Premises

Much has changed in the field of archaeology since Colin Renfrew and colleagues (Reference Renfrew, Todd and Tringham1974) coined the concepts of group-oriented and individualizing chiefdoms. Evolutionary approaches and static categories in general have been thoroughly criticized and abandoned in many circles, and the scholarly focus has shifted from categorization to the study of processes and agency (Thurston & Fernández-Götz, Reference Thurston and Fernández-Götz2021). Yet, Renfrew's two societal categories are worth revisiting, not least because they still resonate in modern literature (Currás & Sastre, Reference Currás, Sastre, Currás and Sastre2020b; Arnold, Reference Arnold and Thurston2021: 119).

Renfrew's group-oriented and individualizing chiefdoms were inferred from empirical observations of several prehistoric societies and ethnographic parallels. Although these societies share certain characteristics (limited territorial extent and population density, absence of towns and permanent institutions), they exhibit a series of contrasting traits and practices. One distinguishing trait sets them in an opposition to each other: whereas group-oriented societies are characterized by the presence of monumental constructions and the absence of visible status and wealth differences in material culture, most of the wealth in individualizing societies was invested in lavish burials or palaces, items indicating high status, rather than in major communal undertakings. The former are attested among the prehistoric societies of western Europe with their megalithic architecture, while the latter in the Bronze Age Aegean or Iron Age Europe, best-known for their palaces and ‘princely’ burials.

There is an unconcealed evolutionism in this perspective (Thurston & Fernández-Götz, Reference Thurston and Fernández-Götz2021: 1–3). Compared to individualizing societies, group-oriented societies are seen at a lower level of economic and technological development. It is the wider range of goods that provided individualizing societies with the material base to convey status and wealth differences. Here, I shall retain the opposition between group-oriented and individualizing societies but discard the evolutionary link, using the two types of society as a conceptual framework to rethink the brief episode of ‘princely’ burials in the central Balkans during the Late Hallstatt period and the radical social transformations they are thought to imply (Blanton et al., Reference Blanton, Feinman, Kowalewski and Peregrine1996; Arnold, Reference Arnold and Thurston2021: 119–20).

Most studies of the ‘princely’ burial phenomenon have adopted an evolutionary perspective, which sees it as the decisive, if not the first, step in the rise of the earliest proto-state societies in the central Balkans. This view does not appear to be fully borne out by the archaeological evidence. On the Glasinac Plateau, the setting of the first ‘princely’ burials in this region, the ‘princely’ burial horizon was followed by a period of rapid decline in settlement numbers and burials (Čović, Reference Čović and Benac1987). Here, I argue that the sudden appearance and disappearance of the ‘princely’ burials was an outcome of the cyclical oscillation of ancient societies between two opposite value systems and ideologies: one espousing strong communal identity and solidarity, and one adopting individual or group status and competition (Morris, Reference Morris and Ruby1999: 60–63; Blanton et al., Reference Blanton, Feinman, Kowalewski and Peregrine1996: 8–12).

Case Studies: Glasinac and Trebenište

Since the defining trait of ‘princely’ burials is their rich assemblages, it is essential to examine how this wealth relates to that of the ordinary burials. Ideally, such an analysis should be based on individual cemeteries formed within relatively short timespans (O'Shea, Reference O'Shea1984: 14, 30–31) but such cases are not easily found. It seems that most ‘princely’ burials did not belong to communal cemeteries in the first place, and, among the few exceptions, the material has rarely been published in its entirety.

The Trebenište burial ground near Ohrid (Figure 1) has a long history of investigations, in four episodes of excavations (Stibbe, Reference Stibbe2003). This brought to light fifty-six individual burials with complete inventories of grave goods spanning the period between the early sixth and the early fifth centuries bc.

The mound burials on the Glasinac Plateau in eastern Bosnia (Figure 1) (Benac & Čović, Reference Benac and Čović1957; Čović, Reference Čović and Benac1987) consist of numerous small mound cemeteries dated to the Bronze and Iron Ages, scattered over an area of a few hundred square kilometres. The ‘princely’ burials in these cemeteries cover the period between the middle of the seventh and the beginning of the fifth century bc, a slightly longer timespan than at Trebenište (Govedarica, Reference Govedarica2017 has redated some of these burials to the eighth century bc, but this revision creates a chronological conundrum that he did not address). Although the Glasinac material does not come from a single cemetery, it was used by close-knit communities that formed a discrete socio-cultural unit. The published assemblages from the Glasinac Plateau includes ninety-seven burials with more or less completely reconstructed sets of grave goods dated to the ‘princely’ burial horizon. Unfortunately, most of this material was excavated in the last decades of the nineteenth century, with standards of excavation and recording lagging behind even those of the earliest excavations at Trebenište. A series of reassessments (Lucentini, Reference Lucentini1981; Govedarica, Reference Govedarica2017: 40) has, however, largely resolved this issue.

Let us briefly consider the nature of the samples analysed here. The small number of burials per generation indicates that they are not even close to representing the communities to which they belonged (Morris, Reference Morris1987: 72–95). The question is not to estimate what proportion of the living population is missing, but to decide which segment of this population is likely to have been underrepresented in the Trebenište and Glasinac samples. The history of research at these burial grounds and their specific geoarchaeological contexts may be of help.

Both burial areas were discovered long ago and had long, if intermittent, research histories. This makes it less likely that too many rich burials are missing from our samples, and this is further supported by the lack of evidence of looting. Although the predominant burial forms and the local geoarchaeological circumstances differ, researchers at both sites have observed that the main factor affecting retrieval is exposure rather than burial. The poor burials at Trebenište are shallow, often lying less than 0.2 m below the present-day ground surface and hence more vulnerable to erosion than the rich burials set in pits 2 m deep. The last archaeologists who worked at Trebenište remarked that much of the surface material was probably derived from burials disturbed by erosion and modern land-use (Malenko, Reference Malenko1975; Kuzman, Reference Kuzman1985). At Glasinac, there is a positive correlation between the size of the burial mound and the wealth of the burial. Small mounds often do not show any evidence of formal burial. Some scholars have even doubted their being burial mounds (Govedarica, Reference Govedarica1978a), but the sparse archaeological material discovered amidst the heaps of cobbles does point to the possibility that these are the remnants of disturbed, poorly constructed graves.

It would therefore seem that most of the missing burials at Trebenište and Glasinac would have qualified as poor, in terms of construction and grave goods. In addition, we should consider the possibility that certain segments of the population were given a different postmortem treatment and are entirely missing from the mortuary record (Morris, Reference Morris1987: 97–109). All this leads to the conclusion that the distribution of grave goods across the original population of burials was probably far more uneven than suggested by the available sample.

Methodology

The supposed social inequality in our study area, as reflected in the distribution of grave goods in Trebenište and Glasinac (Morris, Reference Morris1987: 40–43; Mihajlović, Reference Mihajlović, Currás and Sastre2020), must first be addressed, but poor data and limited space here prevent a discussion of the spatial patterns observed within cemeteries or the construction techniques employed (O'Shea, Reference O'Shea1984; Hodson, Reference Hodson1990; Parker Pearson, Reference Parker Pearson1999). Many scholars believe that the range of grave goods in Iron Age burial assemblages represent the personal belongings of the deceased, thus offering some reflection of social inequality (Čović, Reference Čović and Garašanin1979: 164; Stibbe, Reference Stibbe2003). Be that as it may, my goal here is to examine the distribution of wealth among the population of burials, and compare this distribution to data from the non-mortuary realm, not to reconstruct the structure of Late Hallstatt societies.

First, I analyse the distribution of burials across categories defined by the number of grave goods (e.g. poor, ordinary, rich, etc.; Shennan, Reference Shennan1990) but, to gain additional insights into the distribution of burial wealth, I shall also use a well-established measure of social inequality known as the Gini coefficient. Although originally used to study income distribution in modern societies, the Gini coefficient can be used to index the amplitude of differences in the distribution of virtually any class of objects or attributes (Morris, Reference Morris1987: 42–43; Smith et al., Reference Smith, Dennehy, Kamp-Whittaker, Colon and Harkness2014; Scheidel, Reference Scheidel2017). Here, it is specifically used to measure the inequality in the distribution of grave goods as a reflection of funerary ideology, not social inequality.

When calculating this index, it will be necessary to plot the cumulative frequency distributions of grave goods, the so-called Lorenz curves. Because of the small size of our samples, to compare the differences between the resulting Lorenz curves, a two-sample Kolmogorov–Smirnov test (Shennan, Reference Shennan1990: 57-60) will be applied. The number of burials at Trebenište and Glasinac is uneven and, to facilitate comparison, the burial populations are ranked and broken into quartiles, and the quantity of grave goods expressed in percentages.

Because the range of grave goods, especially in the rich Trebenište burials, is bewildering, they must be summarized quantitatively (O'Shea, Reference O'Shea1984: 61–63). The simplest way of comparing the wealth of individual burials is to count the number of grave goods in each burial, but this can be misleading, not least because different types of goods often have unequal values—a bronze crater is obviously worth hundreds of bronze pins or pendants. Therefore, the individual artefacts represented at Trebenište and Glasinac have been grouped into general categories and each category weighted by a factor ranging from 0.5 to 4, determined by the quantity and quality of the raw material, and the energy invested in their production (Table 2). Large artefacts and those made of rare or precious materials are valued higher than small artefacts made of bronze, iron, or ceramics. Obviously, individual items, not fragments, are counted and for this we must rely on the reconstructions proposed in the publications. Artefacts that appear in very large and loosely determined quantities (amber or glass paste beads) were excluded from the analysis.

Table 2. Rank and weighting of the different categories of grave goods.

This approach not only makes it possible to compare individual graves from the same site but also to see the differential distribution of wealth between burial grounds that belong to different cultural backgrounds or periods. On the downside, the most obvious difficulty is to find a non-arbitrary way of assigning value to the different categories of grave goods. This must take into account not only the physical properties of the artefacts (size, weight, and material) but also the production techniques and the provenance of the objects, which are often elusive. But even if the finely-grained data necessary to make these calculations were available, it would still be impossible to assign a value to the local, symbolic worth of the different artefact types (Parker Pearson, Reference Parker Pearson1993: 207). While I acknowledge these issues, these deficiencies do not necessarily undermine the integrated approach and methodology adopted in this study.

In a final step, I examine the evidence for settlement hierarchy and differential growth outside the realm of mortuary practices (Mihajlović, Reference Mihajlović, Currás and Sastre2020), shifting the focus onto the data available for settlement size. Comparative, multiperiod, and cross-cultural analyses by economic geographers, anthropologists, and archaeologists have shown that complex and hierarchically organized societies regularly exhibit so-called ‘steep’ settlement hierarchies: a small number of large centres and a progressively greater number of settlements of smaller size and rank (Hanson, Reference Hanson2016). It is most easily visualized as a pyramid, with many settlements at its base and a few or one settlement at the top. This pattern is generated by the uneven distribution of the population and, concomitantly, labour, services, and wealth. By contrast, simple agrarian and poorly integrated societies usually have so-called ‘shallow’ hierarchies, that is, although there are differences in size between individual settlements, these are negligible and cannot be assigned to distinct size categories (Currás & Sastre, Reference Currás, Sastre, Currás and Sastre2020b). These slight variations are related to the variable productivity of the respective settlements or the success of individual communities at exploiting their environments and increasing their numbers, not to their status and role in society.

Starting with this premise, data were collected for the size of seventy-three settlements dated to the Late Hallstatt period and for thirty settlements dated to the Classical–Hellenistic period (Table 3). Although known mainly from limited excavations and surveys, the artefact assemblages suggest that these sites were permanently occupied. Only a few can be linked to specific ‘princely’ burials, but all are located in the same general area (Figure 2). Most are fortified hilltops, and the size estimates, extracted from plans or satellite images, generally refer to their walled areas. This is not without problems because, in most cases, the construction date of the enclosure is unknown. Nonetheless, because there is very little evidence that the Late Hallstatt hilltop settlements extended beyond their ramparts, it is likely that size estimates are roughly accurate or slightly higher. This changed in the Classical–Hellenistic period, with evidence of settlements spreading beyond their defences at several sites.

Table 3. Sources used for assessing settlement size.

Figure 2. Distribution of Late Hallstatt and Classical–Hellenistic settlements in the central Balkans.

Glasinac and Trebenište Compared

The differences between the elite burials in these two sites have long been noted in the scholarly literature (Babić & Palavestra, Reference Babić, Palavestra, Ardjanliev, Chukalev, Cvjetićanin, Damjanov, Krstić, Papazovska and Popov2018: 192). Although unaware of the elite burials in the Western Morava Basin at the time, Benac and Čović (Reference Benac and Čović1957: 31; Čović, Reference Čović and Garašanin1979: 163–64) felt compelled to qualify their use of the epithet ‘princely’ for the rich burials on the Glasinac Plateau. They even suggested the label ‘proto-princely’ to indicate their similarity with the preceding horizon of warrior burials and their unimpressive assemblages relative to the other ‘princely’ burials then known (Govedarica, Reference Govedarica2017: 59–60). The Trebenište burials not only contain a much wider range of artefacts than the Glasinac burials (gold is absent there) but also a greater number of imports and larger and more expensive items. The two burial grounds also differ in the number of rich burials (Babić, Reference Babić2004: 72–75). Trebenište, like some other contemporary cemeteries in Macedonia, contains multiple elite burials—about a dozen—whereas most ‘princely’ burials to the north, including those on the Glasinac Plateau, appear isolated or in small groups (Chrysostomou & Chrysostomou, Reference Chrysostomou, Chrysostomou, Tiverios, Nigdelis and Adam-Veleni2012).

However, the differences between these two burial grounds are hardly apparent from the distribution of individual burials across categories defined by the number of grave goods recovered (Figure 3a). At both sites, the proportion of burials that fall into one of the four classes (poor, ordinary, rich, and ‘princely’) is the same. Note, however, the differences in the absolute quantity of grave goods and the variable spectra of the artefact types between the two cemeteries. At Trebenište, both the data range and the standard deviation are much greater than at Glasinac, suggesting subtler variations in burial wealth at the latter site. In other words, the rich burials at Trebenište are richer than their counterparts at Glasinac, and these differences can be quantified.

Figure 3. a) The Glasinac and Trebenište burials compared; b) Lorenz curves for the Glasinac and Trebenište burials.

Figure 3b compares the Lorenz curves for the distribution of grave goods at Trebenište and Glasinac. Individual burials are ranked by wealth and divided into quartiles, from poorest to richest. The top line plots the spread of grave goods in an ideal situation, in which each grave has an equal value of grave goods. In short, the steeper the segments of the Lorenz curves, the sharper the differences between individual burials.

The Lorenz curve for Trebenište lies further away from the ideal case than the Glasinac series, although statistically the distance between the two is insignificant (Table 4). At Glasinac, the bottom half of the burials contain nine per cent of all grave goods included in this analysis, whereas at Trebenište, the two bottom quartiles contain only two per cent. The gap between the two series continues to widen, so that at Glasinac the first three quartiles of burials contain slightly over one quarter of all finds while at Trebenište, they hold only fourteen per cent of all grave goods. The richest twenty-five per cent of the burials include seventy-four per cent of all grave goods and offerings at Glasinac, and eighty-six per cent at Trebenište. The different degree of inequality between the two sites is also evident from the Gini coefficients (Table 4).

Table 4. Gini coefficients of inequality for the cemeteries analysed and results of the Kolmogorov-Smirnov tests. Grey fields indicate that the null hypothesis is rejected.

If these figures have any value as measures of contemporary social, rather than transgenerational or purely symbolic, inequalities, the differences in wealth are remarkable in both societies. These differences must have been even sharper than suggested by the graphs, because the distribution of grave goods is also uneven among the richest quartile of burials.

Comparison with Fifth- to Fourth-Century bc Cemeteries

Notwithstanding some of the qualitative differences between Glasinac and Trebenište, both display similar patterns in the distribution of grave goods, probably reflecting a common mortuary ideology. The occupants of the Trebenište ‘princely’ burials were better connected and supplied than their distant counterparts at Glasinac, but this is a different problem. At this stage, I shall focus on the extent to which these distribution patterns continue into the Classical–Early Hellenistic period, which scholars often see as dominated by proto-state societies (Vranić, Reference Vranić2012).

Finding complete publications of cemeteries from this period that include a sufficiently large number of burials proved challenging. The necropolis at Trebeniško Kale, only a few kilometres north of the Trebenište ’princely’ burials (Figure 1), is among the few partly excavated sites, but only about twenty graves have been published (Lahtov, Reference Lahtov1959). A few burial grounds of this period are known from nearby Pelagonia, but the total number of graves excavated and published does not exceed thirty per cemetery (Crkvište: Mikulčić, Reference Mikulčić1966a, Reference Mikulčić1966b; Varoš-Kaldrma: Kitanoski, Reference Kitanoski1975). These samples are too small to be statistically meaningful, and amalgamating them could introduce other problems. Nevertheless, regardless of whether they are analysed separately or jointly, the deviations from the preceding patterns are obvious.

A first look at the number of burials per categories based on the weighted number of grave goods does not reveal any major differences (Figure 4). With the sole exception of Varoš-Kaldrma the poor burials comprise between sixty and seventy per cent of the total population at each site, echoing the situation in the preceding period. However, the range of grave goods in the Classical–Hellenistic cemeteries is much smaller than in the Late Hallstatt burial grounds and consequently the standard deviation is quite low. In other words, the differences between poor and rich burials were far subtler than in the preceding period, and this is not simply due to the absence of ‘princely’ burials in the Classical–Early Hellenistic cemeteries. Even the richest burials at Crkvište or Kaldrma would have ranked as ordinary among the Trebenište burials.

Figure 4. Distribution of burials by categories at Classical–Early Hellenistic cemeteries.

The differences between the two groups become far more evident once the cumulative distributions of the grave goods are compared (Figure 5a). As in the preceding period, in the Classical–Hellenistic cemeteries, the poorest quartile of burials contains less than one per cent of the total corpus of grave goods but, by the second quartile, the proportion of grave goods rises above ten per cent, whereas at Glasinac and Trebenište it stays below this threshold. In all Classical–Early Hellenistic cemeteries analysed here, the proportion of grave goods is greater than thirty per cent by the third quartile, reaching over forty-five per cent at the necropolis of Kaldrma. By contrast, in Glasinac and Trebenište, the bottom three quarters contained less than one quarter of the total number of grave goods. Consequently, the resulting curves lie much closer to the Lorenz curve for uniformly distributed grave goods than the corresponding curves for Glasinac and Trebenište. Whether taken separately or jointly, Trebeniško Kale, Crkvište, and Kaldrma have lower Gini coefficients than the Late Hallstatt cemeteries (Table 4). The samples are obviously too small for statistical testing, but when the cumulative distribution of grave goods from Crkvište is compared to the data series for Trebenište with the help of a two-sample Kolmogorov–Smirnov test, the differences are significant at a significance level of 0.05. The same holds true for the Lorenz curves for the combined Late Hallstatt and Classical–Hellenistic cemeteries (Table 4; Figure 5b).

Figure 5. a) Lorenz curves for individual Late Hallstatt and Classical–Early Hellenistic cemeteries; b) Lorenz curves for the combined Late Hallstatt and Classical–Early Hellenistic cemeteries.

Despite the small sample size, the changing trends in the distribution of grave goods between these two periods are evident. The mortuary practices in the period that followed the ‘princely’ burials horizon were marked by a relative restraint, at least in the archaeologically visible spectrum of the funerary rites. However, this trend is open to different interpretation, as it is unclear whether these changes in funerary ideology reflect changes in religious beliefs or have wider social implications.

Settlement Size

The most straightforward way to examine the presence of settlement hierarchies is to compare the number of settlements per size range (Figure 6a), but this is potentially misleading given the differences in the range of sizes of settlements in these two periods; namely, the Classical–Early Hellenistic settlements have a much greater range of sizes than those of the Late Hallstatt settlements. The latter range between 0.2 and 4 ha, the former between 0.2 and 20 ha. Whereas, most Classical–Early Hellenistic settlements fall within the same range as the Late Hallstatt settlements, five sites, or less than twenty per cent of the Classical–Early Hellenistic settlements, are two to three times larger than the largest Late Hallstatt settlements and they belong to a settlement tier that was absent in the previous period.

Figure 6. a) Comparison of size ranges of Late Hallstatt and Classical–Early Hellenistic settlements; b) cumulative graphs for settlement size in the Late Hallstatt and Classical–Early Hellenistic periods.

This is manifest in the cumulative graphs for settlement size (Figure 6b). The contribution of each quartile of Late Hallstatt settlements to the total settlement area is consistently greater than the contribution of the respective quartiles of the Classical–Early Hellenistic settlements, but this difference is most pronounced in the top quarter of settlements. The top quartile of the Late Hallstatt settlements accounts for one half of the total settlement area, whereas the Classical–Early Hellenistic settlements of corresponding rank account for almost two-thirds of the total settlement area. Consequently, the cumulative curve for the Late Hallstatt settlements is closer to the distribution of settlements of equal size than the cumulative curve for the Classical–Hellenistic settlements.

The difference in settlement size among the Late Hallstatt settlements is negligible and determined primarily by factors such as the local topography or agricultural productivity. Only in the Classical–Early Hellenistic period is there a true differential growth, with a small group of settlements rising to dominate the system. These differences in settlement size cannot be attributed to local environmental conditions. Their scale is such that they were likely to have been caused by social factors leading to the preferential concentration of people and wealth in a few settlements.