Spatial patterns of social organization in the Early Bronze Age of South Scandinavia

Spatial patterns of social organization in the Early Bronze Age of South Scandinavia

Journal of Anthropological Archaeology 23 (2004) 33–55 www.elsevier.com/locate/jaa Spatial patterns of social organization in the Early Bronze Age of...

5MB Sizes 0 Downloads 60 Views

Journal of Anthropological Archaeology 23 (2004) 33–55 www.elsevier.com/locate/jaa

Spatial patterns of social organization in the Early Bronze Age of South Scandinavia Kasper Lambert Johansen,* Steffen Terp Laursen, and Mads K€ ahler Holst Department of Prehistoric Archaeology, University of Aarhus, Moesg ard, DK-8270 Højbjerg, Denmark Received 17 April 2003; revised 21 August 2003

Abstract An explanatory framework invoking the existence of chiefdoms has governed the interpretation of Early Bronze Age social structure in South Scandinavia since the 1970s. The comprehensive evidence provided by the burial mounds has played a prominent role in advancing this model. A chiefdom organization implies a specific, centralized spatial organization, and in this connection important evidence afforded by the burial mound record has been neglected—namely, the distinct linear structures evident in their distribution. This contribution focuses upon the barrow line structures and their relation to interaction and social organization. Taking as its point of departure a discussion of agency and structure, it proposes a theory of the formation of the barrow line structures. This constitutes the basis of a comparison of network analyses of the barrow distribution with correspondence analyses of artefacts, within a region in Denmark. A distinct correlation between a material expression of wealth, and nodes of high centrality in the network of barrow lines, is demonstrated. Based on these analyses, it is argued that though the Early Bronze Age of South Scandinavia was complex, the chiefdom model is not the most appropriate explanatory model available for understanding the social and spatial organization of this society.  2003 Elsevier Inc. All rights reserved. Keywords: Barrows; Bronze Age; South Scandinavia; Network; Interaction; Agency and structure; Centrality; Social organization

Walter Christaller developed ÔCentral Place TheoryÕ in the 1930s in an attempt to explain the variation in size, function, number, and spatial configuration of contemporary urban settlements in southern Germany (Christaller, 1968; Haggett et al., 1977, pp. 57, 139; Hodder and Orton, 1976, p. 60; Wagstaff, 1986, p. 119). In modeling the optimal exploitation of an isomorphic landscape, Central Place Theory was embraced by archaeologists decades later in the context of the functionalist perspectives characteristic of early processualism (Collis, 1986,

* Corresponding author. E-mail addresses: [email protected] (K.L. Johansen), [email protected] (S.T. Laursen), [email protected] (M.K. Holst).

p. 37). Coinciding with the introduction of neo-evolutionary classifications of prehistoric social organization, Ôcentral placesÕ became a diagnostic feature in identifying chiefdoms, and an important element in the study of complex societies in general (Renfrew, 1977, p. 100; Service, 1971). A ÔchiefdomÕ has been defined as a polity centrally organizing a regional population numbering in the thousands (Earle, 1991, p. 1), characterized by a formalized, personified leadership, kin-based hierarchy, surplus production, a degree of diversification and specialization, and a re-distributive economic system (Service, 1971, p. 133). Economic, social, political, and religious activities are centrally coordinated by the chief from the central place. In few other contexts in prehistoric archaeological research has chiefdom organization been argued as strongly as in connection with the Early Bronze Age

0278-4165/$ - see front matter  2003 Elsevier Inc. All rights reserved. doi:10.1016/j.jaa.2003.10.002

34

K.L. Johansen et al. / Journal of Anthropological Archaeology 23 (2004) 33–55

(EBA) of South Scandinavia, and its spatial correlate, the central place, has been eagerly pursued (Earle et al., 1998; Kristiansen, 1982, p. 262). However, since the chiefdom model was first espoused by researchers concerned with this period, the archaeological record has increased considerably, and still clear evidence of central places is absent. This has rendered the accordance between the chiefdom model and the evidence available less obvious. As the finds compromising the model have appeared gradually, it has been modified incrementally rather than being thoroughly revised and its customary application explicitly challenged (Kristiansen, 1991). The purpose of this contribution is to outline an alternative and more cogent explanatory framework in which chiefs and central places are absent as organizing elements, despite a high degree of social complexity. The first section gives an overview of the EBA of South Scandinavia as it appears today, with an emphasis on recent discoveries and theories which challenge the prevailing model of a chiefdom society. The second section presents a regional case study of the linear distributions of barrows in central Jutland (Denmark). Focussing on the significance of the socially integrating events of barrow construction, a large body of archaeological data is examined from a new methodological and theoretical perspective which embodies network theory, quantitative analysis, and concepts of agency and structure. The barrow lines of central Jutland area are argued to reflect a social interaction network, and based on this spatial structure the social organization of EBA society is reassessed. Finally, a developmental sequence is outlined in which a transformation of agency and a gradual spatial institutionalization of social organization indirectly form a prelude to the rise of regular chiefdoms in South Scandinavia during the Late Bronze Age.

The Early Bronze Age The EBA of South Scandinavia dates from approximately 1700 to 1100 BC corresponding to the Late Early Bronze Age and the Middle Bronze Age of Central Europe. The period is preceded by the Corded Ware Culture (2800–2400) and the Late Neolithic (2400–1700 BC), and followed by the Late Bronze Age (1100–500 BC). A tripartite division of the EBA (Period I–III) was presented by Montelius in 1885 (1885, 1900) and with refinements it is still in use today (Randsborg, 1968, 1993). Throughout the EBA the region is distinguished by a general homogeneity of material culture—especially with respect to metal artefacts which, despite some degree of regional variation, have traditionally been regarded as distinct from those encountered in neighbouring regions (Kersten, 1935; M€ uller, 1914). At the same time, however, the find material also bears

witness to the close incorporation of South Scandinavia in the general European network of exchange of material culture and possibly ideology. In this connection, metal exchange has elicited considerable attention due to the striking discrepancy between a high occurrence of metal artefacts in South Scandinavia and the lack of copper and tin resources available for production of such objects in this area (Kristiansen 1987, 1998a, p. 359; Thrane, 1975). Throughout the period the relations to Central Europe appear remarkably strong, and particularly the ties to the Tumulus Culture are significant (Hachmann, 1958; Randsborg, 1968; Vandkilde, 1996). Research on the EBA of South Scandinavia has traditionally been strongly focussed on metal artefact assemblages from graves and hoards, as these contexts dominate the archaeological record of the period. The vast majority of hoards are probably to be interpreted as votive deposits, and they continue a Neolithic tradition in South Scandinavia of sacrificial activity in wetland areas (Koch, 1998). As regards the graves, the EBA of South Scandinavia is renowned for its immense mound building, and especially the extraordinarily well-preserved oak-log coffin graves found in a limited number of barrows with iron pan encapsulated cores (Boye, 1896; Holst et al., 2001). In Denmark alone, more than half of the 86.000 barrows recorded in the national SMR should probably be ascribed to the EBA (www.dkconline.dk). The South Scandinavian barrow phenomenon is a local manifestation of the second millennium BC pan-European Tumulus Complex. However, it also relates to the local barrow tradition of the Corded Ware Culture and the Late Neolithic (Hansen and Rostholm, 1993), and it has a very distinct expression in regard to the size of the barrows. The mounds, which are constructed of stacked sods, have an average diameter of 15–20 m and are normally demarcated by varying types of kerb stone arrangements (Fig. 1). Usually one or more inhumation graves, in the form of stone supported oak-log coffins containing the deceased and a range of grave goods, are found. A varying number of secondary burials are commonly encountered, some associated with phases of secondary enlargement of the mounds, and others simply dug into the surface of an existing mound (Aner and Kersten, 1973; Jensen, 2002, p. 144). The burial mounds have played an important role in the reconstruction of social structure in the EBA of South Scandinavia. They are often presumed to be monuments containing the bodies of the social elite, namely, the chief and his closest kin (Kristiansen, 1998b, p. 288). An analysis of the age and sex of those interred within mounds indicates that only a proportion of the population is represented, as females and young individuals are highly under represented (Jensen, 1982, p. 169). The discovery of a small number of graves without barrows also demonstrates the existence of a parallel burial custom in the EBA (Broholm, 1940;

K.L. Johansen et al. / Journal of Anthropological Archaeology 23 (2004) 33–55

35

Fig. 1. The Magrethenberg mound as an example of a South Scandinavian EBA burial mound along with a schematic representation of stratigraphic relations (Aner and Kersten, 1973, no. 2186). The barrow phases are apparently all constructed during a single event. The burials both pre-date (A, B1, B2, D, and E) and post-date (C and F) the erection of the mound. Reproduced with the permission of the Danish National Museum.

Feveile and Bennike, 2002). The copious number of mounds in Denmark does, however, indicate that the number of people that were buried in mounds should be counted in the hundreds of thousands (Holst, 1999, p. 120). Notwithstanding mound burials comprise a proportion of the EBA population; it was a proportion so considerable that it raises questions as to the validity of the above mentioned interpretation of the mound burial record. Quantitative studies concerning the weight of gold and bronze demonstrate variation within the grave good assemblages, though rather in the form of gradual differences than distinct groups. Gold ornaments are often encountered in graves with heavy bronze objects and special types of artefacts (Randsborg, 1974, p. 45). Similar qualitative differences are noted by Kristiansen, who believes on the basis of the typology of swords, the patterns of re-sharpening and the combination of objects to be able to differentiate between chiefly burials

and graves associated with a warrior aristocracy (Kristiansen, 1984, p. 195). The relation between the graves and the barrow construction has until recently received less attention than the contents of the graves. Recent studies indicate that there are numerous examples of one phase of construction covering several primary burials, and a few examples of phases containing no graves whatsoever. Successive constructions of empty enlargement phases are also regularly encountered (Holst, 1998, 1999). These observations demonstrate the complexity of the activities associated with barrow construction, and emphasize the importance of the agency of a collective. Calculations on the resource and labour investment that went into mound construction clearly underlines that it was an event which by far surpassed the capacity of a single settlement unit, thus involving the contribution of a large collective of agents (Thrane, 1984, p. 152). Moreover, the occurrence of several primary burials

36

K.L. Johansen et al. / Journal of Anthropological Archaeology 23 (2004) 33–55

covered by the same constructional phase of the barrow raises the question if all the deceased even came from the same settlement unit. Only a vague correlation exists between the number of grave goods and the size of the mound. Moreover, the primary burials are no more richly equipped than the secondary burials (Holst, 1999, p. 123). These observations are inconsistent with the interpretation of the barrow as a monument bound to an individual. Finally, the simple inhumation graves without barrows occasionally contain grave good assemblages of conspicuous wealth, although they are as a rule less richly equipped than those encountered in the burial mounds (Aner and Kersten, 1973; Broholm, 1940; Feveile and Bennike, 2002). Settlements of the EBA have only relatively recently come into focus in South Scandinavia. The lack of evidence for such sites led an earlier generation of researchers to suggest that the EBA society was pastoral, the population leading an essentially nomadic existence (Brøndsted, 1958, p. 311). Following the initial discoveries of EBA houses in the 1970s, and the contemporary advent of large scale open-area excavations, the number of recorded settlements has increased dramatically. EBA settlements are today common, and appear to have been more substantial than was hitherto supposed with regular longhouses occasionally in several succeeding phases (Rasmussen and Adamsen, 1993, p. 137). It has also become evident that the house building of the EBA is related both to that of the Corded Ware Culture and the Late Neolithic in South Scandinavia, and to the general Northwest European building tradition of the period (Boas, 1991; Harding, 2000, p. 45; Nielsen, 1993, 1999). EBA houses do not, however, constitute a homogeneous group. A chronological development occurs, in which three-aisled constructions supplant two-aisled ones. Variation otherwise is only partly determined by chronological and regional factors. The sizes of the houses range from 10 m long to impressive buildings with lengths in excess of 50 m. Considerable variation exists with respect to ground-plan, wall construction, the location and number of entrances, the internal division of space, the distribution of internal features such as fireplaces and storage pits, the presence of byre sections, and the extent to which small adjacent economy buildings are observed (Rasmussen, 1999, p. 283). A number of explanations for this diversity have been proposed in the form of varying access to building materials, differences in the number of social units residing in the houses, and the social status of the households themselves. The larger houses, for instance, have often been interpreted as the dwellings of chiefly families (Andersen, 1990, 1995; Ethelberg, 1995, p. 15; Jensen, 1997, p. 7; Nilsson, 1996, p. 152; Rasmussen, 1999, p. 285; Rasmussen and Adamsen, 1993, p. 137). It is, however, more likely that the observed variation is related to economic specialization and inter-site diversifi-

cation of functions and activities. This is supported by a remarkable contrast between the artefact assemblages from the different sites. Some settlements contain almost no flint artefacts or only a modest representation of different types. Others have a large flint material, evidence of extensive flint knapping, and an assemblage dominated by specific types of implements. Bronze casting and amber work also seem confined to only a small minority of sites (Earle et al., 1998, p. 15; Rasmussen, 1995). An understanding of the overall settlement structure remains elusive. Even on sites with large numbers of houses it has been impossible to convincingly prove the existence of two or more contemporary structures (Ethelberg, 2000, p. 244). The material leaves an impression of a settlement of small, dispersed units sometimes perhaps with a tendency to moderate clustering (Rasmussen and Adamsen, 1993, p. 141). Serious doubts about the existence of regular hamlets or villages in the EBA are nevertheless justified. Settlement instead seems to have been labile, consisting of relatively small social units characterized by a high degree of spatial dynamics within a confined resource space (Rasmussen, 1993, p.182). The settlement structure of the EBA thus has many ties to the Corded Ware Culture and in particular the Late Neolithic of South Scandinavia (Laursen et al., 2003; Nielsen, 1993, 1999). Because settlement sites were discovered quite late in the history of research pertaining to the EBA of South Scandinavia, the burial mounds have played a dominant role in the understanding of general settlement patterns (M€ uller, 1904, p. 56). The relationship between settlements and barrows is, however, one of considerable complexity. In recent years there has been a strong emphasis of the relation between the distribution of barrows and the mobility of the individual settlement units within their resource spaces (Rasmussen, 1993, p. 180). A landscape transformed by human action is a prerequisite to barrow construction, since bare tracts of land are not a naturally occurring phenomenon in South Scandinavia. This is underlined by the very frequent occurrence of ard marks underneath the barrows and the evidence of pollen analyses, macrofossil studies, and pedological examinations, which all situate the barrows in extensively exploited pastureland (Andersen, 1988, p. 54; Holst et al., 1998; Prangsgaard et al., 1999, p. 89; Thrane, 1991). Actual settlement material of EBA date is also regularly encountered within and underneath the barrows (Bokelmann, 1977; Boysen and Andersen, 1983; Ethelberg, 1987; Pedersen, 1986). The barrows are thus now regarded as being more or less randomly distributed within the resource area, their density and degree of clustering reflecting the long-term territorial stability and land-use continuity of the settlement unit (Ethelberg, 2000, p. 262; Rasmussen, 1993, p. 174).

K.L. Johansen et al. / Journal of Anthropological Archaeology 23 (2004) 33–55

Looking at the overall distribution of barrows, however, there are distinct patterns left unaccounted for by a mechanism of random clustering within individual settlement areas. Other organizing principles must therefore be sought, particularly for the inland regions of Jutland (Fig. 2). The distribution of barrows here is characterized primarily by linearity—barrow lines. Two major corridors run from the northern parts of Central Jutland and join into a single line in the south, whence it

37

transgresses the present border and continues in Germany. Minor barrow lines branch out from the more marked bands, and in this way extend in a complex and intricate network throughout the whole of the Jutland peninsula. These barrow lines are, to a large extent, composed of EBA burial mounds, but mounds from earlier periods, particularly the Corded Ware Culture, also contribute significantly to their formation. The barrow lines are thus an accumulated long-term

Fig. 2. The distribution of all prehistoric burial mounds in the southern parts of Jutland (www.dkconline.dk).

38

K.L. Johansen et al. / Journal of Anthropological Archaeology 23 (2004) 33–55

structure; explanations of this structure must perforce recognize that factors which transcend the individual settlement unit were at work during its formation. The attempt to explain how agrarian societies characterized by relatively labile settlement units over more than one millennium constructed barrows in an apparently rule-determined distribution pattern raises the question of what fixed or organized this long-term structure. An early attempt was made by Sophus M€ uller, who argued that the linear distribution of barrows could only have taken form if the monuments were placed along roads (1897, p. 297; 1904). M€ uller envisaged that the settlement occurred along roads, and since the barrows were erected near the settlements, the distribution of barrows would in effect reflect prehistoric infrastructure (1904, p. 53). His ideas rest primarily on the morphological characteristics of the barrow lines, but he was able to support his hypothesis by demonstrating a correlation between the barrow lines and the location of sunken roads, recorded historical roads, natural traffic corridors, and the distribution of fords. In spite of the fact that M€ ullerÕs theory is the only explanation that convincingly takes into account the overall distribution of barrows, it has been neglected or rejected in favour of local studies that emphasize the importance of the individual settlement unit and the accumulation of barrows within its resource area (Andersen, 1983). In addition, the research on prehistoric roads has tended to focus on isolated excavations and constructional details; the overall long-term infrastructure, and association with other features within a larger region, has largely been ignored (Jørgensen, 2001; Th€ orn, 2001, 2002). Only recently has the phenomenon of barrow lines been reintroduced to the debate, and in the light of M€ ullerÕs road hypothesis and crosscultural studies, it has been suggested that roads, crossroads, and fords may have been significant places for EBA society with respect to social relations, liminality, and cosmology (Holst et al., 2001; Rudebeck, 2001). The possible relation of the barrow lines to the overall communication systems and the large geographic scale of the structures make them an obvious but hitherto largely unexploited source of information concerning social interaction and organization in the EBA. The barrow lines occupy a cardinal position in the following analysis of EBA society in South Scandinavia.

Approaches to the study of barrow lines The use of barrow lines in the interpretation of EBA social interaction and organization in South Scandinavia rests on theoretical presuppositions concerning the processes which for more than one millennium contributed to the generation of these distinct structures in the distribution of the barrows. It is obvious that these

lines somehow reflect a stable, long-term structure maintained through a series of single events (the barrow building) in a system characterized by relatively dynamic settlement components. The apparent opposition between dynamics and stability is largely parallel to the duality of agency and structure (Sztompka, 1994, p. 29). The debate on agency and structure can be perceived as originating in the opposing viewpoints of Weber and Durkheim. The former described society as labile, constituted by the totality of individual actions; the latter was of the opinion that the structures of society governed the actions of all, making it stable (Ritzer and Gindoff, 1994, p. 5; Sztompka, 1994, p. 29). In modern social theory, particularly in GiddenÕs theory of structuration, this dualism has led to the notion of the duality of agency and structure. Central to his theory is, on the one hand, the ability of the agents to freely act and produce structures; on the other hand, the disposition of the agents through the internalization of existing structures to act in such a way that the existing structures are reproduced (Giddens, 1984). This permits change to occur, but it also explains the existence of long-term structures since the unintended consequence of action is often the reproduction of structure. The tendency for structures to fixate over time through the actions of individuals is also the essence of BourdieuÕs theory of practice, in which practice is determined by habitus—in effect, the internalization of given structures (Bourdieu, 1977). Despite the similarity, there are also significant differences. Giddens has a strong focus on the ability of agents to act freely, whereas BourdieuÕs emphasis is on the constraints of habitus on practice. Thus, it is particularly BourdieuÕs notion of the reproduction of structures which is of interest to the formation of the barrow lines. The dialectics of agency and structure is processual in nature, and embedded in a sequence of actions (Archer, 1985, p. 72). Structures manifested as lasting patterns of social relations between agents exist only through the long-term repetition of actions. Events of interaction are the forum where agency and structure meet, influenced by intended and unintended consequences of prior actions, and altering the conditions for any future actions (Abrams, 1982, p. 192). This has reintroduced a historical dimension to sociological studies (Sztompka, 1994, p. 43), and it also hints the potential of the application of these concepts in archaeological research. In archaeological studies, it is rarely possible to specify the cause of single events, but in the study of diachronic material manifestation of actions in an accumulated record, archaeology has the advantage of being able to address agency in light of long-term structures (Dornan, 2002, p. 313; Pauketat, 2001, p. 86). Furthermore, when social structures through action acquire a material representation their structuring properties are enhanced and the chance of this significant evidence to be embedded the archaeological record likewise.

K.L. Johansen et al. / Journal of Anthropological Archaeology 23 (2004) 33–55

The formation of the barrow line structures can be understood if regarded from a position that involves structuration theory and the diachronic interplay of settlements, barrows, and roads. These three main components intervening in the formation of the barrow lines are doing so first and foremost in a complex dialectic which secondly leads to causal looping. As the structure physically accumulates this tendency is enhanced, on the one hand transforming the structure, and on the other stabilizing some of its fundamental properties. The causal loop can be outlined by describing some of the componentsÕ characteristics, and then considering their reciprocal relations in the formation of the barrow lines. At a hypothetical starting point only few barrows exist. Their role in the cultural landscape is relatively limited and their influence on the location of settlements and roads is moderate. On the other hand there is a strong interdependence between the location of roads and settlements. Settlements are dependent on the interaction facilitated by the roads and the demand of interaction determines the location of the roads. Settlements, and to some extent roads, have a strong, almost determining, influence on the location of the barrows for a number of reasons, such as various ritual or symbolic factors including the relation between the living and the dead. The land opening qualities of settlement activities also form a prerequisite for the cutting of sods for barrow construction. Due to the strong influence of the labile settlements on particularly the barrow construction and to some extent the location of the roads, these two components inherit some of the dynamic properties of the settlements. Consequently, the scene of barrow construction, road maintenance, and settlement activities are initially subject to frequent relocation. Contrary to roads and settlements the barrows are cumulative in nature. When a collective of agents construct a mound, it is an action that through its material and monumental product has far reaching structural consequences. When a mound is erected the landscape is forever transformed, a fact that all succeeding actions must relate to. Over time, the barrows become an increasingly strong structuring component in the landscape. Their influence on the location of the settlements and the roads increases, and it results in a relatively equal interplay between the three components. As the dynamic properties of the settlements are only partially inherited by the roads, the roads are more spatially confined than the settlements. Consequently, it is the course of the roads which over time becomes most clearly expressed in the distribution of the barrows. This may also have led to an emphasis in the EBA of the link between barrows and roads strengthening the interplay between the two components even further as the barrows continued accumulating. In conclusion, the unintended consequence of the barrow construction was a

39

fixation of the infrastructure over time. This dialectics in which barrows accumulate and infrastructure stabilizes can be conceived as a sort of self-organizing system (Allen, 1982). If we progress from the theoretical considerations concerning the processes generating the barrow structures, to the methodical possibilities of the analyses of the archaeological record, the notion of barrow lines as an accumulated spatial manifestation of an EBA network of social interaction makes it natural to take a closer look at social network analysis. At the heart of social network analysis is the significance of relationships existing amongst interacting units: • •





Actors and their actions are viewed as interdependent rather than independent, autonomous units. Relational ties (linkages) between actors are channels for transfer or ‘‘flow’’ of resources (either material or non-material). Network models focussing on individuals view the network structural environment as providing opportunities for or constraints on individual action. Network models conceptualize structures (social, economic, political, and so forth) as lasting patterns of relations among actors. (Wassermann and Faust, 1994, p. 4).

The attributes of the actors (agents), and the character of their actions, are thus to be understood primarily in terms of their structural and locational properties in a network of social interaction. If the barrow line structures are a spatial manifestation of a significant EBA social interaction network, areas central in the network of barrow lines can be expected to display a material culture inventory that contrasts with those encountered in other regions as special opportunities would be presented to the agents in this environment. Central areas may have been loci of intense interaction and events of overall social importance, perhaps creating a patterned material record. In the field of social network analysis the identification of such central or important positions in a network has a long research tradition in connection with the concept of centrality (Wassermann and Faust, 1994, p. 169). Centrality is a property of all nodes in a network, but the size of the measure varies according to network position, and differing perceptions of what it means to be Ôcentral.Õ Hence a number of different approaches to centrality have been presented of which degree centrality and information centrality are of interest to the study at hand. Degree centrality gives a measure of network activity; it takes into account only direct relations, and high values are attributed to nodes with many and/or strong direct relations (Freeman, 1979; Wassermann and Faust, 1994, p. 178). Indirect relations play an important part in information centrality, which is a measure developed to gauge information control. High information

40

K.L. Johansen et al. / Journal of Anthropological Archaeology 23 (2004) 33–55

centrality is ascribed to nodes located on short paths between many other nodes in the network, thus identifying funnels in the flow of information. It takes into account all paths, attributing differing weight according to length (Stephenson and Zelen, 1991; Wassermann and Faust, 1994, p. 169). It is important to distinguish central places from the concept of centrality as introduced above. Central places refer to a specific social and spatial organization, in this context a chiefdom and a hierarchical settlement pattern. Centrality is an analytical technique which is used in the investigation of social interaction networks in general. A central place will invariably posses high centrality, but high centrality may also exist in a system without central places. Thus without making prior assumptions about social organization, the two concepts of centrality, degree and information, can be used to clarify whether the barrow line structures reflect a social interaction network in the EBA. If they do reflect such a network, then one would expect to find material culture patterning related to centrality in the network of barrow lines. On the basis of such correlations, it may also be possible to single out areas of importance to the overall social organization. To examine if the barrow lines

reflect a social interaction network, a region of southwest Jutland especially apposite for analysis has been subject to investigation. An attempt has been made to correlate material culture patterning with areas of the barrow line structures according to expected centrality in an interaction network.

The barrow lines of southwest Jutland The area of investigation The investigation area covers 4000 km2 of southwest Jutland (Fig. 3). The main stationary line of the Weichselian glacial ice-sheets runs north–south through its eastern parts, resulting here in a gently undulating landscape of clayish moraine deposits. In the western parts, the terrain levels out and is characterized by the presence of sandy diluvial plains of Late Glacial date, with occasional morainic hill formations from the Saale glacial period. Several small rivers flow east–west from the terminal moraine through the diluvial plains and into the North Sea. It is of relevance to this study that among these rivers, the Konge a was well-suited for

Fig. 3. The distribution of all prehistoric burial mounds within the area of investigation.

K.L. Johansen et al. / Journal of Anthropological Archaeology 23 (2004) 33–55

waterborne traffic throughout prehistoric times. The extent and configuration of the North Sea coast in this region during the EBA is largely unknown, but was probably quite different from that which it presently obtains. Only in the vicinity of the present day town of Esbjerg was it presumably very similar to the current coast because of the occurrence of morainic formations. In the investigation area, 8181 prehistoric burial mounds are recorded in the Danish SMR. The vast majority of these were constructed during the Corded Ware Culture period, the Late Neolithic, and in particular during the EBA (www.dkconline.dk). The distribution of burial mounds is strongly correlated with diluvial deposits. These low lying but well drained soils often constitute corridors for the barrows through the surrounding landscape of morainic hills and open moors. The south western part of the investigation area is rather devoid of barrows, probably first and foremost due to a high water table, and an ensuing low settlement density and problems of land transport in this area. As a result of low population density until the 19th century and extensive agricultural exploitation, the representativity of burial mounds in the area is quite high in comparison with the fertile soils of eastern Jutland, where intensive agriculture has predominated through historic times (Baudou, 1985, p. 74). The barrow lines in the investigation area thus branch out in a remarkably well-defined network. Already in 1914, the clarity of these structures caught the attention of Sophus M€ uller. He remarked on the marked intersection of JutlandÕs major barrow lines at the Konge a and the area around Esbjerg, where they meet the coast (1914, p. 212). Applying his hypothesis concerning the relation between barrow lines and prehistoric road systems, M€ uller in this way informally identified areas that in a network of social interaction would be characterized as possessing high centrality.

41

Fig. 4. The barrow lines defined on the basis of all prehistoric mounds within the area of investigation, and the distribution of EBA sites used in the study.

structures to the evidence from the EBA. An index of over-representation is calculated as the percentage of sites within a certain distance of the identified barrow lines to the percentage of area covered by that distance. A random distribution provides an index value of 1, but the prehistoric mounds and the EBA sites both exhibit marked over-representation in the vicinity of the barrow lines (Fig. 5). More important, however, is the fact that the two categories display almost identical patterns of over-representation in relation to the barrow lines. The minor divergences can be attributed to the inclusion of hoards and stray finds in the EBA material. In a v2 test, spearheads and palstaves, which typically derive from votive deposits, give rise to a statistically

The barrow lines A definition of the barrow lines is a prerequisite for the study of the relationship between barrow lines and artefact patterning. This is achieved in the following manner: on the basis of the distribution of the total 8181 mounds in the investigation area from prehistory in general, all barrows within a distance of 2 km or less are connected with lines, which produces a graph where all the major barrow lines become immediately evident as continuous bands. The major barrow lines are traced from this graph (Fig. 4). As the focus is on overall patterns, the minor barrow lines appearing from the graph as less dense and discontinuous are not included in the analysis. Since the barrow lines were generated during both the Corded Ware Culture, the Late Neolithic, and the EBA, it is important to assess the relevance of these

Fig. 5. The relation between distance from the defined barrow lines and over-representation of prehistoric mounds and EBA sites, respectively. A random distribution results in an index value of 1.

42

K.L. Johansen et al. / Journal of Anthropological Archaeology 23 (2004) 33–55

significant difference at 5% level between the EBA artefact assemblage beyond and within 1000 m of the barrow lines (Jensen, 1993, p. 155; Levy, 1982; Willroth, 1985). The v2 test in this way indicates that votive deposits were made in a zone some distance from the barrow lines. In conclusion, the network of barrow lines based on all prehistoric mounds is considered a meaningful spatial framework for the analysis of the EBA artefact assemblage in the investigation area. The barrow lines are accumulated structures with many barrows antedating the EBA, but the result testifies to the relevance of earlier mounds as the structures continued to develop in relation to their distribution. The network of barrow lines can now be examined with respect to centrality (Borgatti et al., 2002). The idea that barrow lines represent EBA infrastructure is plausible only if artefact patterning can be related in a sensible way to varying degrees of network centrality. In order to measure centrality, a graph representation of the network of barrow lines is required (Wassermann and Faust, 1994, p. 92). This is achieved by representing barrow lines as relations, whereas intersections and terminations of barrow lines are conceived as nodes. The points where the barrow lines transgress the borders of the investigation area are also represented as nodes, as eventually these barrow lines will terminate or intersect albeit outside the area of study. The density of barrows on the different barrow lines is thought to reflect network flow and interaction activity, and so the relations of the graph are attributed strengths accordingly. In connection with the analysis of barrow line network centrality, three important source critical points must be kept in mind. First, the use of the barrow lines for analysis presupposes that the processes governing barrow destruction and recovery over time have worked relatively uniformly on all parts of the area. This precondition is probably not fulfilled on a detailed level, but on the general level upon which the analyses are

conducted the representativity of the barrow distribution is regarded as quite high for the reasons stated earlier. Second, only major barrow lines have been included in the analysis. The network in this way represents a subset of a principally larger network. However, in the analyses the network of barrow lines is represented as a valued graph with the relations possessing strengths according to the density of barrows on the barrow lines; as the minor stretches of barrows excluded in the study will yield strengths close to zero, regarding them as absent relations should not seriously affect the results. Third, there are some edge effect problems caused by the arbitrary selection of the investigation area. The barrow lines cover large parts of the peninsula of Jutland, and as such the investigation area constitutes only a segment of the total structure. While this may not influence degree centralities too much because only direct relations are considered, it certainly affects information centralities where transitive relations play an important part. Nevertheless, the investigation area covers the major intersection of the Jutlandic barrow lines at the Konge a, and since there are no land-based short cuts that bypass the area, the patterns identified in terms of information centrality will probably only be enhanced if the region of study were to be enlarged. Because of the source critical problems, it is important to use the centrality analyses only as a general indication of overall structures and not as a detailed representation of prehistoric organization. Bearing these qualifications in mind, the analyses nevertheless reveal some interesting patterns (Fig. 6). Degree centrality as reflection of interaction activity draws attention to two general areas: the major intersection at the Konge a with the intersections following the north banks of this river towards the east, and the region where the barrow lines branch out towards the West Coast. Information centrality displays a similar pattern. Again, the region around the main intersection at the Konge a is pronounced, as is the point of departure of the barrow lines towards the West Coast.

Fig. 6. The nodes of the barrow line network graded according to degree and information centrality.

K.L. Johansen et al. / Journal of Anthropological Archaeology 23 (2004) 33–55

The two measures of centrality are, as a rule, correlated. This indicates that areas of high interaction activity correspond to areas where network flow is forced through particular nodes. The analysis hints that if the barrow line network mirrors social interaction, the EBA artefact assemblages are to exhibit characteristics in the Konge a region and the area where the barrow lines approach the West Coast that are different from those in other regions of the investigation area. The main intersection at the Konge a dominates both centrality measures, and an enlargement of the investigation area is likely only to emphasize this pattern more. Navigable freshwater systems and maritime contacts have been ignored in the network analysis, but if they were to be taken into account they might only strengthen the significance of the areas defined by the barrow lines. In particular, the region around the present day town of Esbjerg may have constituted an important point for maritime contact, as might have the Konge a that in addition probably required land traffic to pass through specific fords (Brøndsted, 1958, p. 91). The artefacts The well-preserved structures in the distribution of barrows were of primary concern in the selection of the area of investigation. The existence of a complete catalogue of all EBA artefact assemblages containing metal (Aner and Kersten, 1973) was another significant factor. With minor reclassifications and the addition of recently discovered finds (Rigsantikvarens Arkæologiske Sekretariat, 1984–1999), these have been recorded in a digital database for the purpose of this study. The material amounts to a total of 1838 objects from 851 sites. These sites include 972 graves (or presumed graves), 45 hoards, and 54 stray finds. Other find categories have not been subjected to analyses. The artefacts belong to a relatively well-defined range of functional types within four main categories: weapons, ornaments, tools, and personal accessories, which primarily comprise toilette equipment. Series of subtypes have also been defined mainly for the purpose of chronological analyses, and the chronological succession of artefact types is supported by changing ornamental styles. Graves of Period I are mainly characterised by pressure-flaked flint daggers and arrowheads, slate pendants, amber beads, and flanged axes, daggers and spearheads of bronze. The Period I grave assemblages thus have close ties to the grave furnishings of the Late Neolithic (Lomborg, 1973). Period II and III graves are furnished with a broader range of bronze artefact types, including palstaves, various types of swords and daggers, and a wide variety of ornaments in the form of rings, studs, and pins. Gold rings and glass beads are also regularly encountered in these graves. In Period III razors and tweezers become an increasingly strong

43

component, marking the beginning of a shift in focus from weaponry to toilette equipment, which is characteristic of the Late Bronze Age (Jensen, 2002). In addition to chronological variation, the EBA grave assemblages in South Scandinavia exhibit a striking gender discrepancy (Sørensen, 1997). Male graves are characterised by weaponry, ornaments, tools, and toilette equipment. Female graves are furnished almost exclusively with ornaments, many of which are types never encountered in male graves (Randsborg, 1974). Variation in what has been presumed to be wealth is another feature of the grave assemblages and it has occupied many researchers concerned with the social structure of EBA society. Three types of wealth indicators can be distinguished within the artefact types. Firstly, wealth may be expressed in the form of relatively exclusive artefact types and the use of exotic raw materials. This group includes unique ritual equipment, gold ornaments, and glass beads. Secondly, certain artefact types are highly raw material demanding in absolute measures. This applies to shaft-hole axes with weights of up to 3 kg and to some extent the palstaves. Finally, there are artefact types which are relatively heavier and technically more difficult to produce than functionally equivalent types. The production of metalhilted swords and octagonal-hilted swords requires considerably more raw material and skill than the production of their various organic-hilted counterparts (Kristiansen, 1984). Similar variation is seen within subtypes of awls, daggers, and knives. Double buttons can also be separated into two well-defined size categories, which may reflect relative wealth, though a functional difference cannot be excluded. The analyses The size of the material and the uneven density of finds, render simple distribution maps of little value in detecting spatial patterning of artefact types within the area of investigation. Areas with high density of finds appear as concentrations of a wide range of artefact types, which, however, is misleading as to the relative over- and under-representation of types. Inasmuch that such relative frequencies are of primary interest to this analysis, a different approach is necessary. The statistical method employed is correspondence analysis, which is an explorative multi-variate analysis related to principal component analysis operating on an abundance or presence/absence matrix of n variables and m units (Baxter, 1994, p. 100; Madsen, 1988; Shennan, 1997, p. 308). Correspondence analysis seeks to capture the most important variation in a dataset and present it on the first few principal axes in an orthogonal vector space that is a linear transformation of the original dataset. As the analysis is symmetric, units and variables can be plotted on the same set of principal axes and can be

44

K.L. Johansen et al. / Journal of Anthropological Archaeology 23 (2004) 33–55

interpreted in relation to each other in terms of proximity: units located close together will tend to be similar in relative terms; variables located close to each other will occur together; variables located close to a particular unit will be characteristic of that unit, and vice versa. Furthermore, variables/units located at the same end of a principal axis are positively correlated, whereas units/variables located at opposite ends are negatively correlated. The origo of a principal axis represents the expected (random) distribution along that axis, whereas any divergence from origo testifies to a statistical overor under-representation. The weight and influence of the different units/variables in the analysis are related to their size and divergence from the expected distribution. This is a considerable advantage. It means that even though units/variables of insufficient sample size may be part of the analysis, they will not exert mentionable influence on the overall structure of the analysis. However, it also means that one should aim at roughly comparable sample sizes to avoid domination. In spite of the complex mathematics involved in a correspondence analysis, the plot of units/variables in a simple manner captures the overall structures of a complex dataset. It shows what occurs together, and what does not occur together. Since the principal axes are constituted by the co-variation between variables/units, they can often be interpreted as a new set of synthetic variables reflecting underlying dimensions of variation in the material at hand. Hence, a principal axis may correspond to abstract concepts such as chronology, spatial patterning, gender, and wealth. Correspondence analysis has been applied in a wide variety of archaeological contexts and for many different purposes (Baxter, 1994, p. 133; Madsen ed., 1988). In this study it has initially been used to detect overall spatial patterning of artefact types in the investigation area. The investigation area has been arranged into regions (the units of the analysis), and counts of artefact types within these regions have been made (the variables of the analysis). As no natural geographic division of regions exist, several different arbitrary definitions have been tested. To compensate for the uneven distribution of finds, and the ensuing unevenness in the relative weight made by the different regions in the correspondence analysis, the geographic size of the individual regions varies to the extent that each one of them always contains approximately the same number of artefacts. As for the number of regions, several possibilities have been explored. In analyses with a large number of regions, and consequently fewer artefacts per region, the structures of the correspondence plot reflect the gender and chronological attributes of individual graves. That is, since each region contains only a couple of grave assemblages, a region with two male graves will have a markedly different artefact assemblage than a region with for instance two female graves. Likewise, a region

with an EBA Period I grave will diverge from a region with an EBA Period III grave. Thus, the correspondence analysis does not transcend the pronounced chronological and gender related characteristics of the EBA artefact material, because of insufficient sample size in the regions. This is accomplished in analyses based on a configuration of fewer, but larger, and in terms of artefacts, more inclusive regions. As a result, however, the correspondence analysis loses any evident structure and takes the form of a random scatter instead. This might be interpreted in two ways: either there is no spatial patterning of relative over- and under-representation of artefact types with respect to the classifications made, or the patterning is blurred because the regions crosscut underlying structures in the data. If artefact patterning is related to the barrow lines, it will most likely be disrupted in an analysis based on a division of the investigation area into arbitrary regions. To examine this last postulate, 19 regions have been formed by segmenting a zone with a radius of 2 km around the barrow lines. The size of the regions has been adjusted so that each one contains between 47 and 61 artefacts. Finds outside the 2 km zone have been excluded in the analysis, reducing the material to 1008 objects distributed across 93 artefact types. The correspondence analysis results in a distinct structure with two well-defined clusters on the first principal axis, and if the regions are mapped accordingly, an interesting spatial pattern emerges (Figs. 7 and 8). One cluster consists of the central regions in the immediate vicinity of the barrow lineÕs major intersection at the Konge a (6, 7, 8, 13) and the regions that follow the course of the Konge a to the east (15, 17). The area where the barrow lines join the West Coast (10), and the stretch of barrows running E–W along the small river Holsted  a (11), also belong to this group. The other cluster is composed of regions along the barrow lines in the periphery of the investigation area (1, 2, 3, 4, 5, 9, 12, 14, 16). Region 18 and 19 tend to be associated with the periphery and the central group, respectively, but they are both poorly accounted for in the analysis. The artefact types characterizing these clusters are interesting. All gold artefact types are found in the central regions cluster, i.e., the spirals, the arm rings, the needle, and a number of unspecified rings. Glass beads and bronze artefacts that require metal in considerable quantity for production are also found in the central regions cluster, such as metal-hilted swords, octagonalhilted swords, palstaves, and shaft-hole axes. Down to detail it can be observed that bronze objects characterizing the central group are more raw material demanding relative to functionally equivalent types in the scatter of the peripheral regions. For instance, the contrast is reflected in the difference between metal handle awls and organic handle awls, and in the affiliations of large and small double buttons.

K.L. Johansen et al. / Journal of Anthropological Archaeology 23 (2004) 33–55

45

Fig. 7. The first and second principal axes of the correspondence analysis of the regions around the barrow lines. Artefact types reflecting wealth are emphasised.

Fig. 8. The distribution of regions graded according to their coordinates on the first principal axis of the correspondence analysis in Fig. 7.

46

K.L. Johansen et al. / Journal of Anthropological Archaeology 23 (2004) 33–55

The first principal axis of the correspondence analysis thus reflects wealth in all the three aspects discussed above. It is a dimension of variation that clearly crosscuts gender and chronology. Throughout the EBA, the central areas of the barrow line structures and the areas where the barrow lines have contact with navigable freshwater systems and the coast are characterized by the occurrence of gold, glass, and bronze artefacts demanding a large amount of metal for their production. The gold artefacts are the most significant contributors to the variation observed on the first principal axis, and on a distribution map the gold rings appear almost like beads strung out on the barrow lines with quantitative and qualitative concentrations at main intersections (Fig. 9). The correspondence analysis of the regions is, in its definition of the analytical units, clearly targeted at detecting overall geographical structures in relation to the barrow lines. It results in distinct patterns and indicates how the material record and which parts of it were influenced by the barrow lines. However, it uses collections of artefacts within arbitrary regions around the barrow lines; these can be somewhat difficult to relate to individual agents or a specific group of agents in EBA society. It is therefore important to attempt to identify the specific graves that contribute to the variation and locate their exact position in relation to the barrow lines.

This implies a change of analytical units from arbitrary regions to individual grave assemblages. The material in this analysis amounts to 336 graves, containing a total of 1086 artefacts distributed across 80 different types. Methodologically the correspondence analysis remains a valid approach. In the initial analysis, the graves and artefacts form two gender-based clusters on the first principal axis, consisting of male and female graves/artefacts, respectively (Fig. 10). The clusters are relatively well defined, although there is a grey area with shared artefact types and graves that are not easily attributable to either gender. Based on the analysis, around 60 female burials can be identified and they are distinguished by the presence of different categories of bronze rings and bands (neck, ankle, arm, and finger/ear), studs, beads (amber, glass, and bronze spiral), collars, belt plates, and metal-hilted daggers. In contrast to their female counterparts, male graves are dispersed along the second principal axis. Artefact types such as slate pendants, flint daggers, amber beads, flanged axes, spearheads, and ceramics, exhibit relatively high scores on this axis. As these objects are typical of the earliest phase of the EBA (Period I), the second principal axis is to some extent related to chronology. Only about a dozen graves can be dated to this early stage; the rest of the male burials belong to Period II and III.

Fig. 9. The distribution of different types of gold artefacts within the area of investigation.

K.L. Johansen et al. / Journal of Anthropological Archaeology 23 (2004) 33–55

47

Fig. 10. The first and second principal axes of the initial correspondence analysis of the grave assemblages.

Gender and chronology are not of primary relevance to this study, and before other dimensions of variation can be uncovered the material has to be subdivided into female, early male, and late male grave assemblages. As the sample sizes of female and early male graves in the investigation area are too limited to permit further study, the analysis progresses exclusively with male grave assemblages of Period II and III. They form an oblong continuous scatter along the first principal axis, with the majority located towards the high end (Fig. 11). Characteristic of the assemblages with low scores are different types of gold spirals, metal-hilted swords, octagonalhilted swords, palstaves, and to some degree, the arm rings of gold. When the different artefact types are examined with respect to their location on the first principal axis, many features from the analysis of the regions defined by the barrow lines are recognized. It is first and foremost the gold spirals which determine the structure

of the correspondence analysis, and in multi-type assemblages, the gold spirals are almost exclusively found in combination with the aforementioned conspicuous metal demanding bronze weapons. Even through the structure clearly relates to a concept of wealth as regards the types of artefacts, it is important to stress that it is a qualitative wealth rather than a quantitative wealth. The individual burials constituting the cluster of ÔwealthÕ do not necessarily contain many artefacts, but are dominated by exclusive types. The graves with the lowest scores on the first principal axis are thus almost only characterized by different types of gold spirals. The distribution of the graves, graded by their coordinates on the first principal axis, reaffirms the relation between gold spirals and location relative to barrow line network centrality as these grave assemblages are concentrated around the key nodes of the network (Fig. 12). In contrast, the graves without gold at the other end of

48

K.L. Johansen et al. / Journal of Anthropological Archaeology 23 (2004) 33–55

Fig. 11. The first and second principal axes of the correspondence analysis of the male grave assemblages from Period II and III. Artefact types reflecting wealth are emphasised.

Fig. 12. The distribution of male grave assemblages from Period II and III graded according to their coordinates on the first principal axis of the correspondence analysis in Fig. 11.

K.L. Johansen et al. / Journal of Anthropological Archaeology 23 (2004) 33–55

the first principal axis are scattered throughout the landscape. The connection to network centrality is even more pronounced in the distribution of the graves containing both gold artefacts and conspicuous bronze

Fig. 13. The distribution of male grave assemblages from Period II and III containing both gold ornaments and metal demanding bronze weapons such as metal-hilted swords, octagonal-hilted swords, and paalstaves. The numbers refer to the grave assemblages in Fig. 14.

49

weapons such as metal-hilted swords, octagonal-hilted swords, and palstaves. They are almost entirely found within the two areas with overt centrality, in the sense of both degree and information (Figs. 13 and 14). These graves correspond to the traditional concept of wealth expressed in the literature concerning the South Scandinavia EBA, and are often associated with the concept of chiefs (Randsborg, 1974, p. 45; Kristiansen, 1984, p. 195). It is, however, important to stress that these burials are only part of a larger pattern of co-variation. The group of burials forming the cluster of ÔwealthÕ and the associated spatial patterns is much larger and more diverse. In conclusion, the results of the analytical process can be summarized in the following points corresponding to the separate analytical steps: • The identified barrow lines are highly relevant to the EBA find material in general (Fig. 5). • There are distinct structures of centrality in the network as it is preserved in the archaeological record with two areas of particularly high centrality (Fig. 6). • A distinct regional variation is associated with wealth in the artefact material in proximity of the barrow lines (Fig. 7).

Fig. 14. Example of two of the male grave assemblages from period II and III containing both gold ornaments and metal demanding bronze weapons. 1: Period II grave from the Esbjerg area with paalstave, octagonal-hilted sword, and large gold spiral (Aner and Kersten, 1973, no. 4069). 2: Period III grave from the Konge a area with metal-hilted sword and two small gold spirals (Aner and Kersten, 1973, no. 4011). Scale 1:3. Reproduced with the permission of the Danish National Museum.

50

K.L. Johansen et al. / Journal of Anthropological Archaeology 23 (2004) 33–55

• This variation correlates with network centrality (Figs. 8 and 9). • The individual graves responsible for the variation can be identified in the analyses of the male grave assemblages of Period II and III of the EBA (Figs. 11 and 14). • The distribution of these graves supports the connection to the structures of network centrality (Figs. 12 and 13).

Discussion The case study establishes that the barrow lines spatially reflect an EBA network of social interaction, evidenced by the recurrent correlation between barrow line network centrality and artefact patterning. Even without arbitrary analytical units like the regions around the barrow lines, a dimension of variation associated with wealth can be isolated. Second to gender and chronology, this dimension of variation is the most important feature of the material. The analyses strongly suggest that the varying investment on grave good furnishings was intimately bound to location in relation to the network of barrow lines. The manifestations of wealth embodied in the distribution of gold, glass, and raw material demanding bronze artefacts are not exclusively confined to areas of high network centrality, but rather observed to be significantly over-represented here. In accordance with the long-term structuration of the barrow lines, the spatial expressions of wealth crosscut chronological variation, bearing witness to the stability of the interaction network and associated centralities throughout most of the EBA. The exotic materials associated with the expression of wealth are dependent upon the existence of a pan-European network of exchange. At the culmination of the EBA, South Scandinavia is indirectly influenced by the Tumuli complex of south and central Germany and particularly the L€ uneburg Group is of significance (Jensen, 1982, p. 166; Kristiansen, 1987, p. 36). The Elbe and Weser rivers seem to have served as important transport corridors in this exchange network, which draws attention to the waterways; the areas highlighted in this analysis are therefore even further emphasized (Fig. 15). The region around present day Esbjerg may have been an important point of maritime contact, and the Konge a has constituted a navigable freshwater system in prehistoric times (Brøndsted, 1958, p. 91). The presumed western origin of EBA gold further accentuates this pattern (Hartmann, 1982; Ottenjann, 1969, p. 77; Thrane, 1985, p. 206), and the accumulation of wealth at the major intersection of barrow lines at the Konge a, and the area where the barrow lines meet the West Coast, may partially be interpreted as a filter-effect associated with a down-the-line exchange through the

Fig. 15. Interpretative model of the overall structure of the interaction network represented by the barrow lines along with possible maritime routes.

regionÕs interaction network. However, the material expression of centrality presumably reflects social interaction of a far greater complexity, embodying exchange, the establishment of marital relations, political alliances, the distribution of specialized production, gift exchange, rituals, competition, and so forth (Cherry and Renfrew, 1986; Renfrew, 1986; Renfrew and Bahn, 1996, p. 364). These elements form the basis of the concept of peer polity interaction, and not only do the areas of high centrality testify to the existence of a social interaction network, they form vital nodes in the overall social organization and facilitate a re-evaluation of conceptual models of EBA society. Social organization in the EBA has been a subject of intense controversy and debate in South Scandinavia for more than a century. Some researchers have argued that the EBA society was unstratified (Broholm, 1944, p. 257), whereas others postulated that a bipartite social organization was established through the invasion of the region by a ruling class of Corded Ware Culture origin (Brøndsted, 1958, p. 10; M€ uller, 1897, p. 401). However, with the advent of processualism and the repudiation of models invoking immigration to account for cultural change, EBA society came to be categorized as a chiefdom in the sense of the term used by neo-evolutionists and structural Marxists. A model incorporating a social hierarchy with a chief maintaining functions of religious, political, and economic nature, supported by a warrior aristocracy exploiting a population of commoners, became widely accepted. The power of the chiefs was thought to rely principally on the control of bronze acquired through a pan-European system of exchange involving prestige-goods and inter-chiefdom alliances, whereas the local economy is described as re-distribution of subsistence goods along with chiefly control of specialized production (Jensen, 1982, p. 169, 2002, p. 220;

K.L. Johansen et al. / Journal of Anthropological Archaeology 23 (2004) 33–55

Kristiansen, 1982, 1984, 1987, 1991, 1998b; Randsborg, 1974, p. 60). Although slight differences between the models of EBA chiefdom organization in South Scandinavia do exist (see for instance Vandkilde, 1996, p. 275), the above does reflect a general consensus that has now lasted for three decades. As mentioned in the introduction, the question is if this widely accepted view remains the most adequate explanatory framework in the light of both new archaeological evidence collected since the introduction of the chiefdom model to the study of the EBA in South Scandinavia, and the structures revealed in the present study. If we consider the postulate that barrows represent the monuments of an elite, there are important observations to bear in mind: first, a significant segment of the EBA population was accorded burial in the barrows; second, the contrast in grave furnishings between mound burials and simple inhumation graves without mounds is vague and inconsequential; third, there is no particular correlation between the size of the mound and the lavishness of the grave assemblage; fourth, primary burials are not differentiated from secondary burials with respect to their grave furnishings; fifth, the quantitative and qualitative differences in grave assemblages are gradual; and finally, many observations indicate that the barrow is not a monument unambiguously bound to an individual. This is in contrast to the burial customs of the Late Bronze Age, where the decline of the tumuli tradition has left burials in monumental barrows extremely rare and a privilege confined to an exclusive segment of the population. Burials in monumental mounds are distinguished from the prevailing burial custom of the period, and the grave good assemblages associated with large mounds are conspicuous. A marked correlation between mound size and the expenditure of wealth on grave furnishings is prevalent, and the association with the individual interred is clear (Menke, 1972; Thrane, 1984; W€ ustemann, 1974). The existence of a regular settlement hierarchy in the EBA can also be questioned, and so far no central places have been identified. Sites maintaining central social functions may be suggested by the major long houses occasionally observed in association with bronze hoards and large storage pits. Evidence of specialized/differentiated site functions might also be taken into account if arguing for a society based on a re-distributive economy (Earle et al., 1998, p. 15; Ethelberg, 1995, p. 12; Nilsson, 1996, p. 150; Rasmussen, 1995, p. 90). These circumstances may nevertheless equally well be accounted for within an alternative explanatory framework, and if this position is assumed, a glaring contrast to the chiefdom society of the Late Bronze Age becomes apparent. The central place of Kirkebjerget of the Late Bronze Age is markedly different from contemporaneous sites inasmuch as it contained large areas of massive cultural

51

deposits and an exclusive artefact assemblage (Berglund, 1982). Among the finds that set this site apart are metal objects, evidence of lur casting and fragments of painted wall plastering (Thrane, 1995). Kirkebjerget is associated with a marked regional accumulation of wealth, expressed in votive deposits, monumental mounds, and rich grave assemblages as known from the famous Lusehøj barrow (Thrane, 1984). Parallels to this complex exist elsewhere in South Scandinavia and northern Germany, forming a discrete chronological horizon of related phenomena (Jensen, 1983; Thrane 1980, 1984, p. 164, 1994.) Evidence associated with chiefdoms, according to Renfrew and Bahn, include a spatial constellation of central place, monumental burial, and regional accumulation of wealth (1996, p. 195). These elements are seemingly present in the Late Bronze Age, and an ascription of chiefdom organization to the society during this period seems justified. A similarly unequivocal indication for the existence of a chiefdom cannot, however, be said to be present in the evidence from the EBA. The society is certainly complex, distinguished by extensive exchange networks, specialization, and some degree of social stratification. Many components ordinarily associated with a chiefdom are indeed present, but there is an absence of the organizing institutional constellation of chief and central place. This poses the crucial question of how a complex society composed of dispersed settlement units was organized, and maintained stabile throughout several centuries, and what initiated the evolutionary trajectory that culminated in the classic chiefdoms of the late Bronze Age? Of paramount importance is the virtual explosion of mound building in the EBA. In this era of Scandinavian prehistory, large numbers of people gathered for the construction of monuments which involved the lavishment of a tremendous amount of labour. It is in these events that the answer to the above question of the organizing principle of EBA society lies. The labour investment involved in barrow construction clearly demonstrates the participation of numerous settlement units; this is furthermore suggested by the periodic occurrence of burials during a single phase of construction that exceeds the realistic contribution of an isolated social unit. Focussing on the collective aspects of mound construction, this event constitutes the pivot of society: an integrative forum of social organization that in opposition to the spatially constrained central place constantly alters its location spatially and temporally according to fluctuating demands of social interaction. The foundation of this alternative explanatory framework of EBA society is a model based on peer polity interaction amongst fairly small, labile social units. This society was probably relatively egalitarian insofar as the social units were peers in a coherent system of differentiated and specialized production.

52

K.L. Johansen et al. / Journal of Anthropological Archaeology 23 (2004) 33–55

Specialization and differentiation are perceived as catalysts of social interaction in a system in which stability, alliances, and exchange is maintained through interdependencies amongst social units with different production. Thus specialization is primarily a non-discursive cultural construct, which creates the channels in which resources at all levels of society flow. Incorporated in this system is the distribution of bronze and other products of non-local origin (Rasmussen, 1995, p. 105). The organization of this system might best be described as an affiliation network: a two-mode network consisting of a set of agents and a set of events in which the agents are integrated through the participation in social events and these events are affiliated by sharing participating agents (Wassermann and Faust, 1994, p. 291). In the EBA, social organization was facilitated through the events of mound building which constituted the ritualized forum for social, political, and economic activities embodied in the concept of peer polity interaction. The long tradition established in sociology which concerns itself with the significance of social events in creating ties between agents has not been pursued with much alacrity by archaeologists, where the focus has been on places rather than events (Wassermann and Faust, 1994, p. 292). The social organization of the EBA, however, clearly demonstrates the importance of the complex interplay between events and places. The event of mound building took place in space and due to the monumental product of these actions, socially significant places emerged (Hirsch, 1995). In this way, the distribution of mounds constitutes a material record which was structured by, and at the same time structured, EBA social interaction. Both mound building and road building rely on, and facilitate, social interaction; this complex interplay led to the formation of the network of barrow lines—in effect, spatial patterns of EBA social organization. Areas of high centrality were of key importance in this network, as they represented focal points of interaction and social organization. The mound building events were intensified here, and associated expressions of wealth testify to the intensity of interaction giving rise to the social importance of these places. The long-term structuration of the network led to a fixation of areas of high centrality, initiating the development of mound groups at major intersections of barrow lines. The forum of mound building as events in space therefore became mound groups where events took place. The mound groups which developed at places of high network centrality stand on the threshold of spatial institutionalization of social organization indirectly forming a prelude to central places. As the correlation between network centrality and variation in wealth testifies to the rise of socially significant places, it also demonstrates a long-term development of agency. Dead agents do not act: agency is to

be ascribed to agents governing the collectives that constructed the mounds and organized associated activities. The burial expenditure, however, must to some extent reflect an interpretation by the living of the terminated agencies of those interred within the barrows. The network of barrow lines is a physical expression of a social structure of interaction which enabled and constrained agency according to network position, the rich grave furnishings bearing witness to expanded possibilities of agents in areas of high centrality. The rich grave assemblages thus reflect a situated agency; it can only be comprehended in the context of the long-term development of the barrow line structures. Expanded possibilities of agency emerge as an unintended consequence of the repetition of events of barrow construction. This transformation of agency is separated from the intentionality of agents. However originally unintended, the constraints and possibilities of agency gradually embedded in the structure as it accumulated probably did at some point partially enter the realm of the discursive knowledge of agents, making the structure an arena where intentional choice of agent participation was subordinated to social, political, and economic strategies. This could have initiated a development from a partially egalitarian system with collectively invested leaders to an organization in which power is seized and formalized by knowledgeable agents exploiting the possibilities of the network structure and controlling key nodes of social interaction. EBA mound building encompasses in a ritual form many of the functions later held by the chiefly controlled central places of the Late Bronze Age, but the system is decentralized and self-organized as opposed to centralized and monopolized. An evolutionary trajectory can be outlined on the basis of the relationship between settlement and mound building. The EBA is characterized by small, dispersed settlement units and an immense but relatively non-exclusive tumulus tradition. In this respect the EBA clearly builds on structures already evident during the Corded Ware Culture and, in particular, the Late Neolithic. However, in comparison to the EBA the mound building during these earlier periods was less monumental, and although it presumably played a role as a forum of social integration, the significance of these events was of a more local character, and the interaction networks less well established. Further, the mound groups which characterise places of high centrality in the social interaction network as reflected in the barrow lines seem to be phenomena of the EBA. During the Late Bronze Age the emergence of central places corresponds with a marked decline of mound building. Monumental mounds are now confined to a very exclusive segment of the population and the developmental sequence thus bears witness to the gradual decline and fall of one social organization and the emergence and rise of a new one. To organize society

K.L. Johansen et al. / Journal of Anthropological Archaeology 23 (2004) 33–55

monuments were constructed; however, the construction of monuments demands organization, the unintended consequence of which eventually led to a formalized leader and a spatial institutionalization of social organization—the chief and the central place. Acknowledgments The authors would like to thank Niels Nørkjær Johannsen, Torsten Madsen, Kristian L.R. Pedersen, Marianne Rasmussen, and Henrik Thrane for comments on earlier drafts of this article. The authors also wish to sincerely thank the Research Foundation of Aarhus University for providing funding. The research of this article is part of a larger research project, ÔThe oak-log coffin barrows of the Bronze Age,Õ funded by the Danish Research Council for the Humanities.

References cited Abrams, P., 1982. Historical Sociology. Cornell University Press, Ithaca. Allen, P.M., 1982. The genesis of structure in social systems: the paradigm of self-organization. In: Renfrew, C., Rowlands, M.J., Segraves, B.A. (Eds.), Theory and Explanation in Archaeology. The Southampton Conference. Academic Press, New York, pp. 347–374. Andersen, S.Th., 1988. Pollenanalyser fra gravhøje. Arkæologiske Udgravninger i Danmark 1987, 54–60. Andersen, S.Th., 1990. Pollen spectra from the bronze age barrow at Egshvile, Thy, Denmark. Journal of Danish Archaeology 9, 153–156. Andersen, S.Th., 1995. History of vegetation and agriculture at Hassing Huse Mose, Thy, Northwest Denmark. Journal of Danish Archaeology 11, 57–79. Andersen, S.W., 1983. Hærvejen og bebyggelsen i Nordslesvig i den ældre bronzealder. Nordslesvigske Museer 10, 53–62. Aner, E., Kersten, K., 1973. Die Funde der €alteren Bronzezeit des nordischen Kreises in D€anemark, Schleswig-Holstein und Niedersachsen. Karl Wachholz Verlag, Neum€ unster. Archer, M.S., 1985. Structuration versus morphogenesis. In: Eisenstadt, S.N., Helle, H.J. (Eds.), Macro-Sociological Theory, vol. 2. Sage, London, pp. 55–88. Baudou, E., 1985. Archaeological source criticism and the history of modern cultivation in Denmark. In: Kristiansen, K. (Ed.), Archaeological Formation Processes. The Representativity of archaeological remains from Danish Prehistory. Nationalmuseet, København, pp. 63–80. Baxter, M.J., 1994. Exploratory Multivariate Analysis in Archaeology. Edinburg University Press, Edinburg. Berglund, J., 1982. Kirkebjerget—a late bronze age settlement at Voldtofte, south–west Funen. Journal of Danish Archaeology 1, 51–63. Boas, N.A., 1991. Late neolithic and bronze age settlements at Hemmed Church and Hemmed Plantation, East Jutland. Journal of Danish Archaeology 10, 119–135.

53

Bokelmann, K., 1977. Ein bronzezeitlicher Hausgrundriss bei Handewitt, Kreis Schleswig-Flensburg. Offa 34, 84–87. Borgatti, S.P., Everett, M.G., Freeman, L.C., 2002. Ucinet for Windows: Software for Social Network Analysis. Analytic Technologies, Harvard. Bourdieu, P., 1977. Outline of a Theory of Practice. Cambridge University Press, Cambridge. Boysen, A., Andersen, S.W., 1983. Trappendal. Barrow and house from the early bronze age. Journal of Danish Archaeology 2, 118–126. Boye, V., 1896. Fund af Egekister fra Bronzealderen i Danmark. Et Monografisk Bidrag til belysning af Bronzealderens Kultur. Andr. Fred. Høst & Søns Forlag, Kjøbenhavn. Broholm, H.C., 1940. Tre Kvindegrave fra den ældre Bronzealder fra Gjedsted Sogn. Aarbøger for Nordisk Oldkyndighed og Historie 1940, 117–140. Broholm, H.C., 1944. Danmarks Bronzealder, vol. 2. Nordisk forlag, København. Brøndsted, J., 1958. Danmarks Oldtid. Bronzealderen. 2. udgave. Gyldendal, København. Cherry, J.F., Renfrew, C., 1986. Epilogue and prospect. In: Renfrew, C., Cherry, J.F. (Eds.), Peer Polity Interaction and Socio-Political Change. Cambridge University Press, Cambridge, pp. 1–18. Christaller, W., 1968. Die Zentralen Orte in S€ uddeutschland. € konomisch-geographische Untersuchung u €ber die Eine o Gesetzm€assigkeit der Verbindung und Entwicklung der Siedlungen mit st€adtischen Funktionen. Wissenschaftliche Buchgesellschaft, Darmstadt. Collis, J., 1986. Central place theory is dead: long live the central place. In: Grant, E. (Ed.), Central Places, Archaeology and History. Department of Archaeology and Prehistory, University of Sheffield, Sheffield, pp. 37–40. Dornan, J.L., 2002. Agency and archaeology: past, present and future directions. Journal of Archaeological Method and Theory 9 (4), 303–329. Earle, T., 1991. The evolution of chiefdoms. In: Earle, T. (Ed.), Chiefdoms: Power, Economy, Ideology. Cambridge University Press, Cambridge, pp. 1–15. Earle, T., Bech, J.-H., Kristiansen, K., Aperlo, P., Kelertas, K., Steinberg, J., 1998. The political economy of late neolithic and early bronze age society: the Thy archaeological project. Norwegian Archaeological Review 31 (1), 1–28. Ethelberg, P., 1987. Early bronze age houses at Højg ard, Southern Jutland. Journal of Danish Archaeology 5, 152– 167. Ethelberg, P., 1995. Den fjerde kæmpehal. Skalk 1995 (3), 11–15. Ethelberg, P., 2000. Bronzealderen. In: Ethelberg, P., Jørgensen, E., Robinson, D. (Eds.), Det Sønderjyske Landbrugs Historie. Sten- og Bronzealder. Haderslev Museum & Historisk samfund for Sønderjylland, Haderslev. Feveile, C., Bennike, P., 2002. Lustrupholm. Et brandgravsfelt fra ældre bronzealder under flad mark. Kuml 2002, 109–139. Freeman, L.C., 1979. Centrality in social networks: conceptual clarification. Social Networks 1, 215–239. Giddens, A., 1984. The Constitution of Society. Polity Press, Cambridge. Hachmann, R., 1958. S€ uddeutsche H€ ugelgr€aber- und Urnenfelderkulturen und €altere Bronzezeit im westlichen Ostseegebiet. Offa 15, 43–76.

54

K.L. Johansen et al. / Journal of Anthropological Archaeology 23 (2004) 33–55

Haggett, P., Cliff, A.D., Allan, F., 1977. Locational Analysis in Human Geography. Edward Arnold, London. Hansen, M., Rostholm, H., 1993. Graves from the single grave culture and the late Neolithic. In: Hvass, S., Storgaard, B. (Eds.), Digging into the Past. 25 Years of Archaeology in  Denmark. Jysk arkæologisk selskab, Arhus, pp. 116–121. Harding, A.F., 2000. European Societies in the Bronze Age. Cambridge University Press, Cambridge. Hartmann, A., 1982. Pr€ahistorische Goldfunde aus Europa II. Spektralanalytische Untersuchungen und deren Auswertung. Studien zu den Anf€angen der Metallurgie, bd. 5. Gbr. Mann. Verlag, Berlin. Hirsch, E., 1995. Introduction. Landscape: between place and space. In: Hirsch, E., OÕHanlon, M. (Eds.), The Anthropology of Landscape, Oxford. Hodder, I., Orton, C., 1976. Spatial Analysis in Archaeology. Cambridge University Press, Cambridge. Holst, M.K., 1998. Interdisciplinary studies on bronze age  burial mounds. In: Anna-Carin, A., Asa, G., Ola, W.J., H akan, K., Magnus V.R. (Eds.), The Kaleidoscopic Past. Gotarc Serie C. Arkeologiska Skrifter No. 16, G€ oteborg, pp. 3–12. Holst, M.K., 1999. Bronzealderens Gravhøje. Unpublished dissertation, University of Aarhus. Holst, M., Breuning-Madsen, H., Olsson, M., 1998. Soil forming processes in and below a bronze age burial mound at Lejrskov, Southern Jutland. Geografisk Tidsskrift. Danish Journal of Geography 98, 46–55. Holst, M.K., Breuning-Madsen, H., Rasmussen, M., 2001. The South Scandinavian barrows with well-preserved oak-log coffins. Antiquity 75 (287), 126–136. Jensen, J., 1982. The Prehistory of Denmark. Methuen & Co, London. Jensen, J., 1983. Et rigdomscenter fra yngre bronzealder p a  Sjælland. Arbøger for Nordisk Oldkyndighed og Historie 1981, 48–96. Jensen, J., 1993. Metal deposits. In: Hvass, S., Storgaard, B. (Eds.), Digging into the Past. 25 Years of Archaeology in  Denmark. Jysk aræologisk selskab, Arhus, pp. 152–158. Jensen, J., 2002. Danmarks Oldtid. Bronzealder 2000-500 f. Kr. Gyldendal, København. Jensen, S., 1997. Bronzealder set fra luften—et usædvanligt hus fra ældre bronzealder. By, marsk og geest 1997 (9), 5–8. Jørgensen, M.S., 2001. Forhistoriske veje. Nogle problemer og perspektiver med udgangspunkt i dansk materiale. In: Lars L. (Ed.), Kommunikation i tid och rum, Report Series No. 82, Institute of Archaeology, University of Lund, Lund, pp. 3–14. Kersten, K., 1935. Zur €alteren nordischen Bronzezeit. Forschung Vor- und Fr€ uhgeschichtlicher Altert€ umer Kiel III, Neum€ unster. Koch, E., 1998. Neolithic Bog Pots from Zealand, Møn, Lolland and Falster. Nordiske Fortidsminder, Serie B, vol. 16. Det kongelige nordiske Oldskriftselskab, København. Kristiansen, K., 1982. The formation of tribal systems in later European prehistory: Northern Europe, 4000–5000 B.C. In: Renfrew, C., Rowlands, M.J., Segraves, B.A. (Eds.), Theory and Explanation in Archaeology. The Southampton Conference. Academic Press, New York, pp. 241–280. Kristiansen, K., 1984. Krieger und H€auptlinge in der Bronzezeit D€ anemarks. Ein Beitrag zur Geschichte des Bronzezeitli-

chen Schwertes. Jahrbuch des R€ omisch-Germanisches Zentralmuseums 31, 187–208. Kristiansen, K., 1987. From stone to bronze—the evolution of social complexity in Northern Europe, 2300–1200 BC. In: Brunfiel, E.M., Earle, T.K. (Eds.), Specialization, Exchange and Complex Societies. Cambridge University Press, Cambridge, pp. 30–51. Kristiansen, K., 1991. Chiefdoms, states, and systems of social evolution. In: Earle, T. (Ed.), Chiefdoms: Power, Economy, Ideology. Cambridge University Press, Cambridge, pp. 16– 43. Kristiansen, K., 1998a. Europe Before History. Cambridge University Press, Cambridge. Kristiansen, K., 1998b. Construction of a bronze age landscape. Cosmology, economy and social organisation in Thy, Northwestern Jutland. In: H€a nsel, B. (Ed.), Mensch und Umwelt in der Bronzezeit Europas. Oetker-Boges Verlag, Kiel, pp. 281–291. Laursen, S.T., Johansen, K.L., Holst, M.K., Rasmussen, M., 2003. Høje, landskab og bosættelse. Rekognosceringer ved Tobøl-Plougstrup-højgruppen. Kuml 2003, 107–127. Levy, J.E., 1982. Social and Religious Organization in Bronze Age Denmark. An Analysis of Ritual Hoard Finds. BAR International Series 124, Oxford. € ber Lomborg, E., 1973. Die Flintdolche D€anemarks. Studien u Chronologie und Kulturbeziehungen des s€ udskandinavischen Sp€atneolithikums. Nordiske Fortidsminder, Serie B— in quarto, bind 1. Det kongelige nordiske Oldskriftselskab, København. Madsen, T., 1988. Multivariate statistics and archaeology. In: Madsen, T. (Ed.), Multivariate Archaeology. Nummerical Approaches in Scandinavian Archaeology. Jutland Archae ological Society Publications XXXI, Arhus, pp. 7–27. Menke, M., 1972. Die j€ ungere Bronzezeit in Holstein. Karl Wachholz Verlag, Neum€ unster. M€ uller, S., 1897. Vor Oldtid. København. M€ uller, S., 1904. Vei og Bygd i Sten og Bronzealderen. Aarbøger for Nordisk Oldkyndighed og Historie 1904, 1–64. M€ uller, S., 1914. Sønderjyllands Bronzealder. Aarbøger for Nordisk Oldkyndighed og Historie 1914, 195–348. Montelius, O., 1885 Om tidsbest€amning inom brons aldern med s€arskild h€ansyn till Skandinavien. Kongl. Vitterhets Historie och Antiqvitets Akademien Handlingar 30, ny f€ oljd, 10. Stockholm. Montelius, O., 1900. Die Chronologie der €altesten Bronzezeit in Norddeuthschland und Skandinavien. Braunschweig. Nielsen, P.O., 1993. The settlement. In: Hvass, S., Storgaard, B. (Eds.), Digging into the Past. 25 Years of Archaeology in  Denmark. Jysk aræologisk selskab, Arhus, pp. 92–95. Nielsen, P.O., 1999. Limensg ard and Grødbyg ard. Settlements with house remains from the early, middle and late neolithic on Bornholm. In: Fabech, C., Ringtved, J. (Eds.), Settle ment and Landscape. Proceedings of a Conference in Arhus, Denmark, May 4–7, 1998. Jutland Archaeological Society, Aarhus, pp. 149–166. Nilsson, T., 1996. Store Tyrrestrup. En vendsysselsk storg ard med bronzedepot fra ældre bronzealder. Kuml 1993–94, 147–154. Ottenjann, H., 1969. Die nordischen Vollgriffschwerter der €alteren und mittleren Bronzezeit, R€ omisch-Germanisches Forschungen 30. De Gruyter, Berlin.

K.L. Johansen et al. / Journal of Anthropological Archaeology 23 (2004) 33–55 Pauketat, T.R., 2001. Practice and history in archaeology. Anthropological Theory 1 (1), 73–98. Pedersen, J.-A., 1986. A new early bronze age house-site under a barrow at Hyllerup, Western Zealand. Journal of Danish Archaeology 5, 168–176. Prangsgaard, K., Andersen, S.Th., Breuning-Madsen, H., Holst, M., Malmros, C., Robinson, D., 1999. Gravhøje ved Lejrskov. Undersøgelse af fem høje. Kuml 1999, 53–97. Randsborg, K., 1968. Von Periode II zu III. Chronologische €ber die € Studien u altere Bronzezeit S€ udskandinaviens und Norddeutschlands. Acta Archaeologica, 39. Randsborg, K., 1974. Social stratification in early bronze age Denmark: a study in the regulation of cultural systems. Pr€ aehistorische Zeitschrift 49, 38–61. Randsborg, K., 1993. Oak-log coffins and the chronology of the bronze age. In: Hvass, S., Storgaard, B. (Eds.), Digging into the Past. 25 Years of Archaeology in Denmark. Jysk  aræologisk selskab, Arhus, pp. 164–165. Rasmussen, M., 1993. Gravhøje og bopladser. En foreløbig undersøgelse af lokalisering og sammenhæng. In: Larsson, L. (Ed.), Brons alderens gravh€ oger, Report Series No. 48, Institute of Archaeology, University of Lund, Lund, pp. 171–185. Rasmussen, M., 1995. Settlement structure and economic variation in the early bronze age. Journal of Danish Archaeology 11, 77–107. Rasmussen, M., 1999. Livestock without bones. The long-house as contributer to the interpretation of livestock management in the Southern Scandinavian early bronze age. In: Fabech, C., Ringtved, J. (Eds.), Settlement and Landscape. Pro ceedings of a conference in Arhus, Denmark, May 4–7, 1998. Jutland Archaeological Society, Aarhus, pp. 281–290. Rasmussen, M., Adamsen, C., 1993. Settlements. In: Hvass, S., Storgaard, B. (Eds.), Digging into the Past. 25 Years of  Archaeology in Denmark. Jysk arkæologisk selskab, Arhus, pp. 136–141. Renfrew, C., 1977. Space, time and polity. In: Friedman, J., Rowlands, M.J. (Eds.), The Evolution of Social Systems. Duckworth, London. Renfrew, C., 1986. Introduction: peer polity interaction and socio-political change. In: Renfrew, C., Cherry, J.F. (Eds.), Peer Polity Interaction and Socio-Political Change. Cambridge University Press, Cambridge, pp. 1–18. Renfrew, C., Bahn, P., 1996. Archaeology. Theories, Methods and Practice, second ed. Thames and Hudson Ltd, London. Rigsantikvarens Arkæologiske Sekretariat, 1984–1999. Arkæologiske udgravninger i Danmark 1984–1999. Det Aræologiske Nævn, København. Ritzer, G., Gindoff, P., 1994. Agency–structure, micro–macro, individualism–holism–relationism: a metatheoretical explanation of theoretical convergence between the United States and Europe. In: Sztompka, P. (Ed.), Agency and Structure. Reorienting Social Theory. Gordon and Breach, Amsterdam, pp. 3–24. Rudebeck, E., 2001. V€ agskel, v€agkorsningar och vadst€allen— liminala platser och arkeologi. In: Larsson, L. (Ed.), Kommunikation i tid och rum. Report Series No. 82, Institute of Archaeology, University of Lund, Lund, pp. 93– 112. Service, E.R., 1971. Primitive Social Organization. An Evolutionary Perspective, second ed. Random House, New York.

55

Shennan, S., 1997. Quantifying Archaeology, second ed. Edinburgh University Press, Edinburg. Stephenson, K., Zelen, M., 1991. Rethinking centrality. Social Networks 13, 1–37. Sztompka, P., 1994. Evolving focus on human agency in contemporary social theory. In: Sztompka, P. (Ed.), Agency and Structure. Reorienting Social Theory. Gordon and Breach, Amsterdam, pp. 25–60. Sørensen, M.L.S., 1997. Reading dress: the construction of social categories and identities in bronze age Europe. Journal of European Archaeology 5 (1), 93–114. Th€ orn, R., 2001. F€ orhistoriske v€ager och innovationer. Reflek€ tioner fr an Oresundsf€ orbindelsen. In: Larsson, L. (Ed.),Kommunikation i tid och rum, Report Series No. 82, Institute of Archaeology, University of Lund, Lund, pp. 51–60. Th€ orn, R., 2002. F€ orhistoriske v€ager i Malm€ o, Sverige. Braut 2. Thrane, H., 1975. Europæiske forbindelser. Bidrag til studiet af fremmede forbindelser i Danmarks yngre broncealder (periode IV–V). Nationalmuseet, København. Thrane, H., 1980. Nogle tanker om yngre broncealders bebyggelse p a Sydvestfyn. In: Thrane, H. (Ed.), Broncealderbebyggelse i Norden. Beretning fra det andet nordiske symposium for broncealderforskning. Odense d. 9.-11. april 1980, pp. 165–173. Skrifter fra historisk institut, Odense Universitet, nr. 28, Odense. Thrane, H., 1984. Lusehøj ved Voldtofte—en sydfynsk storhøj fra yngre broncealder. Fynske studier XIII, Odense. Thrane, H., 1985. Review of: Axel Hartmann: Pr€ahistorische Goldfunde aus Europa II. Spektralanalytische Untersuchungen und deren Auswertung. Studien zu den Anf€angen der Metallurgie Bd. 5. Gbr. Mann Verlag, Berlin, 1982. Journal of Danish Archaeology 4, 204–208. Thrane, H., 1991. Danish Plough-Marks from the neolithic and bronze age. Journal of Danish Archaeology 8, 111–125. Thrane, H, 1994. Centres of wealth in Northern Europe. In: Kristiansen, K., Jensen, J. (Eds.), Europe in the First Millenium B.C. Sheffield Archaeological Monographs 6. J.R. Collins Publications, Sheffield, pp. 95–110. Thrane, H., 1995. Placing the bronze age ‘‘lurer’’ in their proper context. In: Jockenh€ oven, A. (Ed.), Festschrift f€ ur Hermann M€ uller-Karpe zum 70. Geburtstag. In Kommission bei Dr. Rudolf Habelt GmbH, Bonn, pp. 143–152. Vandkilde, H., 1996. From Stone to Bronze. The Metalwork of the Late Neolithic and Earliest Bronze Age in Denmark. Jutland Archaeological Society Publications XXXXII, Aarhus. Wagstaff, M., 1986. What Christaller really said about central places. In: Grant, E. (Ed.), Central Places, Archaeology and History. Department of Archaeology and Prehistory, University of Sheffield, Sheffield. Wassermann, S., Faust, K., 1994. Social network analysis. Methods and applications. In: Structural Analysis in the Social Sciences 8. Cambridge University Press, Cambridge. Willroth, K., 1985. Die Hortfunde der €alteren Bronzezeit in S€ udschweden und auf den d€anischen Inseln. Karl Wachholtz Verlag, Neum€ unster. W€ ustemann, H., 1974. Zur Socialstruktur im Seddiner Kulturgebiet. Zeitschrift f€ ur Arch€aologie 8, 67–107.