Content uploaded by Nicolas Tardy
All content in this area was uploaded by Nicolas Tardy on Apr 25, 2018
Content may be subject to copyright.
Micro-blade production on hyaline quartz during the Late Neolithic of
northern Greece (5400e4600 cal. B.C.): Examples from Dikili Tash and
, J. Vosges
, B. Varoutsikos
eorient, UMR 5133, Maison de l'Orient et de la M
ee, 5 rue Raulin, 69007 Lyon, France
Anthropology Department, Harvard University, 02138 Cambridge, MA, USA
Available online 8 February 2016
The socio-economic processes during the Late Neolithic in northern Greece have been given little
attention compared to earlier phases of this period. However, several studies suggest interesting phe-
nomena such as shifts in settlement patterns and ceramic production, possibly entailing processes of
intense group interactions and increasing territorialization. However, these processes have only been
addressed through the lens of pottery production, thus only looking at a limited aspect of the economic
evolution of these groups. Acknowledging the potential of technological analysis to shed light on such
socio-economic processes, this paper focuses on a lesser known type of material, rock-crystal. The study
presents recent archaeological and experimental results on knapped rock-crystal taking place during the
Late Neolithic in northern Greece. This study emphasizes the need for a new methodological framework
to address this type of raw material, while providing a new approach to lithic production and con-
sumption on Neolithic sites. This analysis is a preliminary step towards a reconsideration of the issues
associated with the identiﬁcation of pressure knapping techniques both on rock crystal, and other types
of raw material. The presence of technical pieces, ﬂakes, debris and cores within these assemblages
questioned the possibility of a “local”pressure debitage in a regional and chronological context where
the products of a pressure blade manufacture are generally considered imported and almost exclusively
present, on consumer sites, in the form of ﬁnished products (unretouched and/or retouched blades).
©2016 Elsevier Ltd and INQUA. All rights reserved.
The Neolithic period represents a dramatic shift in interactions
between humans and their environment. In the Near East, it
approximately started between 12,000 and 10,000 years ago (Bar-
Yosef and Meadow, 2005) and quickly spread across thousands of
kilometres of land and water. The modalities of this transition in the
“Neolithized”territories have been explained following different
models ranging from total population replacement to assimilation
of agricultural technology (Ammerman and Cavalli-Sforza, 1984;
Zvelebil, 1986; Bellwood, 2005). Following the transition, the
Mediterranean area became the host of a rich network of in-
teractions where materials and ideas circulated at a pace still
difﬁcult to comprehend. In the course of this process, large regional
entities appeared in northern Greece. The development of these
entities resulted from complex mechanisms of short to long-term
and long-distance interactions, and the negotiation of knowledge
and technology transfer at different scales.
It has been demonstrated in the course of archaeological and
ethnographic studies that technological practices are strongly
embedded in the cultural and economic fabric of any society. How
technology is carried out, practiced, and transferred is function of a
given social context. Therefore, the identiﬁcation of technological
behaviours has the potential to provide information regarding
some aspects of societal organization.
However, access to such information is limited by our ability to
identify technological processes in the archaeological record.
Despite being the most ubiquitous of archaeological remain, the
potential of chipped tool assemblages to provide fragments of such
behaviours is highly dependent on the type and quality of material,
and an appropriate methodological framework allowing us to
E-mail addresses: firstname.lastname@example.org (N. Tardy), email@example.com
(J. Vosges), firstname.lastname@example.org (B. Varoutsikos).
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/quaint
1040-6182/©2016 Elsevier Ltd and INQUA. All rights reserved.
Quaternary International 424 (2016) 212e231
interpret it. These variables explains the discrepancies found in the
literature regarding raw materials studies in lithic assemblages, and
the lack of methodological framework to address more “complex”
materials such as quartz and rock crystal.
Therefore, the goal of this paper is two-fold. On one hand, we
present the results of a technological analysis on rock crystal as-
semblages from two Late Neolithic settlements in northern Greece
(Dikili Tash and Promachonas-Topolni
ca). On the other hand, we
introduce an experimental and methodological framework
providing new information on quartz-knapping behaviours (rock
crystal). The case study presented here provides new data
regarding the socio-economic developments taking place in
northern Greece during the Late Neolithic.
2. Late Neolithic I and II in northern Greece
2.1. Chrono-cultural context
The chrono-cultural sequence covered by this study roughly
spans from 5400 to 4500 cal. B.C. identiﬁed in the Greek chrono-
cultural sequence as the Late Neolithic I and II periods (Demoule
2.1.1. Ceramic production
In northern Greece, the LN I and II periods are characterized by a
territorial expansion in regions such as eastern Macedonia, Thrace,
and some Aegean Islands, as well as drastic changes in the settle-
ment patterns. Many settlements from the Early Neolithic (EN) and
Middle Neolithic (MN) are abandoned, while we observe in parallel
an increase in settlement density and a shift in settlement patterns.
An overall 20% increase in the number of sites is observed, espe-
cially in Thessaly, half of which are found on new locations
(Demoule and Perl
es, 1993:388). This type of data would suggest
relocation away from the hills towards the alluvial plains, which
might indicate profound socio-economic transformations
(Demoule, 1994:81). The LN periods in northern Greece are also
characterized by a predominance of open-air settlements on ﬂat
surfaces instead of more characteristic tell-like settlements seen in
the southern regions of Greece (Pappa, 2007:270).
Ceramic studies from LN1 sites indicate a strong inﬂuence from
the Balkans in Thessaly, western, central and eastern Macedonia,
and Thrace, while an intensiﬁcation of medium to long distance
contacts may have triggered a certain standardization visible only
in some aspects of the production (Demoule, 1993:7, 1994:82,
Demoule and Perl
es, 1993:392). Pottery productions from LN1 sites
in northern Greece generally show common stylistic traits such as
matt painting on brown decorations, black burnished pottery and
carinated forms, while displaying some degree of regional coher-
ence in the technics and ornamental decoration patterns (Demoule,
1994:82) Overall, these socio-cultural developments seem to point
towards a progressive organization of new cultural entities that will
characterize the next period.
During the second phase of LN (LN2), three large cultural en-
tities emerge in northern Greece, a phenomenon particularly wit-
nessed in the decrease in regional variability of pottery style
(Demoule, 1994:84). While Greek, Thrace and eastern Macedonia
have graphite decorations in common, characteristic of the Marica I
cultural entity, central Macedonia seems to be related to the Vinca
Culture, and Thessaly shows parallels with the “classical Dimini”
style extending to Albania (Demoule and Perl
2.1.2. Lithic production
A strong dichotomy is seen between northern Greece and the
southernmost regions in the lithic production. While lithic as-
semblages in southern LN sites are characterized by imports of
Melos obsidian and low reliance on local raw materials, northern
settlements display a different pattern where local raw materials
are predominant and obsidian raw materials found in very limited
quantities (Kourtessi-Philippakis, 2009). Although Melian obsidian
is distributed in some sites in western Macedonia, especially at Nea
Nikomedia (Rodden, 1962) and Servia (Watson, 1983), it is almost
absent from central and western Macedonia settlement, as well as
in Thrace, as if a socio-economic barrier prohibited the distribution
of this material beyond the Thessalian/Macedonian frontier (Perl
1990:22, 1992:146, Kourtessi-Philippakis, 2009).
Lithic assemblages from LN sites in northern Greece are thus
deﬁned by the use of local or regional raw materials for on-site
production of a major part of the toolkit. These local raw mate-
rials vary widely, depending on the mineral environment in which
the sites are located.
In addition, other types of raw materials are usually present in
small quantities and almost exclusively in the form of retouched
blades and bladelets, a behaviour illustrating a possible impor-
tation of material. This is especially the case with honey ﬂint,
which constitutes a small fraction of the lithic assemblage at each
site (8% in the Dikili Tash I assemblage; 26.39% in the Dimitra II
assemblage and 15% in Promachonas' Greek sector assemblage
(Kourtessi-Philipakkis, 1997:214, 2009:307). At Sitagroi II on the
other hand (73.3%; Tringham, 2003:85, Kourtessi-Philippakis,
2009:307) Tringham (2003) suggests this raw material has been
imported as preformed cores for on-site blade production. The
origin of this honey ﬂint, widespread in Greek sites, is still prob-
lematic. Although outcrops of honey ﬂint have been identiﬁed in
Northeastern B ulgaria and the nort hern bulgarian Thrace, (Gurova
and Nachev, 2008; Bonsall et al., 2010; Gurova, 2012), some less
distant sources may originate from the southern Rhodope
Mountains near Smolyan (Dimitriadis and Skourtopoulou, 2003).
3. Dikili Tash and Promachonas-Topolni
3.1. Regional context
The area discussed in this study is located between the Strymon
(Struma) river in the east to the Nestos (Mesta) in the west, from
Macedonia to Greek Thrace. The landscape in this region is largely
dominated by mountain ranges in the north (Rhodope range), and
basins and plains extending to the southern coastline. As a conse-
quence, this area is constituted of various environmental niches,
and displays drastic climatic variations from a Mediterranean
climate in the south to a more mountainous, continental climate in
Both sites are large settlements ranging from 5 ha for Pro-
ca and 4.5 ha for Dikili Tash. They are located
between the Strymon and Nestos rivers, separated by about
100 km. They indicate different settlement types, as Dikili Tash is a
tell-site situated in the Drama Basin and Promachonas is a ﬂat-site
located in the alluvial plain of the Strymon river (Fig. 1).
3.2. Dikili Tash and Promachonas-Topolni
ca, two LN sites in
Dikili Tash is a tell located in the southeasternpart of the Drama
plain in eastern Macedonia, on the periphery of the modern village
of Krinides (Kavala District), 2.5 km east of the ancient city of
Philippi. The tell lies to the southwest of a marshy pond which was
drained by a small stream that marks its eastern extension. Its
surface is estimated at about 4.5 ha (Lespez et al., 2013:31). The
stratigraphic sequence of the site delivered traces of human occu-
pation from before 6250 cal. B.C. to at least 1200 cal. B.C. (Roque
et al., 2002:630; Lespez et al., 2013:31). The richest layers are
N. Tardy et al. / Quaternary International 424 (2016) 212e231 213
attributed to the Late Neolithic periods I and II (Dikili Tash phases I
and II) which are best represented in areas V and VI, where the rock
crystal assemblages studied in this article come from.
ca is situated in the Strymon valley, at the
northern border of the Serres Basin on the Northeastern slopes of
Mount Cerkine, around 6 km east of the village of Promachonas
(Greece), and 2 km south of the village of Topolni
ca (Bulgaria). The
site is divided by the GreekeBulgarian border, and displays a
stratigraphic sequence divided into four occupational phases and
attributed to the Late Neolithic periods I and II and the Early
Eneolithic (5320e4250 cal. B.C., Koukouli-Chryssanthaki et al.,
2007:50, Vajsov, 2007:82). The rock crystal industry presented
here comes from all the occupational phases of the Greek excava-
tions (Promachon sector).
3.2.1. Lithic assemblage from Dikili Tash: overview
Recent studies conducted on the lithic assemblage from Dikili
Tash are based on the excavated remains from Sectors V (Kourtessi-
Philippakis, 2009) and VI (Kourtessi-Philippakis forthcoming,;
Kostaki forthcoming), areas that have delivered the most signiﬁcant
concentrations of lithic remains. The lithic assemblage from sectors
V and VI consists of a wide variety of raw materials. Overall, the
lithic assemblage is dominated by local raw materials like chal-
cedony, various types of local ﬂint and jasper. These local raw
materials have been preferentially used for the manufacture of
ﬂakes, irregular blades, and bladelets following various knapping
methods (Kourtessi-Philippakis, 2009:306e307). Their reduction
sequences appear to be almost complete at the site, except for the
ﬁrst stages of production.
Indirect percussion seems to play an important role for the
irregular blade and bladelet productions on these local raw mate-
rials. However, within the different varieties of ﬂint represented on
the site, the presence of laminar elements with parallel edges and
ridges and a straight proﬁle on a high quality honey ﬂint may
suggest recourse of pressure ﬂaking. This type of production is only
represented on the site by ﬁnished products (mostly fragmented
blades and bladelets) and can be the result of importation in
increasing amounts from Late Neolithic phases I to II (S
1992: 75, in Treuil, 1992). The presence of obsidian elements in
very small quantities and almost exclusively in the form of frag-
ments of blades and bladelets probably made by pressure knapping
also illustrates import as ﬁnished products from the Island of Melos
In Dikili Tash, the presence of vein quartz is also noteworthy
(5.73%). This raw material is almost exclusively represented on the
site in the form of debris, natural fragments and cobbles and
fragments of cobbles with characteristic traces of percussion ac-
tivities. Blank production does not appear to have been imple-
mented in the site. The small amount of pieces could be explained
by the accidental fragmentation of several cobbles used as per-
cussion tools (Tardy, forthcoming).
3.2.2. Lithics from Promachonas-Topolni
The study of the lithic assemblage from Promachonas-Topolni
is still ongoing. To date, only the composition of lithics from Pro-
machonas phase III (5070e4700 cal. B.C.; Vajsov, 2007:82) of the
Greek sector has been subjected to a detailed analysis (Kostaki
The lithic assemblage is dominated by local raw materials such as
chalcedony, various types of local ﬂints and vein quartz. Chalcedony,
unlike in Dikili Tash assemblages from sectors V and VI, plays a
secondary role in the manufacture of lithic implements, repre-
senting only 13.10% of the lithic assemblage from phase III. However,
it is used in the same way as at Dikili Tash, for production of ﬂakes
and irregular blades and bladelets, but no cores were uncovered.
The vast majority of the lithic assemblage consists of different
varieties of local (44.89%) and non-local ﬂint (25.43%). Among the
latter, a high quality honey ﬂint with and without white spots and
Fig. 1. Geographic situation of the settlements and potential rock crystal sources.
N. Tardy et al. / Quaternary International 424 (2016) 212e231214
predominantly represented by laminar products may indicate an
importation phenomenon, although in larger quantities than at
Dikili Tash. Except for vein quartz, which was used to produce
ﬂakes by direct percussion and bipolar reduction (Tardy
forthcoming), the vast majority of the lithic assemblage is domi-
nated by blades and bladelets. Cores are very rare, but we see a
clear tendency to transform blade cores into splintered pieces
which could partly explain this scarcity (Kostaki pers. comm. 2014).
The use of bipolar reduction is not documented on the rock crystal
3.2.3. Rock crystal assemblages from Dikili Tash and Promachonas-
ca, the analysis of the hyaline quartz
products includes every item discovered in the Greek area. In total,
32 hyaline quartz elements were uncovered and analysed, among
which 15 pieces are micro-blades (Table 1). These micro-blade el-
ements are composed of 3 complete micro-blades and 12 frag-
ments (8 proximal fragments, 3 medial fragments and 1 distal
fragment). The micro-blades' width is between 4.86 and 9.33 mm
while thickness never exceeds 3.18 mm (Fig. 5). Two micro-blades
present traces of use or retouch: a medial fragment has semi steep
inverse retouch on the left edge and a proximal fragment presents
traces of use on the medial part of its right edge (Fig. 2:6).
Among the 11 ﬂakes that compose the assemblage, three are
directly linked with the micro-blade reduction sequence. One may
have played the role of a platform opening ﬂake. This ﬂake's butt as
well as its lateral proximal edges are composed of natural faces of
the prism (Fig. 3:3). The other one is a rejuvenation core ﬂake
(Fig. 3:4). This is a small ﬂake with no debordant back (character-
istic of the core tablets) but with a slightly hinged distal end and a
facetted butt resulting from the micro-blade removals. The last
piece is a crested piece that raises the question of a possible shaping
of the crystals (Fig. 3:5). This crested element is elongated and
presents much wider dimensions than the rest of the products
(40.27 15.75 9.09 mm). Its dimensions, irregular morphology,
and partial and asymmetrical crest suggests a function of platform
opening rather than a function of knapping surface opening.
Fig. 2. Micro-blades from Promachonas-Topolni
ca (1e9) and Dikili Tash (10e11).
N. Tardy et al. / Quaternary International 424 (2016) 212e231 215
Furthermore, the location and inclination of the three residual
natural facets of the prism on its lateral edges evoke three adjacent
surfaces of a pyramidion, which conﬁrms the assumption of an
elongated platform opening ﬂake.
This shaping seems to concern only the apical end of the crystal
in order to create a platform. Taking into account the natural
morphology of a well-developed rock crystal, it is possible to use the
ridges of the natural prism or those of the pyramidion to begin the
micro-blade production immediately after the creation of a plat-
form (Desrosiers, 2007:25) or by starting the production from the
bottom of the crystal in case of crystals that are not bi-terminated
(Desrosiers and Sørensen, 2012:390). The use of a laborious and
consuming shaping phase does not appear necessary, at least in the
case of well-developed rock crystal prisms. Nevertheless, some rock
crystal crested micro-blades found among Palaeoeskimo assem-
blages document the possibility of a shaping of the crystals before
micro-blade production (Desrosiers and Sørensen, 2012: 384).
Two small micro-blade cores are also found in the assemblage
(14.36 13.38 14.26 mm and 15.11 6.52 9.09 mm; Fig. 3:1
and 2). Their method of preparation is similar, based on a faceting of
the striking/pressure platform by the removal of small chips, some
of them being slightly hinged. The striking/pressure platform is
originally slightly inclined, although the faceting process tends to
make it rather orthogonal to the ﬂaking surface.
At Dikili Tash, hyaline quartz only constitutes a small fraction
(2.15%) of the lithic assemblage from the Late Neolithic layers. It is
represented by 21 pieces including 4 micro-blades or fragments
of micro-blades (Table 1). The size of these elements is between
4.1e6.5 mm width and 1.5e2.1 mm thick (Fig. 5). A complete micro-
blade gives an indication of length: 25.3 mm. This complete micro-
blade is also the only element with traces of use in the form of two
lateral blunted edges (Figs. 2: 10 and 4).
No cores have been found during excavations, although 2 nat-
ural crystals and 4 crystal fragments were uncovered. Although
Fig. 3. Other rock crystal elements from Promachonas-Topolni
ca (1e5) and Dikili Tash (6); cores: 1e2; platform opening ﬂake: 3; rejuvenation platform ﬂake: 4; platform opening
(?) crested piece: 5; fragmented monocrystal: 6.
N. Tardy et al. / Quaternary International 424 (2016) 212e231216
these elements show no signs of knapping, they appear to have
been collected for this type of micro-blade production. They all
present similar dimensions (from 19.1 to 25.3 mm long, between
11. 2e17.2 mm wide and 5.2e12.3 mm thick), and the four mono-
crystal fragments show a break at one end which could have played
the role of a striking platform, but no blade removal is present on
the crystals' facets (Fig. 3:6). The rest of the rock crystal elements
consist of 2 ﬂakes, 4 debris and 5 natural fragments.
Some biases are likely to affect the representativeness of hyaline
quartz pieces collected on these two sites. Both sites were partially
excavated and sieving was not systematically carried out. As for
ca, only the hyaline quartz products from the
Greek excavation have been analysed. The material from the
Bulgarian excavation team was not taken into account as it could
not be accessed.
3.2.4. Morpho-technological characteristics of the micro-blades
from Dikili Tash and Promachonas-Topolni
This micro-blade production shows a very similar set of techno-
morphological characteristics on both sites. As each site delivered
only a few rock crystal elements, we chose to gather both micro-
blade assemblages and present them as a whole in order to
deﬁne the modes of micro-blade production on rock crystal during
the Late Neolithic I and II periods in northern Greece.
Among the 19 micro-blades, proximal fragments dominate the
set, although none is present in Dikili Tash's assemblage (8 ele-
ments). Distal fragments are scarce (only 2 pieces) and medial
fragments and complete blades are present in similar proportions
(respectively 5 and 4 elements).
Proximal parts are characterised by a high frequency of diffuse
(6 on 12 pieces) to very diffuse bulbs (5 on 12 pieces) and no lip or
bulbar scars have been observed. Likewise, no ring cracks that could
suggest the use of a copper tip have been identiﬁed on the butts of
The removal preparations mainly consist of a micro-ﬂaking of
the platform. This type of preparation delivers micro-blades with
facetted or dihedral butts (7 elements). Overhang abrasion is also
practiced but not systematically (3 pieces show traces of overhang
abrasion; Fig. 2:1, 4). Finally, butts, when visible, are exclusively in
an orthogonal position with respect to the ﬂaking surface.
The assemblage consists of 9 products with two ridges (Fig. 2:2,
3, 4, 8, 10). Micro-blades with one ridge are nevertheless common
(6 pieces, Fig. 2:1, 7) and some products with more than two ridges
are also reported (4 pieces, Fig. 2:11).
The proﬁles of these micro-blades is either slightly curved (7
pieces; Fig. 2:6, 10) or rectilinear (6 pieces; Fig. 2:1, 4, 7). Some
pieces show a slightly more curved proﬁle on the distal part
(Fig. 2:2, 8, 11). Two micro-blades having a particularly curved and
slightly twisted proﬁle are also present among Promachonas-Top-
ca's assemblage (Fig. 2:9).
Generally speaking parallel edges and ridges which constitute
one of the principles identiﬁcation criteria of a pressure blade
production (Inizan et al. 1995, p.79e80; Tixier, 2012, p.142) are not
systematic (7 elements; Fig. 2:2, 4, 7, 10). In contrast, elements
showing irregular edges and ridges tend to dominate these
archaeological assemblages (Fig. 2:1,5, 6, 8, 9).
The micro-blade elements from Dikili Tash and Promachonas-
ca, numbering 19, have widths between 4.1 and 9.33 mm
Fig. 4. Complete micro-blade from Dikili Tash.
Fig. 5. Comparisons between width and thickness of the complete and fragmented micro-blades from Dikili Tash and Promachonas-Topolni
N. Tardy et al. / Quaternary International 424 (2016) 212e231 217
with a mean of 5.49 mm and thickness is between 1.08 and 3.18 mm
with a mean of 1.89 mm (Figs. 5,14,15).
Among all these micro-blade elements, only 2 pieces show re-
sidual prismatic surface on a debordant edge (Fig. 2:4, 5). In view of
the limited number of pieces showing residual prismatic surfaces, it
is clear that we are not in the presence of all the elements of the
eratoire for this micro-blade production on both sites.
Nevertheless, the presence of a few ﬂakes, technical pieces, cores
and debris indicates that at least part of the production has been
made in situ, especially in the case of Promachonas-Topolni
The acquisition of the crystals is mainly directed towards a
gathering of monocrystals in primary position. However, several
elements from Promachonas-Topolni
ca have residual traces of
rolled surfaces reﬂecting a collection in the form of pebbles in
secondary position in the river alluvium. As to the location of the
rock crystal sources, some pegmatite veins with rock crystals have
been identiﬁed in the Central Rhodopes Mountains in the Arda
region (Bulgaria) (Atanasova-Vladimirova and Sekiranov, 2007),
and the museum of Karst in Chepelare (Bulgaria) provides good
examples of rock crystal pieces found in this crystalline mountain
range. This area is situated around 120 km from Promachonas-
ca and around 60 km from Dikili Tash, and more regional
potential sources are to be expected, especially for the case of
ca. The discovery of a rolled rock crystal
pebble in the Bistrica river near the Neolithic village of Kovacevo
situated around 20 km from Promachonas-Topolni
ca leaves little
doubt as to the presence of this raw material on a regional scale
from the site (Fig. 1).
A phenomenon has been observed on a few hyaline quartz
products: ampliﬁcation, and sometimes, deformation in the
recording of the concentric ripples on the fracture faces of some
products. Thirteen elements are affected by this phenomenon.
Overall, this phenomenon seems to particularly affect thicker
pieces, mostly ﬂakes, but the presence of these altered and
ampliﬁed ripples on the inferior face of a micro-blade created an
irregular edge evoking a ‘natural’denticulation (Fig. 2:1). A core is
also affected on its platform surface (Fig. 3:2). This phenomenon,
which is either diffuse or very marked, is an expression of the
anisotropic nature of rock crystal.
4. Setting the experiment
The small amount of hyaline quartz uncovered among the LN
layers of Promachonas and Dikili Tash indicates a technology based
on micro-blade production. This technology, as well as the use of
rock crystal for the manufacture of lithic implements raises the
question of the knapping techniques employed to produce them.
First, we provide some introductory comments about the
Fig. 6. Techniques and tools used for the experiments (part 1).
N. Tardy et al. / Quaternary International 424 (2016) 212e231218
mineralogy and petrography of hyaline quartz. An experiment was
then conducted on some rock crystals of diverse size and shape
using various knapping techniques (direct percussion using hard
and soft hammers, indirect percussion with a punch, several modes
of pressure ﬂaking) in order to identify the potential techniques
that could generate a micro-blade production. One person (J.V.)
conducted the experiment in order to preserve a relative homo-
geneity within the experimental assemblage thus produced. The
results from the lithic analysis and the experiment lead to reﬂection
on the problems related to the identiﬁcation of pressure knapping
techniques on hyaline quartz as well as on the organization of
technology and lithic production in a Neolithic context.
4.1. Quartz, an introduction
Hyaline quartz, also known as rock crystal, constitutes one of the
crystallized varieties of the mineral quartz under its automorphic
form (Foucault and Raoul, 2005 (1980)). Hyaline quartz occurs, in
its most common conﬁguration, in the form of a prism topped by a
pyramidion. These rock crystals can be bi-terminated or not, and
can also be twinned such as the twins from Gardette or Dauphin
(France), Japan, and Brazil (Boudeulle et al., 1979:18).
Hyaline quartz is not isotropic. Its anisotropy results mainly
from the presence of cleavage planes and diaclasis within it.
Cleavage planes are related to the weakness of some atomic bonds
in the crystal lattice while diaclasis are caused by external factors:
solid, gaseous or liquid inclusions on the crystal facets during its
formation (Novikov and Radililovsky, 1987:594), or mechanical
stresses due to the host rock properties during tectonic movements
(Mourre, 1996:206). This anistropic character inevitably affects the
rock's knapping behaviour, and Neolithic knappers have necessarily
taken into account this particularity.
4.2. Techniques and raw materials
In an attempt to identify the techniques that could generate
micro-blade production on hyaline quartz, an experiment was
conducted using various knapping techniques. Among all technical
Fig. 7. Techniques and tools used for the experiment (part 2).
N. Tardy et al. / Quaternary International 424 (2016) 212e231 219
ways to produce bladelets and micro-blades, we selected direct
percussion with a soft hammer (antler), indirect percussion tech-
nique with an antler punch, and 5 different modes of pressure
debitage: pressure technique in the hand with an antler stick,
without and with the use of a holding device (corresponding to
Pelegrin's modes 1 and 1b; Pelegrin, 2012); pressure technique in
the hand with a shoulder crutch and a holding device (corre-
sponding to Pelegrin's mode 2); pressure technique in a sitting
position with a small abdominal crutch and a holding device rested
on the ﬂoor (corresponding to Pelegrin's mode 3); and pressure
technique in a standing position with a long crutch and a holding
device rested on the ﬂoor (similar to Pelegrin's mode 4) (Figs. 6 and
7). The few publications dealing with micro-blade productions on
hyaline quartz usually mention the use of pressure debitage in a
broad sense, for this type of production (Flenniken, 1981:108,
Honegger, 2011:173), although other techniques are suggested
(Desrosiers and Sørensen, 2012:381e382). There is, however, a
great diversity in the modes of application of a pressure debitage
and we considered 4 modes among the 5 modes proposed to date
(Pelegrin, 1988, 2012) in order to identify the techniques that are
likely to produce these micro-blades. Mode 5 (pressure debitage
with a lever) was directly excluded from the experimental sets as
this technique is usually used to produce long blades. Soft direct
and indirect percussion techniques have been introduced in this
experiment in order to check if a micro-blade production based on
the use of these two techniques was possible.
The main goal of these experiments was to propose techniques
that could generate micro-blades within a range of well-known and
used knapping techniques. Fourteen rock crystal blanks of various
sizes and shapes were then collected for this purpose (Fig. 8).
Shapes comprise well-developed monocrystals of different sizes
(from 4.75 1.2 0.9 cm to 11.1 4.1 3.4 cm), some being bi-
terminated, as well as fragments of crystals, ﬂakes, debris, a pol-
ished block and a rolled pebble. The provenance of the raw mate-
rials used for the experiments were the island of Madagascar for
the largest specimens and the Italian Alps for the smallest bi-
terminated crystals. The choice to use raw materials with a great
diversity of sizes and shapes was motivated by the lack of reliable
information regarding sizes and shapes of the rock crystals
collected by the prehistoric populations of Dikili Tash and Pro-
machonas. The presence of residual prismatic surfaces on a few
archaeological pieces indicates that some prismatic crystals were
collected at the source, while the presence of some pieces with
rolled surfaces might indicate a collection in the form of small
Fig. 8. Raw materials used for the experiments: raw materials from Madagascar (1) and the Italian Alps (2).
Fig. 9. Comparisons of width and thickness between archaeological products and the three non-conclusive experimental techniques.
N. Tardy et al. / Quaternary International 424 (2016) 212e231220
pebbles as well. The only elements concerning sizes of the raw
materials collected by the prehistoric knappers, come from 2
monocrystals and 4 monocrystal fragments that apparently were
not knapped from Dikili Tash. Lengths of these six pieces are be-
tween 1.91 and 2.53 cm, while maximum widths are between 0.93
and 1.72 cm and minimum widths are between 0.52 and 1.23 mm
(Fig. 3:6). Furthermore, two technical elements from Promachonas-
ca (Fig. 3:3, 5) might indicate that larger prisms were also
available and collected.
One of our intentions was also to create an experimental series
for each technique with good control over all the variables. Those
series could then provide statistical information on metrics and
also qualitative dataset to guide more precise determination of
typical and atypical stigmata for each techniques used. However,
the use of speciﬁc variables to allow the identiﬁcation of different
techniques on the rock crystal micro-blade soon became difﬁcult to
exploit. This was mainly due to the small sample size experimen-
tally produced and archaeologically found. However, it appeared
that the dimensional factor, especially the consideration of widths
and thicknesses is a discriminating factor for the identiﬁcation of
We therefore must consider the results from these experiments
as preliminary. Further experimental production for each technique
is needed in order to produce more statistically viable datasets.
4.2.1. Inconclusive techniques
Three techniques have been experimentally tested and dis-
carded: soft direct percussion, soft indirect percussion and pressure
technique with an abdominal crutch in a standing position. The
products manufactured with these techniques did not match the
format witnessed in the archaeological assemblages.
Fig. 9 compares widths and thicknesses between the archaeo-
logical products and the three experimentally tested techniques;
soft direct percussion, soft indirect percussion and pressure tech-
nique with an abdominal crutch in a standing position. It clearly
appears that the majority of the experimental products have much
larger dimensions than the archaeological ones. A few experimental
products, among the smallest, show dimensions in the range of the
archaeological products. However, these experimental products,
numbering 4 (3 soft direct percussion products and 1 indirect per-
cussion product) on a total of 102 pieces represent an extreme of the
production's dimensional variability and certainly not the norm. We
will nevertheless give some insights concerning the technological
and morphological characteristics of the experimental products
from each of these discarded techniques.
70 blades were produced using soft direct percussion, from
which 35 products were complete and 35 products were frag-
mented. Three cores were used for this experiment; a monocrystal
(7.48 4.55 3.7 cm) and two fragments without any prismatic
Fig. 10. Experimental micro-blades and cores using modes 1 (top) and 1b (bottom).
N. Tardy et al. / Quaternary International 424 (2016) 212e231 221
Fig. 11. Experimental micro-blades and cores using modes 2 (top) and 3 (bottom).
Fig. 12. Experimental bladelets, blades and core using mode 3.
N. Tardy et al. / Quaternary International 424 (2016) 212e231222
surfaces (8.3 4.4 2 cm and 6.3 4.42 3.72 cm). The mono-
crystal was knapped using the prismatic edges to guide the ﬁrst
products. The two other cores were shaped prior to bladelet debitage
in order to start the production from a ridge. The striking platform
was slightly inclined for the three cores and overhang abrasion as
well as faceting of the platform was used in similar quantities. The
tools used to produce this set were a soft stone hammer (limestone)
of 474 g and an antler of white-tailed deer of 260 g (Fig. 6:1).
The main technological attributes on this bladelet production
are: the frequent presence of a bulbar scar (37.14% of the bladelets),
the presence of a lip on some products (14.28%) as well as the
presence of radial striations similar to hackles observed on the
bulbs (24.28%). Furthermore, diffuse bulbs are predominant (50%)
although very diffuse bulbs are well represented (31.42%). Promi-
nent bulbs are uncommon (10%) (Table 2).
As for the main morphometrical characteristics, we observed a
predominance of irregular edges (78.57%), small quantities of
convergent and divergent edges (respectively 11.43 and 7.14%) and
no bladelets with parallel edges. The products proﬁle is mostly
regularly curved (64.28%). There was also some twisted pieces
(14.28%) and a few pieces with a rectilinear proﬁle (12.86%)
These laminar products have much larger dimensions than the
archaeological ones (Fig. 9). Widths range from 6.2 to 24.68 mm
with a mean of 14.97 mm, while thicknesses range from 1.14 to
11.62 mm with a mean of 4.06 mm (Fig. 9).
Technological and morphological attributes, as well as dimen-
sional ranges, differ greatly from the archaeological products
(Table 2). This technique was therefore excluded from the potential
techniques that could generate a micro-blade production.
20 pieces have been produced using indirect percussion with an
antler punch. Within these blades, only 3 pieces were complete,
and the 17 other products were fragmented during detachment.
Two cores have been used for this experiment; a monocrystal
fragmented at both ends (9.6 3.45 2.65 cm) and a rectangular
sawn slab (7.8 41.9 cm). The bladelet production started from a
natural ridge to guide the ﬁrst bladelet on both cores. Faceting of
the striking platform was performed on the monocrystal while the
rectangular block was only prepared by overhang abrasion. Striking
platforms were slightly inclined on both cores.
The tools used to produce this set were three antler punches of
52.2 g, 32.3 g and 20.4 g and a wooden indenter (boxwood) of 478 g
(Fig. 6:2). The technological characteristics of the products are: a
frequent presence of a lip (75%), a lesser presence of bulbar scar
(15%) and no radial striations. Bulbs are mainly prominent (45%)
Technological composition of the rock crystal assemblages from both sites.
Micro-blades and fragments Flakes and fragments Debris Cores Natural fragments Monocrystal fragments Complete monocrystals Total
Dikili Tash 4 2 4 0 5 4 2 21
ca 15 11 4 2 0 0 0 32
Total 19 13 8 2 5 4 2 53
List of techno-morphological attributes of the archaeological and experimental products.
Bulbar scar 0 0 0 0 3 8 26 3
Lip 0 3 5 2 34 5 10 15
Radial striations 0 0 0 0 4 0 17 0
Bulbs Very diffuse 5 14 11 2 9 2 22 3
Diffuse 6 8 8 10 34 6 35 8
Prominent 1 0 0 1 2 3 7 9
Absent 7 1 5 0 3 1 6 0
Edges Parallel 7 4 0 7 2 2 0 2
Convergent 1 4 5 3 9 1 8 5
Divergent 4 1 2 3 5 1 5 6
Irregular 1 14 17 0 32 8 55 7
Total 19 23 24 13 48 12 68 20
Proﬁls Straight 6 3 3 11 17 7 9 4
9 11 18 2 28 2 45 11
Twisted 2 8 3 0 2 1 10 1
Irregular 1 1 0 0 1 1 4 4
Undeﬁned 1 0 0 0 0 1 2 0
Total 19 23 24 13 48 12 70 20
Total products 19 23 24 13 48 12 70 20
Number of experimental products obtained with each mode and their respective fracture rate.
Techniques Number of products Complete products Fragmented products
Soft direct percussion 70 35 35
Soft indirect percussion 20 3 17
Mode 1 23 5 18
Mode 1b 24 6 18
Mode 2 13 0 13
Mode 3 42 21 27
Mode 4 12 1 11
N. Tardy et al. / Quaternary International 424 (2016) 212e231 223
although diffuse bulbs are well represented (40%) and very diffuse
bulbs are rare (15%).
The morphometric characteristics indicate a signiﬁcant pres-
ence of divergent and irregular edges (respectively 30% and 35%).
Convergent edges are present (25%) and parallel edges are anec-
dotic (10%) (Table 2).
The widths of these products are between 5.82 and 29.11 mm
with a mean of 14.02 mm, while the thicknesses range between
1.99 and 12.39 mm with a mean of 6.49 mm. The dimensions as
well as technological and morphometric attributes of these prod-
ucts are quite distinct with respect to the archaeological products
Only 12 blades (1 complete blade and 11 fragmented blades)
have been produced using Mode 4: pressure debitage in a standing
position with a long crutch and a holding device rested on the ﬂoor,
using one core from a large and well-formed monocrystal
(11.1 4.1 3.4 cm) with sawn base. The blade production was
undertaken using the ﬂat and polished base as a pressure platform
in an orthogonal position to the ﬂaking surface. The production
started using the prismatic edge to guide the ﬁrst blade. Prepara-
tion prior to debitage consisted mostly of overhang abrasion.
The tools used to produce this experimental series were a large
crutch measuring 840 mm and 650 g. The crutch was armed with a
copper tip (used to produce 10 blades) and an antler tip (used to
produce 2 blades). The holding device consisted of a wooden shoe
measuring 37 cm length. Its morphology is that of a ﬂat sole with 2
uprights parts separated by a central hollow of 2 cm wide. The
uprights allowed support of the core and the hollow enabled the
ejection of the blade. In order to keep the core motionless, an
adjustable small piece of hard wood locked the distal part of the
core through different perforations on the sole. Despite the fact that
this system allows blocking cores of different sizes, it is not suitable
for small cores measuring less than 5 cm length and 3 cm width
The technological characteristics of this production consist of a
high frequency of bulbar scars (8 pieces on 12) as well as the
presence of a lip (5 pieces). No radial striations have been observed.
Bulbs are mostly diffuse (6 pieces) although a few prominent (3
pieces) and very diffuse bulbs (2 pieces) are present.
The morphometric attributes are: mostly irregular edges (8
pieces) although two pieces show parallel edges and two other
pieces show convergent/divergent edges. Furthermore, most pieces
show a straight proﬁle. A few blades show a regularly curved proﬁle
with a curved distal end, and twisted, irregular and undeﬁned
proﬁles are represented by one piece each (Table 2). The dimen-
sional range is probably the most discriminating factor with the
archaeological products, as widths are between 9.99 and 29.34 mm
and thicknesses are between 3.16 and 9.85 mm (Fig. 9).
4.2.2. Conclusive techniques
Four other techniques have successfully produced artefacts that
ﬁt the morphometry of micro-blades from Promachonas-Topolni
and Dikili Tash. We have used Pelegrin's modes (2012) to describe
220.127.116.11. Mode 1: pressure debitage in the hand with a stick eno
holding device (Fig. 10 top). 23 micro-blades were produced using
this mode of pressure ﬂaking corresponding to Pelegrin's mode 1
(Pelegrin, 2012). Two cores in the form of a thick ﬂake
(4.10 2.20 1.42 cm) and debris (2.52 1.58 1.12 cm) were
used to produce these artefacts. After creating a pressure platform
on top of a rectilinear edge on each core, the micro-blade produc-
tion started using the edge as a guide for the ﬁrst micro-blade. The
preparations prior to ﬂaking consisted of a systematic faceting of
the pressure platform by micro-ﬂaking with a pressure tip before
each micro-blade removals. Overhang abrasion was also used when
needed. The striking platform was in an orthogonal position in
respect to the ﬂaking surface on both cores. The ﬁrst core delivered
14 micro blades, the second one produced 9 micro-blades. The tool
used to produce this set was an antler stick, 28 g and measuring
150 mm as well as a piece of leather to protect the palm of the hand
Out of these 23 products, only 5 entire micro-blades were pro-
duced. The 18 other micro-blades were fragmented during
detachment. The widths of the experimental micro-blades ranges
from 3.37 to 7.12 mm with a mean of 4.88 mm, while the thick-
nesses ranges from 0.99 to 2.18 mm with a mean of 1.49 mm. The
distribution of widths and thicknesses of the experimental prod-
ucts is included, almost in its entirety in the dimensional range
given by the archaeological products (Fig. 13). It seems however,
that some archaeological products among the most robust are not
represented in the dimensional diagram of the experimental
products (Fig. 13). A maximum limit in the thickness of the latter is
around 2.18 mm. Six archaeological products exceed this thickness
Fig. 13. Comparisons of width and thickness between archaeological products and the four experimental techniques that could generate micro blade production.
N. Tardy et al. / Quaternary International 424 (2016) 212e231224
threshold. Similarly, experimental micro-blades hardly exceed
7 mm in width as 3 archaeological products have widths situated
among this limit. This suggests that this knapping technique is
capable of producing most of the micro-blades from Promachonas-
ca and Dikili Tash, but not the entire assemblage. Experi-
ments by Pelegrin (2012) and Callahan (1985, cited by Pelegrin,
2012:468) on this pressure knapping technique applied to ﬂint
have delivered dimensional limits similar to ours on hyaline quartz.
Most of their experimental products have widths around 5 mm
with a maximum boundary between 7 and 8 mm (Pelegrin,
2012:468). However, in our case, the statistically small quantities
of both archaeological and experimental products call for caution.
These dimensional diagrams based on limited quantities of ana-
lysed products. It seems therefore difﬁcult, given the quantity of
products to further the analysis by taking only the dimensional
factor into account.
On the other hand, techno-morphological attributes of the
experimental production deliver valuable information for com-
parison with archaeological products. Regarding knapping attri-
butes, there is a total absence of bulbar scars and radial striations
and an anecdotic presence of a lip (3 products). Very diffuse bulbs
seem to characterize this type of production (14 pieces on 23),
although the presence of diffuse bulbs is well observed (8 pieces).
These attributes show a general similarity with the archaeological
Edges and ridges are irregular for most of the pieces (14 pieces).
However, 4 products with parallel edges are present.
By contrast, the products proﬁles of the experimental corpus are
characterized by a dominance of twisted products (8 pieces) as well
as the scarcity of rectilinear products (3 elements only). This in-
dicates strong differences compared to the archaeological assem-
blage. The proﬁle's distribution of the latter is characterized by a
majority of regularly curved products (9 pieces including 2 regu-
larly curved pieces with a sharp distal curvature) and a signiﬁcant
presence of rectilinear products (6 pieces) for only 2 twisted ele-
ments (Table 2).
If the technological attributes are relatively similar to those of
the archaeological micro-blades, the morphological characteristics
such as irregular edges and ridges and twisted proﬁles are quite
distinctive in relation to what is observed on the archaeological
products. Because of the small number of observed products, these
data are to be taken as indications.
18.104.22.168. Mode 1b: pressure debitage in the hand with a stick e
holding device (Fig. 10, bottom). 24 micro-blades were produced
using this technique, which is similar to Pelegrin's mode 1b
(Pelegrin, 2012). Two cores in the form of a proximal fragment of
ﬂake (3.05 3.44 1.28 cm) and a debris (3.22 2.47 2.52 cm)
were used for the production of these micro-blades. On both cores,
the micro-blade production started using a rectilinear edge as a
guide for the ﬁrst product. The pressure platforms were both
slightly inclined and the preparation of the removals systematically
consisted of a faceting of the pressure platform in order to make it
orthogonal to the debitage surface. Overhang abrasion was also
occasionally carried out. The ﬁrst core delivered 14 micro-blades
while the second core delivered 10 micro-blades.
The tools used to produce this experimental series are an antler
stick (l.146 17 11 mm, 30 g) as well as a holding device in the
form of a fragment of a branch measuring 7 cm length and 3.5 cm
diameter. This branch has an interior groove measuring 1.5 cm wide
and 1.5 cm deep on its length. This device allows holding small size
cores from small monocrystals or ﬂakes (Fig. 7:2).
Only 6 micro-blades out of 24 were whole, the 18 others were
fragmented during detachment. The width of the micro-blades
produced using this technique is between 3.59 and 9.19 mm with
a mean of 5.75 mm. The thicknesses range from 0.68 to 3.49 mm
with a mean of 1.73 mm (Figs. 13e15). Dimensional ranges and
means from both this experimental series and the archaeological
products are very similar. The boxplot diagrams (Figs. 14 and 15)
shows better similarity in the dimensional distribution between
these experimental micro-blades and the archaeological ones. If
only the dimensional factor is taken into consideration, this tech-
nique could be considered as the best candidate for the micro-blade
production of the Promachonas-Topolni
ca and Dikili Tash
The techno-morphological attributes of this type of production
The presence of a lip on some products (5 pieces) and the
absence of bulbar scars and radial striations.
The predominant presence of very diffuse bulbs (11 pieces)
although diffuse bulbs are quite present (8 pieces)
Predominantly irregular edges and ridges (17 pieces with
irregular edges) and the absence of pieces with parallel edges
Fig. 14. Boxplot diagram comparing widths between archaeological and experimental products that could produce micro-blades.
N. Tardy et al. / Quaternary International 424 (2016) 212e231 225
A majority of pieces with a regularly curved proﬁle (18 pieces)
and a low presence of pieces with a rectilinear or twisted proﬁle
(3 pieces respectively) (Table 2).
The techno-morphological attributes of these experimental
micro-blades diverge somewhat from those observed on the
archaeological pieces in the absence of elements with parallel
edges and ridges and the relatively small amount of pieces with a
straight proﬁle. However, the presence of signiﬁcant quantities of
very diffuse and diffuse bulbs in addition to the dimensional range
very similar to that of the archaeological products allows us to
consider this pressure knapping technique as one of the most likely
for the production of micro-blades from Dikili Tash and Pro-
The presence of twisted elements, which were predominant in
mode 1, decreases signiﬁcantly in favour of pieces with a regularly
curved or straight proﬁle. As both modes 1 and 1b were carried out
on ﬂakes or debitage core, differences in results could be connected
to the holding device. It seems that the use of a holding device
produces increased regularity observed on the products from this
22.214.171.124. Mode 2: pressure debitage in the hand with a shoulder crutch
and a holding device (Fig. 11, Top). Only 13 micro-blades have been
produced using this technique similar to Pelegrin's mode 2
(Pelegrin, 2012). Two cores in the form of small bi-terminated
monocrystals have been used for this experiment
(4.78 1.3 0.95 cm and 4.75 1. 2 0.9 cm). One monocrystal
has been fragmented, and the fracture surface was then used as a
pressure platform to produce the blanks using a prismatic edge
as a guiding ridge. The other monocrystal has been exploited
directly from one of the pyramidion also using a prismatic edge
to guide the ﬁrst bladelet. Preparation prior to debitage consisted
of a systematic faceting of the pressure platform in order to keep
it orthogonal to the ﬂaking surface, as well as an overhang
abrasion only when needed. The ﬁrst core produced 6 micro-
blades, and the second one produced 7 micro-blades. All of the
micro-blades produced using this technique were fragmented
The tools used to produce this set are a L-shaped wooden
shoulder crutch of 160 g and measuring 300 mm and armed with
an antler tip. The holding device is the same as the one used for the
previous mode (Fig. 7:3).
The widths of the products range from 3.86 to 5.55 mm with a
mean of 4.63 mm. The thicknesses are comprised between 1.01 and
2.31 mm with a mean of 1.71 mm (Figs. 13e15). All of these prod-
ucts have dimensions in the range of the archaeological ones.
However, it became difﬁcult, in the context of our experiment, to
exceed the threshold of 6 mm width when using this technique.
This threshold is far exceeded by some archaeological products as
well as with experiments using this technique on ﬂint (Pelegrin's
products can reach 10 mm width; Pelegrin, 2012:469). These dif-
ferences are due to several factors, among which the morphology of
the cores selected for the experiment could have played an
important role. Indeed the two crystals used were not exceeding
13 mm wide and the dimensions of the cores strongly affected
those of the products removed.
The use of cores with reduced width and the small amounts of
pieces produced during the experiment does not allow us to ac-
count for the limits of this technique. However, micro-blades
similar in dimensions to the archaeological elements can be pro-
duced using this technique.
At this stage of analysis, it is safer to consider this technique as a
possible candidate for the production of micro-blades on hyaline
quartz such as the ones from Dikili Tash and Promachonas-Top-
ca. It is clear that this technique needs further tests to deliver
reliable results. A more complete experiment should allow dis-
cussions of the limitations of this technique on rock crystals in
Techno-morphological attributes of the production are:
A weak presence of a lip (on 2 elements) and the absence of
bulbar scars and radial striations.
Predominance of diffuse bulbs (10 pieces) and a very low pres-
ence of very diffuse bulbs (2 elements).
Edges and ridges predominantly parallel (7 pieces). No pieces
with irregular edges and ridges have been observed.
Straight proﬁles are largely predominant with 11 pieces out of
13, including 3 pieces with a sharply curved distal end. No pieces
with a curved proﬁle were produced (Table 2).
Techno-morphological attributes differ from those observed on
the archaeological products only by the predominant presence of
diffuse bulbs as well as the total absence of pieces with irregular
edges. The other attributes of this production are represented in
similar amounts to the archaeological micro-blades. The small
Fig. 15. Boxplot diagram comparing thicknesses between archaeological and experimental products.
N. Tardy et al. / Quaternary International 424 (2016) 212e231226
amount of experimental micro-blades does not permit a more
thorough comparison. However, this technique, when applied to
small size rock crystals, clearly appears as one of the feasible ways
of producing micro-blades on hyaline quartz.
126.96.36.199. Mode 3: pressure debitage with abdominal crutch in a sitting
position (Figs. 11, bottom and 12). 48 pieces were produced ac-
cording to this pressure mode which is equivalent to Pelegrin's
mode 3 (Pelegrin, 2012). Of these 48 pieces, 21 were complete
bladelets and 27 were fragmented.
Three cores have been used for this experimental production. A
small monocrystal (6.12 1.5 0.78 cm) produced only one bla-
delet, because its small width was not suitable for the holding
device. Two other cores in the form of a fragmented pebble
(3.55 3.2 2.85 cm) and the monocrystal already used for the
production of blades using mode 4 were also employed. The frag-
mented pebble produced 22 bladelets and the monocrystal pro-
duced 25 bladelets. The rolled pebble was shaped in order to
present a suitable edge for the guidance of the ﬁrst bladelet using a
faceted and orthogonal pressure platform. The monocrystal was
exploited from the pyramidion after the detachment of an opening
pressure platform ﬂake. The ﬁrst bladelet was produced using the
prismatic edges of the prism. Both pressure platforms were
orthogonal to the ﬂaking surface and were both prepared by sys-
tematic faceting as well as overhang abrasion when needed.
The tools used to produce this set were a small crutch of 190 g,
measuring 423 mm in length, armed with an antler tip. The holding
device consisted of a thick wooden board of about 20 cm length,
8 cm width and 6.5 cm thickness. A fork-shaped piece of wood was
recessed obliquely into the board so that both forks maintained the
core sufﬁciently. A slight widening of the board at the place where
the distal part of the core rests allowed it to remain motionless
while holding it with one hand (Fig. 7:4).
Widths range from 3.17 to 15.03 mm with a mean of 9.23 mm.
Thicknesses range from 0.83 to 5.87 mm with a mean of 2.29 mm
(Fig. 13). The mean values of the experimental production are much
higher than the mean values provided by the archaeological as-
semblages. More than half of this experimental production has
dimensions within the dimensional range given by the archaeo-
logical micro-blades (26 pieces on 48, Figs. 14 and 15), making this
technique a possible mode of micro-blade production. On the other
hand, a large proportion of the experimental set has much larger
dimensions than the archaeological products, some being bladelets
and blades rather than micro-blades (Fig. 12).
These experimental products frequently present a diffuse bulb
along with a lip (34 pieces). This combination of attributes is not
observed on the archaeological pieces. A few bulbar scars and radial
striations have also been counted (respectively 3 and 4 pieces)
Edges and ridges are predominantly irregular (32 and 26 pieces
respectively) and pieces with parallel edges and ridges are rela-
tively scarce (2 and 4 elements respectively). J. Pelegrin observed an
improvement in the regularity of the blades' edges and ridges in
contrast to the use of modes 1, 1b and 2 with ﬂint raw materials.
This improvement is mainly due to the use of a holding device on
the ground which allows the knapper to be more precise and the
force delivered to be better controlled and in a straight axis
(Pelegrin, 2012:470e473). This increase in regularity was not
observed on the whole experimental production. Products with
small dimensions (micro-blades) tend to be irregular. Products with
larger dimensions (bladelets and blades) generally show a better
regularity. Among the factors that may explain this difference in
regularity observed between micro-blades on one side, and bla-
delets and blades on the other, the morphology of the cores have
played an important role. Two cores have been used for this
experiment using mode 3. One was a well-formed prism with large
dimensions (111 41 34 mm). This prism predominantly pro-
duced bladelets and blades with widths ranging between 7.65 and
15.03 mm (Fig. 5). The other core was a small fragment of smoky
quartz rolled pebble (35.5 32 28.5 mm) that mostly produced
irregular micro-blades and bladelets with widths ranging from 3.17
to 11.69 mm with a mean of 7.17 mm (Fig. 4: bottom). It is clear that
the initial morphology of the blocks selected for knapping has
strongly affected the regularity of the products manufactured. The
products' proﬁles are predominantly regularly curved (28 pieces),
although pieces with a straight proﬁle are relatively well repre-
sented (17 pieces) (Table 2).
It appears from the analysis of the products that this technique
is likely to produce micro-blades as well as bladelets and blades,
depending on the core dimensions. It was also observed that core
morphologies affected the products regularity. Nevertheless this
technique has to be included as a potential candidate for a micro-
blade production on hyaline quartz.
4.3. Comparisons of techniques and conclusions
Several points can be made regarding the result of our experi-
ments. First, we have produced a corpus of data in a controlled
experimental setting that allows us to identify some potentially
relevant quantiﬁable variables to differentiate between the
different pressure ﬂaking techniques deﬁned by Pelegrin as Modes
1,1b, 2, and 3. We can use these data to make suggestions regarding
the nature of the techniques used in Dikili Tash and Promachonas-
ca to produce rock crystal micro-blades. The fragmentary
aspect of the products studied led us to focus on the width/thick-
ness ratio of the products to identify some potential groupings.
However, we will eventually need to account for other qualitative
variables to properly interpret some of the statistical results pro-
On the basis of our sample, the comparison of width/thickness
ratio trimmed mean for all techniques highlights three main
groups. Indeed, t-test analysis supports a highly signiﬁcant differ-
ence between mode 3 and mode 1,1b, and 2 (p<0.001). Modes 1
and 1b can be shown to be statistically similar (p¼0.87). However,
Mode 1 is different than Mode 2 (p¼0.045), while Mode 1b does
not differ signiﬁcantly from Mode 2 (p¼0.06) (Fig. 16).
On the other hand, our experiments led us to suggest the type of
pressure debitage used to the production of rock crystal micro-
blades in both sites. Thus, mainly four ways to produce micro-
blades using pressure techniques were tested in order to deter-
mine more precisely the debitage modalities of our archaeological
Considering the dimensional range of both width and thickness
on experimental corpus, the mode 1, 1b and mode 2 gave the most
valuable results for comparison with archaeological micro-blades.
This is supported by techno-morphological attributes as well as
W/T ratio where both Dikili Tash and Promachonas-Topolni
shown to have similar ratios (p¼0.151), are also signiﬁcantly
different from Mode 3 (p<0.0001).
The results obtained with mode 3 show similar lengths, but a W/
T ratio signiﬁcantly different from the archaeological assemblages
(p¼0.001), while displaying numerous oversized products. How-
ever, some techno-morphological attributes observed on the
products from this mode are absent on archaeological elements
(lips for example, cf. Table 2). Stigmata that could appear more
relevant to the archaeological assemblages can be mostly observed
in the mode 1b and mode 2 series. Taking into account other
metrics and techno-morphological features, either pressure debit-
age in the hand with a stick and holding device (Mode 1b) or in the
hand with shoulder crutch and holding device (Mode 2) represent
N. Tardy et al. / Quaternary International 424 (2016) 212e231 227
the most likely ways to replicate micro-blades from Dikili Tash and
As underlined before, these results need to be taken as in-
dications. Experiments must be continued especially with mode 2
techniques, also with more strictness in the morphologies of cores
and numerically larger series.
5.1. Technological discussion
The experimental results have shown that width and thickness of
the products were not compatible with three techniques (soft direct
percussion, indirect percussion and pressure ﬂaking in a standing
position with a crutch and holding device). On the other hand,four of
the experimental techniques delivered, for all or part of their pro-
duction, widths and thicknesses within the dimensional range given
by the micro-blades from Promachonas-Topolni
ca and Dikili Tash.
These four techniques are all pressure knapping techniques (Modes
Fig. 16. ANOVA analysis and t-test on W/T ratio for modes 1, 1b, 2, and 3, Dikili Tash
(DT) and Promachonas-Topolni
Fig. 17. Expressions of the anisotropic nature of rock crystal. Top left: isotropic and anisotropic directions; Top right: inferior face of a blade detached using mode 4; Bottom left: core
knapped using modes 3 and 4; Bottom right: inferior faces of blades detached using mode 4.
N. Tardy et al. / Quaternary International 424 (2016) 212e231228
1, 1b, 2 and 3). The latter mode displayed more than half of the
production within the dimensional range given by the archaeolog-
ical products. Meanwhile, experimental products from the three
other pressure modes have dimensions that are all included within
that dimensional range given by archaeological micro-blades.
A phenomenon particular to rock crystal i.e., ampliﬁcation of
ripples, has been identiﬁed on the archaeological products as well
as on a large part of the experimental products, regardless of the
technique used. This ripple ampliﬁcation even led to the aban-
donment of the core as it created irregularities sometimes
morphologically close to a hinge fractures on the ﬂaking surface.
This attribute is unknown on ﬂint and ﬂint-like raw materials. It is
the result of the anisotropic nature of hyaline quartz. Although all
the conditions of its formation have not yet fully been identiﬁed,
this stigma seems to develop preferentially but not systematically,
when the rock crystal is knapped in the direction of the length of
the prism, but also in the direction of its thickness. It seems to
disappear on the products knapped in an oblique direction
regarding the crystal. An area appears to be without constraints:
the pyramidion at the end of the prism, or at both ends in case of bi-
terminated crystals (Fig. 17).
The observations on the analysed products presented here allow
reﬂection on the attributes generally used for the identiﬁcation of
blade pressure debitage. These attributes are:
parallel and rectilinear edges and ridges,
a low and constant thickness of the products in the mesial part,
or at least no any abrupt changes,
inferior surface without marked ripples (Inizan et al.
1995:79e80; Tixier, 2012:142).
The experimental data provided by Pelegrin (1988, 2012) on
ﬂint and obsidian highlights the fact that these identiﬁcation
criteria are difﬁcult to apply to products made with modes 1,1b and
2. The author points out a general “poor regularity”of the micro-
blades produced using mode 1 (Pelegrin, 1988:41). Furthermore,
although a gain of regularity and productivity is seen, due to better
stability of the core when a holding device is used as for modes 1b
Fig. 18. Technological attributes seen on various experimental products using modes 3 and 4.
N. Tardy et al. / Quaternary International 424 (2016) 212e231 229
and 2 (p.42), the majority of the micro-blades does not present the
above mentioned criteria used for the identiﬁcation of a pressure
debitage. It is only from the use of an immobilization system of the
core placed on the ground (such as in mode 3) that a signiﬁcant
increase in the regularity of the products is observed (Pelegrin,
2012:470). The criteria for the recognition of a laminar produc-
tion using pressure technique can then be used effectively on ﬂint
and obsidian materials only from mode 3 and in the case of a well-
controlled pressure. The vast majority of micro-blades produced by
modes 1 to 2 do not show the characteristic features of a pressure
debitage as given by Inizan et al. (1995).
These three criteria which are “parallel edges and ridges”,“low
and constant thickness”of the products and an “inferior face
without marked ripples”when applied to an anisotropic area of the
crystal, can create a sort of “natural denticulation”that interferes
with the expected parallelism of the edges and ridges of a blade. On
the other hand, when these ripples are marked as to form wavelets
on the inferior face of a blade, the thickness of the product is not
constant but can vary suddenly. The inferior faces of the products
may have pronounced ripples (Fig. 17).
Thus, our pressure knapping experiments of Pelegrin's modes 3
and 4 on hyaline quartz have delivered very few elements with all
the identiﬁcation criteria of a pressure debitage. Several factors can
explain this observation: the limited nature of these experiments is
one, the morphology of the blocks used is another. Among these
factors, the anisotropic nature of hyaline is certainly one that has
most inﬂuenced the results.
5.2. Archaeological implications
This study also provides new ways of looking at lithic produc-
tion and consumption on Neolithic sites. Binder and Perl
have argued that the Neolithic lithic system is mostly character-
ized on consumer sites by a dichotomy between “skilled imported”
and “expedient local”productions.
The few obsidian and honey ﬂint blades and bladelets with the
attributes of a pressure debitage and represented in Promachonas-
ca and Dikili Tash only in the form of ﬁnished products il-
lustrates the “skilled imported productions”. The absence of cores,
technical pieces, ﬂakes and debris from these chaînes op
suggest import of these products from the sources to the sites.
These raw materials used for the manufacture of standardized
blades and bladelets are exotic. Obsidian probably originates from
southern Greece, especially the island of Melos (Kourtessi-
Philippakis, 2009:307) although a Carpathian origin has been
evoked in the case of Mandalo (Kilikoglou et al. 1996). The honey
ﬂint shows similarities with the “Balkan ﬂint”from which several
sources have been identiﬁed around the Romanian and Bulgarian
Moesian platform (Gurova and Nachev, 2008; Bonsall et al., 2010;
Gurova, 2012). As such, the presence of these imported pressure
ﬂaked products in Dikili Tash and Promachonas-Topolni
sharply with that of the micro-blade pressure production on hya-
The presence (particularly in Promachonas-Topolni
exhausted cores, ﬂakes and fragments, debris in addition to the
micro-blades could illustrate a pressure technique made in situ,at
least partially. These archaeological micro-blades show an overall
weak standardization as edges and ridges are mostly irregular, but a
few pieces show an overall good facture (Fig. 4). On the other hand,
the crystalline rock formation of the Rhodope Mountains, located
between southern Bulgaria and northern Greece is a very suitable
area for the formation of rock crystals. Numerous quartz veins have
been located around Loutra (Greece), a few kilometres from Pro-
ca, and in southern Bulgaria, around the
Neolithic site of Kova
cevo. The discovery of a rock crystal pebble in
the Bistrica river, 20 km from Promachonas-Topolni
that this raw material is present on a regional scale.
The fact that the rock crystal assemblage from Promachonas-
ca is made using pressure debitage on local or regional raw
materials adds a new dimension to the model presented by Binder
es (1990) regarding Neolithic stone tool economies. Does
this rock crystal micro-blade production using pressure techniques
could be considered as a “skilled local production”instead of a
“skilled imported”or “expedient local”production? For now, these
suggestions require further study of lithic assemblages from other
sites in the region in order to be conﬁrmed or refuted.
Ammerman, A.J., Cavalli-Sforza, L.L., 1984. The Neolithic Transition and the Genetics
of Populations in Europe. Princeton University Press, Princeton.
Atanasova-Vladimirova, S., Sekiranov, A., 2007. 'New hope' pegmatite vein from
Arda region, the central Rhodopes, Bulgaria. In: Maleev, M.N. (Ed.), Mineral
diversity research and preservation, IVth International Symposium, 7e10
October 2007. Earth and Man National Museum, Soﬁa, pp. 201e206.
Bar-Yosef, O., Meadow, R.H., 2005. The origins of agriculture in the Near East. In:
Price, T.D., Gebauer, A.B. (Eds.), New Perspectives on the Prehistoric Transition
to Agriculture. School of American Research Press, Santa Fe, pp. 39e94.
Bellwood, P., 2005. First Farmers: the Origins of Agricultural Societies. John Wiley
and Sons, London.
Binder, D., Perl
es, C., 1990. Strat
egie de gestion des outillages lithiques au
eo 2, 257e283.
Bonsall, C., Gurova, M., Hayward, C., Nachev, C., Pearce, N., 2010. Characterization of
‘Balkan Flint’Artefacts from Bulgaria and the Iron Gates Using LA-ICP-MS and
EPMA. Interdisciplinary Studies XXII-XXIII. Archaeological Institute and
Museum Bulgarian Academy of Sciences, Soﬁa, pp. 9e18.
Boudeulle, M., Chermette, A., David, L., Fumey, P., Latreille, G., Michel, P., Mourer, R.,
Nicier, P., Pomarais, P., 1979. Quartz_silice: min
eralogie, gemmologie, industrie.
egionale de Pal
ehistoire et des amis du Mus
Demoule, J.-P., 1993. Anatolie et Balkans: la logique
evolutive du N
Anatolica 19, 1e17.
Demoule, J.-P., 1994. Probl
emes crhon-culturels du N
eolithique de Gr
ece du Nord.
In: Decourt, J.-C., Helly, B., Gallis, K. (Eds.), La Thessalie: quinze ann
recherches archologiques, 1975e1990: bilan et perspectives. Actes du Colloque
International, Lyon 17e22 avril 1990. Minist
ere de la Culture, Ath
Demoule, J.-P., Perl
es, C., 1993. The Greek Neolithic: a new review. Journal of World
Prehistory 7 (4), 355e416.
Desrosiers, P.M., 2007. Palaeoeskimo lithic technology: constraints and adaptation.
Lithic Technology 32 (1), 17e38.
Desrosiers, P.M., Sørensen, M., 2012. Eastern arctic under pressure: from Palae-
oeskimo to Inuit culture (Canada and Greenland). In: Desrosiers, P.M. (Ed.), The
Emergence of Pressure Blade Making, from Origin to Modern Experimentations.
Springer, New York, pp. 375e400.
Dimitriadis, S., Skourtopoulou, K., 2003. Petrographic examination of chipped stone
material. In: Elster, E.S., Renfrew, C. (Eds.), Prehistoric Sitagroi: excavations in
northeast Greece, 1968e1970, The Final Report. Monumenta Archaeologica 20,
vol. II. Cotsen Institute of Archaeology, Los Angeles, pp. 127e132.
Flenniken, J.F., 1981. Replicative System Analysis: a Model Applied to Vein Quartz
Artefacts from the Hoko River. Washington State University Laboratory of An-
thropology Reports of Investigations 59, Pullman, Washington.
Foucault, A., Raoul, J.-F., 2005 (1980). Dictionnaire de g
Gurova, M., 2012. ‘Balkan Flint’eﬁction and/or trajectory to Neolithisation: evi-
dence from Bulgaria. In: Bonsall, C., Gurova, M., Voytek, B., Voric, D. (Eds.),
Balkan Flint in SE European Prehistory: Chronological, Cultural and Social
Perspectives, 15th Annual Meeting of the European Association of Archaeolo-
gists, 15e20 September 2009, Riva Del Garda, Trento, Bulgarian E-Journal of
Archaeology 1, Soﬁa, pp. 15e49.
Gurova, M., Nachev, C., 2008. Formal Early Neolithic ﬂint toolkits: archaeological
and sedimentological aspects. In: Kostov, R., Gaydarska, B., Gurova, M. (Eds.),
Geoarchaeology and Archaeomineralogy, Proceedings of the International
Conference, 29e30 October 2008, Soﬁa, Publishing House St. Ivan Rilski, Soﬁa,
Honegger, M., 2011. L'industrie en silex et en quartz taill
es de l'habitat du Petit
a Sion (Valais). In: Besse, M., Piguet, M. (Eds.), Le site pr
Petit Chasseur (Sion, Valais) 10. Un hameau du N
eolithique Moyen, Cahiers
eologie Romande 124 (Archeologia Vallesiana 6), Lausanne, pp. 165e181.
Inizan, M.-L., Reduron-Ballinger, M., Roche, H., Tixier, J., 1995. Technologie de la
ee. In: Pr
ehistoire de la pierre taill
ee, vol. 4. C.R.E.P., Meudon.
Kilikoglou, V., Bassiakos, Y., Grimanis, A.P., Souvatsis, K., 1996. Carpathian obsidian
in Macedonia, Greece. Journal of Archaeological Science 23, 343e349.
Koukouli-Chryssanthaki, H., Todorova, H., Aslanis, I., Vajsov, I., Valla, M., 2007. Pro-
ca: a Greek-Bulgarian archaeological project. In: Todorova, H.,
N. Tardy et al. / Quaternary International 424 (2016) 212e231230
Stefanovich, M., Ivanov, G. (Eds.), The Struma/Strymon River Valley in Prehistory
ein the Steps of James Harvey Gaul, Proceedings of the International Symposium,
Strymon Praehistoricus, 27.09e01.10.2004, Kjustendil-Blagoevgrad (Bulgaria)
and Serres-Amphipolis (Greece). Soﬁa, vol. 2, pp. 43e78.
Kourtessi-Philipakkis, G., 1997. Les industries lithiques taill
ees de Dimitra. In:
Grammenos, D. (Ed.), Nεο
(Neolithic Macedonia). The Archae-
ological Receipts Fund, Athens, pp. 212e216.
Kourtessi-Philippakis, G., 2009. Lithics in the Neolithic of Northern Greece: terri-
torial perspectives from an off obsidian area. Documenta Praehistorica XXXVI,
Kourtessi-Philippakis, G., 2016. Les industries lithiques taill
ees du N
ecent de Dikili Tash. Soci
eologique (Dikili Tash, village pr
edoine orientale, Gr
enes. Recherches dirig
ees par H. Koukouli-
Chryssanthaki et R. Treuil [1986e2001]) (forthcoming).
Kostaki, A., 2016. Mati
ere, production et organisation de l'espace durant le
ecent en Gr
ece du Nord: Le site de Promachonas-Topolni
Lespez, L., Tsirtsoni, Z., Darcque, P., Koukouli-Khryssanthaki, H., Malamidou, D.,
Treuil, R., Davidson, R., Kourtessi-Philippakis, G., Oberlin, C., 2013. The lowest
levels at Dikili Tash, northern Greece: a missing link in the Early Neolithic of
Europe. Antiquity 87, 30e45.
Mourre, V., 1996. Les industries en quartz au Pal
ethodologie et technologie. Pal
eo 8, 205e223.
Novikov, V.P., Radililovsky, V.V.,1987. Quartz anisotropy in stone age artefacts of the
Hissar. In: Seronie-Vivien, M.-R., Lenoir, M. (Eds.), Le silex de sa gen
Actes du 5e colloque international sur le silex, Cahiers du Quaternaire 17/2.
CNRS, Paris, pp. 593e598.
Pappa, M., 2007. Neolithic Societies: recent evidence from Northern Greece. In:
Todorova, H., Stefanovitch, M., Ivanov, G. (Eds.), The Struma/Strymon river valley
in Prehistory eIn the steps of James Harvey Gaul, Proceedings of the Interna-
tional Symposium 27/09e01/10 2004, Strymon Praehistoricus, Kjustendil-Bla-
goevgrad (Bulgaria) and Serres-Amphipolis (Greece), Soﬁa, vol. 2, pp. 257e272.
Pelegrin, J., 1988. D
erimental par pression ‘du plus petit au plus grand’.
In: Tixier, J. (Ed.), Technologie Pr
ehistorique, Notes et monographies Tech-
niques, vol. 25. CNRS, Paris, pp. 37e53.
Pelegrin, J., 2012. New experimental observations for the characterization of pres-
sure blade production techniques. In: Desrosiers, P.M. (Ed.), The Emergence of
Pressure Blade Making efrom Origin to Modern Experimentation. Springer,
New York, pp. 465e500.
es, C., 1990. L'outillage de pierre taill
eolithique en Gr
ment et exploitation des mati
eres. Bulletin de Correspondance
enique 114 (1), 1e42.
es, C., 1992. Systems of exchange and organization of production in Neolithic
Greece. Journal of Mediterranean Archaeology 5 (2), 115e164.
Rodden, J.R., 1962. Excavation at the Early Neolithic site of Nea Nikomedeia, Greek
Macedonia. Proceedings of the Prehistoric Society 28, 267e288.
Roque, C., Guibert, P., Vartanian, E., Bechtel, F., Treuil, R., Darque, P., Koukouli-
Chryssanthaki, H., Malamidou, D., 2002. The chronology of the Neolithic
sequence at Dikili Tash, Macedonia, Greece: TL dating of domestic ovens.
Archaeometry 44 (4), 613e633.
es, M., 1992. Chapitre IV: L'outillage. In: Treuil, R. (Ed.), Dikili Tash, village
ehistorique de Mac
edoine Orientale, I. Fouilles de Jean Deshayes
(1965e1975), Bulletin de Correspondance Hell
enique Suppl. 24, vol. 1,
Tardy, N., 2016. Les industries en quartz de Kova
cevo (Bulgarie), Madzhari (ARYM),
Dikili Tash et Promachonas (Gr
ece) : Reconstitution des syst
dans le contexte de la N
eolithisation de l'Europe du Sud-Est. Th
ese de Doctorat.
ere Lyon 2 (forthcoming).
Tixier, J., 2012. M
ethode pour l'
etude des outillages lithiques. Centre National de la
eologique du Luxembourg, Mus
ee National d'Histoire et d'Arts.
eoLogiques 4, Luxembourg.
Treuil, R., 1992. Dikili Tash village pr
ehistorique de Mac
edoine orientale. I. Fouilles
de Jean Deshayes (1961e1975). In: Bulletin de Correspondance Hellenique,
suppl. 24, vol. 1. Ecole Française d'Ath
Tringham, R., 2003. Flaked stone. In: Elster, E.S., Renfrew, C. (Eds.), Prehistoric
Sitagroi: excavations in northeast Greece, 1968e1970, The ﬁnal report. Mon-
umenta Archaeologica 20, vol. II. Cotsen Institute of Archaeology, Los Angeles,
Vajsov, I., 2007. Promachon-Topolni
ca. A typology of painted decorations and its use
as chronological marker. In: Todorova, H., Stefanovich, M., Ivanov, G. (Eds.), The
Struma/Strymon River Valley in Prehistory ein the Steps of James Harvey Gaul,
Proceedings of the International Symposium, Strymon Praehistoricus,
27.09e01.10.2004, Kjustendil-Blagoevgrad (Bulgaria) and Serres-Amphipolis
(Greece). Soﬁa, vol. 2, pp. 79e120.
Watson, A., 1983. Chipped stone artefacts from Servia in Greek Macedonia. Journal
of Field Archaeology 10 (1), 120e124.
Zvelebil, M., 1986. Hunters in Transition. Cambridge University Press, Cambridge.
N. Tardy et al. / Quaternary International 424 (2016) 212e231 231