INTRODUCTION
The radiocarbon (14C) record for Later Prehistoric Europe is arguably one of the richest, if not the richest, across the world, both in terms of quantity of dates and density of geographical coverage. That being said, the state of affairs remains far from ideal as numerous documentary gaps persist, generally related to the history of local research and corresponding delays and logistical limits in adopting and using the 14C technique. Without much surprise, perhaps, one of these gaps is centered around the central Balkans, corresponding to most of modern-day Serbia south of the Sava and Danube rivers. While the record for the Mesolithic and Neolithic periods is good, thanks to the presence of famous sites that have attracted continuous scientific attention (e.g. the Danube Gorges and the site of Vinča; see recently Bonsall et al. Reference Bonsall, Macklin, Boroneanţ, Pickard, Bartosiewicz, Cook and Higham2015; Tasić et al. Reference Tasić, Marić, Filipović, Penezić, Dunbar, Reimer, Barclay, Bayliss, Gaydarska and Whittle2016), later periods are less well-covered. This situation is particularly unfortunate as this area lies at the crossroads of many natural communication routes connecting central Europe with the Balkan peninsula and, beyond, the Aegean and the Adriatic seas. The natural corridors include numerous large river valleys (Velika Morava, Južna Morava, Nišava rivers), and seem to have been preferentially settled during later prehistory.
The site of Bubanj, which is the focus of this paper, lies at the heart of one of these corridors, more precisely at the confluence of the Nišava and Južna Morava rivers, approximatively 5 km to the west of the modern city of Niš (Figure 1). The first excavations were conducted during the 1930s and provided important results quickly published by the Austrian Academy of Sciences in Vienna (Orsich de Slavetić Reference Orssich de Slavetich1940). This prehistoric site then covered an extensive surface estimated to approximately 3.5 ha, stretching over three main areas (western, central, and eastern), and appeared to have been almost continously occupied from the Neolithic to the Bronze Age (e.g. existence of stratigraphic profiles up to 3 m high). Further excavations took place during the 1950s, mostly centered upon the central and western areas (Garašanin Reference Garašanin1958). Following this phase of fieldwork, sequence and material culture from Bubanj as an eponymous site for several Eneolithic and Bronze Age archaeological cultures (Garašanin Reference Garašanin1958). It is noteworthy that no samples were then taken for 14C analysis, admittedly still in its early stages.
Figure 1 Location of the site of Bubanj, Serbia(Drawing: A. Bulatović)
The first cultural group named after the site is the Bubanj-Hum I group, which corresponds to the Early Eneolithic in the local terminology. This group belongs to the larger Bubanj-Salcuţa-Krivodol cultural complex (hereafter BSK), documented over much of the Balkans in western Bulgaria, southwestern Romania, Serbia, as well as parts of Macedonia, Albania, and Greece. Although 14C dates for this cultural complex are relatively numerous, there is a great imbalance in their geographical distribution, with so far only a single date available for the central Balkans at the site of Bodnjik (western Serbia; Živanović Reference Živanović2013:54). As a result, precise chronological relationships between the regional variants of this complex remain poorly understood (e.g. lack of dates for Bubanj-Hum I in Serbia or Šuplevac-Bakarno Gumno in Macedonia; good chronology for both Salcuţa I-III in Romania and Krivodol I-II in Bulgaria: Boyadziev Reference Boyadziev1995; Lazarovici Reference Lazarovici2006). An overall attribution to the middle and/or second half of 5th millennium cal BC is widely accepted, on the basis of existing 14C dates and typological comparisons with older complexes (e.g. Kodžadermen-Gumelniţa-Karanovo VI in northern Bulgaria and southern Romania: Boyadziev Reference Boyadziev1995; Nikolov and Petrova Reference Nikolov and Petrova2016; and Gradešnica-Slatino-Dikili Tash in southern Bulgaria and northern Greece: Boyadziev Reference Boyadziev1995; Roque et al. Reference Roque, Guibert, Vartanian, Bechtel, Treuil, Darcque, Koukuli-Chryssanthaki and Malamidou2002).
For the Middle and Late Eneolithic, typological comparisons between Bubanj and other sites concern the Černavoda III and Baden-Boleraz cultures (hereafter ČV III-B-B complex), Koţofeni-Kostolac and Bubanj-Hum II cultures (Jovanović Reference Jovanović1976; Tasić Reference Tasić1979; Garašanin Reference Garašanin1982:153–5; Tasić Reference Tasić1995; Kapuran and Bulatović Reference Kapuran and Bulatović2012). Typologically, the pottery assemblage from Bubanj belongs to the Černavoda III culture, but also presents numerous pottery typological traits echoing the central European Baden culture, especially its Boleraz horizon (Bulatović and Kapuran, forthcoming). Once more, the existing documentation is very unequal, with numerous dates for the Baden-Boleraz culture, mostly clustered in central Europe (Wild et al. Reference Wild, Stadler, Bondár, Draxler, Friesinger, Kutschera, Priller, Rom, Ruttkay and Steier2001; Furholt Reference Furholt2008; Horvath 2008), while dates for the Černavoda III culture are so far only available for the site of Drama-Merdžumekja, in eastern Thrace (Bulgaria; Gleser & Thomas 2012). The Koţofeni-Kostolac culture (hereafter K-K), attributed to the Late Eneolithic, is also sparsely dated, with only a single date for the central Balkans at the site of Belovode in central Serbia, where a K-K layer was dated by chance in course of the analysis of Late Neolithic samples (Borić Reference Borić2009:196, their Table 2). The Bubanj-Hum II group, although named after the site of Bubanj, will not be covered here as it was poorly represented in the recent excavations on which this paper is based (see below).
The situation is similar for the Early Bronze Age, a period for which the site of Bubanj is once more instrumental in the establishment of the relative chronology (cf. Bubanj-Hum III culture; hereafter BH III). The Serbian archaeological literature, in the near absence of any 14C dates (but see Bogdanović Reference Bogdanović1986; Krstić et al. Reference Krstić, Bankoff, Vukmanović and Winter1986:34; Gogaltan Reference Gogâltan1999), relies upon relative chronology and analogies with the material culture of neighbouring, absolutely dated cultures (Bulatović Reference Bulatović2011), such as the Moriş (or Mokrin) culture to the north (O’Shea Reference O’Shea1996) and Armenochori group to the south (Maniatis and Ziota Reference Маniatis and Ziota2011; Gori Reference Gori2015), thus suggesting a date to the last quarter of the 3rd and the beginning of 2nd millennia BC.
Here, we fill this damaging gap by presenting the first 14C dates obtained for the site of Bubanj, covering both the Eneolithic and the Early Bronze Age. These new results, all obtained on samples from recently excavated secure archaeological contexts, are then compared to dates surveyed from the literature for each of the aforementioned archaeological complexes, in order to test the validity of the chronological estimates based upon traditional typological comparisons.
Methods
As previously mentioned, the site of Bubanj originally covered a total surface of approximately 3.5 ha, distributed over a western, central, and eastern area. Unfortunately, since the 1950s, the site has been subject to extensive destruction of various sorts. When a new archaeological field campaign was organized in 2008, only 200 m2 located in the eastern area were available for scientific investigations. Excavations were regularly conducted between 2008 and 2014 by the Archaeological Institute in Belgrade and the National Museum in Niš, in the course of which a broad range of samples was acquired, including bone samples for 14C analysis reported here. Excavations were organized over a total surface of 150 m2, in a 1 × 1 m square grid, coupled with artificial excavation spits of 3–5 cm deep. Archaeological structures were excavated as distinct coherent units.
The stratigraphy covers the Early Eneolithic (BSK), which immediately rests upon the soil substrate, the Middle (ČVIII B-B) and Late Eneolithic (K-K), each represented by two distinct horizons, and the Early Bronze Age (BH III). Younger deposits are absent because of extensive soil erosion and recent activity on the site, although finds from various periods are found at the surface and in the upper, heavily disturbed, meter of the stratigraphy.
A total of 10 samples, covering the entire stratigraphic sequence, were selected for 14C dating. All samples were taken from secure stratigraphic contexts (Figure 2), and were closely associated with typologically well-defined pottery production. To minimize the risks of dating residual samples, animal bones were privileged as part of the sampling strategy, and were all identified by an experienced zooarchaeologist prior to submission for dating. Nine samples were submitted for counting to the SUERC Radiocarbon Dating Laboratory in Glasgow, where they were treated following the standard protocol described by Dunbar and colleagues (Dunbar et al. Reference Dunbar, Cook, Naysmith, Tripney and Xu2016). One supplementary sample was obtained from the Centre de Datation par le Radiocarbone, Lyon. Calibration and further Bayesian modeling (see below) were performed using Oxcal 4.2 (Bronk Ramsey Reference Bronk Ramsey2009), using Intcal 13 (Reimer et al. 2013). All results are reported in Table 1.
Table 1 Newly acquired 14C dates for the site of Bubanj.
Figure 2 Plan of the 2008–2014 excavations in Bubanj(Drawing: A. Bulatović)
The oldest dates were provided by 2 animal bone samples recovered from the center of an Eneolithic pit (structure 69), possibly of ritual function (Bulatović Reference Bulatović2015:30, their fig.3). Its opening is nearly circular (~1.7 m diameter), and its sides, deeply dug into the substrate (depth: 1.5 m) were fire-baked. Its fill includes a mix of pottery fragments (Figure 3: items 1–5), chipped stone tools, daub fragments, animal bones, ash, and other finds (Bulatović Reference Bulatović2015: Pl. II). The pottery typology points to an attribution to the BSK complex. The Middle Eneolithic is dated by 4 samples coming from the 17th artificial excavation spit, 2 pits and a scatter of pottery and animal bones. Pit 108 was dug from the lower horizon of this cultural layer, and probably corresponds to a refuse pit (1.8×1.2 m, max depth 0.6 m). It contained large amounts of pottery fragments (Figure 3: items 7–10), animal bones, daub fragments, considerable quantity of burnt wood, a copper needle fragment, as well as other objects of baked earth and stone. The pottery belongs to the ČV III-B-B complex. Another dated pit (structure 54) has a nearly circular base, and was dug from the upper horizon of ČV III-B-B cultural layer (Table 1, row 4). It contained pottery fragments (Figure 3: item 11) and animal bones. We also obtained a date from a bone sample coming from feature 23 (Table 1, row 6), a scatter of pottery and animal bones found in the lower horizon of ČV III-B-B complex (196.16–196.24 m asl). Once more, the pottery typology clearly points to the ČV III-B-B complex (Figure 3: item 6). The last sample from this horizon (Table 1, row 3) comes from the excavation layer and contained pottery characteristic for ČV III-B-B complex.
Figure 3 1–5: structure 69 (Bubanj-Hum I group); 6: structure 23 (ČV III-B-B complex); 7–10: structure 108 (ČV III-B-B complex); 11: structure 54 (ČV III-B-B complex); 12–16: structure 15 (K-K culture); 17–22: structure 3 (K-K culture); 23–28: structure 1 (Bubanj-Hum III group).
Two samples were obtained from sealed contexts for the K-K horizon. A first sample was taken from the lower floor of a house (structure 15; Table 1, row 7), in direct association with K-K potsherds (Figure 3: items 12–16). The second sample (Table 1, row 8), comes from the base of another house (structure 3/1), from the younger K-K horizon (Figure 3: items 17–22), and is also directly associated with pottery characteristic for this culture. In the absence of secure stratigraphic contexts, it was decided not to take any sample from the layer above, in which material culture belonging to the Bubanj-Hum II group was found.
The Early Bronze Age (BH III group) is dated by finds coming from two secured stratigraphic contexts. The first sample (Table 1, row 10) comes from a deep pit, with a nearly circular base (approximate dimensions: diameter 1.4 m, depth 1.6 m; structure 20). Although the upper part of the pit had been recut by a Medieval grave, it presented an interesting succession of fills, suggesting a refusal function. It contained pottery fragments as well animal bones, daub fragments, and other finds (bone awl, chipped stone tools). The dated sample is identified as Equus species, although it is impossible to determine whether this corresponds to a wild or domesticated animal. The second sample comes from a shallow oval-shaped pit (structure 1) (Table 1, row 9), which contained numerous pottery fragments (Figure 3: items 23–28), and some animal bones.
RESULTS AND DISCUSSION
To improve the precision of our results, we built a Bayesian model in OxCal 4.2. Such an approach allows us to constrain the probability distribution of 14C dates through the inclusion of robust, independent chronological data, known as prior information (Bronk Ramsey Reference Bronk Ramsey2009). In this case, this prior information corresponds to the stratigraphic relationships between the samples. The resulting model includes as few assumptions as possible, with a single sequence of 4 bounded phases. The model presents an overall good agreement (Amodel: 102.9) (Figure 4).
Figure 4 Bayesian modeling of the new 14C dates for Bubanj.
On the basis of the Bayesian modeling, the BSK horizon in Bubanj starts in 4618–4259 cal BC (95.4% probability), possibly 4383–4273 cal BC (68.2% probability) and ends in 4341–3739 cal BC (95.4% probability), possibly 4331–4171 cal BC (68.2% probability), thus pointing to a relatively short phase of use (duration: 0–68 yr, 95.4% probability; 0–30 yr, 68.2% probability). In order to compare this result with the overall development of the BSK complex, we compiled a dataset of all corresponding 14C dates found in the literature (see Table A1 in Appendix). This dataset only lists 38 dates for 13 sites, including Bubanj, showing the reduced size of the documentation, mostly coming from Bulgaria and Romania. This dataset was then used to build another Bayesian model, considering the BSK complex as a single bounded phase (Figure 5). Such an approach provides quantitative estimates for the beginning and end of the modeled phase, which are more robust than mere visual inspections of the range of calibrated 14C dates (Bayliss et al. Reference Bayliss, Bronk Ramsey, van der Plicht and Whittle2007). This model (Amodel: 106.2) suggests that the BSK complex begins in 4573–4422 cal BC (95.4% probability), possibly 4530–4465 cal BC (68.2% probability), and ends in 3947–3806 cal BC (95.4% probability), possibly 3932–3869 cal BC (68.2% probability). Without much surprise given the small sample, the earliest expressions of the BSK complex are located in the well-dated areas, which are both western Bulgaria and Oltenia. Interestingly, the site of Bubanj lies slightly later in the sequence during the third quarter of the 5th millennium cal BC. Further precision is however hampered by the shape of the calibration curve for this period. The background of the emergence of the BSK complex has been discussed at length, especially regarding its relationship with the preceding Vinča culture. Recent Bayesian modeling of this culture, based on a much larger dataset, suggests that its later ceramic phase Vinča D began in 4870–4725 cal BC (95.4% probability), possibly in 4810–4740 cal BC (68.2% probability), and ended in 4515–4360 cal BC (95% probability), possibly in 4490–4415 cal BC (68.2% probability) (Whittle et al. Reference Whittle, Bayliss, Barclay, Gaydarska, Bánffy, Borić, Draşovean, Jakucs, Marić, Orton, Pantović, Schier, Tasić and Vander Linden2016). The comparison between both models suggests an overlap of 1–2 centuries between both ceramic styles.
Figure 5 Bayesian modeling of 14C dates for the BSK complex (dates listed in Table A1)
The second cultural phase represented in the new Bubanj excavations corresponds to the ČV III-B-B complex. Our Bayesian model suggests that this horizon began in 3717–3334 cal BC (91.3% probability), possibly 3434–3345 cal BC (62.8% probability), and ended in 3361–3106 cal BC (95.4% probability), possibly 3359–3258 cal BC (62.7% probability), with a duration spanning 1–3 centuries (duration: 0–275 yr, 95.4% probability; 0–79 yr, 63.7% probability). There is also a noticeable gap between this and the preceding BH I horizon, estimated to have lasted from 2 to more than 9 centuries (Interval BSK-ČV III: 223–983 yr, 92.4% probability; 673–953 yr, 68.2% probability). This situation is widely encountered across the territory of modern-day Serbia, where the archaeological record for the early 4th millennium cal BC is scanty, especially when compared to neighboring countries. It is impossible to assess whether this gap is a byproduct of modern research or reflects a past reality, though the last hypothesis should not be discarded as several recent studies in other European regions point to the existence of fluctuations in demographic regimes and settlement patterns (see review in Shennan Reference Shennan2012).
The nature of the Baden culture, and its relationships with the Černavoda III culture have long been discussed, and chronology has prominently featured in this debate. The Baden culture is generally dated between 3650 and 2800 cal BC (Wild et al. Reference Wild, Stadler, Bondár, Draxler, Friesinger, Kutschera, Priller, Rom, Ruttkay and Steier2001; Furholt Reference Furholt2008; Horváth et al. Reference Horváth, Svingor and Molnar2008). Furholt points to the existence of an early phase, characterized by coarse wares, dated to 3650–3350 cal BC, immediately followed by the Boleraz phase, dated to 3520–3350 cal BC, which is characterized by fine wares and a process of spatial expansion (Furholt Reference Furholt2008). Links between this Boleraz phase and the Černavoda III culture have been suggested on typological grounds (see below). Wild et al. (Reference Wild, Stadler, Bondár, Draxler, Friesinger, Kutschera, Priller, Rom, Ruttkay and Steier2001) rejected this hypothesis by showing, on the basis of few dates, the contemporaneity of the Baden culture with the Černavoda I culture, which supposedly precedes the Černavoda III one (Roman 1999). However, their argument is flawed by the fact that the precise contexts from which these dates were taken are contested (Nikolova Reference Nikolova1999:89).
The overall chronological brackets of the Baden culture are confirmed by our own analysis of the literature, which included 113 dates distributed over 40 sites (Table A2). A simple Bayesian modeling of this dataset (Figure 6), as a single bounded phase, indeed confirms that the whole Baden culture began in 3667–3578 cal BC (95.4%), possibly 3652–3609 cal BC (68.2% probability), and ended in 2886–2824 cal BC (95.4% probability), possibly 2876–2846 cal BC (68.2% probability). We then compared these results with a similar Bayesian modeling of the few existing dates for the Černavoda III culture (Figure 6), which so far only consist of 5 dates from the sites of Drama-Merdžumekja (Bulgaria; Gleser and Thomas 2012), and the 4 dates for Bubanj reported here (Table A3). Given the scarcity of the evidence, we adopted a conservative approach by inserting all dates, although the oldest one (Erl-14441, 4751±57 BP) presents a poor overall agreement (A=55.6). The model suggests that the Černavoda III culture began in 3551–3353 cal BC (95.4% probability), possibly in 3416–3359 cal BC (68.2% probability), and ended in 3369–3166 cal BC (95.4% probability), possibly in 3358–3317 cal BC (68.2% probability). This confirms that the Černavoda III culture starts later than the beginning of the Baden culture, but is strictly contemporaneous with the Boleraz phase. Despite the controversy, such a result should not be surprising, given that numerous sites in the central Balkans present traits from both complexes in close association (e.g. lids of Bratislava type, bowls with funnel-shaped neck channeled on the interior, plastic appliques on rim and elsewhere, Fischbute, ball-shaped smaller pots decorated with vertical channels, bowls with everted S profile featuring accentuated shoulder ornamented with two horizontal rows of pitted dots and others; Bulatović and Kapuran, forthcoming). These new dates thus suggest that the process of spatial expansion associated with the Boleraz phase (Furholt Reference Furholt2008) had wider repercussions, exemplified by the presence of the aforementioned typological traits in the ČV III culture in the central Balkans.
The third cultural horizon represented in the new Bubanj excavation is associated with the K-K culture. According to the Bayesian model, this horizon began in 3335–3018 cal BC (95.4% probability), possibly 3108–3039 cal BC (66.2% probability) and ended in 3108–2326 (95.4% probability), possibly 3083–2821 cal BC (68.4% probability) (duration: 0–225 yr, 95.4% probability; 0–94 yr, 68.2% probability). This third horizon is separated from the previous ČVIII horizon by a compact grey layer, interpreted as a surface leveled on purpose by the settlers associated with the K-K horizon. This reading is partly confirmed by the rapid succession between both phases, estimated to 0–275 yr (95.4% probability), possibly 0–154 yr (68.2% probability). Based on already published dates (Figure 7; Table A4), we estimate the start of the K-K culture to 3352–3037 cal BC (95.4% probability), possibly 3205–3058 cal BC (68.2% probability), and its end to 2875–2667 cal BC (95.4% probability), possibly 2862–2771 cal BC (68.2% probability), thus confirming previous estimates (Nikolić Reference Nikolić2000:78). The dates from Bubanj nearly cover the entire sequence of this culture, and appear to be roughly contemporaneous with both Kostolac (sites of Vučedol, Pivnica and Gomolava; Waterbolk Reference Waterbolk1988; Benko et al. Reference Benkö, Horvath, Horvatinčić and Obelić1989; Bojadžiev Reference Bojadziev1992), and Koţofeni facies (Bojadžiev Reference Bojadziev1992, Reference Bojadžiev1998:357).
Figure 7 Bayesian modeling of 14C dates for the Koţofeni-Kosolac complex (dates listed in Table A4)
Given the aforementioned poor preservation of the upper part of the stratigraphic profile, we were only able to obtain dates from secure contexts for the BH III layer, corresponding to the Early Bronze Age in the local terminology. The extensive damage occasioned to the stratigraphy is confirmed by the chronological gap between this horizon and the preceding one, estimated to 102–1080 yr (95.4%), possibly 575–1036 yr (68.4%). The BH III horizon in Bubanj began in 2624–1946 cal BC (95.4% probability), possibly 2191–1976 cal BC (68.2% probability), and ended by 2121–1652 cal BC (95.4% probability), possibly 2030–1869 cal BC (68.4% probability), and thus lasted a century at most (duration: 0–128 yr, 95.4% probability; 0–47 yr, 68.4% probability). These estimates are older than dates from the oldest horizon at the site of Ljuljaci, attributed to the proto-Vatin (Bogdanović Reference Bogdanović1986), or Bubanj-Hum IV-Ljuljaci group (Bulatović and Stankovski Reference Bulatović and Stankovski2012:343–7), which is considered as the beginning of the Middle Bronze Age in the central Balkans. Dates for the BH III horizon at Bubanj are broadly contemporaneous with the Moriş (or Mokrin) culture from the Hungarian plain (Table A5; O’Shea Reference O’Shea1996), and the Armenochori group in northern Greece (Table A5; Maniatis and Ziota Reference Маniatis and Ziota2011). These three cultural groups indeed share several stylistic and typological pottery traits (Garašanin Reference Garašanin1982; Bulatović and Stankovski Reference Bulatović and Stankovski2012).
CONCLUSION
These radiocarbon dates for the site of Bubanj, one of the key prehistoric locations in southeastern Europe, provide a breakthrough in the chronology of Eneolithic and Bronze Age periods in the central Balkans, as these correspond to the first absolute dates for certain prehistoric periods over this area.
The absolute dates for the BSK, K-K, and B-H III horizons all confirm the previous chronological estimates based on typological comparisons, and reinforce the importance of the site of Bubanj in the typological and cultural sequences of the Eneolithic and Early Bronze Age in the Balkans. The dates for the ČV III horizon are very significant. Not only do they provide the first dates for this culture in the central Balkans (and only the second for this entire culture), but they also indicate to a strict contemporaneity with the Boleraz phase of the Baden culture, thus pointing to the complex cultural interactions happening across central Europe during the late 4th millennium cal BC. Conversely, they cast doubt on the validity of the ČV I-ČV III sequence, although in this case inadequacies in the original dates for the ČV I culture cannot be ruled out.
All in all, this contribution calls for further sampling and dating of Eneolithic and Bronze Age sites across the Balkans. As we have shown here, significant results can be obtained through careful sampling combining robust stratigraphic and typological information, and a relatively limited number of dates, which makes the approach viable, especially in research areas where funding is often a limited resource.
ACKNOWLEDGMENTS
The majority of 14C dates and Vander Linden’s contributions are an output from the European Research Council project EUROFARM, funded under the European Union’s Seventh Framework Programme (FP/2007–2013; ERC Grant Agreement no. 313716). The sample date from structure 20 has been supported by a European Research Council Advanced Investigator Grant (295412), The Times of Their Lives (www.totl.eu), led by Alasdair Whittle and Alex Bayliss.
APPENDIX
Datasets: Tables A1–A5
Table A1 Dataset of all corresponding 14C dates for Bubanj-Salkuta-Krivodol complex.
Table A2 Dataset of 14C dates for Boleraz-Baden culture.
Table A3 Dataset of 14C dates for Černavoda III culture.
Table A4 Dataset of 14C dates for Koţofeni-Kostolac culture.
Table A5 Dataset of 14C dates for cultures of Early Bronze Age in central Balkans.
