Much of the research described here results from the Franco-American collaborative exchange entitled "Aurignacian Genius: Art, daily life and social identity of the first modern humans of Europe", UMI 3199-CNRS-NYU & UMR 5608-TRACES, funded by a three-year grant from the Partner University Fund and the Andrew Mellon Foundation. Additional funding was provided by the Claire G. Goodman and Bertram H. Salwen Funds of New York University. We wish to thank Catherine Cretin, Florent Hautefeuille, Raphaelle Bourrillon, René Castanet and Matthew Sisk for their help at various stages of the project. Unfortunately, Alain Roussot did not live to see this work published. Without his work at the site, we would know almost nothing of its contents and stratigraphy.
1The Abri de la Souquette is a rockshelter located on the western flank of the small valley of Castel-Merle, situated approximately 50 m from the Vézère river (figure 1), in the Dordogne department of south-west France. Celebrated for its extensive Upper Paleolithic deposits, the Castel-Merle valley is one of the numerous karstic dry valleys found in the region, formed by the collapse of a subterranean cavity. Differential erosion rates of the Cretaceous limestone beds in the cliffs flanking the valley resulted in numerous rockshelters, which proved attractive habitation sites during the Paleolithic; the Abri de la Souquette itself faces south-east, is 16 m long and 5 m deep, and today offers a sheltered area of approximately 60 m².
Figure 1 - Left: Localisation of the major Aurignacian sites in the Castel-Merle vallon; right: Map of the Vézère valley in 1910 with the Castel-Merle valley highlighted.
after Hauser 1911
2On the opposing, eastern flank of the Castel-Merle valley lie the seminal Aurignacian sites of Abri Blanchard and Abri Castanet, approximately 150 m southeast of the Abri de la Souquette. Despite its relative lack of renown, however, it is likely that, at the beginning of the 20th century, La Souquette was in fact the richest of all the Castel-Merle Aurignacian sites. In addition, the site once contained a considerable amount of Magdalenian material. An unfortunate series of excavations conducted in the early 20th century, however, removed almost all archeological material from this spectacularly rich site, with only minor vestiges of the original Aurignacian material left. Here we will attempt to reconstruct some of the previous richness by providing a review of the history of archeological investigations at La Souquette and the collections they left behind.
3By 1938, Franck Delage had already described La Souquette as "a perfect example of one of numerous prehistoric sites which have been devastated by excavation for reasons opposed to science" (Delage, 1938: 105). La Souquette had been a victim of antiquarianism and, in a period of less than 40 years, was thought to be almost completely emptied of prehistoric deposits. By the time Delage was writing, there had been no fewer than four different phases of excavation.
4The ‘vicinity of the site’ was first excavated in 1902-1903 by the abbé Michel Antoine Landesque (1838-1905), a priest and geologist (Delage, 1938: 105). Landesque was a reasonably well-regarded figure in his time, but died in 1905 before publishing anything on the site, and so very little is known about his excavations. After his death his collections were broken up and sold all over the world, some to America, simply marked ‘abbé Landesque’, and thus much information was subsequently lost to science. If the death of Landesque was unfortunate for our understanding of the site, the next excavations would be even more so. The site was appropriated between 1903-1910 by a clockmaker and collector of antiquities from Issigeac named Costes, who, working with a local shopkeeper named Letellier, excavated the site. A considerable portion of the site was disturbed, no records were kept, and it appears that only worked flints were retained (Delage, 1938: 105-106). Flint artifacts which did not meet their standards, and seemingly all the bone and ivory industry, were discarded (Delage, 1938: 106).
5In 1910, La Souquette was leased by Antoine Blanchard, the owner, to flint collectors, who in turn sub-leased the site to the Swiss archeologist Otto Hauser, who would excavate much of what remained at La Souquette. Local tradition holds that the goal of the excavation was the acquisition of large Aurignacian blades and of worked flints which might be sold to museums, and Hauser’s workmen were reputedly paid by object recovered. No attention was paid to either spatial or stratigraphic context, and apparently no records kept (figure 2B). Hauser sold large amounts of material to the Museum für Völkerkunde in Berlin (the modern-day Ethnologisches Museum).
Figure 2 - The Abri de la Souquette through the ages.
A: north-facing photo of the Vézère taken from La Souquette; B: disturbed deposits within La Souquette during Hauser’s excavation; C: northwest-facing photo of northern end of La Souquette; D: west-facing photo of La Souquette; E: abri de la Souquette in summer 2011, note Roussot’s section at left.
A-B:Hauser 1911; C: MacCurdy 1921; D: Roussot 1980
6The land was acquired by Marcel Castanet (after whom the Abri Castanet is named) following his return from the First World War in 1919, and he immediately began to investigate the spoil-heaps of previous excavations at La Souquette, recovering huge quantities of discarded material. Although it would come to be known as the ‘Hauser spoil-heap’, the deposits Castanet screened were probably derived from all previous excavations, and they were incredibly rich (figure 3). Castanet was ironically introduced to archeological excavation as a boy during Landesque’s excavations at La Souquette, but had subsequently worked for Louis Didon at the Abri Blanchard, and for Didon and Denis Peyrony at the Abri Reverdit, and he understood the importance of wet-screening for recovery of small items, particularly ornaments.
Figure 3 - A: Aurignacian blade tools from La Souquette recovered by Hauser; B-D: personal ornaments recovered by Marcel Castanet from the spoil-heap of previous excavations; B: perforated ivory sea-shell facsimile; C: perforated steatite pendant with radial incisions; D: perforated red deer vestigial canine.
photos : A: O. Hauser; R. White
7Working with the well-known archeologist, Louis Pradel, Castanet recovered large quantities of flint tools, engraved blocks, pierres anneaux, batons percés, bone and antler points, lissoirs and almost 600 personal ornaments. Given that most of the material recovered by Castanet was of mixed provenience, much of it is difficult to categorize, although some diagnostic Aurignacian and Magdalenian types were recovered.
8The 483 basket-shaped beads (including fragments) recovered may confidently be attributed to the Aurignacian (figure 4), and are accompanied by 187 unfinished bead-blanks. The vast majority are in ivory, with 64 in steatite and another 11 (and 2 blanks) in bone. The perforated fox, wolf, deer and bovid teeth and perforated mollusk shells he recovered may be assigned to either the Magdalenian or Aurignacian, although many of them do exhibit classically Aurignacian perforation techniques. Several pendants also appear ‘Aurignacian’ in style, such as the 5 seashell facsimiles featuring pointillist decoration, a steatite roundel measuring 3.5 cm in diameter, a calcined lignite ‘barrel’ measuring 2.2 cm, and oval and triangular ivory pendants. Delage furthermore attributed to the Aurignacian several fragments of batons de commandement and incised bones based upon technological similarity with material recovered from Aurigacian horizons at Castanet and Blanchard.
Figure 4 - A number of basket-shaped beads in ivory and steatite recovered by Marcel Castanet from previous spoil-heaps at the Abri de la Souquette.
photo: R. White
9The diagnostically Aurignacian lithics recovered by Castanet include large numbers of retouched blades, along with end-scrapers on strangled blades, oblique end-scrapers, double end-scrapers, end-scrapers with burins, large burins with retouched sides, and carinate and nosed end-scrapers (Delage, 1938: 108-109). Castanet also recovered several split-based bone points, along with antler rods indicating in situ manufacture of points (Delage, 1938: 110).
10The Magdalenian material comprises a diverse collection of scrapers, burins, double-burins, perforators, points, knives, a gouger and a micro-burin (Delage, 1938: 115). There is also an extensive microlithic collection, with large numbers of backed bladelets and elongated scalene triangles. Peyrony considered this type characteristic of the Magdalenian II, but Delage sees the assemblage more reflective of the Magdalenian III or IV (Delage, 1938: 117). Barrière (1952) later identified a Tardenoisian-style micro-burin among the Magdalenian IV material, which he thought significant for the origin of that industry.
11Outside of the Aurignacian and Magdalenian, Delage claimed to have identified a Mousterian racloir, along with some Mousterian-like flake tools (Delage, 1938: 108), but these are hardly unknown in the Aurignacian and there is no compelling evidence for a pre-Aurignacian occupation at La Souquette. A single Solutrean laurel-leaf point fragment was also found, apparently in intact deposits in front of the rockshelter, and Delage also regarded a 3 cm long foliate point recovered from the spoil-heap as potentially Solutrean, although no other evidence for a Solutrean occupation has been recovered.
12According to Delage (1938: 108), however, Marcel Castanet did discover and excavate an 80×30×30 cm block of in situ Aurignacian sediment, which revealed some stone tools and several bone or antler points. Some intact Magdalenian deposits were supposedly also found at the southern extremity of the site, while a trench dug 8-10 m in front of the rock-shelter revealed the supposed laurel-leaf point fragment, below a reworked layer that included historic pottery fragments. In fact, Delage himself may have engaged in some small-scale excavations at La Souquette, as there exists a collection bearing his name at the Institut de Paléontologie humaine in Paris (Roussot, 1982a: 2).
13Castanet retained most of this material in the family museum in Sergeac, Dordogne, where much of it remains today. A significant amount of material, however, was sold to the Field Museum of Natural History in Chicago, USA, where one of the authors (RW) has been able to identify 307 individual accessions from the Abri de la Souquette. The majority of the material comprises various lithic materials – mainly flint tools, but also including some engraved limestone or sandstone blocks. 62 accessions of bone or antler, 13 teeth, 3 ivory fragments, 2 shells, 1 echinoderm fossil, 2 fragments of manganese dioxide and 1 piece of red ochre were also identified. Of course, like the material that remained in France, this is not a representative sample of the original archeological record from La Souquette, as archeologists and museums of the period were mainly concerned with acquiring objects thought to be "beautiful" or "typical" of a certain period or region. Castanet applied these same standards, discarding for example, virtually all debitage, cores and river cobbles, as well as most of the fauna.
14By the middle of the 20th century, the Abri de la Souquette was considered something of a lost cause (Sonneville-Bordes, 1960: 105-106). A small number of specialists were aware that it had once contained substantial Aurignacian and Magdalenian deposits, and Castanet’s collections were examined with interest by the likes of Franck Delage and Denis Peyrony: the site itself, however, was considered essentially devoid of further scientific value. In 1980, however, René Castanet, son of Marcel and owner of the site, requested an investigation of the remaining deposits at La Souquette in order to display a stratigraphic section on public visits to the site (Roussot, 1982a: 5). The site was thought to retain little of archeological significance, but test excavations were nevertheless conducted in 1980 by one of the authors (AR), then curator of prehistory at the Musée d’Aquitaine in Bordeaux. Excavations concentrated on a small area at the southern extremity of the rockshelter, which were believed to be the area least disturbed by previous excavations (figure 5). To general surprise, substantial intact deposits were encountered and excavations continued for an additional two years, over a total of approximately 10 m².
Figure 5 - Excavations conducted at the Abri de la Souquette 1980-1982.
A: the site before excavation, photo facing southwest towards intact deposits; B: insertion of test trench into southwestern end of rockshelter; C: cleaned section with Layer 11 highlighted.
photos: A Roussot
15Eleven discrete horizons were identified, one of which (Layer 11) contained Aurignacian material lying directly atop bedrock. These apparently intact Aurignacian horizons were overlain by medieval material, and a medieval silo also truncated the Aurignacian layers and cut into the bedrock towards the back of the rockshelter. Although it was clear that the site had once contained significant Magdalenian deposits, no trace of these was recovered. While it remains possible that these deposits may simply not have extended into this corner of the rockshelter, a considerable amount of reworked lithic material was recovered from the overlying historic layers, and we suspect that the Magdalenian deposits were reworked during the historic occupation of the site (see below). It is unfortunate that no Magdalenian deposits were recovered from these excavations as they provide us with the only stratigraphic profile at La Souquette (figure 6).
Figure 6 - Southwest-facing stratigraphic profile of deposits excavated at Abri de la Souquette in 1982. Highlighted area is shown inset, and shows Layer 11, and the intact overlying Layers 8, 9 and 10. Grid squares = 1 m.
16The excavation was conducted along a meter-square grid, and artifacts recovered in situ had their co-ordinates recorded, while sediments were dry-screened through 2 mm screens. The provenience was recorded for over 850 objects recovered from layer 11, allowing the recent digitization of the excavation archive and projection of materials recovered (figure 7). Unfortunately, the relatively small window of excavation at La Souquette makes generalized assertions about site activities difficult, but these new data nonetheless hint at the potential utility of digitizing old excavation archives.
Figure 7 - Top: Meter-square grid installed at the Abri de la Souquette for object provenience, area of excavation highlighted in red; Bottom: reconstructed projection of material recovered from Abri de la Souquette. Objects re-projected using Abri Castanet/Abri Blanchard grid to enable integration with other Castel-Merle Aurignacian sites. Grid squares = 1 m.
17A large quantity of Aurignacian lithic material was recovered, with 921 flint artifacts and
6 pieces of quartz recovered, 400 of which had their exact provenience recorded. Typological classifications of the lithic tools are listed in table 1, and are dominated by retouched blades, with a significant number of end-scrapers of various types (figure 8). In addition, there were three carinate end-scrapers, and one carinate end-scraper rough-out, often interpreted today as bladelet cores, rather than formal tools in and of themselves. Two conventional bladelet cores were also recovered, but no blade cores.
Table 1 - Sonneville-Bordes (1960) typological classifications of tools found in Layer 11.
Figure 8 - A selection of formal lithic tools recovered from Aurignacian deposits at the Abri de la Souquette.
A: double burin; B: side-scraper; C: end-scraper on retouched blade; D: end-scraper on Aurignacian blade; E-G: carinate end-scrapers; H: double end-scraper on Aurignacian blade.
18Aurignacian and retouched blades can be found right throughout the Aurignacian, however they are most commonly associated with the Aurignacian I, generally referred to today as the Early Aurignacian (Bon, 2002). The same is true of the exclusive use of carinate end-scrapers for bladelet production, as opposed to carinate burins or burins busqués. Given the additional presence – albeit out of context – of a split-based bone point fragment, and the mass of similar material in the Castanet collections, it seems beyond doubt that the La Souquette Aurignacian belongs to this facies. Interestingly, the relative frequencies of the lithic tools at La Souquette closely parallels the rest of the Castel-Merle Aurignacian sites, with significant numbers of retouched blades, a moderate amount of end-scrapers, and a paucity of burins (figure 9). This seems to imply similar activity patterns across the valley, or perhaps indicates a local variant of the Aurignacian complex.
19Faunal identifications are listed in table 2, and are dominated by reindeer. In addition to the taxa listed above, 641 unidentifiable bone fragments, and an additional 455 burnt fragments of 1 cm or less were also recovered from layer 11. The osseous industry from layer 11 included a burnt fragment of a bone or antler point, a fragment of what is described as a baton, and several incised, splintered and percussed bone pieces.
Figure 9 - Relative frequencies of lithic tool type groups across Castel-Merle Aurignacian assemblages.
Table 2 - Faunal identifications as determined by F. Delpech in 1980s (Roussot, 1982a).
20No ornaments were recovered from layer 11, although a perforated red deer canine and a piece of incised ivory were recovered from the backfill of previous excavations (Roussot, 1982a: 14). 142 small fragments of red ochre, weighing 314 g, were recovered, along with two chunks of manganese dioxide and one of yellow ochre. A 30 cm long block of limestone featuring multiple non-figurative engraved lines was also recovered from Layer 11, and may possibly have collapsed from the vault. It may be compared with a similar, although figurative, block recovered by Castanet (figure 10).
Figure 10 - Top: figurative engraved block recovered in spoil-heap at Abri de la Souquette by Marcel Castanet ; bottom: engraved block recovered from Aurignacian deposits in 1981.
photo: R. White; R. Bourrillon
21As mentioned above, significant occupation of the site occurred during historic times. As it seems likely that the absence of Late Upper Paleolithic deposits in the south-western end of La Souquette is due to historic reworking of the rockshelter sediments, it is important to understand these occupations. However, while the prehistoric artifacts from sites like La Souquette were preserved, the overlaying historic layers were not only removed, but also often ignored. Fortunately, the 1980-1982 excavations allow us to understand a little about the site’s historic occupations, and how this likely affected the Paleolithic deposits.
22Several structural features, including "banquettes", post-holes, and roof gutters, were cut into the rockshelter in historic times, and a silo was excavated into the bedrock towards the southern end of the site. Additionally, much of the cliff above the shelter was cut back vertically in quarrying operations, which removed a 5.5 m wide, 5 m high, 1m deep section of limestone (figure 2D, E). This had the effect of weakening the vault of the rockshelter, some of which has since collapsed.
23The historic finds from Roussot’s excavations include 29 bone fragments from modern fauna, 17 pieces of metal, 123 pieces of ceramic, a spindle whorl, and part of a grinding stone. These finds were scattered across the surface as well as through stratigraphic layers 1 to 9/11. Most of the metal was badly corroded iron, although one piece of bronze was also excavated. The corrosion made it difficult in most cases to determine the iron’s original functions, however, some nails were recovered and in other cases evidence that the pieces had at least been worked was preserved.
24The chronology of the historical presence at the site may be determined by the pottery remains, which comprise a homogenous mixture of communal earthenwares of grey, white, and reddish colors. Several large pieces, made of an orangish paste, were refitted and dated to the Carolingian period. In contrast, two glazed fragments from Layer 1 were attributed to the mid and late 14th century, respectively (F. Hautefeuille, pers. comm.). These two fragments derive from the uppermost deposits in the sequence, and likely provide a terminus ante quem for the site’s usage.
25As noted above, in addition to historic artifacts, the 1980s excavations also uncovered a silo; a previously unknown structure inserted into the floor of the southwest corner of the rockshelter (figure 11). This feature truncated Layers 10-11, and extended 20 cm into the bedrock. With a diameter of 1.4 m, the total holding capacity of the silo was approximately 1.54 m3, and historic remains were found in Layer X, the ‘fill’ deposit of this feature.
Figure 11 - Excavation of the silo in 1980-1982, showing truncation of Paleolithic layers and bedrock.
26Silos are a ubiquitous structure in the western medieval world, especially amongst rural populations, and are most simply defined as in-ground pits for grain storage (Conte, 1995: 190). In this capacity, they have served not only medieval populations, but are known as early as Neolithic times (Sigaut, 1979). The artifacts and stratigraphy at La Souquette suggest that the silo here is of an early medieval origin, and was no longer in use by the 14th century. The isolated nature of the silo, as well as the communal nature of the pottery and other artifacts recovered, suggest that La Souquette likely accommodated a small or part-time occupation, either as the home of a family unit or as part of a larger complex that included sites in the immediate area. Regardless of site function, it is important to note that although only a single silo was found at La Souquette, it is not an isolated find; rather, it represents one stage in a large scale agricultural and alimentary process of production, process, and distribution that reflects a vibrant but poorly understood social organization.
27Historic occupations undoubtedly had a great impact on Paleolithic sites, and understanding the cultural activity and context of any prehistoric site must account for all the effects of later populations upon these assemblages. The Abri de la Souquette’s long history of occupation may in fact provide fresh insights into surrounding prehistoric site assemblages, and alerts us to underestimated processes of disturbance. Despite the preoccupation of this paper with prehistoric material, however, it is also important to recognize our responsibility to address, study and preserve later archeological material which we ourselves encounter in our research pursuits.
28Given its current condition, it is difficult to see what can be done to attempt to fit the Abri de la Souquette back together. One possibility is to try to relate it back to its context, and derive inferences from the framework of knowledge we have assembled over the years. The potential relationship of the La Souquette to the Abris Blanchard and Castanet, for instance, may tell us more about the Aurignacian than any one site taken in isolation. The first step in understanding any potential association, however, is to demonstrate a chronological relationship.
29Absolute dating of the Abri de la Souquette began in 2000, when eight pieces from the Roussot and Castanet collections were sent by one of the authors (RW) to the CNRS AMS radiocarbon laboratory in Gif-sur-Yvette, two of which returned successful radiocarbon dates (table 3). Further radiocarbon age estimates were acquired in 2015 using ultrafiltration at the Oxford Radiocarbon Accelerator Unit, and are provided in table 3.
Table 3 - Radiocarbon age estimates on two objects from Layer 11 of the Abri de la Souquette.
30The date of 18790±140 BP (GifA 10054) is clearly incompatible with the object provenience, but it is almost certain that this may be attributed to contamination by a consolidant applied
to the material in the field, rather than infiltration between archeological layers (H. Valladas pers. comm.). The rest of the dates, however, are perfectly coherent with the ultrafiltered dating sequence from the Abri Castanet, which runs from 33650±650 BP through to 32310±190 BP for the Secteur Sud, and 32900±230 BP through to 31960±230 BP for the Secteur Nord (White, Higham, 2008). This is highly suggestive that the Aurignacian occupations at the Abri Castanet and Abri de la Souquette were roughly contemporaneous, and should be analyzed as such.
31In addition to refining our understanding of the chronology of the Castel-Merle Aurignacian, these new dates have important implications for our understanding of the relationship between these sites, which ongoing work will attempt to further. While future excavations of the scant remaining deposits may add to our understanding of this site, the loss of most of the site means comparisons of spatial patterning or activity patterns are unlikely to be adequately realized. What can be compared is the remaining material itself – partial sample though it may be – and future technological analyses of the material collected from La Souquette with that subsequently from Castanet and Blanchard will help us better situate the Aurignacian of the Abri de la Souquette in its local cultural and chronological context.
32What should we do, as archeologists, with an old mess like that left at the Abri de la Souquette? Do we write off these old sites as essentially devoid of or greatly diminished in scientific value? Can technological advances, as well as new theoretical paradigms, glean new meaning from old data? It is certainly true that scarce archeological resources are sometimes better spent on a site with greater integrity, especially in a time of ever more constrained budgets. However, it can sometimes be those key sites that we consider to be ‘pristine’ which benefit most from such contextualization; we may never know exactly what activities were undertaken at La Souquette, for instance, but we may be able to say something about the cultural relationship between the groups occupying that site and those at Castanet. Though once a ‘supersite’ in its own right, the importance of La Souquette may end up being limited to the contribution it can make to our understanding of the geographic placement of Aurignacian habitations on the broader landscape of the Vézère Valley in particular, and the northern Aquitaine in general. We hope that the review presented here has suggested that, despite its troubled history, there is still much to be learned from the Abri de la Souquette.