Flint raw material transfers in the prehistoric Lower Danube Basin: An integrated analytical approach

Journal of Archaeological Science: Reports 5 (2016) 422–441

Contents lists available at ScienceDirect

Journal of Archaeological Science: Reports journal homepage: www.elsevier.com/locate/jasrep

Flint raw material transfers in the prehistoric Lower Danube Basin: An integrated analytical approach Maria Gurova a, Polina Andreeva b, Elitsa Stefanova b, Yavor Stefanov c, Miroslav Kočić d, Dušan Borić e,⁎ a

National Institute of Archaeology with Museum, Bulgarian Academy of Sciences, 2 Saborna Str., 1000 Sofia, Bulgaria Geological Institute, Bulgarian Academy of Sciences, Acad. G. Bonchev Str., bl. 24, 1113 Sofia, Bulgaria Faculty of Geology and Geography, Sofia University “St. Kl. Ohridski”, 15 Tsar Osvoboditel Blvd., 1504 Sofia, Bulgaria d Department of Anthropology, University of Pittsburgh, 3302 WWPH, Pittsburgh, PA 15260, USA e Department of Archaeology and Conservation, Cardiff University, Colum Drive, Cardiff CF10 3EU, UK b c

a r t i c l e

i n f o

Article history: Received 24 June 2015 Received in revised form 5 December 2015 Accepted 14 December 2015 Available online 29 December 2015 Keywords: Flint Outcrops Northern Bulgaria Northeastern Serbia Moesian Platform LA-ICP-MS analysis Petrographic thin-sections Palaeolithic Mesolithic Early Neolithic

a b s t r a c t The paper presents results of a research programme focused on the provenancing of flint raw materials used in the prehistory of the Lower Danube Basin of the Balkans. Field survey encompassed two adjacent regions connected by the Danube River. First, northern Bulgaria where rich flint-bearing Cretaceous deposits are known along with numerous Neolithic sites but with limited pre-Neolithic presence apart from several well-known Middle to Upper Palaeolithic sequences. Second, the Danube Gorges area on the southern, Serbian side of the river, characterized by relatively scarce deposits of flint, but with one of the best preserved concentrations of Mesolithic, transitional and Early Neolithic sites in the wider region of southeastern Europe. Focusing on both of the two selected regions allows one to follow diachronic dynamics in supply and circulation of local and non-local flint raw materials along the examined stretch of the Lower Danube Basin. In order to connect surveyed flint outcrops and different types of raw material used in archaeological contexts, an integrated approach was employed using both petrographic thin sections and LA-ICP-MS trace element chemical finger-printing analyses. © 2015 Elsevier Ltd. All rights reserved.

1. Introduction For all prehistoric periods, the study of flint raw material transfers offers archaeologists one of the best proxies for human mobility, attesting to networks of exchange, sometimes between distant communities, or indicating direct procurement over long distances (e.g., Belardi et al., 2015; Boulanger et al., 2015; Floss, 1994; de Grooth, 1997; Huckell et al., 2011; Lech, 1990, 1997; Nash et al., 2013; Pettitt et al., 2012; Speer, 2014; Zimmermann, 1995). Yet, depending on scholarly traditions, some of the European regions have seen more intense research than others. For instance, southeastern Europe remains an underexplored region regarding flint provenance studies. A notable exception to this general rule is the Lithotheca of the Hungarian National Museum in Budapest, Hungary (Bíro and Dobosi, 1991; Bíro et al., 2000), containing a rich collection of comparative specimens of flint raw

⁎ Corresponding author. E-mail addresses: [email protected] (M. Gurova), [email protected] (P. Andreeva), [email protected] (E. Stefanova), [email protected] (Y. Stefanov), [email protected] (M. Kočić), [email protected] (D. Borić).

http://dx.doi.org/10.1016/j.jasrep.2015.12.014 2352-409X/© 2015 Elsevier Ltd. All rights reserved.

materials from the Carpathian Basin and beyond, serving as an important point of reference. In order to fill this research gap, in 2011, two of us (MG and DB) initiated a programme of fieldwork and archaeometric analyses primarily focused on two adjacent regions of the Balkans — northern Bulgaria and northeastern Serbia (Fig. 1). These regions are connected by the Danube River and have given indications of possibly intense movements of people and objects along this major transitory axis in different prehistoric periods. In this paper, we present preliminary results of our study in identifying flint sources used by prehistoric communities of these areas by employing new fieldwork, thinsection petrographic analyses and LA-ICP-MS chemical finger-printing. A number of flint sources in Bulgaria are well-known and have previously been described in the archaeological and geological literature (e.g., Nachev and Kanchev, 1984; Nachev and Nachev, 1988). Four distinct types of flint in Bulgaria have been recognized: Hemus flint, Dobrudzha flint, Moesian flint and Rhodope flint (Nachev, 2009). Each type has a different geographical distribution, geological age and macroscopically diagnostic features (Fig. 1) (Gurova and Nachev, 2008: Fig. 5). While the perceived quality of stone raw materials used for knapping is not necessarily a precondition for the production of a desirable debitage (e.g. Archer and Braun, 2010; Eren et al., 2014), the mineralogical

M. Gurova et al. / Journal of Archaeological Science: Reports 5 (2016) 422–441

423

Fig. 1. Map of the Lower Danube region showing flint-bearing rocks in Bulgaria and eastern Serbia and geological/primary and secondary raw material outcrops (hexagons with names in italics) and archaeological (triangles with names in italics) sites. The dotted lines indicate boundaries between tectonic zones. Sources for flint-bearing rocks: for Bulgaria, Gurova and Nachev, 2008, Fig. 5; for Serbia, geological maps (Base map by Marko Milošević). 1. Conglomerates, sandstones, argillaceous schists, Dover (D); 2. Upper Jurassic limestones (Oxfordian a age) with siliceous concretions (Jox 3 ); 3. Low Cretaceous (Aptian age) limestones with siliceous concretions (К1); 4. Flysch: thin-bedded limestones, marls and shales (Valanginian and Hauterivian) (K1+2 ); 5. Marls, marly limestones and limestones with chert nodules (Hauterivian) (K21); 6. Upper Cretaceous chalk and chalk-like limestones (Campanian and Maastrichtian 1 age) with siliceous concretions (Кcp-m ); 7. Upper Cretaceous volcanogenous rocks (Coniacian, Santonian and Campanian ages) in Sredna Gora Zone (КCn-Cp ) — Sredna Gora atypical flint; 8. 2 2 Chalcedony veins in Oligocene volcanogenous rocks in Rhodope Zone (Pg3) — Rhodope atypical flint.

comparison of these four types distinguishes Ludogorie (called also Dobrudzha) flint as the most suitable material for knapping (judging by the homogeneity of the raw material and conchoidal fracture pattern) — the unique homogeneity and the size of the nodules permits core preparation and large laminar blanks debitage. The most significant accumulations of siliceous concretions occur on the Moesian Platform and adjacent parts of the Balkan Alpine Orogen. The main lithostratigraphic horizons in which these occur are of Lower Cretaceous (Aptian) and Upper Cretaceous (Coniacian, Campanian and Maastrichtian) age. Both series are represented on the Moesian Platform in northern Bulgaria. The most recent research on the topic has suggested that the area of flint-rich rocks along the Moesian Platform spreads across northern Bulgaria and that it might have been the main acquisition zone for regions farther to the west, i.e. the central and northern Balkans: The Upper Cretaceous flint-rich rocks formed three large areas of outcrops in North Bulgaria (the Moesian Platform and adjacent parts of the Balkan Alpine Orogen), from West to East as follows: the first one between Montana and Lovech, the second — between Pleven and Nikopol and the third — between Shumen and Devnya. In this big territory Moesian flint has large distribution and has formed big deposits. Throughout this big area the Moesian flint has similar features. Only in the Pleven–Nikopol region the Moesian flint is hosted in non-deformed rocks. That is why the flint from these outcrops has a better quality [for knapping purposes]. This fact and the convenient transport connection along the Danube River, determined the big outcrops on the Danube coast near Nikopol and Somovit as the most probable source of

flint raw materials for vast territories in Serbia and Romania (Gurova and Nachev, 2008: 34).

2. Archaeological background Questions about long-distance trade and acquisition of flint from the Moesian Platform in northern Bulgaria relate to two archaeological problems: first, our understanding of long-distance exchange/trade or direct procurement of raw materials in prehistory in general; and, second, the special place in this exchanges of the so-called high-quality yellow-waxy and white-spotted flint also known as “Balkan flint”, the lithic raw material abundantly used for the manufacture of chipped stone artefacts and considered one of the diagnostic features of the spread of Neolithic farming communities across the eastern and central Balkans and parts of the Carpathian Basin. Since the 1970s, a view emerged that this distinctive flint originated from the so-called “PreBalkan Platform” (Moesian Platform) in northern Bulgaria from where it was distributed to neighbouring regions across southeastern Europe (Kozłowski and Kozłowski, 1984; Voytek, 1987). The “Balkan flint” problem especially relates to the supra-regional Karanovo I–Starčevo– Körös–Criş taxonomic unit of the Early Neolithic in southeastern Europe. The question of “Balkan flint” provenancing is inherently linked to the Neolithization debate. In spite of decades of research on the origins and spread of the Neolithic in the Balkans, it has proved difficult to offer a firm identification of “Balkan flint” source(s) in the Moesian Platform and to explain the appearance of standardized (formal)

424

M. Gurova et al. / Journal of Archaeological Science: Reports 5 (2016) 422–441

toolkits made of this flint from ca. 6200 cal BC onwards across the wider Balkan area (e.g., Biagi and Starnini, 2013; Borić, 1999; Gurova, 2008, 2011, 2012; Voytek, 1987). The typological spectrum of the toolkits includes mainly blades with (bi)lateral semi-steep to steep retouch and sometimes pointed or rounded (end-scraperlike) ends. The blades are produced using indirect percussion

(punch) technique. Sickle inserts made on blades and with evidence of multiple posterior re-sharpening are also included in the toolkit. These toolkits were abundant throughout the “classic” Early Neolithic Karanovo I and II periods of the tell Karanovo sequence, as well as among the other key Neolithic sites in Bulgaria ca. 6000 to 5500 cal BC (Fig. 2).

Fig. 2. Early Neolithic artefacts made with “Balkan flint” raw material from Bulgaria and Serbia (Aria Babi): (1) Yabalkovo; (2) Rakitovo; (3) Slatina; (4) Dzhuljunitsa; (5) Kovachevo; (6) Aria Babi; (7) Kovachevo (sickle inserts); (8) Yabalkovo (sickle inserts); (9) Ohoden (Sources: 1, 8 after Gurova, 2012, Fig. 3; 2, 3 after Gurova, 2012, Fig. 4; 5, 7 after Gurova, 2012, Fig. 2; 4, 6, 9 after Gurova, 2012, Fig. 6).

M. Gurova et al. / Journal of Archaeological Science: Reports 5 (2016) 422–441

While the evidence of long distance transport of these flint raw materials is primarily associated with Neolithic and Chalcolithic/Copper Age periods in Bulgaria, there is a chronological gap between the Middle and Upper Palaeolithic sequences and the first Early Neolithic occupation of this territory. This prevents us from understanding diachronic changes in raw material preferences and distribution networks. Fortunately, the adjacent region of the Danube Gorges of northeastern Serbia and southeastern Romania, to the west along the Danube River, hosts a concentration of sites along the Danube banks with continuous occupation throughout the Epipalaeolithic, Mesolithic and up to the end of the Early/Middle Neolithic, which fills this gap (e.g., Bonsall, 2008; Borić, 2011). Building on this regional strength in the availability of flint assemblages for our study of the changing nature of flint raw material transfers in the prehistory of the Balkans, our project design encompassed focused fieldwork in northern Bulgaria and the southern bank of the Danube Gorges region along the Danube River in north-eastern Serbia in order to gain insights into the availability of primary and secondary flint deposits in each of these zones. It is expected that this type of study will allow one to match these deposits with macroscopically selected samples from sites dated to different prehistoric periods: from the Middle Palaeolithic through to the end of the Neolithic and the Copper Age. This is the first step in the planned long-term programme of research focusing on flint raw materials transfers in this part of southeastern Europe. The objectives of our research programme are thus three-fold: (1) to make a reliable distinction between different types of flint raw material found in Cretaceous and Jurassic limestones of northern Bulgaria and northeastern Serbia; (2) to securely identify different flint raw materials used by prehistoric communities in the Balkans by matching archaeological artefacts and materials from known geological deposits, and in this way subsequently test the reliability of macroscopic observations; (3) to reconstruct prehistoric raw material preferences, networks of supply and distribution and observe diachronic changes in the functioning of these networks. To address these objectives, in the course of the project we conducted fieldwork and subsequent archaeometric analyses, the results of which are presented below. 3. Material and methods 3.1. Geological survey and sampling Geological samples for our analyses of primary and secondary deposits were obtained in the course of field surveys conducted in 2011 and 2012. The goals of the surveys were to (a) precisely map known and new flint deposits on topographic and geological maps by recording their GPS coordinates; (b) provide geological descriptions of flint outcrops with a detailed photo documentation of the field conditions; (c) to identify possible archaeological workshops associated with flint outcrops; and (d) to assemble an exhaustive collection of raw material samples for further archaeometric analyses. In 2011, the survey covered the central and western parts of northern Bulgaria (Pleven, Vratsa and Vidin districts) (Figs. 3–4), where 30 flint deposits were discovered as well as six archaeological sites associated with the outcrops (Gurova et al., 2012a, 2012b, 2012c) (Fig. 1). The type of flint raw material found in north-western Bulgaria, i.e. Moesian flint sensu Nachev (2009), is widespread in Pleven and Vratsa districts. In this region, primary deposits consist of flint concretions, which are embedded within limestones of Late Cretaceous (Campanian-Maastrichtian) age (Gurova and Nachev, 2008; Nachev, 2009). Yolkichev (1986) refers to these sediments as the so-called Mezdra Siliceous-Carbonate Formation (Fig. 3a-c) and Rumyantsevo Limestone Formation (Fig. 3d). Secondary eluvium, diluvium (Fig. 3e-g) and (palaeo)alluvial (Fig. 3h) flint deposits in the area are formed due to the destruction and disintegration of the same limestones, containing flint concretions. The latter are often observed as flint pebbles and cobbles within slightly consolidated or

425

unconsolidated breccias and conglomerates. The most relevant sites with regard to the “Balkan flint” outcrops are as follows: 1. Mouselievo. Primary flint deposits are located south of the village of Mouselievo and belong to the Upper Cretaceous Mezdra Formation. Macroscopically, the raw material from Mouselievo corresponds to “Balkan flint” type of material. Significantly, at the foot of the hill a Middle Palaeolithic site and workshop for bifacial leafpoints was excavated in the 1970s and 1980s (Sirakova, 1980, 1990); 2. Zhernov. The location is found north of Mouselievo in the direction of Nikopol, with an exposed and eroded road cut. The site is extremely rich in a secondary deposit of flint concretions and is associated with artefacts (Fig. 5). In our view, a likely large workshop was located here that at present offers no diagnostic chronological markers. A Holocene date for this exploitation of secondary flint deposits, originated from destructed and disintegrated limestone of the Mezdra Siliceous-Carbonate Formation, is likely although no pottery fragments were found in the exposed section. 3. Ali Koch Baba (teke). The location is found farther to the north on the southern outskirts of the town of Nikopol. Primary deposits of flint concretions macroscopically correspond to “Balkan flint.” They occur as chalk-like limestones, belonging to the same Upper Cretaceous Mezdra Siliceous-Carbonate Formation. A series of artefacts were found on the slope but show no diagnostic features for a more precise chronological attribution. The site has been published by Biagi and Starnini (2010, 2011); for a critique see Gurova, 2012). 4. Danube. Southwest from Nikopol along the road to Pleven there are exposed road cuts that show geological sections with flint concretions similar to Ali Koch Baba. Some of these were documented. There were also numerous concentrations of secondarily deposited flint concretions on the Danube's bank near Nikopol. Survey in 2011 also included the Danube Gorges region and its hinterlands (Fig. 1). No systematic archaeometric analyses of flint raw materials used by Palaeolithic, Mesolithic and Neolithic populations from more than 20 known sites/layers in this region and their connection to possible flint outcrops have previously been undertaken. The Danube Gorges region is geologically and geomorphologically a complex area. Here, the Danube connects the Pannonian and Dacian basins and is composed of three smaller valleys and four gorges with distinct geologies (Vulcanescu, 1972; Marković-Marjanović, 1978). The first gorge (c. 400 m wide) starts at the entrance to the Danube Gorges region at Golubac. In this gorge, the Danube cuts Lower Jurassic and Cretaceous limestones, Paleozoic granites and Old Paleozoic metamorphic rocks. The floor of the first valley, Liubcova, is covered by Pliocene alluvial sediments. In the second, Lady's Whirlpool gorge the Danube cuts through sediments of greenish slate, Jurassic sandstones, and quartzporphyry cliffs (Paleozoic). This gorge is followed by the Donji Milanovac valley, filled with Miocene sediments. The third gorge, the Kazan (Cauldron), is very narrow (150–170 m wide). The Danube in this gorge cuts Paleozoic gabbro, Jurassic limestones and Old Paleozoic metamorphic rocks. The Orsava valley, following the Kazan gorge, is covered by Neogene sediments, especially pronounced on the left (Romanian) side of the Danube. The last valley in the Danube Gorges, called the Sip valley ends at the western edge of the Romanian plains. The Danube here cuts the Miroč mountain plateau built of Old Paleozoic rocks, JurassicCretaceous limestones and sandstones, and Sarmatian conglomerates. In the Danube Gorges, the most relevant for our discussion are primary deposits found in the Boljetinska River Canyon, some 15 km from the present-day town of Donji Milanovac (Serbia), with exposed profiles along the river (Fig. 6a) where one finds tectonic contact between Upper Jurassic and Lower Cretaceous deep-water (over 2 km deep in the sea) sediments. A number of samples for further analyses were taken here and at the nearby Pesača geological column (Table 1, Fig. 6; see below) (Vasić and Obradović, 1995). Further samples were taken at the Kremenac outcrop near the present-day city of Niš in southeastern Serbia (Fig. 1). Numerous finds from this location belong to

426

M. Gurova et al. / Journal of Archaeological Science: Reports 5 (2016) 422–441

Fig. 3. (a) Limestones of Mezdra Siliceous-Carbonate Formation (MzFm) containing flint concretions (black arrows), outcropping eastern of Gortalovo village, near to the road GortalovoTodorovo (primary deposits, Pleven district); (b) Primary deposits of irregular in shape dark grey to brown flint concretions from MzFm (Mouselievo village, Pleven district); (c) Ochre to brown in colour flint concretions (black arrows) within light grey limestones of MzFm (primary deposits, 2 km west from Nikopol and very close to the Danube river); (d) Dark grey flint concretions (black arrows) within limestones of Rumyantsevo Limestone Formation (primary deposits, dam lake of Bohot village, Pleven district); (e) (f) Diluvium deposits containing flint cobbles (black arrow) (secondary deposits, north-western of Zhernov village — Site 1, Pleven district); (g) Flint cobble from diluvium deposits (north-western of Zhernov village — Site 2, Pleven district); (h) Alluvium deposits in Osam river containing flint pebbles and cobbles (black arrows) and artefacts (white arrow) (secondary deposits, south of Mouselievo village) (Figure by Polina Andreeva).

knapped specimens that are at different stages of preparation, suggesting that this large outcrop was exploited in the course of the Palaeolithic (Kaluđerović, 1996; Šarić, 2011) but also possibly in later prehistoric periods.

In 2012, our survey extended to northeastern Bulgaria, i.e. the Ludogorie region (Russe and Razgrad districts). In this area, 15 secondary flint deposits were recorded, as well as five archaeological sites, three of which have already been known in the literature and from

M. Gurova et al. / Journal of Archaeological Science: Reports 5 (2016) 422–441

427

Fig. 4. (a) Irregular in shape dark grey to black flint concretions (yellow arrows) (primary deposits, lower levels of MzFm, eastern of Gorna Beshovitsa village, Vratsa district); (b) Small-sized bluish grey flint concretions (black arrows)(primary deposits, lower levels of MzFm, south-western of Reselets village, Pleven district); (c), (d) Large light grey to creamy grey concretions with predominantly oval shape (white arrows) (primary deposits, upper parts of MzFm western of Kunino village, near to the road Kunino-Kameno pole, Vratsa district); (e) Large oval flint concretions (primary deposits, upper parts of MzFm south-western of Kalen village, Vratsa district); (f) Irregular in shape flint concretion with variable colour and whitish spots (primary deposits, MzFm, eastern of Beglezh village, Pleven district); (g) Large flint concretion with irregular shape (primary deposits, MzFm, southern parts of Nikopol town near Ali Koch Baba, Pleven district); (h) Various in colour (beige, ochre, light brown, dark brown) flint cobbles with whitish spots and thick white cortex (secondary deposits, north-western of Zhernov village, Pleven district) (Figure by Polina Andreeva).

the Archaeological Map of Bulgaria (Gurova et al., 2013a, 2013b). In northeastern Bulgaria, flint raw material occurs predominantly in secondary eluvium, diluvium and (paleo)alluvial deposits. This flint is known as Dobrudzha or Ludogorie flint and is of Early Cretaceous (Aptian) age (Nachev, 2009; Nachev and Kanchev, 1984). Based on its

geological characterization, this flint is significantly different from the Moesian flint, and, it also shows a different pattern of use and distribution in prehistory (Andreeva et al., 2014). In total we collected samples from numerous outcrops, 15 of which are most relevant to the study (Table 1, Figs. 3, 4, 6), and 18 archaeological

428

M. Gurova et al. / Journal of Archaeological Science: Reports 5 (2016) 422–441

Fig. 5. Archaeological evidence (core and debitage) of flint workshop at the site of Zhernov with secondary deposits of flint pebbles and cobbles — Moesian flint (Upper Cretaceous, Mezdra formation) (Photos by Maria Gurova).

sites (Figs. 7–8, Table 1). The most significant outcrops of primary and secondary flint-bearing deposits were sampled and the specimens were collected directly from the outcrops. The goal was to sample a range of macroscopically different flint types and to establish a reference source collection for the studied regions. All flint samples were macroscopically described (colour, size, shape, texture, presence of cortex, etc.). Raw material concretions were taken for lithotheque collections, while a selection of flint raw materials as well as a series of representative archaeological artefacts were submitted to LA-ICP-MS analyses and micropetrographic study (Table 1).

3.2. Petrographic thin-sections and LA-ICP-MS analysis The main challenge consisted in combining laser ablation inductively coupled plasma mass spectrometry (hereafter LA-ICP-MS) analysis (chemical finger-printing) and petrographic study of thin sections (mineralogy, rock texture, limestone relics and fossil assemblages, relative proportion of siliceous and carbonate components, etc.) in order to distinguish between different flint types found in primary and secondary outcrops, to identify raw material provenance of the main raw materials used in prehistory and to test the hypothesis of the multiple

M. Gurova et al. / Journal of Archaeological Science: Reports 5 (2016) 422–441

429

Fig. 6. (a) Exposed geological profile of the Boljetinska River canyon with Upper Jurassic and Lower Cretaceous deposits, Danube Gorges, Serbia; (b), (c) location 554–554b, primary deposits of flint concretions in limestones from carbonate sludge and red marls of Upper Jurassic age (Tithonian) in the Boljetinska River canyon; (d) locations 553, 555, black and dark grey flint-bearing concretions in Lower Cretaceous (Berriasian) limestones of the Boljetinska River canyon; (e) location 556, grey Cretaceous flint concretions at the contact between Jurassic and Cretaceous limestones of the Boljetinska River canyon; (f) location 558, radiolarites in contact with Oxford limestones, Pesača, Danube Gorges, Serbia; (g) location 561, Upper Jurassic limestones (Tithonian), Pesača; (h) location 563, chalcedony rocks on the surface of Kremenac near the city of Niš, Serbia (Photos by Dragan Milovanović).

sources of the “Balkan flint” in northern Bulgaria (Moesian Platform). One of the main advantages in choosing LA-ICP-MS over petrographic thin-sections is that it is minimally destructive, which is important when working with archaeological materials. Previous inconclusive results when comparing thin-section analyses of “Balkan flint” raw materials from primary geological deposits and prehistoric artefacts from macroscopically similar materials by Ch. Nachev and M. Gurova (Gurova, 2008), led us to consider other means of identifying the source or sources of provenance for the widespread use of “Balkan flint” among Neolithic communities in Bulgaria and the neighbouring regions of southeastern Europe. A series of archaeological samples from Early Neolithic sites and geological

samples from outcrops of Moesian and Ludogorie flints have been analysed by C. Bonsall and colleagues using LA-ICP-MS and electron probe micro analysis (EPMA) (Bonsall et al., 2010). The combination of these techniques aimed to test the effectiveness of trace-element analysis as a tool for characterizing “Balkan flint”, enabled high precision quantitative chemical finger-printing with a high spatial resolution (up to 1 μm and 25 μm for EPMA and LA-ICP-MS, respectively). Initial results, while promising, were limited by a small sample size, preventing the provenance identification of artefacts with a high degree of certainty. In our study, we have significantly increased the sample of analysed specimens: 68 samples were analysed using the LA-ICP-MS technique while also 66 petrographic thin-sections were made and analysed.

430

M. Gurova et al. / Journal of Archaeological Science: Reports 5 (2016) 422–441

Table 1 Flint samples analysed with petrographic thin-sections and/or LA-ICP-MS. Locale/site name

Longitude

Latitude

Country

Elev. (m)

Geological strata/dating

Raw material samples Zhernov Ali Koch Baba teke (Nikopol) Danube Mouselievo Gortalovo Bohot Lake Kremenac (1/563/S1) Kremenac (2/563A/S2) Boljetinska River Canyon (DG 1/555/S3) Boljetinska River Canyon (DG 2/556/S4)

43° 39′ 58.3″N 43° 41′ 42.5″N 43° 42′ 06.6″N 43° 37′ 33.8″N 43° 19′ 02.6″N 43° 17′ 32.7″N 43° 22′ 58.4 N 43° 22′ 58.4 N 44° 32′ 15.9 N 44° 32′ 28.3 N

24° 51′ 28.4″E 24° 53′ 29.3″E 24° 51′ 32″E 24° 51′ 22.1″E 24° 34′ 27.2″E 24° 40′ 17.5″E 21° 52′ 40.0″E 21° 52′ 40.0″E 22° 1′ 59.0″E 22° 1′ 58.5″E

Bulgaria Bulgaria Bulgaria Bulgaria Bulgaria Bulgaria Serbia Serbia Serbia Serbia

49 70 24 96 207 305 417 417 104 121

Boljetinska River Canyon (DG 3/554/S5) Boljetinska River Canyon (DG 3/554A/S6) Boljetinska River Canyon (DG 3/554B/S7) Pesača (DG 4/558/S8) Pesača (DG 4/558A/S9) Pesača (DG 4/558B/S10) Pesača (DG 5/561/S11) Crni Vrh (DG 6/562/S12)

43° 32′ 10.2 N 43° 32′ 10.2 N 43° 32′ 10.2 N 44° 35′ 48.2 N 44° 35′ 48.2 N 44° 35′ 48.2 N 44° 34′ 23.6 N 44° 5′ 45.1 N

22° 1′ 54.4″E 22° 1′ 54.4″E 22° 1′ 54.4″E 22° 0′ 34.6″E 22° 0′ 34.6″E 22° 0′ 34.6″E 22° 1′ 16.0″E 21° 53′ 57.8″E

Serbia Serbia Serbia Serbia Serbia Serbia Serbia Serbia

84 84 84 114 114 114 117 380

Moesian Flint — K2 (Campanian–Maastrichtian) Moesian Flint — K2 (Campanian–Maastrichtian) Moesian Flint — K2 (Campanian–Maastrichtian) Moesian Flint — K2 (Campanian–Maastrichtian) Moesian Flint — K2 (Campanian–Maastrichtian) Moesian Flint — K2 (Campanian–Maastrichtian) Quaternary deposits Quaternary deposits Upper Cretaceous Lower Cretaceous limestone at the boundary with Jurassic limestones Upper Jurassic limestones (Tithonian) Upper Jurassic limestones (Tithonian) Upper Jurassic limestones (Tithonian) Upper Jurassic limestones (Oxford and Kimmeridge) Upper Jurassic limestones (Oxford and Kimmeridge) Upper Jurassic limestones (Oxford and Kimmeridge) Upper Jurassic limestones (Tithonian) Upper Jurassic limestones (Oxford and Kimmeridge), Geticum

Аrtefacts found at outcrops Zhernov Ali Koch Baba Mouselievo

43° 39′ 58.3 N 43° 41′ 43.1″N 43° 37′ 34.6″N

24° 51′ 28.4″E 24° 53′ 28.9″E 24° 51′ 16.7″E

Bulgaria Bulgaria Bulgaria

48 72 90

Holocene (undetermined period) Holocene (undetermined period) Holocene (undetermined period)

43° 28′ 06″N 42° 31′ 60″N 42° 30′ 51″N 42° 40′ 51″N 43° 08′ 29″N 43° 22′ 70″N 41° 31′ 67″N

23° 28′ 91″E 25° 28′ 01″E 25° 54′ 04″E 23° 22′ 35″E 25° 52′ 71″E 23° 43′ 48″E 25° 23′ 45″E

250 110 198 543 122 192 258

44° 32′ 44.0″N

21° 46′ 08.4″ E

Bulgaria Bulgaria Bulgaria Bulgaria Bulgaria Bulgaria Bulgaria Serbia Serbia Serbia Serbia

44° 36′ 40.0″N

22° 00′ 00.0″ E

Serbia

c. 60–70

Early Neolithic Early Neolithic Early Neolithic Early Neolithic Early Neolithic Early Neolithic Early Neolithic Neolithic Neolithic Neolithic Copper Age? Copper Age? Early Neolithic (S19)

44° 33′ 19.9″N

22° 01′ 38.7″ E

Serbia

c. 60–70

Mesolithic Early Neolithic

44° 32′ 48.3″N

22° 01′ 31.8″ E

Serbia

c. 258

Early Neolithic

Artefacts from archaeological sites Kovachevo (3 specimens) Yabalkovo (2 specimens) Karanovo (3 specimens) Slatina (2 specimens) Dzhuljunitsa (2 specimens) Ohoden (×2) Sedlare (×3) Vrelo-Marice (S13) Lojanik (S14 and S15) Crkvine (S16) Dubočka 1 S17 (Sq. 53/43, 4, 27/07/2011) Dubočka 1 S18 (Sq. 53/43, 1/2, 25/07/2011) Padina S19 (Tr. 1, block 2b, spit 3, inv. 311–314, bag 95, 06/09/1969) Padina S20 (inv. 818, Tr. 2, 2a, spit 4, 1970) Lepenski Vir S21 (inv. 1524, Block F/I, next to oven, July 1966) Aria Babi S23 (Tr. test 6, spit 1, 50 cm from the surface, 06/11/2004) Aria Babi S24 (Tr. 1/2005, bottom of the pit, April 2006) Pavlovac S25 (Context 299, x.2, sector V, sq. F-J/9–10 and F-G/8, 04/10/2011) Pavlovac S26 (context 36, ×6, sector IV, sq. P-T/26–30, 08/09/2011) Peštera Mare S27 (27/10/2004) Peštera Mare S28 (27/10/2004) Tabula Traiana Cave S29 (Sq. C, spit 7, ochre sediment, 02/11/2004) Tabula Traiana Cave S30 (context 206, x.4, Tr. 2/2008, 09/07/2008) Tabula Traiana Cave S31 (context 207/2, x.29, 08/07/2008) Tabula Traiana Cave S32 (context 211, x.11, 11/07/2008) Tabula Traiana Cave S33 (context 41, Tr. 1/2005, sq. B (3/20) 20/08/2005) Tabula Traiana Cave S37 (context 207/2, sq. 4/25, 5 mm, 08/07/2008) Vlasac S34 (Sq. A/17, spit 8) Vlasac S35 (Sq. b/13, spit 4) Vlasac S36 (Sq. b/18, spit 7)

c. 320

Early Neolithic 42° 29′ 07.4″N

21° 51′ 28.8″E

Serbia

c. 390

Early to Late Neolithic Early to Late Neolithic

44° 34′ 38.9″N

22° 01′ 13.2″E

Serbia

c. 250

44° 39′ 8.0″N

22° 18′ 50.9″E

Serbia

c. 92

Copper Age? Copper Age? Upper Palaeolithic Middle Palaeolithic Upper Palaeolithic Middle Palaeolithic Upper Palaeolithic Upper Palaeolithic

44° 32′ 08.1″N

22° 02′ 39.9″E

The LA-ICP-MS system used for analyses consists of 193 nm ArF excimer laser coupled with PE ELAN DRC-e ICP quadruple mass spectrometer based at the Geological Institute of the Bulgarian Academy of

Serbia

c. 62–70

Mesolithic Mesolithic Mesolithic

Science in Sofia, Bulgaria. For controlled ablation of the material, an energy density of about 10 J/cm2 on the sample surface and a laser pulse frequency of 10 Hz were used. Analyses were performed with 75 μm

M. Gurova et al. / Journal of Archaeological Science: Reports 5 (2016) 422–441

431

Fig. 7. Raw material samples (A) and archaeological artefacts from Early Neolithic Bulgarian sites (B) analysed by LA-ICP-MS and petrographic thin-sections. A — (1) Debovo; (2) Zhernov; (3) Mouselievo; (4) Ali Koch Baba (Nikopol); (5) Danube; (6) Gortalovo; (7) Bohot Lake; B — (1) Karanovo; (2) Yabalkovo; (3) Slatina; (4) Kovachevo; (5) Sedlare; (6) Dzhuljunitsa; (7) Ohoden (Photos by Maria Gurova).

beam diameter with ablation usually collecting the background for 40 s and signal from the flint for 50–60 s. External standardization by NIST 610 SRM glass provides relative element concentration ratios that

were transformed into absolute concentrations by internal standardization. We used SiO2 content (99 wt.%) as an internal standard, applying the SILLS software (Guillong et al., 2008) for data

432

M. Gurova et al. / Journal of Archaeological Science: Reports 5 (2016) 422–441

Fig. 8. LA-ICP-MS-analysed specimens from Serbian archaeological sites; (1), (4), (5) Tabula Traiana Cave, Early Upper Palaeolithic; (2), (3) Tabula Traiana Cave, Middle Palaeolithic; (6), (7), (8), Vlasac, Late Mesolithic; (9) Padina, Mesolithic; (10) Padina, Early/Middle Neolithic; (11), (12) Aria Babi, Early/Middle Neolithic; (13), (14) Pavlovac, Early/Middle Neolithic; (15) Lepenski Vir, Early/Middle Neolithic; (16) Peštera Mare, Copper Age; (17) Vrelo-Marice, Neolithic; (18) Lojanik, Neolithic; (19) Crkvine, Neolithic (Photos by Dušan Borić).

reduction. For samples that displayed high intensities of Al, K, Na and Ca during analyses we used that total major element oxides is 99 wt.% for data reduction. We measured 45 elements in total. Based on the macroscopic description and petrographic observations, 68 representative flint specimens (raw materials and artefacts) from Bulgaria and Serbia were analysed using LA-ICP-MS method in order to determine their chemical signatures. There were 32 samples of the so-called “Balkan flint”, which was macroscopically identified

(11 raw material specimens and 21 archaeological artefacts from 10 archaeological sites). Another 36 samples came from geological deposits and archaeological sites in Serbia (12 raw material specimens and 24 archaeological artefacts from 11 archaeological sites). Serbian samples were subjected to LA-ICP-MS procedure while the petrographic analyses of thin sections made on this material are not discussed in this paper. When conducting LA-ICP-MS, at least three measurements were taken per sample. In those instances when a flint sample was differently

M. Gurova et al. / Journal of Archaeological Science: Reports 5 (2016) 422–441

coloured, each coloured portion was identified separately. Complete results of all LA-ICP-MS measurements taken are given in Supplementary Information Appendix Table 1. 4. Results 4.1. Field observations and petrographic description 4.1.1. Bulgaria As mentioned above, the primary flint raw material deposits in northwestern Bulgaria occur within the sediments of Mezdra Siliceous-Carbonate Formation. It consists of thin- to mediumbedded or locally massive to indistinctly layered light grey to creamy grey limestones containing a variety of flint concretions of different shape and colour (Figs. 3a–c; 4a–g). Dark grey, bluish grey and black cherts (Fig. 4a–b) are observed only in the lower levels of the Mezdra Siliceous-Carbonate Formation, which is outcropped near Kunino, Gorna Beshovitsa, Malo Peshtene (Vratsa district). They have irregular shapes, homogeneous texture and variable size (mostly between 3 and 6 cm and only rarely up to 40 cm). White spots are sporadically noted. These flint concretions are predominantly small-sized, often cracked and are not suitable for the preparation of standardized toolkits. The upper parts of the Mezdra Siliceous-Carbonate Formation are characterized by the presence of flint raw material that has better knapping properties. This is represented by large (30 cm to 1 m) light grey to creamy grey concretions of predominantly oval shape, rare white spots and homogeneous or locally observed concentric laminated texture (Fig. 4c–e). These concretions occur in the area of the village of Kalen, and between the villages Kunino and Kameno pole (Vratsa district) where such flints are often observed concentrated in separate levels (Fig. 4c). In secondary deposits (e.g., the villages of Kameno pole, Kunino, Drashan, Reselets, Grivitsa, Mramoren), flint surfaces are often oxidized and are coloured in ochre, orange, brown and dark red. Although this flint type is suitable for tool preparation it is not found among collections of prehistoric artefacts from Bulgaria. What is perceived as good-quality flint raw material is likewise recorded in primary (the villages of Mouselievo, Evlogievo, Nikopol, Beglezh, Todorovo) and secondary (the villages of Gortalovo and Zhernov) deposits of Mezdra Siliceous-Carbonate Formation in the Pleven district. This flint type displays mostly irregular shapes (Fig. 4f–h) and is characterized by a variety of colours (beige, ochre, light brown and dark brown) and size (mainly from 10 to 40 cm). It is of heterogeneous texture, contains common whitish spots and is distinguished by white cortex (up to 2 cm in thickness). Sometimes the flint has zonal texture and is characterized by beige or brown central parts and dark grey periphery. Microscopic examination reveals that the whole of thin-section of this type of flint are composed of chalcedony-quartz groundmass being more than 50% in the framework (chert) with common preserved relics either from limestones and individual carbonate bioclasts: bryozoans, foraminifers, corals, gastropods, calcareous algae (Fig. 9a–f). It is most likely that replaced carbonate rocks were bioclastic limestones. Based on petrographic thin sections, the same type of flint is also found among artefacts from outcrops sites Zhernov, Mouselievo, Ali Koch Baba as well as from archaeological sites Ohoden (2 specimens analysed), Slatina (2), Yabalkovo (2), Dzhuljunitsa (2), Karanovo (3), Kovachevo (3), and Sedlare (3). The first three sites are related to the raw material exploitation (workshops) and were used in different periods during the Pleistocene (Mouselievo) and Holocene (most probably Zhernov and Ali Koch Baba). As for the Early Neolithic settlements of Ohoden, Slatina, Yabalkovo, Dzhuljunitsa, Karanovo, Kovachevo and Sedlare their assemblages contain “Balkan flint” in different proportions in the form of the so-called formal tools kits as well as debitage (Gurova, 2011, 2012; Gurova and Bonsall, 2014). A possible source of flint raw material and artefacts in the area of the village of Bohot (Pleven district) are the poorly outcropped sediments of

433

the Upper Cretaceous (Campanian) Rumyantsevo Limestone Formation (Fig. 3d). This unit is composed of white to light grey argillaceous limestones with glauconite, locally containing flint concretions. The latter are characterized by homogeneous texture, dark brown to dark grey colour and white or orange red cortex. Microscopically, this type of flint is difficult to distinguish from the flint of Mezdra Siliceous-Carbonate Formation and has a similar petrography (Fig. 9g, h). It also consists of chalcedony-quartz groundmass often containing relics from limestones and/or carbonate bioclasts (e.g. corals, bryozoans, gastropods, calcareous algae and foraminifers). 4.1.2. Serbia In the Boljetinska River canyon geological column in the Danube Gorges, at the deepest level, there are several types of carbonate rocks in Upper Jurassic limestones (Tithonian) that were sedimented in different ways: fine, nodular limestones from carbonate sludge and red marls with the evidence of replacement of carbonate materials with siliceous minerals, which created flint concretions in these limestones. There is a lot of siliceous organogenic material (radiolarias) in these rocks. The red colour of flints comes from a highly oxidized system. Marls are very rich in faunal remains (location 554/S5) (Fig. 6b-c). Some flint lenses here are of white colour, keeping the structure of the limestone that they replaced (location 554A/S6). Somewhat higher, on the same exposed geological column of the Boljetinska River, at the contact between Jurassic and Cretaceous limestones there are grey Cretaceous flint concretions (location 556/S4) (Fig. 6d). Black and dark grey flint-bearing concretions are found in Lower Cretaceous (Berriasian) limestones (location 533). Upper Cretaceous black flint-bearing rocks are also found here (location 555/S3) (Fig. 6e). Within the Lady's Whirlpool Gorge of the Danube Gorges area, geological samples were also taken at the Pesača geological column. Here one finds deep-water Upper Jurassic limestones (Oxford and Kimmeridge) with lenses of flint concretions that are of black to grey colour (locations 558/S8 and 558A/S9). At the same location one also finds radiolarites that are in contact with Oxford limestones (location 558B/S10) (Fig. 6f). At nearby location 559, samples are collected of layered calcedonic flints (b20% of radiolarias) and radiolarites (N 20% of radiolarias) from the deepest sediments of Jurassic limestones (Danubicum). At locations 560 and 561, at the boundary with Tithonian, a sample is taken where whole layers are replaced with flint concretions of reddish colour (Fig. 6g). All of the described flint types have relatively poor knapping qualities but in colour and texture are similar to the range of flint types used in the Middle Palaeolithic (Borić et al., 2012) and Mesolithic (Kozłowski and Kozłowski, 1982, 1984) of the region that could have come from both primary and secondary local flint deposits. Samples were also collected at the Kremenac outcrop location (563/ S1 and 563A/S2) near the present-day city of Niš in southeastern Serbia (Fig. 6h). Here surface blocks of chalcedony rocks (opals) with milky white patina and whitish cortex, occasionally over 50 cm in length, are found in Quaternary sediments. These are not organic but chemical sediments that are probably created under the influence of hydrothermal solutions (D. Milovanović, pers. comm., January 2015). These specimens show good knapping properties (Kaluđerović, 1996; Šarić, 2011). 4.2. LA-ICP-MS data Siliceous rocks or “silicates” consist mostly of minerals of the silica dioxide and contain up to 99 wt.% SiO2. Brandl et al. (2014) described three major possibilities to explain the elevated concentrations of trace elements in the silicates: (1) Trace elements that substitute for silicon atoms in the quartz structure. Some of these elements require charge compensation by additional ions; (2) Elements deposited in the pore spaces; (3) Mineral inclusions in the flint samples. Only a few ions such as Al3 +, Ga3 +, Fe3 +, Ge3 +, Ti4 + and P5 + are able to substitute for silicon atoms in the quartz structure. Some of these

434

M. Gurova et al. / Journal of Archaeological Science: Reports 5 (2016) 422–441

Fig. 9. Petrographic thin-sections of “Balkan flint” from Bulgarian outcrops and archaeological sites; (a) Partially replaced by silica minerals carbonate bioclast–bryozoan fragment (white arrow) represented within chalcedony-quartz groundmass (raw material, primary deposit, MzFm, eastern of Gortalovo village, near to the road Gortalovo-Todorovo, Pleven district), CPL; (b) Chalcedony-quartz groundmass containing replaced by silica minerals carbonate bioclast–coral fragments (yellow arrow) (artefact, south parts of Nikopol, near to the Ali Koch Baba, Pleven district), CPL; (c) Partially replaced by silica minerals coral fragment (yellow arrow) within chalcedony-quartz groundmass (artefact, Sedlare), CPL; (d) Chalcedony-quartz groundmass containing carbonate bioclast–bryozoan fragment (white arrow) (artefact, Ohoden, Vratsa district), CPL; (e) Limestone relic (yellow arrow) gradually passing into chert groundmass (raw material, secondary deposit, southern end of Pelishat village, Pleven district), CPL; (f) Limestone relic (yellow arrow) composed by micritic matrix and skeletal grains (artefact, Yabalkovo), CPL; (g) Coral fragment (yellow arrow) partly replaced by silica minerals (artefact, dam lake near Bohot village, Pleven district), CPL; (h) Limestone relic (yellow arrow) clearly passing into chert groundmass (artefact, dam lake near Bohot village, Pleven district), CPL. Note: CPL — cross-polarized light microphotographs) (Photos by Polina Andreeva, Yavor Stefanov).

elements are charge-balanced by additional ions (H+, Li+, Na+, K+, Cu+ and Ag+) in interstitial positions related to structural channels (Goetze et al., 2001).

LA-ICP-MS data show that the elements with the highest concentrations in all analysed samples are Al, K, Na, Ca, Mg, Fe, B and Ti. The measured values of these elements in “Balkan flint” specimens indicate that

M. Gurova et al. / Journal of Archaeological Science: Reports 5 (2016) 422–441

435

Fig. 10. Trace element concentrations in flint raw materials from north-western Bulgaria (Moesian flint) and artefacts from the so called “Balkan flint”(Figure by Elitsa Stefanova).

most of the analysed raw materials and artefacts have similar chemical compositions (Fig. 10). The raw materials from Bohot Lake 2, Mouselievo 3 and Ali Koch Baba have higher contents of Al, Na and K and lower concentrations of B in comparison to other raw materials and

all of the analysed artefacts show comparable concentrations of these elements. The artefact labelled as Sedlare 1 displays the lowest concentrations of Al, Na and K and the highest concentration of B of all the artefacts and raw materials that we have analysed, and cannot be compared to any

Fig. 11. Histogram showing B content in flint raw materials and artefacts from Serbia and Bulgaria. Most of the raw materials and artefacts from the Danube Gorges area in Serbia have B content lower than 50 ppm (Figure by Dušan Borić).

436

M. Gurova et al. / Journal of Archaeological Science: Reports 5 (2016) 422–441

of these raw materials. Often the concentrations of Al, K, Na, Mg and Ca vary significantly in one and the same sample due to the fact that the analysed flint materials are heterogeneous. Elevated concentrations of Al, K, Na and Mg could be explained by tiny clay mineral inclusions. Elevated concentrations of Ca are due to the carbonate mineral inclusions. This is also supported by petrographic observations that relict carbonate components are relatively common for this flint type. There is a strong positive correlation between Ca and Sr concentrations (Fig. 10). As these elements are typical for carbonates, we interpret the elevated concentrations of Ca and Sr as indications of a passing through of small inclusions of carbonates during LA-ICP-MS analyses. Serbian samples are divided into two groups based on their B content. First group has B content up to 50 ppm and the second one contains between 50 and 110 ppm B (Fig. 11). Most of the analysed raw materials and artefacts that come from the Danube Gorges area belong to the first group. In the second group there is only one raw material (S10 Danube Gorges 4/location 558B) and several artefacts (S18 Dubočka 1, S19 Padina, S21 Lepenski Vir, S23 Aria Babi, S24 Aria Babi, S26 Pavlovac, S33 Tabula Traiana Cave and S35 Vlasac). Although the raw material S10 shows similar B content to the artefacts from the group it differs significantly from them based on other trace element contents, with much higher concentrations of Al, K, Na, Mg, Ca and Sr than the rest of the artefacts in this group (Fig. 12), and can be excluded from this group as an outlier. Some of the raw materials from the first group (S4 Danube Gorges 2/location 556, S5 Danube Gorges 3/location 555 and S9 Danube Gorges 4/location 558A) have exceptionally high Ca and Sr content (Fig. 13) so we interpret them as partly silicified limestones, which corresponds with geological observations of these primary deposits (see above). Based on trace element concentrations in raw materials and artefacts from the first group (lower B content) we can distinguish three subtypes among the analysed samples from Serbia: (1) flints with low concentrations of trace elements. Most of the analysed raw materials and artefacts belong to this subtype (Fig. 13); (2) average content of trace elements (three raw materials: S3 Danube Gorges 1; S7

Danube Gorges 3; S8 Danube Gorges 4, and six artefacts: two artefacts from Peštera Mare S27 and S28; two artefacts from Tabula Traiana Cave S29 and S37 and two artefacts from Vlasac S34 and S36); and, (3) high concentrations of trace elements, represented by one raw material (S11 Danube Gorges 5/location 561) and two artefacts from Tabula Traiana Cave (S30 and S32) (Fig. 13). Specimens from the third group macroscopically (Fig. 8.2–3) make a logical grouping – artefacts are found in Middle Palaeolithic contexts (Borić et al., 2012) and suggest the exploitation of locally available quartzite raw materials by Middle Palaeolithic inhabitants of this region. One of the main tasks of the present study was to compare between collected raw materials from northern Bulgaria and artefacts made from “Balkan flint” from Bulgarian and Serbian archaeological sites based on their macroscopic description, petrographic observation, as well as their geochemical composition. In this way we hoped to provide firm evidence about the provenance of archaeologically used raw materials across the study region by tracing raw material transfers in order to understand distances and directions involved in exchange or direct procurement networks. We found that boron (B) is an element that allows discriminating between Bulgarian and Serbian samples (Fig. 11). Most of the Serbian raw materials from the Danube Gorges area and some artefacts have low B content whereas most of the analysed Bulgarian flint materials have higher B content. There is only one raw material specimen from the Danube Gorges area and a few artefacts from Serbia that contain B more than 50 ppm as already mentioned. Except for B this raw material specimen and artefacts also differ significantly in other trace element contents. A number of artefacts with high B content from Serbia could not be compared to any of the analysed raw materials from Serbia. Most of these artefacts with high B content (Padina S19, Lepenski Vir S21, Aria Babi S23 and S24, Pavlovac S26, Vlasac S35 and Tabula Traiana Cave S33) macroscopically are determined as “Balkan flint”. This is also supported by the geochemical data. Al, K, and Na for all of these artefacts are overlapping with the Bulgarian “Balkan flint” raw materials (Fig. 14). Clear discrimination of other artefacts and raw

Fig. 12. Trace elements in one raw material and artefacts from Serbia with B content lower than 50 ppm (Figure by Elitsa Stefanova).

M. Gurova et al. / Journal of Archaeological Science: Reports 5 (2016) 422–441

437

Fig. 13. Trace elements in raw materials and artefacts from Serbia with B content higher than 50 ppm (Figure by Elitsa Stefanova).

materials from Serbia from the raw materials and artefacts from Bulgaria is given not only by B content but also on the basis of other trace elements. 4.3. Statistical analysis LA-ICP-MS data The LA-ICP-MS readings were also submitted to statistical analysis. As at least three readings were taken per sample, and differently coloured parts of the same specimen were measured separately, it was necessary first to normalize the sample as the chemical readings were made on an unstandardized sample with results given in absolute mass (μg/g) (SI Appendix Table 1). In order to normalize the sample to be submitted to statistical analysis, the median was calculated for each sample based on all readings for each element keeping as separate those readings that were made on differently coloured portions of the same specimen. The obtained values were transformed from absolute numbers to percentiles in order to be compared and statistically analysed. The sample was also divided into four groups, which were submitted to statistical analysis both jointly and separately: (1) Bulgarian raw materials; (2) Bulgarian artefacts; (3) Serbian raw materials; and (4) Serbian artefacts. Each group was first analysed individually, and then all groups were analysed together. SPSS v.22 and R (GUI package RKWard) statistical software were used. Distribution of numbers in batches was not ideal, since the biggest problem that arose during the standardization of the sample was that the means had very large standard deviation. In order to visualize how large was the standard deviation at 95% confidence level SI Appendix Fig. 1 shows error bar plot of means with SD taken for B from raw samples. Here it is visible that some of the samples have very large standard

deviations (e.g., B1A, B15A, S22A) that are encompassing almost all of the values present in the sample. This is the reason why batches were transformed to medians of percentiles, in order to negate these vagaries of sampling. In the first batch of analyses on individual groups, we performed descriptive and exploratory statistical analysis. Principal component analysis (Jolliffe, 2002; Dunteman, 1989) was used in order to characterize the samples and reduce the number of components and to show which of the components eigenvalues account for the most variability. Here, a significant number of components is valued from each sample, and, component 1 will account for most variability within the sample (e.g., in the case of Bulgarian artefacts it accounted for 75% of cumulative variability, and in the case of the total sample for 67%). The number of components is delineated by the number of significant variables. Most of the significant variability in the Bulgarian sample was portrayed by two components, suggesting homogenous sample. This was further demonstrated by using proximity analysis. This analysis was done in order to identify relatively homogeneous groups of or variables based on selected characteristics, using an algorithm that starts with each group, or a variable in a separate cluster and combines clusters until only one is left. These dendrograms show more detailed clustering. Yet, with the Bulgarian sample, there are pronounced supra-clusters, with more developed micro-clustering. From the first couple of analysis, there was a pronounced heterogeneity in two sets: largest homogeneity was found among Bulgarian artefact and raw material samples (SI Appendix Figs. 2-3), with the largest heterogeneity found among the Serbian artefact samples (SI Appendix Figs. 4–5). For the complete sample of analysed flint, the dendrogram

438

M. Gurova et al. / Journal of Archaeological Science: Reports 5 (2016) 422–441

Fig. 14. Trace elements in artefacts from Serbia with high B content compared to raw materials from northwestern Bulgaria and “Balkan flint” artefacts (Figure by Elitsa Stefanova).

(SI Appendix, Fig. 6) indicates that there are two large supra groups of samples, one mainly consisting from artefacts, and the other from raw materials, with additional subdivisions within these two clusters. This is clearly visible on the unrotated scatterplot of discriminatory analysis (Fig. 15; the same figure with labels for each sample is given in SI Appendix Fig. 7) when components 1 and 2 were graphed. The same result was seen with graphing of components 2 and 3 since first and second components eigenvalue account for 67% of cumulative variability. It indicates that when using the first two components, which account for 67% of variability in the sample, there is a strong separation of the cluster consisting mostly of “Balkan flint” artefacts from both countries, and a lousily grouped cluster of raw materials with outliers. One very strongly separated cluster is that of four Serbian artefacts that are not coming out of the Neolithic sample. The results of the analyses indicate that while there is a moderately strong correlation between the samples taken out of the artefact category, suggesting a common source, there are only a limited number of raw material sources that correspond to these artefacts. These results suggest that macroscopical analysis and visual identification of yellow white-spotted “Balkan flint” definitely have merits and are useful in the course of archaeological investigation. From the statistical standpoint, analysis would show much clearer results if samples were not too heterogeneous in their source, but more importantly, in their sampling values. While there was a consistently

moderate strength of statement present during the analysis, there were some confidence results that suggested vagaries of sampling. This might be partly due to the fact that during the normalization of sampling, medians from some samples were actually somewhat misleading as vast differences between two readings from the same artefact (probably caused by inclusions) led to some of the median values to be elevated, or decreased from the actual values. This problem is evident in some of the instances when two readings were taken from differently coloured parts of the same sample that do not cluster together (11Gortalovo 2 and 14-Bohot Lake 2). However, on four out of six samples with two readings on differently coloured parts of the same sample these different readings do cluster closely together (23-Ali Koch Baba, S7-Danube Gorges 3, S11-Danube Gorges 5, and S25-Pavlovac). 5. Discussion and conclusions In the present study we combined two analytical methods (petrographic microscopy and LA-ICP-MS) in order to compare raw materials and artefacts from Bulgaria and Serbia. Petrographic observations and the geochemical results indicate a potential for differentiation between “Balkan flint” raw materials and artefacts from Bulgaria and other raw materials and artefacts found in Serbia separated by several hundred kilometres. While we find the potential of LA-ICP-MS analyses highly

M. Gurova et al. / Journal of Archaeological Science: Reports 5 (2016) 422–441

Fig. 15. Scatterplot of all samples. Red triangle: Bulgaria, raw materials; black triangle: Bulgaria, artefacts; cross: Serbia, raw materials; square: Serbia, artefacts (SPSS v.22 and AutoCAD 2015 by Miroslav Kočić).

promising, allowing us to draw robust conclusions on observed patterns, it should be noted that statistical analyses identified some outliers from the expected groupings. This relates to vagaries of sampling likely caused by unavoidable inclusions that skewed some of the median values derived from several readings made on the same sample. In the future, this problem could be addressed by increasing the number of laser ablasion readings per sample, and in this way achieving a higher confidence that the median value is representative of the underlying geochemical signature. The studied raw materials from the Pleven district (the villages of Mouselievo, Evlogievo, Nikopol, Todorovo, Beglezh, Gortalovo and Zhernov) and artefacts from Bulgarian Early Neolithic sites (Ohoden, Slatina, Yabalkovo, Dzhuljunitsa, Karanovo, Kovachevo and Sedlare) are characterized by very similar macroscopic and micropetrographic features. These data allow us to conclude that the main source of flint raw material for tool preparation during the Early Neolithic originated from the Upper Cretaceous sediments of Mezdra Siliceous–Carbonate Formation. As mentioned above, a possible source of flint raw material and artefacts in the area of the village of Bohot (Pleven district) could also be the Upper Cretaceous (Campanian) Rumyantsevo Limestone Formation. There are a few specimens among the analysed artefacts from Serbian sites that show macroscopic similarities to the “Balkan flint” raw materials found in Bulgaria and artefacts found at Bulgarian Early Neolithic sites. Trace element contents of these Serbian specimens overlap with those from Bulgarian outcrops and analysed archaeological sites. These artefacts are also macroscopically similar to a variety of white-spotted flint from Bulgaria that are mostly of yellow colour but can also include white-spotted specimens of greyish colour (Vlasac S35). One of these artefacts from the Early to Late Neolithic site of Pavlovac (S26) is around 300 km away from the nearest outcrop of “Balkan flint” in Bulgaria whereas other specimens from the sites in the Serbian area of the Danube Gorges (Lepenski Vir, Padina, Vlasac, Aria Babi and Tabula Traiana Cave) are at the distance of 250 km from Bulgarian sources as the crow flies, which are located downstream the

439

Danube route that might have facilitated easy access to these flint outcrops. Most of these specimens come from Early Neolithic contexts at the sites of Lepenski Vir, Aria Babi and Padina (Borić, 2011), but it is significant that one out of the three analysed artefacts from the Late Mesolithic levels of Vlasac (S35) might have also come from the Bulgarian outcrops and that another artefact (S33) found in early Upper Palaeolithic levels of the site of Tabula Traiana Cave (Borić et al., 2012) also exhibits geochemical properties that group it with the examined Bulgarian raw materials. At face value, this may suggest that while the intensity of long distance raw material transfers in the wider central and eastern Balkan area very likely increased in the Early Neolithic period regarding the consumption of technologically as well as aesthetically/symbolically desired raw materials from northern Bulgarian outcrops, the route that links the Danube Gorges with Bulgarian outcrops along the Danube River might have been utilized from the earliest phases of the Upper Palaeolithic if not earlier. More robust samples of analysed material in the future are needed in order to confirm these preliminary results. The likely change in the Early Neolithic can be related to the aesthetics of colour of these honey-yellow white-spotted flints as well as their knapping properties in the ability to produce longer blanks and blades as one of the main technological features of Early Neolithic assemblages in the Balkans. There are no raw materials from among the analysed Serbian flintbearing deposits that are comparable to the so-called “Balkan flints” either based on their petrography or geochemistry. Bulgarian “Balkan flint” artefacts have petrographic features similar to analysed raw materials from northwestern Bulgaria. These observations combined with the geochemical data suggest that flint outcrops at Gortalovo, Bohot Lake, Zhernov and various locations along the Danube at Nikopol are likely sources for the “Balkan flint” artefacts found at a representative sample of Early Neolithic sites from across the territory of Bulgaria, Early Neolithic sites in the area of the Danube Gorges between Serbia and Romania as well as farther afield in southern Serbia. It is very likely that elsewhere across the central and northern Balkans (e.g., many Early Neolithic sites in northern Serbia, southern Hungary and Romania linked by the Danube River and its tributaries) where one finds macroscopically distinct examples of “Balkan flint” (Biagi and Starnini, 2013; Bogosavljević and Starović, 2013; Borić, 1999), these specimens also came from northern Bulgarian outcrops, but this needs to be tested and demonstrated in the future. As our study showed, one should not expect that this pattern of raw material transfers to have been confined to the Early Neolithic only and it is very likely that this communication route and knowledge about abundant flint outcrops in northern Bulgaria is of significant antiquity. There is evidence of “Balkan flint” use in Gravettian and Epigravettian sequence of Temnata Cave, where this raw material was described as “wax-yellow BG-T-F4” flint from an unknown source (Pawlikowski, 1992, 286; see Gurova and Bonsall, 2014, 110–111, Fig. 3). It has been stressed before that in the course of the Palaeolithic, the Danube route might have been of key importance for the spread of modern humans into Europe (e.g., Conard and Bolus, 2003), with some key sites along this route in the northern Bulgarian hinterland. It should be noted, however, that there might have been various reasons for diachronic oscillations in the use of this raw material. For instance, it seems it was only sporadically used by the forager communities found in the Danube Gorges area throughout the Mesolithic (e.g., Kozłowski and Kozłowski, 1982; Mihailović, 2004; Radovanović, 1981). This pattern emerges despite the fact that one would expect that at this distance forager groups with embedded provisioning strategies (Binford, 1979) within well organized logistical forays (perhaps by specialized task groups) would have exploited the abundant outcrops of good-quality “Balkan flint” had they known about them. It might have been the nature of largely sedentary existence of forager-fisher communities in the Danube Gorges area that prevented frequent long distance movements outside the maximal band territory, the radius of which is normally estimated at around 125 km (Borić and Cristiani, in press; cf.

440

M. Gurova et al. / Journal of Archaeological Science: Reports 5 (2016) 422–441

Whallon, 2006). Similarly, “Balkan flint” use and distribution among Late Neolithic Bulgarian sites (the second half of the sixth millennium BC) shows a striking decline, and, during this period, local raw materials, which can be characterized as having poorer knapping properties compared to “Balkan flint”, started being used in most of the Late Neolithic settlements (Gurova, 2011). One of the many tasks that remain for future research is to precisely quantify the presence of “Balkan flint” raw material in various lithic assemblages from different sites and periods in the wider region of southeastern Europe, along with a precise characterization and quantification of the “Blakan flint” debitage at each site. In this way, it would be possible to provide more accurate estimates as to the nature and mechanisms involved in raw material transfers. However, it remains an imperative that instead of only guessing distances involved in raw material transfers, future follow-up studies expand the geographical territory under investigation and significantly enlarge the sample of studied specimens, thus allowing us to collect even more robust and widespread comparisons. This can be achieved by employing the suggested integrated approach that pairs and cross-references all currently available analytical methodologies in the study of flint raw material provenance. Acknowledgements We acknowledge the support received for the project through the High-Risk in Archaeology Program of the America for Bulgaria Foundation (11HRAR1) and the American Research Centre (ARCS), Sofia. The authors are grateful to Stefanka Ivanova (Sofia) and Maciej Pawlikowski (Kraków) for their help and fruitful collaboration during the fieldwork in Bulgaria and to Dragan Milovanović and Nebojša Vasić (University of Belgrade, Serbia) for their invaluable expertise regarding the geology of the Danube Gorges area in Serbia. We thank Vera BogosavljevićPetrović (National Museum in Belgrade) for the samples from VreloMarice, Lojanik and Crkvine. For comments on earlier versions of this paper we thank Simon Chenery (British Geological Survey, Nottingham), Dragan Milovanović, and Nebojša Vasić. We are also grateful to Emanuela Cristiani for her help with editing some of the images in this article and Marko Milošević (Geographical Institute “Jovan Cvijić”, Belgrade) for the base map in Fig. 1. Any remaining errors are the sole responsibility of the authors. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.jasrep.2015.12.014. References Andreeva, P., Stefanova, S., Gurova, M., 2014. Chert raw materials and artefacts from NE Bulgaria: a combined petrographic and LA-ICP-MS study. J. Lithic Stud. 1 (2), 25–45. Archer, W., Braun, D.R., 2010. Variability in bifacial technology at Elandsfontein, Western Cape, South Africa: a geometric morphometric approach. J. Archaeol. Sci. 37 (1), 201–209. Belardi, J.B., Cassiodoro, G., Goñi, R., Glascock, M.D., Súnico, A., 2015. Siltstone from Southern Patagonia: its source and archaeological artifact distribution in Santa Cruz Province, Argentina. Geoarchaeology 30 (3), 223–237. Biagi, P., Starnini, E., 2010. A source in Bulgaria for Early Neolithic ‘Balkan flint’. Antiquity 84 (325) (Project Gallery (http://www.antiquity.ac.uk/projgall/biagi325/). Biagi, P., Starnini, E., 2011. First discovery of Balkan flint sources and workshops along the course of the Danube river in Bulgaria. In: Tomičić, Z. (Ed.), Zbornik radova posvećenihKorneliji Minichreiter. Institut za arheologiju, Zagreb, pp. 69–83. Biagi, P., Starnini, E., 2013. Pre-Balkan platform flint in the Early Neolithic sites of the Carpathian Basin: its occurrence and significance. In: Anders, A., Kulcsár (Eds.), Moments in Time. Papers Presented to Pál Raczky on His 60th Birthday. L'Harmattan, Budapest, pp. 47–60. Binford, L., 1979. Organization and Formation Processes: Looking at Curated Technology. J. Anthropol. Res. 35 (3), 255–273. Bíro, K., Dobosi, T.V., 1991. Lithotheca. Comparative Raw Material Collection of the Hungarian National Museum. Hungarian National Museum, Budapest. Bíro, K., Dobosi, T.V., Schleder, Z.S., 2000. Lithotheca. Comparative Raw Material Collection of the Hungarian National Museum 1990–1997. Hungarian National Museum, Budapest.

Bogosavljević, V., Starović, A., 2013. Balkan flint from the Early Neolithic collection at the national museum in Belgrade. ZborMuzBeograd 21 (1), 77–108 (In Serbian with English abstract). Bonsall, C., 2008. The Mesolithic of the Iron Gates. In: Bailey, G., Spikins, P. (Eds.), Mesolithic Europe. Cambridge University Press, Cambridge, pp. 238–279. Bonsall, C., Gurova, M., Hayward, C., Nachev, C., Pearce, N.J.G., 2010. Characterization of ‘Balkan flint’ artefacts from Bulgaria and the Iron Gates using LA-ICP-MS and EPMA. Int. Stud. 22–23, 9–18. Borić, D., 1999. Places that created time in the Danube Gorges and beyond, c. 9000–5500 cal BC. Doc. Praehist. 26, 47–70. Borić, D., 2011. Adaptations and transformations of the Danube Gorges foragers (c. 13,000–5500 cal BC): an overview. In: Krauß, R. (Ed.), Beginnings — New Research in the Appearance of the Neolithic Between Northwest Anatolia and the Carpathian Basin. Verlag Marie Leidorf GmbH, Rahden/Westf., pp. 157–203. Borić, D., Cristiani, E., 2016. Social networks and connectivity among the Palaeolithic and Mesolithic foragers of the Balkans and Italy. In: Raiko, K., Floss, H. (Eds.), Southeast Europe before Neolithisation (Proceedings of the Workshop Held in Tübingen, Germany, May 9th, 2014). Verlag Marie Leidorf GmbH, Rahden/Westf, pp. 73–112 (in press). Borić, D., Dimitrijević, V., White, D., Lane, C., French, C.A.I., Cristiani, E., 2012. Modern human settling of the ‘Danube Corridor’: the Middle to Upper Palaeolithic site of Tabula Traiana Cave in the Danube Gorges (Serbia). Antiquity 334 (86) (Project Gallery) http://antiquity.ac.uk/projgall/boric334/. Boulanger, M.T., Buchanan, B., O'Brien, M.J., Redmond, B.G., Glascock, M.D., Eren, M.I., 2015. Neutron activation analysis of 12,900-year-old stone artifacts confirms 450– 510+ km Clovis tool-stone acquisition at Paleo Crossing (33ME274), northeast Ohio, USA. J. Archaeol. Sci. 53, 550–558. Brandl, M., Hauzenberger, C., Postl, W., Martinez, M., Filzmoser, P., Trnka, G., 2014. Radiolarite studies at Krems-Wachtberg (Lower Austria): Northern Alpine versus Carpathian lithic resources. Quat. Int. 351, 146–162. Conard, N.J., Bolus, M., 2003. Radiocarbon dating the appearance of modern humans and timing of cultural innovations in Europe: new results and new challenges. J. Hum. Evol. 44, 331–371. de Grooth, M.E.T., 1997. Social and economic interpretations of the chert procurement strategies of the Bandkeramik settlement at Hienheim, Bavaria. Analecta Praehist. Leiden. 29, 91–98. Dunteman, G., 1989. Principal Components Analysis (Quantitative Applications in the Social Sciences, Paper 69). Sage, Newbury Park, CA. Eren, M.I., Roos, C.I., Story, B.A., von Cramon-Taubadel, N., Lycett, S.J., 2014. The role of raw material differences in stone tool shape variation: an experimental assessment. J. Archaeol. Sci. 49, 472–487. Floss, H., 1994. Rohmaterialversorgung im Paläolithikum des Mittelrheingebietes. RömischGermanisches Zentralmuseum, Forschungsinstitut für Vor- und Frühgeschichte, Monograph 21. Dr. Rudolf Habelt, Bonn. Goetze, J., Ploetze, M., Habermann, D., 2001. Origin, characteristics and practical applications of the cathodoluminescence (CL) of quartz: a review. Mineral. Petrol. 71, 225–250. Guillong, M., Meier, D.L., Allan, M.M., Heinrich, C.A., Yardley, B.W.D., 2008. SILLS: A MATLAB-Based Program for the Reduction of Laser Ablation ICP-MS Data of Homogeneous Materials and Inclusions: Mineralogical Association of Canada Short Course40, pp. 328–333. Gurova, M., 2008. Towards an understanding of Early Neolithic populations: a flint perspective from Bulgaria. Doc. Praehist. 35, 111–129. Gurova, M., 2011. Prehistoric flint assemblages from Bulgaria: a raw material perspective. In: Niagu, M. (Ed.), East and West. Culture and civilization at the Lower Danube. Editura DAIM, Călăraşi, pp. 96–115. Gurova, M., 2012. ‘Balkan Flint’ — fiction and/or trajectory to Neolithization: Evidence from Bulgaria. Be-JA 1, 15–49 (www.be-ja.org). Gurova, M., Bonsall, C., 2014. Lithic studies: an alternative approach to neolithization. Be-JA 4.2, 107–135 (www.be-ja.org). Gurova, M., Nachev, C., 2008. Formal Early Neolithic flint toolkits: archaeological and sedimеntological aspects. In: Kostov, R.I., Gaydarska, B., Gurova, M. (Eds.), Geoarchaeology and Archaeomineralogy. Proceedings of the International Conference, 29–30 October 2008, Sv. Ivan Rilski, Sofia, pp. 29–35. Gurova, M., Ivanova, S., Andreeva, P., Pawlikowski, M., 2012a. Survey for archaeological settlements in Pleven district. Archaeological Discoveries and Excavations in 2011, pp. 510–513 (In Bulgarian). Gurova, M., Ivanova, S., Andreeva, P., Pawlikowski, M., 2012b. Survey for archaeological settlements in Vidin district. Archaeological Discoveries and Excavations in 2011, pp. 513–515 (In Bulgarian). Gurova, M., Ivanova, S., Andreeva, P., Pawlikowski, M., 2012c. Survey for archaeological settlements in Vratsa district. Archaeological Discoveries and Excavations in 2011, pp. 515–517 (In Bulgarian). Gurova, M., Ivanova, S., Andreeva, P., Kiselinov, H., 2013a. Survey for archaeological settlements in Russe district. Archaeological Discoveries and Excavations in 2012, pp. 525–527 (In Bulgarian). Gurova, M., Ivanova, S., Andreeva, P., Kiselinov, H., Mateva, B., Dilov, D., 2013b. Survey for archaeological settlements in Razgrad district. Archaeological Discoveries and Excavations in 2012, pp. 527–530 (In Bulgarian). Huckell, B.B., Kilby, J.D., Boulanger, M.T., Glascock, M.D., 2011. Sentinel Butte: neutron activation analysis of White River Group chert from a primary source and artifacts from a Clovis cache in North Dakota, USA. J. Archaeol. Sci. 38 (5), 965–976. Jolliffe, I., 2002. Principal Component Analysis. second ed. Springer, New York. Kaluđerović, Z., 1996. Kremenac kod Niša, Paleolitsko nalazište, hronika iskopavanja. Starinar 47, 289–290. Kozłowski, J.K., Kozłowski, S.K., 1982. Lithic industries from the multi-layer Mesolithic site, Vlasac, in Yugoslavia. In: Kozłowski, J.K. (Ed.), Origin of the chipped stone

M. Gurova et al. / Journal of Archaeological Science: Reports 5 (2016) 422–441 industries of the early farming cultures in the Balkans (Prace archeologiczne 33). Panstwowe Wydawnictwo Naukowem, Warszawa-Krakow, pp. 11–109. Kozłowski, J.K., Kozłowski, S., 1984. Chipped stone industries from Lepenski Vir. Preist. Alpina 19, 259–293. Lech, J., 1990. The organization of siliceous rock supplies to the Danubian early farming communities (LBK): Central European examples. In: Cahen, D., Otte, M. (Eds.), Rubané et Cardial. Études et Recherches Archéologiques de l'Université de Liège, Liège, pp. 51–59. Lech, J., 1997. Remarks on prehistoric flint mining and flint supply in European archaeology. In: Ramos-Millán, A., Bastille, M.A. (Eds.), Siliceous Rocks and Culture. Universidad de Granada, Granada, pp. 611–637. Marković-Marjanović, J., 1978. Geologija i stratigrafija. In: Garašanin, M. (Ed.), Vlasac – mezolitsko naselje u Djerdapu II. Srpska akademija nauka i umetnosti, Beograd, pp. 11–27. Mihailović, D., 2004. Chipped stone industry from horizons A and B at the site of Padina in the Iron Gates. In: Late Foragers and Early Farmers of the Lepenski Vir – Schela Cladovei Culture in the Iron Gates Gorges. A metamorphosis of technologies or Acculturations (Acts of the XIVth UISPP Congress, Liege 2001) (BAR Int. Ser. 1302). Archaeopress, Oxford, pp. 61–68. Nachev, C., 2009. The main flint types in Bulgaria as raw material for artefacts. Int. Stud. 20–21, 7–21 (in Bulgarian with English summary). Nachev, I., Kanchev, К., 1984. Aptian and Quaternary flint in north-east Bulgaria. III Seminar on Petroarchaeology (Plovdiv 27–30 of August, 1984, Bulgaria). Institute of Archaeology and Museum-BAS, Plovdiv University, Plovdiv, pp. 65–82. Nachev, I., Nachev, C., 1988. Distribution and evolution of siliceous rocks in Bulgaria. In: Hein, J., Obradovic, J. (Eds.), Siliceous Deposits of the Tethys and Pacific Regions. Springer-Verlag, New York, pp. 81–92. Nash, D.J., Coulson, S., Staurset, S., Ullyott, J.S., Babutsi, M., Hopkinson, L., Smith, M.P., 2013. Provenancing of silcrete raw materials indicates long-distance transport to Tsodilo Hills, Botswana, during the Middle Stone Age. J. Hum. Evol. 64 (4), 280–288. Pawlikowski, M., 1992. The origin of lithic raw materials. In: Ginter, B., Kozłowski, J.K., Laville, H. (Eds.), Temnata Cave. Excavations in Karlukovo Karst Area, Bulgaria (Vol. 1, Part 1). Jagellonian University Press, Kraków, pp. 241–286.

441

Pettitt, P., Rockman, M., Chenery, S., 2012. The British Final Magdalenian: society, settlement and raw material movements revealed through LA-ICP-MS trace element analysis of diagnostic artefacts. Quat. Int. 272–273, 275–287. Radovanović, I., 1981. Ranoholocenska kremena industrija sa lokaliteta Padina u Djerdapu. Arheološki institute, Beograd. Šarić, J., 2011. Lower Palaeolithic site of Kremenac near the village Rujnik (Serbia). Starinar 61, 7–31. Sirakova, S., 1980. Srednopaleolitni Listovidni Dvustranno Obraboteni Orudia v Bulgaria I Tehnite Paraleli v Jugoiztochna i Tsentralna Evropa. Unpublished PhD Dissertation, Sofia. Sirakova, S., 1990. The leaf-points of Muselievo. In: Kozłowski, J.K. (Ed.), Feuilles de pierre des industries à pointes foliacées du Paléolithique supérieure Européenne (ERAUL 42). Université de Liège, Liège, pp. 63–79. Speer, C.A., 2014. LA-ICP-MS analysis of Clovis period projectile points from the Gault Site. J. Archaeol. Sci. 52, 1–11. Vasić, N., Obradović, J., 1995. Sedimentology of the shallow-water sediments from the Pesača unit, eastern Serbia, Yugoslavia. XV Congress of the KBGA, Special Publication of the Society of Greece 4/1. Voytek, B., 1987. Analysis of lithic raw materials from sites in eastern Yugoslavia. In: Bíro, K. (Ed.), Papers for the First International Conference on Prehistoric Flint Mining and Lithic Raw Material Identification in the Carpathian Basin. Budapest-Sümeg, Budapest, pp. 287–295. Vulcanescu, R. (Ed.), 1972. The Iron Gate Complex Atlas. The Academy of the Socialist Republic of Romania, Bucharest. Whallon, R., 2006. Social networks and information: non-“utilitarian” mobility among hunter-gatherers. J. Anthropol. Archaeol. 25, 259–270. Yolkichev, N.A., 1986. Lithostratigraphic units related to Upper Cretaceous series in Northern and Central Fore-Balkan. Rev. Bulgar. Geol. Soc. 37 (3), 49–61 (In Bulgarian). Zimmermann, A., 1995. Austauschsysteme von Silexartefakten in der Bandkeramik Mitteleuropas. Habelt, Bonn.