Review Article

Footprints of the horse-people: new research on Upper Palaeolithic France

Paul Pettitt
Department of Archaeology, University of Sheffield, Sheffield, UK

Antiquity 79 no. 305 September 2005

JEAN COMBIER & ANTA MONTET-WHITE (ed.). 2002 Solutré 1968-1998 (Société Préhistorique Française Mémoire 30). Paris: Société Préhistorique Française

Click to buy

SOPHIE TYMULA 2002. L'art solutréen du Roc de Sers (Charente) (Documents d'archéologie française No. 91) Paris: Maison des sciences de l'Homme

GILLES TOSELLO 2003 Pierres gravées du Périgord Magdalénien: art, symboles, territoires (Gallia Préhistoire Supplément 36) Paris: CNRS

Click to buy

HUGUETTE DELOGE & LOUIS DELOGE (ed.). 2003 Le rocher de la Caille: un site magdalénien de plein air au Saut-du-Perron (Société Préhistorique Française Mémoire 31) Paris: Société Préhistorique Française

Click to buy

Oxbow books logo


Four new publications relating to the French Solutrean and Magdalenian demonstrate admirably how new analysis of old archaeological collections with new methodologies, and in one case new excavations, can shed considerable light on enigmatic areas of Upper Palaeolithic behaviour. Emphasis on Solutrean lithic technology can often mask its artistic output, while the spectacular cave art and engraved bone and antler of the Magdalenian is often appreciated at the expense of engraved stone blocks and pebbles. These beautifully illustrated publications rectify this situation, and while they are not for a general audience, Palaeolithic specialists will find them of considerable value.


For a century and half, Solutré has been an Upper Palaeolithic icon. The large collection of bifacially retouched feuilles de laurier excavated there was used by de Mortillet to define the Solutrean in 1869, and the chronocultural importance of a site containing all major assemblages of the Upper Palaeolithic was recognised by Breuil. The abundance of horse remains in the Gravettian 'magma' (a brecciated mass of horse bones), and in the overlying Solutrean and Magdalenian levels soon gave rise to the notion of horse-driving over the cliffs, which persisted for decades. Archaeological accumulations at Solutré are vast, and although it is difficult to evaluate today the numerous early excavations on the plateau and slope under the cliff, a commendable understanding of accumulations over most of 30 000 years is imparted in the volume edited by Combier and Montet-White. They present the results of excavations conducted in 1968-1976 in a wider context, taking us towards a more sophisticated understanding of sediment accumulation and horse procurement/butchery at Solutré.

Eight principal archaeological levels spanning the Mousterian to Magdalenian accumulated at Solutré in two partially interdigitating slope deposits at the foot of cliffs that have been partly reworked by stream channels and other erosional phenomena. Although the excavation strategy targeted all known chronocultural phases at the site (and its vicinity, producing abundant Mousterian assemblages, detailed in the volume), Gravettian, Solutrean and (especially) Magdalenian materials stand out. In Sector J10, the Gravettian 'magma' was over one metre thick and seems to have been formed by complex water circulation and stream erosion, indicated by a degree of movement given by the orientation of horse long-bones. Sectors N16 and P16 yielded a rich Magdalenian fauna and associated artefacts. Here, a number of articulated elements imply that the remains have not moved far, and a degree of patterning in the discard of tools suggests that butchery areas were cleared in preparation for the next hunt. Turner's meticulous analysis of the Magdalenian fauna shows how horse clearly dominate over reindeer, bison and wolf in the processed fauna. The presence of all stages of butchery and marrow smashing on the horse bones indicates that this species in particular was intensively exploited. The non-selective age distribution suggests that adults from family groups were the preferred quarry. Turner concludes that the tactics of horse butchery remained similar from the Gravettian to the Magdalenian, suggesting a considerable degree of behavioural redundancy at a strategic resource-procurement site. While the location seems to have been of perennial significance for horse hunting, reindeer hunting seems to have been restricted to between November and May, and bison in winter and summer.

The archaeological levels are artefactually variable. Lithics are few in all but the Solutrean levels, and organic artefacts (generally sagaies, baguettes and lissoirs) are only really known from the Solutrean and Magdalenian. The greater abundance of artefacts in the Solutrean levels suggests more intensive use in this period. As the site is largely a kill site, the recovery of art mobilier from the Aurignacian, Solutrean and Magdalenian levels implies that such items were embedded in daily life. A small osseous assemblage from the Aurignacian assemblage includes three broken bâtons percés, rare in the Aurignacian. Small carved animal statuettes from the Solutrean levels recall those from the German Aurignacian, and engraved animal outlines on pebbles look forward to the Magdalenian. In the Magdalenian levels, examples of art are less numerous but more diverse, including fragmentary sagaies and engraved bones, engraved pebbles and items of shell jewellery. Some items can be equated with the Magdalénien Moyen à navettes including a phallic bâton percé.

The project succeeded in its aims of reconstructing environments and chronology for the entire sequence, and the discovery of a rich Magdalenian deposit was a bonus. Some aspects could have been treated more fully, such as a technological analysis of lithic technology, particularly of the feuilles de laurier and pointes à cran, but such omissions do not detract from an important volume on a major kill and butchery site. Two series of 14C dates place the Aurignacian at Solutré ~34-29ka BP; the Gravettian ~28-24ka BP; the Solutrean ~19.5-16.5ka BP and the Magdalenian ~15-12.5ka BP. These samples span some 16 000 radiocarbon years and support the notion that the site was of tactical importance for short-term hunting episodes to various cultural groups over long periods of Upper Pleistocene time. Combier presents two models by which hunter-gatherers could trap migrating horses in the confines of the valley. As one of a number of lateral gorges separating the uplands of the Maconnais to the west and the Saône to the east, it seems to have been important in the seasonal migrations at least of horse and reindeer.

Le Roc de Sers

A very different facet of Solutrean life is presented in Sophie Tymula's volume on the Roc de Sers sculpted frieze, one of the relatively few examples of parietal art that can be confidently attributed to this period as it fragmented and fell off the rear wall of the rockshelter into dated archaeological deposits. Sculpture is rare in Upper Palaeolithic parietal art, and, given the relative paucity of Solutrean parietal art in general, Roc de Sers is doubly important, as, in Tymula's view, it serves as a benchmark for defining Solutrean artistic form and technique.

Abundant evidence was recovered from three main activity areas across a large part of the rockshelter and associated small cave; a rich classic Solutrean lithic assemblage bears witness to the production of feuilles de laurier, pointes à cran, burins and grattoirs. Over 52 items of bone and antler including sagaies, lissoirs and two bâtons were recovered, many bearing engraved decoration. Typically Solutrean personal ornamentation comprises pierced animal teeth (generally fox canines), shells and other pendants. Seventeen engraved stone plaquettes reflect the typical bison-horse-cervid theme of Upper Palaeolithic art, and were, like the other symbolic items, recovered under the line of the sculpted wall, reinforcing the notion that this area of the site at least functioned as a 'sanctuary'. Horse and reindeer dominate the fauna, and all indications are that the occupation took place under cold conditions. With unsatisfying 14C dates between ~19ka BP (bulked bone) and ~17ka BP (AMS) which could make the art contemporary with the Magdalenian art of Lascaux and/or the Badegoulian of Abri Fritsch, the precise chronological position of the site needs clarification. Tymula's inventive and rigorous methodology aims not only to reconstruct the original composition of the parietal frieze, but also the methods of the sculptors, demonstrating that the main artistic objective of sculpture - the interplay between form, light and shadow in the overall work - was fully appreciated by the Solutreans.

Nineteen fragments of the frieze were recovered from the archaeological levels, and a number of these refit, so as to allow a confident reconstruction of their position on the rockshelter wall. It appears that the frieze was c. 10m long, and arranged in two tiers and two main 'sides', with enough continuity of style and theme between them to indicate that the frieze formed a compositional whole. Although the sculptures are considerably eroded, enough detail is preserved to reconstruct the manner of production. The surface of the wall was prepared by pecking, rubbing and scraping, and, in at least some places, charcoal sketches were prepared in advance. The main figures exploited the morphology of the wall and built relief around this, using bas-relief sculpture and engraving to bring out anatomical detail. At least one image was coloured with red ochre. The resulting shading and colour conveys movement, and Tymula estimates that c. 70 per cent of the images are 'animated' in this manner. Overall, some 51 images present a degree of graphic unity. Some are large (one quarter life size), with smaller ones superimposed upon these. Over half are animals: horse and ibex dominate, followed by bison and reindeer, somewhat similar to the engraved plaquettes, presumably placed on the floor under the frieze and reminiscent of Upper Palaeolithic iconography overall. Tymula recognises two ensembles: one (western) dominated by ibex, the other (eastern) by horse, with a mutual exclusion of the two species. Although numerically not as important, bison are found in both ensembles, and in this sense bison is 'all pervasive'. Two humanoid figures are found in the horse ensemble. Clear similarities of style and iconography exist between the sculpted frieze and the engraved stone plaquettes, as well as with the Solutrean art of Parpalló and Fourneau du Diable; these confirm a unity of Soltrean symbolic thought. Tymula's reconstruction and analysis of this important frieze should become a standard reference for the understanding of Solutrean art, as well as for the interaction between parietal art and engraved stone plaquettes and other art mobilier.

Engraved Magdalenian plaquettes of the Périgord

Examples of engraved stone blocks and plaquettes are known in the Western European Upper Palaeolithic from the Aurignacian, but they are rare before the fourteenth millennium (14C) BP and their number increased significantly during the Magdalenian. Despite being recognised since the 1860s, they remain enigmatic. Gilles Tosello's volume presents a major analysis of plaquettes from the Middle (~14-13ka BP) and Late (~13-11ka BP) Magdalenian of the Périgord, and this exhaustive work takes us much further into understanding the dynamics of these pieces. The study is based on over 400 examples from critical sites such as Limeuil, La Madeleine and Laugerie Basse, supplemented by six other sites. After discussing each site in detail, Tosello's synthetic analysis covers the form of supports, compositional framing, subject matter, style, context, associated fauna and art on bone and antler. Similarities in these realms between sites demonstrate a functional continuum, with some change between the Middle and Late Magdalenian. In the Middle Magdalenian (a sample largely comprising La Madeleine and Laugerie Basse, supplemented by Teyjat), horse clearly dominate, followed by bison and reindeer. Iconography is very figurative, albeit with inept animation. Homogeneity between sites in the Périgord and elsewhere (particularly the Pyrenees) is clear. Although objects of bone and antler contained a relatively high degree of geometric decoration, absent on the plaquettes and probably dictated by the form of the osseous media, horse dominates again among the figurative art. Parietal parallels for the Middle Magdalenian are numerous, and again the dominance of horse is clear. The sculpted frieze at Le Cap Blanc is very close thematically, dominated by horse, with bison and cervids (probably reindeer) the main species depicted. Tosello's wider comparisons of lithic and osseous art mobilier and parietal art demonstrate a considerable 'common inspiration' across all of these media and contexts, and emphasise how robust the belief systems underpinning the art were.

Tosello's Late Magdalenian sample is largely taken from Limeuil and the upper levels of La Madeleine, supplemented by the Grotte des Eyzies and Le Soucy. He notes a high degree of fragmentation of the plaquettes - more so than in the preceding sample - and develops a complex chaîne opératoire, from engraving, heating, fragmentation, frequent reuse (re-engraving), to further fragmentation and discard. Clearly, the plaquettes cannot be interpreted as having a brief use life; whatever their function, it took a complex form and was embedded in the complex spatial dynamics of large-site occupation. Reindeer become the dominant species depicted, followed by horse. Engraving is confident and a small number of depictions are animated more competently that in the Middle Magdalenian. Figures were carefully 'framed' within their support (showing an intimate association between support and depiction) and a 'floor line' was often drawn below the hoofs of left/right facing animals. Tosello argues convincingly that animals drawn 'vertically' above such horizons represent animals running towards or away from the viewer, and interprets this as the deliberate depictions of herds and the action of individuals within these. Clearly, these are highly composed, active pictures, not simple individual depictions. Horse still dominate the associated osseous art mobilier, on which fish (very uncommon on the plaquettes) are almost as important as reindeer. These differences suggest something of a growing (and to Tosello, profound) divergence between the plaquettes and art on functional and decorative organic items. More widely, art mobilier on stone and parietal art sees the growing importance of stylised human females and the retention of horse and reindeer as important themes, intriguing examples of the specifics of the divergence between the media.

A number of enigmas remain. Why are indeterminate animals relatively common? Do these represent simply 'badly drawn' animals, or imaginary beings rooted in the Magdalenian imagination? The complex chaîne opératoire of the plaquettes seem to have obliterated any potential spatial patterning, so it may be rash to eliminate Leroi-Gourhan's suggestion that the plaquettes delineated special areas - his 'plaquette sanctuaries'. It is tempting at least to emphasise that, unlike parietal art which must be composed before execution, the plaquettes are by their very nature moveable, and therefore the possibility of rearrangement exists. Their apparently complex movement around major occupation sites suggests that their role/s in Magdalenian life were spatially embedded in more prosaic activities, an all-pervasive mediation between the multiple aspects of the Magdalenian mundane and spiritual world. Tosello's painstaking work is an excellent example of how large collections of material can yield fresh and innovative insights into poorly understood aspects of Upper Palaeolithic behaviour. Beautifully illustrated (if expensive), this volume provides an important conceptual link between parietal art, lithic and osseous art mobilier, faunal assemblages and spatial organisation of sites. The degree of homogeneity between these categories is testimony to the conceptual unity of the Magdalenian world over vast distances.

Le Rocher de la Caille

Engraved plaquettes dating to the Late Magdalenian link Tosello's work with the welcome publication of the Late Upper Palaeolithic site of the Rocher de la Caille by Huguette and Louis Deloge. This is one of a number of Palaeolithic sites in the Saut-du-Perron, on the left bank of the Loire close to where it enters the plain of Roanne. Remains of a substantial Magdalenian site were excavated there in 1979-1982, before the area was flooded. Similar in nature to the Middle and Late Magdalenian open-air camps of the Paris Basin, the Rocher de la Caille seems to date entirely to the first half of the Late Glacial Interstadial. It comprises a primary activity area of some 30m2 defined by lithics and probably representing bladelet production around three large hearths spaced about four metres apart. Burins too were manufactured, but were discarded, presumably after use. Ochre was abundant in the primary activity area. Excavated materials covered some 112m2, and within this area, large blocks of stone deposited by solifluction formed other foci for activity. Cobbles of volcanic rock were scattered across the site and appear to have served as hard hammers, mallets for the working of organic material, and possibly pestles for grinding vegetal matter. The primary activity area also produced 29 fragments of stone bowls, generally of steatite, eleven with engraved decoration. At least two of these served as lamps and were associated with hearths, while most appear to have served as containers, probably for mineral pigments as indicated by spectrographic residue analysis. Some 18 400 lithic items were recovered, of which 1479 are formal tools on high quality locally-available flint. Numerous examples of manufacturing waste and blade/bladelet cores attest production on site. A considerable degree of typological variability is not surprising, given that the site appears to have been a major camp site. Retouched blades and bladelets dominate, particularly lunate, straight and obliquely-truncated forms. Some of the latter are strikingly similar to northern European Cheddar/Creswell/Tjonger points; this is not surprising, as the one 14C date for the Rocher de la Caille (12 210 480: Ly-5645) is contemporary with the British and Belgian Late Magdalenian ('Creswellian'). Burins are particularly diverse in form, presumably because various organic materials were worked. A large number of perçoirs and microperçoirs, some of which are stained with ochre, attest the boring of hides or hard animal tissue, and a number of fragmentary pièces esquilles hint at heavy duty tasks.

The Late Magdalenian community of the Rocher de la Caille was part of a far-flung procurement network characteristic of the period. Although many raw materials of volcanic stone could be (and were) acquired within 10km of the site, flint, which does not outcrop in the gorges of the Loire, was obtained from sources between 5km and 40km distant, and some 66 per cent of the flint assemblage was obtained from the Paris Basin, some 200km further north. The abundance of Paris Basin flint on site, the derivation of a fossil ammonite from at least 70km to the north-east, the general similarity of the site to Late Magdalenian sites of the Paris Basin, and typological similarities of lithic armatures with those of the Late Magdalenian of northern Europe in general, together suggest that the Rocher de la Caille represents a southern extension of groups operating further north.

In the primary activity area and at its southern periphery, forty fragments of engraved schist plaquettes were recovered. In terms of production and style they fall easily within the Middle and Late Magdalenian, as defined in Tosello's volume: they are strikingly similar to those of La Madeleine and Laugerie Basse. Most depictions, since on fragmentary supports, are partial and show a clear similarity with the chaîne opératoire decribed by Tosello; and, in keeping with the 14C date for the site, itself in line with the Middle Magdalenian of the Périgord, horse is the dominant species depicted. The few complete images suggest (to my mind) a degree of compositional 'framing' that Tosello identified in the Périgord. Clearly, the engraving of plaquettes, their use, fragmentation, potential reuse and final discard was not an activity restricted to rockshelters and cave mouths. The data provided by the Rocher de la Caille forms an important geographical and conceptual link between the rockshelter dominated assemblages that Tosello studied, and the large samples of contemporary Middle/Late Magdalenian sites of Gönnersdorf and Andernach. Whether habitation structures were present at the Rocher de la Caille remains an open question, but the repeated use and resulting fragmentation of engraved blocks and plaquettes was clearly a very widespread phenomenon in the Middle and Late Magdalenian. While regional differences in lithic typologies may relate to locally-contingent hunting strategies, these plaquettes present evidence of the uniformity of Magdalenian practice across wide areas of Europe. Given the nature of these pieces, it is difficult to escape a 'ritual' interpretation, and if this is correct, it attests a degree of uniformity of Magdalenian belief.

Oxbow books logo