Download pdf - Preliminary reassessment of the Aurignacian in Banat (South-western Romania)

Preliminary reassessment of the Aurignacian in Banat (South-western Romania)

Valéry Sitlivy a,*, Victor Chabai b, Mircea Anghelinu c, Thorsten Uthmeier d, Holger Kels e, Loredana Nit"a c,Ion B"altean f, Andrei Veselsky b, Cristian Tutu g

a Institute of Prehistoric Archaeology, University of Cologne, Weyertal 125, 50923 Köln, GermanybCrimean Branch of the Institute of Archeology, National Ukrainian Academy of Sciences (NUAS), Yaltinskaya Street 2, 95007 Simferopol, UkrainecDepartment of History and Letters, Faculty of Humanities, Valahia University of Târgoviste, Str. Lt. Stancu Ion 35, 130115 Târgoviste, Romaniad Institute of Prehistoric Archaeology, University of Erlangen, Kochstraße 4/18, 91054 Erlangen, GermanyeDepartment of Geography, RWTH Aachen University, GermanyfHeritage Advice S.R.L., Str. Nicolae Titulescu 1, 280/B/63, 510096 Alba Iulia, Romaniag Str. Tineretului 4, 29/D/80, 130073 Târgoviste, Romania

a r t i c l e i n f o

Article history:

Available online xxx

a b s t r a c t

Despite its richness, the Romanian Paleolithic record has remained for decades relatively poorly known

to the broader scientific community. The situation swiftly changed after the find at Oase Cave, which

brought the Romanian paleoanthropological and archeological record into intensive focus, spurring

several international research projects devoted to the regional Early Upper Paleolithic. The present paper

provides the first summary of recent research undertaken in the neighboring area of Oase, the Romanian

Banat, particularly focused on the Aurignacian open air occupation at Cosava.

A detailed attribute analysis of both old and recently excavated Aurignacian collections from Cosava,

supplemented by a comparative overview of the allegedly similar industry at Românesti-Dumbr"avita, is

presented. An initial chronological assessment of the Banat Aurignacian by means of several OSL samples

from Românesti is also proposed.

The results of the lithic analysis, much like the new chronological estimations (in excess of 30 ka BP)

confirm the early assignment of the Banat assemblages to the Krems-type Aurignacian, but also dismiss

the unusually young chronology initially attributed to these settlements. While several features,

including the constant presence of Krems/Dufour tools, point strongly to an archaic stage of the Auri-

gnacian technocomplex (Protoaurignacian/Aurignacian 0), other elements (carinated forms, twisted

bladelets, and Aurignacian blades) recall more ‘classical’ features conventionally associated to the

Aurignacian I. No coherent chrono-stadial trend or functional requirements can explain these ‘mixed’

features in the Banat industries with the data at hand. Their presence nevertheless points to the internal

variability of these early Aurignacian occurrences.

! 2012 Elsevier Ltd and INQUA. All rights reserved.

1. Introduction

Despite a rich archaeological and paleoanthropological record,to which many generations of scholars have contributed and whichhas been periodically revised (e.g. P"aunescu, 1989, 1993, 2000,2001; Chirica et al., 1996; Cârciumaru, 1999; Chirica, 2001; Borziacet al., 2006; Cârciumaru et al., 2007a; Borziac and Chirica, 2008;Chirica and V"aleanu, 2008), the Romanian Paleolithic has remained

poorly known to the broader scientific community. One of the mainreasons for this is the long period of political isolation, which led todelay in adopting new approaches and models for prehistoricresearch. For the Paleolithic, this resulted in diminishing theoreticaland methodological compatibility between the Romanian data andthe cultural and chronological frameworks commonly used inWestern, Central or Eastern Europe (Anghelinu, 2003, 2006; seealso Horvath, 2009). Although the local Paleolithic record had longattracted scientific attention aimed at incorporating it into theWestern European framework, or correlating, classifying andcomparing lithic assemblages, particularly those belonging to theLate Middle Paleolithic and onset of the Upper Paleolithic (e.g.Gabori, 1976; Allsworth-Jones, 1986; Yevtushenko, 1998; Djindjianet al., 1999; Djindjian, 2000; Chabai et al., 2004; Sitlivy and Zieba,2006), Romania is still perceived as a peripheral zone and almost

* Corresponding author.

E-mail addresses: [email protected] (V. Sitlivy), [email protected]

(V. Chabai), [email protected] (M. Anghelinu), Thorsten.Uthmeier@

ufg.phil.uni-erlangen.de (T. Uthmeier), [email protected] (H. Kels),

[email protected] (L. Nit"a), [email protected]

(I. B"altean), [email protected] (A. Veselsky), [email protected] (C. Tutu).

Contents lists available at SciVerse ScienceDirect

Quaternary International

journal homepage: www.elsevier .com/locate/quaint

1040-6182/$ e see front matter ! 2012 Elsevier Ltd and INQUA. All rights reserved.

http://dx.doi.org/10.1016/j.quaint.2012.07.024

Quaternary International xxx (2012) 1e20

Please cite this article in press as: Sitlivy, V., et al., Preliminary reassessment of the Aurignacian in Banat (South-western Romania), QuaternaryInternational (2012), http://dx.doi.org/10.1016/j.quaint.2012.07.024

systematically excluded from discussing the Middle to UpperPaleolithic transition. The narrow focus on typological descriptionof the lithic assemblages by Romanian authors and the poor chro-nology available has kept much of the regional information out ofthe mainstream debate, while most synthetic interpretations andcritical overviews have come from Western researchers carryingout projects in this country (e.g. Honea, 1984, 1986, 1987; Mertens,1996; Otte et al., 2007; Riel-Salvatore et al., 2008; Horvath, 2009;Noiret, 2009).

In recent decades, several international field projects carried outin various parts of Romania have yielded significant new resultsand fresh perspectives: Prut Valley and Moldavia (e.g. Otte andChirica, 1993; Otte et al., 1996a, 2007; Noiret, 2004, 2009;Tuffreau et al., 2009), South Carpathians (Otte et al., 1996b;Cârciumaru et al., 2000, 2002; Patou-Mathis, 2000e2001; Moncelet al. 2002), Bistrita Valley (Cârciumaru et al. 2006, 2007b;Steguweit et al., 2009), Danube Valley (Alexandrescu et al., 2004)and North-western Banat (Tuffreau et al., 2006, 2007). Systematicdating programs meant to complete the initial dating campaign inthe 1980s (Honea, 1984, 1986, 1987) have also been undertaken(B"alescu et al., 2003), although the amount of available dataremains insufficient.

The situation further improved after the discovery of the oldestAnatomically Modern Human (hereafter AMH) fossil remains inEurope at Oase Cave (Southern Banat) in 2002, followed by directdating, detailed paleoanthropological analysis and publication(Moldovan et al., 2003; Trinkaus et al., 2003, 2005, 2006, 2009). Thelack of an associated archaeological context, in combination withthe traditional view of a long persistence of the Middle Paleolithicin the area and the very late Upper Paleolithic chronology docu-mented across Romania (Cârciumaru, 1999; P"aunescu, 2000, 2001)made the story all the more exciting. The initial Oase discovery ledto systematic research at Oase Cave (Lazarovici et al., 2005; B"alteanet al., 2008), followed by different analyses (Richards et al., 2008)and have spurred several scientific projects: a paleoanthropologicalproject re-examining ‘forgotten’ human fossils from Romaniancaves and their direct dating (Olariu et al., 2002, 2004, 2005;Soficaru et al., 2006, 2007; Harvati et al., 2007; Alexandrescu et al.,2010) and new research in Banat (B"altean et al., 2008), thusbringing the Romanian record into the broader Afro-Eurasiandiscussions on the origins and dispersal of AMH (Trinkaus et al.,2006; Zilhão et al., 2007). Regardless of the disputable author-ships of European ‘transitional’ industries (e.g. see Peresani, 2008;Hoffecker, 2009; Benazzi et al., 2011) and given the time frame ofOasemoderns (ca. 35 ka 14C BP), the earliest cultural entity towhichone would likely assign the initial dispersal of AMH into Europe isthe Protoaurignacian; for at least some authors, the latter extendsthroughout the continent (including the Danube valley andRomanian Banat) to the Near East, matching the Early Ahmarian(Tsanova, 2006; Zilhão, 2007).

As a dispersal following the ‘Danube corridor’ is still one of themain scenarios (Conard and Bolus, 2003; Mellars, 2006; Zilhão,2007) for the appearance of AMH populations and their likelycultural correlate (i.e. Aurignacian sensu lato) in Europe, a properreevaluation of the archaeological record in the proximity of theOase Cave was required. The present paper discusses the prelimi-nary results of the recent research which have taken place in theBanat area, particularly focused on the Aurignacian sites of Cosava Iand Românesti-Dumbr"avita I.

2. The study area

Due to its geographic position and diverse topographies, severaldistinct zones have been identified for Paleolithic cultures inRomania (Djindjian, 2000; see also Anghelinu and Nit"a, submitted

for publication). These were linked to the broad cultural areas ofEurope: Eastern Romania to the North Pontic zone, the westernareas to Central Europe and the southern part to the circum-Mediterranean zone. Banat is a micro-region bordered to thesouth and east by high mountains representing a kind of refugeinfluenced by both Mediterranean and continental climates,reflected in the fauna and flora (Mogosanu, 1972). This region islargely accessible from Central Europe to the west, where theMures, Bistra and Bega river valleys provided natural connectionswith the large Carpathian Basin, while for Oltenia, the likelyconnection was by following the Danube and the northesouthoriented Timis-Cerna corridor, where the Tincova settlement isactually located. The Mures valley bordering the historical Banat tothe north also provides a natural communication passage to CentralTransylvania.

Gradually shifting from a mountain landscape to the east andsouth to open plain to thewest, the Banat geomorphological settingis dominated by hilly piedmonts, plains and local terrace steps.Quaternary deposits are to be found as thin loess-like sedimentcover (see also Kels et al., in preparation). Except for the occur-rences in the Carpathian caves, all Paleolithic settlements yetknown belong to low energy/highly eroded open air depositionalcontexts.

3. The Banat Paleolithic record e a brief outline

The Paleolithic record known so far in Banat is not particularlyrich (Fig. 1). Apart from the Aurignacian, there are few traces ofother industries: ‘Quartzitic Paleolithic’, Mousterian andGravettian/Epigravettian. Nearly all of the Paleolithic sites knownwere discovered during the 1950se1960s, mostly due to theenergy of Ion Stratan, excavated and published by Mogosanu(1972, 1978), and recently inventoried by P"aunescu (2001) andB"altean (2011a, b).

3.1. The Middle Paleolithic

The Middle Paleolithic is represented by small, mostly quartz/quartzite-based assemblages (which traditionally, after Mogo-sanu, were assigned to the so-called ‘Quartzitic Mousterian’ of theSouthern Carpathians, i.e. a variant of the Eastern Charentian).These assemblages have been recovered from uncertain strati-graphic position in two caves and two open-air sites.

Layer I at Hotilor Cave yielded 83 (Mogosanu, 1978) or 130(P"aunescu, 2001; see also B"altean, 2011a, b) debitage products(flakes, often cortical, including naturally backed flakes and 29formless or exhausted cores) and 25 tools (various simple scrapersand notches). A preliminary evaluation based on drawings ofquartzite artifacts suggests non-Levallois Mousterian, with side-scrapers and retouched flakes, lacking bifaces. At Livaditei Cave,only 18 pieces were recovered, including simple, double andconvergent sidescrapers, one modified by semi-Quina retouch(P"aunescu, 2001), as well as a Neanderthal phalanx.

At the open-air site of Românesti-Dumbr"avita I, the lowermostlevel I (beneath the Aurignacian sequence) was classified as LateMousterian (P"aunescu, 2001). This assemblage includes 48quartzite artifacts: three Mousterian unretouched points (?), flakesused as scrapers, simple flakes, two ‘quasi-prismatic’ cores, twoatypical endscrapers (Mogosanu, 1972), sidescrapers (simple,transverse, canted), choppers and naturally backed flakes(P"aunescu, 2001). If the Late Middle Paleolithic assignment iscorrect, a potential ancestor for this industry may come from theopen-air site of Z"abrani in the Arad region, recently re-excavatedand attributed to the Early Weichselian. Curiously, half of the arti-facts from the threeMousterian layers herewere made on quartzite

V. Sitlivy et al. / Quaternary International xxx (2012) 1e202


and show good control of pebble reduction, which was done usingseveral different techniques, depending on the stratigraphic layer:flake unidirectional, bidirectional, centripetal, rare discoidal andlaminar production are all indicated (Tuffreau et al., 2007). Thesequartzite flakes were more often modified into tools than blanks onother raw materials and do not show any ‘archaic’ features. Just7 km to the north, the multi-stratified site of Cladova yieldedbifacial tools made on quartzite, attributed to the Upper Acheulianof the Last Interglacial (Boroneant, 1991) or to the Mousterian ofAcheulian Tradition (P"aunescu, 2001).

In contrast, at the extreme south of Banat at the open-air site ofGornea-Dealu C"aunitei, a short sequence affected by solifluctionyielded an indubitable Levallois industry with correspondingpoints and flakes. The sample of 147 artifacts includes severaldifferent kinds of cores: Levallois, discoidal, polyhedral, globularand ‘quasi-prismatic’/‘quasi-pyramidal’ (Mogosanu, 1978). Theindustry here, labeled as Typical Mousterianwith Levallois debitage(P"aunescu, 2001), represents, to date, the only example of Levalloispoint/flake production in this region.

3.2. The Aurignacian

The Banat Paleolithic record is better known for three undatedAurignacian open-air sites: Tincova, Românesti-Dumbr"avita andCosava (Hahn,1970,1977; Mogosanu,1972,1978,1983), while otherUpper Paleolithic evidence remains quite limited. For Mogosanu,these open-air sites, with one or several occupations, representedshort-term workshops or relais de chasse, preserving exclusivelylithic artifacts. The first Aurignacian stage of the Banat (Cosava elower layer, the single layer at Tincova) was attributed to the last,Würm IIeIII, interstadial. The pollen-based geochronological esti-mations at Românesti suggested a much later, Tardiglacial

chronology (Mogosanu, 1978). The general typological structure ofthe toolkits (carinated, nosed endscrapers, dihedral burins and ontruncation, Aurignacian blades, as well as Dufour bladelets andFont-Yves points) suggested a direct connection to the CentralEuropean, especially Krems-type, Aurignacian (sensu Demidenko,2000e2001). In a typical culture-historical vein, Mogosanuviewed the Banat Aurignacian as a late echo of the Krems Auri-gnacian groups, retreating into the Banat refugia in the face ofexpanding Central European Gravettian populations. Its latesurvival and ‘degeneration’ were used to explain the less and lesscharacteristic toolkits from the upper layers at Cosava and Româ-nesti (Mogosanu, 1978, 1983).

Recent debates on the definition of the Aurignacian that alsoinvoked the Banat settlements have mainly involved Tincova(Teyssandier, 2006, 2007a, b, 2008; Zilhão, 2006; Zilhão et al.,2007; Teyssandier et al., 2010; Tsanova et al., in press). Thissingle-layered workshop (where mostly local opal/‘Banat flint’ wasexploited) contained 2494 artifacts: 2015 waste products (frag-ments, flakes, and cores), 369 blades and bladelets, and 110 tools(Mogosanu, 1978). The toolkit is dominated by endscrapers (31)with carinated, nosed, core-like forms, rabots (all in all 12 pieces)and many Dufour bladelets (22). Font-Yves points are also present(3). Burins are rare (8) and are mostly dihedral (5). Little is knownabout cores: 2 prismatic, 1 pyramidal, 7 globular, 55 core fragmentsand formless specimens are reported (P"aunescu, 2001).

At Românesti-Dumbr"avita I, the Aurignacian (levels II, III, IV andV) is sandwiched between ‘Quartzitic Mousterian’ and Gravettianlevels (Mogosanu, 1972, 1978). The richest assemblage (>5000pieces, including 114 tools) was recovered from level III. End-scrapers number 51 pieces (including many Aurignacian types) andpredominate over burins (26, including 18 dihedral types). EightDufour bladelets and some retouched blades (five of which are

Fig. 1. Geographic map with position of the main sites mentioned in the text.

V. Sitlivy et al. / Quaternary International xxx (2012) 1e20 3


Aurignacian) were also collected from this layer. Among the 77cores mentioned by P"aunescu (2001), there are 14 prismatic single/double platform cores, 8 pyramidal, 25 globular and 30 formlessand fragments. Level IV differs from the previous one due to thepresence of truncated blades/flakes and a decrease in the frequencyof endscrapers, with a corresponding increase in burins. Aurigna-cian artifacts become less common and Dufour bladelets are absent.Level V (consisting of extensive but clustered workshops) containsan assemblage rich in debitage products and few tools (39). Thecomposition of the retouched tool component contrasts with levelsIII and IV. Burins are more common than endscrapers, and Auri-gnacian pieces are comparatively infrequent. It should be notedthat Românesti-Dumbr"avita II yielded small workshop clusterswhich appeared stratigraphically in level V of the main site. Oneworkshop was specialized in Dufour production, associated withalternately retouched bladelets, retouched flakes, two endscrapersand several Krems points (Mogosanu,1978). These Dufour bladeletsare shorter (<3 cm) than those from Tincova (Chirica, 1996).Knappers at the second workshop (in the same stratigraphic posi-tion) also exploited quartz, producing Mousterian-like artifacts. Asimilar case was noted in a test pit outside the main concentrationat Tincova, where quartzite artifacts (sidescrapers and points) werefound in the uppermost loess. Together with the Românesti-Dumbr"avita II workshop, this situation led Mogosanu to theconclusion of a late, Tardiglacial survival of a Middle Paleolithictechnological tradition.

Cosava yielded three Upper Paleolithic industries, of which atleast the two lowermost levels contain Aurignacian tools withoutmixture. According to Mogosanu (1978), the most representativelevel I contains 110e116 tools and contrary toTincova, this toolkit isdominated by carinated and nosed endscrapers, as well as nucléi-

formes and rabots (simple endscrapers are less common), associatedwith abundant retouched blades, including 10 Aurignacian blades(e.g. strangled, notched and denticulated with continuous retouchon one or both sides), rare, exclusively dihedral, burins, as well assingle examples of Dufour bladelets and Font-Yves points. This levelwas usually compared with Tincova and Românesti-Dumbr"avita I,level III. Middle level II comprises 56 tools with a similar compo-sition: a high frequency of endscrapers, particularly carinated,a limited number of dihedral burins and a single Dufour bladelet.Uppermost level III (24 tools) contains Aurignacian types (5 Dufour,2 carinated and one Font-Yves point) as well as some elements(thumbnail endscrapers, blades on exotic obsidian, which normally

appear much later in the Epipaleolithic in this region (Mogosanu,1978; P"aunescu, 2001)). This assemblage together with theuppermost levels of Românesti-Dumbr"avita I are thought topresent a local development of Aurignacian (Mogosanu,1983) or itsfinal stage (P"aunescu, 2001).

In the light of a much needed reevaluation of the Aurignacian inRomania (Anghelinu and Nit"a, submitted for publication;Anghelinu et al., in press), the importance of new, detailed tech-nological and typological studies of available collections appearedevident. Both Românesti-Dumbr"avita and Cosava were selected in2009 for a thorough stratigraphical, chronological and archaeo-logical reevaluation, including test pits, TL, OSL, pollen, sedimen-tological and tephra sampling, correlated to the study of both oldand new archaeological collections. The present paper is focusedprincipally on information from Cosava I, along with a first re-appraisal of the Românesti-Dumbr"avita I site, as the study of bothold and new (2009e2010) assemblages from the latter site is still inprogress.

4. Cosava: current results

4.1. Geographic setting

The Cosava I site (45!51011.920 N, 22!19032.710 E) is located 4 kmnorth of Românesti-Dumbr"avita, which from it is separated by thelarge Bega valley (Fig. 2). The settlement is situated on a plateauspur (up to 282 m a.s.l. and over 90 m above the Bega river level)and slopes of two hills (‘Cuca Mare’ and ‘Cuca Mic"a’, correspond-ingly Cosava I and II), connected by a slightly depressed riverbed-like area (Fig. 3). In contrast to Tincova and Românesti-Dumb-r"avita, both located at the periphery of Poiana Rusc"a Mountains,this settlement is situated on the hill marking a meridian limit ofthe vast Lipova Plateau (Mogosanu, 1967; Stratan, 1970; Kels et al.,in preparation), on the right bank of the upper Bega river (tributaryto the Tisa river) (see also Kels et al., in preparation).

4.2. Previous research

Cosava was first excavated in 1961e1964 (Stratan, 1965) andlater from 1967 to 1969 (Mogosanu and Stratan, 1966, 1969;Stratan, 1970; Mogosanu, 1978). An area of 226 m2 was opened toa maximum depth of around 1.5 m to recover archaeologicalremains. A geological stratigraphic section up to 3 m deep was

Fig. 2. View of Cosava and Românesti-Dumbr"avita (I, II) sites during field campaign in 2009. View to north.



also opened. Lack of field documentation (only one profile isavailable e Mogosanu, 1978, Fig. 41) makes any correlationbetween Mogosanu’s and our excavated areas impossible.According to the original excavators, Cosava represents a hori-zontally extensive site(s) partly destroyed by a sand quarry,though strongly clustered and yielding a rather small lithicassemblage, in which ‘typical’ (tools) and ‘atypical’ pieces (i.e.debitage) are nearly equal in abundance (Mogosanu, 1983).Previous excavations led to the discovery of three Aurignacianoccupations (Mogosanu, 1978):

- level I lies between 85 and 75 cm depth from the modernsurface in the lower part of geological layer 3 (brown-reddishcompact sandy clay, with rich prismatic structure, iron oxideconcretions and rolled pebbles on its lower side e where Tin-cova and levels II, III and IV at Românesti-Dumbr"avita I occur)(Fig. 4);

- level II lies between 60 and 45 cm approximately in the middleof the same layer 3;

- level III (slightly contaminated by Epipaleolithic) lies between35 and 25 cm in the lower part of layer 2 (awind-blown depositof yellow-whitish color).

In SeptembereOctober 2009, research was continued by aninterdisciplinary team at Cosava I (‘Cuca Mare’) in order to verifyPaleolithic remains and collect samples for a dating program (OSL,TL) and geological analysis (5 test pits and several geo-profiles indifferent areas of the site).

4.3. Stratigraphy and archaeological sequence

The uncovered stratigraphic and archaeological sequence in allcomplete test pits correspond in general terms to Mogosanu’sinitial description (see above). Different sedimentary units andoverprinted fossil soils were distinguished (see Kels et al., inpreparation). The loess-like sediments at Cosava show differentsedimentary units which, analogous to the palaeosoil and soildevelopment, reflect climatic changes from the Last Interglacial tothe Holocene. The typical surface soil in the study area is a type ofa Stagnic Albeluvisol, which has a complex genesis, overprintingformer interglacial and interstadial soils of MIS 5 and MIS 3. AtCosava, this soil is well developed as shown by the intense claycoatings visible in the lower parts of all profiles.

Three separate levels with lithic artifacts were distinguished, atan average depth ranging between 15 and 80 cm. The lowermostfinds were documented in geological horizon 4 (GH4), whichmightcorrespond to Mogosanu’s archaeological level I (sandy clay withiron concentrations), an intermediate level with artifacts occurredin GH3 (¼level II in the same geological horizon without ironconcentrations) and uppermost lithic pieces appeared in GH1-2

Fig. 3. View of Cosava site during excavations in 2009. View to north.

Fig. 4. Cosava, Românesti-Dumbr"avita and Tincova stratigraphic sections of the

Aurignacian levels excavated by Mogosanu (1972): modern soil; yellow-whitish fine

dusty layer; reddish clay with prismatic structure; red-yellowish clayish silt; fine sand;

iron-manganese concretions with gravel; terrace alluvium; alluvial fan (erosion cone).



(¼level III in dusty loess-like sediments) (Fig. 5). The first obtainedluminescence dates (61$7 ka and 56$ 6 ka below lowermost levelI and very recent at the top: 4.49 $ 0.52 ka) provide the general,albeit wide, chronological frame for the archaeological assem-blages. Combined with other analyses these show that the genesisof the soil system here and the related post-depositional processeswere much more complex than expected. Nevertheless, theconstant presence of small knapping debris in all recently exca-vated collections point to a rather limited post-depositional impact,at least where vertical sorting of material was concerned.

4.4. Lithic assemblages

We present the first results of attribute analysis (cores, mostlylaminar debitage and tools) of the old Cosava assemblages availablewhich is designed to provide information on lithic technology,typology and raw material exploitation (for attributes see Tixier,1963; Hours, 1974; Marks, 1976; Demars and Laurent, 1989;Inizan et al., 1995; Chabai and Demidenko, 1998; Soriano et al.,2007; Zieba et al., 2008; Le Brun-Ricalens et al., 2009; Sitlivyet al., 2009). New excavations, while small-scale and providingless abundant recovered remains, complement the old data andparticularly the microlithic record, generally lost during Mogo-sanu’s excavations.

4.4.1. Raw materialsThe most common raw material in this region is opal (‘Banat

flint’), variable in quality (from very good to poor), homogeneity(with/without inclusions) and colors (light, dark brown, reddishand combination). Large and medium-sized oblong well-rounded

cobbles with alluvial cortex as well as nodules with fresh cortexof opal were used. Most of these (often poor quality) one can findnow in the river gravels and on old eroded terraces. Nevertheless,the exact sources of good quality opal and some other ‘exotic’raw materials are still unknown. The latter (radiolarite, jasper,flint, quartz, quartzite, chalcedony and sandstone) are present inmuch smaller quantities. However, the range of raw materialsused in Cosava is higher than in other sites. Regarding the rawmaterial, there are no significant diachronic changes within theCosava sequence: opal remains the main raw material categoryused for knapping. However, opal cores show significant declinefrom the base to the top (79.1%e65.2%e42.8%) in favor ofdifferent rocks, especially radiolarite (although the pre-cores areof opal and sometimes of quartz). The tools display a stablechoice of opal across all layers, but the frequencies vary signifi-cantly from tool-flakes (73.8%e63.4%e75%) to tool-blades/bladelets (56.2%e48.8%e58%). Unretouched blades in lower-most level (54.4%) display the same trend (56.2% of opal wasmodified into tools). This is not the case in the two uppermostlevels: the frequency of opal laminar debitage increases (from59.7% to 69.5%), while opal tool-blades/lets decline in level II(48.8%) or increase in level III (58%).

As for the state of preservation, artifacts recovered by excava-tions, as well as recently collected ones from eroded surfaces (notpresented in this analysis) are usually fresh, non-rolled, and non-patinated with rare gloss and some mechanical damage, whichmost likely occurred during excavation and storage. Heated/burntmaterial is not rare.

4.4.2. The general structure of the lithic assemblagesAll three of Mogosanu’s assemblages are dominated by large

debitage products (flakes e from 44.3% to 51.1% and blades efrom 17.4% to 24.8%). Relative abundances remain nearlyunchangeable throughout the entire sequence (Table 1). Theproportion of retouched tools is rather high, especially in

Fig. 5. Cosava I, Trench 2, Profile S (2009). Aurignacian levels and corresponding

Geological Horizons: I/GH4, II/GH3 and III/GH2-1.

Table 1

Cosava. Artifacts.

Mogosanu’s excavations Level I Level II Level III

n % %ess n % %ess n % %ess

Pre-cores 5 0.7 0.7 3 0.5 0.6

Cores 43 5.8 6.3 23 4.2 4.4 7 2.3 2.3

Flakes 331 44.3 48.3 269 49.1 51.1 158 51.0 53.0

Blades 130 17.4 19.0 120 21.9 22.8 77 24.8 25.8

Bladelets 26 3.5 3.8 20 3.6 3.8 16 5.2 5.4

Micro-blades 2 0.3 0.3

Tools 145 19.4 21.2 91 16.6 17.3 39 12.6 13.1

Tools/cores 3 0.4 0.4

Burin spalls 1 0.3 0.3

Chips 2 0.3 2 0.4 2 0.6

Small fragments 22 2.9 7 1.3 3 1.0

Chunks 38 5.1 13 2.4 7 2.3

Total 747 100 100 548 100 100 310 100 100

2009 excavations GH4/Level I GH3/Level II GH1-2/Level III


Cores 1 1.1 3.8 1 0.8 2.6 1 0.5 1.3

Flakes 17 18.7 65.38 8 6.3 20.5 40 20.5 50.6

Blades 2 2.2 7.7 6 4.7 15.4 12 6.2 15.2

Bladelets 4 4.4 15.4 10 7.9 25.6 12 6.2 15.2

Micro-blades 1 1.1 3.8 8 6.3 20.5 2 1.0 2.5

Tools 1 1.1 3.8 5 3.9 12.8 9 4.6 11.4

Pre-forms 1 0.8 2.6 3 1.5 3.8

Chips 53 58.2 72 56.7 81 41.5

Small fragments 12 13.2 14 11.0 32 16.4

Chunks 2 1.6 3 1.5

Total 91 100 100 127 100 100 195 100 100



lowermost level I e 19.4% (or 21.2% in essential counts, i.e.without chips and chunks), where they are even more commonthan blades. Tools decrease slightly in frequency to 13.1%(essential counts) at the end of the sequence. Cores and chunksare quite numerous in the lowermost layer and graduallydecrease to the top, while bladelets, always rare, become slightlymore abundant toward the top of the sequence. The tool/coreratio is moderate in all assemblages: from 3:1 up to 5.5:1. Theblank to core ratio progressively increases from the bottom to thetop and shows rather low productivity in level I (10.1:1) anda higher ratio in level III (35.8:1). This observation is influencedby the lack of small laminar products in the old collections, quiteapparent in comparison with the new assemblages. Chips(<15 mm), small fragments (<25 mm) and micro-blades(W < 7 mm) are nearly absent in all assemblages. Thus, a fewdifferences in artifact composition were documented in the oldinventories.

Recently recovered assemblages from three geological horizons(GH4 correlated with level I, GH3/level II and GH1-2/level III) differdramatically from the collections from older excavations in thedominance of small-sized artifacts (especially chips: from 58.2% to41.5%) over large pieces, even taking in consideration biases due tounequal numbers of artifacts. The presence of many chips alongwith large items confirms that there was little geological orhydrological sorting of the material. While the general structure ofnew assemblages appears different from Mogosanu’s collections(due to different sieving practice), changes in artifact categoriesthroughout the sequence are almost negligible.

Cores are frequent in the old assemblages (especially in the twolowermost levels) and dominated by carinated and prismatic types(Table 2). Cores for flakes (discoidal, polyhedral and orthogonal) arerare (5 items) and all but one occur in lowermost level I. Pre-coresare also rare and were documented in the two lowermost levels (5and 3, correspondingly). These are represented by (a) narrow-facedbladelet pre-cores on flakes, (b) blade pre-cores on chunk/nodule/pebbles and (c) failed (hinge fractures) on flakes and chunks. Thereduction of cores (whether flat or voluminous) was usually aimedat blade, bladelet and micro-blade production. Carinated coresdisplay a certain variability in regard to final products/reductionstages (blade/micro-blade, blade/bladelet, flake/bladelet), debitagedirection, platform(s)/working surface(s), quantity and position(uni-/bidirectional, orthogonal and their combination: adjacent,alternate) and extension of flaking surfaces (narrow, partly turned/semi-circular and turned/circular) (Fig. 6). They are often made onmassive flakes. ‘Regular’ laminar cores are frequently narrow-faced(often through reduction of flake slices/edges). These ‘burin-like’cores occur in all levels and except for two cases are unidirectional(Fig. 7: 3e5). There are also some partly turned (semi-tournant)cores on nodules and chunks (Fig. 7: 1) bearing some remnants of

Table 2

Cosava. Cores (including cores on tools).


n % n % n %

Pre-cores 5 3

Cores 46 100 23 100 7 100

Carinated bladelet/Micro-blade: 13 30.2 8 36.4 4 57.1

Unidirectional 8 6 3

Bidirectional-adjacent 1

Bidirectional 2 1

Orthogonal-adjacent 2 1

Orthogonal-alternate 1

Carinated blade/Bladelet: 2 4.7 1 4.5

Unidirectional 1 1


Carinated flake/Bladelet: 4 9.3 1 4.5

Unidirectional 1

Bidirectional-alternate 1


Bidirectional 1

Perpendicular 1

Blade: 2 4.7 1 4.5

Unidirectional 2


Blade/Bladelet: 6 14.0 3 13.6

Unidirectional 3

Unidirectional, narrow flaking surface 1 2

Bidirectional, narrow flaking surface 1

Bidirectional 1 1

Bladelet/Micro-blade: 7 16.3 7 31.8 3 42.9

Unidirectional 2

Unidirectional, narrow flaking surface 2 4 1

Orthogonal-adjacent, narrow flaking surface 1

Bidirectional 1

Bidirectional, narrow flaking surface 1

Unidentifiable 4 1

Blade/Bladelet core on tool: 3 7.0

Unidirectional, narrow flaking surface,

on thick endscraper

3

Flake/Bladelet, semi-polyhedral 1 2.3

Flake: 5 11.6 1 4.5

Semi-polyhedral 1

Polyhedral/discoidal 1

Discoidal 1

Bidirectional-transverse 1

Orthogonal, trifacial 1

Unidentifiable 1

Unidentifiable 3 1

Core Main Groups n %ess n %ess n %ess

Carinated 19 44.2 10 45.5 4 57.1

Prismatic & Narrow/Burin-like laminar 19 44.2 11 50.0 3 42.9

Flake cores 5 11.6 1 4.5

Total (identifiable) 43 100 22 100 7 100

Fig. 6. Cosava I. Carinated cores: 1 e blade/bladelet unidirectional with bilateral

narrowing on massive flake (level I); 2 e blade/bladelet bidirectional-alternate (level

I); 3 e bladelet orthogonal-adjacent (level III); 4 e bladelet unidirectional (GH3).



crests. Three cores recovered in 2009 include a carinated bladeletcore e GH3, (Fig. 6: 4), a narrow-faced flake/blade core on chunk eGH4 (Fig. 7: 2) and a bidirectional transversal flake core made ona pebble e GH1.

Debitage structure of the old inventories shows the dominance offlakes (from 62.5% to 57%) over laminar blanks, which includeblades (from 32.6% to 34.7%), bladelets (from 4.6% to 6.9%) andmicro-blades (only two in lowermost level I and four in uppermostlevel III, including tools). The changes across the sequence arenegligible (Table 3). Newly recovered assemblages display a shift inthe structure in favor of small laminar products e bladelets andmicro-blades. Between 68% and 72% of flakes and especiallylaminar products have no cortex. The most frequent cortex positionis lateral and forming a back. Such removals play a significant rolein the creation and maintenance of flaking surfaces on prismaticcores. While blades with 100% and >50% of cortex are uncommon,primary flakes are much more numerous across the sequence(9.8%e12.1%e3.6%). The ratio of corticated (primary) flakes to coresis quite high and suggests on-site preparation of at least somecores: from the bottom to the top e 39 corticated flakes to 48 cores,38 to 26 and 6 to 7. Not surprisingly, blades with unidirectionaldorsal scars dominate all assemblages (63.7%e60.6%e76%) overblades with convergent (15.2%e7.5%e13%) or crested scar patterns(8.8%e17.5%e9%). Bladelets more often show convergent scars

(19.4%e19%e17.4%). Crested flakes and core tablets are commonand appear in similar proportions.

Table 4 shows some changes in lateral profiles of laminarproducts throughout the sequence: generally incurvate blade/letand flat (rectilinear) profiles decline slightly from the base to thetop, while twisted profiles (as frequent as incurvate) increasemarkedly, especially for bladelets (38.7%e42.9%e59.1%). Interest-ingly, no major differences in profiles between blades and bladeletswere recorded, except for the uppermost level III, where blades aremuch less often twisted than bladelet (44.7% contra 59.1%). Theabundance of triangular sections may also be explained by thefrequent use of narrow-faced cores to obtain narrow laminarblanks, nevertheless trapezoidal (similar frequency) and multiplesections (also for bladelets) confirms the intensity of both blade andbladelet production. This is also true for micro-blades recoveredduring recent excavations.

With respect to platform preparation of laminar debitage, plainbutts dominate in all Cosava levels (from 53.8% to 72.8%), accom-panied by linear butts, especially frequent for bladelets (Table 5).Bladelet punctiform butts show a sudden increase in the upper-most level (15.4%). In total laminar blanks with plain, linear, andpunctiform butts account for between 76% and 100% (the latterfigure refers to the bladelets in level III). Dihedral and crudely-facetted platforms were also recorded, but account fora maximum of 13.9% of specimens. Butt lipping of laminar productsis common in all levels (Table 6). This attribute displays severaltendencies among these laminar products: (a) blade lipped buttsdominates over bladelet, (b) bladelet unlipped butts increase, while(c) blade unlipped butts display a slight decrease. As for bulbs,a stable trend was recorded for all laminar blanks (including toolsmade on them): (a) diffused bulbs dominate (from about 52.6% to63% for blades and from 50% to 71.4% for bladelets) over blankswithout bulbs (ca. 30.9% for blades and slight decreasing from 25%to 14.3 and 23.1% for bladelets) and over developed bulbs (from16.5% to 13.2% for blades and from 25% to 15.4% for bladelets); (b)there is an increase within the sequence of diffused bulbs, whiledeveloped bulbs decline (Table 7). Table 8 displays the typicaldominance of obtuse interior flaking angles over right angles andthe frequency of inverted angles for all laminar products. Abrasion

Fig. 7. Cosava I. Unidirectional cores: 1 e blade partly turned on chunk (level I); 2 e

flake/blade narrow-faced on chunk (GH4); 3, 5 e bladelet narrow-faced (“burin-like”)

on flake (3 e level II, 5 e level III); 4 e blade/bladelet narrow-faced (“burin-like”) on

flake (level I).

Table 3

Cosava. Debitage structure (including tools).


n % n % n %

Flakes 396 62.5 315 63.0 166 57.0

Blades 207 32.6 164 32.8 101 34.7

Bladelets 29 4.6 21 4.2 20 6.9

Micro-blades 2 0.3 4 1.4

Total 634 100 500 100 291 100

Laminar products:

Blades 207 87.0 164 88.6 101 80.8

Bladelets 29 12.2 21 11.4 20 16.0

Micro-blades 2 0.8 0.0 4 3.2

Total 238 100 185 100 125 100

2009 excavations GH4/Level I GH3/Level II GH1-2/Level III

n % n % n %

Flakes 17 68.0 10 27.0 42 56.0

Blades 3 12.0 8 21.6 16 21.3

Bladelets 4 16.0 11 29.7 13 17.3

Micro-blades 1 4.0 8 21.6 4 5.3

Total 25 100 37 100 75 100

Laminar products:

Blades 3 37.5 8 29.6 16 48.5

Bladelets 4 50.0 11 40.7 13 39.4

Micro-blades 1 12.5 8 29.6 4 12.1

Total 8 100 27 100 33 100



of the butt edges was frequent, although a gradual decrease in thistechnique, both on blades (61.1%e51.4%e40.5%) and on bladelets(37.5%e30.8%e23.1%) was documented. Tools-on-blades displaythe same trend, though with a sudden decrease in the upper layer(74.1%e70.6%e33.3%). Also, abrasion is more frequent on bladesmodified into tools. Interestingly, trimming of the overhang bysmall removals became more commonly used over time, accom-panied by or replacing abrasion. Thus, trimming of butt edgesincreased on blades (55.8%e61.1%e79.5%), on blades used as toolblanks (66.7%e66.7%e77.8%) and especially on bladelets (43.8%e61.5%e92.3%). This fact suggests that different techniques wereused to eliminate overhang on blade cores.

4.4.3. Core maintenance products

These products, which include crested blades/lets and flakes,tablet-flakes, débordant flakes and flank-flakes from laminar coresare abundant throughout the Cosava sequence, especially in thetwo lowermost levels. Technical flakes are numerous in all levels,often showing core platform rejuvenation by tablets, as well as coreworking surface maintenance by various crests and débordant

removals (though some of these might derive from polyhedral anddiscoidal core reduction). Crested blades are both primary andsecondary. Primary crested blades with two prepared slopes/versants (central crest position) are rather rare. Instead, most haveone preparation surface (partial or complete) with flakes directedfrom a flat natural surface: these come from the sides of cores(lateral crests). Secondary crested blades are frequent, includingmostly lateral examples, some neo-crests and combined items withoverpassed ends. Crested blades (Fig. 8: 4; 11: 5) were also selectedfor tool production together with crested/débordant flakes. Onlythree crested bladelets were observed: one-sloped (level I),primary and secondary (level III).

4.4.4. ToolsOpal plays a major role in tool manufacturing (55%e65%).

Radiolarite, flint and chalcedony were also regularly used for toolproduction, while tools on jasper, quartz, and quartzite occur

episodically. Tools are rather abundant (from the base to the top:19.4%e16.6%e12.6%). In the two lowermost levels tools are madeon both flakes (level I e 44.8% and II e 50.5%) and blades (level I e53.1% and II e 48.4%). Blades (61.5%) together with small laminarblanks (bladelets e 7.7% and micro-blades 10.3%) were selected fortool production much more often in uppermost level III. The rarityof tools on bladelets and micro-blades can be explained by pastexcavation practices and the lack of lithic concentrations duringrecent fieldwork. Analysis of the old Banat collections revealedmore retouched tools in all levels than were published by Mogo-sanu (this is also true for most Cosava samples). Inconsistencies infrequencies of some tool types between Mogosanu’s publishedinventories (Mogosanu, 1978: 80) and recently re-studied material(Table 9) are due to the loss of some artifacts and illegible labeling,as well as differences in classification practices (e.g. carinatedendscraper/core or retouched blade/retouched piece on blade).Despite these biases, the main tool categories display quite similarproportions.

Endscrapers are still dominant among the retouched tool typesthroughout the sequence: from 40% to 50% (Mogosanu’s data) andfrom 25.9% to 16.2% (our analysis: essential counts) (Table 9; Fig. 8).Combining endscrapers with carinated cores, which were mostlyclassified as tools by previous workers, produces counts similar toMogosanu’s. The decline in numbers of these tools in uppermostlevel III can be explained by the loss of some pieces (12 circa 6).Carinated and thick endscrapers were usually made on massiveflakes, while the simple ones mostly on blades. Carinated (withlamellar sub-parallel/parallel retouch) and thick endscrapers(mostly scalar modifications) are frequent in two lowermost levels.However carinated pieces (tools/cores) and thick endscrapers werealso recorded in uppermost level III. Carinated and thick end-scrapers show variability, including nosed, shouldered, and doublespecimens on lateral/bilateral retouched blanks, including Auri-gnacian blades (Table 10). Some of them were distally/proximallythinned or truncated.

Burins occur in small numbers (7 and 3 items; Mogosanu’s data:9 and 5 items) in the two lowermost levels and only a single piece

Table 4

Cosava. Blank lateral profiles (including tools).


Blade Blet/Micro-blade Blade Bladelet Blade Blet/Micro-blade

n % n % n % n % n % n %

Convex 4 2.2

Flat 46 25.6 7 22.6 23 14.6 3 14.3 25 26.6 4 18.2

Incurvate 64 35.6 12 38.7 53 33.8 9 42.9 27 28.7 5 22.7

Twisted 66 36.7 12 38.7 79 50.3 9 42.9 42 44.7 13 59.1

Irregular 2 1.3

Total 180 100 31 100 157 100 21 100 94 100 22 100

Table 5

Cosava. Butts (including tools).


Blade Blet/Micro-blade Blade Bladelet Blade Bladelet

n % n % n % n % n % n %

Cortical 4 4.3 4 5.6

Plain 67 72.8 8 57.1 49 68.1 8 61.5 27 71.1 7 53.8

Linear 7 7.6 4 28.6 3 4.2 3 23.1 5 13.2 4 30.8

Punctiform 1 1.1 1 7.1 3 4.2 2 15.4

Dihedral 6 6.5 3 4.2 2 15.4 4 10.5

Crudely-facetted 7 7.6 1 7.1 10 13.9 2 5.3

Total 92 100 14 100 72 100 13 100 38 100 13 100



in level III. These are represented by carinated transverse (2 in levelI), as well as by single examples of carinated transverse double(level II) and carinated dihedral double burins (level III) on flakesand blades. Angle burins on snap and/or on retouched blanks andon truncation (oblique and straight) were documented in levels Iand II (one is angle on snapmultifacetede level II) (Table 11; Fig. 9).

Retouched blades (with continuous non-marginal quite invasiveretouch) are numerically significant in all levels and equal to end-scraper frequencies in levels I and II. Mogosanu’s category probablycontains our ‘retouched pieces on blade’ (with partial, discontin-uous and/or marginal retouch), e.g. in level I. Regardless, retouchedblades are significant in all Cosava assemblages and include Auri-gnacian types in all levels: with scalar lateral/bilateral retouch,some pointed (Fig. 10: 7, 8) and strangled items (Fig. 11: 6). Abun-dant unidentifiable retouched blades, often small medial fragments(20 in level I and 12 in level II) might increase the Aurignacian bladecategory.

Retouched pieces on blades and on flakes with light shortdiscontinuous or partial retouch (while non-marginal) arecommon. These pieces on blades have obverse retouch (contra toolson bladelets with inverse and alternate modification).

Sidescrapers in the two lowermost layers (10 and 12 respec-tively, i.e. ca. double the number in comparison with publisheddata) were seen as proof for the archaism of these industries(Chirica et al., 1996). These tools were produced mostly on flakesand are variable, including single lateral and transverse, doublebilateral, convergent and canted forms, some with thinned back,base or truncated base (Fig. 11: 1e3).

Notched and denticulated pieces (Fig. 11: 4e5) are present insmall quantities and were made more on flakes in level I and moreon blades in levels II and III (they also occurred in assemblages fromthe new test pits). Only one denticulate on flake per level wasrecorded in Mogosanu’s collection.

Non-geometric microliths on bladelets were documented in alllevels (e.g. Hahn,1977, Tafel 164: 3, 10; 165: 13): they are rare in thelowermost levels (3e1 or 2.2%e1.1%) and more frequent in upper-most level III (8 items or 21.6%). Additional tools were recoveredfrom test pits in 2009 (1 in GH3 and 3 in GH1-2) (Fig. 15: 1e4).Recorded ‘small-sized’ tools are represented by three types:Dufour bladelets/micro-blades of Dufour subtype (alternate orinverse fine/micro-scalar, semi-steep retouch), pseudo-Dufourbladelets/micro-blades (idem obverse retouch) and Font-Yves

points (idem obverse convergent retouch). In the lowermostlevels bladelets were used as blanks to produce pseudo-Dufours,while in level III micro-blades (W < 7 mm) are as frequent as bla-delets as ‘supports’ for all three types of ‘micro-tools’. Curiously,one blade (W¼ 15.7mm)was alternatively retouched in the Dufourmanner. A single Font-Yves point was recorded by Mogosanu inlevel I (unfortunately absent from published drawings and thestudied collection) as well as in level III (Fig. 15: 14). In terms offragmentation mode and metrics, only the Font-Yves point and oneDufour bladelet are complete: 37.8; 9.3; 3.4 mm and 30.8; 7.9;2.6 mm respectively. The remaining Dufours and pseudo-Dufoursare broken (proximal, medial and distal parts) with max.L > 29.4 mm. Despite the quantitative difference between levels,these tools exhibit similar morphological, technological and typo-logical patterns: 1) exclusively ‘on-axis’ detachment; 2) unidirec-tional, rarely convergent dorsal scars; 3) incurvate and rectilinear/flat lateral profiles e exclusively in the lowermost levels andcombined with less frequent twisted, i.e. 4 contra 7incurvated þ straight); 4) plain and linear butts, lipped andunlipped, with developed, diffused bulbs and without them; 5)right and obtuse internal butt angles; 6) dorsal butt edge reduction/overhang elimination by small removal trimming and/or abrasion;7) continuous semi-steep fine/micro-scalar retouch; 8) use of localopal together with rarer or exotic raw materials (brown/red radi-olarite, jasper and white/grey flint). Thus, there no evidence fora clear shift of retouched bladelet/micro-blade Dufour and pseudo-Dufour into the Roc-de-Combe sub-type.

Truncated and thinned pieces on blades and flakes appear at thebeginning and at the end of Cosava sequence and might representpre-forms, e.g. burins on truncation.

4.4.5. Lithic technology: summary observations

Core reduction was mainly aimed at production of medium-sized blades, bladelets and micro-blades throughout thesequence. Although flakes appear as the dominant debitage cate-gory, their production was marginal and unsystematic (fivedifferent cores in level I and a single in level II). Apart fromexhausted pieces, only one mixed polyhedral/discoidal core mightsuggest systematic use of a Middle Paleolithic manner of flaking.Abundant flakes were obtained during different stages of bladecore reduction: cortex removal, initial platform preparation/re-preparation, crest installation, flaking surface maintenance (e.g.

Table 6

Cosava. Butt lipping (including tools).



n % n % n % n % n % n %

Lipped 50 53.8 3 20.0 35 47.3 4 28.6 16 42.1 5 38.5

Semi-lipped 24 25.8 7 46.7 29 39.2 6 42.9 15 39.5 1 7.7

Unlipped 19 20.4 5 33.3 10 13.5 4 28.6 7 18.4 7 53.8

Total 93 100 15 100 74 100 14 100 38 100 13 100

Table 7

Cosava. Bulbs (including tools).



n % n % n % n % n % n %

Developed 16 16.5 4 25.0 8 11.0 2 14.3 5 13.2 2 15.4

Diffused 51 52.6 8 50.0 46 63.0 10 71.4 21 55.3 8 61.5

None 30 30.9 4 25.0 19 26.0 2 14.3 12 31.6 3 23.1

Total 97 100 16 100 73 100 14 100 38 100 13 100



narrowing) and exhaustion. On the other hand, it appears thatmassive flakes were brought to the site to be exploited in differentways: 1) transversally from ventral face (striking platform) ewide-faced cores, carinated scraper pattern (exploitation of supportthickness); 2) longitudinally on narrow slice/edge e narrow-facedcores, burin pattern (exploitation of narrow support length).

Blade production was principally based on longitudinal volu-metric nodule/chunk/pebble reduction (prismatic cores), while bla-delet technology was more variable and resulted from: (a)advanced blade core reduction, (b) longitudinal flake slice reduction

(narrow-faced cores, including exploitation of flattish fragments ortool recycling) and (c) transversal flake/flat fragment reduction

(carinated cores/tools). The scarcity of cortical blades and corticalbutts in comparison with primary flakes shows that many fully

shaped laminar cores were prepared by short removals prior toblade production. Nodule pre-forming was usually accomplishedthrough the creation of a lateral (often partial) one-sloped crest(rarely central/frontal two-sloped). Other cores were unpreparedand together with a series of primary and secondary removals withhigh triangular and lateral steep cross-sections document directexploitation following natural convexities/ridges of the initial non-cortical chunk. Thus as in many blade industries, Cosava debitagewas (a) prepared and (b) direct (without crest installation).

Striking platforms are usually single-blow (acute or right),sometimes crudely-faceted. Core maintenance was achieved byadditional lateral crests, re-preparation (neo-crests), narrowing,back flattening, and systematic platform rejuvenation by largeflake-tablets (partial or total), platform edge abrasion and trimmingby short elongated removals that resulted in sub-cylindrical cores.Longitudinal uni-/bidirectional reduction of narrow parts of flattishchunks/plaquettes (with/without crest installation) was commonpractice as well as exploitation of narrow flake edges (burin-likepattern). All but one of these narrow-faced cores-on-flakes also

Table 8

Cosava. Butt angles (including tools).



n % n % n % n % n % n %

Inverted 11 12.2 9 13.6 2 20.0 5 14.3 1 14.3

Obtuse 64 71.1 5 50.0 45 68.2 6 60.0 26 74.3 5 71.4

Right 14 15.6 5 50.0 12 18.2 2 20.0 3 8.6 1 14.3

Acute 1 1.1 1 2.9

Total 90 100 10 100 66 100 10 100 35 100 7 100

Table 9

Cosava. Tool types.

Mogosanu’s

excavations

Level I Level II Level III


Endscraper 36 24.8 25.9 22 24.2 25.0 6 15.4 16.2

Borer 1 1.1 1.1

Burin 7 4.8 5.0 3 3.3 3.4 1 2.6 2.7

Retouched blade 32 22.1 23.0 18 19.8 20.5 9 23.1 24.3

Aurignacian 12 10 2

Retouched pieces

on blade

16 11.0 11.5 13 14.3 14.8 6 15.4 16.2

Notched piece 7 4.8 5.0 1 1.1 1.1 2 5.1 5.4

Denticulated piece 1 0.7 0.7 2 2.2 2.3 1 2.6 2.7

Sidescraper 10 6.9 7.2 12 13.2 13.6 2 5.1 5.4

Microlith 3 2.1 2.2 1 1.1 1.1 8 20.5 21.6

Dufour 1 2

Pseudo-Dufour 3 5

Font-Yves point 1

Backed piece 2 1.4 1.4

Backed blade 1

Backed flake 1

Retouched piece

on flake

18 12.4 12.9 15 16.5 17.0 1 2.6 2.7

Truncated piece 4 2.8 2.9

Thinned piece 2 1.4 1.4 1 2.6 2.7

Bifacial piece 1 0.7 0.7

Unidentifiable 6 4.1 3 3.3 2 5.1

Total 145 100 100 91 100 100 39 100 100

SUB-Total tools

on blades

77 53.1 44 48.4 24 61.5

SUB-Total tools

on bladelets

3 2.1 1 1.1 3 7.7

SUB-Total tools

on micro-blades

4 10.3

SUB-Total tools

on flakes

65 44.8 46 50.5 8 20.5

Total 145 100.0 91 100.0 39 100.0

Fig. 8. Cosava I. Endscrapers: 1 e carinated shouldered on flake (level II); 2 e carinated

nosed on flake (level II); 3 e thick ogival distally thinned on flake (level I); 4 e cari-

nated on crested bilaterally retouched blade (level I); 5 e thick on bilaterally retouched

blade (level I).



show pre-forming traces (crest, lateral narrowing and distal thin-ning) or direct unidirectional reduction. Flaking surfaces wereplaced on distal or lateral edges, with platforms correspondingly onlateral or distal/proximal parts of a blank. Final products are mid-sized, small blades and/or bladelets variable in shape (commonlyrectangular), with dominant ‘on-axis’ detachment pattern, withmore frequent flat or incurvate than twisted lateral profiles. Cari-nated reduction sequences are quite variable judging from cores.Usually unidirectional transversal exploitation began with the flat,thick part of a massive flake using the ventral face or naturally flatsurface of a chunk as a platform producing first generation oflaminar products (some blades and bladelets). With progress of

debitage, two tendencies are observed. The first shows continuousparallel or convergent unidirectional circular/semi-circular or distalfrontal reduction (combined with bilateral narrowing by means ofordinary flakes and rejuvenation removals) and consequentlyresulted in wide- or narrow-fronted cores. Combination of flakingsurfaces is another trend, in which platform re-orientation orsecond platform preparation resulted in (a) bidirectional removalsfrom one or two core faces (placed adjacently or alternatively) or(b) change in core orientation. Additional flaking surfaces placed indifferent plans one to another create wide-fronted carinatedorthogonal-adjacent/-alternate cores. Platforms of carinated coresare commonly plain but sometimes crudely-faceted (a result ofrejuvenation by partial tablets). Overhangs were often eliminatedby abrasion and/or trimming using small elongated removals. Finalproducts are blades, bladelets and micro-blades. It is likely thatwide-fronted carinated cores produced different laminar blankswith more incurvate profiles and pointed shapes, while narrow-faced coring with nosed pattern (including nosed scrapers)offered more twisted ‘of-axis’ products. Also, twisted blanks mightbe obtained from flanks of wide carinated pieces (as in Champ-Parel, Dordogne e Chadelle, 2005). The increase in the twistedpattern, especially for bladelets/micro-blades throughout theCosava sequence from the base to the top (38.7%e42.9%e59.1%) fitswell with the frequency of carinated bladelet/micro-blade cores(44.2%e45.5%e57.1%).

4.4.6. Reduction sequencesLaminar production in Cosava includes three co-existing

systems with continuous reduction of (a) ‘regular’ prismatic cores(b) narrow-faced (burin-like) cores and (c) carinated pieces (coresand tools). Continuity of debitage systems is confirmed by blade/let

Table 10

Cosava. Endscrapers.


n % n % n %

Endscraper on blade: 18 50.0 9 40.9 4 66.7

Carinated, on bilaterally retouched blade 1

Thick, nosed, on bilaterally retouched blade 1

Thick, on bilaterally retouched Aurignacian blade 2 1

Thick, on bilaterally retouched blade 4

Thick, on laterally retouched blade 1 1

Simple, on laterally retouched blade 3 1 1

Simple, on laterally retouched Aurignacian blade 1

Simple, on bilaterally retouched blade 5 1

Simple 1 3 2

Fan-shaped 1

Unidentifiable 1

Endscraper on flake: 18 50.0 13 59.1 2 33.3

Carinated, double 1

Carinated, thinned base 1

Carinated, lateral-proximal 1

Carinated, on laterally retouched flake 1

Carinated, shouldered 1

Carinated, nosed 1 1

Shouldered (flat) 1

Thick, shouldered 2

Thick, on bilaterally retouched flake 1

Thick 2 5

Thick, nosed, on laterally retouched flake 1

Thick, nosed, with truncated-faceted base 1

Thick, on bilaterally retouched flake 1

Thick, lateral, thinned base 1

Thick, ogival 1

Thick, ogival, distaly thinned 1

Thick, unidentifiable 1

Simple 3

Fan-shaped, on bilaterally retouched flake 2 1

Ogival 1 1

Total 36 100 22 100 6 100

Table 11

Cosava. Burins.


n % n % n %

Burin on blade: 2 28.6 2 67

Angle, on snap, on laterally retouched blade 1 1

Angle, on bilaterally retouched blade

Angle, on oblique truncation 1

Carinated, transverse, double, on laterally

retouched blade

1

Burin on flake: 5 71.4 1 33 1 100

Angle, on snap 1

Angle, on snap, on bilaterally retouched flake 1

Angle, on snap, multifaceted 1

Transverse, on straight truncation 1

Carinated, transverse 1

Carinated transverse, on laterally retouched flake 1

Carinated, diheadral, double 1

Total 7 100 3 100 1 100

Fig. 9. Cosava I. Burins: 1, 2 e carinated transverse on flake (level I); 3 e carinated

transverse double on laterally retouched blade (level II); 4 e carinated dihedral double

(level III).



scars on working surfaces of mid-sized prismatic, narrow-faced(burin-like) cores and large carinated cores (e.g. mixed blade/bla-delet, see Table 2). The overall composition of the study sample (thedominance of large knapping products in the old collection andsmall blanks and waste in the new assemblages, numerous cores atdifferent reduction stages and high tool frequencies), various deb-itage products (with/without cortex, core maintenance pieces andblanks) and metric data, suggests on-site reduction of at least someopal cores from the very beginning. This includes cortex removal,crest preparation, maintenance of debitage by neo-crests and oftenplatform rejuvenation by flake-tablets. Reduction sequences wereadvanced and often successful as suggested by 27 items or 69.2%fully shaped cores intermediate between pre-cores and exhaustedones in good condition for subsequent flaking in level I, as well asthe rarity of initial and failed cores in general. Metric data ofcomplete cores in level I is in agreement with this statement:maximum dimensions (L; W; Th) are 93.6; 47.5; 42.7 mm (bladecore), while averages are 44.6; 40.2; 42.8 mm and 45.2; 40.8;43.2 mm (including pre-cores). However, duration of laminarproduction varied, with long sequences for prismatic and carinatedcores, and short ones for narrow-faced cores-on-flakes. At the sametime, good quality raw material is scarce in the site’s nearbysurroundings, and on-site gravels provide mostly quartz andquartzite which were also knapped. Several other aspects,including the lack of nodules and rarity of tested pieces, low/moderate frequency of opal cortical elements provide evidence foroff-site pre-shaping of opal cores (testing, partial pre-forming/cortex removal, production of large flakes for carinated andnarrow-faced cores). This is also true for some tools, e.g. burins (asingle jasper burin spall in level III) and especially for many artifact

Fig. 11. Cosava I. Tools: 1 e sidescraper transverse (level I); 2 e sidescraper straight-

concave (level II); 3 e sidescraper convergent (level II); 4 e denticulated piece

lateral on core tablet (level I); 5 e notched piece on crested blade (GH2), 6 e strangled

blade (level I).

Fig. 10. Cosava I. Retouched blades: 1, 2, 3, 5, 6 e bilaterally retouched (1 e level I; 2, 3,

6 e level II; 5 e GH4); 4 e bilaterally retouched with distal oblique truncation (GH3); 7,

8 e pointed (level I); 9 e laterally retouched (9 e level III).

Fig. 12. Românesti-Dumbr"avita I, Trench 4, GH3. Technological refitting (A e flake and

micro-blade with core; B e micro-blade with notched piece on blade) and conjoining

of broken retouched piece on blade (C).



categories of rare/exotic chalcedony, several varieties of flint andradiolarite, jasper, etc. Many items of these raw materials (oftentools) seem to have been transported to the site in finished form.

4.4.7. Techniques

Hard stone hammer percussion was normally used at thebeginning of laminar debitage (testing, core preparation) and formaintenance of blade production surfaces (platform rejuvenation,restoring of working surface), as well as for extraction of largeflakes or more rarely in flake core reduction (discoidal, polyhedral).Laminar detachment commonly started from creation of a widestriking platform (single-blow cores) or without any preparation,using naturally flat surfaces on non-cortical chunks or flake ventralfaces. The ‘easy’ beginning was followed by very careful mainte-nance (elimination of overhang by abrasion and/or trimming by

small elongated removals) and platform restoration by flake-tablets(partial or complete). Blanks commonly have abraded and/ortrimmed proximal/dorsal parts and plain/linear butts. Duringproduction stages striking blows were aimed mostly close to theedge of the core platform (marginal percussion), which resulted innarrow flat butts of final products. These laminar products(including many flakes) clearly display weak/absent bulbs oftencombined with butt lipping and obtuse or right (less common)interior platform angles. The use of direct soft hammer percussionis well-documented. Seemingly, soft stone percussion was accom-panied by organic tools (indicated by invisibility of impact points),which should be clarified in the future. Available data are too scarceto reconstruct the difference in percussors used for blade and bla-delet manufacture. Curiously, not a single hammerstone wasrecorded in Cosava or Românesti.

Fig. 13. Cosava I and Românesti-Dumbr"avita I: (1) main core and (2) tool groups.



5. Românesti-Dumbr"avita: site potential and comparativeoutline

The newly obtained data from Românesti I (test pits and exca-vations in 2009e2010) and the ongoing study of old collectionsprovide an interesting set of comparative information and gives

a first impression of this site’s potential (Sitlivy et al., inpreparation). These assemblages will not be described in detailbut used to show some differences between the two Banat sites.Românesti-Dumbr"avita I is a huge open-air settlement, which, incontrast to Cosava, revealed a set of high density clusters(Mogosanu, 1978). One of these, discovered in 2009 and excavated

Fig. 14. Cosava I and Românesti-Dumbr"avita I bladelet attributes: (1) lateral profiles, (2) butt lipping and (3) bulbs.



over a small area (5 m2), accounted for 253e396 artifacts recordedin situ per square from all Paleolithic occupations (reaching a totalof 7505 Aurignacian artifacts with water sieving). This clusteringstrongly differs from the entire Cosava record, including bothnew (413 items in 2009) and old collections (1605 items).

In contrast to Cosava, the Românesti site differs in several ways,including morphological position, finer sediment and slightlydissimilar soil development (see Kels et al., in preparation). ForRomânesti I, the lowermost quartz industry (level I, belonging toGH4) might not be older than 57 $ 5.4 ka (luminescence). TheAurignacian assemblages from GH3 most likely date to MIS 3, withages between 45.1 $ 4.9 ka and 35.5 $ 3.9 ka BP (luminescence)from the middle part of the analyzed section. Toward the top of theprofile and from the lower part of the sediments of GH2, theluminescence age of 19.2$ 2.3 ka dates loess sedimentation duringthe Upper Pleniglacial (MIS 2), which fits to the archaeological levelwith a Gravettian/Epigravettian inventory.

The newly excavated Aurignacian inventories occur continu-ously in the different levels of GH3 showing no sterile, but variablevertical artifact density, clearly suggesting repeated occupations.Similar to Mogosanu’s stratigraphic record, the Aurignacianassemblages were found in different levels of GH3 (rarely GH4),sandwiched by less abundant Epigravettian/Gravettian at the topand an industry with isolated, mostly quartz artifacts (Middle/Upper Paleolithic?) at the base of the sequence. The horizontaldistribution of artifacts appears to indicate their primary positionas both conjoining of broken artifacts and technological refittingwere possible (Fig. 12).

Preliminary observations made on a new sample of 7505 arti-facts from GH3, trench 4 and A (2009/10) and 2654 items ofMogosanu’s level III from the same geological horizon allow us topoint out some differences between the lowermost Aurignacianindustries of the two neighboring sites, notwithstanding theircommon features.

With respect to raw material exploitation, the assemblages atCosava exhibit broader diversity in knapped stones and frequencyof rare/‘exotic’ rocks, usually of better quality. The general artifactstructure shows that cores/pre-cores (6.4%) and tools (19.4%) aremore abundant in Cosava, level I (contra 1.1%/0.3% of cores and 6.1%/2.3% of tools in Românesti, level III/GH3, respectively). The corecomposition is similar: carinated and laminar prismatic ones areabundant in both sites, while in Românesti, GH3 (new excavation)carinated cores occur episodically, replaced by blade prismatic andespecially blade/let narrow-faced cores-on-flakes (burin-like)(Fig. 13: 1). Moreover, two recycled tools were used as cores withthe same burin-like technique.

Some differences have also been recorded in bladelet produc-tion. The bladelet lateral profiles vary (Fig. 14: 1): rectilinear (flat)profiles are more common in level III in Românesti (32.2%) than inCosava lowermost level I (22.6%), regardless of the high percentageof twisted blade/lets in both assemblages (37.1/37.7%e36.7/38.7%for blade/lets); the new sample (371 pieces) in GH3 at Românestialso shows a dominance of flat (49.3%) and twisted profiles (37.5%)for bladelets as well as for blades (247 pieces e 43.7% flat and 38.1%twisted profiles). Bladelet butt lipping is more common for GH3 inRomânesti (32.7%) (Fig. 14: 2), while the bulb pattern showsstronger bulbs in Românesti, level III (developed bulbs rise to 31.8%,while this attribute in Cosava comprises 25%) (Fig. 14: 3). Bladeletbutt edge abrasion is well represented in GH3 at Românesti (60.9%)and much less frequent in level III of Românesti (32.4%) and inCosava level I (37.5%). However, this techniquewas accompanied bytrimming of butt edges with small elongated removals in Româ-nesti, level III (51.5%) and to a lesser extent in Cosava (43.8%).

For the toolkits, the absence of burin spalls in Cosava (excepta single one in the uppermost layer) is correlated to the low

frequency of burins (G > B at minimum 5 times; in level III inRomânesti these tools appeared in similar numbers); burins arealso more diverse in the latter site. Retouched blades, includingAurignacian, are more common in Cosava I (23% contra 10.8% inRomânesti, level III and 3.5% in GH3 e Fig. 13: 2), while microlithsshow an opposing pattern (2.2% contra 12.8%). Finally, combined(endscraper/burin) and mixed tools (i.e. burins) occur in smallquantity only in the Aurignacian sequence of Românesti. Moreover,preliminary analysis of the tool sample (141 identifiable pieces)from the Românesti cluster in GH3 differs dramatically from allother assemblages by its very high microlithic component (56%,including 64 Dufour bladelets/micro-blades with a systematicpattern of fine semi-steep/abrupt retouch, i.e. direct on the left edgeand inverse on the right) and rare (6) often fragmented Krems andFont-Yves points (Fig. 15: 5e13). Small tools were accompanied bysome burins, marginally or partially directly retouched pieces onblades/flakes, few retouched blades (including Aurignacian),notches, sidescrapers, and only two simple endscrapers (one ofwhich is on exotic flint). Thus, while having a general Aurignacianbackground and many common characteristics, the assemblages ofboth sites demonstrate a certain degree of technological andtypological variability.

6. Discussion and conclusion

The examination of old and new assemblages from Cosavaconfirms the slight contamination of uppermost level III by somelater Upper Paleolithic (e.g. isolated small round and nail-shaped

Fig. 15. Non-geometric microliths on blade/lets/micro-blades. Cosava I: 1, 2, 3 e

Dufours (GH1-2), 4 e pseudo-Dufour (GH3), 14 e Font-Yves point (level III); Româ-

nesti-Dumbr"avita I, GH3: 5e11 e Dufours, 12 e Font-Yves point, 13 e Krems point.



endscrapers, and two non-patinated blades of black obsidian). It iscommonly accepted that exotic obsidian (of Hungarian/Slovakian eNorth -Tokaj, Oas, or South Mediterranean origin) occurred in thisregion much later, during the Epipaleolithic (Mogosanu, 1983).However, several micro-artifacts (chips, bladelet and a micro-bladewith characteristic grayish thin matte patina) made from obsidianwere also recorded during our excavations at Cosava and Românestiin theAurignacian inventories, includingGH3and lowermostGH4, aswell as one small unworked obsidian pebblewith neo-cortex in GH1.

Our observations, as well as the revision of the RomanianAurignacian (Anghelinu and Nit"a, submitted for publication), makethe Aurignacian attribution of both Cosava I (levels I, II and III) andRomânesti I (at least level III) indisputable. However, anotherimportant question arises: to which facies and chronological phaseof the Aurignacian technocomplex may these industries beassigned? Can these assemblages, especially the lowermost, beattributed to an initial, emergent stage of the Aurignacian tech-nocomplex (and tentatively to the first AMH entering Europe alongthe Danube from the Mediterranean zone), or are they rather late(that is, contemporary or later) presences of the Central Europeantype Aurignacian, as the initial excavator thought (Mogosanu,1978)?

The general typological structure of the toolkits of the Banatinventories clearly suggested a direct connection to the CentralEuropean, especially the Krems type Aurignacian (e.g. Kozlowskiand Kozlowski, 1975; Mogosanu, 1978; Demidenko, 2002). Apartfrom the Krems-Hundssteig collection, whose integrity is actu-ally questionable (Teyssandier, 2008), other similar assemblagesin Central and Eastern Europe can also be assigned to this group:Siuren I rock shelter, Units H and G (e.g. Kozlowski andKozlowski, 1975; Demidenko et al., 1998; Demidenko,2000e2001; Demidenko and Otte, 2007; Demidenko et al.,2012) and Beregovo I (Usik, 2008), as well as the number of“Archaic Aurignacian/Protoaurignacian assemblages with Dufourbladelets” in Western Europe (Demidenko and Noiret, 2012:343e347). To current knowledge, the cultural assignment to theKrems-Dufour type, proposed long ago by Mogosanu (1978),generally fits Cosava (all levels) and at least the main concen-tration at Românesti, level III.

However, the recent refinement in the chrono-technologicalsubdivision of the Aurignacian (e.g. Chiotti, 1999; Lucas, 2000;Bon, 2002; Bordes, 2002; Teyssandier, 2003, 2007a, 2007b, 2008;Le Brun-Ricalens, 2005; Zilhão, 2006, 2007; Teyssandier et al.,2010) makes the interpretation of these inventories less straight-forward. For instance, Tincova was recently seen as a good candi-date for inclusion into the Protoaurignacian (Teyssandier, 2003;Zilhão, 2006) on the basis of the dominance of continuous bladecore reduction sequences, which resulted in blades (first generationof blanks) and then elongated bladelets with a straight profile(second generation), modified into points and Dufour sub-typebladelets; carinated scrapers are rare, Aurignacian retouch isabsent (Teyssandier, 2008).

For the Banat sites, a proper reevaluation of the all of theassemblages involved is required in order to allow a detailedcomparison both with other Protoaurignacian and ‘classical’ Auri-gnacian assemblages. Although the stratigraphic (and presumablythe chronological) position of the lowermost industries of the Banatsites is nearly identical, technological and typological features arenot similar. At first view, from a ‘Protoaurignacian perspective’,Cosava, level I differs from Tincova by the higher frequency ofcarinated pieces (19 pieces/44.2% of carinated cores and 6 items/4.3% of carinated endscrapers were recorded in Cosava), the pres-ence of Aurignacian blades (12 pieces or 8.6% in essential counts)and the quite high frequency (about 38.7%) of twisted bladelets.Thus, these features are not ‘archaic’ enough to place Cosava within

the Protoaurignacian as defined above, but rather in the range of theclassic Aurignacian phases.

However, recent attribute analysis of Banat assemblagesdemonstrates the presence of ‘classical Aurignacian features’ in theTincova inventory as well: dominance of carinated cores overprismatic, similar frequency of Aurignacian blades, and highertwisted bladelet pattern. In addition, 12 carinated pieces/cores(including atypical, nosed, core-like and rabot) were alreadyrecorded and partially illustrated in earlier publications (Hahn,1977, Tafel 170: 9e10; Mogosanu, 1978, Fig. 16: 1e8).

On the other hand, the newly excavated cluster in GH3 atRomânesti apparently expresses some ‘archaism’ in core types:dominance of prismatic and bladelet narrow-faced core-on-flakessimilar to bladelet technology at Kozarnika, level VII (Tsanova,2006), but also at Mitoc-Malu Galben, unit I (Noiret, 2009:Fig. 309); higher frequency of bladelet rectilinear profiles; andespecially the abundance of Dufour sub-type bladelets with ‘clas-sical’ retouch placement. Aurignacian blades are rare at Românesti.There is also no evidence for a clear shift from retouched Dufourand pseudo-Dufour bladelet/micro-blade to the Roc-de-Combesub-type, as documented in several Mediterranean sequences(Bon, 2002; Bordes, 2002; Demidenko and Otte, 2007). A similarcase can be made for the Cosava sequence. As for pointed bladelets(Font-Yves and Krems), these appear in all Banat industries in smallquantity, including Cosava uppermost level III. In brief, while thenon-geometric microliths suggest an ‘early Aurignacian’ sensu lato/Protoaurignacian pattern, other criteria (carinated forms, inten-sively retouched blades) rather point to a classical Aurignacian (orAurignacian I, sensu Mellars, 2006).

Thus, having a general Aurignacian background, the assem-blages of Cosava and Românesti, as well as Tincova, display a certaindegree of technological and typological variability due either to (a)chronological and/or (b) functional impact within the broad timeframe suggested by the luminescence chronology at Românesti(45.1 $ 4.9 ka and 35.5 $ 3.9 ka BP, see Kels et al., in preparation).While the available chronological data are obviously too distant andscattered for accurately following the diachronic changes sepa-rating these assemblages, the second causal factor for the vari-ability recorded (functionality) is already better supported by theBanat Aurignacian record.

Cases of certain specialization, variable activities and clusteringof lithic remains in the context of a larger settlement were alreadyrecorded by Mogosanu, but also during the new excavation atRomânesti. For example, the neighboring Românesti II spot yieldeda compact area with numerous retouched bladelets/micro-blades,including Dufour, rare Krems points and a few non-modified bla-delets (Hahn, 1977, Tafel 169: 17e28; Mogosanu, 1978). Regardlessof their high stratigraphic position and metric parameters (shorterthan those from other industries), these laminar products appearedrather ‘non-twisted’ and seemingly ‘archaic’. Obviously, assigningthe Banat Aurignacian inventories, including this concentration anda similar but older locus in GH3 at Românesti I (trench 4, 2009e10),as well as the Cosava I assemblages (with high retouched toolfrequency) to the ‘workshop’ category is much too simplistic, if notplainly false. None of the settlement/layers involved fit the exclu-sive ‘workshop’ definition. Even the allegedly local raw materialwas actually brought to settlements from a certain distance,sometimes as pre-worked nodules. Other activities (e.g. recycling,resharpening, hunting, domestic tasks) are evidenced by curatedlithic items, high percentage of tools (e.g. Cosava), and abundantmobile hunting micro-tools (Românesti, Tincova). Noticeablechanges in assemblage structure and typological range between thelower levels (I and II) and the uppermost level III at Cosava are noless telling for such a functional noise. Varying topographicconcentrations suggest at least repeated or longer duration



occupations (Românesti I), as opposed to short halts (Românesti II,Cosava). As for Cosava and Românesti, they might be either strictlycontemporaneous with different functions (as expected given theirproximity, which place each settlement in the exploitation range ofthe other), or simply distant in both time and function. As such,they may belong to the same chrono-cultural stage or not. In thelack of a better chronological control and basic subsistence data, notto mention the poor segregation of living floors (at Românesti atleast), we are currently far from being able to elaborate on theseaspects.

Unfortunately, although highly elaborated, most of the availablechrono-cultural frameworks acknowledged for the Aurignaciansuffer from similar weaknesses. For objective (i.e. lack of additionalinformation) or rather subjective (i.e. theoretical/methodological)reasons, they retain preferentially the stylistic-normative aspects intheir description of the Aurignacian lithic technology. The risk isobvious: unless strongly supported by absolute chronology,stylistic-based approaches may separate in evolutionary stages andthus explain away a rather fluid, likely functional and possiblysynchronous variability. The Banat situation seems to providea cautionary tale in this respect. Despite some clear variability(expressed in frequencies of almost the same tool types and somelaminar attributes), these assemblages have nevertheless a numberof common characteristics and appear rather homogeneous withina single technocomplex. At least in the case of the settlementsdiscussed above, the alleged succession of the Protoaurignacian andAurignacian I appears at best unclear: ‘archaic’ (e.g. straight Dufour/Krems implements) and ‘classic’; (i.e. carinated forms, Aurignacianblades, and twisted bladelets) appear, albeit in various amounts, inall assemblages.

Summing up, the Aurignacian (Krems-type) attribution ofCosava I (levels I, II and III), Românesti I (at least level III), includingthe newly recovered samples from both sites, and apparently Tin-cova, is correct and supports Mogosanu’s original interpretation.Technologically and typologically, these inventories exhibit a rangeof features that tentatively link them to the Protoaurignacian/Aurignacian 0 (Tincova, Românesti), but also to an early Aurigna-cian/Aurignacian 1 (Cosava). In fact, the Banat assemblages, espe-cially Românesti I, feature rather a combination of Proto- and earlyAurignacian traits, similar to that observed recently, e.g. in theassemblage C 4c4 at Isturitz (sandwiched between Proto- and earlyAurignacian industries and dated to 37,180 420 BP e Szmidt et al.,2010). The techno-typological variability of the Banat industriesseems to have partially chronological (throughout the Cosavasequence) and/or functional explanations (probably for all inven-tories) within the broad period from 45.1$4.9 ka and 35.5$ 3.9 kaBP (luminescence), the chronology obtained so far only for Româ-nesti I and which likely encompasses all these archaeologicalsequences. Still, depending strictly on the five-phase stylistically-based succession currently developed for Western Europe isperhaps unproductive for an accurate estimation of the regionalvariability of the Aurignacian in South-Eastern Europe and partic-ularly in Romanian Banat.

Given the isolation and the distant comparative data currentlyavailable for the Banat Aurignacian, only a fresh chronology sup-ported by absolute dates will enable the reconstruction of theregional time distribution and functional patterns of these indus-tries, as well as their integration in the Eurasian record. Hopefully,the TL samples recovered from Românesti and currently understudy will shed more light on these issues.

Acknowledgements

Fieldwork at Cosava and Românesti was supported by theUniversity of Cologne and “Valahia” University of Târgoviste. The

research in Romania was carried out in the framework of the CRC806 ‘Our way to Europe’, project B1 ‘The ‘Eastern Trajectory’: LastGlacial Palaeogeography and Archeology of the Eastern Mediter-ranean and of the Balkan Peninsula’, supported by the DFG (Deut-sche Forschungsgemeinschaft). Many colleagues have contributedto the fieldwork organization (M. Odangiu, B. Joita), and providedaccess to the archaeological collections in Lugoj Regional Museum(R. Pinca) and the Institute of Archaeology, Academy of Science inBucharest (R. Dobrescu). Drawings were made by F. Dumitru(“Valahia” University of Târgoviste). Luminescence dating wascarried out by Alexandra Hilgers and Anja Zander (Institute ofGeography, University of Cologne). Corrections and polishing of thetext were kindly made by S. L. Kuhn, followed by fruitful discus-sions. Further language editing was done by R. Miller. Pierre Noiretand an anonymous reviewer for Quaternary International furthercontributed to the improvement of our paper. To all of them weexpress our deepest gratitude.

References

Alexandrescu, E., B"alescu, S., Tuffreau, A., 2004. Nouvelles données chronologiques,technologiques et typologiques sur le Paléolithique supérieur ancien de laPlaine roumaine du Danube: le gisement de Giurgiu Malu Rosu. L’An-thropologie 108 (3e4), 407e423.

Alexandrescu, E., Olariu, A., Skog, G., Stenstr}om, K., Hellborg, R., 2010. Os fossilshumains des grottes Muierii et Cioclovina, Roumanie. L’Anthropologie 114,341e353.

Allsworth-Jones, P., 1986. The Szeletian and the Transition from Middle to UpperPaleolithic in Central Europe. Clarendon, Oxford.

Anghelinu, M., 2003. Evolutia gândirii teoretice în arheologia din România. Con-cepte si modele aplicate în preistorie. Cetatea de Scaun, Târgoviste.

Anghelinu, M., 2006. O paleontologie a “omului etern”: arheologia paleoliticului dinRomânia. Cercet"ari Arheologice XIII, 135e158.

Anghelinu, M., Nit"a, L. What’s in a name: the Aurignacian in Romania. QuaternaryInternational, submitted for publication.

Anghelinu, M., Nit"a, L., Sitlivy, V., Uthmeier, Th., B"altean, I. Looking around PesteraCu Oase: the beginnings of Upper Paleolithic in Romania. Quaternary Interna-tional, in press.

Benazzi, S., Douka, K., Fornai1, C., Bauer, C., Kullmer, O., Svoboda, J., Pap, I.,Mallegni, F., Bayle, P., Coquerelle, M., Condemi, S., Ronchitelli, A., Harvati, K.,Weber, G.W., 2011. Early dispersal of modern humans in Europe and implica-tions for Neanderthal behaviour. Nature 479, 526e529.

Bon, F., 2002. L’Aurignacien entre Mer et Océan. Réflexion sur l’unité des phasesanciennes de l’Aurignacien dans le sud de la France. Mémoire XXIX de la SociétéPréhistorique Française, Paris.

Bordes, J.-G., 2002. Les interstratifications Chatelperronien/Aurignacien du Roc deCombe et du Piage (Lot, France). Analyse taphonomique des industries lith-iques: consequences archeologiques. Ph.D. dissertation, Institut de Préhistoireet Géologie du Quaternaire, Université de Bordeaux I, Bordeaux.

Boroneant, V., 1991. L’acheuléen supérieur de Cladova (Roumanie) et la question dupaléolithique quartzitique. Anthropologie (Brno) XXIX/1e2, 29e39.

Borziac, I., Chirica, V., 2008. Periodisation culturelle, chronologie relative et radio-metrique des facies du Paleolithique superieur de l’espace carpato-dniestreen.In: Chirica, V., V"aleanu, M.C. (Eds.), Etablissements et habitations pré-historiques. Structure, organisation, symbole. Pim, Iasi, pp. 7e50.

Borziac, I., Chirica, V., V"aleanu, M.-C., 2006. Culture et société pendant le Paléo-lithique supérieur à travers l’espace carpato-dniestréen, Institut d’Archéologie.PIM, Iasi.

B"alescu, S., Lamothe, M., Mercier, N., Huot, S., B"alteanu, D., Billard, A., Hus, J., 2003.Luminescence chronology of Pleistocene loess deposits from Romania: testingmethods of age correction for anomalous fading in alkali feldspars. QuaternaryGeochronology 22 (10e13), 967e973.

B"altean, I., 2011a. Paleoliticul superior din Banat în contextul paleoliticului superiordin spat iul central-est european. Unpublished PhD. thesis, ArchaeologicalInstitute, Romanian Academy. Iasi.

B"altean, I., 2011b. The Palaeolithic in Banat. In: Tasic, N., Drasovean, F. (Eds.), ThePrehistory of Banat; I. The Palaeolithic and Mesolithic. Academiei, Bucuresti.

B"altean, I.C., Petrescu, S.M., Cinca, A., Negrei, D., Zilhão, J., 2008. Human presenceand prehistoric settlements in the Minis Karstic System an overview. In:Chirica, V., V"aleanu, M.C. (Eds.), Etablissements et habitations préhistoriques.Structure, organisation, symbole. Pim, Iasi, pp. 183e196.

Cârciumaru, M., 1999. Le Paléolithique en Roumanie. “Préhistoire d’Europe”, 7. J.Millon, Grenoble.

Cârciumaru, M., Moncel, M.-H., Cârciumaru, R., 2000. Le Paleolithique moyen de lagrotte CioareieBorosteni (commune de Pestisani, department de Gorj, Rou-manie). L’Anthropologie 104, 185e237.

Cârciumaru, M., Moncel, M.-H., Cârciumaru, R., 2002. The CioareieBorosteni Cave(Carpathian Mountains, Romania), Middle Paleolithic finds and technologicalanalysis of the lithic assemblages. Antiquity 76, 681e690.



Cârciumaru, M., Anghelinu, M., Steguweit, L., Nit"a, L., Fontana, L., Brugère, A.,Hambach, U., M"arg"arit, M., Dumitrascu, V., Cosac, M., Dumitru, F., Cârstina, O.,2006. The Upper Palaeolithic site of Poiana Ciresului (Piatra Neamt, North-eastern Romania) e recent results. Archäologisches Korrespondenzblatt 3,1e20.

Cârciumaru, M., Anghelinu, M., Nitu, E.-C., Cosac, M., Mur"atoreanu, G., 2007a. Geo-archéologie du paléolithique moyen, paléolithique supérieur, épipaléolithiqueet mésolithique en Roumanie. Cetatea de Scaun, Târgoviste.

Cârciumaru, M., Anghelinu, M., Nit"a, L., 2007b. The Upper Palaeolithic in the BistritaValley (Northeastern Romania). A preliminary review. Annales d’UniversitéValahia Târgoviste, section d’Archéologie et d’Histoire VIIIeIX, 107e124.

Chabai, V.P., Demidenko, Yu.E., 1998. The classification of flint artifacts. In:Marks, A.E., Chabai, V.P. (Eds.), The Middle Paleolithic of Western Crimea, vol. 1.ERAUL 84, Liège, pp. 31e51.

Chabai, V.P., Marks, A.E., Monigal, K., 2004. Crimea in the context of the EasternEuropean Middle Paleolithic and Early Upper Paleolithic. In: Chabai, V.,Monigal, K., Marks, A.E. (Eds.), The Middle Paleolithic and Early Upper Paleo-lithic of Eastern Crimea, vol. 3. ERAUL 104, Liège, pp. 419e460.

Chadelle, J.-P., 2005. Productions “intriquées” de lames et de lamelles dans l’Aur-ignacien de Champ-Parel locus 3 (Bergerac, Dordogne). In: Le Brun-Ricalens, F.(Ed.), Bordes, J.-G., Bon, F. (collab.), Productions lamellaires attribuées à l’Aur-ignacien. Chaînes opératoires et perspectives technoculturelles. MNHA,Luxembourg, pp. 193e208.

Chiotti, L., 1999. Les industries lithiques des niveaux aurignaciens de l’Abri Pataud,Les Eyzies-de-Tayac (Dordogne): étude technologique et typologique. Ph.D.dissertation, Institut de Paléontologie Humaine, Musée National d’HistoireNaturelle, Paris.

Chirica, V., 1996. Les habitats de Banat. In: Chirica, V., Borziac, I., Chetraru, N. (Eds.),Gisements du Paléolithique supérieur ancien entre le Dniestr et la Tissa. B. A. I.V. Helios, Iasi, pp. 125e138.

Chirica, V., 2001. Gisements paléolithiques de Mitoc. Helios, Iasi.Chirica, V., V"aleanu, M.C. (Eds.), 2008. Etablissements et habitations préhistoriques.

Structure, organisation, symbole. Pim, Iasi.Chirica, V., Borziac, I., Chetraru, N., 1996. Gisements du Paléolithique supérieur

ancien entre le Dniestr et la Tissa. Helios, Iasi.Conard, N.J., Bolus, M., 2003. Radiocarbon dating the appearance of modern

humans and timing of cultural innovations in Europe: new results and newchallenges. Journal of Human Evolution 44, 331e371.

Demars, P.-Y., Laurent, P., 1989. Types d’outils lithiques du Paléolithique supérieuren Europe. Cahiers du Quaternaire 14. Paris: CNRS.

Demidenko, Yu.E., 2000e2001. The European Early Aurignacian of Krems-Dufourtype industries: a view from Eastern Europe. European Prehistory 16e17,147e162.

Demidenko, Yu.E., 2002. Siuren I rock-shelter (Crimea): industrial-chronologicalsequence of the site and Aurignacian complexes. Archeological Proceedings(Rostov-na-Donu) 2, 29e67 (in Russian).

Demidenko, Yu.E., Noiret, P., 2012. The Siuren-I Aurignacian of Krems-Dufour typeindustries in the context of the European Aurignacian. In: Demidenko, Yu.E.,Otte, M., Noiret, P. (Eds.), Siuren I Rock-shelter. From Late Middle and EarlyUpper Paleolithic to Epi-Paleolithic in Crimea. The Paleolithic of Crimea, IV.ERAUL 129, Liege, pp. 343e357.

Demidenko, Yu.E., Otte, M., 2007. Siuren I (Crimea) as a key site for Aurignacianindustries of Krems-Dufour type in Eastern Europe. In: Bicho, N.F. (Ed.), Fromthe Mediterranean Basin to the Portuguese Atlantic Shore: Papers in Honor ofAnthony Marks. Promontoria Monografica 07. Centro de Estudos de Patrimonio,Departamento de Historia, Arqueologia e Patrimonio, Universidade do Algarve,pp. 101e107.

Demidenko, Yu.E., Chabai, V.P., Otte, M., Yevtushenko, A.I., Tatartsev, S.V., 1998.Siuren-I, an Aurignacian site in the Crimea (the investigations of the 1994e1996field season). In: Otte, M. (Ed.), Anatolian Prehistory. At the Crossroads of TwoWorlds, vol. 1. ERAUL 85, Liège, pp. 367e413.

Demidenko, Yu.E., Otte, M., Noiret, P. (Eds.), 2012. Siuren I Rock-Shelter. From LateMiddle Paleolithic and Early Upper Paleolithic to Epi-Paleolithic in Crimea. ThePaleolithic of Crimea, IV. ERAUL 129, Liège.

Djindjian, F., 2000. Compte-rendu, Cârciumaru, M. 1999. Le Paléolithique en Rou-manie, collection “L’Homme des origines” serie “Prehistoire d’Europe”, n! 7,Grenoble: Editions Jerome Millon. Bulletin de la Société Préhistorique Française97 (2), 309e324.

Djindjian, F., Kozlowski, J., Otte, M., 1999. Le Paléolithique supérieur en Europe.Armand Colin, Paris.

Gabori, M., 1976. Les civilisations du Paléolithique moyen entre les Alpes et l’Oural:esquisse historique. Akademiai Kiado, Budapest.

Hahn, J., 1970. Recherches sur l’Aurignacien en Europe centrale et orientale. L’An-thropologie (Paris) 74, 195e220.

Hahn, J., 1977. Aurignacien das ältere Jungpaläolithikum in Mittel-und Osteuropa.B}ohlau, Fundamenta A/9, K}oln & Wien.

Harvati, K., Gunz, P., Grigorescu, D., 2007. Cioclovina (Romania): affinities of an earlymodern European. Journal of Human Evolution 53 (6), 732e746.

Hoffecker, J.F., 2009. The spread of modern humans in Europe. Proceedings of theNational Academy of Science U S A 22 (106), 16040e16045.

Honea, K., 1984. Chronometry of the Romanian Middle and Upper Palaeolithic:implications of current radiocarbon dating results. Dacia XXVIII (1e2),23e39.

Honea, K., 1986. Rezultate preliminare de datare cu carbon radioactiv privindPaleoliticul mijlociu din Pestera Cioarei de la Borosteni (jud. Gorj) si Paleoliticul

superior timpuriu de la Mitoc-Malu Galben (jud. Botosani). Studii si Cercet"ari deIstorie Veche si Arheologie 37 (4), 326e332.

Honea, K., 1987. The chronology of Romania’s Palaeolithic. In: Chirica, V. (Ed.), Lagenèse et l’évolution des cultures paléolithiques sur le territoire de la Rouma-nie. B.A.I, Iasi, pp. 49e61.

Horvath, I., 2009. The Early Upper Palaeolithic in Romania: past and currentresearch. In: Camps, M., Szmidt, C. (Eds.), The Mediterranean from 50 000 to25 000 BP: Turning Points and New Directions. Oxbow Books, Oxford,pp. 137e162.

Hours, F., 1974. Remarques sur l’utilisation de listes-types pour l’etude du Paléo-lithique supérieur et de l’Epipaléolithique du Levant. Paléorient 2 (1), 3e18.

Inizan, M.-L., Reduron, M., Roche, H., Tixier, J., 1995. Technologie de la pierre taillée.Préhistoire de la pierre taillée 4. CREP, Meudon.

Kels, H., Protze, J., Sitlivy, V., Hilgers, A., Zander, A., Anghelinu, M., Lehmkuhl, F.Genesis of loess-like sediments and soils at the foothills of the Banat Moun-tains, Romania e examples from the Paleolithic sites Românesti and Cosava.Quaternary International, in preparation.

Kozlowski, J.K., Kozlowski, S.K., 1975. Pradzieje Europy od XL do IV tysiaclecia p. n. e.Panstwowe Wydawnictwo Naukowe, Warszawa.

Lazarovici, G., B"altean, I.C., Trinkaus, E., Cinc"a, A., Zilhao, J., Petrescu, S.,Moldovan, O., Rodrigo, R., Rougier, H., Pendea, F., Constantin, S., Quiles, J.,Soficaru, A., Petrea, C., Milota, S., Gherase, M., Sarcina, L., Codrea, V., 2005.Archaeological excavation report, NAR-code 50905.01. In: Cronica cercet"arilorarheologice din România. CIMEC.

Le Brun-Ricalens, F. (Ed.), Bordes, J.-G., Bon, F. (collab.), 2005. Productions lamel-laires attribuées à l’Aurignacien. Chaînes opératoires et perspectives tech-noculturelles. MNHA, Luxembourg.

Le Brun-Ricalens, F., Bordes, J.-G., Eizenberg, L., 2009. A crossed-glance betweensouthern European and Middle-Near Eastern early Upper Palaeolithic tech-nocomplexes: existing models, new perspectives. In: Camps, M., Szmidt, C.(Eds.), The Mediterranean from 50 000 to 25 000 BP: Turning Points and NewDirections. Oxbow Books, Oxford, pp. 11e34.

Lucas, G., 2000. Les industries lithiques du Flageolet 1 (Dordogne): approcheéconomique, technologique et analyse spatiale. Ph.D. dissertation, Institut dePréhistoire et Géologie du Quaternaire, Université de Bordeaux I, Bordeaux.

Marks, A.E., 1976. Glossary. In: Marks, A.E. (Ed.), Prehistory and Paleoenvironmentsin the Central Negev, Israel, vol. 1. SMU Press, Dallas, pp. 371e383.

Mellars, P., 2006. Archaeology and the dispersal of modern humans in Europe:deconstructing the “Aurignacian”. Evolutionary Anthropology 15, 167e182.

Mertens, S.B., 1996. The Middle Paleolithic in Romania. Current Anthropology 37,515e521.

Mogosanu, 1967. Descoperirile paleolitice ale Muzeului din Lugoj. Revista MuzeelorIV (6), 555e559.

Mogosanu, F., 1972. Information générale sur le Paléolithique du Banat (Sud-Ouestde la Romanie). Dacia N.S 16, 5e27.

Mogosanu, F., 1978. Paleoliticul din Banat. Biblioteca de Arheologie, 32, Bucuresti.Mogosanu, F., 1983. L’Aurignacien du Banat. In: Otte, M. (Ed.), Aurignacien et

Gravettien en Europe I. ERAUL 13, Liège, pp. 223e237.Mogosanu, F., Stratan, I., 1966. Noi descoperiri paleolitice în Banat. Studii si Cercetari

de Istorie Veche 17 (2), 335e344.Mogosanu, F., Stratan, I., 1969. Rezultatele ultimelor s"ap"aturi arheologice privind

palaeoliticul din nordul Banatului. Revista Muzeelor 1, 84e90.Moldovan, O., Milota, S., Sarcina, L., Trinkaus, E., B"altean, I., Soficaru, A., Rajka, G.,

2003. The oldest modern humans in Europe. Theoretical and Applied Karstol-ogy 16, 77e81.

Moncel, M.-H., Cârciumaru, M., Anghelinu, M., 2002. Le paléolithique moyen desCarpates Méridionales (Roumanie) et la grotte Cioarei-Borosteni. Des témoig-nages d’une fréquentation de la moyenne montagne a la faveur d’améliorationclimatique par des groupes de Néandertaliens? Anthropologie (Brno) 40, 11e32.

Noiret, P., 2004. Le Paléolithique supérieur de la Moldavie. L’Anthropologie 108,425e470.

Noiret, P., 2009. Le Paléolithique supérieur de Moldavie. ERAUL 121, Liège.Olariu, A., Hellborg, R., Stenström, K., Skog, G., Faarinen, M., Persson, P., Erlandsson, B.,

Alexandrescu, E., 2002. Datingof someRomanian fossil bones by combined nuclearmethods. Journal of Radioanalytical and Nuclear Chemistry 253 (2), 307e311.

Olariu, A., Skog, G., Hellborg, R., Stenström, K., Faarinen, M., Persson, P., Alexan-drescu, E., 2004. Dating of Two Human Fossil Bones from Romania Bay Accel-erator Mass Spectronomy. Report Wp1 IDRANAP 80-04/2004 of the EuropeanCenter of Excellence.

Olariu, A., Skog, G., Hellborg, R., Stenström, K., Faarinen, M., Persson, P.,Alexandrescu, E., 2005. Dating of two Paleolithic human fossil bones fromRomania by accelerator mass spectrometry. In: Olariu, A., Stenström, K.,Hellborg, R. (Eds.), Applications of High-Precision Atomic and Nuclear Methods.Academiei, Bucuresti, pp. 222e226.

Otte, M., Chirica, V., 1993. Atelier aurignacien à Mitoc Malul Galben (Moldavieroumaine). Préhistoire Européenne 3, 55e66.

Otte, M., López Bayón, I., Noiret, P., Borziac, I.A., Chirica, V., 1996a. Recherches sur lePaléolithique supérieur de la Moldavie. Bulletin de la Société Royale Belge,Anthropologie et Préhistoire 107, 45e80.

Otte, M., Ulrix-Closset, M., Cârciumaru, M., Beldiman, C., 1996b. Comportementstechniques au Mousterien de la ‘‘Pestera Cioarei” (Oltenie). In: Bietti, A.,Grimaldi, S. (Eds.), Reduction Processes for the European Mousterian. Qua-ternaria Nova VI, pp. 83e92.

Otte, M., Chirica, V., Haesaerts, P. (Eds.), 2007. L’Aurignacien et le Gravettien deMitoc-Malu Galben (Moldavie Roumaine). ERAUL 72, Liège.



Patou-Mathis, M., 2000e2001. Les grands mammiferes de la grotte Cioarei (Bor-osteni, Roumanie): repaire de Carnivores et halte de chase. PréhistoireEuropéenne 16e17, 57e63.

Peresani, M., 2008. A new cultural frontier for the last Neanderthals: the Uluzzian inNorthern Italy. Current Anthropology 49, 725e731.

P"aunescu, Al, 1989. Le Paléolithique et le Mésolithique de Roumanie (un brefaperçu). L’Anthropologie 93 (1), 123e158.

P"aunescu, Al, 1993. Ripiceni-Izvor. Paleolitic si Mezolitic. Academiei, Bucuresti.P"aunescu, Al, 2000. Paleoliticul si mezoliticul din spatiul cuprins între Carpati si

Dun"are. Agir, Bucuresti.P"aunescu, Al, 2001. Paleoliticul si epipaleoliticul din spatiul transilvan. Agir,

Bucuresti.Richards, M.P., Pacher, M., Stiller, M., Quilès, J., Hofreiter, M., Constantin, S., Zilhão, J.,

Trinkaus, E. , 2008. Isotopic evidence for omnivory among European cave bears:late Pleistocene Ursus spelaeus from the Pestera cu Oase, Romania. PNAS 105 (2),600e604.

Riel-Salvatore, J., Popescu, G., Barton, M., 2008. Standing at the gates of Europe:human behavior and biogeography in the Southern Carpathians during the LatePleistocene. Journal of Anthropological Archaeology 27 (4), 399e417.

Sitlivy, V., Zieba, A., 2006. Eastern and Central Europe before 30 kyr BP: Mousterian,Levallois & Blade industries. In: Chabai, V., Richter, J., Uthmeier, Th (Eds.), KabaziII: the 70,000 Years Since the Last Interglacial, The Palaeolithic Sites of Crimea,2. Simferopol-Cologne, Shlyakh, pp. 361e419.

Sitlivy, V., Zieba, A., Sobczyk, K., Kolesnik, A., 2009. Lithic assemblages. In: Sitlivy, V.,Zieba, A., Sobczyk, K. (Eds.), Middle and Early Upper Palaeolithic of KrakowRegion. Ksiecia Jozefa, Musées Royaux d’Art et d’Histoire, Bruxelles, pp. 19e169.

Sitlivy, V., Chabai, V., Anghelinu, M., Uthmeier, T., Kels, H., Hilgers, A., Schmidt, C.,Nit"a, L., B"altean, I., Vesselsky, A., Hauck, T. The Aurignacian sequence atRomânesti-Dumbr"avita I site in Banat (Southwestern Romania). 40 years later,in preparation.

Soficaru, A., Dobos, A., Trinkaus, E., 2006. Early modern humans from the PesteraMuierii, Baia de Fier, Romania. Proceedings of the National Academy of Sciences103, 17196e17201.

Soficaru, A., Petrea, C., Dobos, A., Trinkaus, E., 2007. The human cranium from thePestera Cioclovina Uscat"a, Romania: context, age, taphonomy, morphology, andpaleopathology. Current Anthropology 48, 611e619.

Soriano, S., Villa, P., Wadley, L., 2007. Blade technology and tool forms in the MiddleStone Age of South Africa: the Howiesons Poort and post-Howiesons Poort atRose Cottage Cave. Journal of Archaeological Science 34, 681e703.

Steguweit, L., Cârciumaru, M., Anghelinu, M., Nit"a, L., 2009. Reframing the UpperPalaeolithic in the Bistrita Valley (Northeastern Romania). Quartär 56,139e157.

Stratan, I., 1965. Asezarea paleolitic"a de la Cosava. Revista Muzeelor II, 412.Stratan, I., 1970. Contributii la cunoasterea palaeoliticului din Banat. Tibiscus I, 7e18.Szmidt, C.C., Normand, C., Burr, G.S., Hodgins, G.W.L., LaMotta, S., 2010. AMS 14C

dating theProtoaurignacian/EarlyAurignacian of Isturitz, France. Implications forNeanderthal-modern human interaction and the timing of technical and culturalinnovations in Europe. Journal of Archaeological Science 37 (4), 758e768.

Teyssandier, N., 2003. Les débuts de l’Aurignacien en Europe. Discussion à partir dessites de Geissenklösterle, Willendorf II, Krems-Hundssteig et Bacho Kiro. Thèsede doctorat, Université de Paris X-Nanterre.

Teyssandier, N., 2006. Questioning the first Aurignacian: mono or multi culturalphenomenon during the formation of the Upper Paleolithic in Central Europeand the Balkans. Anthropologie, International Journal of the Science of ManXLIV (1), 9e29.

Teyssandier, N., 2007a. L’émergence du Paléolithique supérieur en Europe: muta-tions culturelles et rythmes d’évolution. Paléo 19, 367e390.

Teyssandier, N., 2007b. En route vers l`Ouest. Les débuts de l’Aurignacien en Europe.In: BAR International Series 1638. Oxford.

Teyssandier, N., 2008. Revolution or evolution: the emergence of the UpperPaleolithic in Europe. World Archaeology 40 (4), 493e519.

Teyssandier, N., Bon, F., Bordes, J.-G., 2010. Within projectile range. Some thoughtson the appearance of the Aurignacian in Europe. Journal of AnthropologicalResearch 66, 209e229.

Tixier, J., 1963. Typologie de l’Epipaléolithique du Maghreb. Mémoires du Centre deRecherches anthropologiques et préhistoriques et ethnographiques 2. AMG,Alger, Paris.

Trinkaus, E., Milot"a, S., Rodrigo, R., Mircea, G., Moldovan, O., 2003. Early modernhuman cranial remains from the Pestera cu Oase, Romania. Journal of HumanEvolution 45, 245e253.

Trinkaus, E., B"altean, I.C., Constantin, S., Gherase, M., Horoi, V., Milot"a, R.S.,Moldovan, O., Petrea, C., Quiles, J., Rodrigo, R., Rougier, H., Sarcina, L.,Soficaru, A., Zilhão, J., 2005. Asupra oamenilor moderni timpurii din Banat:Pestera cu Oase. Banatica 17, 9e27.

Trinkaus, E., Zilhão, J., Rougier, H., Rodrigo, R., Milota, S., Gherase, M., Sarcina, L.,Moldovan, O., B"altean, I., Codrea, V., Bailey, S.E., Franciscus, R.G., Ponce DeLéon, M., Zollikofer, C.P.E., 2006. The Pestera cu Oase and early modern humansin Southeastern Europe. In: Conard, N.J. (Ed.), When Neanderthals and ModernHumans Met. Kerns Verlag, Tübingen, pp. 145e164.

Trinkaus, E., Soficaru, A., Dobos, A., Constantin, S., Zilhão, J., Richards, M., 2009.Stable Isotope evidence for early modern human Diet in Southeastern Europe:Pestera cu Oase, Pestera Muierii and Pestera Cioclovina Uscat"a. Materiale siCercet"ari Arheologice, V, (N. S.), 5e14.

Tsanova, T., 2006. Les débuts du Paléolithique supérieur dans l’Est des Balkans.Réflexion à partir de l’étude taphonomique et tehcnoéconomique des ensem-bles lithiques des sites de Bacho Kiro (couche 11), Temnata (couches VI et 4) etKozarnika (niveau VII). Thèse de doctorat de l’Université Bordeaux 1.

Tsanova, T., Zwyns, N., Eizenberg, L., Teyssandier, N., Le Brun-Ricalens, F., Otte, M. Leplus petit dénominateur commun: réflexion sur la variabilité des ensembleslamellaires du Paléolithique supérieur ancien d’Eurasie. Un bilan autour desexemples de Kozarnika (Est des Balkans) et Yafteh (Zagros central). L’An-thropologie, in press.

Tuffreau, A., Boroneant, V., Goval, E., Boroneant, A., Dobos, A., Lefevre, A.,Popescu, G., 2006. L’utilisation du quartz dans l’industrie moustérienne deZ"abrani (Banat, Roumanie). WS 15- Technological analysis on quartziteexploitation. Congrès U.I.S.P.P., Lisbonne.

Tuffreau, A., Boroneant, V., Goval, E., Lefevre, B., Boroneant, A., Dobos, A.,Popescu, G., 2007. Le gisement paléolithique moyen de Z"abrani (Départementd’Arad). Materiale si Cercet"ari Arheologice (NS) III, 5e18.

Tuffreau, A., Chirica, V., B"alescu, S., Haesaerts, P., 2009. Nouvelles recherches sur legisement paléolithique de Mitoc-Valea Izvorului (dép. de Botosani), Fouilles2003e2004. Materiale si Cercet"ari Arheologice (NS) V, 21e31.

Usik, V.I., 2008. The Upper Paleolithic of Transcarpathia: chronology and culturalaffinity of Beregovo I Aurignacian. In: Materials and Investigations of Carpa-thian and Volyn Archeology, vol. 12, pp. 49e67 (in Russian).

Yevtushenko, A.I., 1998. The industries of the Eastern Micoquian: some approachesto typological variability. In: Otte, M. (Ed.), Préhistoire d’Anatolie, Genèse dedeux mondes I. ERAUL 85, Liège, pp. 113e123.

Zilhão, J., 2006. Neandertals and Moderns mixed, and it matters. EvolutionaryAnthropology 15, 183e195.

Zilhão, J., 2007. The emergence of ornaments and art: an archeological perspective onthe origins of ‘behavioral modernity’. Journal of Archaeological Research 15,1e54.

Zilhão, J., Trinkaus, E., Constantin, S., Milota, S., Gherase, M., Sarcina, L., Danciu, A.,Rougier, H., Quilès, J., Rodrigo, R., 2007. The Pestera cu Oase people, Europe’searliest modern humans. In: Mellars, P., Boyle, K., Bar-Yosef, O., Stringer, C.(Eds.), Rethinking the Human Revolution, New Behavioural and BiologicalPerspectives on the Origin and Dispersal of Modern Humans. McDonald Insti-tute, Cambridge, pp. 249e262.

Zieba, A., Sitlivy, V., Sobczyk, K., 2008. The lithic assemblages. In: Sitlivy, V.,Zieba, A., Sobczyk, K. (Eds.), Middle and Early Upper Palaeolithic of KrakowRegion. Piekary IIa. Musées Royaux d’Art et d’Histoire, Bruxelles,pp. 57e148.

