of game in the other places mentioned does not tell us very much, in view of the small numbers of remains found there. It is not impossible that in those places too game contributed to a minor extent to the meat diet of the inhabitants. The most abundant species among the wild mammals is red deer. In all complexes where game is represented red deer occurs. The natural biotope of red deer is on the edge of forest. Yet sometimes red deer are forced to live in open woodland and on areas of heathland (Burton 1976). Roe deer, at present the commonest species of deer in Western Europe, was found only in the legionary fort and in Druten III. Roe deer prefer open forest, but nowadays they can be found almost anywhere where sufficient cover is available (Burton 1976). The elk, of which remains were found only in the 4th-century defensive ditch in Nijmegen, lives in thin forest with a lot of undergrowth including willow, birch and rowan. In the summer it is also to be found in boggy places, while in the winter it often moves to higher and drier areas (Burton 1976). The fact that elk has been shown to be present only in the 4th century is probably accounted for by the more favourable environment for this species at that time. As mentioned above, the area of land in a natural state was probably much larger at that time than previously. In addition the rising water level and the frequent flooding will have resulted in the availability of wetter areas suitable for the elk. In the description of the Hercynian Forest in Caesar's De Bello Gallico (VI.27) there is a fantastic story about elk hunting. According to this story, the elk, since it does not have joints in its legs, sleeps leaning against trees, and once it has fallen down it cannot stand up again. Hunters look for these trees and saw the trunks halfway through. When the elk rests against the tree the tree falls down together with the elk, which is then an easy prey for the hunters. Part of this story may originate in the observation that young elk knock down young trees to be able to get at the leaves and twigs. Remains of aurochs have been found in five complexes: in the canabae legionis and the 4th-century settlement in Nijmegen, Heteren I, Ewijk II and Druten III. During the entire Roman period in the Eastern River Area aurochs appear to have been present, or at least hunted by people in the region. The aurochs became extinct in 1627, and consequently we cannot accurately determine its natural biotope. Historical information, from Conrad Gessner's Historia animalium of 1606 (cited in Requate 1957) and archeozoological research (Degerbol and Fredskild 1970; Von den Driesch and Boessneck 1976) give us an idea of its biotope: open deciduous forest with oaks and with grassland in the vicinity. Gessner also mentions that in the summer the aurochs visit farmland and cause much damage there. In the above-mentioned story about the Hercynian Forest it is stated that the Germani catch aurochs in pitfalls, after which they kill them (Caeaar IV.28). The presence of wild boar has been demonstrated only in 4th-century Nijmegen and in Druten II and III. The natural habitat of wild boar is open deciduous forest (Burton: 1976). One fragment of brown hare was found in the Nijmegen castra. The brown hare is able to occupy a variety of habitats, including open ground, heath, farmland, rough pastures, marshes and woodland (Burton 1976). If we look for the common denominator of the biotopes of the hunted animals then we find this to be deciduous or mixed forest. Such forest will have been 145 situated mainly on the Pleistocene soils of the Rijk van Nijmegen and the adjacent Reichswald, North Brabant and the Veluwe. In the summer the low-lying damp region in the middle of the Eastern River Area will have provided a suitable habitat for the elk. The predominance of red deer and the relatively scarce occurrence of the roe deer, for example, suggest that in Roman times the forest vegetation was still very dense. The red deer prefers to live on the edge of thick forest, while the other species feel more at home in more open woodland. Naturally the relative abundance of red deer in the settlements could also be explained by a special interest in hunting red deer, for example for their antlers. Moreover the hunting of these animals did not necessarily take place in the above-mentioned forest regions. All these wild species frequent cultivated land to feed on crops. Thus the remains of game animals in the settlements are not necessarily indicative of pleasant hunting forays but could just as well be the result of the farmers' efforts to protect their crops from their voracious competitors. As mentioned at the beginning of this section, the material provides little information about wildfowl. Table 65 gives an overview of the various species that have been found at the different settlements. The group of wild or domesticated birds includes the greylag goose, Anser anser, the wild or tame duck, Anas platyrhynchos, and a pigeon of indeterminate species. The greylag goose breeds on moorland and in marshy areas. At the present time this species breeds rather seldom in the Netherlands and occurs only as a migrant in fairly large numbers. The wild or tame duck, on the other hand, is still the most common and widely distributed species in the Anatidae group, and breeds near marshes and still or slow-moving water (Heinzel et al. 1972). In view of the measurements of the pigeon bone found in the castra the species represented could be either the stock dove {Columba oenas), the rock dove (Columba livia) or the domestic pigeon that is descended from it (see 3.2.2). The other three species of bird, that are certainly all wild, were all found in Druten. The crane, Grus grus, a species that is now threatened with extinction (Austin and Singer 1975), breeds in Northern Europe and spends the winter particularly in subtropical regions. Thus it now only occurs in the Netherlands during its migration: in the autumn from mid-September until November and in spring from the end of February until April (Wessels 1978). Yet the oldest known document (± 1635) on hunting in the Dutch language, the 'Jacht-bedrijff' by Cornells Jacobz. van Heenvliet, indicates that the crane was a breeding species also in the Netherlands at that time (Swaen 1948): '..., because they [i.e. bustards] used to feed and breed here in the natural bogs, like the Cranes, but since the country has become so populated there are none remaining...' (p. 67). '[Cranes] also used to live here in the natural bogs,... and that is the reason that both species have been put in the proclamations, since the boglands have become so populated, and none remain...' (p.74). The cormorant, Phalacrocorax carbo, can be present throughout the whole year. It is a bird of coastal regions, though it also occurs on lakes inland. Finally, the long-eared owl, Asio otus, occurs in the Netherlands both as a breeding species and as a migrant. It lives in forests, in areas with scattered trees, on heaths and marshes. It is almost certain that most of the animals of which remains were found had been eaten. They were found in the refuse pits of the settlements, and moreover all kinds of birds were relished in the Roman world. In the cookery book of Apicius Caelius nearly twenty species of birds are mentioned including goose, duck, various species of pigeon, and crane (see 3.3.4). Varro (III.2.14) tells of great numbers of geese, pigeons, cranes and peacocks kept in captivity at the villa of Seius. And Pliny (X.30.60) mentions for instance the increasing popularity of 146 The occurrence of fish in the hand-collected material plus a few sieve-samples from Nijmegen (x: presence established). TABLE 67 Nijmegen Nijmegen Nijmegen Ib-c castra canabae Esox lucius Tinea tinea Leueiseus cephalus or Leuciscus idus Leueiseus cephalus or Blieea bjoerkna Scardinus erythrophthalamus Cyprinidae unidentified Anguilla anguilla Perea fluviatilis Siluris glanis Alosa alosa or Alosa Jallax Salmo salar cf. trutta Sphyraena sp. unidentified I*: 2*: 3*: 4*: 5*: Meinerswijk sieve-samples from Nijmegen - pike - tench - chub or orfe - chub or white bream - rudd - eel perch European catfish allis shad or twaite shad - Atlantic salmon or sea trout - barracuda - X - samples of sieving experiment (see 2.1.3); mesh width 5 mm; date Ib-c. samples with find number Nijm.172/13; mesh width 2.5 mm; date Ib-c. samples with find number Nijm.173/1 ; mesh width 2.5 mm; date Ib-c. samples with find number Nijm.155/16; mesh width 5 mm; date Id-IIa. samples from 4th-century ditch (see 3.3.2.3) mesh width 2.5 mm; date IV. cranes as table birds. As for the owl, it is unlikely that this had been eaten; this find probably represents a natural fossil that became preserved accidentally. Alternatively, we may be concerned with an owl that had been killed or captured, possibly for use as a scarecrow on newly sown arable land. 4.5 FISHING In the Eastern River Area, where there is such an abundance of water, fishing will certainly have taken place. This is testified by the finds of net-sinkers (Willems 1981a; 197-201). Archeozoological evidence for fishing in the region is extremely scarce. In the hand-collected material from the settlements investigated the total number of fish bones found was only 6: in the Nijmegen castra and the canabae legionis and in Meinerswijk (see table 67). An additional hand-collected find was that of a large bone of a European catfish, noticed incidentally in part of the early Roman settlement on the Valkhof and surroundings, Nijmegen Ib-c, that was otherwise not investigated archeozoologically (find no. Nijm. 40/12). As has been discussed earlier (2.1.2), only sieved samples can provide meaningful information on fishing practices and the consumption of fish. Thus most of the fish data come from sieved samples (table 67, sample groups 1-5). Only the samples of group 5, from the large 4th-century defensive ditch in Nijmegen, come from archeological features from which also the hand-collected material was studied (3.3.2.^5). The 147 Fig. 38 European catfish: vertebra II-IV and right and left os Suspensorium (Nijmegen 40/12). •. ,3 other samples come from excavation trenches from which the hand-collected material is not dealt with in this study. Nevertheless the data on fish from these samples have been included to give a more complete picture of fishing practices. The samples of group i are discussed in section 2.1.3. The fish remains were identified, or at least their identification was checked, by D.C. Brinkhuizen (BAI). All fish remains that could be identified to species level or as one of two possible species were found to be fish species that still occur today in the upper course of the Waal. They are all present in the list of 34 indigenous species that were caught during research carried out in 1962/1963 and 1970/1971 in the Waal and waters in the Waal basin in the region between the point where the Rhine splits up and Nijmegen (Oomen and van Wijck 1978). The fishing tackle used in this research included stow net (on anchors), fyke nets, various kinds of seine nets, lift nets and scoop nets. Also included in the list given by Oomen and Van Wijck are a few species that were not caught in the course of the research but that had been caught in the same waters after the Second World War, according to the professional fisherman who was involved with the research. One of these species is the European catfish, of which only one specimen had been caught, in 1969, probably a fish that had been put out in the mid-sixties. The bone fragment of a catfish that was found in early Roman Nijmegen (Nijm.40/12) consists of the fused vertebrae II-IV (the Weberian ossicles) plus the right and left os Suspensorium (fig. 38). D.C. Brinkhuizen (BAI) has estimated that this bone fragment came from a fish that must have measured about i m in length and weighed 8-10 kg. Although it used to be thought that the European catfish was an exotic species introduced by monks in the Middle Ages, archeozoological research has proven that the catfish is an indigenous species in the Netherlands, where it can attain a maximum length of 2 m (Brinkhuizen 1979b). The majority of the fish mentioned in table 67 are freshwater species. Only those of the genera Alosa and Salmo are anadromous species, that seek out fresh water to spawn (Nijssen and De Groot 1980). One find was made of an exotic fish, belonging to the genus Sphyraena, that could 148 not be identified to species level. The fish-bone concerned is a fairly large vertebra ± i cm long (fig. 39), that was found in the 5-mm sample of find no. Nijm. 812/6, from the large 4th-century defensive ditch of Nijmegen. On the dorsal surface of the vertebra there is a distinctive tube (the neural arch), in which respect it resembles vertebrae of the eel. The vertebra could not be identified with the aid of the comparative collection of the BAI, which contains at least all the indigenous freshwater species of fish now occurring in the Netherlands as well as all marine fish that are brought into the Netherlands (see Clason, Kosters and Jacobs 1983). Consequently D.C. Brinkhuizen (BAI) came to the conclusion that the vertebra must come from an exotic species. A more detailed investigation by Dr. K. Rosenlund (Universitetets Zoologiske Museum, Copenhagen) has led to the provisional conclusion that it could be a vertebra of a fish of the genus Sphyraena. It was not possible to identify the bone to species level with the aid of the Copenhagen Museum collection. All species belonging to the Sphyraenidae family are predatory fish living in warm seas, almost all of which are regarded as excellent and tasty fish for the table (Herald 1970; Müller 1983). One of the species that could be the one represented in Nijmegen is Sphyraena sphyraena, that occurs in the eastern Atlantic Ocean, the Mediterranean and the Black Sea. Rosenlund (personal communication) mentions 5. guachancho and 5. viridensis as alternative possibilities. In any case the vertebra from Nijmegen IV certainly comes from an exotic species of fish. If we assume that this vertebra was not brought, as a bone, unintentionally from elsewhere, then this single find suggests that fish products may have been brought from faraway southern coasts. Seeing that the transport of fresh fish over great distances is unlikely to have occurred, we should be thinking in terms of fish Fig. 39 Vertebra ofa fish of the genes Sphyraena (?). A dorsal, B lateral, C ventral, D cranial »li 4cm 1 149 products like liquamen, or more probably the residue of the liquamen preparation, allée, or possibly also salted fish (Forbes 1965). Taking heed of the warning given by Wheeler (1978), not to make too much out of too few data, we should not attach too much importance to the find of this single exotic bone. 4.6 MOLLUSCA Shells of oysters (Ostrea edulis) were found in a few settlements, namely the Nijmegen castra, the castellum of Meinerswijk, the military vicus of the castellum in Kesteren and the villa at Druten (Druten II and III). Oyster is also represented in the sieve samples of 4th-century Nijmegen. The fact that oysters have been found only in a military context and at the Romanized villa of Druten could indicate that in the Eastern River Area the oyster is a typically Roman element, that would not have been eaten by the native population as a general rule. In fact it is debatable whether we can justifiably speak of a 'typically Roman element' with regard to the 4th century, in view of the late point in time within the Roman period and the absence of comparative material from late Roman times. In the Roman period oysters were often transported over great distances (see for example Thiiry and Strauch 1984). No attempt has been made to ascertain the place of origin of the oysters on the basis of the epifauna or isotope analysis (Duncan Waugh 1965; Strauch and Thiiry 1985). Clearly the oysters must have come from along the Dutch coast. Another representative of the coastal area is the common whelk (Bueeinum undatum), found in Kesteren. This species of mollusc is edible too. In the course of the sieving experiment a mussel was found in Nijmegen Ib-c. This could not be identified to species level, and could have been either a marine or freshwater mussel. The few mollusc finds show that shellfish suitable for consumption were brought to the Eastern River Area from along the coast. Oysters were probably not eaten by the native population and must be regarded as a Roman element in the local diet. 4.7 THE SLAUGHTERING OF LIVESTOCK AND THE PROCESSING OF MEAT To obtain information about butchery practices and the processing of meat the butchery marks present on bones from a few of the settlements were recorded. The method used to record this information is described in section 2.2.9, 'butchery marks'. The butchery marks are described in the appendix called 'Butchery mark code'. When in this section reference is made to the butchery mark code, the code number of the mark is given between square brackets: [no.]. Butchery marks were recorded for bone material from the settlements Nijmegen la, Nijmegen Ib-c, Nijmegen castra, Nijmegen canabae, Druten I, II and III, Heteren, Kesteren, Meinerswijk and Eist. The aim of this investigation of butchery marks is to make a preliminary inventory of the information that the butchery marks can provide about the butchering of livestock and the processing of various animal products in these settlements. To carry out a thorough investigation of butchering technique and the processing of animal products it is necessary to take into consideration also the fragmentation of the bone. Most of the studies that have devoted attention also to butchery techniques are based mainly on these fragmentation data. As the material from the Eastern River Area is described according to the 'Knocod' 150 TABLE 68 Overview of numbers of bones with butchery marks per skeletal element per species. species skeletal element horn-core/antler skull mandibula hyoid scapula humérus radius ulna metacarpus pelvis femur tibia fibula astragalus calcaneus metatarsus phalanx fragment of long bone* atlas axis vertebrae costa unidentified fragment* bones with marks total number marks marks per bone number of bones "/„ bones with marks no size assignment sheep cattlepig horse size size horse pig sheep/ goat 13 6 cattle aurochs red deer roe deer 18 5 73 3 138 64 36 7 - I II - I - - " " 20 I 25 1.25 2 I 13 8 69 28 6 10 34 II wild boar 13 I I 3 I I 9 2 22 I 40 16 28 16 98 184 3 - 2 - — I 8 155 I 56 — — -~ ~ 8 8 56 56 292 84 90 36 41 780 I 314 32 37 1012 I I I 1.00 1.00 1.08 1.16 1.07 1.14 1.30 I.DO I.GO 310 749 7 3136 1020 700 5 2 3 934 9 5568 9 14 50 6 17 7 3 93 22 — — 2.00 6 17 unclassified butchery marks. system of Uerpmann (1978), the data on the fragmentation are available for closer investigation. Other sources of information that can make a valuable contribution to a study of this kind are historical written and pictorial sources relating to butchering and processing, as well as handbooks, dating from later historical times, for the butchers', tanners' and bone-workers' trades. Also grave gifts in the form of food can provide information. In carrying out such a study one can focus attention on the butchering techniques in general, but also on such aspects as differences between civilian and military complexes, between urban and agrarian complexes, and any changes that may have occurred with regard to butchering techniques and the use of animal products in the course of time. Additional interesting information can be obtained if the age and, where possible, also the sex of the animals are taken into consideration. Such a thorough study can provide a more detailed picture of the use of animal products and the reason why these animals were kept, and can give us greater insight into the economy of the settlements investigated. Table 68 gives an overview of the numbers of bones with butcherv marks per 151 TABLE 69 Butchery marks found on bones of site skeletal element mark no.* Nijmegen Ib-c radius 16 Nijmegen castra humérus metatarsus III 7 16 iiuxacs. 17 29 ** 30 Druten I pelvis astragalus Druten II mandibula pelvis tibia Druten III scapula humérus pelvis femur tibia 14 16 I 16 20 29 31 17 ** 33 9 24 15 2X 17 27 2 X 18 5 26 metatarsus III 7 Heteren Kesteren pelvis 16 16 ** 29 tibia astragalus 5 14 20 8 mandibula pelvis phalanx I 16 16 2 * see appendix 'Butchery mark code' ** marks on one and the same bone >^ne>i. 152 skeletal element per species. In the table the data for sheep (2 horn-cores and i scapula with butchery marks) are included with the data for sheep/goat. Among the bonesldentifiable to species level, in the 'sheep-pig size' group the only bones with butchery marks present are ribs. These data will belong to pig and/or sheep/goat. In the 'cattle-horse size' group the bones on which butchery marks were found to occur are vertebrae as well as ribs. Seeing that the number of bones identified as horse is small with respect to the number identified as cattle (i8% with respect to cattle), and that horse bones are generally less fragmented and thus more easily identifiable, while the percentage of horse bones with butchery marks is much smaller than that of cattle bones, it is very likely that almost all the bones with butchery marks in the 'cattle-horse size' class come Fig. 40 Butchery marks found on bones of horse. I chopped through, 2 sawn through, 3 chop mark, 4 cut mark, 5 chop or cut mark on side not visable. 1 s -^ 3 r> 5 from cattle. This is especially so with ribs, as ribs of horse are relatively easy to recognize. 4.7.1 DOG None of the 68 bones of dog found showed any traces of butchery. Seeing that the dog bones that were found were often complete, it is unlikely that any products of dog were used after death. 4.7.2 HORSE On 32 out of the 1020 bones one or more butchery marks were present (table 69). No butchery marks were observed for Nijmegen la, Nijmegen canabae, Meinerswijk (no horse bones present) or Eist. Fig. 40 provides an overview of the butchery marks on the various skeletal elements. To ascertain whether horse meat was used for human consumption in the settlements investigated we can compare the data for horse with those for cattle. This approach is based on the assumption that animals of about the same size are slaughtered and their carcasses cut up in more or less the same way. In the first 153 TABLE 70 Nijme- Reste- Druten Druten Druten ren III gen la I II The occurrence of gnaw marks made by dogs on bones of horses*. mandibula scapula humérus radius ulna metacarpus pelvis femur tibia astragalus calcaneus metatarsus phalanx I phalanx II Total _ - _ - Beteren total I 3 _ - _ - _ - I I I 3 - 4 2 - I 2 I 4 4 2 I 10 I 2 - I I I I 3 4 - - - I - I I - 6 8 5 I I - - - 7 4 II I I 3 2 3 - - I 2 I 3 5 I - I 20 10 53 I 3 * Only the settlements for which the butchery marks have been studied; in Nijmegen Ib-c, castra and canabae no gnaw marks were found on bones of horses. place it is noteworthy that no butchery marks occur on the vertebrae and ribs, skeletal parts that are divided into smaller pieces when large animals are butchered for human consumption. This is in contrast to cattle: on ribs of cattle butchery marks were frequently observed (table 68). As mentioned above, probably almost all vertebrae with butchery marks and certainly the ribs in the 'cattle-horse size' class are of cattle. If we compare the percentages of horse bones and cattle bones with butchery marks, we see that butchery marks occur on cattle bones about 5 times as frequently as on horse bones. Also the number of butchery marks present on bones is smaller in the case of horse. Finally the degree of fragmentation of horse bones is less than that for cattle (table 75). From these observations it has to be concluded that horse meat, in contrast to beef, was not used for human consumption. The question remains as to how the butchery marks present on the horse bones can be explained. The butchery marks on the metatarsus among the bones from the castra have nothing to do with butchering but are the result of bone-working. One bone, a diaphysis fragment, has been sawn through at the proximal end of the diaphysis [16], while another is a distal fragment that has been sawn off the distal end of a diaphysis [17]. Moreover, from the latter bone the dorsal and the plantar side of the distal articulation surface have been struck off [29], [30]. Of the remaining 11 butchery marks observed on the hind legs 7 are cut marks. All of these cut marks are present in the area from the diaphysis of the tibia down to and including the proximal epiphysis of the metatarsus: tibia [14], [20], [26], [31]; astragalus [8]; metatarsus [i], [7]. As this is precisely the part where the leg becomes narrower and thus the surrounding skin tighter, it is reasonable to assume that these cut marks occurred in the process of skinning the animal, when the skin was cut loose from the leg. The only chop mark in this part of the skeleton is an astragalus [i] that has been chopped through. A part of the skeleton where chop marks occur frequently is the pelvis [14], [16] X 4, [17], [27] X 2, [29] X 2. Evidently the hind leg was intentionally detached from the trunk. This may have been done to make it easier to bury the horse 154 carcasses in a pit or ditch, notably after rigor mortis had already set in. It is also possible that parts of the legs were fed to dogs, for in almost all of the settlements gnaw marks made by dogs are present on bones of horse (table 70). In a few cases horse bones show both butchery marks and gnaw marks: in Druten I two pelvis fragments [14], [16], in Druten III a pelvis [27], and in Heteren a pelvis [16] and a tibia fragment [20]. It is possible that we should regard butchery marks on the foreleg bones also as the result of cutting loose large pieces of bone with meat attached for feeding to dogs or of cutting off (parts of) the leg to facilitate burial. The cut marks on the scapula [33] from Druten III are noteworthy. With cattle this type of butchery mark is possibly associated with the smoking of meat. To draw the same conclusion here on the basis of a single find is unjustifiable. Moreover such a conclusion would be in disagreement with the clearly deduced fact that horse meat was not eaten. In Kesteren and Druten II there are indications that the lower jaw was removed from the skull [16], [20]. 4.7.3 CATTLE In all the settlements investigated butchery marks have been found on bones of cattle. Table 71 gives an overview of the frequency of occurrence of the various butchery marks on the skeletal elements. As it has been pointed out above, we may assume that most of the data relating to bone fragments of 'cattle-horse size' should be attributed to cattle. Comparing cattle and horse, we may conclude that the meat of cattle, in contrast to horse meat, was used for human consumptiçn. In contrast to horse, with cattle butchery marks occur on all parts of the skeleton. Moreover the percentage of bones with butchery marks is higher and bones with more than one butchery mark occur more often. Cattle bones are also more fragmented (table 75), which means that the difference between horse and cattle in terms of the percentage of original, complete bones with butchery marks is even greater than what is indicated in table 68. The large number of butchery marks present on the ribs is very characteristic of the pattern of meat consumption. The ribs are often broken up into pieces [7], indicating that the meat attached to the longer ribs was divided into portions. Also among the bone fragments of 'cattle-horse size' it is mainly mark [7] that occurs on the ribs. Also most of the vertebrae can be viewed in this light. A few horn-cores show traces that indicate that the horn-cores were struck off [i], [3]. This was probably done to make use of the horn. The saw mark [8] that occurs three times also points in this direction. If one wants to saw off a piece of horn neatly then one can do this more easily and with less risk of fracture by sawing off the horn while it is still attached to the horn-core; the horn-core then functions as a support. A group of cut marks occurring regularly on the bones indicate that the cattle were skinned. These include in the first place the relatively frequently abundant marks [i], [3] and [7] at the proximal end of the metapodials, in particular the metatarsus. Mark [15], that occurs relatively often on the metapodials, can be associated with the practice of smashing the bone to permit the extraction of the marrow. In view of the frequent occurrence of complete metapodials it is evident that marrow extraction from the metapodials was not a regular part of the butchering procedure. Also the marks [8], [14] and [16] occurring on the calcaneus could be indicative of skinning the carcass; the same can be said of the cut marks [6], [8] and [12] on the astragalus and the cut marks [23], [27] and [35] on the distal part of the tibia. On the mandibula butchery mark [4] occurs especially, a mark that is produced when the almost meatless oral part of the lower jaw is struck off. The chop marks 155 Butchery marks found on bones of cattle. TABLE 71 mark no.* I horn-core skull mandibula hyoid scapula humérus radius ulna metacarpus pelvis femur tibia astragalus calcaneus metatarsus phalanx I atlas axis other vertebrae 6-8---i3-------___ ---1---1--1 !______ Costa 2 3 4 5 6 7 8 9 10 II 12 13 14 15 16 17 18 -3 126--2-41-1-298-1 I 2---------------13743-2337--17942--621----3-1-15822615 27183-1-1-27-2---I-3--2--I--I-----_ --4----------1 6--4848173-1---33399--71-51------1--11 2413-1--7231----2I2-I-5-I---ii-____ 62-1 1--26-2--2-2-10-124-12--121-17--- -6-5-3---3----____ -7-62-521 I 1---5222 33-2-2-223-5-1---5i--i------_______ 3--6-- 148 ----------- ''see appendix 'Butchery mark code' [i5]j [16], [18], [19] and [21-24] were produced when the lower jaw was removed from the skull. The cut marks [i] and [2] on the hyoid were probably produced when the tongue was cut loose. On the scapula a very large number of butchery marks were observed. This phenomenon has already been discussed in section 3.2.4 in connection with the concentration of shoulder blades that was found in the western canabae. To summarize, we can say that the observed butchery marks [i] - [23], and possibly also [24] - [30], can be associated with the removal of the shoulder blade from the pectoral girdle, mark [40] with the hanging up of shoulders of meat for the purpose of smoking and marks [32] - [38] with the cutting away of the meat that had become very firmly attached to the bone as a result of the smoking process [see 3.2.4]. The butchery marks that are indicative of the smoking of shoulders of meat occur to a greater or lesser degree in almost all of the settlements investigated. The butchery marks on the long bones, apart from those that have already been discussed above in connection with the skinning of carcasses, can mostly be divided into marks that are indicative of the removal of the various parts of the skeleton, the removal of meat from the bone and the dividing up of the meat into portions, and the smashing of the bone to permit the extraction of the marrow. Traces of the removal of the various parts of the skeleton are represented by the following butchery marks: humérus [i], [2], [16], [18], [22], [29-36]; radius [1-5], [12], [14], [16], [33], [34]; ulna [i], [3], [6], [9]; femur [3], [4], [28], [29], [32], [36]; tibia [1-4], [27], [32], [33], [35-38]. The butchery marks that are referred to as 'shaving marks', superficial traces of the removal of bone shavings, were probably produced when the meat was being boned. With butchery marks of this kind it is impossible to see whether they were produced by a knife or a meat axe. Butchery marks that can be associated with the boning of meat are: humérus 156 19 20 21 22 23 24 25 26 27 28 29 3° 31 32 33 34 35 36 37 3« 39 4° total _____________--------- i8 ---------------------- 4 8-35422-1---1--------- 84 ---------------------- 3 3513--1----2-2-10287 I 14 -1-6 2-21-3-2--1655151031---4-2----2-1----12------ 254 93 46 ___________---_-_----- 8 ___________-__--_----- II 65--1-163-1----------3---11---11--1--22----2121---1---116-7612-__________------------ 86 29 57 12 -----------------_-_----!-_-----------_--_-_---_------------ 24 42 17 - - 36 --------------------!---_-------------------------------------- - - _ - - - - - - - - - - - - - - - - - 23 8 157 [12-15]; femur [19]; tibia [9-11], [21-23]. This category may possibly also include the chop marks on the diaphysis that have not resulted in the bone becoming chopped right through: humérus [7], [9]; radius [19], [21]; femur [17-18]; tibia [6], [17]. In some cases the diaphysis had indeed been chopped right through; there may be various explanations for this. Firstly, in the process of boning the butcher may have struck so hard that the bone broke. Equally, it may have been the butcher's intention to break the bone in two, together with the meat attached, in order to divide the meat into portions. Finally, it is possible that the bone was smashed to allow the extraction of the marrow. Butchery marks that could be accounted for by these various explanations are, among others: himienis [11], radius [26], tibia [12]. The butchery marks that were observed on the pelvis can all be associated with the cutting away of the pelvis from the skeleton or the dividing up of the meat attached into portions. 4.7.4 SHEEP/GOAT With the exception of Meinerswijk and Druten III, where only 8 and 17 bones of sheep/goat were found respectively, in all of the settlements investigated butchery marks occur on the bones. The frequency of occurrence of the butchery marks on the different parts of the skeleton is indicated in table 72. Some of the ribs of 'sheep-pig size' will certainly have come from sheep/goat, and some from pig. In contrast to cattle only a few data on butchery marks are available. Most of the butchery marks were the result of the removal of the various parts of the skeleton: mandibula [15], [16]; all marks on the scapula; humérus [16], [22], [24]; radius [12]; pelvis [i], [5], [20]; femur [25]; tibia [31]. A few butchery marks on the middle of the diaphysis of bones that were not chopped right through indicate 157 Butchery marks found on bones of sheep/goat. TABLE 72 mark no.* horn-core mandibula scapula humérus radius metacarpus pelvis femur tibia metatarsus atlas costa I 2 3 5 6 7 9 12 13 15 16 17 19 20 -I I I - 2 - 22 24 25 26 28 total 31 1--121- I -----I--- 3 - I I * see appendix 'Butchery mark code' that the meat was divided into portions: humérus [7], [9], tibia [15], [17]. Also the humérus that was chopped right through [12] fits in with this picture. The radius with shaving marks [28] and the rib with a cut mark [6] probably indicate the cutting away of meat from the bone. Among the ribs of 'sheep-pig size' cut mark [6] or the chop mark variant [5] occurs five times. Most of the butchery marks of 'sheep-pig size' indicate that the meat was cut up into portions [7] x 45. On the metapodials mainly cut marks occur, at the proximal end [i], [3], [5], once on the middle part of the diaphysis [13], and once on the distal part of the diaphysis [24]. As with cattle this is probably indicative of the cutting loose of the skin for the removal of the hide. Also the cut mark on the distal end of the tibia [31] can be interpreted likewise. On the horn-cores there are two instances of a butchery mark at the base. In one case the horn-core has been struck off [i], while in another case it has been sawn off [2], indicating that the horn was required for some specific purpose. Butchery marks found on bones of pig. TABLE 73 mark no.'* I mandibula scapula humérus radius ulna pelvis femur tibia fibula astragalus calcaneus metapodial atlas other vertebrae ---------1--1-----11-2------1---1--1--- * see appendix 'Butchery mark code' 158 2 3 5 6 7 8 9 10 II 14 15 16 17 18 -----2-3-1--1-1 19 20 23 25 26 1---1 ---------I 1----2--2-I I _______________ --2--1-1---141--1-21 --------1----1---1 I- ----I 1---1-3-2--1--I ___________________ ______ _I ___ I I _____________ !_________________ I ________________ __________ I ________ TABLE 74 Butchery marks found on bones of red deer. " mark no.* i 2 3 4 5 6 7 8 18 21 27 31 32 total antler scapula humérus radius. metacarpus metatarsus 12-1-147----________---i------1-111-__________i-i __i_______--____j______-- 16 I 4 2 i i * see appendix 'Butchery mark code' 4.7.5 PIG Table 73 shows the butchery marks that were observed on the bones of pig. With the exception of Druten I and Heteren, where only 17 and 15 bones of pig were found respectively, in all the settlements investigated butchery marks were found to be present. As with cattle and sheep/goat most of the butchery marks are indicative of the cutting away or chopping up of the various skeletal elements. Only a few butchery marks indicate that the meat was divided into portions or boned. Compared with the bones of cattle the bones of both sheep/goat and pig show few butchery marks that are associated the dividing up of the meat into portions. This is explained by the fact that the bones of these animals are much smaller than those of cattle, and thus it is not always necessary to chop the bone with the meat attached into pieces. With these smaller species of animal the bone material is thus fragmented to a much lesser extent than the bones of cattle. 4.7.6 WILD MAMMALS 28 29 30 31 40 - 7 - I total - II 1-222 13 - 14 I I 2 - 4 - 6 2 15 9 ^^ ^ In the settlements investigated the most abundant species of wild animal represented by the bone remains is red deer (93 bones). The most frequently occurring skeletal elements of red deer are antler fragments, partly because also isolated antlers were collected (see 4.4 and 3.6.1). Consequently most of the butchery marks were found to occur on these parts of the skeleton (table 74). Antler of red deer is used as a raw material for bone-working. This is indicated by the finds of shed antlers from e.g. Heteren (see 3.6.1), but especially by the fact that most of the marks present are saw marks: [2], [4], [6], [8]. Such sawn-off fragments of antler were also found in Nijmegen IV (3.3.2.2). Also mark [i] indicates that the antler was chopped off the skull, and mark [7] that the antler was divided into pieces, probably for further processing. A few objects made out of antler were also found (fig. 41). The butchery marks on the scapula, the humérus and the radius [27] indicate that the skeletal elements were cut away from the carcass and that the meat was divided into portions. The cut marks on the distal part of the radius [32] and on the proximal part of the metapodials [3], [5] indicate that the skin was cut loose here to facilitate the removal of the hide. Of the six bones identified as roe deer there is one, from the Nijmegen castra, on which butchery marks are also present. This is a femur with the marks [6] and [16], cut marks at the proximal end, that were probably inflicted when the femur was cut away from the pelvis. 159 2 cm Fig. 41 Knife-handle and die made of antler of red deer. On a scapula fragment of an aurochs from Druten III the butchery mark [23] is present; this was probably produced in the process of separating the scapula and humérus. Also one of the two bones of wild boar found in the settlements investigated shows butchery marks. On the scapula fragment from Druten II chop marks are present on the distal part of the margo thoracalis [12], that can be associated with the cutting loose of the scapula and the humérus. 4.8 THE MENU Up until now, whenever food consumption has been discussed only animal products have been mentioned. An attempt has been made to gain insight into the relative proportions of the various species in that part of the local diet constituted by animal products. Concerning the mammals that were used for human consumption, namely cattle, sheep, goat, pig and the wild species, it became evident that there are many uncertain factors that play a role in the determination of the individual contribution of each species to the menu. For fowls, fish and shellfish it was even impossible to make an estimate. A question that may be of greater importance is to what extent animal products contributed to the total menu. In other words: what was the ratio between animal and vegetable products? Archeozoological research, even in combination with archeobotanical research, can only give very vague indications of this ratio. For the Eastern River Area, on the basis of the data on age at the time of slaughter of by far the most frequently eaten species, namely cattle, we may conclude that cattle were kept primarily for traction power and for the supply of manure. The conclusion that follows is that agriculture was evidently more important than keeping cattle for the supply of milk and meat. For the composition of the menu this suggests that meat probably played a very minor role in the daily diet. In the future we will probably be able to obtain a clear answer to such questions by means of chemical research, for example the sterol analyses of human faeces found in archeological contexts (Knights et al. 1983), and the investigation of trace elements in human and animal bone (Runia 1985). As far as the Roman army is concerned we may conclude from the written sources that grain products, mainly wheat, formed the most important part of the 160 diet (Webster 1969: 254-255; Davies 1971). Webster concludes that meat was eaten on special occasions, on high days and holidays, but was not regarded as a daily requirement. There may have been an increase in the consumption of meat as result of the influx of Teutonic elements in the army, notably in the third and fourth century. Davies (i971), on the other hand, concludes from the evidence provided by written sources that in peacetime the basic diet consisted of grain products, bacon, cheese and probably vegetables, with sour wine to drink and with salt and olive oil as condiments. Moreover he maintains that contrary to popular belief the Roman army in the Empire ate meat at all times as part of its diet. Concerning the diiferences in the conclusions drawn by Webster and Davies as regards how frequently meat was eaten I am unable to form my own opinion on account of my lack of knowledge of the written sources. The conclusion drawn by Davies that bacon formed part of the basic diet does not seem likely in view of the archeozoological evidence. From the surveys given by King (1984) of the bone material found in eighty British and Continental military sites it appears that in general cattle was the most abundant species present. In 42% of the British military settlements the frequency percentage of the cattle bones is 70% or more, while in the military settlements on the Continent the frequency percentages for cattle are generally considerably higher than those for pig. If we assume that frequency percentages for cattle represent far too low an estimate of the meat percentage then this means that on the basis of the bone data beef rather than pork belonged to the staple diet of the Roman army. Naturally it is possible that preserved boneless pork was brought into the military settlements from outside. After all, the preservation of pork by means of salting, smoking and drying was not unusual (Hauger 1921: 106-108), and such preserved pork was a normal commercial product (White 1970: 320-321). If the supply of such preserved pork was indeed a more important part of the diet than beef, then this supply of pork must have been organized on a big scale, certainly if pork featured on the daily menu. If this was the case, then in the written sources on the Roman army one would expect to find in addition to the frequent references to the grain provisions also frequent mention of the supply and storage of pork. Yet the sources referred to by Davies (1971) indicate that the supply of meat consisted of live pigs and other animals that were slaughtered by the army itself. In other words, these supplies, including those of pork, would have left their traces in the form of bones from which the relative importance of the different species for the diet could be estimated in terms of percentages of bone weights and frequencies. It is true that for the Nijmegen castra these estimates (see 4.1 and 4.2) do show relatively high pig percentages with respect to the region as a whole, but even here cattle are still by far the greatest source of meat. To conclude, we may state that, contrary to what has been suggested by Davies (1971), in the military settlements in the Eastern River Area, and probably also in other military settlements where high percentages of cattle bones have been found, beef was the most important kind of meat eaten. Pork and mutton or goat meat came in second or third place on the menu, depending partly on the nature of the local environment and partly on the extent of the Roman or military preference for pork. On a small scale the meat diet was supplemented by game. Concerning the proportion of fish, fowl and shellfish in the diet no estimates can be made. On the basis of the data on age of animals at the time of slaughter, it is evident that milk and cheese formed at most a quantitatively unimportant part of the diet of the people living in the settlements investigated. In view of the age data available for cattle it is evident that vegetable products formed the most important part of the diet. It is debatable whether meat formed part of the daily menu. 161 5 The animals not used for human consumption 5.1 HORSE The proportion of horse in the bone material found varies considerably from one settlement to another: from nothing or almost nothing in the castellum of Meinerswijk and in the two periods in Eist to about a third of the material in Druten III (table 75). It is remarkable that many more horse bones were found in the agrarian settlements than in the military settlements and camp villages. Also in the civilian settlement of the 'Valkhof, Nijmegen Ib-c, the quantity of horse bones is small. This difference cannot be explained by proposing that horses were used more in the agrarian settlements. The reason for this difference probably lies in the different way in which horse carcasses were disposed of. Horse meat was not eaten (see below), so after the death of a horse its owner was saddled with the problem of how to dispose of a few hundred kilograms of rotting, stinking meat. This problem can be solved by digging a large refuse pit within the settlement or by taking the carcass away from the settlement and leaving it or burying it elsewhere. In the agrarian settlements there will have been plenty of room available, so dumping rubbish will not have been a problem. In the more densely populated civilian and military settlements the amount of space available for dumping large quantities of animal refuse, like a carcass, will have been limited. Clearly animal carcasses from civilian and military settlements will have been taken away from the settlement. For the military vicus of Kesteren we have archeological evidence of this. Of the bones from the settlement only 7.7% are of horse. Outside the settlement, on the other hand, in the cemetery 'De Prinsenhof, 18 horse graves were found (Prummel 1979b). Also the high percentage of horse bones from the 4th-century defensive ditch of Nijmegen IV, that probably come from complete skeletons (see 3.3.2.2), can be interpreted along these lines. At the time when this big, dry ditch was no longer functional as such, but before it was filled in, for a short while it served as an enormous refuse pit at a very convenient spot. Unfortunately no archeozoological data are available for the habitation area within the defensive ditch, where we would expect to find a low percentage of horse according to the reasoning outlined above. The practice of dumping horse carcasses outside the settlement is also known for other military and civilian settlements. The inhabitants of the castellum in Zwammerdam disposed of carcasses of horses and dogs by throwing them into the Rhine outside the settlement (Van Wijngaarden-Bakker 1970). In other settlements horse carcasses were taken to the knacker's yard, where they were skinned and disposed of in refuse pits. An example of this can be found at the castellum of Stanwix, the base of a cavalry unit, the Ala Petriana Milliaria (Birley 1961: 206; Davies 1969: 432). Also in Paris a similar knacker's yard has been found, dating from the late 3rd and 4th century (Poulain-Josien 1962). The withers height of the horses varies from 124 to 163 cm, with a mean value of 143.4 cm (n= III). The withers heights are discussed further in section 6.2.3. 162 The occurrence of horse in the settlements (hand-collected material; not allocated) (n: number; g: weight in g). TABLE 75 horse 0/ /o weight-'X, Nijmegen la Nijmegen Ib-c Nijmegen castra Nijmegen canabae Nijmegen IV Meinerswijk Resteren Heteren I Heteren II Ewijk I Ewijk II Druten I Druten II* Druten III Eist pre-temple Eist tempel period 1.8 3-3 g/n cattle g/n total number identified 36.8 40.8 29.5 276 854 1623 34-3 1971 1565 131 272 276 2.9 3-0 8.7 9-7 3-5 7-2 530 94-4 74-9 67.7 22.3 28.8 141.0 109.1 0 0 0 69.0 7-7 16.3 104.1 52.7 12.7 23.1 99-7 76.0 86.6 9-3 234 14.9 38.4 14.7 30.6 32.8 122.2 10.8 20.9 83.6 53-3 34-4 58.0 37.6 54.0 47.6 32.1 37-1 115.2 109.9 0 0 0 43-7 O.I 0.1 35.0 32.9 19.9 50.0 211 horse number 5 28 47 69 349 weight cattle number 192 7071 592 24138 816 24071 265 2643 3521 4674 1643 56430 49201 971 105953 6140 10433 0 0 89 21 2186 198 35 3488 3191 8838 175 131 42 weight 9324 4507 48555 1102 102 I169 4517 134 290 1042 274 36 54 335 38590 522 45 0 0 42 13796 57370 1836 804 I 35 744 24500 242 501 13692 4400 837 777 29203 7230 excluding foundation sacrifices Data on the age of horses at the time of death is given in table 76. The age data are based on the state of fusion of the epiphyses. The actual number of individuals that determines the pattern of ages at death is smaller or much smaller than the number indicated in the table showing the age data. This was also the case with cattle (see 4.2.3), but with horse the discrepancy is even greater. For Druten II, for example, all the data on the foundation-sacrifice animals are included, and also the material from Nijmegen IV consists to some extent of a number of bones from the same individuals (see 3.3.2.2). This means that we may regard the data only as a rough indication of the pattern of ages at the time of death. The mortality rate for horse was 8% in the first year of life, 7% and 4% in the first and second halves of the second year, respectively, and 4% in the third and first half of the fourth year; 78% of the horses died at an age exceeding 3V2 years. The occurrence of horses that died at an immature age in Nijmegen IV, Kesteren, Druten II and Druten III indicates that horses were probably bred in these settlements. Conversely, the absence of animals that died at a young age in the other settlements could be an indication that horses were brought to these settlements that had been born and raised elsewhere, or, in the case of saddle horses, previously trained at stud farms. According to written sources, in Roman times, just as today, horses were trained to jump only when they had reached the age of four years, since only then is the skeletal system strong enough (Davies 1969: 445). However, it is impossible to draw any such conclusions here, in view of the small quantity of age data available for these settlements. Horse was not eaten in the Eastern River Area, as was the general rule in Roman times (Luff 1982: 248). Clear indications of the absence of horse meat on the menu are the absence or rare occurrence of butchery marks on the bones of horse, and the greater average weight values of the bone fragments found (table 75) as compared with cattle. The few cut and chop marks present on horse bones can be associated with the practice of skinning the animals and cutting them up into 163 TABLE 76 Ages at death of horse, based on the degree of fusion of the epiphyses. In the totals (data for all the sites combined) it is indicated what percentage of the horses died before and after a given age and during the interval between the given age and the preceding on (f: fused; nf: not fused; p: proximal; d: distal). Nijmegen Nijmeg en Nijmegen Nijmegen Nijmegen la Ib-c canabae castra IV age (months) 10-15 bone and epiphysis 10-12 10-12 12-15 12-15 12-15 scapula d phalanx Up metacarpus d metatarsus d phalanx I p total f nf f nf f nf f nf f nf 8 0 - - I 0 0 I 0 I 0 I 0 - - I 0 I 0 2 0 I 0 I 0 16 I 0 I 0 I 0 - - 8 I - - I 0 - - I 0 5 0 2 0 6 0 4 0 2 0 38 I 0 16 I 0 0 6 22 4 5 - 16 2 15-18 15-18 humérus d radius p total - — - - _ I 0 3 0 I 0 0 0 3 0 2 0 3 3 6 24 24 tibia d - - I 0 - 42 42 42 42 42 42 humérus p radius d ulna p femur p femur d tibia p total - - - — 3 - 0 42 - 2 0 6 0 I 0 I 0 3 0 - - 4 0 I 2 0 I 0 I 10 I 5 - - - I 0 - 5 3 0 - - 0 I I 0 7 0 6 0 9 15 46 4 2 16 15-18 pieces before burying them (see 4.6). Sometimes some meat may have been cut off a horse carcass to feed dogs, which according to Luff (1982: 248) may also explain the presence of butchery marks on at most a few horse bones in various settlements. Also in the Eastern River Area dogs were fed with horse meat, or at least the dogs were able to get at horse meat, in view of the regular occurrence of gnaw marks made by dogs: Nijmegen la (i instance = 20% of the bone material), Nijmegen Ib-c (0 = 0%), Nijmegen castra (0 = 0%), Nijmegen canabae (0 = 0%), Nijmegen IV (22 = 6%), Heteren (12=13%), Kesteren (6 = 2970)) Druten I (3 = 8%), Druten II (11 =20%; foundation sacrifices: 0 = 0%), Druten III (20=6%). The horses will have been kept for use as pack animals or saddle horses, or as draught animals. 5.2 DOG Bones of dog were found in small numbers in all of the settlements investigated, with the exception of Eist. The presence of dog is testified not only by dog bones but also by the occurrence, in Eist too, of distinctive gnaw marks on the bones of other species. As a general rule dog was not eaten in Roman times (Luff 1982). The absence of cut and chop marks on the bones found makes it clear that this also applied in the Eastern River Area. Yet the consumption of dog meat was not entirely unusual in Roman times. Pliny (XXIV. 14.58) mentions that the meat of puppies is of high quality, and that it was regularly eaten at dinners given in honour of the gods and at inaugural banquets. Indications of dog meat having been eaten have been found i.a. in Roman excavation at Pompeii, Arae Flaviae (Kokabi 1982: 94) and Vemania (Piehler 1976). Long before the Roman period already the dog existed in a variety of forms and 164 Heteren Heteren III nf f Kesteren nf --20 1 O - - f nf 22 10 Druten I Druten II Druten III f nf f 2 2 0 17 - 4 9 7 0 8 I 40 24 29 118 91 89 97 92 f nf f I 0 I 0 2 0 - o 3 5 4 7 nf I O TOTALS killed after 2 (%) nf (%) 89 2 100 0 II 0 4 3 II - - - 3 0 I 0 II I 10 2 0 6 2 0 19 3 33 4 2 2 10 10 5 o O - 2 O 0 I 0 I 9 9 i8 4 5 32 30 62 91 5 7 79 85 I o 5 I 8 4 39 - - 2 0 2 I 0 0 12 12 II II 10 10 I 21 II 15 15 7 81 9 19 4 12 92 8 69 63 75 83 I 10 0 4 0 7 6 22 2 0 I 2 5 0 I 0 18 I 0 0 o o 3 6 O I 4 6 5 20 II 19 32 108 23 9 3 8 - 3 (%) 3 8 I 2 killed between 10 - o o killed before I 9 31 38 82 3 6 4 7 78 31 22 25 17 18 was used for different purposes (Zeuner 1967). Also the Roman authors describe this variability in external appearance and function. On the basis of the Roman literature Hauger (1921) and Schlich (1957) mention three functional groups. As guard dogs for the house and homestead preference was given to robust, loud-barking, dark-coloured animals. Sheepdogs had to be fast and strong, with a more elongated build than the squat house dog, and preferably white, to make them distinguishable from wolves in the dark. For hunting, middle-sized to large dogs were used. Piehler (1976) points out that Strabo and Martial mention also small breeds of dog that were kept in the town merely for pleasure, even as genuine lap dogs. There is also a great deal of variation in the remains of dogs from Roman times found in the Netherlands. Following the estimation method of Harcourt (1974) (see 2.2.6), Clason (1978) found that dogs in Rijswijk had withers heights ranging from 30.8 cm to 64.8 cm (n= 8), on the basis of the measurements of tibiae. The data from Valkenburg (Clason 1967) indicate heights varying from 26.1 to 60.8 cm (n = 15). The few bones from the Eastern River Area that could be used for an estimation of withers height were all found in Nijmegen. For the canabae legionis a withers height of 41.0 cm could be estimated from the maximum length of a tibia; a radius provided a withers height of 52.2 cm. For 4th-century Nijmegen the following values of withers height were obtained: 51.9 cm (humérus), 55.1 cm (radius) and 42.5 and 54.7 cm (tibia). The size of the dogs found in Nijmegen lie in the range of small breeds like the beagle (40 cm) to middle-sized breeds like the Irish setter (55-65 cm). 165 6 Withers height of the farm animals 6.1 INCREASE IN THE SIZE OF CATTLE: ROMANIZATION IN STOCK-BREEDING PRACTICES The increase in the size of cattle in the Eastern River Area in Roman times can be regarded as the result of the Romanization of stock-breeding practices. The withers height has been taken as a measure of size. For the estimation of withers height the factors recommended by Von den Driesch and Boessneck (1974) have been used for the bones of the stylopodium and zygopodium, with the recommended average values being used for the metapodials: 6.15 for the metacarpus and 5.45 for the metatarsus (see also 2.2.6). An overview of the estimated withers heights for the various groups of finds, as based on particular skeletal parts, is shown in table 77. Fig. 42 gives the total picture per group of finds. The withers heights vary from 103.2 cm in Kesteren to 154.6 cm in 4th-century Nijmegen. In fig. 43 the groups of finds that yielded more than 10 data are set out. These consist of the combination of the Nijmegen Fig. 42 Withers heights of cattle per group of finds. ?: possibly pre-Roman Iron Age. 160' 65 number 150- .1 ..10 140- 130- 120- 110- I 2 1 a ? X range 20 I standard deviation mean 10 — + 100- 166 9 Ib-c Id-IIa Id-IIa Id-IIa Id-IIa Ilb-d IIIA IV TABLE 77 Withers heights (cm) of cattle estimated per skeletal elemem per site. bone metatarsus metatarsus metatarsus metacarpus radius metatarsus Nijmegen canabae metacarpus metacarpus Nijmegen IV metatarsus humérus* radius femur* tibia metacarpus Druten I metatarsus metacarpus Druten II Druten III metacarpus metatarsus humérus (GL) radius tibia Heteren I metatarsus Kesteren tibia Meinerswijk (IIIA) metacarpus Nijmegen la Nijmegen Ib-c Nijmegen castra range mean I - 3 114.3-119.2 110.5 116.8 113.0 113.7 124.1 116.7 III.O 125.9 128.4 136.6 125-I 126.7 120.7 X04.6 108.2 number I - 5 106.2-118.1 I - I I 23 3° 4 6 112.5-141.1 114.7-142.6 123.5-154-6 110.7-144.8 I - 3 9 114.6-119.1 118.4-134.7 I I I I I - 3 4 115.6-139-3 113.6-129.4 I — I I 139-7 116.9 127.3 116.7 130.2 119.3 109.2 103.2 116.4 s 2.45 4.72 7.89 7.04 13-97 12.08 2.25 5-87 12.82 7.14 - * maximum length of caput. data for the ist century and the beginning of the 2nd century (Nijmegen la, Ib-c, castra and canabae), the data for Druten III, dated Ilb-d, and those of 4th-century Nijmegen. In Nijmegen la-IIa the withers heights vary from 106.2 cm to 124. i cm with a mean value of 114.95 cm (n= 13; 5 = 4.52). The centred character of the distribution could indicate that we are concerned with a homogeneous population. The data for Druten III vary from 113.6 cm to 139.3 cm with a mean value of 124.03 cm (n = 20; 5 = 8.19). The diagram is double-peaked with a gap between the sizes 119.6 and 124.4 cm. For 4th-century Nijmegen withers heights of no.7 to 154.6 cm have been calculated, with a mean value of 127.59 cm (n = 65; s = 8.49). The distribution of the data is bell-shaped, which means that the data probably come from a single population. The data of the above-mentioned groups of finds cannot be tested for normality (Blalock i960: 240), because the expected frequency in two of the four cells in the chi-square table is less than 5 (Schefler 1979). It is possible to discern a distinct increase in withers height in the course of time. The mean value for lib-dated Druten III is 9.08 cm (8%) higher than that for la-IIa dated Nijmegen. Because the data of Druten III and Nijmegen la-IIa have unequal variances (F = 3.28, p = 0.04), a one-tailed t-test based on separate variance estimate has been carried out. The withers height in Druten III is very significantly higher than that for Nijmegen la-IIa (t = 4.09, df=30.4, a<o.ooi). The increase from Nijmegen la-IIa to Nijmegen IV is 12.64 cm (11%), which is also a very significant increase (1 = 7.72, df=3i.9, a<o.ooi). The increase from 2nd-century Druten to 4th-century Nijmegen of 3.56 cm (3%) appears to be not 167 m N Nijmegen IV —1 yk-yK<W:-> •:•:-:•:•:-:•:•:-:•:•:•:-:•:•:•:•:•:•: :•:•:•:•:•:•:•:•:•:•:•:•:•:•:•:•:•:•:• j D Drutenllb-d b Nijmegen la-lla 2KÜ 1.20 1.60m Fig. 43 Withers heights of cattle. A Nijmegen IV, B Druten III (dated Ilb-d), C combination of Nijmegen Ia, Ib-c, castra and canabae (dated la-IIa). I one measurement, 2 data of metapodia, m: mean. 168 significant (t= 1.65, df=83, a > 0.05). In this last test t was based on the pooled-variance estimate (F= 1.08; p = o.89). If we look at the total picture of the withers height data then we see an increase in height in the course of time (fig. 42). The cattle that possibly date from the pre-Roman Iron Age are smaller than no cm. In the ist century and at the beginning of the 2nd century the withers height increased to around 115 cm. In the following period the animals are very significantly bigger with a mean withers height of about 125 cm. Between the 2nd and the 4th century there appear to be no significant differences, although the mean value increases slightly. So far we have disregarded the double-peaked character of the Druten III values. In view of the gap between the sizes 119.6 and 124.4 cm, it appears that we have a group (a) with a mean withers height of 116.87 cm (n = 10, s = 2.13) and a group (b) with a mean value of 131.9 cm (n= 10, s = 4.8i) (fig. 42 and 43). Testing using a one-tailed t-test based on separate variance estimate (F = 5.ii, p = o.02) shows that the animals of group b are significantly bigger than those of group a (t = 8.61, df= 12.4, a<o.ooi). Such a double-peaked distribution can be explained in two ways. Either the two peaks represent different sexes or there are two populations involved. The first explanation seems unlikely because length measurements, on the basis of which the estimations were made, show litde or no sexual dimorphism. Moreover, if we look at the data relating to the sex of the metapodials, then according to the indices of Howard (1963) in the left group (a) there are 4 cows and i bull and in the right group (b) 6 cows and i cow/bull. Thus it has to be concluded that here we are concerned with two different populations, one of small cattle (a) and one of large cattle (b). The mean withers height of the group of small animals (a) is slightly greater (1.9 cm) than that of the cattle of Nijmegen la-IIa, though the difference is not significant (t= 1.35, df= 17.9, a = 0.096; separate variance estimate (F = 4.52, p = o.03)). The difference of 10.6 cm between the group of small animals from Druten and that from Nijmegen IV is very significant (t = 8.58; df = 57.6, a<o.ooi; separate variance estimate (F= 15.68, p<o.ooi)). The mean withers height of the group of large animals (b) from Druten with a difference of 16.2 cm is very significantly greater than that of the animals from Nijmegen la-IIa (1 = 8.31, df=2i, a<o.ooi; pooled variance estimate (F= 1.13, p = 0.824)). The animals of group b from Druten are also bigger (3.7 cm) than cattle from Nijmegen IV, yet this difference is not significant (t = 1.33, df = 73, a = 0.093; pooled variance estimate (F = 3.07, p = 0.073)). If we once again consider the overall picture presented by the withers-height data then we see an increase in height in the course of time (fig. 42 and 43). The cattle that possibly date from the pre-Roman Iron Age are smaller than no cm. In the ist century and early 2nd century the withers height has increased to around 115 cm. In the following period there are two populations present in Druten: one group of small animals that are only 2 cm higher than previously and one group of much larger animals that measure about 131 cm in height. In the 4th century the height is 127.5 cm, larger than animals of group a and slightly, but not significantly, smaller than animals of group b from 2nd-century Druten. The increase in size of cattle notably in the ist and 2nd century can be explained by assuming that a different exploitation technique came into use at that time. This could have consisted of improved nutrition and better treatment of the animals, as well as breeding schemes applied to native cattle aimed at increase in size with a view to obtaining more traction power. The fact that this change in size occurred immediately after the pre-Roman Iron Age makes it appear very likely that this is directly connected with the Roman presence in the region and with the accompanying spread of Roman know-how in agricultural practices. 1 w^ 54 58 62 66 70 cm Fig. 44 Withers heights of sheep/goat; total data. I one measurement, 2 data of metapodia, m: mean. This view is supported by the fact that in the terpen region (i.e. the marine-clay region in the north of the Netherlands, outside the Roman Empire, where people lived on terpen - raised dwelling mounds) no differences in size occur during the Roman period and that the mean withers height remains at a level of not more than no cm (Todd 1975). Also the data from the terpen Paddepoel I, II and III (200 BC-AD 250) show a similar picture: the withers heights vary from 96.7 to 115.6 cm with a mean value of around 107 cm (n = 38) (Knol 1983). In 2nd-century Druten, in addition to the improved cattle there is a separate group of much larger animals. As we are concerned here with a second population an explanation of simply a changed exploitation technique is insufficient. This second group can be explained more satisfactorily as a population, imported from elsewhere, of large animals that were kept in addition to the improved native cattle. Elsewhere too it has been shown that large cattle were imported in Roman times (Zeuner 1967; Boessneck et al. 1971). The population of large cattle of the 4th century, that were considerably bigger than the improved native cattle of the 2nd century and slightly smaller than the imported animals, could have been the result of a gradual interbreeding of the native and imported cattle. Seeing that the difference between the large cattle of Druten and those of Nijmegen IV is not significant, it is more correct to assume that in the 2nd century and later breeding regimes were practised principally with the group of larger animals. A few problems arise, however, when we compare the 2nd- and 4th-century cattle. In the first place no data are available for the intermediate period, so that the developments cannot be traced. It is questionable whether we can assume a pattern of continuity in cattle-farming practices from the 2nd to the 4th century. Moreover, the situation prevailing in Druten could be specific for the Romanized character of the settlement and does not necessarily represent a general phenomenon in the Eastern River Area. 6.2 THE WITHERS HEIGHTS OF THE OTHER FARM ANIMALS In this section the withers heights of the other farm animals are discussed. With these animals, sheep/goat, pig and horse, no development in withers height can be TABLE 78 Withers heights (cm) of sheep/goat estimated per skeletal element per site. bone Nijmegen la Nijmegen Ib-c Nijmegen castra Nijmegen canabae Druten II Druten III Heteren II Meiners wijk humérus metatarsus metacarpus metatarsus humérus metacarpus metatarsus metacarpus metatarsus metacarpus metatarsus humérus radius metacarpus tibia metatarsus tibia metatarsus number range mean s I I - - 3 55.5-61.8 I I - 4 60.0-66.9 I 2 - 57.9 64.3 59-1 57-9 64.2 62.5 61.6 64.1 50.0 56.8 59-5 62.2 62.2 54-1 65.8 57-7 58.5 59.2 I 2 2 59.3-68.9 56.0-57.5 57.4-61.5 I - 4 56.2-68.2 I - I I I I 3.26 3.16 6.79 1.06 2.90 4.90 169 TABLE 79 Withers heights (cm) of pig estimated per skeletal element per site. bone Nijmegen la Nijmegen castra Nijmegen canabae Nijmegen IV Meinerswijk astragalus calcaneus femur astragalus calcaneus astragalus number withers height 61.0 65.4 63.7 65.2 69.8 69.1 observed. In view of the absence of a sufficient quantity of data from the Eastern River Area it is impossible to study any such development. 6.2.1 SHEEP/GOAT The withers height of sheep/goat have been estimated on the basis of Teichert's (1975) factors for sheep, as recommended by Von der Driesch and Boessneck (1974). The entire quantity of material studied contained only 29 bones for which the withers heights could be estimated: 20 metapodials and 9 bones of the zygopodium and stylopodium. An overview of the estimated withers heights per site is given in table 78; fig. 44 shows the overall picture for the Eastern River Area. The withers heights vary from 54.1 cm to 68.9 cm with a mean value of 60.65 cm (n = 29, s = 3.77 cm). LuflF(i982; table 5:27) gives a survey of withers heights, estimated from measurements of the metapodials according to Teichert (1975), for 30 sites dating from the Roman period in the Roman northwestern provinces. In all of these sites the mean withers heights are higher than in the Eastern River Area. Only for Lorenzberg and Marseilles are minimum values mentioned that are lower than those in the Eastern River Area (52.7 and 53.4, respectively). Also in the castellum of Valkenburg (Clason i960), which site is not included in Luff's survey (1982), the mean value is higher (62.19 cm, n= 15, 8 = 4.00 cm), although the minimum value is lower (53.35 cm). To summarize, we can say that the sheep/goat in the Eastern River Area was small in size compared with other sites in the Roman period. The scarce data per site give no indication that there was any development in size during the Roman period. Any possible differences in withers height between the various sites are impossible to discern in view of the small quantity of data. 6.2.2 PIG The withers heights of pig have been estimated according to the factors mentioned by Teichert (1969) (see also Von den Driesch and Boessneck 1974). As most of the complete bones found come from immature animals, the withers heights could be estimated for only six bones in all (table 79). Five of the six bones are small bones, astragali and calcanei, which means that a certain measure of inaccuracy has to be taken into account. The mean withers height is 65.69 cm, the minimum and maximum values are 61.04 cm and 69.77 cm respectively. The animals in the Eastern River Area are small compared with the mean withers height of 74.6 cm estimated by Teichert (1970) for sites dating from the Roman period. 170 TABLE 80 Withers heights of horse per individual bone. For each bone the length measurement (cm) is given, together with the size class according to Vitt and the estimated withers height (cm). bone Nijmegen Ib-c radius metacarpus Nijmegen castra radius metacarpus radius tibia humérus radius Nijmegen canabae Nijmegen IV metacarpus femur tibia 3305 21.35 21.58 36.96 22.08 36.50 41.16 29.27 32.03 32.59 32.67 33.27 33-28 3346 33.61 33.80 21.73 22.15 22.21 22.51 22.79 23.72 24.04 40.08 40.22 40.32 3358 34-94 36.23 36.46 37.00 37.20 class withers height 5 4 4 6 5 6 8 5 4 4 4 5 5 5 5 5 4 5 5 5 5 6 6 136.20 132.53 133.76 151.84 136.43 150.00 162.64 137-08 132.12 134.36 134.68 137.08 137.12 137.84 138-44 139.20 134.56 136.80 137.12 138.72 140.21 5 5 5 4 5 5 5 6 6 6 6 6 4 radius 37-65 37-91 38.10 25.80 26.02 27.21 27.46 27.48 27.86 27.98 22.42 31.50 31.60 36.70 metacarpus 36.75 23.26 6 6 6 6 5 23-37 23.50 24.81 24.89 41.85 42.65 42.84 6 6 6 6 6 6 metatarsus Druten I Druten II measurement metacarpus humérus femur 5 5 5 5 6 6 5 5 145-17 146.88 140.32 140.88 141.28 132.32 137.76 142.92 143.84 146.00 146.80 148.60 149.64 150.40 134-93 136.II 142.45 143-79 143.89 145.92 146.56 138.24 146.00 146.40 150.80 151.00 142.72 143.31 144.00 150.99 151.41 147.40 150.60 151.36 171 bone tibia metatarsus Druten III humérus radius metacarpus tibia metatarsus Heteren I Heteren II metatarsus metacarpus Kesteren* metatarsus humérus radius metacarpus femur tibia * ** All data for Kesteren, with the exception of the metatarsus indicated by two asterisks, come from the horse graves (Prummel 1979)- 172 metatarsus measurement 34-25 37.00 37.42 3796 26.92 28.78 28.88 28.84 33-04 35-48 36.00 22.03 22.38 22.60 22.62 23.18 23.60 24.14 24.17 34.40 35.20 37.26 39.06 24.81 25.83 25-94 27.18 27.20 27.50 27.98 25.10 19.83 23.31 28.56 29.77 30.20 34.90 34-97 21.71 22.51 23.02 23.05 23.60 23.80 39.29 40.35 41.00 41.08 36.24 36.71 24.24** 26.80 26.96 27.82 28.37 class withers height 4 6 6 6 135.00 146.00 147.68 149.84 140.91 150.83 151.36 135-36 136.16 145.92 148.00 136.16 138.03 139.20 139.31 142.29 5 6 6 4 5 6 6 5 5 5 5 5 6 6 6 4 5 6 7 4 4 4 5 5 6 6 144-53 147.41 147.57 135.60 138.80 147.04 154.24 129.65 135.09 135.68 142.29 142.40 144.00 146.56 4 3 5 6 131.20 5 5 5 5 4 5 5 5 6 6 139.08 5 5 6 6 5 6 3 5 5 6 6 124.43 142.99 149.65 140.80 143.60 143.88 134-45 138.72 141.44 141.60 144-53 145.60 137.16 141.40 144.00 144-43 142.96 144.84 126.61 140.27 141.12 145-71 148.64 50% 6.2.3 HORSE 40 The withers heights of horse (fig. 45) are presented according to the division into nine classes made by Vitt (Von den Driesch and Boessneck 1974). Table 80 gives the measurement data per site that have been used for the indication of the withers height. Also included are the bones from the horse graves of Kesteren (see 3.5.1), measured by Prummel (see 1979b). I shall restrict myself to this broad description of the withers heights. Statistical analyses of the data, as carried out by Prummel (1979b), for example, are somewhat dubious for the skeletal material of horse. Unlike the cattle bones, for example, the horse bones cannot be assumed to come from different individuals as a general rule. The horse bones were often found as whole skeletons or parts of skeletons. If more than one bone of such a complete or incomplete skeleton is involved in the estimations, then these individuals greatly outweigh the individuals of which only one bone was found. The majority of the estimated withers heights fall in the classes 5 (136-144 cm) and 6 (144-152 cm), namely 43 and 39% respectively. Class 4 (128-136 cm) accounts for 14%. Only two bones (2%) indicate a withers height of 120-128 cm (class 3), one bone a withers height of 152-160 cm (class 7) and one a withers height of 160-186 cm (class 8). On the basis of Vitt's classification the interpolated values of the withers heights give mean values of 143.4 cm with a minimum of 124 cm and a maximum of 163 cm (n= II i). On average the greatest values for withers height are found in the Nijmegen castra and canabae and the villa in Druten (Druten II), which could indicate that the Roman army used relatively large horses. Yet there is certainly no general increase in withers height in the region during the Roman period. It is noteworthy that the values of withers heights of horse from the Eastern River Area are high compared with those from other parts of the Netherlands in the same period: in the native settlement dating from the Roman Iron Age in Rijswijk the mean withers height is 131.4 cm (n = 22) and in the castellum in Valkenburg 140.6 cm (n= 19) (Prummel 1979b). Also the horses of Xanten (Colonia Ulpia Traiana), with a mean withers height of 137.5 cm (n = 34) are smaller in size (estimated, as above, using the data of Waldmann (1967)). •30 20 3 4 5 6 7 8 class Fig. 45 Withers heights of horse; total data. class 3 4 5 6 7 8 height (cm) 120-128 128-136 136-144 144-152 152-160 160-168 173 Summary ANIMALS IN ROMAN TIMES IN THE DUTCH EASTERN RIVER AREA This archeozoological study concerns the animal remains that were found during excavations of settlements dating from Roman times in Nijmegen and its surroundings. The aim of the study was to acquire detailed information about the animals, in particular as regards their function. Faunal remains from the following settlements were included in the study: native farmsteads in Ewijk and Heteren, the villa in Druten, civilian settlements in Nijmegen, military forts (castella) in Nijmegen and Meinerswijk and a camp village, belonging to such a fort, in Kesteren, the mid-Roman legionary camp (castra) and the camp village surrounding it (canabae legionis) in Nijmegen. In addition an investigation was made of animal bones found in the Gallo-Roman temples in Eist and a fourth-century cemetery in Nijmegen. The mammal species that were represented can be divided into three groups. The first group consists of farm animals that were eaten: cattle, sheep, goat and pig. The second group consists of wild animals: aurochs, elk, red deer, roe deer, wild boar and hare. Also these animals were eaten, in view of the butchery marks that are present on the bones. In addition shed antlers were collected incidentally for the purpose of making particular objects out of them. The third group is formed by horse and dog, domesticated animals that were not eaten. If we compare the first two groups it is clear that the hunting of larger game animals was of very little significance for the pattern of meat consumption. Within the group of farm animals the most important meat-providing species is cattle. Pig and sheep/goat come in second and third place in turn. Relatively high percentages of pig can be associated on the one. hand with the better facilities for grazing pigs on the Pleistocene soils, and on the other hand with the military or Roman character of settlements. Higher percentages of sheep/goat can be associated especially with an environment suitable for sheep in the Holocene area. The remains of birds and fish come for the most part from only a few sieve-samples. The data obtained for these groups will therefore be very incomplete. The most common species among the poultry is the domestic fowl. At some sites also goose, duck and pigeon were found. Concerning these last three species it is not certain whether the remains found represent domesticated or wild forms. The remains that were found of crane, cormorant and long-eared owl certainly came from wild birds. Fishing was also practised, and the species caught include pike, rudd, orfe or chub, perch, allis shad or twaite shad, eel, salmon and catfish. An exotic fish is represented by the find, dating from the fourth century, of a vertebra of the genus Sphyraena, a kind of barracuda, that only occurs in warm seas. This specimen 174 probably arrived here as a fish product, in the liquamen or allée, or as salted fish. At a few sites remains of oysters and whelks were found, which must have been brought inland from the coast. The data on cattle, sheep/goat and pig present a picture of an agrarian economy based on mixed farming with the emphasis on agriculture. Cattle were not kept primarily for milk production, nor primarily as a source of meat, despite the fact that beef was the kind of meat most frequently eaten. Rather they were kept in the first place as a source of traction power and of manure for agricultural purposes. In addition smaller numbers of pigs and sheep were kept for their meat, and in the case of sheep also for the supply of wool and milk (cheese). Also the production of manure for fertilizing the soil will have played a role in keeping these smaller farm animals. Agricultural products were much more important than meat for the diet of the local inhabitants. It is even questionable whether meat formed part of the daily menu. Horses were kept as pack animals, for riding or as a supply of traction power. Cattle, sheep and horses were skinned for their hides. The bones were used only incidentally for the purpose of making particular objects. The fact that in the first and second century the size of cattle increases considerably can be explained by the development of improved exploitation techniques as a result of the availability of Roman know-how with regard to agriculture. This could have involved better nutrition and better treatment of the animals and the use of a more refined breeding regime with native cattle with the aim of obtaining more traction power. Large cattle were probably imported as well. A description of the individual settlements and the bone material found there is given in chapter 3. The bone refuse of the early Roman castellum (Nijmegen la) appeared to have been removed intentionally away from the fortification. Most of this material was found outside the castellum in the neighbourhood of the western entrance. Also in the settlement on the Valkhof (Nijmegen Ib-c) the bone refuse is concentrated in a few places: more than half the material was found in three refuse pits. In terms of the composition the material from these refuse pits did not differ from the bone material found elsewhere. As for the two associated settlements, the mid-Roman canabae legionis and the castra, there was no evidence of any differences between them relating to the incidence of slaughtering animals and the consumption of meat. The finds of indisputably primary butchery refuse indicate that cattle were brought into the camp on the hoof and were ultimately slaughtered there. In certain places in the canabae bone concentrations were found that are indicative of specialization in meat processing. In the western canabae the smoking of shoulders of beef probably took place. In the eastern canabae skulls of cattle were processed for the production of a kind of brawn. These meat products may have been intended for the inhabitants of the army camp. In the large fourth-century defensive ditch in Nijmegen the Ijone material was found mostly concentrated at one spot. It is likely that after the ditch had lost its defensive function it was used as a rubbish dump. The concentration of the material could indicate that at that spot a road ran up to the ditch; alternatively there may have been a bridge over the ditch here. The bone finds from the fourth-century cemetery come from meals for the dead, provided on dishes or plates at the time of burial, to sustain the deceased during 175 the journey to the next world. The meat for the dead consisted mainly of pork and chicken, while the meat diet of the living consisted mainly of beef. Comparison with other settlements and cemeteries from Roman times shows that this was a general phenomenon. If we take into consideration the information given by the epicurean cookery book of Apicius, then we must conclude that the dead were provided with a more sumptuous meal than that to which the living were accustomed to eating. A remarkable feature of the castellum of Meinerswijk is that no bones of horses were found there. Carcasses of animals that were not eaten were usually buried outside the settlement, however, as was also the case in Kesteren; for example. At the entrance of each of the two most important buildings (i and 11) of the villa in Druten (II) two horse skeletons were found. These skeletons probably represent foundation sacrifices. In contrast to Druten III, in Druten II there is a distinct difference in the bone material found in the pars urbana and the pars rustica of the settlement. The presence of foundation sacrifices, the predominance of pig over sheep/goat, and the greater species diversity in the pars urbana can be associated with the higher status of the occupants in this part of the villa. The deviating age of the cattle found at the temples in Eist can be associated with the function of cattle as sacrificial animals as proposed by Bogaers (1955). The fact that the cattle from the period before the temples were built show the same pattern of age at the time of slaughter provides support for the view that this site was already a place of cultic significance before the building of the temples. The find of a unique combination of the skulls of a pig, a sheep and an ox, bearing in mind the sex and size of the animals they came from, provides an extra argument for regarding these bones as the remains of a suovetaurilia sacrifice. The function of animals kept in the Eastern River Area in Roman times was to a great extent conscerned with agriculture. They provided traction power and manure. In addition they were a source of meat, for both the living and the dead, and were used as sacrificial offerings. Hides and wool were made use of, and occasionally bone was used as a raw material for making particular objects. 176 Samenvatting DIEREN IN DE ROMEINSE TIJD IN HET OOSTELIJK RIVIERENGEBIED In dit archeozoölogisch onderzoek zijn de dierlijke resten onderzocht die gevonden zijn tijdens opgravingen van nederzettingen uit de Romeinse tijd in Nijmegen en omgeving. Doel van het onderzoek was een idee te vormen over de dieren en de functie die zij hadden. Faunaresten uit de volgende nederzettingen zijn in het onderzoek opgenomen: \ inheemse boerderijen uit Ewijk en Heteren, de hereboerderij (villa) uit Druten, burgerlijke nederzettingen uit Nijmegen, militaire forten {castella) uit Nijmegen en Meinerswijk en een bij zo'n fort behorend kampdorp uit Kesteren, de midden-Romeinse legioensvesting {castra) en het daaromheen liggende kampdorp {canabae legionis) uit Nijmegen. Verder is het dierlijk bot onderzocht dat gevonden is bij de Gallo-Romeinse tempels in Eist en een vierde eeuws grafveld uit Nijmegen. De gevonden zoogdiersoorten zijn in drie groepen te verdelen. De eerste groep bevat de landbouwhuisdieren die gegeten werden: rund, schaap, geit en varken. De tweede groep is wild: oeros, eland, edelhert, ree, wild zwijn en haas. Ook deze dieren werden gegeten, getuige de slachtsporen die op de botten voorkomen. Incidenteel werden ook afgeworpen geweien verzameld om er voorwerpen van te maken. De derde groep wordt gevormd door paard en hond, huisdieren die niet voor de consumptie gebruikt werden. Als we de eerste twee groepen vergelijken is duidelijk dat de jacht op het grotere wild voor de vleesconsumptie geen rol van betekenis heeft gespeeld. In de groep van de landbouwhuisdieren is rund overal de belangrijkste vleesleverancier. Varken en schaap/geit komen beurtelings op de tweede en derde plaats. Relatief hoge percentages varken kunnen worden geassocieerd met enerzijds de betere mogelijkheden op de Pleistocene gronden om varkens te mesten, anderzijds met het militaire of Romeinse karakter van nederzettingen. Hogere percentages schaap/geit kunnen vooral worden geassocieerd met een voor schapen geschikt milieu in het Holocene gebied. Het materiaal van vogels en vissen komt voornamelijk uit slechts enkele zeefmonsters. De gegevens over deze groepen zullen daarom zeer onvolledig zijn. Als pluimvee komt het huishoen, de kip, algemeen voor. Op çommige vindplaatsen is ook gans, eend en duif aangetroffen. Van deze laatste soorten is het niet zeker of het dieren van de boerderij zijn of dat het hun wilde soortgenoten betreft. Zeker uit het wild komen de kraanvogel, de aalscholver en de ransuil. Er werd ook gevist, en wel op snoek, rietvoorn, winde of kopvoorn, baars, fint of elft, paling, zalm en meerval. Exotisch is de vondst van een vierde eeuwse wervel 177 van een Sphyraena, een baracuda-achtige, die alleen voorkomt in warmere wateren. Waarschijnlijk is dit exemplaar als visproduct, in de liquanten of allee, of als gezouten vis hier terecht gekomen. Importen vanuit de kust zijn de op enkele vindplaatsen aangetroffen overblijfselen van oesters en wulken. De gegevens over rund, schaap/geit en varken leveren het beeld op van een agrarische economie die gebaseerd is op een gemengd bedrijf met de nadruk op akkerbouw. Runderen werden niet primair gebruikt voor de melkproductie noch, hoewel rundvlees de meest gegeten vleessoort was, primair voor die van vlees. Ze zullen voornamelijk gehouden zijn voor trekkracht en mest ten dienste van de akkerbouw. Daarnaast werden kleinere aantallen varkens en schapen gehouden om hun vlees, en in het geval van het schaap ook om de wol en de melk (kaas). Ook de productie van mest voor de akkerbouw zal een rol hebben gespeeld bij het houden van kleinvee. Akkerbouwproducten waren voor de voeding van veel groter belang dan vlees. Het is zelfs de vraag of vlees deel uitmaakte van de dagelijkse kost. Paarden werden gehouden als leveranciers van kracht in de vorm van last-, rij-, of trekdier. Runderen, schapen en paarden werden gevild voor hun huiden. De beenderen werden slechts incidenteel gebruikt voor het vervaardigen van voorwerpen. Dat de grootte van het rund in de eerste en tweede eeuw aanzienlijk toeneemt is te verklaren door de opkomst van betere exploitatietechnieken als gevolg van het beschikbaar komen van de Romeinse know-how op het terrein van de landbouw. Te denken valt aan betere voeding en behandeling van het vee en aan een op grootte gerichte fok van het inheemse vee met het oog op een grotere trekkracht. Waarschijnlijk werden ook grote runderen ingevoerd. Een beschrijving van de afzonderlijke nederzettingen en het daar gevonden botmateriaal wordt gegeven in hoofdstuk 3. Het botafval van het vroeg-Romeinse castellum (Nijmegen Ia) bleek grotendeels uit de versterking verwijderd te zijn. Het merendeel werd gevonden buiten het castellum in de buurt van de westelijke uitgang. Ook in de nederzetting op het Valkhof (Nijmegen Ib-c) komt het botafval geconcentreerd op enkele plaatsen voor: meer dan de helft van het materiaal werd gevonden in drie afvalkuilen, die overigens wat de samenstelling van de inhoud betreft niet afwijken van het elders gevonden materiaal. Tussen de twee bij elkaar behorende nederzettingen, de midden-Romeinse cabanae legionis en de castra, werd geen duidelijke slacht-consumptie relatie gevonden. De vondsten van duidelijk primair slachtafval duiden erop dat runderen op de hoef het kamp in werden gevoerd om daar geslacht te worden. In de canabae zijn plaatsen met botconcentraties die duiden op specialisaties in de verwerking van vlees. In de westelijke canabae werden waarschijnlijk schouders van runderen gerookt. In de oostelijke canabae werden schedels van runderen verwerkt tot een soort hoofdkaas. Mogelijk waren deze vleesproducten bestemd voor de bewoners van de legerplaats. In de grote vierde eeuwse gracht in Nijmegen is het bot voornamelijk geconcentreerd op één plaats gevonden. Waarschijnlijk werd (Ie gracht nadat het zijn verdedigende functie had verloren gebruikt als stortplaats voor afval. De concentratie van het materiaal kan er op duiden dat op die plaats een weg leidde naar de gracht of dat er mogelijk een brug bestond. De botvondsten uit het 4e eeuwse grafveld zijn afkomstig van maaltijden die de doden, op schalen of borden, werden meegegeven voor hun reis naar de andere wereld. Aan de doden werd voornamelijk varkensvlees en kip meegegeven, terwijl 178 de levenden vooral rundvlees aten. Vergelijking met andere nederzettingen en grafvelden uit de Romeinse tijd laat zien dat dit een algemeen beeld is. Betrekken we hierbij de gegevens van het luxueuse kookboek van Apicius, dan moeten we concluderen dat de doden een feestelijkere maaltijd werd meegegeven dan de levenden gewoon waren te eten. Opvallend in het castellum van Meinerswijk is, dat er geen beenderen van paarden gevonden zijn. Kadavers van dieren die niet werden gegeten werden echter meestal buiten de nederzetting begraven, zoals bijvoorbeeld ook in Resteren het geval was. Bij de ingang van elk van de twee belangrijkste gebouwen (i en ii) van de villa in Druten (II) zijn twee skeletten van paarden gevonden. Deze skeletten zijn waarschijnlijk bouwoffers. In tegenstelling tot Druten III bestaat er in Druten II een duidelijk verschil in het botmateriaal gevonden in de pars urbana en de pars rustica van de nederzetting. De aanwezigheid van bouwoffers, de dominantie van varken boven schaap/geit, en de grotere soortenrijkdom in de pars urbana kan in verband worden gebracht met de hogere status van de bewoners in dit deel van de villa. De afwijkende leeftijd van de runderen gevonden bij de tempels in Eist kan in verband worden gebracht met de door Bogaers (1955) geopperde functie van het rund als offerdier. Dat de runderen uit de periode voor de bouw van de tempels dezelfde slachtleeftijd hebben ondersteunt de vooronderstelling dat voor de bouw van de tempels het terrein reeds cultusplaats was. Het geslacht, de grootte en de unieke combinatie van de schedels van een varken, een schaap en een rund zijn extra argumenten om deze te beschouwen als restanten van een suovetaurilia offer. De functie die de dieren in de Romeinse tijd in het Oostelijk Rivierengebied voor de mens vervulden was voor een belangrijk deel gericht op de akkerbouw. Ze leverden trekkracht en mest. Daarnaast werden dieren gebruikt voor de consumptie, als offerdier en als maaltijd voor de doden. Huiden en wol werden benut, en af en toe werd been gebruikt voor de vervaardiging van voorwerpen. 179 Appendix: butchery mark code i8i Cranium, Bos taurus lateral (i) and aboral (2) view CRANIUM aboral oral basal dorsal 2 3 chop cut transversal on frontale lateral 4 5 chop cut on temporale chop as 4, cut through lateral 7 8 chop cut on maxillare aboral 9 10 chop cut on lateral side of occipitale aboral chop as 9, cut through aboral chop condylus occipitalis cut off 10 II 182 hole in aboral part of frontale dorsal lateral 9 chop HORN-CORE Horn-core, Bos taurus lateral view _^ frontale pariétale aboral I proximal 1 ^y [ 2 "^sj 3 ƒ I 5 CV__y' ^V ••It lateral lateral dorsal i chop 2 sawn 3 chop horn-core cut off through frontale and pariétale on dorsal side of frontale just below horn-core 5 chop 6 cut 7 8 chop sawn base of horn-core cut through horn-core fragment transversally cut through 183 ANTLER AND PEDICLE Antler and pedicle, Cervus elaphus lateral view brow tine beam I 1 2 chop sawn pedicle cut through 2 3 4 3 4 chop sawn beam cut through above burr 5 6 chop sawn brow tine cut off 7 8 chop sawn beam cut through 9 10 chop sawn tine cut off 5 6 9 lO 184 Ï Mandibula, Bos taurus lateral view MANDIBULA oral aboral basal lateral chop corpus cut through behind last molar lateral chop corpus cut through ± between M2 and M3 lateral chop corpus cut through ± between P4 and Mi lateral chop diastema cut through in front of premolars lateral 5 6 chop cut on lateral side of the corpus lateral 7 8 chop cut on lateral side of the diastema chop on basal side of the diastema lateral lateral shaving marks on basal side of diastema/ramus lateral 12 lateral 13 12 13 chop angle of ramus cut off chop cut on lateral side of the basal half of the ramus 14 lateral chop ramus horizontally cut through 15 lateral chop processus coronoides cut off 16 17 lateral chop cut horizontally underneath the processi 18 lateral chop vertically; tip of processus coronoides cut off'and/or chop between processi 19 medial chop horizontally underneath the processi 16 17 19 20 185 MANDIBULA ib 22 dorsal chop lateral side of processus articularis cut off dorsal chop medial side of processus articularis cut off dorsal chop longitudinal on dorsal side of the processus articularis dorsal chop latero-medial on dorsal side of the processus articularis ** j^i 23 24 ^ 25 26 iV^ l-Ti^^^ 27 28 ^ 33 186 chop cut on medial side of the corpus medial 27 28 chop cut on medial side of the diastema chop symphysis longitudinally cut through \i (Sus) basal Ü p Ifw (Sus) basal 30 31 chop cut longitudinally on aboral side of the symphysis (Sus) basal 32 33 chop cut latero-medial on basal side of the symphysis // 32 25 26 <y 29 30 31 medial Hyoid, Bos taurus medial view HYOID dorsal aboral oral medial cut on proximal end medial cut on middle part 187 Atlas, Bos taurus dorsal (i) and ventral (2) view ATLAS m I f) 7 ns cranial caudal dorsal chop median on arcus dorsalis dorsal chop as I, cut through dorsal chop lateral on arcus dorsalis dorsal chop as 3, cut through dorsal chop on cranial part of processus transversus dorsal chop lateral on processus transversus dorsal chop as 6, cut through dorsal chop diagonal on processus transversus dorsal chop as 8, cut through chop diagonal on caudal part dorsal chop as 10, cut through (^ m 10 / f> ">\ 12 n'ïï^^ci ventral chop median on arcus ventralis 13 -i^'i^C ' ventral chop as 12, cut through 14 o-ï^^c' ventral chop lateral on arcus ventralis ventral chop as 14, cut through 15 188 ATLAS i6 ventral chop on cranial part 17 j^^J^cX ventral chop as 16, cut through i8 f%^^^fç\ ventral chop on caudal part of processus transversus 00 189 Epistropheus, Bos taurus cranial (i) and lateral (2) view I dorsal 2 EPISTROPHEUS ventral ft. cranial chop on lateral side of the facies articularis cranialis cranial chop lateral part of facies articularis cranialis cut off cranial chop on ventral part of the facies articularis cranialis cranial chop as 3, cut through cranial chop completely sagittally cut through chop cut on ventral side of dens ventral chop as 6, cut through lateral chop on caudal side of the facies articularis cranialis lateral chop proximal part cut off lateral chop processus transversus cut off lateral chop combination of 9 and 11 lateral chop dorso-ventral on lateral side lateral chop as 13, cut through # * m ventral -\r-J a 13 14 m .-/'X a 190 6 7 Vertebrae, Bos taurus cervical (i), thoracic (2), lumbar (3), dorsal, lateral, cranial view VERTEBRAE I cranial caudal ventral ventral cranial chop processus transversus cut off dorsal chop transversally cut through dorsal chop diagonally cut through dorsal chop processus articularis cranialis cut off dorsal chop cranial part of processus articularis cranialis cut off dorsal chop lateral part of processus articularis cranialis cut off dorsal chop processus articularis caudalis cut off dorsal chop caudal part of processus articularis caudalis cut off dorsal chop lateral part of processus articularis caudalis cut off dorsal chop dorsally on processus spinalis lateral chop on dorsal part of processus spinalis 12 lateral chop on ventral part of processus spinalis 13 lateral chop processus spinalis cut off 14 lateral chop processus spinalis dorso-ventrally cut through 15 lateral chop facies terminalis cranialis cut off 10 191 VERTEBRAE l6 lateral chop facies terminalis caudalis cut off 17 cranial chop corpus vertebrae horizontally cut off 18 cranial chop lateral part of corpus vertebrae cut off 19 lateral chop dorso-ventrally on lateral side 192 COSTAE Costa, Bos taurus caudal view dorsal lateral / / medial ventral 1 chop 2 cut 5 6 on ventral side of dorsal part chop coUum costae cut off chop epiphysial part cut off chop cut on corpus costae chop as 5, cut through 193 SCAPULA Scapula, Bos taurus lateral view cranial distal proximal caudal distal caudal chop on lateral side of spina distal caudal chop on distal side of spina distal caudal chop spina cut off from distal side chop cut on cranio-lateral side of the distal articulation chop as 4, cut through distal 4 5 distal distal lateral 7 8 chop chop combination 2 and 6 combination 3 and 6 9 jQ chop j-m on caudo-lateral side of the distal articulation 9 10 C7~S ^-^V distal II (iT^-^V^ distal chop as 9, cut through 12 /^Jj distal lateral chop cut from distal side on margo thoracalis 13 )^^ "^ distal lateral chop combination 11 and 12 14 15 f/~r) —^ distal chop cut on caudal side of the distal articulation 16 yj j\ distal chop as 14, cut through 17 18 ^—\ ^-^ distal chop cut on processus coracoideus or cranial side of the articulation 19 A ) distal chop as 17, cut through 20 (fH) distal 20 chop on medial side of the distal articulation 21 cut 21 194 14 15 17 18 SCAPULA c^ distal 23 24 ^ distal 25 ^ chop as 20, cut through chop cut in the middle of the lateral side; radial distal blow distal articulation latero-medial cut through 26 distal chop lateral edge(s) of articulation smoothed 27 28 distal lateral 27 28 chop cut on the lateral side of the distal end 29 30 distal medial 29 30 chop cut on the medial side of the distal end 31 medial chop blade cranio-caudal cut through in the middle 32 medial chop cranio-caudal cut mark(s) on the medial side of the blade 33 medial cut cranio-caudal cut mark(s) on the medial side of the blade 34 medial cut longitudinal cut mark(s) on the medial side of the blade medial cut cut mark(s) on margo cervicalis 22 35 r' 36 medial 37 medial 38 medial 39 lateral 40 lateral 23 24 shaving mark(s) on margo cervicalis cut cut mark(s) on margo thoracalis shaving mark(s) on margo thoracalis cut longitudinal cut mark(s) on caudal side of spina hole in the blade on caudal side of spina 195 Humérus, Bos taurus caudal view HUMERUS medial distal proximal lateral 2 chop chop caudo-medial side of the proximal articulation cut off as I, not cut through proximal caudal 3 4 chop chop lateral side of the proximal articulation cut off as 3, not cut through proximal 5 6 chop chop caudal side of the proximal articulation cut off as 5, not cut through middle caudal 7 8 chop cut on caudal or lateral side of the middle part middle cranial 9 chop cut on cranial or medial side of the middle part 10 chop middle part of the diaphysis cut through proximal 9 to 'i I middle caudal middle caudal shaving mark(s) on medial side of the diaphysis 13 middle caudal shaving mark(s) on lateral side of the diaphysis 14 middle caudal shaving mark(s) on caudal side of the diaphysis 15 middle cranial shaving mark(s) on cranial side of the diaphysis 16 17 18 19 20 21 0 distal caudal 16 17 chop cut on lateral side of the distal end of the diaphysis distal cranial 18 chop cut on cranial side of the distal end of the diaphysis distal medial 20 chop cut on medial side of the distal end of the diaphysis chop longitudinal on proximal side of fossa olecrani chop cut diagonal on medial side of the distal end 19 21 distal caudal 23 24 196 distal medial 23 24 HUMERUS li ëà distal medial 25 26 chop cut transversal on medial side of the distal articulation distal cranial 27 28 chop cut various cut marks on cranial side of the trochlea 0 distal medial chop distal end of the medial part of the trochlea cut off a distal medial chop cranial part of the trochlea (partly) cut off chop trochlea (partly) cut off (^ distal medial distal cranial chop medial part of the trochlea (and the epicondylus medialis) cut off ^ distal cranial chop lateral part of the trochlea (and the epicondylus lateralis) cut off distal cranial chop trochlea cut through in the middle ^ 35 i^) distal medial chop angle of epicondylus medialis cut off 36 6^ distal lateral chop angle of epicondylus lateralis cut off 25 26 27 28 29 30 31 32 33 1:1 34 197 Ulna, Bos taurus lateral view ULNA volar proximal distal dorsal proximal lateral chop cut on lateral side of the proximal end chop as I, cut through chop cut on the proximal articulation (and the lateral side) chop as 4, cut through chop cut on the lateral side of the processus anconeus chop as 7, cut through 10 chop longitudinal on the volar-lateral side of the proximal end 11 cut 1 2 proximal lateral proximal lateral # 4 5 proximal lateral proximal lateral 7 8 proximal lateral 10 II proximal lateral proximal lateral 13 14 proximal medial 15 distal lateral \ 198 chop as 10, cut through 13 chop longitudinal on the medial side of the proximal end 14 cut chop distal end cut through Radius, Bos taurus dorsal view RADIUS lateral proximal distal medial (TS^ proximal chop dorso-medial of proximal articulation cut off (C^^ proximal chop volar-medial part of the proximal articulation cut off V proximal volar chop medial part of the proximal articulation cut off S^ proximal chop lateral part of the proximal articulation cut off proximal chop proximal articulation (and diaphysis) cut through in the middle proximal 6 7 chop cut longitudinal cut mark(s) on the proximal articulation proximal dorsal 8 9 chop cut on dorsal side of the proximal end 10 II proximal volar 10 11 chop cut on volar side of the proximal end 12 12 13 chop cut on medial side of the proximal end 13 proximal medial 14 15 proximal lateral 14 15 chop cut on lateral side of the proximal end 16 proximal volar chop volar side of the proximal end partly smoothed 17 ig middle dorsal 17 18 chop cut on lateral side of the middle part of the diaphysis 19 middle dorsal 19 20 chop cut on medial side of the middle part of the diaphysis 20 21 22 middle dorsal 21 chop cut on dorsal side of the middle part of the diaphysis 22 23 24 middle volar 23 24 chop cut on volar side of the middle part of the diaphysis 199 RADIUS 25 dorsal chop proximal end of the diaphysis cut through 26 dorsal chop diaphysis cut through in the middle 27 dorsal chop distal end of the diaphysis cut through 28 middle volar 29 30 distal dorsal 29 30 chop cut on dorsal side of the distal end 31 32 distal volar 31 32 chop cut on volar side of the distal end 33 distal chop medial part of the distal articulation cut off 34 distal chop distal articulation (and diaphysis) cut through in the middle 200 shaving marks on the medial-volar side of the diaphysis Metatarsus iii-iv. Bos taurus dorsal view METAPODIA lateral distal proximal ü^ | = medial f ^ (7' proximal dorsal I 2 cut chop on dorsal side of proximal end proximal plantar 3 4 cut chop on plantar side of the proximal end proximal lateral 5 6 cut chop on lateral side of the proximal end proximal medial 7 8 cut chop on medial side of the proximal end proximal dorsal chop (part of) dorsal side of the proximal articulation cut off proximal plantar chop (part of) plantar side of the proximal articulation cut off proximal plantar chop process(es) on plantar side of the proximal end cut off proximal cut on proximal articulation cut chop on middle part of diaphysis 13 14 13 14 middle 15 middle chop as 14, cut through i6 proximal dorsal saw proximal end of diaphysis sawn through I? distal dorsal saw distal end of diaphysis sawn through Ig dorsal 19 dorsal 20 distal dorsal 21 shaving marks on dorsal side of the diaphysis 20 21 chop longitudinally cut through on median cut chop on dorsal side of distal end of the diaphysis 201 METAPODIA 22 distal dorsal 22 23 cut chop on dorsal side of the distal articulation distal plantar 24 25 cut chop on plantar side of distal end of the diaphysis 25 26 27 distal plantar 26 27 cut chop on plantar side of the distal articulation distal dorsal chop (part of) the distal end of the distal articulation cut off 23 24 28 ai 29 •mr distal chop (part of) dorsal side of the distal articulation cut off 30 ML distal chop (part of) plantar side of the distal articulation cut off 202 Phalanx l, Bos taurus dorsal view PHALANGES peripheral proximal distal axial I volar 2 chop cut on proximal half of the volar side 2 I 3 4 volar 3 4 chop cut on distal half of the volar side 5 6 dorsal 5 6 chop cut on proximal half of the dorsal side dorsal 7 8 chop cut on distal half of the dorsal side 9 chop cut on axio-volar side 10 II 12 chop cut on proximal epiphysis 9 axial 10 II 12 proximal 203 Pelvis, Bos taurus ventral view PELVIS lateral caudal cranial medial ventral chop rim of acetabulum cut off from cranial side ventral chop rim of acetabulum cut off from caudal side ventral chop rim of acetabulum cut off from medial side ventral chop acetabulum on cranial side cut through ventral chop acetabulum on caudal side cut through ventral chop acetabulum on medial side cut through 7 8 chop below acetabulum on ventral side 9 chop lateral lateral 9 10 II 12 ventral 13 ventral 14 15 lateral i6 lateral 17 18 dorsal é^-— 19 dorsal 20 21 ventral 204 cut 10 cut II 12 chop 14 15 17 18 20 21 below acetabulum on dorsal side on ventro-medial side of coUum of ilium cut chop as 11, cut through chop cut on lateral side of coUum of ilium chop as 14, cut through chop cut on dorsal side of ilium chop as 17, cut through chop cut on ventral side of ilium 22 ~) ventral 23 ventral 24 25 ventral 26 ventral 27 28 ventral 29 "^ ventral chop as 20, cut through shaving marks on attachment point of sacrum 24 25 27 28 chop cut on ventral side of pubis chop as 24, cut through chop cut on ventral side of ischium chop as 27, cut through 205 Femur, Bos taurus cranial view FEMUR lateral proximal distal medial proximal chop cut on proximal side of caput femoris proximal cranial chop caput femoris cut off proximal cranial chop caput femoris and part of diaphysis cut off I 2 9 10 proximal medial 5 6 chop cut on medial part of caput femoris proximal medial 7 8 chop cut longitudinal on medial part of caput femoris proximal caudal 9 chop cut on caudal side of trochanter major 10 chop trochanter major latero-medially cut through chop cut longitudinal on caudal side of trochanter major chop trochanter major longitudinally cut through chop cut on medial side of the proximal end of the diaphysis II proximal caudal 12 proximal caudal 13 12 13 14 proximal caudal 15 i6 proximal medial I? cranial chop on lateral side of the middle part of the diaphysis cranial chop on medial side of the middle part of the diaphysis 19 cranial shaving mark(s) on the lateral side of the middle part of the diaphysis cranial shaving mark(s) on the medial side of the middle part of the diaphysis cranial 2o6 15 i6 chop diaphysis cut through in the middle distal lateral 23 chop cut on lateral side of the distal end of the diaphysis 23 24 25 distal medial 24 25 chop cut on medial side of the distal end of the diaphysis 26 27 distal medial 26 chop cut on caudo-medial side of the distal epiphysis (see also 29-30) 28 distal medial chop caudal part of the distal epiphysis cut off 29 30 distal medial chop cut on caudo-medial side of the distal epiphysis (see also 26-27) 31 distal medial chop caudo-distal part of the distal epiphysis cut off 32 distal cranial chop lateral part of the distal epiphysis cut off 33 distal cranial chop medial part of the distal epiphysis cut off 34 35 distal chop cut on distal side of the distal epiphysis 36 distal caudal chop distal part longitudinally cut through 22 22 27 29 30 34 35 207 TIBIA Tibia, Bos taurus dorsal view lateral proximal distal medial 13 14 17 18 208 chop dorsal part of proximal articulation cut off proximal chop plantar part of proximal articulation