the BARMAKy, LeVeL 2 epiGRAVettiAn ASSeMBLAGe: the cORe RedUctiOn StRAteGieS

The Barm­aky cultural deposits accum­ulated under the harsh clim­atic conditions of the periglacial zone at about 19 kyr cal BP. It is difficult to overestim­ate the m­eaning of lithic raw m­aterial supply as a com­ponent of the technological organization of m­obility patterns of hunter-gatherer groups. The applied core reduction strategies are an integral part of the technological chain of raw m­aterial m­anagem­ent. The two discrete core reduction strategies were defined for the flint assem­blage of Barm­aky, level 2: blade and bladelet / m­icro-blade. The form­er is based on the reduction of the sub-, cylindrical and narrow flaking surface cores; the latter is based on the exploitation of the narrow flaking surface cores. The blade cores were the source of lam­ellar products for the «dom­estic» toolkit; the bladelet / m­i-cro-blade cores reduction aim­ed at the insets for distant weapons.

At the same time, except for V. Lozovski and O. Lozovskaya's studies of Mezhyrich lamellar assemblage, the Mid-Dnieper Epigravettian special technological investigations are unknown (Lozovski, Lozovskaya 2014).The Mid-Dnieper Epigravettian assemblages' core reduction is defined as «prismatic».Such a definition was made in the frames of the general emotional characteristic of Epigravettian, as the «impoverished version of Gravettian» (залізняк 2021, с. 9; Нужний 2015, с. 442).The continuous cores reduction aimed the blades with a straight lateral profile and width range of 9-19 or 12-15 mm was stated for Desna valley Epigravettian and such assemblages as Semenivka 3, Mezhyrich and Dobranichivka (Гаврилов 2016, с. 55;Нужний та ін. 2017, с. 23-24;Shydlovskyi et al. 2022, p. 213).The cores' typological variability and extensive débitage studies were left beyond the research procedure.Obviously, such technological studies were difficult to recognize as enough detailed to be the background for further investigations of subsistence strategies of the Epigravettian population on the edge of the Scandinavian Ice Sheet.Thus, the main aim of the proposed paper is the analytical reconstruction of core reduction strategies of Barmaky, level 2 assemblage.
Methods.The proposed studies are based on the investigations of technologically significant attributes of core-like pieces and débitage.Such attributes were adopted for the number of Middle and Upper Paleolithic stone assemblages in the Crimea, Portuguese Estremadura, Levant, Romanian Banat, Central Asia, and Altai (Chabai, Demidenko 1998;Demidenko 2012;Almeida 2000;Monigal 2002;Sitlivy et al. 2012;Колобова 2014;Derevianko et al. 2018).The characteristic feature of this approach consists of the analysis of each artifact through the number of attributes, such as conditions, shapes, the axis of detachment, lateral and distal profiles, cross-sections, dorsal scar patterns, cortex placements and percentages of corticated areas, striking platforms characteristics, measurements, as well as the blank dimensions.
The core-like pieces typology is based on V. Gladilin's classification, which gives the possibility for the precise description of core types' variety (Гладилин 1976).The definition of the difference between burin spalls and microblades is one of the vaguest typological research problems (Demidenko, Chabai 2012, p. 269).In present studies, in addition to the attributes proposed by J. Tixier (1974), the burin spalls were defined as the aggregation of the following attributes: the lateral steep or bilateral steep midpoint cross-sections; the oblique, not abraded, untrimmed, finely faceted or plain striking platform.
The ENTRER 4 program and Mitutoyo digital caliper were applied to analyze the proposed attributes.
Raw material characteristics.The extremely rich flint outcrop is situated about 300 m south of the site.The flint nodules originate in the Turonian chalk / marl deposits.The color of flint is intensively grey or black; the cortex is whitish or light grey; the structure is fine-grained.In Prehistoric studies, such kind of raw material is widely known as a high-quality Volhynian flint.
Rounded nodules, plaquettes, and natural flakes represent the shapes of flints in the Barmaky outcrop.Also, the big primary flakes were often used as the blanks for pre-, cores production.Such variability of raw material shapes might be combined into two groups: 1, the plate-like pieces -plaquettes with cortex on two opposite surfaces, the natural and primary flakes with cortex on one surface; 2, the rounded nodules.The main characteristic feature of plate-like pieces consists in the width-to-thickness ratios: width several times prevails over thickness.The average thickness of plate-like pieces is rarely exceeding 5 cm.The maximum length and width of plaquettes and natural flakes compose up to 20 cm.The maximum dimensions of rounded nodules compose up to 35-40 cm.
sample.The 152,617 artifacts found in Barmaky, level 2, excavated on 72 m 2 during the 2018-2020 field campaigns (table 1), represent the flint assemblage.Due to the cryoturbation processes and anthropic activity, the level 2 assemblages were subdivided into several horizons and artificial objects (чабай та ін.2022).There are no significant differences in raw material exploitation between studied horizons and objects.There are no indications of imported raw materials.The flints covered by patina, as well as artifacts with naturally damaged edges, are solitary.The general structure of artifacts suggests the on-site cores reduction, tools produc-tion, use, and rejuvenation under the condition of flint abundance, id est the site-workshop model of raw material exploitation was defined (чабай та ін.2020b).The toolkits are represented by burins (≈45 %), truncated pieces (≈20 %), and microliths (≈25 %), while the end-scrapers, perforatorborers, pointed blades, raclettes, and composite tools are not numerous.Thus, Barmaky, level 2 supposed to be a reliable source for technological studies.
hammer-stones.There are two relatively complete hammer-stones: the fragment of sandstone pebble and a flint pebble.In addition, there are small chunks and flakes of sandstone and quartzite: 692 pieces.Such pieces might be the spalls from crashed hammer-stones.It is likely that some of the flint chunks also originated from hammer stones destruction, given that sandstone and / or quartzite pebbles are absent in the local rivers and flaking instruments therefore rare.The deficit of reliable material for hammer-stones is usual for Volhynian Paleolithic sites (Савич 1975, с. 56-99).The source for sandstone / quartzite pebbles is the Sarmatian deposits available on the Volhynian upland.
chunks.Fragments of raw material without clear traces of flaking represent chunks.Some pieces exhibit a few negatives, which do not show the clear flaking surface or striking platform.About 20 % of chunks are big enough (≥5 cm in max dimensions) to be attributed as the reserve of raw material.
unidentifiable débitage.This category represents the fragments with definite dorsal and ventral surfaces (table 1).The degree of fragmentation disables its precise definition.The relatively high portion of unidentifiable blanks reflects the application of hard hammer percussion and/ or the evidence of intensive trampling.Latter appears to be the evidence of intensive exploitation of the living floor surface.
chips.The blanks with flake proportions and maximum dimensions ≤15 mm represent the chips (Marks 1976).The chip component in Barmaky, level 2 assemblage composes more than 80 % (table 1).It might be evidence of both intensive raw material reduction and satisfactory preservation of artifacts in cultural deposits.
core-like pieces.The core-like pieces are subdivided into the pre-cores and cores.Nodules, plaquettes, and big flakes with either formed striking platform (s) or partly shaped flaking surface (s) represent the former.The latter is represented by nodules, plates, and big flakes with a well-shaped striking platform (s) and flaking surface (s) (fig.1-4).The number of cores is three times bigger than the pre-cores number (tables 1-3).The common feature for all of the core-like pieces is the application of the trimming method of overhang between the striking platform and flaking surface (fig.1: 1-4; 2; 3: 2; 4: 1, 3).Based on the width of negatives on the flaking surface (s), the pre-core assemblage is subdivided into five main groups.These groups reflect the assumed pre-core function: the flake pre-cores; blade pre-cores; blades / bladelets precores; bladelets / micro-blades pre-cores; and unidentifiable.The flake pre-cores consist of a single fragmented piece.
Five pieces belong to the blade pre-cores.All of them are unidirectional (table 2).Two pre-cores are sub-cylindrical and made of nodules and natural flakes.The latter shows the flaking surface prepared by the lateral crested ridge.The only blade pre-core made on nodule exhibits a narrow flaking surface.Two more blade pre-cores are unidentifiable because of fragmentation.2).The unidirectional narrow flaking surface pre-cores are the dominant type of pre-core assemblage.Nearly all of them are made of natural flakes (table 2).Also, the natural flakes are the main initial blank for blade / bladelet pre-cores production.Fifteen items demonstrate the crested preparation of the flaking surface, back, and distal extremity.The size variations are the following: length -from 46.14 to 107.65 mm; width -from 14.29 to 39.76 mm; thickness -from 23.98 to 90.63 mm.The average value of length is 71.08 mm; width -24.99 mm; thickness -60.61 mm.The width-to-thickness ratio reflects the dominance of narrow flaking surfaces among the blade / bladelet cores.
Seven pre-cores were made for the purpose of bladelet / micro-blade production.Six of them are unidirectional; one piece -is bidirectional; all of them show a narrow flaking surface (table 2; fig.3: 1).Four bladelet / micro-blade pre-cores were made on natural flakes; two -on plaquettes; one -on flake.The single piece exhibits the crested ridge on the flaking surface.The size variations are the following: length -from 45.39 to 108.04 mm; widthfrom 12.30 to 25.03 mm; thickness -from 34.58 to 106.90 mm.The mean dimensions are the following: length -63.78; width -19.33; thickness -67.56 mm.The width-to-thickness ratio reflects the dominance of narrow flaking surface pieces.In addition, there are seven fragmented precores (table 2).One of them clearly shows the narrow flaking surface.Two flaking surfaces exhibit traces of the crested ridge preparation.There is no sense to mention the dimensions of unidentifiable pre-cores because of fragmentation.
In sum, the preferable blanks for pre-cores were the plate-like primary flakes, natural flakes, and plaquettes (fig.5: 3).These led to the appearance of the narrow flaking surface pieces, which prevail over the pre-cores with the sub-cylindrical flaking surface (fig.5: 2).The pre-cores reduction presumes the surface / back crested ridge (s) preparation.The single-platform unidirectional items dominate the pre-core assemblage (fig.5: 1).The narrow flaking surface pre-cores on flakes and plaquettes serve for blade / bladelet / microblade production, while the sub-cylindrical precores on nodules serve for blade production.The mean values of pre-core sizes are represented by the following numbers: length -67.98; width -29.17; thickness -58.27 (fig. 6: 7-9).The widthto-thickness ratio reflects the orientation of platelike pieces of raw material towards the narrow surface flaking.In general, 27 of 34 identifiable pre-core blanks belong to the middle-size platelike, relatively thin forms: plaquettes, primary and natural flakes (table 2; fig.5: 3).The remainders of pre-cores were made on roundish elon-gated nodules close to the cylindrical shape.The former aimed the bladelet / micro-blade production, while the latter is supposed to be the blanks for blade cores.Thus, the careful selection of the initial blank for core reduction predetermines the further results of the knapping process.
According to the assumed function, six main groups present the cores' assemblage (table 3).The most numerous are the cores for blades (36.64 %), cores for blades / bladelets (34.35 %), and cores for bladelets / micro-blades (20.61 %).These groups consist of types' variety that reflects the number of orientations, arrangements, and shapes of the flaking surfaces and striking platforms.Such variety was not initially intentional.The cores types' variety reflects the necessity to solve the numerous unexpected problems during the process of cores reduction.
Nearly equal numbers represent the unidirectional and bidirectional cores for blades: 21 and 27 pieces, respectively (table 3; fig.5: 4).The bidirectional blade cores show the clear difference between striking platforms: the main striking platform is arranged at the nearly right angle to the flaking surface, while the opposite supplementary platform is arranged at a very sharp angle toward the flaking surface (fig.1: 2; 2: 2, 3).In addition, the main striking platforms are much wider and thicker than supplementary platforms  3: 3; 4: 3).Along with the bladelet / micro-blade cores with pyramidal / pencilshaped working surfaces, the wedge-shaped items were recognized as the probable evidence of indirect percussion technology and origin of pressure flaking (Svoboda 1995, p. 651;Нужний 2008, с. 202).The following ranges represent the bladelet / micro-blade cores' dimensions: length -from 36.66 to 123.49 mm; width -from 14.56 to 35.76 mm; thickness -from 14.56 to 35.76 mm.The mean bladelet / micro-blade core dimensions are represented by the following values: length -55.13 mm; width -23.17 mm; thickness -33.83 mm.The average width and thickness values reflect the im-fig.5. Barmaky, level 2, the attributes of core-like pieces: 1-3 -lamellar pre-cores; 4-6 -blade cores; 7-9 -blade / bladelet cores; 10-12 -bladelet / micro-blade cores portance of narrow flaking items in bladelet / microblade core assemblage.
In sum, both pre-core and core assemblages are characterized by a relatively significant portion of relatively short (50.00-65.99mm) pieces (fig.6: 1, 2, 7).The width and thickness measurements demonstrate the binary distribution of values.The widths of approximately 80 % of cores and pre-cores belong to the 15.00-39.99mm interval.The second metrical group shows the range of width measurements in 40.00-69.99mm intervals (fig. 6: 3, 4, 8).The same trend is characteristic of the distribution of length values.About 50 % of pre-cores demonstrate the thickness in 55.00-79.99mm intervals.Three-quarters of cores are thinner than 49.99 mm (fig.6: 5, 6, 9).The presented size characteristics suggest the orientation of core reduction strategies toward the production of middle-length and both narrow and relatively wide lamellar products.
Thus, the following characteristic features of the lamellar core-like assemblage were defined.
1.The purposeful selection of plate-shaped pieces of raw material for narrow flaking surface pre-cores and cores production.
2. The purposeful selection of rounded nodules of raw material for sub-, cylindrical pre-cores and cores production.
4. The blade production was based on the sub-, cylindrical and narrow flaking surface cores reduction; mean cores' measurements -65.57× 40.01 × 41.10 mm.
5. Both core reduction strategies were based on the exploitation of mainly unipolar cores by hard hammer direct percussion from the right-angled striking platform with an application of the overhang trimming procedure.
6.The main reasons for core discard are transverse breakage and hinge fractures on the flaking surface.
7. The main method of cores' flaking surface repair is the detachments from the opposite supplementary platform.
The trapezoidal and rectangular pieces detached off-axis prevail among the «regular» flake assemblage (fig.7: 1, 17).More than half of «regular» flakes show differently incurve lateral profiles (fig.7: 5).About two-thirds of «regular» flakes exhibit the feathering distal extremities (fig.7: 9).The sum of multiple and trapezoidal cross-sections amount the half of the assemblage (fig.7: 13).The unidirectional scars dominate the types of dorsal patterns (fig.8: 1).The lateral and distal cortex placements prevail among the variants of cortex placement.The primary pieces compose 10.41 % of «regular» flakes.The totality of items with cortex on the dorsal surface reaches up to 55 % of «regular» flakes (fig.8: 5).The flakes' striking platforms are mainly plain, without overhang trimming, and not abraded (fig.8: 9, 13).The dominating right-angled (73.15 %) and un-lipped (88.25 %) butts correspond with pronounced bulbs (79.21 %).Latter is supposed to be the indication of direct hard hammer percussion (Pelegrin 2000, p. 75-76).The «regular» flakes are short, often transversal, and relatively thin (table 5).Mentioned characteristics suggest the origin of «regular» flakes from the pre-cores and cores shaping and maintenance procedures.The «regular» blades, bladelets, and micro-blades show more similarities than differences.On the other hand, the lamellar débitage is distinct from the «regular» flake assemblage by the number of attributes.Due to the increasing portion of shapes, such as crescent, leaf, rectangular, and triangular, the lamellar blanks look geometrically more regular (fig.7: 2-4).Moreover, about half of them were detached on-axis (fig.7: 18-20).Compared with «regular» flakes, the lamellar regular débitage shows the more significant component of blanks with straight profiles and feathering distal ends (fig.7: 6-8, 10-12).The lamellar blanks with unidirectional negatives dominate among the dorsal scars types (fig.8: 2-4).The percentages of primary lamellar products are insignificant.The pieces with mainly lateral and distal cortex on the dorsal surfaces compose about half of the blades, one-third of bladelets, and a quarter of micro-blades (fig.8: 6-8).Along with the plain, the linear striking platforms become more important (fig.8: 10-12).Compared with flakes, the role of overhang trimming increased three times for lamellar pieces (fig.8: 14-16).At the same time, the values of unlipped (88-95 %), pronounced bulbs (75-81 %), not abraded (96-97 %), and right-angled striking platform (77-85 %) did not change much.The relatively high amount of crushed striking platforms (16-20 % for all types of débitage), pronounced bulbs, and un-lipped butts are characteristic of the hard hammer direct percussions.
The mean values of length and width of lamellar débitage and lamellar cores are well corresponding (table 5; fig.6).The elongation indexes for blades, bladelets, and micro-blades compose 2.58, 2.79, and 3.32, accordingly.That is, the «regular» lamellar blanks are both not very long and not very elongated.At the same time, the lamel-lar pieces are relatively thick; the width-to-thickness ratios for blades, bladelets, and micro-blades compose 3.68, 3.97, and 3.46, accordingly.
In sum, the «regular» lamellar assemblage attributes suggest an application of mainly unipolar, hard-hammer direct detachment, supplemented by overhang trimming to obtain medium size, relatively thick blanks.Compared with «regular» flakes, it is necessary to underline the more careful preparation of striking platforms, the increasing roles of straight profiles, feathering distal ends, and on-axis «regular geometrical shapes» of the lamellar assemblage.
technological débitage.One of the most characteristic features of Barmaky, level 2 débitage assemblage is the high percentage of technological blanks: crested pieces, different types of débordante blanks, core tablets, Kantenabschläge, and primary pieces.In sum, the technological blanks comprise 25-35 % of the different classes of blanks (table 4).
The Kantenabschläge and primary flakes play a significant role in débitage assemblage (table 4).The Kantenabschläge is the result of the preliminary stage of crested ridges and striking platforms preparation on plate-shaped pieces of raw material (Richter 1997, p. 186-187).The Kantenabschläge shows the distally / medially incurved lateral profile, a blunt distal end, which plane is nearly parallel to the plane of the acute striking platform (fig. 9: 13, 15, 16).
The Kantenabschläge dorsal surfaces are completely or partially covered by the cortex; the shapes are mainly transversally trapezoidal, detached off-axis.The specific type of Kantenabschläge dorsal scar pattern is defined as a transversal crested ridge, perpendicular to the axis of detachment.The Kantenabschläge is characterized as the widest and thickest type of débitage, including the platform dimensions (table 5).No specific attributes of primary flakes were recognized.The primary flakes' average dimensions are not sufficiently different from the measurements of «regular» flakes (table 5).The primary pieces with maximum dimensions >50 mm compose only 7.72 % of primary flakes assemblage.It looks like, most numbers of the primary flakes originate from the repeated stages of cores shaping.
Having in mind the number of cores, the number of core-tablets is not impressive (table 4).Table 5 represents the average dimensions of core-tablet pieces.The mainly transversal proportions of core-tablet flakes suggest their origin from sub-, cylindrical cores striking platform shaping (table 5).A few flakes and lamellar coretablets might be associated with narrow flaking surface core striking platforms maintenance.
The débordante blanks are the detachments from the lateral sides of cores.The main attribute of the débordante pieces is the lateral steep crosssection (fig. 9: 3-5, 7-9, 14).The «steep» side of the débordante piece is arranged at a nearly right angle to the ventral surface (Boëda, Geneste, Meignen 1990).These pieces are also known as «corner» blanks (Giria, Bradley 1998).The main technological reason for débordante detachment consists in the creation of the needed transversal convexity of the core flaking surface.Depending on the conditions of the «steep» lateral side the débordante pieces were subdivided into cortical -covered by the cortex, natural -exhibiting the natural surface, and lateral -exhibiting the negatives from previous detachments (Derevianko et al. 2018, p. 379).The débordante to regular pieces ratios constitute 1 to 7.31 for flakes, 1 to 3.55 for blades, 1 to 4.22 for bladelets, and 1 to 4.41 for micro-blades.Such significant ratios, especially for lamellar items, reflect the big role of débordante blanks in narrow flaking surface core reduction.
Half of the débordante pieces belong to the lateral type; the other half is equally divided between natural and cortical types (table 4).Except for the declared lateral-steep cross-section and dimensions, the attributes for débordante flakes and lamellar pieces are nearly similar to those mentioned for corresponding «regular» types of débitage.In fact, in Barmaky, level 2 assemblage the differences between flake and lamellar débordante consists in blank proportions, only.
Thus, the débordante pieces are a little bit longer and wider, but sufficiently thicker than their «regular» analogs (table 5).The same is true for débordante and regular débitage striking platforms measurements.In addition, the length / width proportions for débordante flakes are bigger than for «regular» flakes.That is, the efficiency of the débordante method of flaking surface maintenance might be evaluated as problematic, but still important.
The plateaus of both débordante and regular lamellar width values correspond to the width of bladelet / micro-blade pieces and cover about 60 % of all available lamellar débordante (fig.11: 2, 3).In sum, the plateau of the width of both fig.10.Barmaky, level 2, the length dimensions distributions by metrical intervals: 1 -the lamellar crested pieces; 2 -the lamellar débordante pieces; 3 -the lamellar «regular» pieces débordante and regular lamellar is defined in the 4.00-12.99mm range and reflects the dominating role of bladelet / micro-blade production.The 1.00-2.99mm thick pieces compose about half of the lamellar débordante and regular débitage (fig.12: 2, 3).
The crested débitage composes 8.31 % of all identifiable blanks.The crested débitage consists of flakes (32.66 %), blades (39.00 %), bladelets (20.55 %), and micro-blades (7.78 %).In general, the attributes for crested flakes and lamellar pieces are nearly similar to those mentioned for corresponding «regular» and débordante types of débitage.Based on the negative arrangements on dorsal surfaces the crested pieces are subdivided into several types (table 6).These types reflect the crested ridges applications during the primary (one-side, double-sides, alternating) and subsequent (combinations of crested ridges and uni-, bidirectional negatives) stages of pre-cores and cores reductions (fig.9: 1, 2, 6, 10-12).In total, the pieces of subsequent stages compose 66-76 % of all crested pieces.It suggests the repeated preparation of crested ridges during the fig.11.Barmaky, level 2, the width dimensions distributions by metrical intervals: 1 -the lamellar crested pieces; 2 -the lamellar débordante pieces; 3 -the lamellar «regular» pieces  Thus, the technical meaning of débordante and crested pieces are similar: the construction of working surface transversal convexity by detachments of lateral blanks.The typology and diagrams of length and width values of both débordante and crested pieces suggest its detachment during all stages of cores reduction.Also, the presence of débordante and «primary» crested bladelets and micro-blades underline the existence of special bladelet / micro-blade pre-cores and cores.
discussion.Thus, the proposed studies of technologically significant attributes of core-like pieces and débitage might be summarized as follows.2. The special methods of flaking surface and striking platforms preparations for blade and bladelet / micro-blade cores were not defined.Unidirectional flaking is the main method of cores reduction.The flaking surface shaping is based on the crested pieces and débordante detachments.The striking platforms preparation is based on the extensive use of the overhang trimming method.The isolated striking platforms were not found.
3. The main method of cores' flaking surface repair is the creation of an opposite supplementary platform for the removals from the conjoined or alternate flaking surface.The core exhaustion led 5.The bladelet / micro-blade core reduction resulted in the middle-size (av.25.65 × 9.18 × 2.31 mm) bladelets / micro-blades, mainly of rectangular shapes, straight lateral profiles, and feathering distal ends.
6.The results of blade cores reduction were used for all types of tool production, except microliths.The results of bladelet / micro-blade cores reduction were used for backed microlithic insets production (fig.13: 1-3).
Thus, the proposed reduction strategies were applied to the production of different toolkits (fig.14).The production of the «domestic» toolkit was the main aim of the blade core reduction.
While the production of distant weapon insets was the purpose of bladelet / micro-blade cores reduction.Along with the reconstructed site-workshop model of raw material exploitation, the presence of dwelling structure, and pits, the application of double-purpose core reduction strategies might be evaluated as the characteristic features of the base camp (чабай та ін.2022).Quantitatively the summarized amount of bladelet and microblade pieces prevails over the blade number.The microlithic insets constitute about a quarter of the tool kit.Moreover, some number of the insets were produced for off-site use and lost in hunting episodes outside the camp area.
Such a dual approach to the reduction strategies contradicts the concept of continuous reduction strategy, proposed for the Epigravettian assemblages from Desna valley, and such Mid-Dnieper assemblages as Semenivka 3, Mezhyrich, and Dobranichivka (Гаврилов 2016, с. 55; Нужний, Шидловський, Лизун 2017, с. 23-24).Also, the method of butt abrasion known from the Mezhyrich assemblage is not found in Barmaky (Loz-fig.14.Barmaky, level 2, the core reduction strategies: I -bladelet / micro-blade core reduction; II -blade core reduction ovski, Lozovskaya 2014).The meaning of the difference between mentioned assemblages might be of chronological and / or stylistic values.At least, the relatively early age of Barmaky assemblage in the frames of Mid-Dnieper Epigravettian supports the hypothesis of the chronological meaning of technological variability.The chronological meaning of continuous and discrete reduction strategies has been reported as a characteristic feature for the beginning of Middle Magdalenian in southwestern France, between 19 and 17.5 kyr cal BP (Langlais et al. 2016).For the region of the Northern Mediterranean Basin, discrete blade and bladelet flaking were reconstructed for Early Epigravettian, while blade cores characterize the Late Epigravettian assemblages (Naudinot et al. 2017).Also, the presence of micro-blade wedgeshaped narrow flaking surface cores is supposed to be the characteristic feature of Early Epigravettian assemblages from Moravia and Silesia, «probably dated into the time span of the several millennia around the Last Glacial Maximum» (Svoboda 1995, p. 655).The Hungarian Early Epigravettian site Vác 1 assemblage is characterized by bladelet core reduction and backed pieces, as the main aim of production (Béres, Demidenko 2021).On the other hand, the discrete blade and bladelet / micro-blade reduction strategies were studied for Prut / Dniester Epigravettian assemblages (Noiret 2009, p. 458-461).Taking into account the spatial and chronological proximity of Barmaky to Dniester valley Epigravettian and the typological similarity between Barmaky and Mizyn assemblages, the proposed studies might open a new direction in the investigations of the late Last Glacial Maximum industries' origin of the Mid-Dnieper basin.
Acknowledgments.The proposed paper is a part of the studies conducted in frames of the National Ukrainian Academy of Sciences project 0121U110296 «Матеріальна культура давнього Причорномор'я як індикатор впливу глобальних міграційних процесів» (Material culture of the ancient Black Sea region as an indicator of the influence of global migration processes) and Deutsche Forschungsgemeinschaft, project DFG-392605832 «Social networks and environmental conditions before, during and after Last Glacial Maximum in Volhynia, Western Ukraine».

V. P. Chabai, D. V. Dudnyk the BARMAKy, LeVeL 2 epiGRAVettiAn ASSeMBLAGe: the cORe RedUctiOn StRAteGieS
This paper is devoted to the analytical reconstruction of the core reduction strategies of the Barmaky site, level 2 assemblage.Studying the technological features of knapping is significant to determining the characteristics of lithic industries and techno-complexes.In Eastern, Southern, and eastern parts of Central Europe, the late Last Glacial Maximum industries mainly belong to the Epigravettian techno-complex.Along with typological descriptions, the characteristic of core reduction strategies is essential for Epigravettian techno-complex definition.At the same time, the 57 ISSN 2227-4952 (Print), ISSN 2708-6143 (Online).Археологія і давня історія України, 2022, вип. 4 (45) Chabai, V. P., Dudnyk, D. V.The Barmaky, Level 2 Epigravettian Assemblage: the Core Reduction Strategies comprehensive description of Epigravettian assemblages' core reduction of the Mid-Dnieper basin is still unclear.Excavated at 2018-2020 field campaigns, the exceptionally reach Barmaky, level 2 artifact assemblage gave such an opportunity.The 152,617 flint artifacts, including 174 core-like pieces and 14,988 débitage items, originate from 72 m 2 of the excavation area (tables 1-3).The pre-cores and cores' typological attributes and measurements suggest the presence of two discrete core reduction strategies: blade and bladelet / micro-blade (fig.1-6).The unidirectional sub-, cylindrical, and narrow flaking surface cores were the sources for blades.The bladelet / micro-blade assemblage is the result of narrow flaking surface cores exploitation.Both core reduction strategies are based on the intentional selection of differently shaped pieces of raw material: plate-like pieces (plaquettes, natural and primary flakes) for narrow surface flaking cores and rounded nodules for sub-, cylindrical cores.The middle-size blade (av.46.80 × 18.15 × 4.93 mm), mainly of rectangular shapes, straight lateral profiles, and feathering distal ends represents the results of blade core reduction (fig.7-14).The bladelet / micro-blade core reduction results are represented by the middle-size bladelets / micro-blades (av.25.65 × 9.18 × 2.31 mm), mainly of rectangular shapes, straight lateral profiles, and feathering distal ends.The detachment of such technological blanks as débordante and crested pieces are characteristic of both reduction strategies (tables 4-6; fig.9).The results of blade cores reduction were used for all types of tool production, except microliths.The blade cores were the source of lamellar products for the «domestic» toolkit; the bladelet / micro-blade cores reduction aimed at the insets for distant weapons.The existence of two discrete reduction strategies at Barmaky, level 2 is consistent with the technological characteristics of the Early Epigravettian of the Northern Mediterranean Basin and Prut / Dniester basins.Consequently, the spatial and chronological proximity of Barmaky to Dniester valley assemblages and the typological similarity of Barmaky and Mizyn assemblages raises several problematic issues in the investigations of the late Last Glacial Maximum industries' origin of the Mid-Dnieper basin.
whole stage of core exploitation.Relatively often (21.62-34.02%) the crested ridges were made on débordante blanks (fig.9: 2, 6, 10).The crested flakes, blades, bladelets, and micro-blades are sufficiently longer, wider, and thicker than «regular» and débordante analogs (table 5).The pattern of length distribution of the values of crested lamellar pieces is quite different from the same of «regular» and débordante lamellar products (fig.10: 1-3).The plateau of length values of crested lamellar pieces shows nearly equal percentages distribution in the range up to 54.99 mm (fig.10: 1).The plateau of width values shows a relatively equal distribution of values in the range from 4.00 up to 19.99 mm (fig.11: 1).The peak-like diagram of thickness values reflects that the crested pieces are thicker than the rest of lamellar items (fig.12: 1).

table 5 .
Barmaky, level 2, débitage:the mean values of measurements including tools, in mm

table 6 .
Barmaky, level 2: the typological structure of crested pieces including tools