Female representations were first discovered at the end of the 19th century (e.g. archaeological sites of Grimaldi, Liguria in Italy; Brassempouy, Landes or Laugerie-Basse, Dordogne in France). Since then, they have been studied from different perspectives, formal, technical and symbolic as well as anthropological (Piette, 1984; Begouen, 1934; Pales, 1972; Delporte, 1979; Leroi-Gourhan, 1971; White, 2006; Dupuy, 2007; etc.). Researchers have paid particular attention to full-body figures (figures can be "whole" or fractured), like Gravettian sculptures. Representations which H. Delporte called "synecdoches" - female genitals with no other anatomical details - have thus been neglected. This study aims at encompassing the image of woman as a whole with a diachronic then a synchronic approach, allowing the analysis of permanencies and variations with regard to both shape and symbolism. Female representations (whole or segmented) are known from the early Aurignacian in France in Perigord (Blanchard, Castanet, etc.) and Ardeche (Chauvet) and in Germany (Hohle-Fels). These figures show great diversity while retaining an undeniable unity of form throughout the Upper Palaeolithic. Synchronic and diachronic analyses of this stylistic variability using statistical tools (Correspondence Analysis) show the existence of both universal invariants in form and variations that may indicate distinct cultural traditions. The evidence of graphic convergences and divergences in female representations demonstrates the circulation of conventions over vast territories, but also raises questions about the semiotic status of these formal variations over long periods of time. First, a set of anatomical and sexual criteria is defined in order to identify human figures. The corpus established on the basis of these criteria shows a total of 988 entities, which can be subdivided into 729 female figures (portable 665, parietal 64) and 259 isolated genitals (portable 115, parietal 144). Although the female theme is known throughout the European continent, certain sites are quantitatively important (Combarelles, La Marche, Grimaldi, Gonnersdorf, Kostienki, etc.). In a second step, a description of the figures is proposed, using a formal repertoire, classified according to attributes/values. The data can then be processed using statistical tools (Correspondence Analysis). This first analysis allows us to observe variations from a diachronic point of view. The results concerning the full-body figures highlight the permanencies of shape for both two-dimensional figures (engraving, painting, bas-relief) and three-dimensional figures (sculptures) and show two formal conceptions of womanhood, independently of the techniques used. Regarding the representations of female genitals, these are organized according to a chronostylistic seriation, pointing to a dichotomy of form between the Aurignacian, Gravettian and Solutrean periods on the one hand, and the Magdalenian period on the other. In the course of the first three periods, pear-shaped and sub-oval forms apparently prevailed, except in the Ardeche region where triangular forms seem to have been privileged, forms that dominated during the Magdalenian period. The third step consists of the restitution of these graphic sets on a regional and chronological level that enriches the first set of data. Stylistic comparisons can then be made between sites and regions on a synchronic level. The evidence of graphic convergences and divergences for a similar period also allows us to observe their spatial evolution. For example, concerning the full-body figures, in the Gravettian period, stylistic influences seem to come from Central Europe, whereas during the Magdalenian period, but in a different form (schematic female figures), the subject finds itself anchored in South Western France and appears to spread quickly to Northern Europe. During the Aurignacian and Gravettian periods, a stylistic opposition concerning the representation of female genitals can be perceived between South Western France and the Cantabrian region on the one hand, and the Ardeche region on the other. During the Magdalenian period, however, formal homogenization spread throughout Europe. Finally, this analysis provides useful elements of reflection on the symbolic representations of women that persist throughout the Upper Palaeolithic within the European territory. Does the evidence of three different graphic conceptions of woman (detailed tendency, non-detailed tendency, isolated female genitals) reveal a symbolic dichotomy and different meanings? At the same time, how does one explain the persistence of certain stylistic traits over such a long period? As for the symbolic signification of these representations, as shown in this analysis, it can only be considered under a particular angle. It seems that the forms may have served as "containers" expressing different contents, probably of a mythical nature (Leroi-Gourhan, 1964; Levi-Strauss, 1983). To conclude, the symbolism of these figures cannot be disconnected from Palaeolithic graphic manifestations, in a close relationship with the socio-economic functioning of the group.
