ArticlePDF Available

Abstract and Figures

- Previous archaeological research in the Banat area (South-western Romania) resulted in the definition of a chronologically late Krems-Dufour type Aurignacian, followed by the isolated find of several considerably old anatomically modern human (AMH) remains at Oase Cave, several decades later. The last find set the stage for new stratigraphic, chronological and archaeological reassessment of Banat Aurignacian settlements at Tincova, Coşava and Româneşti- Dumbrǎviţa. This study presents the attribute analysis of the Aurignacian lithic assemblage at Româneş ti-Dumbrǎviţa I, involving both old and recently excavated collections. Alongside the more accurate identification of the main technological and typological features, pointing to a Protoaurignacian/Early Aurignacian assignation of the Early Upper Palaeolithic industry here, new chronological landmarks, much older than previously considered, became available. Preliminary thermoluminescence results point to an estimated age between 45 and 40 ka for the main accumulation in GH3 at Româneşti, thus indicating a possible contemporaneity of the Banat Aurignacian and the Oase AMH finds. A brief comparative outline of the Banat Aurignacian settlements is also provided, followed by and attempt at placing the local Aurignacian into the European Early Upper Palaeolithic landscape.
Content may be subject to copyright.
Quartär 59 (2012) : 85-130
85
doi: 10.7485/QU59_4
The earliest Aurignacian in Romania:
New investigations at the open air site
of Româneşti-Dumbrăviţa I (Banat)
Das früheste Aurignacien in Rumänien – Neue Untersuchungen an der
Freilandfundstelle Româneti-Dumbrăvia I (Banat)
Valéry S
1
*, Victor C
2
, Mircea A
3
, Thorsten U
4
, Holger K
5
,
Alexandra H
6
, Christoph S
6
, Loredana N
3
, Ion B
7
, Andrei V
2
&
Thomas H
1
1
Institute of Prehistoric Archaeology, University of Cologne, Weyertal 125, D-50923 Cologne
2
Crimean Branch of the Institute of Archeology, National Ukrainian Academy of Sciences (NUAS) Yaltinskaya Street, 2, 95007
Simferopol, Ukraine
3
L Department of History and Letters, Faculty of Humanities, Valahia University of Târgovte, Str. Lt. Stancu Ion, 34-36,
130105 Târgovte, Romania
4
Institute of Prehistoric Archaeology, University of Erlangen, Kochstraße 4/18, D-91054 Erlangen
5
Department of Geography, RWTH Aachen University, Wüllnerstraße 5b, D-52056 Aachen
6
Institute of Geography, University of Cologne, Otto-Fischer-Str. 4, D-50674 Cologne
7
Heritage Advice S.R.L., Str. Nicolae Titulescu, 1, 280/B/63, 510096 Alba Iulia, Romania
Abstract - Previous archaeological research in the Banat area (South-western Romania) resulted in the definition of a
chronologically late Krems-Dufour type Aurignacian, followed by the isolated find of several considerably old anatomically
modern human (AMH) remains at Oase Cave, several decades later. The last find set the stage for new stratigraphic,
chronological and archaeological reassessment of Banat Aurignacian settlements at Tincova, Coşava and Româneşti-
Dumbrăviţa. This study presents the attribute analysis of the Aurignacian lithic assemblage at Româneşti-Dumbrăviţa I,
involving both old and recently excavated collections. Alongside the more accurate identification of the main technological
and typological features, pointing to a Protoaurignacian/Early Aurignacian assignation of the Early Upper Palaeolithic industry
here, new chronological landmarks, much older than previously considered, became available. Preliminary thermoluminescence
results point to an estimated age between 45 and 40 ka for the main accumulation in GH3 at Româneşti, thus indicating a
possible contemporaneity of the Banat Aurignacian and the Oase AMH finds. A brief comparative outline of the Banat
Aurignacian settlements is also provided, followed by and attempt at placing the local Aurignacian into the European Early
Upper Palaeolithic landscape.
Z - Bisherige Forschung zum Beginn des Jungpaläolithikums im Banat ergaben widersprüchliche Ergebnisse.
Nachdem die von dort bekannt gewordenen Aurignacien-Freilandfundstellen Tincova, Coava und Româneti-Dumbrăvia
zunächst chronologisch an das Ende dieses Technokomplexes gestellt worden waren, wurde nach der Entdeckung der Überreste
früher anatomisch moderner Menschen in der Oase-Höhle angenommen, es handele sich um ein Proto-Aurignacien. In dem
vorliegenden Artikel werden diese Widersprüche anhand neuer Grabungen und erster absoluter Datierungen sowie einer
detaillierten Analyse der Alt- und Neufunde an der Fundstation Româneti-Dumbrăvia aufgelöst. Demnach handelt es sich an
diesem Fundplatz um eine Steingeräteindustrie, die sowohl Merkmale des Proto-Aurignacien als auch des klassischen Aurignacien
aufweist. Erste Thermolumineszenz-Alter zwischen 45 kyr BP im Liegenden und 40 kyr BP im Hangenden deuten auf eine frühe
Zeitstellung von Româneti-Dumbrăvia innerhalb des älteren Jungpaläolithikums und eine zeitliche Überschneidung mit den
Menschenresten aus der Oase-Höhle. Vor diesem Hintergrund wird die Bedeutung des Banat im Rahmen der Ausbreitung des
frühen modernen Menschen nach Europa diskutiert.
Keywords -
Banat region, Protoaurignacian, Stratigraphy, Absolute Dating, Technology, Lithic analysis
Banat Region, Protoaurignacien, Stratigraphie, Absolute Datierung, Technologie, Artefaktanalyse
*corresponding author:
vsitlivy@uni-koeln.de
Quartär 59 (2012) V. Sitlivy et al.
86
Introduction
The rich Romanian Palaeolithic record has been
reflected in a number of comprehensive publications
in recent decades (e.g. Chirica et al. 1996; Cârciumaru
1989, 1999; Păunescu 1989, 2000, 2001). Several other
recent contributions, mostly devoted to a critical
reassessment of published data, particularly focused
on the Middle-to-Upper Palaeolithic transition and
the onset of the Upper Palaeolithic (e.g. Riel-Salvatore
et al. 2008; Noiret 2009; Barton et al. 2011; Anghelinu
et al. 2012; Anghelinu & Niţă in press). Notwithstanding,
the Southwestern Romania record (i.e. the historical
region of Banat - Mogoşanu 1978) was for a long time
overlooked by detailed studies (but see Hahn 1970,
1977) and rarely incorporated into the broader frame-
work of the European Upper Palaeolithic (e.g.
Kozlowski & Kozlowski 1975; Djindjian et al. 1999),
remaining, at best, a subject of regional overviews
(Păunescu 2001; Băltean 2011a, b).
The discovery of anatomically modern human
remains (AMH) at Oase Cave (Trinkaus et al. 2003),
as well as recent advances in Aurignacian studies
(e.g. Bon 2002; Bordes 2002; Chiotti 1999; Lucas
2000; Le Brun-Ricalens et al. 2005, 2009; Nigst 2006;
Nigst et al. 2008; Nigst & Haesaerts in press) promp-
ted, however, a renewed interest in the Early Upper
Palaeolithic in this region (Teyssandier 2003, 2007a, b,
2008; Teyssandier et al. 2010; Tsanova et al. in press;
Zilhão 2006, 2007). The lack of material culture
Fig. 1. Palaeolithic sites in Banat, south-western Romania and selected loess sections of the lowland (Projection: UTM 34 WGS 1984,
Cartography: R. Löhrer).
Abb. 1. Paläolithische Fundstellen im Banat und ausgewählte Löss-Bereiche in den Niederungen (Projektion: UTM 34 WGS 1984, Kartierung:
R. Löhrer).
Quartär 59 (2012)Earliest Aurignacian in Romania: the open air site of Româneşti-Dumbrăva
87
associated with the early-dated fossil remains at Oase
focused attention initially towards the single-layered
Aurignacian settlement at Tincova, located in
the same region as Oase Cave. Reassigned to the
Protoaurignacian, Tincova was rapidly included in the
intense debate regarding the initial dispersal of AMH
into Europe (Teyssandier 2003; Zilhão 2007; Tsanova
et al. in press). The Aurignacian assemblages at Coşava
and Româneşti-Dumbrăviţa in the northern part of
this region were rarely invoked in recent studies,
despite the fact that the industries of these sites had
always been compared and presented together
with Tincova as “classical” or Krems-Dufour type
Aurignacian/culture (Mogoşanu 1972, 1978; Kozlowski
& Kozlowski 1975; Hahn 1970, 1977; Demidenko 1999,
2000 - 2001; Demidenko & Otte 2000 - 2001;
Demidenko & Noiret 2012).
In the light of a much needed re-evaluation of the
Aurignacian in Romania (Anghelinu & Niţă in press),
the importance of new, detailed reassessment of these
sites appeared evident (Sitlivy et al. in press), all
the more so since, challenging the lithic analogies
connecting the Banat assemblages with the Krems-
Dufour Aurignacian (see also Chirica et al. 1996), the
original pollen-based geochronological estimations –
Herculane/Tursac, for the single layer at Tincova and
Herculane II/Laugerie, for the main concentration,
layer III at Româneşti (Mogoșanu 1978; Cârciumaru
1989, 1999) – have indicated an unexpectedly young
chronology. Clarifying this chronological status,
considerably younger than any known Eurasian
Aurignacian occurrence, has become crucial as well.
As a consequence, both Româneşti-Dumbrăviţa
and Coşava were selected in 2009 for detailed
stratigraphic, chronological and archaeological
reevaluation, including excavation of new test pits,
and TL, OSL, pollen, sedimentological and tephra
sampling, correlated with the study of old and new
archaeological collections. This article presents the
first results of such recent research, focusing on both
Mogoşanu’s and newly excavated (2009 - 2010) lithic
assemblages from Româneşti-Dumbrăviţa I, along with
comparative reappraisal of regional technological and
typological data.
Previous work
Româneşti (Timiş district) is located on the periphery
of Poiana Ruscă Mountains, in the eastern part of the
historical region of Banat. The Palaeolithic site of
Româneşti (local toponym: Dumbrăviţa) is situated at
the confluence of the Bega Mare and Bega Mica rivers,
4 km N from the site of Coşava, from which it is
separated by the large Bega valley (Figs. 1 2). The
settlement is situated in a rather short loess-like
sequence (c. 3 m) on the 10 m river terrace (45°49’02.41”
N, 22°19’15.12” E) and, based on lithic surface scatters,
extends about 4 hectares.
Fig. 2. Românti-Dumbrăva I and II sites during field campaign in 2009-2010: 1 – view from the east;
2 – view from the north-west.
Abb. 2. Die Fundstellen Româneti-Dumbrăvia I und II während der Ausgrabungen 2009-2010:
1 – Sicht von Westen; 2 – Sicht von Nordwesten.
1
2
Quartär 59 (2012) V. Sitlivy et al.
88
Mogoşanu’s excavations
The history of investigations in Banat has been recently
described in detail (e.g. Băltean 2011a, b), and will not
be extensively discussed here. We focus mainly on the
original stratigraphic subdivision and artifact
descriptions (Mogoşanu 1972, 1978, 1983).
F. Mogoşanu excavated Româneşti-Dumbrăviţa I
site in 1960 - 1964 and 1967 - 1972, over a large area
of about 450 m² (Fig. 3). The stratigraphy was
represented in one combined synthetic profile for
adjacent zones Româneşti I and II (Mogoşanu 1978)
(Fig. 4: 2, 3):
1) modern soil: 0-15 cm;
2) fine loess-like sediment with iron oxide: 15-35 cm;
3) intermediate yellow-reddish layer with brown
“stains”: 35-50 cm;
4) brown-reddish clay with prismatic structure,
rich in iron oxide concretions, especially in the lower
part: 50-110 cm;
5) fine, reddish clay with vertical greyish-bluish
veins: 110-180 cm;
6) clay mixed with rolled pebbles and iron oxide:
180-200 cm;
7) clay mixed with fine pebbles: 200-250 cm;
8) compact horizon of iron oxide and rolled
pebbles: 250-280 cm;
9) fine, reddish clay, mixed with pebbles:
280-320 cm.
From bottom to the top, six archaeological layers
were identified in the upper part of the sequence at
Româneşti I (Mogoşanu 1972, 1978, 1983):
- Layer I was present at a depth of about 115-105 cm
from the modern surface at the upper limit of reddish
clay and contained 48 quartzite/quartz artifacts;
- Layer II lied at the base of brown-reddish clay at
a depth of 95-90 cm, documented as a thin layer (5 cm
thick) across a small excavated area of 8 m² and
contained rare artifacts, including, endscrapers,
burins, blades with fine retouch, sidescraper and
flakes;
- Layer III was found at a depth of about 86-70 cm
in the same sediment, yielding a rich industry
Aurignacian industry of more than 5000 artifacts,
including 114 tools (end-scrapers, including carinated
ones, predominate over burins, associated with eight
Dufour bladelets and some retouched blades,
comprising some typical Aurignacian forms);
- Layer IV was located in the upper part of brown-
reddish clay, at a depth of 67-60 cm (20 m²). 61 tools,
about 30 % of which are truncated pieces on blades
and flakes, were recorded, with fewer endscrapers
and Aurignacian blades and more burins, including
burins on truncation;
- Layer V was found in a transitional zone between
the same brown-reddish clay and the uppermost loess
(at a depth of 50-40 cm), representing extended but
clustered work-shops with an industry rich in
knapping waste and only 38 tools, dominated by
burins and with less common Aurignacian pieces.
- Layer VI is located in the uppermost loess (30-20 cm
in depth from the modern soil) and attributed to the
Epipalaeolithic (Mogoşanu 1978) or the Gravettian
(Mogoşanu 1983; Chirica et al. 1996).
Thus, according to Mogoşanu, the Aurignacian
(layers II, III, IV and V) was stratified between the
“Quartzitic Mousterian” (layer I) and the Gravettian
(layer VI).
As far as the horizontal distribution of artifacts is
concerned, Mogoşanu published only very broad
information in form of a schematic plan (Fig. 3). What
can be said, apart from the scattered distribution of
layer VI (“Gravettian/Epipalaeolithic”), with some
Fig. 3. Româneşti-Dumbrăva I site map: Mogoşanu‘s and 2009-10 excavations.
Abb. 3. Româneti-Dumbvia I. Plan der Ausgrabungsflächen von Mogoanu und der neueren
Untersuchungen 2009-2010.
Quartär 59 (2012)Earliest Aurignacian in Romania: the open air site of Româneşti-Dumbrăva
89
certainty is that only layer III was observed in almost
all trenches. The widely spread occurrence of this
layer corresponds to the absolute number of lithics.
Layers I (“Mousterian”) and IV (“Aurignacian”) were
found over comparably large, only partly intersecting
areas in the southwestern part of his excavation,
whereas layers V (“Aurignacian”) was located in small,
and at the same time distant, clusters. Layer II
(“Aurignacian”) was only recognized once almost in
the centre of Mogoşanu´s main trench. With
exception of layer V and II, the mapping is better to be
understood as information about the presence and
absence of layers rather than actual artifact concen-
trations in conventional sense. This is even more so as
no structures, e.g. fire places, knapping areas, etc., are
reported from Româneşti-Dumbrăviţa I. While both
sites lack bone preservation, it should be noted that
Româneşti-Dumbrăviţa II yielded small workshop
clusters which appear to have been stratigraphically
parallel to layer V of the main site. One workshop was
specialized in Dufour production, associated with
alternately retouched bladelets, retouched flakes, two
endscrapers and several Krems points (Mogoşanu
1978).
Fig. 4. Româneşti-Dumbrăviţa I, stratigraphic sections: 1 – 2009 field campaign; 2, 3 – Mogoşanu’s
excavations (after Mogoşanu 1978).
Abb. 4. Româneti-Dumbrăvia I, Stratigraphie. 1 – Ausgrabung 2009 ; 2, 3 – Ausgrabungen durch
Mogoanu (nach Mogoanu 1978).
Quartär 59 (2012) V. Sitlivy et al.
90
History of interpretations
Despite many references to Româneşti I, interpretation
of this site was quite limited and, apart from the
hypothesized workshop function, focused on two
related issues: chronology and cultural attribution.
The initial excavator, F. Mogoşanu, had rapidly
noted the similarities between Tincova, Coşava layer I,
Româneşti-Dumbrăviţa I layer III, and the UP
collection at Krems-Hundssteig (Austria), a correlation
which was subsequently confirmed by others (e.g.
Kozlowski 1965; Hahn 1970, 1977; Demidenko 1999;
Demidenko & Otte 2000 - 2001; Demidenko & Noiret
2012). Unfortunately, much like the eponymous site at
Krems, the Banat Aurignacian assemblages, altogether
lacking organic material, remained for decades
undated. As a consequence, most interpretations relied
on formal aspects of the published lithic collections.
Due to the inferred late geochronology, Mogoșanu
saw the Banat Aurignacian as an echo of its Central
European counterpart, postdating the emergence of
the Gravettian technocomplex. To him, the Banat
acted as a geographic refugium for late Aurignacians
retreating from an “expanding” Gravettian (Mogoșanu
1978, 1983). Further references in the Romanian archaeo-
logical literature (Cârciumaru 1999; Păunescu 2001)
only reiterated the Krems-Dufour analogy and
generally acknowledged the late chronology of the
phenomenon (but see Chirica et al. 1996).
More recent debates on the definition of the
Aurignacian have usually exclusively included Tincova,
attributed to the Protoaurignacian, Early Ahmarian or
Kozarnikian (e.g. Teyssandier 2003; Tsanova et al. in
press). Other reassessments of the Banat sites,
including Tincova, Româneşti (layers IIIII) and Coșava
(layers III), however, point to the similarity between
their lithic technology and the Protoaurignacian,
tentatively pushing the chronology of Româneşti layer
I and Tincova to the Hengelo-Arcy interstadial
(Băltean 2011a, b). Hopefully, the arguments in the
present contribution will help resolve such lingering
taxonomical and chronological issues.
Outline of the present study
The present study focuses on the following issues:
1. re-evaluation of lithic assemblages excavated by
Mogoșanu with regard to changed methodology
(in classification, attribute analysis) and reconstruction
of raw material reduction (technology);
2. re-assessment of the stratigraphical information
by small scale test pits with special focus on vertical
and horizontal artifact distribution using up to
date excavation techniques (e.g. three-dimensional
measurements);
3. elucidation of the frequency of small-scale
artifacts by the application of wet sieving;
4. evaluation of the absolute age of the Aurignacian
layers by using OSL- (on sediment samples) and TL-
(on burned flint samples) dating methods;
5. clarification of the status and context of the
assemblages in frames of the Early Upper Palaeolithic
and the first dispersal of modern humans into Europe.
The following sections reflect the outline of the
study and first present the main characteristics of the
Aurignacian assemblages from Mogoşanu´s
excavations layers II, III, IV and V. It will be followed by
the presentation of our own field work, comparisons
between the two data sets and, finally, a discussion
in how far the results allow for a classification as
Aurignacian or Protoaurignacian.
Results of the analysis of Mogoşanu’s lithic
assemblages
Methods and samples
The lithic analysis, designed to provide information
on technology, typology and raw material exploitation,
was based on attribute analysis of cores, laminar
debitage and tools in the available old and newly
excavated assemblages (for attributes see Tixier 1963;
Hours 1974; Marks 1976; Demars & Laurent 1989;
Inizan et al. 1995; Pelegrin 2000; Chabai & Demidenko
1998; Soriano et al. 2007; Le Brun-Ricalens et al. 2009;
Sitlivy et al. 2009; Demidenko 2012; Sitlivy et al. in
press). The subdivision of carinated pieces into cores
and endscrapers was made sensu Yu. Demidenko: “In
addition to its classical characteristics (Sonneville-
Bordes & Perrot 1954: 332; Movius & Brooks 1971:
255), a carinated endscraper should always have in its
typical form a front-edge scraper width greater than
the length of lamellar (bladelets sensu lato) retouch
facets which created this front-edge.” (Demidenko
2012: 97).
The analysis of old collections revealed biases in
some artifact categories published in the past. For
example, Mogoşanu’s (1978) layer III contains more
than 5000 artifacts, including 114 tools. The study
sample of 2654 lithic artifacts differs from previously
published data by (a) a larger tool-kit (161 contra 114),
(b) lesser representation of endscrapers (18 contra 51,
including 24 various carinated/core-like items) and (c)
burins (17 contra 26), (d) more abundant Dufour
bladelets (11 contra 8), (e) sidescrapers (7 contra 2) as
well as retouched pieces on blades/flakes and broken
tools. At present, 29 cores and 6 pre-forms have been
recorded, above all carinated and narrow-faced types
(often previously attributed to endscrapers and
burins). On the other hand, the laminar debitage
appears to be more representative: 719 blades,
168 bladelets (W is 7 - 12 mm; sensu Tixier 1974) and
7 micro-blades (W<7 mm sensu Amirkhanov 1986)
contra 788 non-separated laminar products. Inconsis-
tencies in frequencies and numbers of some artifact
types between Mogoşanu´s published assemblages
(Mogoşanu 1978: 80; Păunescu 2001: 188) and recently
restudied material are due to the loss of some
artifacts, illegible labels, incompleteness of former
Quartär 59 (2012)Earliest Aurignacian in Romania: the open air site of Româneşti-Dumbrăva
91
studies, differences in classification practices, or
omitting of a certain number of broken tools and
debitage products. Despite these biases, all main tool
categories are present and generally reflect similar
priorities in tool structure. Moreover, debitage,
except “micro” artifacts, is abundant and informative
for an application of a technological approach.
Raw material
Opal (“Banat flint”) is the most common raw material
used at Româneşti-Dumbrăviţa and neighboring sites.
This rock is variable in quality (generally mediocre),
homogeneity (often with inclusions) and colors
(brownish, reddish and their combination). Large and
medium-sized oblong well-rounded cobbles with
alluvial cortex as well as nodules with fresh white
cortex and non-cortical chunks of opal were used.
Most of these (often of poor quality) are present in
the Bega river gravels, on old uppermost eroded
terraces and slopes in the vicinity of the excavated
area. Nevertheless, the exact sources of good quality
opal and other “exotic” rocks (e.g. “black” flint) remain
unknown, but local and meso-local origins are likely.
Flint, radiolarite, quartzite are present in much smaller
quantities (<10 %). Chalcedony, jasper, quartz and
obsidian occur sporadically or as isolated pieces. In
addition, there are no significant changes within the
Româneşti-Dumbrăviţa I Aurignacian-Gravettian
Layer II Layer III Layer IV
n % % esse n % % esse n % % esse
Pre-forms 1 0.31 0.32 6 0.23 0.24
Cores 3 0.93 0.95 29 1.09 1.14 17 1.58 1.65
Flakes 165 51.08 52.22 1448 54.56 56.87 663 61.45 64.37
Blades 109 33.75 34.49 719 27.09 28.24 234 21.69 22.72
Bladelets 19 5.88 6.01 168 6.33 6.60 43 3.99 4.17
Micro-blades 2 0.62 0.63 7 0.26 0.27 3 0.28 0.29
Blank fragments
Tools 16 4.95 5.06 161 6.07 6.32 67 6.21 6.50
Tools/cores 1 0.04 0.04
Burin spalls 1 0.31 0.32 7 0.26 0.27 3 0.28 0.29
Debris 23 0.87 8 0.74
Chips 3 0.93 58 2.19 19 1.76
Chunks 4 1.24 27 1.02 22 2.04
TOTAL: 323 100 100 2654 100 100 1079 100 100
Layer V GH 3 GH 4
n % % esse n % % esse n % % esse
Pre-forms 1 0.13 0.14 2 0.03 0.08
Cores 22 2.87 3.06 19 0.25 0.71
Flakes 452 58.93 62.95 1136 15.14 42.74 24 17.02 52.17
Blades 162 21.12 22.56 260 3.46 9.78 5 3.55 10.87
Bladelets 38 4.95 5.29 471 6.28 17.72 5 3.55 10.87
Micro-blades 1 0.13 0.14 472 6.29 17.76 7 4.96 15.22
Blank fragments 40 0.53 1.50
Tools 41 5.35 5.71 169 2.25 6.36 3 2.13 6.52
Tool/Core 1 0.01 0.04
Burin spalls 1 0.13 0.14 88 1.17 3.31 2 1.42 4.35
Debris 389 5.18 4 2.84
Chips 24 3.13 4440 59.16 89 63.12
Chunks/Pebbles 25 3.26 18 0.24 2 1.42
TOTAL: 767 100 100 7505 100 100 141 100 100
Fig. 5. Românti-Dumbrăva I. Artifacts: Mogoşanu’s (layers II-V) and 2009-2010 (GH3 and 4)
excavations.
Abb. 5. Româneti-Dumbvia I, Artefakthäufigkeiten. Ausgrabungen Mogoanu (Schichten II-V) und
Ausgrabungen 2009-2010 (Schichten GH3 und 4).
Quartär 59 (2012) V. Sitlivy et al.
92
sequence: opal remained the main raw material
category used for knapping (about 85 - 95 %) (for
details, see also Băltean 2011a, b).
General structure of Mogoşanu’s layers II, III, IV and V
The general composition of all four Mogoşanu’s
assemblages remains nearly unchanged throughout
the entire sequence, and is dominated by large
debitage products: flakes (51.1 % to 61.4 %) and bla-
des (21.1 % to 33.7 %). The frequency of bladelets
(4.0 % to 6.3 %), tools (5.0 % to 6.2 %), and especially
cores, is quite low (Fig. 5). The number of cores
increases slightly from bottom to top (0.9 % to 2.9 %),
also true for chunks, which become slightly more
abundant toward the top of the sequence. Chips (<15
mm) and small debris/fragments (<25 mm) occur in
small quantities due to different sieving practices.
Consequently, micro-blades are nearly absent in all
assemblages together with burin spalls, while burins
are common in all layers. The tool/core ratio is
moderate in the lowermost layers (5.3: 1 and 5.5: 1),
progressively decreasing from 3.9: 1 to 1.8: 1. The
blank-to-core ratio is high, showing significant core
reduction; it also decreases from bottom to top
(80.7: 1; 55.4: 1 and 29.6: 1 in layers III, IV and V
correspondingly).
The debitage structure throughout the sequence
shows the dominance of flakes (normally >60 %
including tools on flakes) over blades, which are
1.8 times less common at the bottom and 2.5 times at
the top, showing progressive decrease in laminar
blanks (from 38.8 % to 26.5 %). The laminar structure
reflects the stable and absolute dominance of blades
(>80 %) over smaller laminar products: bladelets (from
14.6 % to 18.4 %) and micro-blades (never exceeding
1.5 %) (Fig. 6).
Despite quantitative differences between layers,
cores and tools exhibit similar morphological, techno-
logical and typological patterns. In addition, the
presence and absence of tool types is more or less
identical throughout the sequence. The main tool
categories comprise endscrapers, burins, retouched
blades, retouched pieces on blades/flakes and non-
geometric microliths. In the richest layer III, these
tool types occur in similar frequencies, while in the
overlying layers endscrapers and especially burins are
more numerous than non-geometric microliths. The
latter were recorded in all levels, but are more
frequently in levels III and IV.
In sum, the assemblages from layers II, III, IV and V
do not exhibit major changes. Consequently, the
following section presents a comparative description
of the main classes of lithics, whereas data for the
separate assemblages is to be found in the cited tables.
Cores
Cores are rare although their frequency progressively
increases towards the top of the sequence (to 3.1 %).
These belong to three main groups: (a) blade/
bladelet/micro-blade carinated cores, (b) blade/let/
micro-blade “regular”, i.e. prismatic, including
burin-like cores, and (c) flake cores. Taking into
consideration only identifiable cores, significant
changes in their quantitative representation can be
Fig. 6. Româneşti-Dumbrăviţa I. Debitage and laminar structures: Mogoşanu’s (layers II-V) and 2009-2010 (GH3) excavations.
Abb. 6. Româneşti-Dumbrăviţa I, Grundformhäufigkeiten. Ausgrabungen Mogoşanu (Schichten II-V), Ausgrabungen 2009-2010 (Schichten
GH3 und 4).
Layer II Layer III Layer IV Layer V GH3
n % n % n % n % n %
DEBITAGE STRUCTURE
Flakes 165 52.9 1448 57.9 663 65.4 452 65.0 1136 44.0
Tool on flakes 3 1.0 62 2.5 15 1.5 17 2.4 31 1.2
Blades 109 34.9 719 28.7 234 23.1 162 23.3 260 10.1
Tool on blades 12 3.8 81 3.2 45 4.4 22 3.2 47 1.8
Burin spall on blades 1 0.1 3 0.1
Bladelets 19 6.1 168 6.7 43 4.2 38 5.5 471 18.2
Tool on bladelets 1 0.3 15 0.6 5 0.5 2 0.3 19 0.7
Burin spall on bladelets 1 0.3 2 0.2 28 1.1
Micro-blades 2 0.6 7 0.3 3 0.3 1 0.1 472 18.3
Tool on micro-blades 3 0.1 2 0.2 57 2.2
Burin spall on micro-blades 1 0.1 57 2.2
TOTAL: 312 100.0 2503 100.0 1013 100.0 695 100.0 2581 100.0
LAMINAR STRUCTURE
Blades 121 84.0 800 80.6 280 83.6 184 81.4 310 21.9
Bladelets 21 14.6 183 18.4 50 14.9 40 17.7 518 36.6
Micro-blades 2 1.4 10 1.0 5 1.5 2 0.9 586 41.4
TOTAL: 144 100.0 993 100.0 335 100.0 226 100.0 1414 100.0
Quartär 59 (2012)Earliest Aurignacian in Romania: the open air site of Româneşti-Dumbrăva
93
observed. While in lowermost layer II only two
carinated cores were recorded, in the overlying layer
III these are nearly as frequent as “regular” laminar
cores. The latter category becomes dominant towards
the top (Fig. 7). The core reduction was usually aimed
at laminar production. Pre-cores are rare and were
documented in uppermost layers IV and V (4 and 3
items). Carinated cores occur in all layers with
different frequencies showing variability in final
products/reduction stages (blade/bladelet, flake/
bladelet and bladelet/micro-blade), debitage direction,
platform(s)/working surface(s) number and position
Layer
II
Layer
III
Layer
IV
Layer
V
GH
3
CARINATED BLADELET/MICRO-BLADE:
unidirectional 2 2
bidirectional 1 1
orthogonal-adjacent 1
orthogonal-alternate 1
CARINATED BLADE/BLADELET:
unidirectional 1 2 2 2
bidirectional 1
bidirectional-adjacent 1
CARINATED FLAKE/BLADELET:
unidirectional 2
BLADE:
unidirectional 2 1
unidirectional, narrow flaking surface 2
bidirectional 1 2
bidirectional-adjacent 1
BLADE/BLADELET:
unidirectional 2 1
unidirectional, narrow flaking surface 2 1
multiridectional, narrow flaking surface 1 1
bidirectional, narrow flaking surface 1 1
bidirectional 2 1
BLADELET/MICRO-BLADE:
unidirectional 2
unidirectional, narrow flaking surface 1 3
orthogonal-adjacent, narrow flaking surface 1 1
bidirectional, narrow flaking surface, on flake 1
BLADE/BLADELET ON TOOL
change orientation, narrow flaking surface, on scraper 1
unidirectional, narrow flaking surface, on scraper 1
bidirectional, narrow flaking surfaces, on scraper 1 2
FLAKE/BLADELET, sub-polyhedral 2 1 1
FLAKE:
semi-polyhedral 1
discoidal 1 1
semi-discoidal 1
crossed, on scraper, Kombewa 1
orthogonal, trifacial 1
UNIDENTIFIABLE 1 6 4 2 4
TOTAL: 3 30 13 19 17
Fig. 7. Româneti-Dumbrăvia I. Cores: Mogoanu’s (layers II-V) and 2009-2010 (GH3) excavations.
Abb. 7. Româneti-Dumbvia I, Häufigkeiten von Kernformen. Ausgrabungen Mogoanu (Schichten II-V),
Ausgrabungen 2009-2010 (Schichten GH3 und 4).
Quartär 59 (2012) V. Sitlivy et al.
94
Fig. 8. Româneşti-Dumbrăviţa I, Mogoşanu ‘s excavations, layer III. Core: carinated blade/bladelet, unidirectional, narrow flaking
surface, keeled.
Abb. 8. Româneti-Dumbrăvia I, Ausgrabungen Mogoanu, Schicht III. Unidirektionaler gekielter Klingen-/Lamellenkern mit schmaler
Abbaufläche.
(e.g. unidirectional, bidirectional, orthogonal and
their combination: adjacent – two opposed or 9
striking platforms where two flaking surfaces are
adjacent, or alternate – where two flaking surfaces are
opposite, isolated) (Figs. 8; 9; 10: 1, 2, 3). Carinated
unidirectional blade/bladelet and bladelet/micro-
blade cores are the most representative. They are
generally made on massive flakes, but also on nodules,
reflecting different reduction stages, mostly the full
debitage and the initial stages, with corresponding
rather large sizes. Striking platforms are usually plain
(sometimes crudely faceted), with acute angles.
Narrowing of core flanks by flake removals is common.
Regular” laminar cores comprise (a) prismatic and
(b) narrow-faced with rectangular or triangular/keeled
flaking surface(s) on the thin slice/edge of a core blank.
Prismatic cores are quite diverse in regard to final
products/reduction stages (blade, blade/bladelet,
bladelet/micro-blade, flake/blade and flake/bladelet),
debitage direction and platform(s)/working surface(s)
placement. These cores were made on chunks,
nodules and pebbles bearing some remnants of crests
(Figs. 11: 1, 2, 3; 12: 1). Narrow-faced cores are
represented by unidirectional on flakes (often “burin-
like), but also bidirectional, multidirectional and
orthogonal-adjacent on flattish fragments/plaquettes
(including recycling of tools) (Figs. 12: 2, 3; 13: 1).
These cores show a rather advanced reduction stage
Quartär 59 (2012)Earliest Aurignacian in Romania: the open air site of Româneşti-Dumbrăva
95
Fig. 9. Româneşti-Dumbrăviţa I, Mogoşanu’s excavations, layer III. Core: carinated blade/bladelet,
unidirectional, narrow flaking surface, keeled.
Abb. 9. Româneti-Dumbrăvia I, Ausgrabungen Mogoanu, Schicht III. Unidirektionaler gekielter
Klingen-/Lamellenkern mit schmaler Abbaufche.
(extension of narrow working surface to the wide side
or several reduction zones) with or without crest
remnants. Striking platforms are plain or crudely
faceted; angles are almost acute. Flake cores are rare
and include discoidal (Fig. 10: 2), semi-discoidal,
orthogonal and Kombewa (4 pieces).
Laminar debitage
Quantitative analysis of laminar debitage is based on
the layer III sample, which yielded by far the most
numerous assemblage of the old collection. However,
only minor differences throughout the sequence were
recorded. Therefore, the debitage of layer III can be
regarded as representative for all Aurignacian layers
at Româneşti-Dumbrăviţa I. In layer III, blades are the
second most common artifact class (after flakes)
(n=719), but only 13.8 % (or 15.4 % including tools) of
blades are complete. Bladelets are numerous (n=168),
with more complete pieces (23.8 %, or 24.9 %
including tools) and few micro-blades). The metrical
data are as follows: (1) complete blade, max. size is
108.6, 41.6, 14.7 mm (an average of 34.8, 19.3, 5.6 mm);
complete bladelet max. size is 52.7, 11.5, 2.8 mm (an
average of 23.9, 10.2, 3.2 mm); complete micro-blade
max. size – 30.5, 6.8, 1.6 mm (an average of 20.7, 6.4,
2.5 mm). Many removals are lacking cortex: flakes
(78 %), blades (86.8 %) and bladelets/micro-blades
(93.8 %). Among the primary flakes, cortical ones
(>76 % of cortex) are quite rare (n=57, < 4 %), while
there are only two such primary blades. Cortex
position is typically lateral (>60 %). Only 11 bladelets/
micro-blades have <25 % cortex. A unidirectional scar
pattern is dominant for all laminar products: blades/
tool-blades accounting for 66.7 % and bladelets/
micro-blades for 70.5 %. The second most frequent
blank scar pattern is convergent: 15.6 % and 16.5 % for
blades and bladelets respectively. Other laminar
dorsal scars are less common (about 5 - 6 %: crested,
unidirectional-crossed) or rare (e.g., bidirectional).
Blade shapes are mostly rectangular (up to 40 %), and
trapezoidal (21.6 %); other shapes are rare (irregular
– 13.4 %, ovoid – 10.3 %, triangular – 8 % and crescent
Quartär 59 (2012) V. Sitlivy et al.
96
6.8 %). Rectangular bladelets keep priority (up to
45 %), while triangular (14.6 %) and ovoid (14.6 %)
shapes slightly increase. Debitage symmetry (on-axis)
of laminar products is dominant (68.1 % of blades and
63.9 % of bladelets) over off-axis products, however
the latter are still quite common. Blade/let profiles are
twisted (37.1 %/37.7 %), curved (33.2 %/28.4 %) and
flat (25.9 %/32.2 %). Twisted profiles increase in
Fig. 10. Româneşti-Dumbrăviţa I, Mogoşanu’s excavations, layer III. Carinated bladelet cores: 1, 2 – bidirectional, sub-cylindrical;
3 – unidirectional, sub-pyramidal.
Abb. 10. Româneti-Dumbrăvia I, Ausgrabungen Mogoanu, Schicht III. Gekielte Lamellenkerne: 1, 2 – bidirektional, sub-zylindrisch;
3 – unidirektional, sub-pyramidal.
Quartär 59 (2012)Earliest Aurignacian in Romania: the open air site of Româneşti-Dumbrăva
97
Fig. 11. Româneşti-Dumbrăviţa I, Mogoşanu’s excavations, layer III. Blade/bladelet prismatic cores:
1 – bidirectional; 2, 3 – unidirectional, keeled.
Abb. 11. Româneti-Dumbrăvia I, Ausgrabungen Mogoanu, Schicht III. Prismatische Kerne:
1 – bidirektional; 2, 3 – unidirektional, gekielt.
overlying layer IV, while curved profiles decrease
markedly (Fig. 14). Blade/let distal ends are mostly
feathered (59.8 %/62.7 %), less blunt (20.5 %/14.5 %),
hinged (13.7 %/21.7 %) and rarely overpassed
(6 %/1.2 %). Blade cross-sections are principally
trapezoidal (41.4 %) and triangular (31.3 %), followed
by lateral steep (scalene) (18.5 %) and rare multiple
(8.0 %) ones. On the other hand, triangular sections
are more frequent (44.8 %) for bladelets, although
trapezoidal (34.9 %) and lateral steep (16.1 %) sections
are common. The abundance of triangular sections
may also be explained by the frequent use of narrow-
faced cores to obtain narrow laminar blanks. Laminar
removals with lateral steep sections also maintain
working convexity during debitage. With respect to
platform preparation, single blow platforms are
dominant. Plain butts are the most numerous among
blades (68.4 %), followed by some dihedral and linear;
other butts are rare (Fig. 15). Bladelets show a rise in
linear (33.3 %) and punctiform (10.8 %) butts and a
decrease in flat platforms (49 %). Blade butt lipping is
common (including semi-lipped, which show the
combination of butt lips and bulb of percussion on the
blank’s ventral surface), while bladelets show a decline
Quartär 59 (2012) V. Sitlivy et al.
98
Fig. 12. Româneşti-Dumbrăviţa I, Mogoşanu’s excavations, layer III (1, 3) and layer V (2). 1blade
core, bidirectional, sub-cylindrical; 2 – bladelet narrow flaking surface core, unidirectional, keeled;
3 – blade/bladelet narrow flaking surface core, unidirectional, keeled.
Abb. 12. Româneti-Dumbrăvia I, Ausgrabungen Mogoanu, Schicht III (1, 3) und Schicht V (2).
1 – Klingenkern, bidirektional, sub-zylindrisch; 2 – unidirektionaler gekielter Lamellenkern mit schmaler
Abbaufläche; 3 – unidirektionaler gekielter Klingen-/Lamellenkern mit schmaler Abbaufläche.
Quartär 59 (2012)Earliest Aurignacian in Romania: the open air site of Româneşti-Dumbrăva
99
Fig. 13. Româneşti-Dumbrăviţa I, Mogoşanu’s excavations, layer III. Cores: 1 – bladelet narrow flaking surface, unidirectional, made on
scraper; 2 – discoidal.
Abb. 13. Româneti-Dumbrăvia I, Ausgrabungen von Mogoanu, Schicht III. Kerne: 1 – unidirektionaler Lamellenkern mit schmaler Abbaufläche
an Schaber; 2 – diskoider Kern.
in lipped butts with more unlipped butts (Fig. 16). As
for bulbs, the diffused dominate over the developed
ones for blades and bladelets; the bulb absence is
common (Fig. 17). Split/shattered bulbs are rare. The
domination of obtuse and inverted angles was
recorded for both blades and bladelets, while right
angles show the same low frequency (Fig. 18). The
abrasion of the blade butt edges was frequent (53.2 %
as well as butt reduction by faceting – up to 60 %),
while this practice declines for bladelets (32.4 %). On
the other hand, bladelet trimming of the overhang by
faceting (small removals) was more common (51.5 %).
Thus, two techniques were practiced to eliminate
overhang on laminar cores as well as their combi-
nation.
Core maintenance products are frequent: in layer
III these are represented by débordant flakes (69),
crested flakes (29), crested blades (50), crested
bladelets (10), tablet-flakes (29), tablet-blades (4),
tablet-bladelets (1), core flank-flakes (56) and a flank-
blade (1). Crested blades have one or two prepared
slopes (6) with a central position on the core (5),
sometimes partially prepared (4) or complete and
include primary, secondary and neo-crests.
Quartär 59 (2012) V. Sitlivy et al.
100
Fig. 15. Româneşti-Dumbrăviţa I. Laminar butts: Mogoşanu’s (layers III-IV) and 2009-2010 (GH3) excavations.
Abb. 15. Româneti-Dumbrăvia I, Klingenproduktion, Häufigkeiten der unterschiedenen Formen der Schlagflächenreste. Ausgrabungen
Mogoanu (Schichten III-IV), Ausgrabungen 2009-2010 (GH3).
Layer III Layer IV GH3
Blade Bladelet Blade
Bladelet/
Micro-blade
Blade Bladelet Micro-blade
n % n % n % n % n % n % n %
cortical 5 1.4 1 0.8 1 4.5 4 3.7 2 1.0 1 0.6
plain 242 68.4 50 49.0 81 65.9 11 50.0 67 62.0 102 50.0 44 28.2
plain-abraded 3 0.8
punctiform 15 4.2 11 10.8 5 4.1 1 4.5 6 5.6 9 4.4 16 10.3
linear 27 7.6 34 33.3 10 8.1 9 40.9 17 15.7 77 37.7 94 60.3
dihedral 35 9.9 4 3.9 16 13.0 10 9.3 11 5.4 1 0.6
crudely-faceted 13 3.7 2 2.0 6 4.9 4 3.7 2 1.0
fine faceted 9 2.5 1 1.0 4 3.3
spur 5 1.4
abraded 1 0.5
TOTAL: 354 100.0 102 100.0 123 100.0 22 100.0 108 100.0 204 100.0 156 100.0
Layer III Layer IV
Blade Bladelet Blade
Bladelet/Micro-
blade
n % n % n % n %
convex 12 1.7 1 0.5 2 0.8 1 2.0
flat 181 25.9 59 32.2 69 26.6 15 29.4
incurvate 232 33.2 52 28.4 66 25.5 6 11.8
twisted 259 37.1 69 37.7 114 44.0 29 56.9
irregular 14 2.0 2 1.1 8 3.1
TOTAL: 698 100.0 183 100.0 259 100.0 51 100.0
GH3, blade GH3, bladelet
broken &
complete
complete
broken &
complete
complete
n % n % n % n %
convex 2 0.8 1 4.3 4 1.1 4 8.5
flat 108 43.7 4 17.4 183 49.3 14 29.8
curved 42 17.0 7 30.4 43 11.6 3 6.4
twisted 94 38.1 11 47.8 139 37.5 25 53.2
irregular 1 0.4 2 0.5 1 2.1
TOTAL: 247 100.0 23 100.0 371 100.0 47 100.0
GH3, micro-blade
broken &
complete
complete
n % n %
convex 1 0.2 1 2.3
flat 246 56.3 16 37.2
curved 41 9.4 9 20.9
twisted 148 33.9 17 39.5
irregular 1 0.2
TOTAL: 437 100.0 43 100.0
Fig. 14. Româneşti-Dumbrăviţa I. Laminar lateral profiles: Mogoşanu’s (layers III-IV) and 2009-2010
(GH3) excavations.
Abb. 14. Româneti-Dumbrăvia I, Klingenproduktion, Häufigkeiten der unterschiedenen Längsprofile.
Ausgrabungen Mogoanu (Schichten III-IV), Ausgrabungen 2009-2010 (GH3).
Quartär 59 (2012)Earliest Aurignacian in Romania: the open air site of Româneşti-Dumbrăva
101
Layer III Layer IV GH3
Blade Bladelet Blade
Bladelet/
Micro-blade
Blade Bladelet Micro-blade
n % n % n % n % n % n % n %
lipped 156 43.1 25 23.1 69 55.6 7 29.2 57 51.8 67 32.7 28 17.6
semi-lipped 125 34.5 40 37.0 33 26.6 7 29.2 25 22.7 101 49.3 82 51.6
unlipped 81 22.4 43 39.8 22 17.7 10 41.7 28 25.5 37 18.0 49 30.8
TOTAL: 362 100.0 108 100.0 124 100.0 24 100.0 110 100.0 205 100.0 159 100.0
Fig. 16. Româneşti-Dumbrăviţa I. Laminar butt lipping: Mogoşanu’s (layers III-IV) and 2009-2010 (GH3) excavations.
Abb. 16. Româneti-Dumbrăvia I, Klingenproduktion, Häufigkeiten der unterschiedenen Ausprägungen der Schlaglippen. Ausgrabungen
Mogoanu (Schichten III-IV), Ausgrabungen 2009-2010 (GH3).
Layer III Layer IV GH3
Blade Bladelet Blade
Bladelet/
Micro-blade
Blade Bladelet Micro-blade
n % n % n % n % n % n % n %
developed 92 25.3 35 31.8 28 22.2 11 44.0 22 19.6 53 24.9 61 38.4
diffused 197 54.3 57 51.8 74 58.7 11 44.0 60 53.6 105 49.3 64 40.3
split/shattered 8 2.2 2 1.8 3 2.4 1 4.0 2 1.8 10 4.7 2 1.3
absent 66 18.2 16 14.5 21 16.7 2 8.0 28 25.0 45 21.1 32 20.1
TOTAL: 363 100.0 110 100.0 126 100.0 25 100.0 112 100.0 213 100.0 159 100.0
Fig. 17. Româneşti-Dumbrăviţa I. Laminar bulbs: Mogoşanu’s (layers III-IV) and 2009-2010 (GH3) excavations.
Abb. 17. Româneti-Dumbrăvia I, Klingenproduktion, Häufigkeiten der unterschiedenen Ausprägung der Bulben. Ausgrabungen Mogoanu
(Schichten III-IV), Ausgrabungen 2009-2010 (GH3).
Layer III Layer IV GH3
Blade Bladelet Blade
Bladelet/
Micro-blade
Blade Bladelet Micro-blade
n % n % n % n % n % n % n %
inverted 67 20.9 8 11.9 14 12.6 21 22.8 15 9.8 4 5.2
obtuse 178 55.5 43 64.2 79 71.2 11 61.1 57 62.0 107 69.9 48 62.3
right 72 22.4 16 23.9 18 16.2 7 38.9 14 15.2 30 19.6 25 32.5
acute 4 1.2 1 0.7
TOTAL: 321 100.0 67 100.0 111 100.0 18 100.0 92 100.0 153 100.0 77 100.0
Fig. 18. Româneşti-Dumbrăviţa I. Laminar butt angles: Mogoşanu’s (layers III-IV) and 2009-2010 (GH3) excavations.
Abb. 18. Româneti-Dumbrăvia I, Klingenproduktion, klassierte Winkel zwischen Schlagfläche und Ventralfläche. Ausgrabungen Mogoanu
(Schichten III-IV), Ausgrabungen 2009-2010 (GH3).
Tool-bladelets are present in small quantity, and their
frequency declines throughout the sequence. Tool-
micro-blades are rare. The scarcity of tools on
bladelets and micro-blades can be explained by past
excavation practices, considerably different from the
present-day.
Endscrapers in layer III were made on blades as
well as on flakes, while in the uppermost layer V all
seven tools were made only on flakes (Fig. 20).
Carinated and thick endscrapers were usually made
on massive flakes (Figs. 21: 1, 2; 22: 1), while simple
ones mostly on (quite massive) blades (Figs. 22: 3;
23: 1, 3, 6). Both carinated (with sub-parallel/parallel
retouch) and thick endscrapers (mostly with scalar
modifications, including oval and shouldered ones)
are frequent in layer III and IV. Only simple and thick
endscrapers on flakes were recorded in uppermost
layer V. Carinated (Fig. 23: 5) and thick endscrapers as
Secondary crested blades are frequent (26), often
including lateral examples (13) from the sides of the
core and some neo-crests (6) showing repeated core
maintenance. Crested blades/lets were also selected
for tool production, as well as some tablets and
débordant flakes.
Tools
Tool manufacture in all Aurignacian layers of
Româneşti-Dumbrăviţa I relied mostly on opal of
various qualities (e.g. about 87 % of tool-blades, tool-
bladelets in layers III and IV). Different types of flint
occur sporadically, as well as quartzite. Isolated pieces
were made on chalcedony, jasper, radiolarite (the last
raw material being better represented in laminar
debitage). Tools were made on both blades and flakes,
more on blades, however without proportional
changes towards the top of the sequence (Fig. 19).
Quartär 59 (2012) V. Sitlivy et al.
102
Layer
II
Layer
III
Layer
IV
Layer
V
GH 3 GH 4
n % esse n % esse n % esse n % esse n % esse n
Endscraper 1 8.3 18 12.2 7 12.3 7 20.0 2 1.4
Borer 1 0.7
Burin 2 16.7 17 11.5 10 17.5 12 34.3 15 10.6
Combined tool 1 8.3 1 0.7 1 2.9
Retouched blade 1 8.3 16 10.8 11 19.3 4 11.4 5 3.5
Retouched pieces on blade 2 16.7 24 16.2 7 12.3 4 11.4 16 11.3
Notched piece 1 8.3 9 6.1 3 5.3 1 2.9 4 2.8
Denticulated piece 3 2.0
Sidescraper 7 4.7 2 3.5 3 8.6 3 2.1
Retouched piece on flake 22 14.9 4 7.0 12 8.5
Truncated piece 3 25.0 5 3.4 3 5.3 2 1.4
Thinned piece 5 3.4
Pieces esquillées 1 0.7 3 5.3 1 2.9 3 2.1
Non-geometric microlith 1 8.3 19 12.8 7 12.3 2 5.7 80 56.3 3
Unidentifiable tool 4 13 10 6 27
TOTAL: 16 100.0 161 100.0 67 100.0 41 100.0 169 100.0 3
tools on blades 12 75.0 81 50.3 45 67.2 22 53.7 47 30.5
tools on bladelets 1 6.3 15 9.3 5 7.5 2 4.9 19 12.3
tools on micro-blades 3 1.9 2 3.0 57 37.0 3
tools on flakes 3 18.8 62 38.5 15 22.4 17 41.5 31 20.1
TOTAL: 16 100.0 161 100.0 67 100.0 41 100.0 154 100.0 3
Tools on Aurignacian blades 1 3 4 3 2
Fig. 19. Româneşti-Dumbrăviţa I. Tool types: Mogoşanu’s (layers II-V) and 2009-2010 (GH3 and 4) excavations.
Abb. 19. Româneti-Dumbrăvia I, Häufigkeiten der Werkzeugklassen. Ausgrabungen Mogoanu (Schichten II-V), Ausgrabungen 2009-2010
(GH3 und 4).
Layer
II
Layer
III
Layer
IV
Layer
V
GH
3
ENDSCRAPER ON BLADE:
carinated 1
thick 1
double 1
simple 1 7 3 2
unidentifiable 1
ENDSCRAPER ON FLAKE:
carinated 3
thick 5 2 3
simple 2
fan-shaped 1
divergent (ovoid) 1
unidentifiable 1
TOTAL: 1 18 7 7 2
Fig. 20. Româneşti-Dumbrăviţa I. Endscrapers: Mogoşanu’s
(layers II-V) and 2009-2010 (GH3) excavations.
Abb. 20. Româneti-Dumbrăvia I, Häufigkeiten der Kratzer.
Ausgrabungen Mogoanu (Schichten II-V), Ausgrabungen 2009-
2010 (GH3).
well as simple ones show variability, including oval,
shouldered, and double specimens on lateral/bilateral
retouched blanks, including Aurignacian blades
(Figs. 22: 2, 4; 23: 2, 4). Some were truncated.
Burins are numerically significant in all layers and
highly variable throughout the sequence (Figs. 24; 25;
26: 1, 3, 6). These were produced more often on
blades than on flakes. Angle burins on snap are the
common type in all layers, as well as dihedral, with
only rare occurrences of carinated, dihedral angle
(angle on transverse burin facet), double, mixed and
isolated busked types. Burins on retouched blades
(lateral, bilateral and also Aurignacian) and various
truncations are common. Transverse and flat burins
are also observed. In one case (layer V), a double
angle burin on snap was made on a retouched bladelet.
The scarcity of burin spalls in all layers (12 in total)
contrasts with the high frequency of burins.
Combined tools were documented in different
layers: simple endscraper + angle burin on snap on
blade (layer II), thick shouldered endscraper on
retouched flake + transverse burin on snap (layer III)
(Fig. 22: 1) and simple endscraper + straight truncation
on retouched piece on blade (layer V) (see Hahn 1977,
Plate 169: 8).
Retouched blades. For unknown reasons, the study
sample of retouched blades and retouched pieces on
Quartär 59 (2012)Earliest Aurignacian in Romania: the open air site of Româneşti-Dumbrăva
103
Fig. 21. Româneşti-Dumbrăviţa I, Mogoşanu’s excavations, layer III. Endscrapers on flakes: 1 – thick, double alternate, ogival/shouldered;
2 – thick.
Abb. 21. Româneti-Dumbvia I, Kratzer an Abschgen. Ausgrabungen Mogoanu, Schicht III: 1 – Doppelkratzer an massivem Abschlag: oval/
Nasenkratzer; 2 – massiv.
Quartär 59 (2012) V. Sitlivy et al.
104
blades is much bigger in comparison with the
previously published data. Regardless, retouched
blades (with continuous, non-marginal, quite invasive
scalar lateral/bilateral, convergent/pointed obverse,
inverse and alternate mostly semi-steep retouch) are
numerically significant (n=32), including Aurignacian
types (n=11). These and other tool types made on
Aurignacian blades occur in all layers (Fig. 27).
Retouched pieces on blades and on flakes with
light, short discontinuous or partial retouch (while
non-marginal) are common throughout the sequence
(Fig. 19). Pieces on blades have usually obverse lateral
semi-steep retouch.
Notched pieces are well represented in layer III
(n=9), less in layer IV (3) and were made more often on
blades than on flakes. The notches are often lateral,
but can also be bilateral, proximal, distal and lateral/
distal. The retouch (usually scalar, semi-steep and
steep) is mostly obverse and rarely inverse. Notched
Aurignacian bilateral (Fig. 27: 1) and lateral blades
were documented in layers III and IV (Hahn 1977, Plate
169: 2).
Denticulated pieces on flakes were recorded only
in layer III: distal, dorsal (2) and lateral dorsal (1).
Sidescrapers occur in all assemblages (except layer
II) in comparable proportions and in small numbers
(Fig. 19) and were produced on flakes by continuous
steep, semi-steep and flat retouch. They are of two
main types: lateral and transverse with convex or
straight working edge, except for one angle (lateral/
transverse) type. The retouch is mostly obverse,
while ventral, alternate and bifacial retouch occurs
episodically (Fig. 26: 4).
Truncated pieces appeared in all layers, except
uppermost layer V. Two of these are distal on
retouched blades, including Aurignacian retouch, and
Fig. 22. Româneşti-Dumbrăviţa I, Mogoşanu’s excavations, layer III (1, 3, 4) and layer IV (2). Endscrapers: 1 – thick, shouldered, on laterally
retouched flake with transverse burin on snap; 2 – thick shouldered, on Aurignacian laterally retouched blade; 3 – simple on blade;
4 – double, on Aurignacian alternatively retouched blade.
Abb. 22. Româneti-Dumbrăvia I, Ausgrabungen Mogoanu Schicht III (1, 3, 4) und Schicht IV (2). Kratzer: 1 – massiver Nasenkratzer an kanten-
retuschiertem Abschlag und Transversalstichel an Bruchkante am gegenüberliegenden Werkzeugende; 2 – massiver Nasenkratzer an Klinge mit
Aurignacien-Retusche. 3 – einfach an Klinge; 4 – doppelt, an retuschierter Klinge.
Quartär 59 (2012)Earliest Aurignacian in Romania: the open air site of Româneşti-Dumbrăva
105
Fig. 23. Româneşti-Dumbrăviţa I, Mogoşanu’s excavations, layer III: 1, 6 – simple endscraper, on blade; 2 – endscraper, on Aurignacian
laterally retouched blade; 3 – simple endscraper, on bilaterally retouched blade; 4 – thick endscraper, on Aurignacian bilaterally retouched
blade; 5 – carinated endscraper, on blade.
Abb. 23. Roneti-Dumbrăvia I, Ausgrabungen Mogoanu, Schicht III: 1, 6 – einfacher Kratzer an Klinge; 2 – Kratzer an Klinge mit Aurignacien-
Retusche; 3 – Kratzer an beidseitig retuschierter Klinge; 3 – massiver Kratzer an Klinge mit beidseitiger Aurignacien-Retusche; 5 – Kielkratzer an
Klinge.
one is proximal on blade (layer II), 3 are proximal
straight and 2 are distal (layer III) (Fig. 26: 2, 5) and 2
are distal and 1 is proximal (layer IV) (Hahn 1977,
Plates 169:3; 168:3). They were made on laterally/
bilaterally retouched and non-retouched blades with
oblique, straight and concave truncation (except one
case on flake).
Thinned pieces (sensu Geneste 1985) occurred
only in layer III and were produced on flakes (3) and
blades (2). These are backed, lateral, proximal, and on
snap by dorsal and ventral flat retouch.
Pieces esquillées (scaled tools sensu Marks 1976 or
splintered pieces) on both flakes and blades, are rare
and generally uncommon for the Banat Aurignacian.
Non-geometric microliths (sensu Hours 1974) were
recorded in all layers, more frequently in layers III and IV
(12.8 % or 19 and 7 in number). The 29 “micro-tools”
were made on bladelets and only 5 on micro-blades.
These are represented by: (a) Dufour bladelets/micro-
blades (alternate or inverse fine/micro-scalar, semi-steep
retouch), pseudo-Dufour bladelets/micro-blades
(idem obverse lateral and bilateral retouch) (c) Font-
Yves point (idem obverse lateral retouch) and (d) varia
(“small-sized” modified bladelets: 2 lateral notches,
lateral/proximal denticulate, proximal oblique
truncation, burin and bladelet with invasive flat
retouch – some of them with possible projectile
impact traces) (Figs. 28; 29). In the most representative
Dufour sample (layer III), modified lamellar blanks
were mostly symmetrical (on-axis) with nearly equally
represented flat (7), twisted (6) and curved (4) profiles.
In terms of fragmentation mode and metrics, only 9 of
Quartär 59 (2012) V. Sitlivy et al.
106
29 microliths are complete: max. size is 59.5, 10.5,
5.5 mm; min. 18.7, 9.3, 2.9 mm; average 35.9, 9.9,
2.9 mm; average W is 9.6 mm and T is 2.8 mm. One
ventral Dufour in layer III was made on a small
proximally broken blade (>33.2, 13.6, 4.6 mm) by fine
alternate semi-steep bilateral/distal retouch (Fig. 29:
11). Blanks of all non-geometrical microliths have
unidirectional (17) and some convergent (5) dorsal
scars and show mostly “on-axis” detachment (22).
Twisted (14), flat (10) and curved (5) lateral profiles
were recorded. Only 2 micro-notches and 1 Dufour
(layers II, III and IV) show an “off-axis” pattern
combined with twisted profiles. As for butts, these are
usually plain and linear (Th< 2 mm), often lipped/
semi-lipped, with diffused bulbs, obtuse internal butt
angles and quite rare overhang reduction by small
removal trimming (7) or abrasion (3).
Results of the 2009 - 2010 excavations
New excavations aimed at collecting samples for the
dating program, coupled with comprehensive geo-
logical analyzes and a more accurate contextualization
of the lithic assemblages from the preserved stratified
part of the settlement. The new researches focused
on the character of soils and sediments, palaeo-
environment and chronostratigraphy, state of preser-
vation, and the actual content of the archeological
remains. In comparison to the already large size of the
area previously excavated, the new researches were
much more restricted. Field work concentrated on
well documented punctual survey trenches near to
Mogoşanu’s main trench where all of his layers
were recorded to be present, albeit with different
clustering. Pre-requisites for the selection of the
spatial position of the surveys were therefore the
localization of Mogoşanu´s trenches and the identifi-
cation of a more or less horizontal position to secure
preservation of sediments. The interdisciplinary field
research at Româneşti-Dumbrăviţa I took place in
2009 - 2010. After the localization of numerous old
trenches, a grid system was established and two test
pits were excavated until the lowermost archaeo-
logical horizon (layer I according to Mogoşanu) was
reached. One of these, Trench 4 to the West of the
central part of the former main trench, opened a high-
density cluster. The excavation methodology
comprised 3D recording of all artifacts regardless of
their size and wet sieving of sediments recovered from
each quarter of 1 m². The new excavations, while
small-scaled, provided numerous lithic remains,
which complemented the old data, especially the
microlithic component, generally lost during
Mogoşanu´s investigation.
Stratigraphy
In general, the sedimentary cover at Româneşti is
comparably thin and artifacts are buried close to the
surface, which both complicates geo-archaeological
analysis. The combination of different methods of
sedimentology, geochemistry and luminescence
dating nevertheless showed that all findings above the
lowermost layer I (“Quartzitic industry”) belong to the
last glacial cycle (for details see Kels et al. subm.).
When compared to the original description of
sediments by Mogoşanu (1978), it can be said that
during re-excavation the same main geological
features were detected (Fig. 4: 1). However, it turned
out that the upper part of the sequence, including
most of the archaeological layers, is dominated by
the surface soil, here a Stagnic Albeluvisol. This soil
developed under moderate climate and temporary
soil wetness and is quite common on comparable
sediments in flat or dell positions of the Banat foothills
(Ianoş 2002; Mavrocordat 1971). The soil can be
subdivided into two general horizons below the humic
horizon, which has been further subdivided into a
plough-horizon (Fig. 4: 1 - GH 1) and a humic horizon
below (Fig. 4: 1 - GH 2): a bleached, light brown to
grey (albic) horizon (Fig. 4: 1 - GH 3), followed by a
brownish to reddish, weakly clay illuviated horizon
below (Fig. 4: 1 - GH 4), which is more or less rich on
Layer
II
Layer
III
Layer
IV
Layer
V
GH
3
BURIN ON BLADE:
angle, on snap 6 5 6 5
angle, on butt 1
angle, on truncation 1 2
angle double ,on snap 1
angle double, on trunca-
tion
1
transverse 1
transverse, on natural
truncation
1
dihedral 1 1
dihedral, on truncation 1
dihedral busqué 1
double mixed 1 2
carinated 1
BURIN ON FLAKE:
flat 1
angle, on snap 1 1 1 2 2
angle busqué, on trunca-
tion
1
angle, on truncation 1 1
transverse 1 1
dihedral 1 1
carinated 1 1
multiple/bladelet core (?) 1
TOTAL: 2 17 10 12 15
Fig. 24. Româneşti-Dumbrăviţa I. Burins: Mogoşanu’s (layers II-V)
and 2009-2010 (GH3) excavations.
Abb. 24. Româneti-Dumbrăvia I, Häufigkeiten der Stichelformen.
Ausgrabungen Mogoanu (Schichten II-V), Ausgrabungen 2009-
2010 (GH3).
Quartär 59 (2012)Earliest Aurignacian in Romania: the open air site of Româneşti-Dumbrăva
107
Fig. 25. Româneşti-Dumbrăviţa I, Mogoşanu’s excavations, layer III (1, 2, 6, 7) and layer IV (3, 4, 5). Burins: 1 – carinated, dihedral, multifaceted;
2 – busked, dihedral, with lateral notch; 3 – carinated, dihedral; 4 – angle, on straight truncation; 5 – angle, on snap, on bilaterally retouched
blade; 6 – double mixed, dihedral asymmetric/transverse, on bilaterally retouched blade; 7 – transverse, on natural truncation – crest.
Abb. 25. Româneti-Dumbrăvia I, Ausgrabungen Mogoanu, Schicht III (1, 2, 6, 7) und Schicht IV (3, 4, 5). Stichel: 1 – Kielstichel; 2 – Bogenstichel;
3 – Kielstichel; 4 – Stichel an Endretusche; 5 – Stichel an Bruch an retuschierter Klinge; 6 – Doppelstichel: Mehrschlagstichel/Transversalstichel an
retuschierter Klinge; 7 – Stichel an Bruch.
Quartär 59 (2012) V. Sitlivy et al.
108
Fig. 26. Românti-Dumbrăva I, Mogoşanu’s excavations, layer III (2, 4, 5) and layer V (1, 3, 6): 1 – double opposed burin, on snap/truncation,
on laterally retouched Aurignacian blade; 2 – truncated piece, on laterally inversely retouched blade; 3 – burin carinated, double opposed;
4 – scraper, double-convex, alternate; 5 – truncated piece, on laterally obversely retouched blade; 6 – double mixed burin, transverse/
dihedral asymmetric, on laterally retouched blade
Abb. 26. Româneti-Dumbvia I, Ausgrabungen Mogoanu, Schicht III (2, 4, 5) und Schicht V (1, 3, 6): Doppelstichel an Bruch/an Endretusche
an Aurginacien-Klinge; 2 – Endretusche an lateral ventral retuschierter Klinge; 3 – Kielstichel/Kielstichel; 4 – Doppelschaber: konvex/konvex,
alternierend; 5 – Endretusche an retuschierter Klinge; 6 – kantenretuschierter Doppelstichel: Transversalstichel, asymmetrischer Mehrschlagstichel.
Fig. 27. Româneşti-Dumbrăviţa I, Mogoşanu’s excavations, layer III (1, 3, 6), layer IV (2, 4) and layer V (5). Blades: 1, 4, 5, 6 – Aurignacian blades,
bilaterally retouched; 2 – bilaterally retouched blade; 3 – Aurignacian blade, laterally retouched.
Abb. 27. Româneti-Dumbrăvia I, Ausgrabungen Mogoanu, Schicht III (1, 3, 6), Schicht IV (2, 4) und Schicht 5 (5). Klingen: 1, 4, 5, 6 – Aurignacien-
Klingen, bilateral retuschiert; 2 – bilateral retuschierte Klinge; 3 – Aurignacien-Klinge.
Quartär 59 (2012)Earliest Aurignacian in Romania: the open air site of Româneşti-Dumbrăva
109
Fig. 27.
legend on
previous
page
Quartär 59 (2012) V. Sitlivy et al.
110
iron and manganese spots and concretions. In its lower
part, unearthed in an additional geological trench and
not visible in Fig. 4, GH 4 becomes more gleyic. From
the bleached horizon, albeluvic tonguing interfingers
into the brownish horizon. It is important to point out
the soil formation has reached deep below the recent
surface and overprinted the sediments of Mogoşanu
Aurignacian layers. Therefore, sedimentological changes
recorded by previous fieldwork seem less relevant for
the understanding of their formation process. Our
field studies and the analytics show a much more
complex differentiation and development of
sediments and soils. One of the major, yet preliminary
results is that the surface soil shows a polygenetic
development, overprinting pre-weathered sediment
belonging to former horizons not visible in the field.
The latter horizons may represent weak soil develop-
ments (of interstadial character?) whose relation to the
Aurignacian artifacts in GH 3 is - at the momentary
stage of investigation - not clear cut.
Artifact distribution
Aurignacian assemblages originate from two trenches
(Fig. 3: trenches 4 and A, 2009/2010; 7 m²). Both show
the same stratigraphical sequence described above,
with Geological Horizon 3 (GH3) being quantitatively
much more important. Whereas a total of 7505 lithics
comes from GH3, artifacts from the underlying GH4
do not differ from the GH3 industry, but are much less
abundant an therefore only described very briefly:
3 Dufours on micro-blades (Fig. 24: 5, 32), 2 burin
spalls, 5 blades, 5 bladelets, 7 microblades, 24 flakes,
1 chunk, 1 pebble and waste (89 chips, 4 small
fragments <25 mm). The uppermost GH2 and GH1
contain a quite abundant Epigravettian industry,
sometimes mixed with pottery fragments of different
periods.
Only Trench 4 allows for an admittedly restricted
hypothesis about horizontal and vertical distributions
of finds in the different GHs. While GH3 is – at least in
this part of the open air site - a high density zone, the
Layer
II
Layer
III
Layer
IV
Layer
V
GH
3
Font-Yves point, lateral, dorsal, on micro-blade 2
Font-Yves point, lateral, dorsal, on bladelet 1
Font-Yves point, bilateral, dorsal, on blade 1
Font-Yves point, bilateral, dorsal, on bladelet 1
Krems point, alternate, on blade 1
Krems point, lateral, ventral, on bladelet 1
Dufour alternate, on blade 1
Dufour alternate, on bladelet 4 1* 5
Dufour alternate, on micro-blade 3 38
Dufour lateral, ventral, on blade 1
Dufour lateral, ventral, on blade with distal inverse retouch 1
Dufour lateral, ventral, on bladelet/distal oblique truncation 1
Dufour lateral, ventral on bladelet 1 2 5
Dufour lateral, ventral on micro-blade 14
Pseudo-Dufour, bilateral, on bladelet 5
Pseudo-Dufour, bilateral, on micro-blade 1 1
Pseudo-Dufour, lateral, on bladelet 6 1 1
Pseudo-Dufour, lateral, on bladelet/proximal truncation 1
Pseudo-Dufour, lateral, dorsal, on micro-blade 1 2
Notch, lateral, dorsal, on bladelet 1 1
Denticulate, lateral/proximal, dorsal, on bladelet 1
Truncated bladelet, proximal, oblique, dorsal 1
Bladelet with lateral, dorsal, flat invasive retouch 1
Burin double on snap, on retouchet bladelet 1*
TOTAL: 1 19 7 2 80
* with projectile impact
Fig. 28. Româneşti-Dumbrăviţa I. Non-geometric Microliths: Mogoşanu’s (layers II-V) and 2009-2010
(GH3) excavations.
Abb. 28. Româneti-Dumbrăvia I. Häufigkeiten der fein retuschierten Lamellen. Ausgrabungen
Mogoanu (Schichten II-V), Ausgrabungen 2009-2010 (GH3).
Quartär 59 (2012)Earliest Aurignacian in Romania: the open air site of Româneşti-Dumbrăva
111
uppermost GH1 and GH2 evidence a low density zone
with some concentrations. The lowermost GH4
contains only isolated lithics. While the excavated
trench is too small (5 m²) to securely assess the
horizontal distribution, it can still be said that the lithic
material was dispersed equally across the entire
excavated area, leaving some small round spaces
without stone items. Burnt artifacts are common,
forming several clusters, e.g. one in square 85/220,
which delivered the highest amount of burnt material
(38 pieces, including samples for TL dating). Cores
were found mostly in the central part (10 out of 19),
while tools, including microliths, were dispersed more
equally (max. 58 and 24 pieces respectively in
southern square 85/220).
As for the vertical distribution, characteristic
meaningful artifacts were documented in the most
concentrated part of GH3. Microliths, including
Dufours, occurred in the middle of GH3 (47 out of 80),
disappearing progressively toward the top and the
bottom of the sequence. Cores (of all types) also
occurred in the most concentrated spits of GH3, as
well as in higher positions of this geological unit. No
carinated pieces were recorded in the lowermost or in
the uppermost part of the sequence. Twisted and
rectilinear bladelets display the same trend, occurring
together with the same frequency mostly in the
central part of GH3. Skewed (off-axis) bladelets
appear in equally small numbers at both ends of the
sequence, being more frequent in the middle part.
Thus these attributes do not show any significant
technological changes across this succession. The
Fig. 29. Româneşti-Dumbrăviţa I, Mogosanu’s excavations, layer III (2, 3, 4, 5, 8, 10, 11), layer IV
(1, 7) and layer V (9). Non-geometrical microliths: 1 – Dufour, on inversely retouched bladelet;
2, 3, 8, 10 – Dufour, on alternatively retouched bladelets; 11 – Dufour, on alternatively retouched
blade; 4, 5 – Dufour, on alternatively retouched micro-blades; 6- pseudo-Dufour, on obversely bilateral
retouched micro-blade; 9 – Font-Yves point, on laterally retouched crested bladelet; 7 – bladelet with
oblique proximal truncation.
Abb. 29. Româneti-Dumbrăvia I, Ausgrabungen Mogoanu, Schicht III (2, 3, 4, 5, 8, 10, 11) Schicht IV (1, 7)
und Schicht V (9). Fein retuschierte Lamellen: 1 – Dufour, ventral retuschiert an Lamelle; 2, 3, 8, 10 – Dufour,
an alternierend retuschierten Mikroklingen; 11 – Dufour, an alternierend retuschierter Klinge; 4, 5 – Dufour,
an alternierend retuschierter Mikroklinge; 6 – Pseudo-Dufour, an beidseitig dorsal retuschierter Mikro-
Klinge; 9 – Font-Yves Spitze, an Lamellen mit Kernkante, 7 – Lamelle mit proximaler Endretusche.
Quartär 59 (2012) V. Sitlivy et al.
112
presence of many chips alongside the large items, the
vertical and horizontal distribution of finds (Fig. 30), as
well as the number of refitted artifacts within GH3
(Figs. 31 & 32), confirm that there was little geological
or hydrological sorting of the material.
Trench A yielded similar assemblages, but in much
less quantity (periphery/low density area).
Absolute dating
OSL-dates
Luminescence dating was applied at the site Româneşti I
in order to establish a chronology of the sedimentation
processes. Luminescence measurements followed
single-aliquot regenerative dose protocols analyzing
the postIRIR
290°C
signal of potassium-rich feldspars
extracted from bulk sediments (for technical details
and discussion see Thiel et al. 2011; Buylaert et al.
2012; Fig. 33). Annual dose rates were calculated
from the radionuclide contents determined by high-
resolution gamma-ray spectrometry of bulk
sediments. Dose attenuation by soil moisture and
contribution of cosmic dose according to the sampling
depth was accounted for. The lowermost OSL sample
proved to be difficult to date by luminescence. The
postIRIR
290°C
signal was already close to saturation and
hence not suitable for dating (for more details see Kels
et al. subm.). The IRSL
50°C
age of the lowermost sample
of about 57.9 ± 5.4 ka provides only a minimum age
for deposition of the sediments, because it is not
corrected for any impact of anomalous fading of the
signal. Connected to this layer is the layer of the lower-
most “Quartzitic industry” (layer I, belonging to the
lower part of GH 4), which is separated from the
Aurignacian layers. Thus, the lowermost “Quartzitic
industry” might not be younger than ~58 ka, but could
be considerably older. Here, luminescence dating is
unable to provide more accurate results.
Although parts of the profile show fine discordances,
the Aurignacian assemblages from GH 3 most likely
date to MIS 3, with luminescence ages of 45.1 ± 4.9 ka
and 35.5 ± 3.9 ka from the middle part of this section
(Fig. 33). Some of the findings could be connected to
the buried weak soil in the middle part of the
reference profile, which was not obvious in the field,
but detected by multi-elemental analysis. Beneath this
palaeosol of interglacial character we detected three
layers of fossil root channels, which could belong to
former palaeosurfaces, too.
Toward the top of the profile from the lower part
of the sediments of GH 2 and connected to the
bleached horizon a luminescence age of 19.2 ± 2.3 ka
dates loess sedimentation into the Upper Pleniglacial
(MIS 2). This fits to the archaeological layer initially
attributed to the Gravettian, which can now be
conventionally reassigned to a post-LGM Epigravettian.
With these results, a first chronology of the sedimen-
tary development at the Banat foothills was possible,
offering first correlations to loess sections of the
Romanian and Serbian Banat (Fig. 1) and giving new
insights into the palaeoecology in different altitudes
of the region (for further details see Kels et al. subm.).
TL-dates
The excavations yielded 12 heated artifacts from GH2
(1 piece) and GH3 for thermoluminescence (TL)
dating. Due to early onset of dose saturation and
scarcity of sample material (small size of artifacts),
single-aliquot regenerative-dose (SAR) protocols
were employed for palaeodose estimation in combi-
nation with or instead of conventionally used
multiple-aliquot additive-dose (MAAD) techniques.
Analysis of glow curves and dose response behavior
during SAR measurements allowed us to distinguish
two types of samples which possibly also reflect
different mineralogical composition and thus different
raw material sources. Dose recovery tests and internal
checks for the quality of the dose estimates led to the
discard of data from one type of samples. The
reduced data set gives the following preliminary dates
of the last heating event.
The only sample from GH2 yielded a SAR age of
15.2 ± 1.3 ka and a MAAD age of 16.1 ± 1.5 ka, thus
falling into the Epigravettian. However, since this age
estimate is based on one sample only, care must be
Fig. 30. Româneşti-Dumbrăviţa I, 2009-2010 field campaigns,
GH3: horizontal and vertical distributions of artifacts with refitted
blocks (1-5).
Abb. 30. Româneti-Dumbrăvia I, Ausgrabungen 2009-2010,
Artefakte aus GH 3: Horizontale Verteilung, Profilprojektionen und
Zusammenpassungen (1-5).
Quartär 59 (2012)Earliest Aurignacian in Romania: the open air site of Româneşti-Dumbrăva
113
Fig. 31. Românti-Dumbrăva I, 2009-2010 field campaigns, GH3: 1ab, 1a, 1b – core and refitted debitage (block 2); 2a, 2b, 2ab – conjoining
of broken retouched piece on blade (block 4); 3a, 3b, 3ab – refitted micro-blade with notched piece on blade (block 3); 4ab, 4a, 4b –
conjoining of broken crested flake (block 5).
Abb. 31. Roneti-Dumbrăvia I, Ausgrabungen 2009-2010, GH 3: 1ab, 1a, 1b – Kern mit aufeinandergepassten Artefakten des Kernabbaus
(block 2); 2a, 2b, 2ab – Aneinanderpassung eines gebrochenen retuschierten Stücks an Klinge (block 4); 3a, 3b, 3ab – Aufeinanderpassung einer
Mikro-Klinge mit gekerbtem Sck auf eine Klinge (block 3); 4ab, 4a, 4b – Aneinanderpassung eines gebrochenen Abschlags mit Kernkante (block 5).
Quartär 59 (2012) V. Sitlivy et al.
114
taken when assessing its significance. Furthermore, the
TL results indicate that the last heating of artifacts
from GH3 most probably occurred between 40.0 ±
1.4 ka (SAR error-weighted mean) and 45.0 ± 1.5 ka
(MAAD error-weighted mean), while the younger age
clearly represents a minimum age, as discussed in
detail in Schmidt et al. (subm.). All single ages
contributing to these data are listed in Figure 33.
Analysis of the GH3 lithic assemblage (2009 - 2010)
Raw material
The new excavations show a raw material composition
similar to Mogoşanus assemblages in both raw
material types and frequency (i.e. opal dominance).
Additionally, an “exotic” black flint of very good
quality appeared in Trench 4. A heavily reduced core
on that flint, to which one micro-blade and one flake
was refitted (Fig. 31: 1ab, 1a, 1b), as well as some
isolated debitage products and Dufours were found.
General structure of the lithic assemblage
Chips are the dominant artifact category in the new
assemblage (59.2 %), followed by a much lower
frequency of flakes (15.1 %). Curiously, micro-blades
and bladelets occur in the same number (472 and 471)
and proportion (6.3 %), while blades are nearly twice
as less frequent (Fig. 5). Small debris (<25 mm) are
more representative than chunks and pebbles (5.2 %
contra 0.2 %). The same disproportion is true for tools
(2.3 %) and cores (0.3 %). Pre-forms/tested blocks are
nearly absent (2 pieces). Burin spalls are abundant (88
items or 1.2 %) in comparison with the corresponding
tools. The tool/core ratio is quite high (8.8: 1), as is the
blank to core ratio (125: 1), reflecting very high pro-
ductivity.
Flakes (45.2 % including tools-on-flakes) dominate
Fig. 32. Româneşti-Dumbrăva I, 2009-2010 field campaigns, GH3: 1 – retouched piece on blade;
2 – blade/bladelet core, bidirectional, narrow flaking surface; 3 – refitting of the same core and blade
(block 1).
Abb. 32. Româneti-Dumbvia I, Ausgrabungen 2009-2010, GH 3: 1 – retuschierte Klinge; 2 – bidirek-
tional Klingen-/Lamellenkern mit schmaler Abbaufläche; 3 – Aufeinanderpassung desselben Kerns und
der Klinge (block 1).
Quartär 59 (2012)Earliest Aurignacian in Romania: the open air site of Româneşti-Dumbrăva
115
the debitage structure over micro-blades and blade-
lets (22.7 % and 20.1 % correspondingly), while blades
are nearly twice as less frequent (Fig. 6). Nevertheless,
the laminar orientation of GH3 industry is evident,
since all laminar removals (54.8 %, including tools on
laminar blanks and burin spalls) dominate over flakes
and tool-flakes (45.2 %). Laminar structure shows a
clear priority for small blank production, while blades
appeared to be less important. Moreover, micro-
blades were the most desired blanks for modifications
– 57 tools on micro-blades.
Cores
A total of 19 cores were found, showing a very low
frequency in the whole assemblage. Identifiable cores
(13 items) have been subdivided into the following
categories: pre-cores (2), carinated (2), prismatic (4),
narrow-faced (5) and flake cores (2) (Fig. 7). Both
pre-cores are on quartz and sandstone pebbles with
several unidirectional scars. Carinated cores are
unidirectional, with bladelet/micro-blade scars. The
first core is on chunk with a wide sub-pyramidal
working surface (Fig. 34: 2). The second core-on-flake
has a working surface on hinged end and a
platform on the débordant side (Fig. 34: 4). Prismatic
cores comprise 1 blade unidirectional sub-cylindrical,
1 blade/let unidirectional core on flake, 1 bladelet
orthogonal-adjacent (with wide and narrow working
surfaces) core on flake (Fig. 34: 3) and 1 exhausted
core (Fig. 31: 1ab). Striking platforms are flat. A single
crest remnant was recorded. All of these cores show
advanced reduction or exhaustion. Narrow-faced
core-on-flakes (5 items) are represented by 3 uni-
directional (Fig. 34: 1), 1 bidirectional (Fig. 32: 2, 3) and
1 core with changed orientation (Fig. 34: 5). Three
cores are initial, and two cores are almost exhausted.
Two cases of scraper recycling were recorded.
Striking platforms are usually crudely prepared. The
angles (between platform and flaking surface) are
acute. Narrow unprepared working surfaces have a
Sample Layer Protocol Temperature interval [°C] Age [ka]
Rom17 GH2
MAAD, silex, 100-200 µm 270-380 16.1 ± 1.5
SAR, silex, 100-200 µm 300-350 15.2 ± 1.3
Rom35 GH3
MAAD, silex, 100-200 µm 230-250 47.7 ± 3.6
MAAD, silex, 100-200 µm 300-375 45.6 ± 1.9
SAR, silex, 100-200 µm 270-310 42.3 ± 3.2
SAR, silex, 100-200 µm 340-400 37.9 ± 3.7
Rom72 GH3
MAAD, silex, 100-200 µm 325-410 41.7 ± 3.0
SAR, silex, 100-200 µm 350-400 41.0 ± 3.6
Rom116 GH3 SAR, silex, 100-200 µm 350-410 37.6 ± 3.2
Rom239 GH3 SAR, silex, 100-200 µm 340-400 39.2 ± 3.2
Rom346 GH3 SAR, silex, 100-200 µm 340-390 41.8 ± 3.5
Rom1-3
lower part
of GH4
SAR-pIRIR290, K-rich
feldspars, 63-100 µm
close to
saturation
SAR-IR50, K-rich
feldspars, 100-200 µm
>>57.9 ± 5.4
Rom1-4a GH3
SAR-pIRIR290, K-rich
feldspars, 63-100 µm
45.1 ± 4.9
Rom1-4b GH3
SAR-pIRIR290, K-rich
feldspars, 63-100 µm
35.5 ± 3.9
Rom1-5
lower part
of GH2
SAR-pIRIR290, K-rich
feldspars, 63-100 µm
19.2 ± 2.3
Fig. 33. IRSL (feldspar) and preliminary TL (silex) ages. Individual TL dates were used for calculation of the
error-weighted mean ages; discarded data (e.g. statistical outliers) are not shown. For some samples,
two ages were generated by separate evaluation of single TL peaks passing the plateau test. The term
MAAD stands for the multiple-aliquot additive-dose protocol and SAR for the single-aliquot regenera-
tive-dose protocol. Dated materials and respective grain size ranges are also given; all ages are shown
with their 1 sigma uncertainty. For further details on measurements, see main text, Schmidt et al. (subm.)
and Kels et al. (subm.).
Fig. 33. IRSL- (Feldspat) und vorläufige TL (Silex)-Alter. Aus den gezeigten TL-Daten wurden die fehler-
gewichteten Mittelwerte berechnet; verworfene Daten (z.B. statistische Ausreer) sind nicht enthalten.
Für einige Proben wurden zwei Alter berechnet, indem TL-Peaks, die den Plateautest bestanden haben,
getrennt ausgewertet wurden. MAAD steht für das multiple-aliquot additive-dose-Protokoll und SAR für
das single-aliquot regenerative-dose-Protokoll. Datierte Materialien und zugehörige Korngrößenfrak-
tionen sind ebenfalls vermerkt; alle Alter sind mit 1 sigma Standardabweichung angegeben. Zu weiteren
Einzelheiten der Messungen siehe Haupttext, Schmidt et al. (eingereicht) und Kels et al. (eingereicht).
Quartär 59 (2012) V. Sitlivy et al.
116
triangular shape, sometimes extending to the ventral
or dorsal sides. The two flake cores include a semi-
discoidal and one on a lateral side-scraper (recycling)
with crossed scars, belonging to the Kombewa type.
Laminar debitage
Laminar products are abundant (1414), although
complete items are rare: blades (6.5 %), bladelets
(9.6 %), and micro-blades (8.1 %). The tools show a lesser
Fig. 34. Româneşti-Dumbrăva I, 2009-2010 field campaigns, GH3: 1 – blade core, unidirectional, narrow flaking surface, made on flake;
2 – bladelet carinated core, unidirectional, sub-pyramidal; 3 – bladelet core, orthogonal-adjacent, made on flake; 4 –bladelet carinated core,
unidirectional, made on flake; 5 – bladelet core, change orientation, narrow flaking surface, made on scraper.
Abb. 34. Româneti-Dumbrăvia I, Ausgrabungen 2009-2010, GH 3: 1 – unidirektionaler Klingenkern mit schmaler Abbaufläche an Abschlag;
2 – Gekielter unidirektionaler Lamellenkern, sub-pyramidal; 3 – Lamellenkern an Abschlag; 4 – Gekielter unidirektionaler Lamellenkern an
Abschlag; 5 – Lamellenkern mit schmaler Abbaufläche an Schaber, mit einem Wechsel von Schlag- und Abbaufläche.
Quartär 59 (2012)Earliest Aurignacian in Romania: the open air site of Româneşti-Dumbrăva
117
degree of fragmentation: complete tool-blades
(12.8 %), tool-bladelets (10.5 %) and tool-micro-
blades (8.8 %). The metrical data are as follows: (1)
blade max. size is 78.7, 37.6, 18.7 mm and average
of 17 complete blades is 44.1, 17.6, 7.0 mm; bladelet
max. size is 45.3, 11.9, 9.6 mm; average of 45 complete
bladelets is 24.8, 9.5, 3.1 mm; micro-blade max. size is
21.1, 6.9, 3.9 and average of 38 complete micro-blades
is 13.1, 4.9, 1.5 mm.
Flakes with different proportions of cortex
(29.5 %), dominate over blades (11.7 %); bladelets and
micro-blades with cortical surfaces are the least
Fig. 35. Româneşti-Dumbrăviţa I, 2009-2010 field campaigns, GH3: 1, 3 – angle burin, on snap, on blade; 2 – angle burin, on snap, on laterally
obversely retouched blade; 4 – endscraper, on bilaterally obversely retouched blade; 5 – dihedral angle burin, on distal convex truncation,
on blade; 6 – dihedral angle burin, 7 – transverse burin, on retouched blade; 8 – sidescraper, on core-tablet.
Abb. 35. Româneti-Dumbvia I, Ausgrabungen 2009-2010, GH 3: 1, 3 – Stichel an Bruch; 2 – Stichel an Bruch an Klinge mit dorsaler Kantenretu-
sche; 4 – kantenretuschierter Kratzer an Klinge; 5-6 – Mehrschlagstichel; 7 – Transversalstichel an retuschierter Klinge; 8 Schaber an Kernscheibe.
Quartär 59 (2012) V. Sitlivy et al.
118
frequent (about 3 %). Cortical flakes (>76 % of cortex)
account for 30 items (5.2 %), while primary blades
comprise three pieces and only a single bladelet.
Laminar products with unidirectional dorsal scars are
the most representative: blades (75.2 %), bladelets
(74.4 %) and micro-blades (81.4 %), following by
convergent (10.6 %, 13.8 % and 14.4 % respectively);
other patterns are rare.
Blade shapes are mostly rectangular (46.7 %), more
rarely trapezoidal (16.7 %), irregular (12.5 %) and
triangular (10.4 %). Bladelets with rectangular shape
dominate (51 %) over triangular (13 %) and trapezoidal
(12.7 %). The micro-blades are usually rectangular
(52.5 %), while triangular (20.7 %) and crescent
Fig. 36. Româneşti-Dumbrăva I, 2009-2010 field campaigns, GH3: 1 – pointed blade, bilateral, dorsal; 2 – alternatively retouched piece on
blade; 3, 7 – proximally truncated blades; 4, 6 – Aurignacian blades; 5 – notch on blade, lateral, dorsal; 8, 9, 10 – retouched pieces on blade;
11 – retouched blade, bilateral, dorsal.
Abb. 36. Româneti-Dumbrăvia I, Ausgrabungen 2009-2010, GH 3: 1 – Spitze an bilateral retuschierter Klinge; 2 – alternierend retuschierte
Klinge; 3, 7 – Klinge mit proximaler Endretusche; 4, 6 – Aurignacien-Klingen; 5 – gekerbtes Sck an Klinge; 8, 9, 10 – retuschierte Klingen;
11 – bilateral retuschierte Klinge.
Quartär 59 (2012)Earliest Aurignacian in Romania: the open air site of Româneşti-Dumbrăva
119
Fig. 37. Româneşti-Dumbrăva I, 2009-2010 field campaigns, GH3 and GH4 (5, 32). Non-geometric microliths: 1 – Font-Yves
point, on bilaterally retouched micro-blade; 2 – Font-Yves point, on laterally retouched micro-blade; 3 – Krems point, on
alternatively retouched micro-blade; 4-6, 8-27, 29-36, 41 – Dufour, on alternatively retouched micro-blades; 7, 28 – Dufour,
on inversely retouched micro-blade; 37 – Font-Yves point, on laterally retouched blade; 38 – Krems point, on inversely
retouched bladelet; 39 – Krems point, on alternatively retouched blade; 40, 48 – pseudo-Dufour, on bilaterally obversely
retouched bladelet; 42-45, 49, 50 – Dufour, on alternatively retouched bladelet; 46, 47, 51 – Dufour, on inversely retouched
bladelet; 52 – Dufour, on alternatively retouched blade.
Abb. 37. Româneti-Dumbvia I, Ausgrabungen 2009-2010, GH 3: fein retuschierte Lamellen: 1 – Font-Yves Spitze an
beidseitig retuschierter Mikroklinge; 2 – Font-Yves Spitze an beidseitig retuschierter Mikroklinge; 3 – Krems-Spitze an alter-
nierend retuschierter Mikroklinge; 4-6, 8.27, 29-36, 41 – Dufour, an alternierend retuschierter Mikroklinge; 7, 28 – Dufour, an
invers retuschierter Mikroklinge; 37 – Font-Yves Spitze an kantenretuschierter Klinge; 38 – Kremser Spitze an ventral retuschierter
Lamelle; 39 – Kremser Spitze, an alternierend retuschierter Klinge; 40, 48 – Pseudo-Dufour, an bilateral retuschierter Lamelle;
42-45, 49, 50 – Dufour, an alternierend retuschierter Lamelle; 46,47,51 – Dufour, an ventral retuschierter Lamelle; 52 – Dufour,
an alternierend retuschierter Klinge.
Quartär 59 (2012) V. Sitlivy et al.
120
(11.6 %) shapes are less frequent. The debitage
symmetry of laminar products shows some variation
within three main groups throughout size ranges. The
blades show a near-balanced frequency of on-axis/
off-axis products (56.2 %/43.8 %), while smaller
removals became more symmetrical, i.e. on-axis
bladelets (67.9 %) and especially on-axis micro-blades
(74.9 %). Blade/let and micro-blade profiles are
similarly twisted (38.1 %/37.5 % and 33.9 %), showing a
progressive decline of curved profiles through the
laminar categories (17 %/11.6 % and 9.4 %). On the
other hand, flat profiles increase with the diminishing
of laminar parameters (Fig. 14). It is worth mentioning
that the complete laminar products show different
profile frequencies, expressed particularly in higher
twisting rates (Fig. 14). However, data based on intact
products do not confirm the domination of one
particular profile type.
As for the distal ends, they become more
feathered from bigger to smaller laminar removals:
53.3 % (blades), 66.4 % (bladelets) and 75.6 % (micro-
blades). Consequently, hinged fractures slightly
decline from about 21 % (blade/lets) to 16 % (micro-
blades), as well as blunt distal parts. Overpassed
removals are rare, occurring mostly among blades (9)
and flakes (12). Interestingly, triangular cross-sections
of all laminar removals are more frequent than
trapezoidal cross-sections. Lateral steep cross-
sections progressively decline from blades (14.9 %) to
micro-blades (6.4 %); multiple sections are rare and
display a similar trend. This trend fits well with the
common presence of narrow-faced cores and burins.
Figure 15 shows the dominance of single blow
platforms of laminar products as well as the decline in
flat platforms in favor of lineal throughout the laminar
size groups. As for butt lipping (which is always high),
the highest frequency of lipped and semi-lipped
platforms was documented for bladelets (Fig. 16). The
analysis of laminar products also revealed (a) a decline
in lipped butts from blades to micro-blades and vice
versa, and (b) an increase in semi-lipped butts from
the smallest to the biggest laminar products. Diffused
bulbs dominate among all laminar removals
(max. 53.6 % for blades). However, this attribute
shows a gradual decrease from blades to micro-blades
(min. 40.3 %), while the frequency of developed bulbs
increases. Bulb absence is considerable for all
removals (c. 20 % - 25 %); the split/shattered pattern
occurs sporadically (Fig. 17). Obtuse interior flaking
angles dominate with similar frequency for all laminar
products, while the frequency of right angles doubled
for micro-blades in comparison with blades (Fig. 18).
As for overhang reduction, abrasion and faceting were
applied separately with similar intensity (>60 %) for
both blades and bladelets. On the other hand, these
techniques were used independently in nearly the
same lower proportions for micro-blades. Moreover,
mutual application of both techniques was often
attested, mostly for blades (48.2 %) and bladelets
(38.4 %) and much less so for micro-blades (16.6 %).
Thus, cores designated for obtaining bigger laminar
products were generally more often restored (as for
overhangs) in comparison with the “micro” items.
Core Maintenance Products include 60 pieces:
débordant/rejuvenation flakes (8), crested flakes (8),
crested blades (11), crested bladelets (5), crested
micro-blades (1), tablet-flakes (18), tablet-blades (1),
and core flank-flakes (8). Crested removals are not
very common, but represent all variations; lateral
crests are dominant.
Tools
Tool production in GH3 was based mainly on opal
(90.7 %), especially for the tools on flakes and on
blades (up to 96 - 97 %), while several micro-blade
tools were made of flint (8 pieces). Tools (n=169) were
made on different kinds of blanks, as follows: micro-
blades (37 %), blades (30.5 %), flakes (20.1 %) and
bladelets (12.3 %) (15 cases of unidentifiable tool
blank type). The toolkit is dominated by non-geometric
microliths (56.3 %). Burins and retouched pieces on
blades and on flakes are present in considerably lower
percentages. Other tools are rare (Fig. 19).
Endscrapers include only two simple types on
retouched blades (Figs. 20; 35: 4).
Burins (n=15) were made on blades (9) and flakes
(6). These are angle burins on snap (7), including 2 on
truncation, as well as dihedral (2), transverse (2 – one
on truncation), flat (3) and unidentifiable (1). In four
cases, the blanks were laterally (usually dorsally)
retouched (Figs. 24; 35: 1-3, 5-7). Carinated burins are
absent. In addition, 88 burin spalls were recorded.
These are on blades (3), bladelets (28) and micro-
blades (57); 19 have retouch/truncated remnants.
Retouched blades are rare (n=5), but also diverse,
including pointed (1), laterally retouched (2) and two
Aurignacian blade fragments (Fig. 36: 1, 4, 6, 11).
Retouched pieces on blades and on flakes with
light short discontinuous or partial semi-steep retouch
(while non-marginal) are common (16 and 12 items of
each). These pieces on blades usually have lateral
obverse retouch, rarely bilateral, alternate and alter-
nating (Fig. 36: 2, 8-10). Retouched pieces on flakes
often exhibit lateral obverse, but also inverse retouch.
Lateral notched pieces (n=4) are all on blades
(Fig. 36: 5), with scalar, fine or marginal semi-steep and
steep obverse retouch (1 case of alternating/double
notch).
Sidescrapers (n=3) were produced on flakes and a
flake-tablet (Fig. 35: 8) by continuous scalar semi-
steep retouch and fall into the lateral (straight and
concave) and transverse convex varieties.
Truncated pieces (n=2) are proximal, on retouched
blades, with straight oblique and concave truncations
(Fig. 36: 3, 7).
Pièces esquillées (n=3) are on flakes (2) and one on
blade and have unidirectional scalar flat bifacial
negatives (2); one piece is broken.
Quartär 59 (2012)Earliest Aurignacian in Romania: the open air site of Româneşti-Dumbrăva
121
Non-geometric microliths are abundant (n=80)
and represent more than half of the toolkit
(Figs. 19; 28; 37). These micro-tools were often made
on micro-blades (57), more rarely on bladelets (19),
but also on small blades (4); curious cases of 2 Dufours,
Font-Yves and Krems points on blades (Fig. 37: 37, 39,
52) with width between 12.6 - 13.7 mm. Fragmenta-
tion is very high: only 7 tools are complete (6 Dufours
and 1 pseudo-Dufour). The sizes of complete pieces
are as follows: max. 40.5, 11.2, 4.2 mm; min. 15.4, 6.5,
1.5 mm and average 25.2, 7.0, 2.1 mm. Average metrics
of all microliths is >13.8, 6.7, 1.7 mm. Non-geometric
microliths contain (a) Font-Yves points on blade/let
and micro-blades (4), Krems points on blade/lets (2),
Dufours on blade/let/micro-blade (64) and pseudo-
Dufours on bladelet/micro-blade (10). The dominant
type is represented by the Dufour with alternate
retouch on micro-blades (=38 pieces), followed by
pieces with lateral ventral retouch on micro-blades
(14); both the Dufour and the laterally retouched type
are less numerous when using bladelets as blanks. Half
of the pseudo-Dufours were attributed to the
bilateral dorsal type on bladelets, but some could
belong to the Font-Yves points (7 pieces lack the distal
parts). The selected blanks have unidirectional (63)
and convergent (12) dorsal patterns. “On-axis” lamellar
detachment is the usual trend (56), while “off-axis”
removals are much less common (15). Flat lateral profiles
dominate (38) over twisted (20) and curved (9); the
association of twisting and “off-axis” detachments is
rare (5). As for the butts, these are usually linear (19)
and flat (7), with a slight difference between types,
often semi-lipped/lipped (15/5), and also unlipped
(7). Bulbs are m