The Iron Gates in Prehistory:

New perspectives
Edited by

Clive Bonsall Vasile Boronean Ivana Radovanovi

BAR International Series 1893 2008

This title published by Archaeopress Publishers of British Archaeological Reports Gordon House 276 Banbury Road Oxford OX2 7ED England
bar@archaeopress.com www.archaeopress.com

BAR S1893

The Iron Gates in Prehistory: New perspectives

the individual authors 2008

Cover and Title Page illustration: The Nymph drawn by N. Mitri!, based on a gural sculpture from Lepenski Vir now in the National Museum collection in Belgrade ((with permission)

ISBN 978 1 4073 0373 4

Printed in England by Blenheim Colour Ltd

All BAR titles are available from: Hadrian Books Ltd 122 Banbury Road Oxford OX2 7BP England
bar@hadrianbooks.co.uk

The current BAR catalogue with details of all titles in print, prices and means of payment is available free from Hadrian Books or may be downloaded from www.archaeopress.com

Faunal assemblages from the Early Neolithic of the central Balkans: methodological issues in the reconstruction of subsistence and land use
Haskel J. Greeneld

Abstract: The earliest food producing societies in Europe north of the Aegean littoral appear in the central Balkans during the Early Neolithic. As a result, they are frequently included in discussions of the evolution of animal domestication in Europe. However, relatively little zooarchaeological research has been undertaken in the central Balkans. This paper has two goals. First, it introduces and discusses the nature, quality and problems pertaining to each of the major Early Neolithic zooarchaeological samples from the central Balkans to demonstrate the difculty of inter-assemblage comparison. Second, it explores the potential of these assemblages for reconstructing animal exploitation strategies and land use patterns during the Early Neolithic, which are essential for understanding the economic processes involved in the evolution of early food producing societies in Southeast Europe. It is shown that it would be misleading to uncritically accept these assemblages as high quality sources of information for reconstructing the origins of animal domestication in Southeast Europe and animal exploitation strategies and land use patterns during the Early Neolithic. These assemblages have a very limited potential for understanding the economic processes involved in the evolution of early food producing societies in Southeast Europe. Key words: fauna, Neolithic, southeast Europe, subsistence, land use patterns

The earliest food producing societies in temperate Southeast Europe north of the Mediterranean littoral appear in the central-north part of the Balkan peninsula (hereafter referred to as the central Balkans Fig. 1) during the earlier half of the Neolithic. The Early Neolithic (Starevo culture) of the central Balkans, c. 73006400 BP (61505350 cal BC), witnessed the earliest spread of farming adaptations in Europe north of the Mediterranean littoral (Tringham 1971; Barker 1985). It is within this region that food production strategies based essentially upon an eastern Mediterranean complex of plants and animals are modied before spreading to the temperate climatic zones of Central and Northern Europe (Champion et al. 1984; Barker 1985). Many scholars have proposed explanatory models, often including economic or subsistence components (with a paucity of supporting data), for the spread of the Early Neolithic food producing cultures of temperate Southeast Europe (e.g. Ammerman & Cavalli-Sforza 1971; Garaanin 1973, 1983; Bknyi 1974a; Barker 1985; Whittle 1985). While the Early Neolithic of the central Balkans is considered one of the most crucial periods in European prehistory, it nonetheless remains one of the most poorly investigated in terms of prehistoric economics. But, if economic processes are recognized as fundamental to the emergence of food producing society, economic data must be collected and used to test models for their evolution. The appearance of early food producing societies in temperate Southeast Europe is often assumed by archaeologists to be the result of the migration of previously established food producing communities from the south (e.g. Childe 1929, 1958; Ammerman & Cavalli-Sforza 1971; Tringham 1971: 70ff.; Champion et al. 1984: 100, 120; cf. Dennell 1985: 153ff.; Whittle 1985: 54). Two often unstated assumptions are used to support this view of the spread of food producing societies into temperate Southeast Europe:

1. If domestic fauna are found to be dominant in Early Neolithic sites in the region, these sites belong to intrusive colonists colonizing an essentially open nearly uninhabited environment; and 2. Since terminal Mesolithic sites are poorly represented in the region (with the exception of agriculturally marginal areas such as the Danubian Iron Gates or Montenegrin highlands), signicant indigenous Mesolithic huntinggathering populations survived only in such areas, and slowly adopted domestic economies. Therefore, any sites with domesticated plants and animals found outside of these refugia were assumed to belong to intrusive agricultural populations. The rst Early Neolithic zooarchaeological studies from the region demonstrated a preponderance of domestic fauna in Early Neolithic sites outside of the Iron Gates (e.g. Bknyi 1974b, 1976, 1988; Clason 1980). These were used in support of this circular reasoning. Recent studies indicate a more complex picture. As will be shown in this paper, the frequency of domestic fauna in sites cannot be used to argue for the presence or absence of indigenous populations in and outside of agriculturally marginal areas. Rather, the frequencies of faunal remains in sites are a reection of differential resource exploitation, recovery methodology, and bone assemblage attrition. The Early Neolithic of the central Balkans has been extensively investigated because of the interest in the origins of food production in Europe. But, little effort has been placed upon the systematic recovery of faunal remains from such sites. This has resulted in a dramatically biased zooarchaeological database, and one that has been uncritically accepted in the literature to explain the origins of food production in Southeast Europe (e.g. Murray 1970; Barker 1985; Whittle 1985; Halstead 1988). Neither the few systematic zooarchaeological studies nor the secondary discussions employing such data have paid attention to the taphonomic history of

103

The Iron Gates in Prehistory

Figure 1. Map of the central Balkans, showing Early Neolithic sites with analyzed faunal assemblages mentioned in text. Site name abbreviations used in all gures and tables are: A=Anza; B=Blagotin; BC=Bukovaka esma; D=Divostin; F=Foeni-Sala; G=Golukot; HV=Hajduka Vodenica; LV=Lepenski Vir; LB=Ludo Budak; M=Madari; N=Nosa; O=Obre; P=Padina; RB=Rug Bair; S=Starevo; V=Vlasac.

samples or to the methodological problems connected with the faunal samples from this period and region (cf. Gifford 1981; Greeneld 1991, 1993). In the region, there is a dearth of published systematic analyses of Early Neolithic fauna. Most published accounts consist of species-frequency lists or preliminary reports, usually lacking detailed supporting data. These shortcomings in the database are signicant inuences during reconstruction of prehistoric economies on the basis of inter-site patterning in bone assemblages. It is only after methodological and taphonomic issues are addressed that animal exploitation strategies can be discussed. Some of the more important Early Neolithic faunal samples from the central Balkans are discussed here in order to demonstrate

some of the difculties involved in inter-assemblage comparison and their potential for reconstructing animal exploitation strategies and land use patterns.

The region, archaeological culture and sites

The region The central Balkans encompasses the eastern half of the former Yugoslavia. It includes the countries presently known as Serbia, Bosnia, and Macedonia. The term Balkan was originally a Turkish word, meaning a chain of (forested) mountains (Naval Intelligence Division 1944; Klai 1982). The

104

Haskel Greeneld: Early Neolithic faunal assemblages from the central Balkans

Balkan Peninsula is a topographically complex environment with several interconnected mountain systems coursing through the area. Highlands and lowlands are juxtaposed, with more than 37% of the land above 500m (Turnock 1989: 8). This variation in landforms within a small area has a strong inuence on local climate. The regional climate is transitional between that of temperate Central Europe and the more arid Mediterranean basin. Climate and plant and animal communities take on different characteristics not only in a general northsouth gradient, but also with increasing altitude. Neighbouring valleys often exhibit very different combinations of regional environmental variables, yet retain the general pattern of environmental diversity within the area as a whole (Pounds 1969). The Early Neolithic of the central Balkans The advent of the Neolithic in this region is generally connected to the appearance of food producing technology and adaptations, such as settled communities, ceramics, polished stone tools, domestic plants and animals. These are found in large quantities for the rst time with the Starevo culture. The Early Neolithic of the central Balkans coincides for the most part with the spatial extent of the Starevo culture c. 73006400 BP (61505350 cal BC cf. Tringham 1971; Gimbutas 1976; Garaanin 1983).1 Geographical variants of the Starevo culture are known from Romania (Cri) and Hungary (Krs). The Starevo culture has been temporally subdivided into several sub-phases. Several competing chronological systems have been proposed on the basis of, primarily, the seriation of ceramic materials found at stratied and unstratied sites throughout the region (e.g. Miloji 1949; Aranelovi-Garaanin 1954; Gimbutas 1976; Garaanin 1979, 1983; Srejovi 1988). It is considered to be the archaeological manifestation of the earliest Neolithic populations in the region. Large-scale excavations have demonstrated that Starevo settlements are not internally differentiated into functionally distinct areas (such as domestic houses, specialized work areas, cemeteries). There is little evidence for socioeconomic differentiation among houses or burials that would imply signicant status distinctions within communities. Remains of structures include both semi-subterranean and ground-level dwellings, often with associated hearths and refuse/storage pits. Artefacts include coarse and ne ceramics, small ceramic altars, amulets, chipped and ground stone tools, bone and antler tools, and unmodied animal bones and shells. The ceramics are thick-walled and decorated with a variety of typical Starevo decorative motifs, including mono- and polychrome painting, interior burnishing, pinching, impressing, channelling, and barbotine surface decorations (Tringham 1971; Garaanin 1979, 1983; Srejovi 1979; Lekovi 1985; Bogdanovi 1988; McPherron & Srejovi 1988; Chapman 1989; Greeneld 2000; Jongsma & Greeneld 2001). The sites Most of our knowledge of Early Neolithic subsistence derives from four groups of sites. Each is located in a different environmental context. The nature of the subsistence data differs between each of these environments.

1. The rst are found along the banks of the Danube in the gorge known as the Iron Gates (Lepenski Vir: Bknyi 1971; Padina: Clason 1980; Hajduka Vodenica: Greeneld, this volume). The sites in the gorge share access to a similar range of resources and have similar faunal proportions. 2. The second group is located in the northern end of the Mediterranean environmental zone. They are also at the southern end of the Starevo range. Sites include Anza and Rug Bair in a small Macedonian upland basin (Bknyi 1976; Schwartz 1976). 3. The third group includes those found in the rolling hills and river valleys of central Serbia and Bosnia, such as Divostin and Blagotin nestled among the rolling hills of central Serbia, Bukovaka esma in the Morava river valley, and Obre I in a valley deep in the mountains of northern Bosnia (Bknyi 1974b, 1988; Greeneld 1994, Greeneld & Jongsma n.d.). This is the transition zone between the Mediterranean and the Central European temperate climatic system. River valleys, such as the Morava, Bosna and Vardar were important routes for the spread of Early Neolithic adaptations from the Mediterranean to Central Europe and a centre for Starevo settlement (cf. Barker 1975; Srejovi 1979). As such, they represent an important link in understanding the spread of early farming adaptations out of the southern Balkans into the rest of Europe. 4. The fourth group is found in the Pannonian Plain (a nonpolitical term for the great lowlands encompassing Hungary, eastern Romania, and the northern part of the former Yugoslavia). These sites are at the northern end of the Starevo range. Sites from this group include the type site of Starevo on the Danube along the south edge of the plain and those further north and east, in the interior of the plain, such as Foeni-Sala, Ludo-Budak, Nosa, and Golukot (Bknyi 1974a; Clason 1980; Lazi 1988; Greeneld et al. n.d.). They are found close to the Hungarian or Romanian borders and are located in a more Central European climatic system (Pounds 1969).

Environmental variability in site location

Based on vegetation, climate and topography, the region can be divided into four major zones: 1. A Mediterranean zone in the southern half of the Balkan Peninsula, with low annual precipitation, high summer temperatures, and semi-steppe vegetation (Anza and Rug Bair). 2. A temperate transition zone in the hills of Serbia and Bosnia, transitional in many of its environmental characteristics between the Mediterranean and Central European systems. This zone is characterized by high, year-round precipitation, deciduous forests and strong soil development. The sites in this zone are located on hills overlooking river oodplains and surrounded by agriculturally fertile soils (Blagotin, Obre I, Divostin, Bukovaka esma). 3. A temperate highland zone in the Iron Gates gorge of the Danube, surrounded by the mountains of eastern Serbia, thick deciduous forests, soils with poor agricultural po-

105

The Iron Gates in Prehistory

tential, and overlooking the Danube and its rich aquatic resources (Hajduka Vodenica, Lepenski Vir, Padina); and 4. A temperate Central European lowland zone in the plains of Pannonia (Starevo, Golukot, Foeni-Sala, Nosa, and Ludo-Budak). These sites overlook the oodplains of rivers, streams and marshes, with ora and fauna adapted to soils with high moisture levels.

Some methodological issues in inter-assemblage comparison

Although consideration of the taphonomic history of each sample is a necessary precondition to the reconstruction of prehistoric subsistence and land use, most of the samples were not excavated, analyzed or published in a manner that allows a systematic reconsideration of their taphonomic history. Consideration of the methodological problems underlying the faunal studies can be used, however, to enhance our understanding of some of the inter-assemblage variability. Eventually, we may be able to differentiate patterning in the samples attributable to methodology from that of prehistoric human behaviour. Some of the more important methodological issues in inter-assemblage comparison are discussed next. The issues chosen for discussion are those that can be re-evaluated given the limited nature of most faunal publications. Recovery methodology It has long been a truism of faunal analysis that recovery methodology dramatically affects the outcome conclusions of the analysis (Payne 1972; Casteel 1973). As a result, collection methodology must be considered before any conclusions concerning human behaviour can be made. In general, bone collection from this region has been unsystematic. All of the samples were unsieved except for Anza, Blagotin, Foeni-Sala and Rug Bair. Only Foeni-Sala and Rug Bairs fauna were entirely sieved. The sieved and unsieved samples from Anza and Blagotin were not separated during analysis. Hand recovery affects the results of all the analyses in the following manner: 1. The bones of larger animals will be more completely collected than smaller animals. This will result in overrepresentation of larger animals in assemblages. For example, cattle remains were collected more frequently than sheep remains, over-representing them in the assemblage relative to their original frequencies. Such is the situation at Obre (Bknyi 1974b; Clason & Prummel 1977; Greeneld 1986, 1991). 2. The size of the bone is also an important variable. The larger the bone of the same species, the more likely it is that it will be collected. Large bones are more frequently collected than the smaller bones of the same taxon. Thus, reconstruction of preferential selection of body parts for use and/or disposal is compromised. If excavators preferentially collect large unbroken long bones (e.g. humerus) and ignore the smaller bones (e.g. phalanges, tarsals and carpals), then one may mistakenly assume that the prehistoric occupants disposed of the missing bones elsewhere. However, the reason that the smaller bones are missing from the assemblage is due more to the behaviour of the

archaeologists conducting the excavation than to the behaviour of the prehistoric occupants (Lyman 1994). While most faunal studies from the region do not list the type and frequency of various element categories, some studies do (Clason 1980; Greeneld 1994, n.d. a, b, this volume). There is a systematic recovery bias against small bone elements. When the distribution of element categories are compared in the unsieved samples, there is a total or near absence of small elements (phalanges, carpals and tarsals) for the medium- and small-sized mammals (e.g. ovicaprines, pigs, dogs, beaver). They are relatively more common among large mammals, both domestic and wild. In the sieved samples, the small element categories are more common. It is also possible to note the effect of recovery technology upon overall sample size by comparing the Iron Gates samples (Lepenski Vir, Padina, Vlasac, and Hajduka Vodenica Bknyi 1971, 1978; Clason 1980; Greeneld, this volume). The excavators of Lepenski Vir, Padina, and Hajduka Vodenica openly acknowledge that bone recovery was limited and biased toward what was considered to be tools or otherwise modied bones (Greeneld, this volume; B. Jovanovi, pers. comm.). The result was extremely low recovery rates when compared to Vlasac (Srejovi & Letica 1978) where efforts were made to avoid the mistakes of the earlier excavations in the gorge. The other assemblages are dominated by large mammal taxa (red deer, cattle). At Vlasac, contrary to the other excavations in the gorge, the remains of sh and various small- and medium-sized taxa dominate the assemblage in all phases. Quantication Only two methods have been used to quantify species representation from Early Neolithic faunal assemblages in the central Balkans: number of individual specimens (NISP) and minimum number of individuals (MNI) (see Bknyi 1970; Greeneld 1986, 1991 for reviews of procedures in the Balkans). In this study, only NISP counts are employed for several reasons. First, the problems with MNI seem to be more severe than with NISP (Grayson 1984). For example, both MNI and NISP appear to be equally predictive of species abundance in large samples (>10,000 fragments). Some of the samples, however, are relatively small, and NISP is more useful for the analysis of small samples, especially those with unequal species frequency ratios (Gilbert et al. 1982). Also, MNI counts are not given for all of the samples, being available only for the total remains of each species from six of the sites (Divostin, Lepenski Vir, Ludo-Budak, Nosa, Obre and Anza IIII). In each instance, the values were calculated in the same manner (Bknyi 1970). Finally, NISP counts are available for all of the sites, making the samples more comparable. In the NISP counts from Blagotin, FoeniSala, Hajduka Vodenica and Bukovaka esma, whole and partially articulated skeletons and limbs were not doublecounted. Articulated specimens were counted only once. Antler and horn fragments were few and mostly attached to cranial elements (Greeneld 1986). It is not known if the NISP counts from the other sites were calculated in exactly the same way since there is no discussion of NISP methodology in their publication. Even though the question of individual groups of articulated elements was

106

Haskel Greeneld: Early Neolithic faunal assemblages from the central Balkans Table 1. Frequency distribution of burnt and unburnt bones from Early Neolithic levels of each site.
Site Blagotin Context N Exterior Pit house - central Pit house - peripheral Plough zone above pithouse Subtotal All If grey bones are deducted All 20 319 482 2 823 1 497 2 Burnt % 2.27% 1.82% 3.58% 1.38% 2.57% 0.18% 6.06% 1.79% N 861 17228 12982 143 31214 543 7700 110 Unburnt % 97.73% 98.18% 96.42% 98.62% 97.43% 99.82% 93.94% 98.21% Total N 881 17547 13464 145 32037 544 8197 112

Bukovaka esma Foeni-Sala Hajduka Vodenica

Table 2. Frequency distribution of weathering stages from Early Neolithic levels of each site.
Weathering Stages Site Blagotin Context type Exterior in cultural horizon Exterior - in ploughzone above ZM02 Subtotal - Exterior Interior (Pit house - central) Interior (Pit house - peripheral) Subtotal - Interior Grand total Mixed Exterior in cultural horizon Interior (Pit house) Grand total Mixed 50 4 54 10761 2194 12955 13009 544 485 632 1117 27 Light 14.49% 2.17% 10.21% 61.27% 16.30% 41.77% 41.24% 100.00% 14.52% 12.20% 13.09% 24.11% Medium 133 135 268 5893 9994 15887 16155 0 2592 4375 6967 29 38.55% 73.37% 50.66% 33.56% 74.27% 51.22% 51.21% 0.00% 77.58% 84.46% 81.67% 25.89% 162 45 207 908 1268 2176 2383 0 264 183 447 56 Heavy 46.96% 24.46% 39.13% 5.17% 9.42% 7.02% 7.55% 0.00% 7.90% 3.53% 5.24% 50.00% Total 345 184 529 17562 13456 31018 31547 544 3341 5190 8531 112

Bukovaka esma Foeni-Sala

Hajduka Vodenica

not directly addressed in the previously published analyses of any of the sites except those analyzed by the author, the articulated remains of whole skeletons were not found at Anza, Obre, Divostin, Hajduka Vodenica and Bukovaka esma. Only the analyses conducted by the author have recognized the importance of documenting the presence of articulated limbs in order to modify NISP or MNIs. Articulated limbs were found at Blagotin, Bukovaka esma and Foeni-Sala (Bknyi 1974b, 1976, 1988; Greeneld 1994, this volume, n.d. a, b). Data publication Each of the published assemblages has been published in ways that limit reanalysis. Even though the samples from Anza, Divostin and Obre were excavated by both natural and articial stratigraphic units, the Early and Late Neolithic samples were not distinguished for publication except for initial summary statistics. For example, while species proportions over time are noted, changes in age and sex proportions are not reported. As a result, it is impossible to identify shifts in culling strategies. Curation Saving and storing bone collections in safe and accessible facilities is fundamental to the pursuit of the discipline. Advances made in one generation can be improved upon if

the next generation can access the materials. However, the central Balkans is plagued by poor storage facilities and an inadequately developed ethos in relation to the storage of zooarchaeological materials. The faunal remains from most of the largest collections were discarded immediately after analysis (Obre, Lepenski Vir, Divostin personal communications from Sandor Bknyi, Dragoslav Srejovi, Svetozar Stankovi). Only the bone tools were kept since they were considered to be artefacts, in addition to small samples of other bones. The faunal remains from Anza and Rug Bair were kept for several years in a storage shed outside of the local museum. By the time I tried to re-examine the Anza remains in 1982, rats had gotten into the shed. All of the bags and tags were destroyed, and the faunal remains subsequently discarded by the curators. The remains from Hajduka Vodenica and Padina are stored at the central storage facility for the Serbian Iron Gates sites at Karata. It is extremely difcult to access material in this facility since crates of material are stored one on top of the other, without any central cataloguing system. The remains from Starevo and Blagotin are stored in Belgrade (at the National Museum and the University of Belgrade, respectively), while those from Foeni-Sala are in Timioara, Romania (Muzeul Banatului). Selected samples from Hajduka Vodenica, Blagotin and Foeni-Sala are in Winnipeg, Canada (University of Manitoba). The fate of the other collections is

107

The Iron Gates in Prehistory Table 3. Faunal remains from central Balkan Early Neolithic sites. The number of taxa column includes all bones identied to a class or ner taxonomic level. Unidentied ovicaprines (i.e. those that are not identied to either Ovis or Capra) are not counted if Ovis or Capra are present. If Ovis or Capra are not present, then unidentied ovicaprines would be counted only once. All invertebrates, bird, and sh remains, respectively, are each treated as a single taxon owing to the low level of analysis accorded their remains (Bknyi 1970, 1974, 1976, 1988; Clason 1980; Lazi 1988; Moskalewska 1986; Schwartz 1976; Greeneld n.d. a, b).
Site Anza IIII Blagotin Bukovaka esma Divostin I Foeni-Sala Golukot Hajduka Vodenica Lepenski Vir III Ludo-Budak Madari Nosa Obre I Padina B Rug Bair Starevo Figure abbreviations A B BC D F G H L LB M N O P R S Number of taxa 14 18 11 14 17 10 11 21 18 17 16 19 19 9 19 138 719 166 203 416 893 211 1765 572 185 736 1700 2656 13 1029 Wild 4.25% 8.29% 61.48% 8.45% 19.87% 68.38% 84.40% 74.50% 20.91% 6.44% 71.88% 20.75% 92.64% 1.88% 25.91% Domestic 3112 7959 104 2198 1678 413 39 640 2163 2689 229 6491 211 680 2943 95.75% 91.71% 38.52% 91.55% 80.13% 31.62% 15.60% 27.02% 79.09% 93.56% 22.36% 79.25% 7.36% 98.12% 74.09% NISP 3250 8678 270 2401 2094 1306 250 2369 2735 2874 1024 8191 2867 693 3972 b Notes a a

a. Sieved and unsieved samples were not separated during analysis. b. Sieved fauna from the American excavations.

unknown at present, but I suspect that they have been discarded. The various analysts also had an inuence on curation. In general, Sandor Bknyi advised his Yugoslavian hosts that they could discard the remains after he nished his analysis. In contrast, all other analysts encouraged that their collections be curated. However, since the advent of the Balkan wars of the 1990s, the condition of any of the collections is uncertain.

Physico-chemical assemblage attrition

Most analyses ignore the issue of assemblage attrition through weathering and other destructive forces. The controversy surrounding the schlepp effect is a good example (Lyman 1994). Without a discussion of taphonomic variables, it is impossible to judge the value of assemblages for subsistence reconstructions. Burning Most Early Neolithic faunal reports do not mention burnt bones at all. Only a very small fraction of burnt bones were found at Blagotin (2.57%), Bukovaka esma (0.18%), Foeni-Sala (6%) and Hajduka Vodenica (1.8% Table 1). The highest values come from Foeni-Sala, which is two to three times higher than the values found in all the rest. Similarly low frequencies of burnt bones were found in Late Neolithic and post-Neolithic assemblages from the region (Greeneld 1986, 1991). Why are so few bones burnt? There are several possible reasons. Most bones probably were still covered with meat

when exposed to re, were boiled in a stew-like concoction, or had their meat removed prior to cooking. In each case, there would have been limited contact between the bone and the ame. Also, the burnt bone may have simply disintegrated in the soil, since burnt bone breaks up into small fragments more readily than uncooked bone (Boneld & Li 1966). Another reason that few pieces of burnt bone are found is differential disposal patterns. Cooking areas (hearths) and nearby middens are the depositional contexts in which burnt bone most likely would have been discarded. Other than in the Iron Gates sites, few hearths have been found. At Bukovaka esma, almost all of the bones around the two excavated hearths (trench 5, N=40; trench 8, N=11) were not burnt. At Anza, most debris was acknowledged to derive from exterior middens (Bknyi 1976), where the burnt bone would have broken up and disintegrated relatively quickly. However, due to the nature of excavation methodology at most of the sites that gathered together disparate pieces of bone from large-scale excavation units, no spatial analyses of bone distributions can be conducted. Why were so many more burnt bones found at Foeni-Sala than at the other sites? Given the absence of any evidence for large-scale burnt destruction of the settlement, and the fact that most of the burnt bone remains were from small unidentiable fragments, it would appear most likely that the extensive dry and wet sieving operation would have been responsible for the differences. Bone weathering and depositional context The degree of weathering of samples can also profoundly affect the level of reconstruction. Analyses that avoid this issue

108

Haskel Greeneld: Early Neolithic faunal assemblages from the central Balkans

Figure 2. Semi-logarithmic scale scatter-plot showing relationship between number of taxa identied and identied taxa assemblage size in each assemblage.

are in danger of reconstructing the patterns of preservation rather than any prehistoric behaviour. There are profound differences in preservation between the samples analyzed by the author. One major source of bone weathering is soil pH. Soil pH levels, however, were not measured for most sites, since they were excavated prior to an interest in taphonomy in the region. However, pH was measured at Blagotin and FoeniSala. At both, the Early Neolithic deposits ranged from pH 5.0 to 7.0, depending upon the context relatively normal values for the region. While pH may have had less of an inuence, depositional context probably had a great inuence. There appear to be regional variations in bone preservation. The material from the Iron Gates seems to be much more poorly preserved than the other regions. Most of the Hajduka Vodenica sample was heavily eroded (50% Table 2), with characteristic weathering cracks, pitted bone

surfaces, and the loss of delicate features. Based on a study of some of the bone tools from Lepenski Vir and Vlasac curated at the University of Belgrade, the state of preservation of the Hajduka Vodenica sample is probably characteristic of the other Iron Gates samples. In the central Serbian sites, bones tend to be much better preserved. The Bukovaka esma sample is remarkable for the preservation of even delicate morphological features (Greeneld 1994). All of the bones were very well preserved. At Divostin, bones were well preserved when buried in the ashy and alkaline soil of pits. Bones from the cultural horizon (which is composed of Smonica-type soils, with a heavy clay content and high water retention), however, were so fragile and soft that they fragmented or disintegrated during excavation (Lyneis 1988: 301).2 At nearby Blagotin, there were similar patterns of preservation to those found at Divostin (Greeneld 2000, n.d. a; Greeneld & Jongsma n.d.). Most of the bones were lightly (41%) or moderately (51%) weathered (Table 2). However, when the remains are divided between those found within features (such as pits) and those that lay in the open, the degree of preservation is very different. In the exterior deposits, whether in the plough zone or in the cultural deposits, most of the remains were either moderately (51%) or heavily (39%) weathered. Very few were lightly weathered. In the interior deposits, most of the remains were moderately (51%) or lightly (41%) weathered. There are some differences between specic deposits, but the effect of depositional context on preservation remains (Meadow 1978). Only one sample had information from Pannonia Foeni-Sala (Greeneld n.d. b; Greeneld et al. n.d.). It had a very different pattern of preservation. Most of the bones (81.6% Table 2) were only moderately preserved, regardless of whether bones were recovered in pits or in the exterior deposits. This is a reection of the sandy loam deposits of the site. It is difcult to gauge the pattern of bone preservation at tell sites (which had deeply stratied deposits that were advantageous for bone preservation Obre I, Anzabegovo) since there is no recorded information on these sites. In general, preservation is worst in the Iron Gates (where they were heavily eroded), those from Pannonia were better, and those from central Serbia were the best. Bukovaka esma had the best preserved assemblage of all. Little is known about preservation at the other sites in the region.

Discussion
Despite all of the above issues, can anything be said about subsistence practices? Before this can be accomplished, two other issues must be examined. Sample size and taxonomic diversity The samples under consideration range quite widely in size. While large samples are best for reconstruction of subsistence practices, even small assemblages should be considered. They can contribute information otherwise unobtainable from other sources. Even though the number of samples and the diversity of environmental settings of Starevo sites with analyzed fauna is extremely small, each 109

Figure 3. Normal scale scatter-plot showing relationship between number of taxa identied and identied taxa assemblage size in each assemblage.

The Iron Gates in Prehistory

Figure 4. Normal scale scatter-plot showing relationship between percentage of wild animals in the assemblage and identied taxa assemblage size.

adds a new dimension to our understanding of Starevo adaptations. Species counts are available by separate Starevo phases only for Anza. At Obre, it is difcult to correlate the species frequencies with major occupational phases or strata (cf. Bknyi 1974b; Greeneld 1991). The faunal samples from most sites were not published by separate levels, horizons, or phases within the Starevo layer(s) (except at Starevo, itself). As a result, it would be impossible to re-analyze the material, if and when the sub-phasing of the Starevo culture, in general, or excavation units at particular sites, becomes more rened. The total fragment count of unidentied specimens (Table 1) is unknown from half of the sites (Anza, Divostin, Lepenski Vir, Obre). Since total fragment count cannot be used to examine the impact of total number of fragments upon taxonomic diversity, a proxy measure must be used instead i.e. NISP. Only Obre and Blagotin have substantial samples of identied taxa (8000+ NISP). The rest have medium (20004000 bones) or small samples (c. 1000 or less). When the number of identied taxa are plotted against the number of bones identied to a genus or species in the sample on a semi-logarithmic scale, all the sites fall into a single large cluster (Fig. 2). A semi-logarithmic scale was chosen because the sample sizes are so disparate. It would appear that a curvilinear relationship exists between the two variables (y = 4005.6 Ln (x) 7917.9). Statistically, when a curvilinear relationship appears and there is a low correlation (R2 = 0.1951), it is possible that the nature of the scale was incorrect. When the two variables are plotted on a normal scale, a different pattern emerges. Two separate distributions appear (Fig. 3). In the lower half of the graph, there is a strong positive correlation between identied sample size and species

diversity (R2 = 0.73). In other words, the greater the size of the identied sample, the greater is the number of identied taxa. The small number of identied taxa in small assemblages is, therefore, a reection of sample size. Rare species may appear in small data sets, but less common forms appear regularly only when the sample reaches certain minimal size levels (usually several thousand identied fragments: Casteel 1973; Grayson 1984). Although the absence of rare species in small samples does not dramatically affect the overall analysis of subsistence systems, they are necessary for other types of analysis (e.g. ecological reconstruction: Hesse & Wapnish 1985). Several factors, such as recovery and preservation, may intervene to affect the normally linear relationship between the sample size of identied mammal bones and the number of mammal species. In the upper half of the graph, a different pattern appears, with the data from Blagotin and Obre. Major increases in taxonomic diversity can occur even when there are only modest increases in sample sizes. In these two cases, increases in sample size did not signicantly alter taxonomic diversity. It would appear that once sample sizes reach a certain level (above 5000 identied specimens), other variables affect taxonomic diversity. In these cases, location would have limited the range of taxa available for exploitation. Obre is found overlooking a major river, while Blagotin is found in the highlands and away from any major water source. Hence, location and available surrounding resources can be a compounding issue. Domestic:wild proportions Is there a variable governing the relative frequency of domestic and wild animals in assemblages? Initial comparison of the domestic:wild species percentages from Starevo sites reveals that half of the assemblages have relatively high and the other half relatively low percentages of wild animals. There is no overlap between the two groups. All three of the Iron Gates sites have very high wild species frequencies (>61%). But three of the sites outside of the Iron Gates (Ludo-Budak and Golukot in the Pannonian plain, and Bukovaka esma in central Serbia) also have very high wild species counts. The remaining eight sites outside of the Iron Gates have a preponderance of domestic animals (>74%). How can these distributions be interpreted? In the past, sites outside of the Iron Gates were assumed to be those of intrusive food producers because most of the fauna was domestic. In contrast, sites in the Iron Gates were assumed to be from indigenous hunter-gatherers because: 1. Even when food production appeared, it was never a mainstay of the local economy (cf. Bknyi 1971, 1978), and 2. The human skeletal record has been consistently interpreted to represent population continuity in the Iron Gates from the Terminal Palaeolithic through the Mesolithic and into the Early Neolithic (cf. ivanovi 1975, 1976, 1979).3 In fact, the reality is more complex; sites with high and low frequencies of wild fauna are found outside the Iron Gates. If we continue to use the logic of most researchers on the Early Neolithic of the region, does this mean that even sites outside of the Iron Gates belong to hunter-gatherers? This is a point

110

Haskel Greeneld: Early Neolithic faunal assemblages from the central Balkans

to be returned to later. When the number of identied fauna from each of the sites is plotted against the percentage of wild animals identied in the assemblage, three clusters appear (Fig. 4). Using the xaxis, the samples can be divided on the basis of percentage of wild animals one group with a high percentage of wild animals and two groups with low percentages. Each of the major groups can be subdivided on the basis of sample size: 1. The high wild percentage group contains sites with both low and medium sample sizes. This group comes from a variety of environmental settings. It includes the sites from the Iron Gates. The other sites in this group are found in the region beyond the narrow seclusion of the Iron Gates, in a variety of environmental contexts. The implications of these distributions appear fairly obvious. The high percentage of wild fauna found in the Iron Gates sites reect the isolation, topographic ruggedness and abundance of wild fauna, and the lower energy returns of pastoralism versus hunting-shing. The topography and environment of the gorge with its steep slopes and thick forests make pastoralism difcult and less protable as a subsistence strategy when compared with the hunting of wild ungulates. The high percentage of wild fauna in the sites outside the Iron Gates (Bukovaka esma, Golukot, and Nosa) is clearly inuenced by the location of the sites near or on major riverine environments. These sites are more similar to those from the Iron Gates. This is a very different pattern than in the other sites found outside of the Iron Gates sites rather than the generally accepted pattern for sites outside of the Iron Gates. 2. The low wild percentage group can be divided into two sub-groups: (i) one sub-group is composed of sites with small and medium sample sizes; (ii) the other sub-group is composed of sites with large sample sizes. On the surface, it would appear that the difference in wild:domestic ratios may be affected by sample size. The smaller samples tend to have higher percentages of wild animals. The situation is more complex. One variable is recovery method (sieving or hand recovery). The partially sieved samples include both those with large (Blagotin) and medium sample sizes (Anza) among the low wild group. Sieving does not appear to have much of an effect on wild: domestic ratios. The preponderance of wild fauna cannot be attributed entirely to recovery method since wild animals are generally not larger than domestic animals. If anything, many wild taxa are smaller (e.g. beaver and sh) and would be more difcult to recover by hand. Hand collection would under-represent the wild component of the assemblage. Fish remains, in particular, are poorly collected by hand and therefore could be predicted to have been present in larger quantities before excavation. For example, they already represent 10% of the Bukovaka esma sample, a hand collected site located above the oodplain of the Morava river. They were probably present in even greater frequencies before excavation. Another possibility is the surrounding environment and its relationship to the subsistence economy. Most of the sites with high frequencies of wild fauna are found in a limited set of environments in the temperate environmental zone, and close to rivers, streams and oodplains that even today abound with wildlife. The sites with smaller sample sizes also 111

happen to be in such environments. Therefore, their high wild frequencies may not be a reection of sample size, but of their exploitation of the rich wild resources in the surrounding wetlands. The faunal remains of Bukovaka esma, Golukot and Nosa are more similar to those from the Iron Gates than from other sites spread through the hill country of Serbia, Bosnia or Macedonia because they also overlook resource-rich aquatic environments. Much of the source of variation between samples can therefore be attributed to a limited set of variables: recovery methodology, sample size and environmental location. Therefore, the percentage of wild or domestic fauna may still be used as a general reection of overall environment. Recovery can alter the details signicantly, nonetheless. Much of the source of variation between assemblages can therefore be attributed to a limited set of variables: recovery methodology, sample size and environmental location. The quantity of wild remains in sites outside of the Iron Gates can be relatively high indicating that traditional assumptions concerning the nature and rate of spread of the transition to a food producing economy rate in Southeast Europe need to be re-evaluated. The spread of food production may be a more complex process than previously envisioned. Some thoughts on the transition to food production in Southeast Europe If we accept the assumption that sites with wild resources in the Iron Gates are the remains of indigenous foragers, then it may even be that some of the sites outside of the Iron Gates with a predominance of wild resources may also be the remains of acculturating foragers. In fact, we know too little about the relationship between Mesolithic and Early Neolithic populations in this region to be able to make such blanket assumptions. While Early Neolithic sites are widespread across the region, so much is assumed in Mesolithic studies on the basis of a few highly concentrated localities (Iron Gates, Montengro). For much of the Early Neolithic of Central and Northern Europe, indigenous Mesolithic foragers probably coexisted with initial and intrusive food producers. All we need to do is to look at the evidence from other better-studied regions for comparison. For the rst phase of the Early Neolithic of Central and Northern Europe, indigenous Mesolithic foragers coexisted with initial and intrusive food producers (e.g. Bogucki 1988). There is also evidence for acculturation by indigenous foragers to an agricultural lifestyle (Price 1987). In contrast, most studies of Southeast European prehistory implicitly assume that either the indigenous population immediately adopted food production upon its arrival or the area was not occupied by foragers at the advent of the Neolithic. The Mesolithic forager sites are not visible in the areas of intensive Early Neolithic settlement (in the great river valleys, beyond the Iron Gates and Montenegrin highlands). The widespread presence of Mesolithic sites (in areas that have been systematically surveyed, i.e. Iron Gates Whittle 1985; or in regions with little vegetative cover, e.g. Montenegro Djurii 1991) and the demonstrable continuity between Mesolithic and Early Neolithic human populations in the more mountainous and less accessible areas of the peninsula (i.e. the Iron Gates ivanovi 1975, 1976,

The Iron Gates in Prehistory

1979; yEdynak 1978) indicate that it is likely that foragers were also present in the lowlands at the advent of the Neolithic. If indigenous foragers were present at the advent of the Neolithic, it might be more realistic to assume that they coexisted with early food producers for a period of time, only slowly abandoning foraging subsistence economies? During this transition, they may have retained much of their commitment to foraging while at the same time incorporating some domestic food producing strategies and technologies into their subsistence round (as occurred in Southwest, Northwest and Northeast Europe Price 1987; Rimantiene 1992), such as domestic animals, pottery production, or other hallmarks of Starevo sites. If this is the case, we might expect foraging communities to retain their mobility, while adopting and incorporating domestic animals into their mobile subsistence system before adopting other Early Neolithic accoutrements (such as pottery) or vice versa as in northern and Western Europe (cf. Zvelebil & Dolukhanov 1991). If this was the case, many of the sites dened as Mesolithic on the basis of lithic typology and the absence of ceramics may, in fact, date from the Early Neolithic. The paucity of radiocarbon dates from Mesolithic-type sites makes this a difcult hypothesis to test immediately. However, there are no cases where domestic animals are found in clear association with a Mesolithic-type stone tool technology (and where Early Neolithic pottery is absent) outside of the Iron Gates. The few early Holocene sites without pottery, but with a Mesolithic stone assemblage, do not contain any domestic animals. But there are several sites outside of the Iron Gates with Early Neolithic material culture, but with assemblages dominated by wild animals (Greeneld, this volume). The question then arises to whom do these settlements actually belong? 1. Are they settlements of indigenous foragers slowly acculturating to a food producing economy by selectively incorporating components of typical Starevo material culture? 2. Are they a specialized wild food collection site for a larger community committed to domestic economies? 3. Do they represent new settlements established by recently arrived food producers who have yet to build up the agricultural component of their local economy? 4. Or, could they be settlements of food producers who have moved into an agriculturally marginal area? Each alternative has an emphasis upon wild resources as a logical explanation for the patterning in the data. How to choose between alternatives? The second and third choices seem less likely considering the size and degree of permanency of the settlements, as reected by the material culture and domestic architecture, and the absence of any evidence for specialized production and exchange between contemporary sites involving foodstuffs at this time or even later in the Neolithic. Estimations of the seasonality of death of various species based upon the ageat-death of the various species at several sites also indicate year-round occupation (Clason 1980; Arnold & Greeneld 2006). It is more difcult to distinguish between the rst and fourth choices. The choice varies depending upon the evidence chosen to consider. In the Iron Gates, there is evidence

for human population, material culture, and subsistence continuity between Mesolithic and Early Neolithic levels, even with the appearance of new subsistence strategies and material culture. But comparable studies have not been undertaken upon assemblages from Mesolithic sites outside of the Iron Gates. Also, virtually all sites with a Mesolithic-type artefact assemblage from the region outside of the Iron Gates are not dated by radiocarbon techniques. They are assumed to date earlier than Starevo on the basis of the similarity of their material cultural assemblages to other Mesolithic sites in the Iron Gates and on the absence of Starevo ceramic material culture.

Conclusion
Previous studies of the earliest food producing communities in the more temperate zones of Southeast Europe, such as the central Balkans, have generally attempted to show that subsistence economies were largely orientated towards domestic animal economies (e.g. Murray 1970; Barker 1985). Faunal data from Early Neolithic sites in the central Balkans are reconsidered here to demonstrate that animal exploitation strategies during the Early Neolithic of the region was a relatively complex matter and that taphonomic issues must be considered if prehistoric economies are to be understood. Even though such faunal assemblages can increase our understanding of Early Neolithic subsistence strategies, it is not an easy leap from bones to human behaviour. A number of studies have demonstrated the obvious and not so obvious pitfalls of extrapolating from patterning in bone assemblages to prehistoric human behaviour (e.g. Gifford 1981; Greeneld 1986, 1988a, 1988b, 1991; Koster & Chang 1986). All of the analyses of Early Neolithic faunal assemblages and the ensuing comparative use of their results have ignored the possible distorting effects of bone assemblage attrition (e.g. Champion et al. 1984; Lazi 1988). There is a great deal of variability from site to site which affects the representativeness of the sample and utility of the nal analysis, such as variability of the surrounding environmental conditions (temperature, precipitation), local sedimentary conditions (including soil type and pH), recovery procedures, and methods of quantication. As a result of the re-evaluation of the quality of the Early Neolithic faunal data derived from such an approach, it is hoped that this will force a reconsideration of the signicance and value of conclusions derived from previously analyzed faunal assemblages. Notes
1. Manson (1990, this volume) established an age range of 61005100 cal BC, based on the archaeomagnetic intensity analysis of Starevo and Krs ceramic sherds. 2. But note that Bknyis (1988) faunal report does not deal with this issue. 3. Skeletal series from Starevo sites outside of the Iron Gates are few and far between. However, they are now being re-analyzed and show continuity with Mesolithic populations (see this volume).

Acknowledgements
I should like to express my sincere gratitude to my numerous friends

112

Haskel Greeneld: Early Neolithic faunal assemblages from the central Balkans and colleagues throughout the central Balkans who made available their assemblages for analysis or who helped in the excavation and analysis. Unfortunatel, there are too many to list separately here. Funding for this research came from a variety of sources, including the National Science Foundation (#BNS8105358), the WennerGren Foundation for Anthropological Research (#4210), the Fulbright-Hays Foreign Language and Area Studies Program (Program no. 84.022; Project no. 022AH10048), the International Research and Exchanges Board (198182, 199394), the University of Manitoba, Winnipeg (199099), and the Social Science and Humanities Research Council of Canada (19901996). I am grateful to all of them for supporting the eldwork, analysis and writing up of this paper. In particular, I thank my wife, Tina Jongsma, who shared much of the eld and laboratory analysis that led to this paper. Any errors, however, are my responsibility. Press. 1958: The Prehistory of European Society. Harmondsworth: Penguin. Clason, A.T. 1980: Padina and Starevo: game, sh and cattle. Palaeohistoria 22: 142173. Clason, A.T. & Prummel, W. 1977: Collecting, sieving and archaeozoological research. Journal of Archaeological Science 4: 171175. Dennell, R. 1984: European Economic Prehistory: A New Approach. London: Academic Press. Djurii, L. 1991: Surveying Paleolithic and Mesolithic Sites (in Serbian). Glasnik Srpsko Arheolosko Drutvo (Journal of the Serbian Archaeological Society Belgrade) 7: 4351. Garaanin, M. 1973: Praistorija na Tlu SR Srbija, 2nd edition. Belgrade: Srpska Knievna Zadruga. 1979: Centralbalkanska zona, In Basler, D., Benac, A., Gabrovec, S., Garaanin, M., Tasi, N., Covi, B. & VinskiGasparini, K. (eds) Praistorija Jugoslavenskih Zemalja II: Neolitsko Doba. Sarajevo: Akademija Nauka i Umjetnosti Bosne i Hercegovine, 79212. 1983: The Stone Age in the central Balkans. In Cambridge Ancient History, volume 3, part 1. Cambridge: Cambridge University Press, 75135. Gifford, D.P. 1981: Taphonomy and paleoecology: a critical review of archaeologys sister disciplines. In Schiffer, M.B. (ed.) Advances in Archaeological Method and Theory, volume 4. New York: Academic Press, 365438. Gilbert, A.S., Singer, B.H. & Perkins, D. 1982. Quantication experiments on computer-simulated faunal collections. Ossa 8: 7994. Gimbutas, M. (ed.) 1976: Neolithic Macedonia as Reected by Excavation at Anza, Southeast Yugoslavia. Los Angeles: The University of California, Los Angeles (Monumenta Archaeologica, vol. 1). Grayson, D.K. 1984: Quantitative Zooarchaeology. New York: Academic Press. Greeneld, H.J. 1986: The Paleoeconomy of the Central Balkans (Serbia): A Zooarchaeological Perspective on the Late Neolithic and Bronze Age (45001000 BC). BAR International Series 304. Oxford: British Archaeological Reports. 1988a: Bone consumption by pigs in a contemporary Serbian village: implications for the interpretation of prehistoric faunal assemblages. Journal of Field Archaeology 15: 473479. 1988b: The origins of milk and wool production in the Old World: a zooarchaeological perspective from the central Balkans. Current Anthropology 22: 573593. 1991: Fauna from the Late Neolithic of the central Balkans: issues in subsistence and land use. Journal of Field Archaeology 18: 161186. 1993: Zooarchaeology, taphonomy, and the origins of food production in the central Balkans. In Jamieson, R.W., Abonyi, S. & Mirau, N.A. (eds) Culture and Environment: A Fragile Coexistence. Proceedings of the 24th Chacmool Conference. Calgary: University of Calgary Archaeological Association, 111117. 1994: Faunal remains from the Early Neolithic Starevo settlement at Bukovaka esma, Starinar n.s. 43/44: 103114. 2000: Integrating surface and subsurface reconnaissance data in the study of stratigraphically complex sites: Blagotin, Serbia. Geoarchaeology 15: 167201. n.d. a: Vertebrate Fauna from Blagotin. Unpublished manuscript on le, University of Manitoba, Department of Anthropology. n.d. b: Vertebrate Fauna from Foeni-Sala. Unpublished manuscript on le, University of Manitoba, Department of Anthropology.

References
Ammerman, A. & Cavalli-Sforza, L.L. 1971: Measuring the rate of spread of early farming in Europe, Man n.s. 6: 674688. Aranelovi-Garaanin, D. 1954: Starevaka Kultura. Ljubljana: Universitet v Ljubljana. Arnold, E.R. & Greeneld, H.J. 2006: The Origins of Transhumant Pastoralism in Temperate Southeastern Europe: A Zooarchaeological Perspective from the Central Balkans. British Archaeological Reports, International Series 1538. Oxford: Archaeopress. Barker, G. 1975: Early Neolithic land use in Yugoslavia. Proceedings of the Prehistoric Society 41: 85104. 1985: Prehistoric Farming in Europe. Cambridge: Cambridge University Press. Bogucki, P. 1988: Forest Farmers and Stockherders: Early Agriculture and its Consequences in North-Central Europe. Cambridge: Cambridge University Press. Bknyi, S. 1970: A new method for the determination of the number of individuals in animal bone material. American Journal of Archaeology 74: 291292. 1971: Animal remains from Lepenski Vir. Science 167: 17021704. 1974a: History of Domestic Mammals in Central and Eastern Europe. Budapest: Akadmiai Kiad. 1974b: The vertebrate fauna. In Gimbutas, M. (ed.) Obre I and II. Neolithic Sites in Bosnia. Wissenschaftliche Mitteilungen des Bosnisch-Herzegowinischen Landesmuseum, Arheologie 3, part B: 55154 (Sarajevo). 1976: The vertebrate fauna from Anza. In Gimbutas, M. (ed.) Neolithic Macedonia as Reected by Excavation at Anza, Southeast Yugoslavia. Los Angeles: The University of California, Los Angeles (Monumenta Archaeologica, vol. 1), 313363. 1978: The vertebrate fauna of Vlasac. In Srejovi, D. & Letica, Z. (eds) Vlasac: A Mesolithic Settlement in the Iron Gates, vol. 2. Belgrade: Serbian Academy of Science and Arts, 3565. 1988: The Neolithic fauna of Divostin. In McPherron, A. & Srejovi, D. (eds) Divostin and the Neolithic of Central Serbia. Pittsburgh: University of Pittsburgh, Department of Anthropology (Ethnology Monographs no. 10), 419446. Boneld, W. & Li, C.H. 1966: Deformation and fracture of bone. Journal of Applied Physics 37: 869875. Casteel, R.W. 1973: Some biases in the recovery of archaeological faunal remains. Proceedings of the Prehistoric Society 39: 382388. Champion, T., Gamble, C., Shennan, S. & Whittle, A. 1984: Prehistoric Europe. London: Academic Press. Childe, V.G. 1929: The Danube in Prehistory. Oxford: Clarendon
1

113

The Iron Gates in Prehistory Greenfield, H.J. & Jongsma, T.L. n.d.: Early Neolithic Blagotin: a summary of recent research. In Bailey, D.W., Whittle, A. & Cummins, V. (eds) Sedentism in the Central and East European Neolithic. Oxford: Oxbow Books. Greenfield, H.J., Jongsma,T.L. & Jezik, S. n.d.: Sedentary pastoral gatherers in the Early Neolithic architectural, botanical, and zoological evidence for mobile economies from Foeni-Sala, SW Romania. In Bailey, D.W., Whittle, A. & Cummins, V. (eds) Sedentism in the Central and East European Neolithic. Oxford: Oxbow Books. Halstead, P. 1989: Like rising damp? An ecological approach to the spread of farming in south east and central Europe. In Milles, A., Williams, D. & Gardner, N. (eds) The Beginnings of Agriculture. BAR International Series 406. Oxford: B.A.R., 2353. Hesse, B. & Wapnish, P. 1985: Animal Bone Archaeology: From Objectives to Analysis. Washington D.C.: Taraxacum Press. Jongsma, T. & Greeneld, H.J. 2001: Architectural technology and the spread of early agricultural societies in temperate southeastern Europe. In Tupakka, S., Gillespie, J. & de Mille, C. (eds) Untrampled Ground Untrammelled Views: Human Exploitation of, and Settlement Patterns on, New Landscapes. Proceedings of the 31st Annual Chacmool Conference, 1998. Calgary: University of Calgary, Department of Archaeology, 181200. Klai, B. 1982: Rjecnik Stranih Rijeci. Zagreb: Naklodni Zavod Matice Hrvatske. Koster, H. & Chang, C. 1986: Beyond bones: toward an archaeology of pastoralism. In Schiffer, M.B. (ed.) Advances in Archaeological Method and Theory, volume 9. Orlando: Academic Press, 97148. Lazi, M. 1988: Fauna of mammals from the Neolithic settlements in Serbia. In Srejovi, D. (ed.) The Neolithic of Serbia. Belgrade: University of Belgrade, 2438. Lekovi, V. 1985: The Starevo mortuary practices new perspectives. Godinjak (Sarajevo: Centar za Balkanaloka Istraivanja) 15: 157172. Lyman, R.L. 1994: Vertebrate Taphonomy. Cambridge: Cambridge University Press. Lyneis, M.M. 1988: Antler and bone artifacts from Divostin. In McPherron, A. & Srejovi, D. (eds) Divostin and the Neolithic of Central Serbia. Pittsburgh: University of Pittsburgh, Department of Anthropology (Ethnology Monographs no. 10), 301319. Manson, J.L. 1990: A Reanalysis of Starevo Culture Ceramics: Implications for Neolithic Development in the Balkans. Unpublished Ph.D. dissertation. Southern Illinois University, Carbondale, IL. Meadow, R. 1978: Effects of context on the interpretation of faunal remains. A case study. In Meadow, R. & Zeder, M.A. (eds) Approaches to Faunal Analysis in the Middle East. Cambridge, MA: Harvard University Peabody Museum Bulletin no. 2, 1521. Miloji, V. 1949: Chronologie der Jungeren Steinzeit Mittel- und Sdosteuropas. Berlin: Gebruder Mann. Murray, J. 1970: The First European Agriculture: A Study of the Osteological and Botanical Evidence until 2000 B.C. Edinburgh: Edinburgh University Press. Naval Intelligence Division. 1944: Jugoslavia. Geographical Handbook Series, vol. 1. London: Naval Intelligence Division. Payne, S. 1972: Partial recovery and sample bias: the results of some sieving experiments, In Higgs, E.S. (ed.) Papers in Economic Prehistory. Cambridge: Cambridge University Press, 4964. Pounds, N. 1969: Eastern Europe. Chicago: Chicago University Press. Price, T.D. 1987: The Mesolithic of western Europe. Journal of World Prehistory 1: 225306. Rimantiene, R. 1992: The Neolithic of the eastern Baltic. Journal of World Prehistory 6: 97143. Schwartz, C. 1976: The vertebrate fauna from Rug Bair. In Gimbutas, M. (ed.) Neolithic Macedonia as Reected by Excavation at Anza, Southeast Yugoslavia. Los Angeles: The University of California, Los Angeles (Monumenta Archaeologica, vol. 1), 364374. Srejovi, D. 1979: Protoneolit kultura Lepenskog vira. In Basler, D., Benac, A., Gabrovec, S., Garaanin, M., Tasi, N., Covi, B. & Vinski-Gasparini, K. (eds) Praistorija Jugoslavenskih Zemalja II: Neolitsko Doba. Sarajevo: Akademija Nauka i Umjetnosti Bosne i Hercegovine, 3378. Srejovi, D. & Letica, Z. (eds) 1978: Vlasac: A Mesolithic Settlement in the Iron Gates, 2 volumes. Belgrade: Serbian Academy of Arts and Sciences. Tringham, R. 1971: Hunters, Fishers and Farmers of Eastern Europe, 60003000 B.C. London: Hutchinson University Library. Turnock, D. 1989: The Human Geography of Eastern Europe. London: Routledge. Whittle, A. 1985: Neolithic Europe: A Survey. Cambridge: Cambridge University Press. yEdynak, G. 1978: Culture, diet and dental reduction in Mesolithic forager-shers of Yugoslavia. Current Anthropology 19: 616618. ivanovi, S. 1975: A note on the anthropological characteristics of the Padina population. Zeitschrift fr Morphologie und Anthropologie 66: 161175. 1976: Cromagnon in the Iron Gate Gorge of the Danube. Nature 260: 518. 1979: Further evidence on Cro-Magnon in the Iron Gate Gorge of the Danube. Current Anthropology 20: 805806. Zvelebil, M. & Dolukhanov, P.M. 1991: The transition to farming in eastern and northern Europe. Journal of World Prehistory 5: 233278.

114