Kulbulak is an open-air site (41°00′31″N–70°00′22″E), c. 1040m asl, at the foot of the Chatkal range, on the western edge of the Tian-Shan mountains, a few kilometres from the town of Angren (Akhangaran valley, Tashkent region, Uzbekistan; Figure 1).
From 1962 until the 1990s, several excavation campaigns have been carried out in Kulbulak (Kasymov 1972; Kasymov & Grechkina 1994; Anisutkin et al. 1995). An area of around 600m2 in total has been excavated and one of the trenches reached a maximum depth of 19m below ground surface (on 3m2). According to Kasymov & Grechkina (1994), the stratigraphy of Kulbulak corresponds to a long sequence that spans the Lower to Upper Pleistocene; it contains 49 archaeological horizons corresponding (from bottom to top) to 22 'Acheulean' levels, 24 Mousterian levels and three Upper Palaeolithic levels. During these excavations, more than 70 000 artefacts (without systematic sieving) were recovered.
While several scholars have pointed out the importance of Kulbulak (e.g. Ranov 1988; Vishnyatsky 1999), they also stressed the weaknesses and poor quality of the available information, making it difficult to verify the chronological and cultural interpretations proposed by the excavators. In 2007, an international collaboration was established to undertake new excavations at Kulbulak. At present, three campaigns have been successfully carried out. This paper reports on the preliminary results of these new excavations.
The excavation surface corresponds to 36m2, in three different areas, and two stratigraphic profiles have been cleaned on the sides of Kasymov's former trench (corresponding to the upper part of his 19m-deep trench). The current stratigraphy is 13m deep and consists of 27 geological layers (Figure 2), some of which contain archaeological material (layers 2.1, 2.2, 3 & 12–18).
At present, the chronostratigraphy is based on two absolute dates only: optically stimulated luminescence (OSL) dating of sedimentary quartz from the base of layer 7b, and the top of layer 8 yielded results of 53±10ka (GLL-070304) and 50±5ka (GLL-070305), respectively. Although the luminescence characteristics of the quartz samples require further investigation, these initial results are consistent with the stratigraphic position of the samples and with the geological and archaeological expectation. Further luminescence investigations, including the dating of additional samples, are in progress.
A very large assemblage was found in layers 2.1 and 2.2 (more than 40 000 artefacts, including those recovered by sieving), in a green clayey deposit related to a spring (proluvium deposit). Most of the finds come from a concentrated archaeological horizon within layer 2.1. A second (much less abundant) horizon is present in underlying layer 2.2. The main part of the assemblage corresponds to knapping activities made on local raw material (flint outcrops and river pebbles available within a 2km range). Blank production corresponds mainly to blades (from volumetric cores) and abundant bladelet and microblade production using different methods. Among these methods, the most striking is microblade debitage from carinated or nosed 'endscrapers' (Figure 3). Some of these microblades were retouched (lateral direct retouch on one or two edges, one partially backed by abrupt retouch, one with inverse retouch). Two geometric triangles were also found, but one of them out of context. Apart from these retouched bladelets, the tool-kit is dominated by endscrapers (on blades) and splintered pieces.
Several arguments support the attribution of these assemblages to the Upper Palaeolithic. They show similarities (bladelet production from carinated 'endscrapers', kinds of retouched bladelets, presence of geometric triangles, importance of endscrapers and splintered pieces) with a recently excavated Uzbek open-air site, Dodekatym II, where one of the levels has been 14C dated to around 23ka BP (Krivoshapkin et al. 2005). Bladelet production from 'carinated endcrapers' is also present in the Kara Kamar Cave sequence (northern Afghanistan, Davis 2004; Otte & Kozłowski 2005), dated between 35ka and 25ka BP.
Fauna is very poorly preserved in this layer, but a human tooth was discovered in layer 2.1, at the base of the archaeological horizon described above. This lower left second premolar (Figure 4) is currently under study. It is one of the rare human remains probably attributable to the central Asian Upper Palaeolithic.
In layer 3, a strongly reworked industry has been discovered, probably corresponding to a heterogenous assemblage within high-energy colluvium. This industry shows a mixed Upper and Middle Palaeolithic technology (flake production from discoidal cores and bladelet carinated cores) and a tool-kit containing some side-scrapers. The assemblage also includes many pseudo-tools (denticulate-like) due to edge damage during colluvial reworking processes; this observation contradicts the cultural interpretations ('Mousterian with denticulates', 'Tayacian', and a long tradition of denticulates in the Kulbulak sequence) made by Kasymov (Kasymov & Grechkina 1994: 12). Layers 4 to 11 appear to be archaeologically sterile (based on observations while cleaning Kasymov's trench), but underlying layers (12–18) once again contain artefacts. At least one clear archaeological horizon has been observed in this part of the stratification (layer 16), which contains a small concentration of debitage products with some refitting. All other excavated deposits in this part contain isolated artefacts. This currently lowermost industry is marked by the presence of blade production (blades showing thick, sometimes facetted, platforms; Figure 5). This kind of early blade production may be compared with Obi-Rakhmat rockshelter (Uzbekistan, several layers estimated between 100ka BP and 40ka BP; Derevianko 2004). Up to now, there is no trace of a Lower Palaeolithic or 'Acheulean' industry as originally proposed by Kasymov (1972; Kasymov & Grechkina 1994).
The new excavations at Kulbulak have provided new data enabling us to re-evaluate this site and its place in the Palaeolithic of central Asia. The presence of a Lower Palaeolithic industry is not demonstrated and with respect to the Mousterian artefacts, the importance of the 'denticulates' seems to be taphonomic, related to colluvial reworking processes. Moreover, the interpretation of the Upper Palaeolithic assemblages at Kulbulak has been deeply altered by the recognition of the importance of microblade production. This new project has also provided the first absolute dates and the discovery of a rare human remain. These results now make it possible to include the Kulbulak industries in discussion of Middle Palaeolithic blade assemblages and Upper Palaeolithic industries with microblade carinated cores in central Asia. The site will be playing a role in addressing several related issues: the Middle to Upper Palaeolithic transition, the presence of modern humans and other hominins and hypotheses about the Aurignacian complex in central Asia (Vishniatsky 2004; Otte & Kozłowski 2005; Derevianko 2009; Krause et al. 2010).
Thanks to everyone who participated in the excavations and helped us to organise this project. Special thanks to Andrey Krivoshapkin (IAET SB RAS) without whom the realisation of this project would have been impossible. This project is funded by the Belgian Science Policy, the Russian Academy of Sciences, the Russian Foundation for Humanities, and the Institute of Archaeology and Ethnography (Russian Academy of Sciences, Siberian branch). Luminescence research is funded through the Research Foundation — Flanders (FWO — Vlaanderen; DV: Postdoctoral Fellow). We are grateful to Natalia Vavilina for her assistance with lithic illustrations. Thanks to Rebecca Miller for corrections of the English text.