Abstract
Lithic weapon points occasionally found in Middle Palaeolithic Neanderthal sites are large and do not differ in size, shape or type from those used in other activities such as butchering or plant gathering. The presence in a same assemblage of various types of projectile armatures, some of which are microlithic and designed for this purpose, has only been documented in Modern Humans sites. Recent studies indicate that light projectile points, which would become a key element in Upper Palaeolithic lithic industries, were already present in its formative stages. However, they remain marginal in debates regarding the Middle to Upper Paleolithic transition. We present the initial findings of a traceological search for weapon heads in the oldest layers of the Obi-Rakhmat rock shelter in Uzbekistan, dating back around 80 ka. The lithic industry of this settlement is forming part of the Levantine Early Middle Paleolithic continuity but with several innovative traits. This site, located in the western foothills of the Tian Shan Mountains, northeastern Uzbekistan, has yielded throughout 10 meters of Pleistocene deposits covering 40,000 years a lithic industry characterized by the systematic production of blades (regular thick narrow blades from unipolar and bipolar sub-prismatic and narrow-faced cores, thin and wide blades from flat-faced Levallois-like cores) along with shorter pieces from convergent or centripetal Levallois cores, and bladelets from burin-cores and other small cores. Three types of projectile armature are identified over a selection of 20: retouched points, bladelets and more particularly unretouched triangular micropoints which had previously gone unoticed due to their fragmentary state. According to the fundamental principles of hunting weapon design these micropoints are too narrow for having been fitted to anything other than arrow-like shafts. They resemble the armatures described in a pioneer settlement by Sapiens in the Rhône Valley, France, 25,000 years later.
Introduction
Archaeological perspective
Instrumented hunting is a distinctive trait of the Homo genus. Given the impact of meat consumption on hominization, both cognitively and behaviourally [1], the search for archaeological evidence of past weaponry is of primary importance, with a particular attention to the oldest occurrences.
Increasing studies show that middle or small sized lithic points which are part of the typological characterisation of the Initial or Early Upper Palaeolithic assemblages were projectile heads [2–5], probably mechanically delivered [6–8]. They mark a technical break with the Middle Palaeolithic; from then on, projectile armatures will become the central structuring element of lithic industries (Bordes and Teyssandier, 2011). In spite of this, they remain marginal in the debates regarding the Middle-to Upper Palaeolithic transition. Already suspected of being present in Obi-Rakhmat sequence despite its age [9], light projectile points deserve attention.
We present here the first results of a search for weapon points in the oldest levels of the Obi-Rakhmat rock shelter in Uzbekistan at around 80 ka. The lithic industry of this settlement is forming part of the continuity of the Levantine Early Middle Paleolithic but with several innovative traits [10].
Weapons in focus
Various criteria have been used to recognise prehistoric hunting weapons. The first is analogy with objects of comparable shape known from ethnographic records or from modern sporting or play practices. This is the case with javelins and throwing sticks from the ancient Palaeolithic [11]. More sophisticated approaches employ various acuteness indices, such as TCSA and TCSP [12–17], calculated for lithic points based on ethnographic and experimental data. However, these indices only represent theoretical potential [18, 19]. As F. Bordes wrote: “ It could just as easily be argued that the sockets of bronze spearheads were used to cut rounds out of pie dough” [20]. Furthermore, these indices fail to distinguish between simply tapered heads and those with cutting capacity [21], which makes a significant difference in real-life hunting, nor do they take into account the incidence on the penetration of the hafting device which depends on the morphology of the basal part. A more reliable basis is provided by functional clues. The most obvious evidence is when the tip of a perforating projectile is found stuck in a bone, though such finds are rare [22–31]. It is far more prevalent to find lithic or bone points that have sustained impact damage. However, this damage is subject to variation depending on the projectile design, the ballistic parameters and the impacted target. In contrast to the direct and invariable cause-and-effect relationship that characterises tools used for cutting, scraping, drilling, etc., where the same causes invariably produce the same effects, the identification of a projectile point or insert constitute an extrapolation based on the direction of the violent stress that resulted in the artefact fracture. It is evident that axial compressive stresses can be induced by factors other than being at the tip of a spear or arrow, such as knapping accident [32–34] (S1 Fig: 2), certain types of shaping [35], hard butchering, use as a chisel, accidental dropping, and so forth. In certain instances, the differentiation is simple to make at the level of the artefact itself, in others it can prove more challenging if a series of criteria is not given full consideration. A compelling illustration is given by obsidian points from the Ethiopian rift dated to 279 ka years ago. These points were interpreted as javelin tips on the basis of their shape, apical removals and velocity-dependent microfracture features [36]. This last innovative criterion is physically relevant, but the presumed cause of the energy involved was likely not. A more extensive study on analogous assemblages, founded upon a technological analysis, posits that the recurrent apical removals on such pointed artifacts result from a rejuvenation by the lateral tranchet blow technique [37]. More commonly confusing are the minor damages that often occur at the extremity of pointed tools, which is their most exposed and fragile part. The negative of a tiny burin spall can turn a Levallois triangular flake into a projectile head [38, Fig 5].
Fig 1. Location of Obi-Rakhmat - N41°34'08.8" and E70°08'00.3", 1,250m asl – (basemaps courtesy of the U.S. Geological Survey/ https://usgs.gov/) and view from the site.
https://doi.org/10.1371/journal.pone.0328390.g001
Fig 2. Stratigraphy of Obi-Rakhmat rock shelter and map of the excavation.
https://doi.org/10.1371/journal.pone.0328390.g002
Fig 3. Terminations of bending fractures.
Left: experimental Solutrean leaf point used as dart head, broken at impact, with long step-terminating bending fracture (TFPS Collective Research Program). Right: profile of the different types of termination according to the resultant of compressive and bending forces [39, 40] and their degree of relevance for the recognition of projectile points.
https://doi.org/10.1371/journal.pone.0328390.g003
Fig 4. Obi-Rakhmat: Impacted points and bladelets. The unbroken Levallois point in the white frame (00 – OP X7) illustrates what the ideal type of micropoint was likely to be.
https://doi.org/10.1371/journal.pone.0328390.g004
Obi-Rakhmat
The Obi-Rakhmat rock-shelter is located in the Paltau valley, at the south-western end of the Talassky Alatau range of the Tien Shan mountains, in northeastern Uzbekistan, 100 km of Tashkent. (N41°34’08.8“ and E70°08’00.3“) (Fig 1). Carved into the Palaeozoic limestone at an altitude of 1,250 m, it takes the form of a niche measuring 20 m in width and 9 m in length, with a southern orientation. Since the 1960s, several excavation campaigns have been conducted, initially by the Institute of History and Archaeology of the Uzbek Academy of Sciences [41], and later in collaboration with the Institute of Archaeology and Ethnography, Siberian Branch of the Russian Academy of Sciences [42, 43].
The exposed stratigraphy (Fig 2) consists of 21 sedimentary levels spanning a depth of 10m [44], all of which contain archaeological material. The lithic industry, made from local silicified limestone, is homogeneous and characterized by the production of large blades from unipolar or bipolar and narrow-faced cores and bladelets from core-burins and bladelet cores of various morphology. These reduction strategies coexist alongside Levallois concept which is manifested by the presence of convergent or centripetal Levallois cores and flat-faced cores (Levallois-like). The typological tools include blades—often pointed and/or retouched— splintered pieces, burins, end- and side scrapers, denticulated, borers and retouched flakes. Notably, Levallois (mostly elongated) and Mousterian points are also present, the morphology of which is various and corresponds to the typology of retouched points in the Levantine Early Middle Paleolithic [45, 46]. This composition has led to comparisons between the Obi-Rakhmat lithic assemblage and the Late Middle Palaeolithic, Initial and Early Upper Palaeolithic blade industries [43, 47] from Near East [48–51] and the Siberian Altai [52–55]. Current research considers the Obi-Rakhmat industry exclusively within the Middle Palaeolithic framework [10]. Attempts to date the site have yielded heterogeneous results, as is often the case when applying different methods (14C AMS, U-series, ESR). However, they fix a chronological range from 90 ka for the deepest strata to 40 ka for the uppermost levels [56–58].
The faunal spectrum at Obi-Rakhmat is limited and shows little variation throughout the stratigraphy (Fig 2B). It is dominated by the Siberian ibex ( Capra sibirica) and red deer ( Cervus elaphus), with smaller contributions from wild boar ( Sus scrofra), roe deer ( Capreolus capreolus), golden jackal ( Canis aureus), fox ( Vulpes vulpes), marmot ( Marmota sp.) and hare ( Lepus sp.). This assemblage reflects a combination of steppe and forest environments [59, 60], consistent with palynological data [61]. Remains of large carnivores, including cave lions, hyenas and bears, are rare. In 2003, human remains were discovered at the site: 6 left maxillary teeth and 121 skull fragments from a single juvenile individual (9–12 years old). While the dental morphology aligns more closely with Neanderthal populations, the mosaic of cranial morphological features prevents a definitive attribution, leaving open a classification as an archaic Homo sapiens [47, 62–64]. Finally, elements of bone industry have been identified among the faunal remains [65, 66].
Materials and methods
The analysed sample comes from the collection of the 2001–2002 and 2007–2011 excavation campaigns led by A.I. Krivoshapkin, covering 20 m2, which is currently under study. It comprises typological pieces and small triangular flakes that have been recovered from the bags of lithic debris from layers 20–21 stored at the National Center of Archeology in Tashkent, Uzbekistan. The initial sorting, which sought to identify impact damage, was done with the naked eye. Thereafter the selected specimens were examined with a stereoscopic microscope (Wild M1B/ x7, x14 magnification). Subsequently those exhibiting minimal erosion were analysed with a reflective optical microscope (Olympus BHM, bright field with DIC/ x50, x100, x200, x500 total magnification) to discern microscopic linear impact traces – MLIT [67].Single shots photomacrographs of the impact traces were captured in raw format using a Canon EOS 60D camera equipped with a Canon EF-S 60 mm f/2,8 Macro USM lens and photomicrographs using a Nikon D750 on the phototube of the microscope. Multi-focus shots were processed with Helicon Focus©..3D scanning was performed using a Solutionix D700 structured-light scanner, which makes it possible to create high-resolution non-textured 3D models of micropoints. A standard protocol for structured-light scanning of artifacts was followed [68]. Visualisation was carried out using Artifact-3D software [69]. The 3D PDFs of the supplementary data were produced in Acrobat Pro 9 from the STL files converted in MeshLab 2021.10. To allow each specialist to make their own judgement, the 3D model and photographs of each piece are provided in supplementary information.
As most of the artefacts are unsuitable for microscopic analysis due to the raw material and surface alteration, the determinations are essentially based on the morphology of the fractures, which depends on the ratio of compressive and flexural stresses that caused the material to break by buckling or percussion (Fig 3) , and lateral damage, as described in 40 years of publications [e.g., 2, 3, 39, 70–74].
Our personal experience is based on a corpus of over 500 flint points and barbs of various shapes used as arrow, dart and thrusted spear heads (Fig 3) on medium and large-sized mammals [75–82]. Notwithstanding their ease of use, gelatin targets, even when loaded with bone, cannot be recommended due to their inability to adequately reflect the stresses to which lithic points penetrating an animal’s body were exposed and which determined their design (S1 Note) [21, 83, 84]. Additional experiments (S21 Fig, S2 Table) were conducted with the local silicified limestone to test the debitage scheme hypothesised on the basis of the shape of the micropoints and potential cores seen within the lithic assemblage. This step enabled the observation of the knapping accidents that could mimic impact damage. Twelve of the micropoints thus produced were fixed to 8 mm-diameter commercial wooden arrow shafts using a mixture 80/20% of bitumen (natural outcrops exist in the region) and pulverized charcoal, without binding (as this would be a technical absurdity on continuous cutting edges) for being shot with a modern laminated bow (36 lbs) on a complete uneviscerated carcass of a small ungulate hung in anatomical position in order to start documenting impact breakage and microscopic linear impact traces for the shape and raw material under study. Experiments demonstrated that silicified limestone exhibits comparable macroscopic fracture properties to flint, while MLIT are only observable on the most highly crystallized variants.
The Obi-Rakhmat set has been interpreted after the fundamental principles of projectile head design, as outlined in S1 Note, and compared with earlier and contemporary collections of projectile points, as well as with more recent assemblages featuring certain morphological similarities.
In order to reconstruct the production technologies of point and micro-point production, we performed an attributive analysis on cores, truncated-faceted pieces, and both points and micropoints. Within the Obi-Rakhmat débitage assemblage, points (>3 cm) and micro-points (≤3 cm) were distinguished. Layers 20–21 yielded a total of 194 point specimens. For micro-point analysis, 193 specimens were selected from the small flake category (1–3 cm). The study examined all cores from layers 20–21 (n = 96), a sample of points and micro-points exhibiting impact traces (n = 20), as well as those without impact traces (n = 326). Additionally, 36 truncated-faceted pieces from layers 20–21 were analyzed.
The following attributes were recorded for the cores: core type, morphology and measurements of the last negatives, number of negatives, angle between the striking platform and the flaking surface, striking platform type, and presence of overhang trimming. For the points and micropoints, the examined attributes included: type, morphology, scar pattern, overhang trimming, lip type, bulb type, striking platform type, striking platform width and thickness, and the angle between the striking platform and the ventral surface.
The reduction sequences of cores and truncated-faceted pieces were reconstructed through scar-pattern analysis [85, 86]. Comparative analysis of selected metric parameters was conducted using a statistical approach, including the Mann–Whitney U test for pairwise comparisons.
Inclusivity in global research: Additional information regarding the ethical, cultural, and scientific considerations specific to inclusivity in global research is included in the Supporting Information (S1 Inclusivity). The archaeological materials analyzed in this study are curated by the National Center of Archaeology of Uzbekistan. This research was conducted under the Center’s institutional protocols for scholarly access to collections, in collaboration with their research staff and as part of our ongoing academic partnership with the Institute of Archeology and Ethnography SB RAS (Novosibirsk, Russia). As this study exclusively employed non-destructive analytical methods and complied fully with the Center’s established guidelines for collection use, no additional permits were required under current Uzbek cultural heritage legislation
Results
Impacted armatures
On the basis of the macroscopic criteria commonly used in publications and of our own experimental corpus, we have selected 20 pieces (Fig 4) that can be regarded as projectile armatures from the ongoing lithic inventory of the deepest levels (20–21) of Obi-Rakhmat.
Morphological recurrences and a diversity of distinctive traces are already apparent in this small preliminary sample. According to an agent-based modelling, it could even be considered high [87]. Projectile points and inserts are unusual artefacts in that they are not produced, used and discarded in the same place, and their rejection in the habitat differs depending on whether they are basal fragments brought back attached to the shafts or apical fragments returned in the carcasses [88]. Their density therefore depends on the type of site and the area excavated. In any case, the collection is sufficient to distinguish 3 classes of projectile armature: more or less massive points (between 25 and 35 grams for almost complete specimens), micropoints (between 1 and 4,5 grams but broken) and bladelets (Fig 5).
The large points (Fig 6) are represented by 2 almost complete retouched points (Fig 6: 28, S1 Fig, S1 File, S2 Fig, S2 File) (38 mm and 41 mm wide) and an apical half (Fig 6: 22, S3 Fig, S3 File), all 3 crushed at the tip by an axial impact. An apical fragment of retouched point from layer 19 is similar, having been badly chipped by a tangential impact to its right edge along the same axis (Fig 6: 104, S4 Fig, S4 File). The proximal thinning on both sides of the complete specimens suggests a hafting layout. A thicker but more elongated point (Fig 6: 24, S5 File, S5 File) (30 mm wide) has an apical fracture with a wavy feather termination on the lower side from the same type of compressive stress.
One large proximal fragment (Fig 6: 20, S6 Fig, S6 File) which may be from the same type of point as the previous one, also has a fracture attributable to its use as spear tip, but it is lateral, which can only occur under the leverage of a long shaft.
Two apical fragments from more slender retouched points complete the set. One has a bending fracture with an atypical feather termination but here slightly twisted (S7 Fig, S7 File), while the other has a long spin-off fracture (S8 Fig, S8 File).
The second category consist of 9 unretouched micropoints and 1 retouched micropoint (Fig 7, S9–16 Figs, S9–16 Files), produced in a variety of ways, from simple triangular flakes to typical Levallois points. Two specimens have both fracture from longitudinal stress and MLIT (Figs 8, 9). The average width is 18.2 mm (minimum 15,0 – maximum 23,7 mm). The average weight of the fractured specimens is 1.4 grams (minimum 0.7 g – maximum 2.5 g) (S3 Table). An unbroken Levallois micropoint (Fig 7: 00, S17 Fig, S17 File) measures 21.8 mm long x 17.3 mm wide and weights 1.1 grams; its edge damage is atypical (trampling or crushing). There is no basal shaping despite prominent bulbs. The location on the lower side of the fracture hinge for 6 specimens, half of which have elongated endings, whereas a flat surface is not conducive to their extension, suggests a mounting on the shafts that does not compensate for the prominence of the bulb, i.e., a mounting that is not perfectly in line with the axis (which is not optimal for the wake drag coefficient). The shafts were probably made of wood only, as no potential intermediate nor apical bone elements were found [65].
Fig 5. Three types of lithic weapon armature identified in layers 20 and 21 of Obi-Rakhmat: 1 – Large retouched point, 2 – micropoint (Levallois) and 3 – bladelet.
https://doi.org/10.1371/journal.pone.0328390.g005
Fig 6. Large retouched points crushed or broken at impact.
https://doi.org/10.1371/journal.pone.0328390.g006
Fig 7. Micropoints broken or crushed at impact.
https://doi.org/10.1371/journal.pone.0328390.g007
Fig 8. Micropoint n° 7 (5 – ОР 01-08 СЛ 21.1) with macroscopic and microscopic impact traces.
Centimetric scale.
https://doi.org/10.1371/journal.pone.0328390.g008
Fig 9. Micropoint n° 8 (6 – ОР 2001-2008 СЛ 21.1) with macroscopic and microscopic impact traces.
Centimetric scale.
https://doi.org/10.1371/journal.pone.0328390.g009
The 3rd category (Fig 10) is the least represented here because it was not included in our sorting criteria, which focused on axial armatures. Four incidental findings came from this sorting. These are raw bladelets. Without retouch or MLIT it is not possible to distinguish between those accidentally broken during knapping [32, 34] and those broken by an axial use. They were therefore excluded from the current inventory. Two other bladelets were selected on the basis of the following criteria: one (Fig 10: 19, S18 Fig, S18 File), whose breakage by bending is equivocal, has a very discreet retouch along one edge caused by pressure that appears to be postdating the fracture; the second sample (Fig 10: 15, S19 Fig, S19 File), probably burnt from dehafting, has been crushed by a tangential contact with a hard material, which is typical of a projectile lateral insert. A more discreet crushing is also observed on a genuine backed bladelet (Fig 10: 31, S20 Fig, S20 File).
Fig 10. Bladelets crushed of broken at impact.
https://doi.org/10.1371/journal.pone.0328390.g010
The set, although small, indicate the use of bladelets in the design of hunting weapons at Obi-Rakhmat, 80 ka ago. We could imagine that they complemented the micropoints, to extend the length of the cutting edge, thus forming part of a more robust projectile head compatible with dart foreshafts, but nothing in the design of the micropoints would ensure a continuity to prevent the bladelets from being pulled out at the penetration. Recent archaeological examples show the extreme attention paid to the continuity of lithic inserts [89]. A larger sample and therefore specific research would be required to understand the function(s) of the bladelets, which are a notable element of the lithic production at Obi-Rakhmat.
Technological characterization of production
Point production.
The primary reduction system in Obi-Rakhmat layers 20–21 exhibits significant diversity. This phenomenon is evident in the morphology of the cores themselves, as well as in the typology of the technical flakes and blanks. The industry places significant emphasis on the Levallois techniques, including Levallois for points and centripetal Levallois, as well as blade production through the exploitation of flat faced, sub-prismatic, and narrow-faced cores. In addition, bladelets were also produced using various core types.
The diagnostic blanks include blades, bladelets, and flakes in nearly equal proportions. Depending on the layer, points constitute up to 10% of all flakes (S1 Table). Notably, according to the tool assemblage, pointed blank morphology was of particular importance (Fig 11). A significant proportion of the tools in layers 20–21 consist of various types of retouched pointed implements (S1 Table).
Fig 11. Convergent spalls from layers 20-21. 1-6 Bladelets; 7-10, 17-18, 21-23 Levallois points; 11-13, 16 Axial points; 14-15, 19-20 Elongated points.
https://doi.org/10.1371/journal.pone.0328390.g011
Given the broad diversity of reduction sequences employed in the assemblage and the high proportion of retouched points among the tools, we propose that blanks for pointed tools were not limited to predetermined Levallois points but also included suitable technical flakes with pointed morphology. This observation is applicable to both pointed tools in general and the sample of impact-fractured points.
In the assemblage of layers 20–21, Levallois point cores (Fig 12A) constitute 5–30% of the total (S1 Table). Scar-pattern analysis indicates that pointed blanks were produced both as predetermined products and technical spalls from blade cores. The collection includes flat-faced unidirectional (Fig 12: 2) and bidirectional (Fig 12: 3) cores, as well as a considerable proportion of narrow-faced cores (Fig 12: 4-5), which frequently exhibit pointed negatives. Another technological method (Fig 12D), which is extensively represented in the collection, is associated with subprismatic unidirectional (Fig 12: 6) and bidirectional (Fig 12: 7) cores. Among the subprismatic cores, there are asymmetrical (semi-rotated) cores (Fig 12: 8) that combine exploitation of both large and narrow core faces.
Fig 12. Technological chains of point and elongated blade production.
A. 1 Levallois point core. B Flat faced cores: 2 Flat faced, unidirectional; 3 Flat faced, bidirectional. C. Narrow-faced cores: 4-5 Narrow-faced unidirectional. D Sub-prismatic cores: 6 Sub-prismatic unidirectional; 7 Sub-prismatic bidirectional; 8 Sub-prismatic asymmetrical ( Semi-tournant).
https://doi.org/10.1371/journal.pone.0328390.g012
Micropoint and bladelet production.
While micropoint production has not been a focus of previous studies in the Obi-Rakhmat assemblage, the identification of impact-damaged specimens in the use-wear sample now necessitates an examination of this reduction sequence. Initial investigations do not reveal a clearly defined chaîne opératoire dedicated exclusively to micropoint production. We have systematically examined all cores for the presence of small triangular negatives corresponding to the dimensions of impact-crushed micropoints – with particular attention to the characteristics of the proximal zone.
As with the larger blanks in the Obirakhmatian industry, we observe considerable diversity in the cores aimed at producing small blanks – bladelets and small flakes (Fig 13). Based on the morphology of the débitage itself, the dimensions and morphology of the primary and secondary negatives, we conclude that micropoints could have been produced by several reduction sequences. One of the predominant reduction sequences appears to have involved: small Levallois cores present in the collection (Fig 13A); and small flat faced cores-on-flakes (Fig 13B) that yielded small flakes and bladelets
Fig 13. Technological chains of micro-point and bladelet production.
A. 1-2 Small Levallois point cores. B: 3 Flat faced, bidirectional core on flake. C: 4 Burin-cores, 5 Narrow-faced core. D: 6-7 Sub-prismatic, unidirectional cores. E: 8-10 Carinated cores.
https://doi.org/10.1371/journal.pone.0328390.g013
While truncated-faceted pieces (Fig 14) have been interpreted as micro-point cores in other contexts [6, 90], we argue that within the Obi-Rakhmat technocomplex, most such artifacts functioned mainly as tools [46].
Fig 14. Truncated faceted pieces from layers 20-21: 1-3.
https://doi.org/10.1371/journal.pone.0328390.g014
This interpretation is supported by several key observations: (1) their highly standardized morphology emphasizes deliberate working edge formation rather than core reduction features; (2) they rarely exhibit the triangular negatives characteristic of micropoint production; and (3) although they often possess prepared striking platforms, these show no evidence of flake removals. Crucially, metric analysis reveals that truncated-faceted pieces differ significantly from formal cores in striking platform angles (Fig 15). Moreover, the platform angles of these truncated-faceted pieces show no correlation with the residual angles observed on the analyzed points, in marked contrast to all other core types in the assemblage. This discrepancy (confirmed by Mann-Whitney U tests, p < 0.01) further demonstrates their technological separation from systematic core reduction activities. Fig 15 presents boxplots showing: (1) angles between flaking surfaces and striking platforms of different core types, (2) angles of truncated-facetted pieces, and (3) striking platform angles preserved after point detachment (measured from points in our sample).
Fig 15. Comparative analysis of the angles between the flaking surface and the striking platform of cores, as well as the dorsal angle of points.
Residual point angles were calculated using the formula: 180° minus the angle between the striking platform and the ventral surface. Raw data are in S4 Table.
https://doi.org/10.1371/journal.pone.0328390.g015
Scar-pattern analysis showed that flakes with the required triangular morphology could also be produced during bladelet production. A developed bladelet production is widely represented in the lithic industry (S1 Table). The assemblage of lower layers is comprised of 20–30% bladelets (S1 Table), with the proportion of bladelet cores reaching 50% (S1 Table). Th
