Magdatis project

In this paper we present all the AMS radiocarbon dates obtained for the 19 sites of the MAGDATIS project (69 results, including 4 human remains) and the approach we took to do this. This large-scale project involved a substantial multidisciplinary re-assessment of the Middle and Upper Magdalenian of the western Aquitaine/Pyrenean region of France. Our principal questions in this project regarding this period and region related to the renewal (i.e. expansion and disappearance) of the ungulate community and the timing of major techno-economical innovations. Dating sites was a key aspect of this project. Many of these sites were excavated a long time ago, often without the benefit of taphonomic assessment at the time. As a result, careful selection of artifacts for dating was key to obtaining accurate and reliable results and thus our methodology is emphasized in this paper. Among specific results obtained through this project are that (1) based on a re-analysis of the techno-typology of the lithic and osseous industries, in conjunction with the new dates done, the Magdatis project is proposing a new finer subdivision of the Middle and Upper Magdalenian into 4 phases (early and late sub-phases in each) using Bayesian modelling to determine boundaries; (2) that the Middle Magdalenian began earlier than previously suspected (ca 18 800 vs 18 300 cal BP); (3) that there is a hiatus in occupation in the northern Aquitaine during the Late Middle Magdalenian whereas occupation is continuous closer to the Pyrenees. (4) We will also be presenting briefly in this paper the theory behind, approach to, and results of, the dating of Magdalenian human remains done in this project.

The impact of environmental conditions and climate change on human societies has long been a key issue in hunter-gatherer studies, both ethnological and archeological. However, these approaches are mostly efficient for the most recent part of the prehistoric archeological record. For more ancient periods, the lower resolution of paleoenvironmental, archeological and chronological data, and their unequal availability depending on periods and regions, often hamper any precise assessment of the links between cultural changes and coeval environmental variations. The Magdatis project was an attempt at overcoming this difficulty by concentrating on a favorable case: the Middle and Upper Magdalenian, ca. 19-14 calendar millennia before present, one of the oldest periods where the resolution of both the environmental and the archeological data is high enough to insert them into a common, relatively precise chronological framework based on calibrated 14C dates. The aim of this project was therefore to evaluate to which extent the pattern of technical, economical and social evolution observed in the Magdalenian archeological record could be interpreted as a series of adaptive responses to environmental fluctuations, concentrating on a specific, high-resolution study area: the western Aquitaine basin. A large part of the results were published within the course of the project. The 9 papers in this Quaternary International issue add further to this body of data, and most of them provide a synthesis of the project’s results in a specific field: lithic and osseous technology, zooarcheology, palynology, radiocarbon dating, etc. The aim of this introduction is to provide a quick overview of some of the main results in these 32 papers.

Under the Magdatis project a new evolutionary model has been proposed for the Middle and Upper
Magdalenian based on a review of several lithic assemblages from southwest France. The Lower
Magdalenian is not addressed in this article. Single lithic assemblages and stratigraphic sequences have
been compared according to several parameters: the origin of the siliceous raw materials, the interconnection
between domestic tool production and hunting armatures, and the typo-technology of the
microliths. Alongside the new techno-economic data, the geographical distribution of certain practices
and particular artefacts in southwest France has led to a discussion of the cultural geography of the
different phases of the Magdalenian between 19,000 and 14,000 cal BP. The Early Middle Magdalenian
covers an extensive European cultural territory, demonstrating social interactions over vast distances
both in the circulation of raw materials and in the distribution of certain morphotypes of lithic armament.
The Late Middle Magdalenian appears to correspond to a phase of population contraction in which
cultural emulation is observed among lithic tool kits in the socio-economic value placed on large blades.
The Early Upper Magdalenian coincides with the renewed settlement of the north Aquitaine plains.
Among the technological innovations observed is the development of lithic points on small blades,
leading to a first geographical division of the southwest (Aquitaine Basin versus Pyrenees). Finally, the
Late Upper Magdalenian confirms this evolution of lithic equipment with new developments in lithic
armaments in parallel with some easing off of the technical constraints associated with blade production.
This period thus coincides with the progressive dilution of Magdalenian norms. With these four phases
of the Middle and Upper Magdalenian, it is possible to gain a clearer understanding of the sequence of
technical and economic changes observed in an environment in recomposition. At the same time,
population fluctuations over these five millennia can be associated with developing phenomena of
regionalism, particularly in the southwest of France.

The excavations of 1961, led by Robert Arambourou in layer 4 of Duruthy rockshelter at Sorde-l'Abbaye, Landes in France, revealed an exceptional assemblage in two adjacent squares. According to the lithic and the bone industry and the radiocarbon dating, this was attributable to a late phase of the Middle Magdalenian. This assemblage included three horse sculptures in sandstone, marl limestone, and ivory, and somebody ornaments and tools made of both mineral and organic materials. A concentration of 23 horse mandible fragments found near or in direct contact with the three horse sculptures comprised a selection of bones from old horses (MNIc ¼ 11) that had been subject to particular treatment. All the bones had scraping marks aimed at the removal of the connective tissue down to the alveolar bone, possibly to reveal the teeth entirely. On three pieces, saw marks, made with flint, and deep incisions were observed. One piece had deep grooves in an almond shape, and a mandible fragment had an engraved design on it that was unique among the geometrical designs of the Pastou group of sites. By drawing upon concepts taken from Descola's anthropology of nature, we have been able to consider the informative potential of these pieces regarding the ontology of their authors.

This paper presents an overview of unpublished or recent data obtained from taphonomic and zooarchaeological studies carried out on several Magdalenian osseous assemblages in the Gironde, most of which were performed or updated as part of the Magdatis project. These assemblages have been attributed to techno-complexes from the Lower Magdalenian (Saint-Germain-la-Rivi ere lower complex), Middle Magdalenian (Roc de Marcamps 2, Moulin-Neuf layer 2, Saint-Germain-la-Rivi ere upper complex and Fongaban) and Upper Magdalenian (Morin and row 24 of Faustin). They document, from a diachronic and synchronic standpoint, the relationships between Magdalenian hunter gatherers and their prey, principally ungulates. Our study focuses on the butchery of carcasses, from their transport to their disposal, through an investigation of the transportation methods, resources obtained, intensity of butchery, and techniques used, as well as any recurrences in the chaînes op eratoires. We also discuss the variations in the modes of consumption in relation to the size of the prey, the seasons of procurement, and the function of the sites. Irrespective of the size of the hunted games, the carcasses are incomplete. The post-cranial axial skeletons have been abandoned at the kill site. The limb and mandible bones are common, but differences are observed between the species exploited, reflecting either logistical constraints or nutritional contingencies. The resources used are diverse at the residential sites (skin, meat, marrow, tendons and hoofs). The marrow has been particularly intensely exploited regardless of the season; this advanced use does not however extend to the extraction of the fat. The results of this study indicate that, despite palaeoenvironmental and cultural changes, there were no real differences during the whole Magdalenian period in Gironde and that human groups remained faithful to the same traditions regarding the acquisition and processing of carcasses.

The Middle and Upper Magdalenian have yielded rich and diverse assemblages of osseous hunting technology, including projectile points (barbed or not), foreshafts, and spearthrowers. Within the Magdatis project, the evolution of this technology at French sites was reassessed by building an extensive dataset of 14C dates on Magdalenian hunting implements, selecting assemblages with good archeological homogeneity and taking into account questions of manufacture, hafting, and use. A finer four-stage chronology was established in coordination with lithic studies. A steady evolution was identified, marked by the intensification of antler exploitation and the increasing diversity and complexity of osseous armatures. This evolution is the continuation of a long-term trend that began in the Badegoulian and Lower Magdalenian and that culminated in the Late Middle/Early Upper Magdalenian, showing a special emphasis on composite projectile tips and evidence of the long-distance circulation of osseous materials and objects. A simplification process then takes place, with much of the technical innovation now centered on barbed implements to the detriment of the rest of the weapon kit, a trait that may announce the Azilian.

During the Tardiglacial, the significant changes in plant communities relating to climate changes were
responsible for faunal recompositions perceptible throughout Europe. In this article, by comparing all the
AMS radiocarbon dates obtained on reindeer bone and the faunal communities derived from bone assemblages
dated between ca. 19,000 cal BP and 11,700 cal BP, we examine the disappearance of reindeer
from the southwest of France. The new dating shows that the species disappeared slightly earlier in the
Pyrenees, at ca. 14,000 cal BP, than in the northern Aquitaine where reindeer remained until ca.
13,800 cal BP. In the southwest of France, the natural range of reindeer began to fragment very early,
from the Bølling period, and by the end of the GI-1e only residual reindeer populations remained in the
Dordogne. These results are consistent with those observed throughout the rest of France and
Switzerland, where reindeer also disappeared at the GI-1e/GI-1ca transition. Further north (Belgium,
Germany, and Denmark), the species found favourable conditions for its development throughout the
GS-1. In England, reindeer remained present until the beginning of the Holocene at very low latitudes
compared to what has been observed on the continent. These results clearly illustrate the gradual
withdrawal of reindeer towards the north and east of Europe and probably the local extinction of
reindeer in Britain.

Because of its rich record of Magdalenian sites, the Arudy Basin is a key case for the study of resettlement dynamics in the Pyrenean valleys after the Last Glacial Maximum. Interpretations are, however, hampered by the ancient age of many excavations. The recently acquired data from Laa 2 cave contribute to filling this gap, although the excavation of the Paleolithic layers was restricted to a very small surface. Layer C5 documents a very early – but still poorly defined – reoccupation of the cave (20000-19500 cal BP) that probably took place shortly after the deglaciation of the lower valley. Layers C4 to C2 (18500-15000 cal BP) document the evolution of the Middle and Upper Magdalenian in a cool, humid and open landscape, although evidence of reforestation appears in layer C2. The species hunted include a large proportion of mountain ungulates in the earlier layers, but horse and reindeer increase in proportion in the later layers. Birds and foxes are also exploited, and the question of the presence of dogs is raised, although not conclusively. Lithic sources seem to diversify in the Upper Magdalenian, along with an evolution in the dimensions of lithic armatures. The osseous industry also shows evidence of changes in the design of armatures (presence of forked implements) and includes a single evidence of portable art (an engraved scapula).

RÉSUMÉ L'une des questions qui anime le débat en archéologie paléolithique est de comprendre les relations qui existent entre l'évo-lution des techniques, des cultures, ou encore des territoires occupés par les groupes humains, et les changements du climat, de la faune et de la flore. Le projet Magdatis avait pour objectif d'apporter des éléments de réponse par l'étude d'un cas privilégié : le Magdalénien moyen et supérieur (vers 19 000-14 000 cal. BP) de l'ouest du Bassin aquitain. Cette région présente de forts contrastes écologiques qui en font un laboratoire idéal pour comparer les comportements des groupes humains dans des milieux différents à la fin des temps glaciaires. Pour cela, un bilan paléoenvironnemental détaillé, des études archéologiques multidisciplinaires et des datations radiocarbone ont été réalisés. Les résultats du projet ont significativement modifié notre vision de l'espace ouest-aquitain au Paléolithique, de ses contraintes environnementales et de son peuplement humain. La Gironde qui se présente alors comme une étendue steppique, semble avoir été désertée pendant deux millénaires, sans doute à cause de dégradations environnementales. Dans les Landes, un désert périglaciaire règne pendant toute la période. Cela explique sans doute la faiblesse de la présence humaine dans cet espace probablement très contraignant pour la vie. En revanche, les zones côtières étaient plus souvent fréquentées et plus intensément exploitées que ce que nous pensions jusqu'ici, attestant de l'existence d'une économie littorale. Dans les basses vallées pyrénéennes, la déglaciation, achevée précocement, libère des paysages ouverts qui sont rapidement occupés par les groupes humains. Ce sont ainsi les variations du milieu qui dessinent la carte des espaces attractifs et répulsifs. En revanche, les industries lithique et osseuse montrent l'existence de réseaux de diffusion dont l'extension n'est pas conditionnée par les obstacles naturels et relève de logiques sociales. La dynamique culturelle observée à cette échelle ne peut pas être reliée simplement aux changements environnementaux. ABSTRACT A CHICKEN-OR-EGG QUESTION. FEEDBACK ON THE MAGDATIS PROJECT " THE MAGDALENIAN FROM THE ATLANTIC FACADE FACING ENVIRONMENTAL CHANGES " One of the main issues in Paleolithic archeology is to understand the relationship that the evolution of techniques, cultures, and the territories occupied by human groups, has with changes in the climate, flora, and fauna. The aim of the Magdatis project was to provide elements of answer through the study of a key case: the Middle and Upper Magdalenian (ca. 19.000-14.000 cal. BP) from the Western Aquitaine basin. This region shows sharp ecological contrasts that make it an ideal context to study the behavior of human groups in different environments at the end of the last glaciation. A precise paleoenvironmental survey, multidisciplinary archaeological studies and radiocarbon dating were realized. The results of this project significantly changed our perception of the Western Aquitaine basin in the Paleolithic, of its environmental constraints and human settlements. The Gironde, then mostly a steppe-like region, was apparently deserted for two millennia, probably because of environmental degradation. A periglacial desert laid in the Landes during the whole period considered. This probably explains the small human presence in this region that was a constraining environment for life. Conversely, the coastal areas were more often visited and more intensively exploited than previously thought, attesting the existence of a seashore economy. In the Pyrenean lower valleys, the deglaciation was completed early and opened new landscapes that were rapidly settled by human groups. The environmental changes hence permit to draw a map of attractive and repulsive areas. The lithic and osseous industry, however, show the existence of distribution networks not conditioned by natural obstacles: their extent is instead linked to social dynamics. The cultural dynamics observed on this scale cannot be simply related to environmental changes.

Abstract: The identification of an early phase of the Magdalenian (Lower Magdalenian) between the Badegoulian and the Middle
Magdalenian has breathed new life into research concerning the emergence of Magdalenian societies in Western Europe. Moreover,
the initial phase of the Middle Magdalenian, or ‘Early Middle Magdalenian’ or EMM, plays a unique role for our understanding of the
Late Paleolithic given converging technological, economic and symbolic changes and innovations. Such developments are reflected,
for example, in individual burials (e.g. Saint-Germain-la-Rivière, Laugerie-Basse, Lafaye, Chancelade…), rockshelters with friezes of sculpted animals (e.g. Roc-aux-Sorciers, Chaire-à-Calvin…), development of human representations (engraved slabs, decorations on
bone and antler implements). Historically, this period has been characterised by different ‘facies’ defined by the presence of particular
objects: the Magdalenian with Lussac-Angles points, the Magdalenian with navettes or the Magdalenian with scalene bladelets.
The distribution of these three so-called entities of the EMM overlaps from the Poitou to the Périgord, and especially in the Gironde,
which is the subject of the present work. Building on the work of M. Lenoir, the re-evaluation of several assemblages North of the Sable
des Landes as part of the ‘Magdatis’ Project has revitalised questions concerning the characterisation of EMM lithic and osseous implements
as well as the regional demography of these groups. In addition to revising the ‘classic’ assemblages assigned to the Magdalenian
with navettes — Roc-de-Marcamps (original collections) — the analysis was expanded to included Saint-Germain-la-Rivière (Blanchard
Collection and Trécolle excavations), Moulin-Neuf (Lenoir excavations), Roc-de-Marcamps 1 and 2 (Lenoir excavations) and the Grotte
des Fées (Daleau Collection). While clearly biased by differing excavation methods, this database nevertheless allows varying techno-
economic behaviours reflected in the composition of the flint and osseous industries from sites attributed to the different entities
of the EMM to be addressed. The identification of Lussac-Angles points at the Grotte des Fées, located just above Roc-de-Marcamps
(Magdalenian with navettes), opened the debate concerning the contemporaneity of these two sites.
New radiocarbon dates obtained on diagnostic objects (Lussac-Angles points, ‘navette’, pierced phallic batons) and identifiable faunal
material (reindeer, saïga, bison) and human remains demonstrate some degree of contemporaneity between the different EMM entities.
Beyond the presence or absence of particular objects, comparing data from lithic and osseous industries in terms of the organisation of
the chaînes opératoires for the production of domestic tools and hunting weaponry revealed elements common to the different entities,
suggesting a relative cultural permeability. Additionally, new data concerning the circulation of lithic raw materials further reinforces the
idea of dynamic cultural interactions between the valley of the Cher, the Poitou, the Charentes, the Périgord and the Chalosse, exposing for
the first time a north-south axis of communication. Comparisons of lithic and osseous material from several assemblages dated to approximately
19 and 17,5 ky cal. BP thanks to a series of targeted radiocarbon dates has allowed traits particular to the EMM of southwestern
France to be identified. Future research goals include better understanding this important period which, while clearly marked by ideas from
the Lower Magdalenian, sees an acceleration in the emergence of the ‘classic’ Magdalenian. This general increase in the pace of cultural
innovations is clear both in symbolic manifestations as well as techno-economic changes. The integration of these different data forms
should help refine the mechanisms underlying the development of a Pan-European Palaeolithic culture with distinct regional variations.

La partie centre-ouest du Bassin aquitain est caractérisée par une vaste couverture sableuse, aujourd'hui recouverte par la forêt des Landes de Gascogne. Cet espace correspond à un véritable erg sableux, dont la formation remonte au début du Quaternaire. Au cours du Tardiglaciaire, le Sable des Landes constitue déjà une grande unité naturelle au sein du Sud-Ouest de la France. Quel a été l'impact de ce désert sur la mobilité et les réseaux d'échanges des groupes humains vivant en Aquitaine au Magdalénien ? En l'état des données et malgré des prospections, seuls deux gisements bien caractérisés du Magdalénien supérieur sont connus dans ce contexte sableux : Tizon au sud et la Honteyre au nord. En l'état des découvertes, le Sable des Landes apparaît donc visiblement contraignant pour les groupes de chasseurs-collecteurs en relation sans doute avec une faible biomasse végétale et animale. La réévaluation du site de la Honteyre a pour objectif de mieux comprendre les stratégies techniques et économiques mises en oeuvre par les chasseurs-collecteurs magdaléniens dans ce contexte de désert sableux, a priori contraignant. Ce site de plein air a livré un niveau unique relativement peu dilaté mais clairement homogène. L'étude des matières premières indique une aire d'approvisionnement résolument tournée vers le nord avec en particulier l'introduction de blocs à peine dégrossis transportés depuis la Charente (> 100 km ; i. e. « grain de mil »). Ce comportement économique prévisionnel illustre parfaitement la stratégie de planification des besoins d'un groupe venu occuper un espace dépourvu en volumes de silex suffisants pour des débitages laminaires. De plus, la présence de nucléus à lamelles en silex local (i. e. Villagrains) témoigne d'une connaissance relative de ce milieu mais soulignerait plutôt une implantation relativement courte dans ce territoire. La production lamino-lamellaire soignée a permis de fournir des supports d'outils et des armatures de chasse. Certains morphotypes permettent de préciser les liens culturels entretenus par les occupants de la Honteyre avec des gisements situés de part et d'autre du Sable des Landes. En outre, la présence de burins « bec-de-perroquet » et de pointes de Laugerie-Basse permet de tracer des relations évidentes entre la Vienne et les Pyrénées occidentales, espace intégrant désormais le Sable des Landes via ce jalon que constitue la Honteyre. De plus, une production lamellaire particulière livrant des supports naturellement pointus confirme ces liens inter-sites et appuie l'identité culturelle de ces groupes du Magdalénien supérieur récent. Ce gisement, perdu au milieu d'un erg sableux, est donc parfaitement inté-gré à un réseau culturel matérialisé notamment par la diffusion d'objets et de comportements techniques spécifiques. On peut alors mieux appréhender cet espace contraignant comme un lieu de passage de groupes transitant du nord au sud, comme semble l'indiquer le réseau plus large de circulation des matières premières entre ces différents gisements ouest-aquitains, lors de l'amélioration climatique du GI 1e. Abstract: The mid-west part of the Aquitaine basin is characterized by a vast sandy terrain, today covered by the Landes de Gascogne forest. This geological formation was recently the subject of new geomorphological and chro-nostratigraphic analyses, demonstrating that the formation indeed involves an erg that began to form in the early Qua-ternary. During the Late Glacial, the Sable des Landes was already a large natural unit within the southwest of France. How did Magdalenian human groups appropriate this unique area? And what was the impact of this desert on the mobility and exchange networks of these groups? To date, surveys of the area have only documented two sites within this sandy desert clearly identified as belonging to the Upper Magdalenian: Tizon to the south and La Honteyre to the north. Compared to the more densely populated regions of Entre-Deux-Mers to the north and the western Pyrenees to the south, this area therefore appears to have been more constraining for hunter-gatherer groups, doubtlessly due to the low plant and animal biomass. The open-air site of La Honteyre, discovered in 1990 by G. Belbeoc'h and excavated between 1997 and 2000 by M. Lenoir, presents a single level with a relatively low thickness. We here propose a re-evaluation of the site in order to gain a better understanding of the technical and economic strategies implemented by Magdalenian hunter-gatherers in

La Gironde, occupe une place cruciale dans la recherche sur la genèse du Magdalénien moyen et de ses identités du fait de sa position géographique à l’interface des faciès à navettes, à pointes de Lussac-Angles et à lamelles scalènes. Les fouilles menées dans les années 1980 au Roc-de-Marcamps 2 (Prignac-et-Marcamps, Gironde) par Michel Lenoir, ont livré un ensemble de vestiges attribué au Magdalénien moyen. Dans le cadre d’une révision collective du gisement, ce travail présente en détail les résultats de l’étude typotechnologique et technoéconomique de l’industrie lithique du Roc-de-Marcamps 2. Plusieurs dates radiocarbone raisonnées placent ce gisement autour de 18 800 – 18 500 cal. BP, soit lors des premiers temps du Magdalénien moyen. La reconstitution des chaines opératoires a permis de montrer une dissociation des chaines laminaires et lamellaires. Les lames sont destinées à la fabrication de l’outillage. Les lamelles servent de supports à la fabrication des armatures de chasse. Cette étude a donc permis de confirmer l’attribution du Roc-de-Marcamps 2 au Magdalénien moyen. Elle montre aussi que dès le début du Magdalénien moyen, l’ensemble des traits techniques existe déjà (normalisation des supports, optimisation de la production…). L’analyse des matières premières a permis d’établir des liens forts entre la Gironde et la Charente alors que les études précédentes suggéraient un approvisionnement local.
______________________________________________________________________________

The Gironde occupies a crucial place in research on the emergence of the Middle Magdalenian and its identities due to its geographic position at the interface of facies “à navettes”, with Lussac-Angles points and with scalene bladelets. Excavations led in the 1980’s at Roc-de-Marcamps 2 (Prignac-et-Marcamps, Gironde), by Michel Lenoir, uncovered material attributed to the Middle Magdalenian. Within the framework of a collective revision of the site, this work presents detailed results of the typo-technological and techno-economical analysis of the lithic industry of Roc-de-Marcamps 2. Several targeted 14C dates place this site around 18 800 - 18 500 cal. BP, or during the first stages of the Middle Magdalenian. The reconstruction of the chaînes opératoires highlights a dissociation of the blade and bladelet production. Blades are intended for the manufacture of the tools. Bladelets are used as blanks for hunting weapons. This study thus confirms the attribution of Roc-de-Marcamps 2 to the Middle Magdalenian. It also shows that from the beginning of the Middle Magdalenian, all the technical features already exist (standardization of blanks, optimization of the production…). The analysis of raw materials shows strong links between the Gironde and the Charente while previous studies suggested a local supply.

Since the beginning of the 2000s, the classic picture of Pleistocene hunter-gatherers as exclusively “inland hunters” is gradually changing thanks to a positive reassessment of the role of seashore resources. This article provides an updated picture of this question, centering on one case study: the outskirts of the Bay of Biscay during the Middle and Late Magdalenian. The evidence of coastal exploitation is reviewed and shows that several types of activities were recurrent and intensive enough to be documented in a number of sites and to yield abundant archeological remains spanning a long time period: the use of mollusks as food on the Cantabrian coast and the use of their shells as raw material for personal ornaments that circulate widely into the inland; the use of cetacean teeth and bones to manufacture a variety of objects circulating mostly in the northern Pyrenean range. The use of other resources (invertebrates other than mollusks, fish, birds, sea mammals as alimentary resource, algae, maybe drifted pumice) is documented but seems more occasional. Taken together, this evidence allow us to suggest the probable existence, at least in the Middle and Late Magdalenian on the southern outskirts of the Bay of Biscay, of a specific seashore economy established on a narrow coastal strip that is now all but submerged. Evidence from the eastern coast of the bay is much rarer, probably because of the greater distance of the sites from the Paleolithic coast but also maybe because of the presence of a coastal desert in Landes. The possible existence of a specific coastal equipment is discussed (Cantabrian harpoon heads?). These data are briefly put in a broader chronological and geographic perspective and their implications for our conception of the Magdalenian hunter-gatherers and the transition from Pleistocene to Holocene hunter-gatherers are discussed.

Over the last decade, biomedical 3D-imaging tools have gained widespread use in the analysis of prehistoric bone artefacts. While initial attempts to characterise the major categories used in osseous industry (i.e. bone, antler, and dentine/ivory) have been successful, the taxonomic determination of prehistoric artefacts remains to be investigated. The distinction between reindeer and red deer antler can be challenging, particularly in cases of anthropic and/or taphonomic modifications. In addition to the range of destructive physicochemical identification methods available (mass spectrometry, isotopic ratio, and DNA analysis), X-ray micro-tomography (micro-CT) provides convincing non-destructive 3D images and analyses. This paper presents the experimental protocol (sample scans, image processing, and statistical analysis) we have developed in order to identify modern and archaeological antler collections (from Isturitz, France). This original method is based on bone microstructure analysis combined with advanced statistical support vector machine (SVM) classifiers. A combination of six microarchitecture biomarkers (bone volume fraction, trabecular number, trabecular separation, trabecular thickness, trabecular bone pattern factor, and structure model index) were screened using micro-CT in order to characterise internal alveolar structure. Overall, reindeer alveoli presented a tighter mesh than red deer alveoli, and statistical analysis allowed us to distinguish archaeological antler by species with an accuracy of 96%, regardless of anatomical location on the antler. In conclusion, micro-CT combined with SVM classifiers proves to be a promising additional non-destructive method for antler identification, suitable for archaeological artefacts whose degree of human modification and cultural heritage or scientific value has previously made it impossible (tools, ornaments, etc.).

Magdalenian Dog remains from Le Morin rock shelter (Gironde, France). Socio-economical implications of a zootechnical innovation.

"We present in this paper new remains and direct radiocarbon dates of small canids from Le Morin rock shelter (Gironde, France) which constitute a major discovery with respect to the question of wolf domestication during the European Palaeolithic.
In this study a multi-proxy approach has been employed, including species identification and a consideration of the archaeological and chronological context. The canids’ remains have all been studied regarding their morphology, biometry and surface attributes. All dental and postcranial remains of canids were attributed to a species by using a thorough biometric database built from fossil and modern data from Europe. The morphometry of seven remains is outside the size range variability of wolves and therefore can be securely attributed to dog (Canis lupus familiaris). Nineteen are attributed to wolf (Canis lupus) and six could not not be securely attributed to one species or the other (Canis sp.). More than 50% of these Canis remains bear anthropogenic marks that demonstrate the utilization of both wolves and dogs by late glacial human groups. Two of the dog remains from Le Morin rock shelter were directly dated and indicate that Magdalenian groups lived with dogs. A discussion is therefore developed in this article regarding the development of this domestication through time and space.
"

This paper presents a spearthrower made of reindeer antler from the Middle Magdalenian of Isturitz (Layer II, Saint-Périer's 1932 excavations in the Great Hall of the cave). This object, now almost complete, was only restored recently in 2008 when two fragments preserved in various museums were joined. Its morphometric and technical characteristics remind a group of 14 other spearthrowers from the sites of El Castillo (1), El Mirón (1), Roc de Marcamps (3), Combe Saunière I (1), Placard (6) and Garenne (2). We suggest to name this group 'Type 2a', defined by five criteria: a short spearthrower; usually made from an antler splinter, rather than from an antler's section; generally with a single-bevelled proximal end; with a hook of a specific shape, resulting from a specific shaping process; bearing no decoration, except occasionally short and straight incisions. Thus, the spearthrower of type 2a differs from the other spearthrowers of type 2 and types 1, 3 and 4 by the typological and technological aspects as well as its geographical distribution. In chronological terms, it seems to be older and may well be the first known model of spearthrower dating from the Palaeolithic. Its precise dating remains difficult for several sites, but in some cases (El Mirón, Le Roc de Marcamps, La Garenne), it probably dates back to the 19,000-
18,000 cal BP phase (around 15,500-15,000 BP). Depending on the region and the research traditions, this phase has
received many names: Cantabrian Lower Magdalenian, Magdalenian III, Magdalénien à navettes, Magdalenian with
Lussac-Angles points, and the most recent, Early Middle Magdalenian.

Detailed stratigraphic analysis and numerical dating (OSL, IRSL, ESR, 14C) of Pleistocene coversands in
southwest France enable the construction of a renewed chronostratigraphic framework for sand deposition.
The chronological data obtained from sandsheet units testify to the development of transgressive
dunefields since at least the Middle Pleistocene (MIS 10). Three main phases of accumulation occurred
during the Last Glacial. The oldest one (64e42 ka) is associated with wet sandsheet facies, histic horizons
and zibar-type dune fields, which reflect deposition in a context strongly influenced by the groundwater
table. The Late Pleniglacial (24e14 ka) corresponds to the main phase of coversand extension in a drier
context. Silty gley horizons suggest, however, local interruptions of sand drifting during GS 2.1. Lateglacial
stabilization of the coversands may not have occurred before GI-1c (Allerød), which was typified
by the development of cumulic arenosols. These were covered by parabolic dunes during the Younger
Dryas. The variations in extent of the emerged continental shelf during the glacialeinterglacial cycles
may explain the uneven geographical distribution of sand deposition through time. Because of coastline
retreat up to 100 km north of 45
N during the LGM lowstand, the coversands were unable to reach the
northern part of the basin. Comparison with other European regions highlights stronger affinities of the
French record with Portugal than with the Netherlands and Great Britain, probably because of reduced
influence of permafrost.

Since 2012, Tastet Cave (Sainte-Colome, Pyrénées-Atlantiques) is the subject of a new research program led by one of the authors (J.-M. P.). This program offered the opportunity to restudy the parietal art from this site. The parietal art in Tastet Cave has been known since the 1970s; its restudy by three of the authors (D.G., O.R. and R.B.) offers an updated view of the figures, slightly different from that of the previous publications (Omnès 1983; Blanc and Marsan 1984). In particular, it has been possible to replace this art within the chronological and stylistic framework of the Middle and Upper Magdalenian from the Pyrenean and Cantabrian region.

The Magdalenian site of Saint-Germain la Rivière (Gironde, France) has often figured in debates and discussions about the Late Upper Palaeolithic of western Europe. Most of the discussions are based on the material excavated by R. Blanchard - in particular the now well-known burial - as well as the more rigorous excavations by G. Trécolle in the slope deposits. After the detailed study of all this material by M. Lenoir, which allowed the material from the slope and terrace deposits to be attributed to the Lower and Middle Magdalenian, many other studies were done by various researchers on specific aspects of the collections (lithics, personal ornaments, fauna). It became clear, however, that the stratigraphic framework and chronology of the site needed to be reexamined. In order to understand the rise of the classic Magdalenian in Gironde, members of the ANR Magdatis project reexamined many early Middle Magdalenian sites and collections of that area. The thorough work of Trécolle at Saint-Germain-la-Rivière was pivotal to this reassessment. Initially, this involved the analysis of his fieldnotes which led to an archaeostratigraphic reattribution of most artefacts into two principal units, with a potentially problematic interface between these. Once these new archaeostratigraphic divisions had been determined, the analysis of the artefacts could proceed. The comparison of the lithic and bone/antler industries to those of other sites have afforded us new insights into the nature of technical and economic behaviours of this initial Magdalenian of western Europe. In terms of lithic and osseous raw material, we have found a good degree of continuity between the units. The differing treatment of domestic versus hunting flint tools is also an important aspect to consider in the discussion of the changes that took place in hunter-gatherer societies at this time. The bone/antler industry shows some continuity between units (for example the use of the groove and splinter technique), but also some differences (in the upper unit: less diversity in tools; higher frequency of decorated objects). The systemization of a standardized blade production in the upper unit is accompanied by an increase in size of antler points. The production of different lithic projectile morphotypes (backed microbladelets versus scalene bladelets), in parallel with a change in base types within antler points, in addition to the presence or absence of grooves, demonstrate that changes occurred in the toolkits used. A reassessment of the faunal remains also helped to clarify the practices having occurred in each of the two principal stratigraphic units; combined with a taphonomic analysis, the specific subsistence practices in each could be identified. Different hunting strategies were noticed regarding saiga antelope (hunting within female groups versus mixed herds) as well as differing carcass treatment (degree of intensity of marrow extraction, carcass processing techniques). Our analysis also revealed that small-game hunting had been practised (birds and small mammals) already in the earliest Magdalenian phases at the site, thus providing a more complete picture of these reindeer and saiga antelope hunters. Taking into account all of these data and practices allows us to be more specific about the type and pace of technical and subsistence changes having occurred during this period. New 14C dates done directly on bones of identified species or on characteristic organic artefacts also aids us in shedding new light on the two main phases of the Magdalenian. Thea Trécolle excavation and its contents can thus be placed within the overall Saint-Germain-la-Rivière site sequence and compared to other contemporaneous sites in western Europe which have also benefited from a renewed stratigraphic, archaeological and chronological assessment and framework.

Excavated since the 1860s, the Arudy basin has yielded a number of rich Magdalenian sites that enable us to address the question of the human reoccupation of the lower Pyrenean valleys at the end of the Last Glacial Maximum. Until the 1990s, it was considered that at Arudy this reoccupation started ca. 16,000 cal BP in the Middle Magdalenian and continued into the Upper Magdalenian until ca. 14,000 cal BP. However, since the mid-2000s, this picture has been modified through a series of new analyses, excavations (caves of Laa 2 and Tastet), 14C dates and reassessment of ancient collections. Evidence of human occupation ca. 20,000-19,000 cal BP, although limited, has been identified at Laa 2 and Tastet; this is consistent with numerous data obtained at Pyrenean scale showing a deglaciation of the north Pyrenean foothills as early as the end of the global LGM. The number of sites with evidence of Magdalenian has increased, showing that human presence in the basin was more intensive than previously thought. The Middle and Upper Magdalenian occupations have been redated to at least 18,000-17,500 cal BP to 15,000-14,500 cal BP, with the transition between the two phases probably occurring ca. 16,600-16,200 cal BP. The faunal records of these sites document an emphasis on horse hunting, which is not common in the northern Pyrenees, but also witness an evolution over time in the range of species hunted. Furthermore, the Pyrenean affinities of the Arudy groups and their relations with neighboring regions, such as Spain and the northern part of the Aquitaine basin, can be addressed through new evidence: in particular lithic and osseous raw materials and specific types of projectile tips. These results significantly change our view of the nature, rhythm and chronology of the Late Glacial settlement in this area.

Les couches paléolithiques de la grotte d’Isturitz ont livré 119 vestiges humains, découverts lors des fouilles conduites par E. Passemard et R. et S. de Saint-Périer entre 1912 et 1959. Très fragmentés et provenant à 90 % de la tête osseuse, ces vestiges portent dans la grande majorité des cas des modifications anthropiques indiquant un même traitement du cadavre, en tout cas de la tête (décharnement, désarticulation, fracturation intentionnelle, parfois façonnage et/ou gravure des os). Ces restes humains présentent donc un intérêt majeur pour la connaissance des comportements mortuaires paléolithiques. Leur attribution culturelle est cependant problématique. Car même si 55 % des pièces proviennent des couches du Magdalénien moyen (II, E, Eω, SI), une partie des pièces ont également été découvertes dans les autres couches du Paléolithique supérieur : Aurignacien (4 %), Gravettien (29 %), Solutréen (2 %), Magdalénien supérieur et Azilien ancien (8 %). Cette récurrence indique, soit la persistance d’un même comportement pendant toute la période, soit l’existence de mélanges stratigraphiques. Plusieurs indices nous font privilégier la seconde hypothèse : l’ancienneté des fouilles ; la complexité de la stratigraphie et la difficulté à distinguer les différents niveaux, problèmes visibles à la lecture des publications d’E. Passemard et R. et S. de Saint-Périer ; l’existence de raccords entre fragments provenant de couches différentes ; et la mise en évidence de mélanges dans d’autres catégories de vestiges (industrie osseuse et art mobilier notamment). Dans le but de tester cette idée, des datations 14C par SMA ont été effectuées sur un fragment de pariétal et un fragment de frontal humains de la couche II (Magdalénien moyen), et un fragment de pariétal humain de la couche III (Gravettien) ; les vestiges proviennent d’individus adultes et sont issus de trois individus différents. Les résultats indiquent qu’un des vestiges de la couche II et le vestige de la couche III se situent dans un intervalle correspondant au Magdalénien moyen : respectivement 14750 ± 50 BP, soit 18461-17652 cal BP à 2 sigmas (GrA-45328), et 14640 ± 50 BP, soit 18043-17540 cal BP à 2 sigmas (GrA-45332). Ceci confirme l’hypothèse de la contamination de la couche III par la couche II. Le second vestige de la couche II est associé à un intervalle contemporain du Magdalénien supérieur : 13035 ± 45 BP, soit 16351-15174 cal BP à 2 sigmas (GrA-45329). Ce résultat suggère un comportement pérenne par-delà la transition du Magdalénien moyen au Magdalénien supérieur, alors que les autres vestiges archéologiques indiquent que cette transition se caractérise par des changements importants : modification de la panoplie d’armature osseuses, du spectre des espèces chassées, des modalités d’approvisionnement en silex, des modalités de débitage lamellaire, de la fréquence et du type d’éléments de parure et de pièces d’art mobilier. Bien que d’autres datations soient nécessaires pour poursuivre la discussion, ces nouvelles dates confirment l’appartenance de la majorité des vestiges humains issus des fouilles anciennes au Magdalénien.

The importance of coastal resources in the Late Upper Paleolithic of western Europe has been reevaluated in recent years thanks to a growing body of new archeological evidence, including the identification of more than 50 implements made of whale-bone in the Magdalenian level of the Isturitz cave (western Pyrenees). In the present study, the assemblages of osseous industry from 23 Magdalenian sites and site clusters in the northern Pyrenees were investigated, systematically searching for whale-bone implements. The objective of this research was to determine if, and how, tools and weapons of coastal origin were circulated beyond Isturitz into the inland, and if similar implements existed on the eastern, Mediterranean side of the Pyrenees. A total of 109 whale-bone artifacts, mostly projectile heads of large dimensions, were identified in 11 sites. Their geographic distribution shows that whale-bone in the Pyrenean Magdalenian is exclusively of Atlantic origin, and that objects made of this material were transported along the Pyrenees up to the central part of the range at travel distances of at least 350 km from the seashore. This phenomenon seems to have taken place during the second half of the Middle Magdalenian and the first half of the Late Magdalenian, ca. 17500–15000 cal BP. The existence of a durable, extended coastal-inland interaction network including the circulation of regular tools is thus demonstrated. Additionally, differences between the whale-bone projectile heads of the Middle Magdalenian and those of the Late Magdalenian document an evolutionary process in the design of hunting weapons.

Currently, investigations about siliceous resources exploited during prehistoric times in the Cantabrian Mountain and Western Pyrenees are still scarce and, generally, they did not employ methods which go deeply into the provenance characteristics. A review of the studies of lithic resource exploitation offered by historiography indicates that the theme has been examined in a generalized way in most cases. A model with a clear difference between the Eastern and Western territory of the Cantabrian Coast was created: an area with flint and an area without flint. This model needs to be qualified, because in recent years siliceous outcrops have been discovered in zones of theWestern Cantabrian Mountains (Asturias). Information from the investigations in the Cantabrian Mountains, Basque-Cantabrian Basin, and Western Pyrenees indicates diverse patterns. There is a preference for lithic raw material found near the occupations, together with the inclusion of exotic or distant flints, always of good quality, to a greater or lesser extent, according to the chronological periods and the geographical location.

Le Renne, qu’il soit sauvage ou domestique, constitue une source alimentaire de grande importance pour certains groupes humains vivant dans les régions les plus froides du globe. À la lumière de données issues d’observations de boucheries pratiquées de manière traditionnelle par ces peuples, nous avons cherché à discuter des pratiques culinaires au Paléolithique supérieur. Les données récemment acquises sur cinq séries archéologiques, documentant les différents technocomplexes du Paléolithique supérieur, montrent en effet que le Renne occupait une place importante dans la vie quotidienne de ces sociétés. La confrontation des données issues de l’ethnographie avec celles des séries archéologiques nous permet ainsi d’apprécier et de discuter de plusieurs aspects cruciaux de l’alimentation aux temps préhistoriques.

This article synthesises recent zooarchaeological studies and techno-typological analyses of Azilian assemblages from southwestern
France. This data, occasionally complemented by information from bone tools and mobiliary art, allows the variability of the
various Azilian industries to be discussed and sheds new light on the diachronic evolution of Lateglacial societies in the region. Situated
chronologically between the Early Azilian and the Laborian, faunal assemblages from the Late Azilian and its regional manifestations on
both sides of the Garonne Plain are investigated in order to better understand questions concerning the exploitation of the environment.
 

 
  




  
 






  -
lying Magdalenian in regards both the lithic industry (bipoints and backed points interpreted as hunting weapons, and small, regular
blades with scalariform retouch) and the faunal spectrum (reindeer replaced by horse). Overlying this Early Azilian occupation, the
less dense Late Azilian occupation is represented by less well-prepared, backed points on blades associated with a faunal assemblage
dominated by rabbit and red deer.
The long stratigraphic sequence of Pont d’Ambon includes several layers attributed to the Magdalenian, Early Azilian, Late Azilian
and Laborian. Differences in knapping techniques suggests that the Late Azilian can be divided into two assemblages representing
 




 






 


!


" 

 


 
"
 
#



 
  

$
"%
!









  




 

"
 
$
"
  


debitage modes differ between the two assemblage types. The Laborian is characterised by several particular artefacts – backed points
with truncated bases on small, very regular blades and tools on fairly regular blades – demonstrating a higher technical investment than
is evident with the Late Azilian assemblage.
More general aspects of the Early, Late Azilian and Laborian lithic industries are also discussed by taking into account others sites of
the northern Aquitaine region such as Rochereil, Roc d’Abeilles, Pégourié, Pagès, Le Morin and Le Cuzoul de Gramat. Compared to
the Early Azilian, Late Azilian assemblages show simpler production modes, although some technical traditions persist (blade production
and châine opératoires). In terms of faunal exploitation, the combination of ungulates such as red deer and smaller game such as






 
 
 & 
+  
   


#
 ;



#

 




 


&
!




 


 




<==
>







 




 


 $
"

$
">

 
 
 
-

$
">
 
$
" !

" 

"
 

?



 

K
!#

$
"#
 
!
 
& 
<==
>








 
 
!  !



! 





!


>


Margineda (Early Azilian, an Azilian with harpoons, Final Azilian). A major difference with Troubat lies in the preferential use of local

#



$
  
[
 
[
 

 %$
 
#






 

phases of Rhodes II (level ‘Foyer 5’: Early Azilian, level ‘Foyer 6’ intermediate, ‘Foyer 7’: Late Azilian with harpoons), but remains
consistent at Balma Margineda (level 10 and 9: Early Azilian, level 8: Typical Azilian, level 7: Final Azilian). Although the Azilian is
easily distinguishable from other regional industries, its internal variability remains to be documented and its early and late phases are



  

Laborian sites in the central Pyrenees are less typical than those from the northern Aquitaine with the best-preserved assemblages being
those from Gouërris, Manirac and Buholoup. Laborian material is present on other regional sites such as Troubat, Mas d’Azil or La
Tourasse; however, no clear occupation layers have been documented. As in the northern Aquitaine, red deer remains the major prey




]


 


" #% 

 
 


-
larly at Troubat, were rarely captured and consumed. The presence of other game in the faunal spectrum such as the ibex, indicative of
more mountainous terrains, demonstrates the exploitation of a variety of different biotopes.

During the last glacial period, a large part of the Aquitaine basin (southwest France) was a periglacial desert comprising coversands with low-relief dune fields surrounded by loess accumulations. OSL and radiocarbon dates show that the phase of maximum sand deposition coincides with Marine Isotope Stage (MIS) 2. Peats and gleyic palaeosoils intercalated within the sands at some sites indicate that vegetation cover was able to develop locally during short events, possibly DeO interstadials, due to raised groundwater levels in interdunal depressions. Few Palaeolithic sites have been discovered in the coversand area in contrast to the peripheral loess region. Systematic survey along a future highway corridor demonstrates that this paucity of sites is not the result of insufficient survey nor deep site burial, but rather reflects an archaeological reality. This strongly suggests that the sand area was not attractive for hunteregatherer populations due to its reduced levels of water resources, and available vegetation and animal biomass. The distribution of cultural markers such as art items and projectile points also shows that the coversand area probably acted as a barrier separating two different cultural sub-areas, one in the Pyrenees and Cantabria, the other in the Périgord. As a consequence, the commonly accepted view that southwest France, as a whole, served as a refugia during the cold and arid phases of the Pleistocene should be replaced by a more complex one that reflects the fact that a large part of the territory was almost unoccupied and that human populations were concentrated along alluvial valleys.

Dans le cadre du projet Magdatis, nous avons réalisé trois films pédagogiques de 2 minutes chacun, destinés à valoriser le travail d'étude et de datation effectué sur les collections paléolithiques du Musée d'archéologie nationale et du Musée d'Aquitaine. Chaque film est construit autour de la datation directe d'une ou deux pièces d'industrie osseuse choisies parmi les objets emblématiques de la fin du Paléolithique supérieur. Ces films ont été réalisés par la société Get In Situ, grâce à un financement de l'ANR et au travail de chercheurs CNRS de Traces et de Pacea, sur la base d'objets prêtés par le Musée d'archéologie nationale et par le Musée d'Aquitaine. Tous les participants ont donné leur accord pour que ces films soient librement et largement diffusés. Seule l'exploitation commerciale en est interdite.

In South-West of France, new analyses of lithic and bone equipment and a series of 14C dates allow us to address the question of the genesis of the Magdalenian. For the Lower Magdalenian, the re-examination of lithic and bone assemblages of the lower levels of Saint-Germain-la-Rivière (Trécolle excavation), Gironde, clarifies the techno-economic behavior of these groups. Comparisons are made with other sites such as Gandil shelter in Bruniquel or Scilles cave in Lespugue. This “first” Magdalenian is dated between about 21 and 19 ky calBP. New results in the MAGDATIS project also clarified the definition of an “Early Middle Magdalenian” in Gironde, dated between 19 and 18 ky calBP. Assemblages from this period include Roc-de-Marcamps and Moulin-Neuf (Gironde) but also the upper levels of Saint-Germain-la-Rivière (Trécolle excavations) and those of the ongoing excavations at Petit Cloup Barrat (Lot; Castel and Chauvière excavations). Data from these sites shed a new light on the early days of the Classic Magdalenian, including the so-called “facies” with Lussac-Angles points, antler “shuttles” or scalene bladelets. Between Lower and Early Middle Magdalenian, techno-economic characterization of antler working, direct dating of diagnostic objects and research on lithic equipment all show technical changes. These data lead to specify the rhythms of transformation of the Lower Magdalenian and the slow genesis of Classic Magdalenian through new behaviors, both technical, economic and symbolic. 

Dans le Sud-Ouest de la France, de nouvelles analyses menées sur les équipements lithiques et osseux et un corpus de dates 14C permettent d’aborder la question de la genèse du Magdalénien. Pour le Magdalénien inférieur la révision des séries lithiques et osseuses de l’ensemble inférieur des fouilles Trécolle de Saint-Germain-La Rivière en Gironde permet de préciser les comportements techno-économiques de ces groupes. Des comparaisons sont proposées avec d’autres gisements comme l’abri Gandil à Bruniquel ou les Scilles à Lespugue. Ce « premier » Magdalénien est daté entre environ 21 et 19 ky calBP. De nouveaux résultats obtenus dans le cadre du projet MAGDATIS ont également permis de préciser la définition d’un Magdalénien moyen ancien en Gironde daté entre 19 et 18 ky calBP. Il s’agit notamment des sites de Marcamps et de Moulin Neuf en Gironde mais aussi des ensemble supérieurs provenant d’une part des fouilles Trécolle de Saint-Germain-La Rivière et d’autre part, des fouilles en cours menées au Petit Cloup Barrat dans le Lot. Ces sites alimentent la problématique des premiers temps du Magdalénien classique avec les dits « faciès » à pointes de Lussac-Angles, navettes ou lamelles scalènes.
Cette communication sera donc l’occasion de présenter une nouvelle synthèse revisitant les différents scénarios actuellement envisageables pour comprendre les débuts du Magdalénien. La poursuite des fouilles tant dans les Pyrénées que dans le Quercy livrent aujourd’hui de nouvelles données complémentaires à la révision de séries anciennes. Entre Magdalénien inférieur et Magdalénien moyen ancien, la caractérisation techno-économique des procédés de débitage du bois de cervidé ainsi que la datation directe d’objets diagnostiques mettent en lumière l’existence de traditions techniques distinctes. Parallèlement, la poursuite des recherches sur les équipements lithiques montre des changements techniques. Ces données permettant ainsi de préciser les rythmes de transformation du Magdalénien inférieur et la lente genèse du Magdalénien classique, au travers de nouveaux comportements tant techniques qu’économiques ou symboliques.

Le Roc de Marcamps (Prignac-et-Marcamps, Gironde) est un site du Magdalénien moyen connu notam-ment par la découverte de nombreuses navettes dans les fouilles anciennes du secteur 1. Les travaux menés par Michel Lenoir dans le secteur 2 durant les années 1980 n'ont pas livré ce type de marqueur osseux mais un ensemble de vestiges attribués également à cette période. Plusieurs dates radiocarbone raisonnées placent ce gisement autour de 18900-18600 cal. BP, soit lors des premiers temps du Magdalénien moyen. Cette phase de transition apparaît synchrone de l'événement climatique d'Heinrich 1 marquant une péjoration climatique avec des conditions froides et steppiques. Elle se caractérise en outre par d'importants changements techno-économiques ou symboliques. À l'interface des « faciès » à navettes, à pointes de Lussac-Angles ou à lamelles scalènes, le Roc de Marcamps 2, et plus largement la Gironde, occupe une place privilégiée pour la compréhension de la mosaïque géoculturelle structurant la genèse du Magdalénien moyen. Dans le cadre d'une révision collective du gisement par des membres du projet « Magdatis », cet article présente une approche croisée, détaillant les analyses taphonomiques et archéozoologiques des restes de faune (ongulés, méso-mammifères et avifaune) ainsi que les études typotechnologiques des industries lithique et osseuse et de la parure. La stratigraphie du Roc de Marcamps 2 comprend un ensemble supérieur mêlant dépôts récents, déblais de carrières, céra-miques et vestiges paléolithiques brassés par des animaux fouisseurs. L'ensemble inférieur apparaît mieux conservé et renferme les vestiges magdaléniens. Les restes fauniques documentent une chasse principalement axée sur les bisons, les antilopes saïga, les chevaux et les rennes durant la mauvaise saison et au début de la bonne saison. L'exploitation de ces ressources animales est intensive et orientée vers la récupération de la moelle en complément de la viande. L'étude des matières premières siliceuses, préférentiellement d'origine locale, atteste toutefois un apport de silex de Saintonge (silex « grain-de-mil »). La production lithique est principalement composée de débitages lamellaires voués à la confection d'éléments d'armatures de chasse. Le débitage laminaire fournit des supports normalisés pour la fabrication d'outils. L'industrie en matières dures d'origine animale est assez pauvre et comprend un équipement en os (aiguilles, lissoirs, retouchoirs) et en bois de renne (essentiellement des armatures de projectile) ainsi que des éléments de parure sur coquillages (en majorité des dentales) et sur dents (en majorité des incisives de renne sciées). L'étude typotechnologique des industries lithique et osseuse et de la parure suggère des rapprochements tant avec le « Magdalénien à navettes » qu'avec le « Magdalénien à pointes de Lussac-Angles », deux faciès dont le Roc de Marcamps 2 est contemporain. Ceci doit conduire à s'interroger sur la pertinence de l'utilisation de ces faciès pour définir des « cultures exclusives » qui finalement se recouvrent chronologiquement et géographiquement, notamment en Gironde.