On the eastern side of the Rhône delta, the wide pebbly plain of la Crau spreads, formed by old deposits of the Durance river which flowed, during the Pleiostocene, directly into the Mediterranean sea. A survey of the different morphological units of this area and a sedimentological study of the pebbly formations show that there are two distinct cycles with quite antithetic petrographical and mineralogical characteristics. The first cycle, named "Aries pebbly formation" is made up of calcareous cobbles and smectite from the subalpine areas (lower Villafranchian). That formation is the extension of the highest terraces of the lower and middle Durance basin. In Crau, as well as in the Durance basin, one may find reworked materials of these pebbly formations in the same stratigraphie position : "Terraces of Terrefranche and la Vermillière". The pebble gravels of the second cycle, better preserved because younger (upper Pleistocene), show cobbles of endogenic rocks and illite from the Alps. These are the "Luquier and Miramas pebbly formations". The "Miramas pebbly formation" overlying a molluscs community of the end of the Rissian glaciation can be dated Riss-Wurm interglacial or early Würm (Würm I). Lithic tools of late Acheulean were discovered by H. de Lumley (1965) on the top of the "Luquier pebbly formation", the age of which is therefore Rissian. Between pebble gravels of the first and the second cycle, we found some intermediate deposits (mid-Pleistocene) : "Terraces of les Glauges, Brays and Entressen". Above the last one, we discovered one early acheulean lithic tool. The discovery of the "Terrace of les Glauges" is important because by some characters (bedding pattern, elevation, flexur-ed longitudinal profile) it is similar to the first cycle deposits, while its petrographical and mineralogical associations are typical of the second cycle. It's the first evidence of a new origin of the clastic materials. In the middle and lower part of the Durance basin, one may find the same duality between the two sedimentary cycles, and also observe the same transitional layers. But, we do not take part in the embarrassing problem of assigning these terraces to definite glacial episodes. We just propose a model of correlations. Most of the earlier authors readily connected the lower and the middle terraces with "Luquier and Miramas pebbly formations". While the petrographical and mineralogical characteristics of the first cycles are found only in the highest terrace of the Durance (lower basin) or in the upper levels of the "Riez-Valen-sole conglomerates" (middle basin). Moreover, the first level from which we can observe the materials of the second cycle is, in the lower part the the Durance basin, the "Terrace of Lauris", or, in the middle part, the "Terraces of Brigadel — Ruffe and Villard-St-Laurent", described by M. Dubar (1975). These last terraces are in homology with the "Terrace of les Glauges". Therefore, because of the elevation of the "Terrace of les Glauges", we may think that the transition between the first and the second cycle occurs very early in the Pleistocene. The advent of the second cycle can result only from profound modifications in the upper part of the Durance basin such as overthrusting movements in the subalpine chains (Digne over-thrust...), or such as the erosive actions due to the first glaciations which deeply marked the Southern Alps (Donau,Gunz ?). Finally, during all the Pleistocene, the Durance river flowed through la Crau directly into the sea. During the Würm climax of regression (Würm III) only, the Durance has been captured by a left bank affluent of the Rhône.
On admet généralement que deux nappes de cailloutis, liées à des cours successifs de la Durance, colmatent la dépression de Crau. Il nous a été possible de distinguer dans chacune de ces nappes plusieurs formations originales et de mettre en évidence des lambeaux de terrasses intermédiaires. Les éventails minéralogiques et pétrographiques de ces formations résultent de deux cycles d'alluvionnement bien différents dont la dualité constitue le caractère majeur de la sédimentation durancienne, aussi bien en Crau que dans la basse et moyenne vallée. Le 1er cycle (Villafranchien inférieur), caractérisé par des cailloutis à calcaires subalpins et smectite dominante, correspond à la "Crau d'Arles ou d'Eyguières" équivalente des très hautes terrasses duranciennes. Des faciès de remaniement de ces formations ont été reconnus en Durance ("Terrasses de la Vermillière") et en Crau ("Terrasse de Terrefranche"). Le 2e cycle (Pleistocène supérieur), caractérisé par des cailloutis à roches endogènes alpines et illite dominante, correspond aux "Crau de Miramas et du Luquier", prolongement des basses et moyennes terrasses duranciennes. Les terrasses intermédiaires (Brays, Entressen, les Glauges) se rattachent au 2e cycle par leurs cortèges minéralogique et pétrographique, mais ont été mises en place avant le Riss. Il semble que la transition entre le 1er et le 2e cycle se soit produite très tôt dans le Pléistocène. Les causes sont à rechercher dans un élargissement du bassin durancien supérieur sous l'effet de facteurs tectoniques et climatiques. Enfin, la permanence du passage du cours de la Durance en Crau est maintenant manifeste.
