My experimental research into the working of ivory falsified several naive misconceptions about proboscidean tusks as raw material. My first mistake was to presume that modern African elephant tusk was a suitable experimental analogue for woolly mammoth tusk. Although, superficially, the differences do not appear great, the two forms of ivory exhibit quite different fracture patterns that are the product of significant differences in the angle of intersection of Schreger lines (known as Schreger angles).
Contrary to my initial presumption, fresh tusks proved extremely difficult to break into, and to reduce to usable parts. Fresh ivory does not fracture along the large scale concentric laminae observable in prepared sections, as we had so naively imagined. Nor can a fresh elephant tusk be sectioned longitudinally by beginning a split at the thin, hollow proximal end (pulp cavity) and propogating it toward the distal end of the tusk.
Neither Alaskan permafrost ivory collected in the 1920's, nor fresh African elephant ivory could be worked effectively by direct percussion. Attempts at direct percussion produced the same results that one might expect of hitting a stout piece of hardwood with a hammerstone! Indeed, there are distinct similarities in the response of hardwood and ivory to percussion and wedging, based upon similarities in structure and grain. Sizeable flakes were removed by percussion only where the tusk had already been partially split by dessication.
Neither fresh nor artificially dessicated (kiln-dried) African elephant ivory could be worked by splitting-and-wedging. Upon dessication, elephant ivory developed incipient concentric fractures that conformed to the broad internal laminae of the tusk. However, these were too poorly developed to serve as points of access for wedges. Overheating in a kiln made modern elephant ivory so brittle that it was breakable by hand, and structurally weaker than many 35,000 year-old Aurignacian artifacts.
At the time I received it, the medial segment of Alaskan permafrost mammoth tusk showed a quite developed concentric fracture at a lamellar boundary near its external surface. Attempts to exploit this dessication fracture by percussion-driven wedging from the distal extremity of this medial segment produced broad flakes that hinged outward part way down the segment's length. However, a split could not be propogated through the entire length of the segment, presumably because the lines of force were directed outward when they encountered the interfaces between "superimposed cones" of ivory. With enough labor however, it is certainly possible to scrape, grind and polish such large flakes of fresh ivory into desired forms. Thus, their employment by Aurignacian ivory-workers cannot be entirely excluded.
In addition, the slightly dessicated Alaskan permafrost tusk showed radial fractures that cross-cut the concentric laminar fractures (Figure 8). While these radial fractures did not provide an effective purchase for percussion-driven wedges to split the tusk longitudinally, this is probably attributable to the extremely "fresh" state of the permafrost tusk. Similar attempts with more highly dessicated lamellar fragments of mammoth tusk led to considerable success in creating long, workable splinters by longitudinal splitting and wedging.
Splitting-and-wedging constituted a fundamental Aurignacian strategy (Knecht 1993) for working organic materials, but I conclude that they are/were not especially appropriate to the structure of fresh or even slightly dessicated proboscidean tusks. The use of sub-fossil tusks as raw material has been hypothesized previously by Hahn (1986) for the German Aurignacian and by Phillipov (1983) for the Upper Paleolithic of the Russian Plain. I think it nearly certain that, absent some form of artificial drying or purposeful curing of fresh tusks, "sub-fossil" or at least mammoth tusks some years old were sought as raw material by Aurignacian ivory workers.
Even then, the production of long rods of ivory by splitting and wedging of tusks is not suggested by the structure of ivory, and is difficult to achieve. These rods seem to have been a pre-existing goal of Aurignacian ivory workers even if they required the solution of significant mechanical problems. My best explanation for this determination by Aurignacian I ivory workers to obtain pencil-like rods of ivory is that it is a technique highly conducive to standardization. Cylindrical rods of roughly the same diameter lead easily to the production of standardized bead-blanks and ultimately standardized beads. I have shown elsewhere (White 1989) the high degree of standardization in Aurignacian beads, a standardization that I believe to be motivated by the anticipated arrangement of uniform, sewn-on beads on garments. In other words, the visual impact of Aurignacian beads was not as individual objects, but as arrangements constituted of uniform elements.
Laboratory experiments using Aurignacian blades of Bergerac flint, a fine limestone grinding stone, powdered red ocher and liberal applications of water (essential to softening and lubricating the surface of the ivory), produced stigmata and highly polished surfaces similar to those observed on Aurignacian ivory beads. Significantly, while most Aurignacian beads show traces of hematite, few if any of them are profoundly stained. In my experience working permafrost mammoth tusk, indelible staining only occurred when powdered hematite was mixed with fat or oil. The superficial nature of hematite deposits on Aurignacian beads supports the use of water rather than fat as a softening/lubricating agent.
In my experience, there is no archaeological evidence before Sungir (dating to ca. 28-25,000 BP or earlier) for the preparation of whole tusks by softening (heating, boiling), which would have allowed the ivory to have been more easily worked. Even at Sungir however, the operational chain for bead production (White 1993) seems not to have employed thoroughly softened ivory. Simple soaking of tusks in water has only superficial effects. However, once the tusk is reduced to much thinner fragments, such soaking can penetrate the entire thickness, making drilling, scraping and gouging much easier. This soaking also works well on subfossil tusk fragments. Comparison of our experimental sample with actual Aurignacian production debris indicates clearly the use of water in the final stages of bead production.
In my experience with extant archaeological collections, there is no evidence in the Aurignacian for the oft-cited "groove-and-splinter" technique, so common in later periods. While grooving and splintering of fresh tusks is a feasible (if enormously tedious and time-consuming) approach, the splinters that exist in Aurignacian assemblages show no traces of having been incised out of the surface of the tusk. The French sites that have produced the greatest quantities of ivory ornaments and production débris have yielded few significant tusk segments that might yield insight into Aurignacian approaches to tusk reduction.