Navigation – Plan du site

AccueilNuméros63The beginnings of the Aurignacian...

The beginnings of the Aurignacian in a European framework: where do we stand?

Abridged version
Nicolas Teyssandier
Cet article est une traduction de :
Les débuts de l’Aurignacien dans leur cadre européen : où en est-on ? [fr]


Résumé. Cette contribution est une synthèse sur les premières phases de l’Aurignacien à l’échelle européenne. Elle se focalise d’abord sur le Sud-Ouest français afin d’établir un modèle évolutif en deux phases successives : le Protoaurignacien et l’Aurignacien ancien. Une fois défini sur des bases archéologiques en compilant les données des industries lithiques et en matières osseuses, mais également des parures et des représentations graphiques quand ces éléments sont disponibles, le modèle établi est confronté à d’autres macro-régions d’Europe afin d’envisager sa validité pan-européenne. Cet article suggère que la succession Protoaurignacien/Aurignacien ancien est valide à l’échelle continentale et propose ensuite des pistes de recherche pour interpréter cette dernière en termes de dynamique de peuplement en intégrant au raisonnement les données paléoanthropologiques et paléogénomiques ainsi que des éléments relatifs aux industries dites de « transition » du Paléolithique moyen au Paléolithique récent qui précèdent strictement l’émergence du Protoaurignacien à l’échelle continentale.

Haut de page

Texte intégral


1The origin and global dispersal of Homo sapiens are central themes in international debates on prehistory and human evolution. This conquering human lineage, which appeared in Africa nearly 300,000 years ago (Hublin et al. 2017), dispersed outside the original cradle, gradually replacing all the preceding so-called archaic species with various forms. These events took a particular turn between 50,000 and 40,000 years before the present, as several biological populations were last observed on the Eurasian continent during that period.

2In Europe, south-western France is a particularly propitious terrain of study, as it comprises the most abundant and best-studied archaeological sequences. This owes as much to the wealth of documentation in the region as to its major role in the development of the discipline of prehistory. Indeed, the evolutionary frameworks of the Late Palaeolithic were constructed at the beginning of the twentieth century in the Aquitaine Basin, based on the succession of two major technocomplexes: the Châtelperronian and the Aurignacian.

3More recently, major methodological developments have revised the Aurignacian of this region, and research no longer focuses solely on tool morphology, but also on the processes by which they were made and used, through techno-economic analyses of stone and bone equipment. Even today, the Aurignacian of the Aquitaine Basin is an essential reference for any attempt at comparisons, and the geographical area of reference for any attempt to reposition Aurignacian assemblages in an evolutionary sequence.

4This state of knowledge, which led to a focus on this study region, has given rise to a number of reactions in recent years. The main question that has arisen is whether the classic evolutionary model established for the Aquitaine Basin and its margins is still valid. For some, this model of diachronic evolution is solid and the chrono-typo-technological evolution established in the southwest generally follows the same path elsewhere (Teyssandier 2007, Zilhão 2007, 2013, Teyssandier et al. 2010). For others, the validity of this model on a supra-regional scale is questionable and requires a more detailed definition of regional Aurignacian signatures (Falcucci et al. 2017, 2020, Tafelmaier 2017, Bataille et al. 2018). According to the latter authors, the identified phases bear no chronological value and instead constitute functional or adaptive facies.

5With this in mind, we first present an updated summary of the archaeological signatures of the early Aurignacian period in south-western France. The supra-regional validity of the established evolutionary scheme is then compared with recent data obtained in other regional areas, in France, the Iberian Peninsula, Italy and the Danube corridor. Finally, the beginnings of the Aurignacian are examined and compared with the so-called transitional industries from the Middle Palaeolithic to the Late Palaeolithic, in order to present an overall view of the processes at play at this pivotal time when the first anatomically modern humans emerged in Europe and the last Neanderthal societies disappeared.

Geographical distribution, stratigraphy and chronology

6There are many Aurignacian sites in south-western France, mainly in the north of the Aquitaine Basin and the foothills of the Pyrenees. The vast majority are caves or rock shelters, although the open-air sites in the Bergerac and Serres regions are notable exceptions.

7Clear stratigraphic recurrences have been observed and, based on the principle of archaeostratigraphy, it is possible to establish a relative chronology of the various successive phases of the Aurignacian. Two phases can be distinguished for the early Aurignacian period: the Protoaurignacian (PA) and the Early Aurignacian (EA). In all cases where they are observed within the same stratigraphic sequence, their order is constant, with the PA always preceding the EA (fig. 2). There are no known cases of interstratification between these two phases, so we can confidently state that the PA strictly predates the EA. In some of these same deposits, it is also possible to observe the stratigraphic position of the PA in relation to the local preceding technocomplexes. At Le Piage, Brassempouy and Gatzarria, the Châtelperronian systematically precedes the PA, which leads us to conclude that in south-western France, the situation is clear and unambiguous, with a Châtelperronian/PA/EA stratigraphic succession. This archaeostratigraphic sequence is also confirmed by absolute chronology data. According to Bayesian modelling incorporating stratigraphic constraints, the PA is dated between 41,700 and 40,500 cal BP and the EA between 40,500 and 38,500 cal BP, i.e., during the very cold phase of the Heinrich stadial 4.

Fig. 2 – Archaeostratigraphic sequences and calibrated 14C dates of the Châtelperronian (salmon rectangles), Protoaurignacian (light green rectangles) and Early Aurignacian (dark green rectangles) in southwest France.

Fig. 2 – Archaeostratigraphic sequences and calibrated 14C dates of the Châtelperronian (salmon rectangles), Protoaurignacian (light green rectangles) and Early Aurignacian (dark green rectangles) in southwest France.

The Chatelperronian/PA/EA sequence is constant and is repeated at all sites with a relatively complete archaeostratigraphy of the Middle Palaeolithic to the Late Palaeolithic transition. Due to the low number of multi-stratified sites, the Cottés sequence, located in the southwest of the Paris Basin, has been added.

The PA: archaeological definition

Lithic industry

8Large bladelets with a slightly curved or rectilinear profile are a consistent feature of the series attributed to the PA. These bladelets are frequently retouched into Dufour bladelets by inverse or alternating retouch, a Dufour sub-type. In all cases, the location of inverse retouch is almost systematically on the right edge (Falcucci et al. 2016). These bladelets are also commonly retouched and pointed by fine direct retouch, in which case they are typologically similar to Krems points.

9Bladelets are also produced in similar ways, making it possible to define common standards. Here, a remarkable feature is the continuity of the production of blades and then bladelets, frequently on pyramidal-shaped cores. Independent parallel bladelet productions most often correspond to the exploitation of the edges of thick flakes or blades. During the PA, blades are usually light and only marginally retouched.

Osseous industry

10In the PA, osseous industry assemblages are scarce, making interpretation difficult.

Elements of adornment

11The elements of adornment found disparately at south-western PA sites bear symbolic connotations. Here again, the most persuasive evidence comes from Isturitz, with around 15 perforated shells and a soft stone pendant interpreted as a feminine, zoomorphic or phallic representation.

2.4. The author of the PA in south-western France

12The only recognised human remains are three perinatal remains discovered in layers K and J of the Piage site during the Champagne and Espitalié excavations.

EA: archaeological definition

13The EA has been identified at a much larger number of sites than the PA, and its presence has also been confirmed in more diverse types of sites; long sequences in caves or shelters and open-air sites in several regions.

Lithic industry

14EA lithic productions are very similar in the whole of south-western France. The main characteristic is the almost systematic dissociation of blade and bladelet debitage: on the one hand, blades are obtained from prismatic cores and, on the other, bladelets are obtained from the famous carinated cores (“end scrapers”).

15Blades are usually wide and thick, obtained by soft hammer direct percussion using a tangential gesture that hooks the edge of the striking platform. Butts are generally facetted, and sometimes even spurred. Debitage is strictly unipolar and core shaping is usually quite minimal. Blades vary in size, but the most commonly sought-after blanks for domestic tools were sturdy blades with a regularly curved profile, which were transformed into end scrapers, laterally retouched or even strangled blades.

16Bladelet production is mainly carried out on 2-to-3-cm-wide carinated “end scrapers” made from laminar production waste. The obtained bladelets are short, fairly wide and generally curved in profile. They are rarely, if ever, intensely transformed, and when they are, retouch is discreet and inverse.

Osseous industries

17The rich and diversified bone industry of the EA is characterised first and foremost by the individualisation of the operating chains and types of activity according to the raw materials used (e.g., Tartar et al. 2006). As with lithic production, production chains for weapons are clearly dissociated from the other main functional categories. This dissociation is accentuated by a clear differentiation in raw material management: reindeer antler is used to make hunting weapons, ivory is used primarily for adornment and bone is used for the main categories of domestic objects. The EA bone equipment includes the famous split-based points used as chronological markers from very early on.


18Elements of adornment also developed considerably with far more categories than in previous periods. For the first time, the corpus is abundant with multiple materials: shells, as was already the case in the PA, varieties of soft stone, ivory, bone, reindeer antler, animal teeth and even human teeth. One type of bead is particularly common and abundant, especially in sites in the Périgord and the pre-Pyrenees: basket beads, issued from a standardised manufacturing process on a regional scale.

Graphic representations

19Figurative graphic representations also appeared for the first time in the region during the EA, adding to the innovative character of this period. Thirteen Périgord rock shelters have yielded fragments of vaults or semi-mobile blocks bearing figures with two colours, black, engraved or pecked animal outlines, as well as basins and rings (Bourrillon and White 2015). Recent discoveries in stratigraphic context confirm the positioning of these limestone blocks in the EA, as is the case with the block engraved with an aurochs and a series of 72 dots at Blanchard (Bourrillon et al. 2018).

The author of the EA

20The rare human remains are systematically isolated and scattered across the habitat, as is the case at Brassempouy, La Ferrassie, La Combe and Isturitz. The Grotte des Hyènes at Brassempouy yielded the largest series, with a mandible fragment, a cranial fragment, two distal phalanges of the hand and a dental series, in a reliable stratigraphic context. Non-metric characters on the teeth confirm the attribution of these fossils to Homo sapiens (Bailey and Hublin 2005), as do certain remains from layer 3 of the Quina-aval rock shelter. There is therefore every reason to attribute the EA to anatomically modern humans.


Is the Early Aurignacian sequence from south-western France valid elsewhere in Western Europe?

21Is the Early Aurignacian sequence structured around the PA/EA succession valid elsewhere or specific to south-western France?

The Paris Basin

22The Cottés sequence (Vienne) begins with a still poorly understood Mousterian in SU 08 and continues with an early Late Palaeolithic assemblage in the form of the Châtelperronian in SU 06. The overlying stratigraphic unit 04 is subdivided into two groups separated by a sterile level: SU 04 inf. attributed to the PA and SU 04 sup. to the EA (Roussel and Soressi 2013). The evolution of the lithic equipment here clearly reflects this PA/EA succession, with large rectilinear bladelets in lower SU 04 and smaller curved bladelets and more robust laminar supports with the presence of Aurignacian blades in upper SU 04.

The north of the Iberian Peninsula

23The emblematic and clearest site is the Labeko Koba Cave in Spanish Basque Country (Arrizabalaga et al. 2003). Above a scant Châtelperronian level, four superimposed layers make up the Aurignacian sequence and are attributed respectively to the PA (layer VII) and the EA (layers VI, V and IV). The dates obtained with ultrafiltration are very consistent and in perfect agreement with those obtained in the north of Basque Country at Isturitz (Wood et al. 2014). The chronological and typo-technological characteristics of the Labeko Koba sequence replicate those obtained in south-western France, giving it extra-regional value.

Southeast France

24Data have been acquired over the last 30 years with, in particular, the very clear definition of a PA at the base of local Late Palaeolithic sequences. This is the case, for example, at Esquicho Grapaou and La Laouza in the Gard, Mandrin in the Drôme and the Observatoire Cave in Monaco. In all cases, large Dufour bladelets of the Dufour sub-type are found in lithic assemblages where bladelets are predominant, with examples of continuous blade/bladelet production but also autonomous bladelet debitage, particularly at Mandrin and Observatoire. In all cases, the classic EA tools - carinated “end scrapers” or robust blades with lateral retouch - are absent.

The Italian Peninsula

25The Fumane site is the reference sequence for the end of the Middle Palaeolithic and the beginning of the Late Palaeolithic in Italy. The Upper Pleistocene sequence covers the Mousterian, Uluzzian, PA and an Aurignacian sensu lato that has yet to be analysed in detail. At the base of the Aurignacian sequence, level A2, dated between 41,200 and 40,400 cal BP, contains a rich assemblage typical of the PA, with bladelets of various sizes. There are many large rectilinear bladelets, as well as bladelets with converging edges obtained from specific pyramidal cores, frequent in the PA of south-western France. Another typical PA feature is the large number of bladelets interspersed with blade debitage. While there is no doubt that the A2 level can be attributed to the PA, the upper part of the Aurignacian sequence is less clear. There is an increase in the number of carinated cores, a reduction in the size of bladelets and a decrease in the frequency of retouched bladelets. Despite these features, the entire Aurignacian sequence has been attributed by some authors to the PA (Falcucci et al. 2020), even though two split-based points, typical of the EA, are present in A1 and D3b alpha, in a comparable chronology to the development of the EA in the southwest. It seems more pertinent to emphasise the trend towards shorter bladelets and the development of keeled technology, while at the same time expressing a reservation about the validity of the entire Aurignacian archaeosequence from Fumane, where no archaeostratigraphic projection or positioning of connected and/or refitted pieces has been published. As a result, it is plausible that separate stopovers of Early Aurignacian groups form an occupation palimpsest, but this was not recognised during excavation. This is also consistent with the wider archaeological context of the Italian Peninsula, where associations between split-based points and typical EA lithic industry have been observed.

Central Europe

26Two Central European sites are central to the debate on the existence of a very early Aurignacian in Central Europe: Geissenklösterle in Germany and Willendorf 2 in Austria. Below, we summarise the arguments compiled to assess the emergence of the Aurignacian in Central Europe two to three millennia before it appeared in Western Europe.


27The Aurignacian sequence in this cave comprises six archaeological levels (base to summit IIIb, IIIa, III, IId, IIb and IIa), defined on the basis of geological observations combined with the horizontal and vertical distribution of artefacts. Following several extensive refitting campaigns, J. Hahn (1988) and then ourselves deduced that there were two main archaeological groups: AH III, comprising layers IIIb to IId, and AH II, comprising layers IIb and IIa. It should be pointed out, however, that the Aurignacian sequence clearly corresponds to a palimpsest of occupations, mixing human occupations and occupations by cave bears, the latter being the main faunal taxon in AH III. Based on Bayesian modelling of the 14C dates, T. Higham et al. (2012) consider phase AH III to be archaeologically valid, and use it to show that the EA at Geissenklösterle predates similar cultural manifestations in Western Europe by almost two millennia.

28However, it turns out that this archaeological complex underwent major post-depositional disturbances, with refits containing elements scattered throughout the Aurignacian sequence (levels IIIb to Iia; fig. 54). Nevertheless, an analysis of the horizontal and vertical distribution of the remains and the position of certain hearth structures show that the abundant lithic remains associated with a hearth in layer IIIa are clearly related and display typical EA technological characteristics. As a result, dates obtained from samples from IIIa displaying traces of human activity should be favoured over those from non-hunted species or samples from secondary levels linked to the dismantling of the original IIIa layer (in the case of IIIb, III or IId). Only two measurements meet these conditions and indicate that the oldest EA at Geissenklösterle would be between 42 and 39,500 cal BP, which differs from the conclusions of Higham et al. (2012), which place the start of AH III between 42,940 and 42,180 cal BP. Thus, Bayesian modelling in a secondarily reconstructed sequence where it is not possible to control the actual association between dated faunal samples and the associated archaeological context is meaningless (e.g., Discamps et al. 2015). In our view, the oldest Aurignacian from Geissenklösterle is chronologically broadly contemporaneous with technically comparable finds from layer 14 of the Pataud rock shelter, with an EA without split-based points dated between 41,500 and 39,000 cal BP. It is followed locally in AH II by an EA associated with split-based points dated between 40,000 and 38,000 cal BP, then by a recent Aurignacian with twisted bladelets dating from around 38,000-36,500 cal BP and providing a convincing cultural context for the animal statuettes, ivory anthropomorph and flutes.

Fig. 54 – Archaeostratigraphic distribution of refits from Geissenklösterle within 20 groups of refits.

Fig. 54 – Archaeostratigraphic distribution of refits from Geissenklösterle within 20 groups of refits.

Note that, in many cases, the finds are scattered across five or six archaeological layers of the Aurignacian sequence identified during excavation.

After Teyssandier 2007a.

Willendorf II: an “Early” Aurignacian, not as early as some say

29Nigst et al. (2014) define the first Aurignacian at Willendorf as synonymous with the first Homo sapiens in the region. Following new excavations undertaken at the site, they conclude that the EA developed as early as 43,500 cal BP, several millennia before it developed in Western Europe.

30We disagree with these conclusions and the data used to reach them, and thus reviewed all the data published since the early twentieth century on this major site. Following these analyses, we categorically challenge the association between samples dated to 43,500 cal BP and Willendorf Layer 3 attributed to the EA (Teyssandier and Zilhão 2018).

31Layer 3 is included in the so-called C8-3 sedimentary lens for which no radiocarbon age has been obtained. The age attributed to layer 3 is in fact derived from terminii obtained from dating the surrounding geological horizons. Furthermore, this superposition with the surrounding geological horizons is not the result of observation but of inference, i.e., this superposition is not observed anywhere in situ but is the result of observations made in different areas of the site.

32The archaeological content of layer 3 comes mainly from excavations carried out at Willendorf at the very beginning of the twentieth century. Nigst et al. (2014) therefore sought to recontextualise this previously excavated level in the light of observations from their new excavation. Our observations show that the stratigraphic interpretation and correlations of Nigst et al. leave two fundamental questions unanswered: 1) are the charcoals dating the horizon overlying layer 3 and thus providing a terminus ante quem really in situ? 2) Do the knapped stones from the early twentieth-century excavations, which make up the bulk of the collection in layer 3, all come from lens C8-3 as defined during recent excavations? A re-reading of all the stratigraphic data gathered since the beginning of the last century does not suggest so. As a result, we have been able to conclude that the archaeological finds in layer 3 from the excavations conducted at the beginning of the twentieth century are stratigraphically and archaeologically heterogeneous.

33Furthermore, the attribution of the entire archaeological content of layer 3 to the EA is problematic. It is certainly clear that this complex contains diagnostic finds from the Aurignacian in general, and from the EA in particular. But this is not the case for all the pieces in this small collection of just 48 elements. Two scrapers and 12 laterally retouched blanks, which may well correspond to scraper fragments, were also identified. Yet these types of objects are generally rare, if not absent, in EA contexts. In addition, data from the underlying archaeological level, layer 2, show that the Willendorf finds are clearly affected by post-depositional problems. These include the presence of a blade detached with a soft organic striker in a “pre-Aurignacian” assemblage, and a bladelet core with a refitting bladelet, which itself comes from the overlying layer 5, attributed to the Early Gravettian (Moreau 2012). There is thus a strong possibility that the archaeological assemblage in layer 3 contains a mixture of undoubtedly Aurignacian finds and others from different technocomplexes.

34Moreover, during their new excavation, Nigst et al. (2014) discovered a small assemblage which they attribute to layer 3 of the earlier excavations and which they unequivocally relate to the EA, thus reinforcing their diagnoses. In our view, there is no empirical basis for this conclusion. The assemblage in question only contains 32 lithics, none of which, in our opinion, is diagnostic of the Aurignacian, even in the broadest sense of the term.

35In short, it appears that the relationships established by Nigst et al. (2014) between finds from their excavations and the early twentieth-century excavations which, it should be remembered, contain the only diagnostic EA objects, are rather tenuous. Consequently, it is improbable and erroneous to conclude that an EA at Willendorf precedes all other similar cultural manifestations in Europe by several millennia. We also show that other radiocarbon dates for a sub-unit of Layer 3, around 39,000 cal BP, are perfectly consistent with certain diagnostic EA objects and with all known European contexts. There is therefore no proven temporal shift or chronological antecedent for the presence of modern humans in Central Europe.

On the significance of the early Aurignacian period in terms of settlement dynamics

36Based on the available data, it is reasonable to associate the PA and EA with anatomically modern humans (Benazzi et al. 2015, Hublin 2015). As far as Eurasia is concerned, in this chronological context at the beginning of the Aurignacian, i.e., between around 42,000 and 37,000 BC, we know that biologically modern fossils systematically show variable proportions of genetic introgression from an archaic human type, in this case Neanderthals. On the basis of the complete genome obtained from the femoral diaphysis of Ust Ishim in western Siberia, with a modern morphology, the Neanderthal gene flow has been dated back 230 to 400 generations before the life of this individual, i.e., between 50 and 60,000 years before the present (Fu et al. 2014). Very different data have been obtained on fossils of relatively similar ages from Zlatý Kůň (Prüfer et al. 2021) and Peştera cu Oase 1 (e.g., the mandible; Fu et al. 2015). For Zlatý Kůň in the Czech Republic, direct 14C dates are problematic due to heavy contamination of the specimen with an animal-based consolidating glue, and the oldest age obtained after hydroxyproline treatment dates back to 34,000 cal BP. However, this is considered to be a minimum age, and the alleged antiquity of the lineage represented by this fossil, based on the length of the Neanderthal DNA segments of its nuclear genome, must in fact be more than 45,000 years before the present (Prüfer et al. 2021). For Zlatý Kůň, the last interbreeding with Neanderthal in the family tree of this modern human dates back 70 to 80 generations before him. For Peştera cu Oase 1, we would be even closer to one of the episodes of interbreeding between the two human lineages since this fossil would have a higher proportion of Neanderthal genes than all the other specimens, between 6 and 9%, and the Neanderthal/Homo sapiens mix would have taken place six generations before the birth of Peştera cu Oase 1 (Fu et al. 2015). This indicates that both population groups inhabited Central Europe around 40,000 cal BP, which is a minimum age for this fossil given that its dating has not been subjected to hydroxyproline treatment. In this way, there may well have been a flow/introgression of Neanderthal genes into biologically modern groups after they left Africa.

37From a palaeoanthropological point of view, it had long been pointed out, before the results of palaeogenomics, that biologically modern humans from the Late Palaeolithic combined mixed modern and archaic features, suggesting interbreeding between sapiens and Neanderthals. We now know, and it was logical to expect, that the closer we get to introgression with Neanderthals, the greater the proportion of Neanderthal genes in the sapiens genome, ranging from an average of 5.70–4.30% to 2.20–1.10% today (Fu et al. 2016).

38In terms of settlement dynamics for the early Eurasian Late Palaeolithic period, the notion of a single wave of settlement from Africa via the Near East no longer seems relevant and the situation is far more complex than was thought at the time of the not so distant publications of simple arrow maps (e.g., Mellars 2004). In fact, the study by Fu et al. (2016) shows that the analysed individuals, of early Late Palaeolithic age, are all identified within “clusters” where they are alone and isolated, unlike clusters grouping together several human fossils of recent Aurignacian or Gravettian age. In this way, whether we are talking about Ust Ishim, Oase 1, Kostienki 14 or Goyet, dated between 45 and 35,000 BP, and therefore all included within an earlier or near-contemporaneous chronological interval of the early phases of the Aurignacian or certain transitional industries, each belongs to a distinct lineage and therefore shows no affinities. Furthermore, significant differences are observed between these four fossils. The Kostenki and Goyet individuals, dated to around 37,000 and 35,000 cal BP respectively, are related in that they share more alleles with present-day Europeans than with East Asian populations (Fu et al. 2016). On the other hand, individuals from before 40,000 cal BP (Ust Ishim dated to around 45,000 cal BP; Fu et al. 2014; Oase 1) show no genetic affinity with later European populations (Fu et al. 2014), suggesting that they correspond to short-lived modern migrations that died out without leaving any descendants.

39This is a major result for the chronological context in question: the first Homo sapiens potentially associated with the Middle Palaeolithic/Late Palaeolithic transition period left no subsequent traces and did not contribute genetically to the current gene pool, whereas from 37,000 cal BP onwards, and probably even earlier if the Oase fossils were re-dated using hydroxyproline, all the analysed individuals are ancestors of present-day Europeans. We can therefore postulate that pre-Aurignacian modern humans, who were responsible for transitional industries, interbred with Neanderthals and left no descendants, a situation that would explain why these technical traditions were abandoned from the Proto-Aurignacian onwards. On the other hand, from the PA and EA onwards, modern populations would have left their genetic imprint in later populations, which would account for a certain archaeological continuity within the Aurignacian phylum.

40Another recently analysed individual, from the Bacho Kiro Cave in Bulgaria and dated between 45 and 42,500 cal BP, is thought to have interbred with Neanderthal less than six generations earlier (a similar “situation” to that of Oase 1). Palaeogenomic results indicate that he would have contributed, not to western Eurasian populations, but to populations found today in East Asia and the Americas (Hajdinjak et al. 2021). In other words, it is becoming increasingly clear that between roughly 47/45,000 and 42/40,000 cal BP, the rare anatomically modern individuals analysed by palaeogenomics indicate the existence of several populations on the European continent that did not pass on their genetic heritage to subsequent European Late Palaeolithic populations. We could therefore assume that there was a major change in the settlement of Europe at the end of the early Aurignacian phases, a hypothesis that some interpret in the light of the Campanian Ignimbrite (CI) volcanic mega-irruption, dated to around 39,000 cal BP.

41Finally, we now realise that the chronological resolution of 14C measurements for the early Late Palaeolithic is no longer sufficient, given certain data provided by palaeogenomics, since we are talking about human groups (and counting the number of generations between a fossil and an episode of interbreeding) that can be more accurately determined as being, in some cases, very close to the contact point between the last Neanderthals and the first sapiens. We are talking here about individuals such as Bacho Kiro or Oase, who had a very close Neanderthal ancestor in their family tree. Differentiation should therefore be based more on cultural factors than on a simplistic biological dichotomy, with some individuals being classified as Neanderthals and others as sapiens.

42Ultimately, in the light of the most recent palaeobiological data, there is no contradiction in seeing the PA as the manifestation of pioneering anatomically modern human groups in the territories where they settled. Indeed, this does not presuppose that the PA was the first cultural manifestation of Homo sapiens on European territory, since the latest palaeogenomic data show that between 45,000 and 40,000 cal BP, biologically modern humans populated Europe by interbreeding with Neanderthals, without leaving any subsequent traces of their descendants.


43Following this overview of the beginnings of the Aurignacian, initially centred on south-western France and then extended to a large part of Europe, our results point to the archaeostratigraphic validity of the Aquitaine model and an evolutionary structuring into two distinct phases, the PA preceding the EA. At the scale of southern France, we propose that the PA represents the arrival of pioneering populations, while the EA corresponds to a phase of territorial anchorage with a clear regionalisation, particularly evident in the types of adornment produced by these populations. This hypothesis does not imply that the authors of the PA were the first modern humans to arrive on the European continent and then in the southern half of France. Indeed, it is thought that earlier sapiens populations existed and, at least in the case of Bacho Kiro (Hublin et al. 2020), that they produced industries around 45,000 cal BP (Bachokirian), that some consider to belong to the so-called IUP movement. Palaeogenomic data indicate that these populations, which preceded the PA, left no trace in later genomes contemporary with the Aurignacian. These early pre-Aurignacian sapiens populations thus seem to have died out without any genomic influence on more recent groups in the same territories or, we assume, their cultural tradition. In this scenario, the Protoaurignacian would correspond to the first stage of a pioneering migration of anatomically modern populations to Europe.

44Archaeologically, the relationship between the first Aurignacian and the preceding transitional cultural complexes seems to have differed from one region to another. In the regions of Western Europe where it is well attested, the Chatelperronian always precedes the PA in stratigraphy, reflecting strict chronological anteriority. In agreement with several researchers, we see no reliable argument for the acculturation of Chatelperronians through contact with Protoaurignacians or early Aurignacians. Indeed, after a critical review of the Aurignacian assemblages potentially contemporaneous with the Chatelperronian, the beginning of the Aurignacian is clearly later than the Chatelperronian. In this sense, the Chatelperronian is indeed an independent development of populations currently consensually described as Neanderthals on the sole basis of the data from the Renne d’Arcy-sur-Cure Cave. We cannot resolve this debate at this stage, but we have nevertheless pointed out new biases in our understanding of the archaeosequence of Renne Cave, which make it impossible to confidently guarantee that all the Neanderthal human remains unearthed in the Chatelperronian layers of this site are undoubtedly associated with Chatelperronian cultural equipment. Only new discoveries will enable us to overcome this bias.

45From a technological point of view, the Chatelperronian appears to mark a clear break with the preceding Mousterian assemblages, in terms of both lithic and bone equipment and the wide range of items of adornment. If we only consider lithic production methods, the contrast between the last Neanderthals, with mostly discoid flake-based traditions, and the laminar technologies typical of the Upper Palaeolithic Chatelperronian is striking (Bachellerie 2011, Bordes et Teyssandier 2011, Roussel 2011). In this sense, we can propose that if the Chatelperronian is indeed the work of Neanderthals, then there was a clear break in learning and transmission modes compared to previous populations. If there was such a break in transmission methods, it seems to have been much more marked with the last Mousterians than with the first Aurignacians.

46The differentiation between the Early Aurignacian and other transitional facies is also marked. Nothing in the Bohunician of Central Europe, the Bachokirian of the Balkans, the Uluzzian of Italy or the LRJ of north-western Europe expresses any technological kinship with the Protoaurignacian. In this sense, the Protoaurignacian appears to have been intrusive on the European continent, and its origins must be sought elsewhere. In this vein, it has been proposed for some 15 years that the precursor of the AP could be found in the Early Ahmarian assemblages of the Near East (Bar-Yosef 2006, Mellars 2006, Teyssandier 2008, Zilhão 2013, Hublin 2015). These assemblages display unmistakable typological and technological similarities, in terms of blade and bladelet production and the presence of el Wad points, which are extremely similar to Krems points and large Dufour sub-type bladelets. In any case, whatever the outcome of these debates, which require a return to the primary data and the detailed analysis of European and Near Eastern assemblages using unified analytical frameworks, the fact remains that the group of industries with large retouched bladelets, into which the PA fits, covers immense geographical areas and transcends differentiated ecosystems at the very beginning of the Late Palaeolithic.

47The archaeological situation is undoubtedly different with the EA, which marks a probable period of demographic growth, given the frequency and intensity of exceptionally rich occupations in the north of Aquitaine, as is the case in particular in Dordogne, in the Vézère valley region. As well as demographic growth, the EA would seem to mark a phase of regionalisation for groups who then laid claim to territorial identities, as is clearly shown in types of body adornment, in particular (e.g., Vanhaeren and d’Errico 2006). The necessary ensuing reformulation of social relations would have led in parallel to an explosion of the quantity and diversification of conspicuous production, and probably also to the emergence of figurative graphic production, which continued to develop throughout the Late Palaeolithic. This would be the last stage of the Aurignacian “conquest”, as the dates of the Chauvet black paintings and the chronostratigraphic position of the ivory statuettes from the Swabian Jura seem to indicate.

Haut de page


Arrizabalaga A., Altuna J., Areso P., Elorza M., Garcia M., Iriarte M.J., Mariezkurenna K., Peman E., Mujika J., Tarrino A., Uriz A., Viera L., Straus L. G. 2003 – The Initial Upper Paleolithic in Northern Iberia: new evidence from Labeko Koba, Current Anthropology, 44 (3), p. 413–421, [online]

Bachellerie F. 2011 – Quelle unité pour le Châtelperronien ? Apport de l’analyse taphonomique et techno-économique des industries lithiques de trois gisements aquitains de plein air : le Basté, Bidart (Pyrénées-Atlantiques) et Canaule II (Dordogne), PhD thesis, Bordeaux, Université Bordeaux I.

Bailey S., Hublin J.-J. 2005 – Who made the early Aurignacian? A reconsideration of the Brassempouy dental remains, Bulletin et Mémoire de la Société d’Anthropologie de Paris, 17 (1–2), p. 115–121, [online]

Bar-Yosef O. 2006 – Neanderthals and Modern Humans: a different interpretation, in Conard N.J. (ed.), When Neanderthals and Modern humans met, Tübingen, Kerns, p. 467–482.

Bataille G., Tafelmaier Y., Weniger G.-C. 2018 – Living on the edge — A comparative approach for studying the beginning of the Aurignacian, Quaternary International, 474 (Part A), p. 3–29, [online]

Benazzi S., Slon V., Talamo S., Negrino F., Peresani M., Bailey S. E., Sawyer S., Panetta D., Vicino G., Starnini E., Mannino M. A., Salvadori P. A., Meyer M., Pääbo S., Hublin J.-J. 2015 – The makers of the Protoaurignacian and implications for Neandertal extinction, Science, 348 (6236), p. 793–796, [online]

Bordes J.-G., Teyssandier N. 2011 – The Upper Paleolithic nature of the Châtelperronian in South-Western France: archeostratigraphic and lithic evidence, Quaternary International, 246 (1–2), p. 382–388, [online]

Bourrillon R., White R. 2015 – Pratiques symboliques aurignaciennes en abri-sous-roche dans la vallée de la Vézère : à la recherche d’une identité ?, P@lethnologie, 7, p. 119-139, [online]

Bourrillon R., White R., Tartar E., Chiotti L., Mensan R., Clark A. E., Castel J.-C., Cretin C., Higham T., Morala A., Ranlett S., Sisk M. L., Devièse T., Comeskey D. J. 2018 – A new Aurignacian engraving from Abri Blanchard, France: Implications for understanding Aurignacian graphic expression in Western and Central Europe, Quaternary International, 494, p. 46–64, [online]

Discamps E., Gravina B., Teyssandier N. 2015 – In the eye of the beholder: contextual issues for Bayesian modelling at the Middle-to-Upper Palaeolithic transition, World Archaeology, 47 (4), p. 601–621, [online]

Falcucci A., Conard N. J. Peresani M. 2017 – A critical assessment of the Protoaurignacian lithic technology at Fumane Cave and its implications for the definition of the earliest Aurignacian, PLoS One, 12 (12), e0189241, [online]

Falcucci A., Conard N. J., Peresani M. 2020 – Breaking through the Aquitaine frame: a re-evaluation on the significance of regional variants during the Aurignacian as seen from a key record in Southern Europe, Journal of Anthropological Sciences, 98, p. 99–140, [online]

Fu Q., Li H., Moorjani P., Jay F., Slepchenko S. M., Johnson P. L. F., Petri A. A., Prüfer K., de Filippo C., Meyer M., Zwyns N., Salazar-Garcia D., Kuzmin Y. V., Keates S. G., Kosintsev P. A., Razhev D. I., Richards M. P., Peristov N. V., Lachmann M., Douka K., Higham T. F. G., Slatkin M., Hublin J.-J., Reich D., Kelso J., Viola T. B., Pääbo S. 2014 – The genome sequence of a 45,000 year old modern human from Western Siberia, Nature, 514, p. 445–449, [online]

Fu Q., Hajdinjak M., Moldovan O. T., Constantin S., Mallick S., Skolglund P., Patterson N., Rohland N., Lazaridis I., Nickel B., Viola B., Prüfer K., Meyer M., Kelso J., Reich D., Pääbo S. 2015 – An early modern human from Romania with a recent Neanderthal ancestor, Nature, 524, p. 216–219, [online]

Fu Q., Posth C., Hajdinjak M., Petr M., Mallick S., Fernandes D., Furtwängler A., Hakk W., Meyer M., Mittnik A., Nickel B., Peltzer A., Rohland N., Slon V., Talamo S., Lazaridis I., Lipson M., Mathieson I., Schiffels S., Skoglund P., Derevianko A. P., Drozdov N., Slavinsky V., Tsybankov A., Grifoni Cremonesi R., Mallegni F., Gély B., Vacca E., González Morales M. R., Straus L. G., Neugebauer-Maresch C., Teschler-Nicola M., Constantin S., Teodora Moldovan O., Benazzi S., Peresani M., Coppola D., Lari M., Ricci S., Ronchitelli A., Valentin F., Thevenet C., Wehrberger K., Grigorescu D., Rougier H., Crevecoeur I., Flas D., Semal P., Mannino M. A., Cupillard C., Bocherens H., Conard N. J., Harvati K., Moiseyev V., Drucker D. G., Svoboda J., Richards M. P., Caramelli D., Pinhasi R., Kelso J., Patterson N., Krause J., Pääbo S., Reich D. 2016 – The genetic history of Ice Age Europe, Nature, 534, p. 200–205, [online]

Hahn J. 1988 – Die Geissenklösterle-Höhle im Achtal bei Balubeuren I. Fundhorizonbildung und Besiedlung im Mittelpaläolithikum und im Aurignacien, Stuttgart, Konrad Theiss Verlag.

Hajdinjak M., Mafessoni F., Skov L., Vernot B., Hübner A., Fu Q., Essel E., Nagel S., Nickel B., Richter J., Moldovan O.T., Constantin S., Endarova E., Zahariev N., Spasov R., Welker F., Smith G. M., Sinet-Mathiot V., Paskulin L., Fewlass H., Talamo S., Rezek Z., Sirakova S., Sirakov N., McPherron S. P., Tsanova T., Hublin J.-J., Peter B. M., Meyer M., Skoglund P., Kelso J., Pääbo S. 2021 – Initial Upper Palaeolithic humans in Europe had recent Neanderthal ancestry, Nature, 592, p. 253–257, [online]

Higham T., Basell L., Jacobi R., Wood R. E., Ramsey C. B., Conard N. J. 2012 – Testing models for the beginnings of the Aurignacian and the advent of figurative art and music: The radiocarbon chronology of Geissenklösterle, Journal of Human Evolution, 62 (6), p. 664–676, [online]

Hublin J.-J. 2015 – The modern human colonization of western Eurasia: When and where?, Quaternary Science Reviews, 118, p. 194–210, [online]

Hublin J.-J., Ben-Ncer A., Bailey S., Freidline S. E., Neubauer S., Skinner M. M., Bergmann I., Le Cabec A., Benazzi S., Harvati K., Gunz P. 2017 – New fossils from Jebel Irhoud, Morocco and the pan-African origin of Homo sapiens, Nature, 546, p. 289–292, [online]

Hublin J.-J., Sirakov N., Aldeias V., Bailey S., Bard E., Delvigne V., Endarova E., Fagault Y., Fewlass H., Hajdinjak M., Kromer B., Krumov I., Merreiros J., Martisius N. L., Paskulin L., Sinet-Mathiot V., Meyer M., Paääbo S., Popov V., Rezek Z., Sirakova S., Skinner M. M., Smith G. M., Spasov R., Talamo S., Tuna T., Wacker L., Welker F., Wilcke A., Zahariev N., McPherron S. P., Tsanova T. 2020 – Initial Upper Palaeolithic Homo sapiens from Bacho Kiro cave, Bulgaria, Nature, 581, p. 299–302, [online]

Mellars P. 2004 – Neanderthals and the modern human colonization of Europe, Nature, 432, p. 461–465, [online]

Mellars P. 2006 – Archeology and the dispersal of modern humans in Europe: deconstructing the “Aurignacian”, Evolutionary Anthropology, 15 (1), p. 167–182, [online]

Moreau L. 2012 – Le Gravettien ancien d’Europe centrale revisité : mise au point et perspectives, L’Anthropologie,116 (5), p. 609-638, [online]

Nigst P. R., Haesaerts P., Damblon F., Frank-Fellner C., Mallol C., Viola B., Götzinger M., Niven L., Trnka G., Hublin J.-J. 2014 – Early modern human settlement of Europe north of the Alps occurred 43,500 years ago in a cold steppe-type environment, PNAS, 111 (40), p. 4394–14399, [online]

Prüfer K., Posth C., Yu H., Stoessel A., Spyrou M. A., Devièse T., Mattonai M., Ribechini E., Higham T., Veleminsky P., Bruzek J., Krause J. 2021 – A genome sequence from a modern human skull from Zlaty kun in Czechia, Nature Ecology & Evolution, 5, p. 820-825, [online]

Roussel M. 2011 – Normes et variations de la production lithique durant le Châtelperronien : la séquence de la Grande-Roche-de-la-Plématrie à Quinçay (Vienne), PhD thesis, université Paris Ouest Nanterre – La Défense.

Tafelmaier Y. 2017 – Technological variability at the beginning of the Aurignacian in northern Spain. Implications for the Proto- and early Aurignacian distinction, Mettmann, Wissenschaftliche Schriften des Neanderthal Museums.

Tartar E., Teyssandier N., Bon F., Liolios D. 2006 – Équipement de chasse, équipement domestique : une distinction efficace ? Réflexion sur la notion d’investissement technique dans les industries aurignaciennes, in Astruc L., Bon F., Léa V., Milcent P.-Y., Philibert S. (eds.), Normes Techniques et Pratiques Sociales. De la simplicité des outillages Pré- et Protohistoriques. XXVIe rencontres internationales d’Archéologie et d’Histoire d’Antibes, Antibes, Éditions APDCA, p. 109–117, [online]

Teyssandier N. 2007a – En route vers l’Ouest – Les débuts de l’Aurignacien en Europe, Oxford, Archeopress (BAR International Series 1638).

Teyssandier N. 2008 – Revolution or Evolution? The emergence of the Upper Paleolithic in Europe, World Archaeology, 40 (4), p. 493–519, [online]

Teyssandier N., Zilhão J. 2018 – Early, not old: on the entity and antiquity of the Aurignacian at Willendorf II, Journal of Paleolithic Archeology, 1 (1–2), p. 107–138, [online]

Teyssandier N., Bon F., Bordes J.-G. 2010 – Within projectile range. Some thoughts on the appearance of the Aurignacian in Europe, Journal of Anthropological Research, 66 (2), p. 209–229, [online]

Vanhaeren M., d’Errico F. 2006 – Aurignacian ethno-linguistic geography of Europe revealed by personal ornaments, Journal of Archaeological Science, 33 (8), p. 1105–1128, [online]

Wood R. E., Arrizabalaga A., Camps M., Fallon S., Iriarte-Chiapusso M. J., Jones R., Maroto J., Rasilla M. de la, Santamaria D., Soler J., Soler N., Villaluenga A., Higham T. 2014 – The chronology of the earliest Upper Palaeolithic in northern Iberia: New insights from L’Arbreda, Labeko Koba and La Viña, Journal of Human Evolution, 69 (1), p. 91–109, [online]

Zilhão J. 2007 – The Emergence of Ornaments and Art: An Archaeological Perspective on the Origins of “Behavioral Modernity”, Journal of Archaeological Research, 15, p. 1–54, [online]

Zilhão J. 2013 – Neandertal-modern human contact in western Eurasia: Issues of dating, taxonomy, and cultural associations, in Akazawa T., Nishiaki Y., Aoki K. (eds.), Dynamics of learning in Neanderthals and modern humans, vol. I: Cultural perspectives, Tokyo, Springer, p. 21–57, [online]

Haut de page

Table des illustrations

Titre Fig. 2 – Archaeostratigraphic sequences and calibrated 14C dates of the Châtelperronian (salmon rectangles), Protoaurignacian (light green rectangles) and Early Aurignacian (dark green rectangles) in southwest France.
Légende The Chatelperronian/PA/EA sequence is constant and is repeated at all sites with a relatively complete archaeostratigraphy of the Middle Palaeolithic to the Late Palaeolithic transition. Due to the low number of multi-stratified sites, the Cottés sequence, located in the southwest of the Paris Basin, has been added.
Fichier image/jpeg, 273k
Titre Fig. 54 – Archaeostratigraphic distribution of refits from Geissenklösterle within 20 groups of refits.
Légende Note that, in many cases, the finds are scattered across five or six archaeological layers of the Aurignacian sequence identified during excavation.
Crédits After Teyssandier 2007a.
Fichier image/jpeg, 135k
Haut de page

Pour citer cet article

Référence électronique

Nicolas Teyssandier, « The beginnings of the Aurignacian in a European framework: where do we stand? »Gallia Préhistoire [En ligne], 63 | 2023, mis en ligne le 17 janvier 2024, consulté le 12 avril 2024. URL : ; DOI :

Haut de page


Nicolas Teyssandier

CNRS, Laboratoire TRACES, UMR 5608, Université de Toulouse, Université Jean Jaurès, Maison de la Recherche

Haut de page

Droits d’auteur


Le texte seul est utilisable sous licence CC BY 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search