lable at ScienceDirect

Journal of Archaeological Science 42 (2014) 1e14

Contents lists avai

Journal of Archaeological Science

journal homepage: http : / /www.elsevier .com/locate/ jas

Ivory debitage by fracture in the Aurignacian: experimental andarchaeological examples

Claire E. Heckel a,*,1, Sibylle Wolf b

aDepartment of Anthropology, New York University, USAbAbteilung für Ältere Urgeschichte und Quartärökologie, Universität Tübingen, Germany

a r t i c l e i n f o

Article history:Received 11 February 2013Received in revised form17 October 2013Accepted 21 October 2013

Keywords:AurignacianMammoth ivoryOsseous material technologiesExperimental archaeologyIvory structure

* Corresponding author. Present address: 25 WaverlUSA. Tel.: þ1 704 463 1959.

E-mail addresses: ceheckel@gmail.com, claire.heck1 Temporary mailing address (until 12/13): PO Box

USA.

0305-4403/$ e see front matter � 2013 Elsevier Ltd.http://dx.doi.org/10.1016/j.jas.2013.10.021

a b s t r a c t

The recent focus on methods of osseous material transformation in the study of Upper Paleolithictechnologies has shown that approaches to these materials vary between phases of the Upper Paleolithic.In the absence of the groove-and-splinter technique of blank extraction first widely documented in theGravettian, production of ivory, bone, and antler blanks in the Aurignacian relied on processes of splittingand percussive fracture. The technological treatment of bone and antler in Aurignacian contexts hasbenefitted from renewed attention, but ivory processing and blank-production remains poorly under-stood in spite of the unique place that ivory occupies in many Aurignacian assemblages. In order to clarifythe diagnostic features of ivory debitage, a series of experiments was conducted to produce ivory flakesunder varying knapping conditions. These diagnostic features are products of the application of force tothe complex internal morphology of proboscidean tusks, as explained in this article. Improved criteria forthe identification of ivory flakes and manufacturing byproducts in the archaeological record are pre-sented, and are illustrated with examples from two Aurignacian sites well known for ivory processing:Abri Castanet (Dordogne, France) and Hohle Fels Cave (Swabian Jura, Germany). A better understandingof ivory structure and improved identification of the products of ivory debitage in the Aurignacian willaid in the recovery and analysis of ivory artifacts and further efforts to reconstruct technological ap-proaches to this complex material.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

The systematic exploitation of osseous raw materials is adefinitive characteristic of the Upper Paleolithic. Approaches toexploitation of these materials changed throughout prehistory inways that coincide with the acknowledged technocomplexes of theUpper Paleolithic (Aurignacian, Gravettian, Magdalenian, etc). Thediverse techniques used to process these raw materials haverecently received increased attention. Many studies have exploredcontext-specific processes of transforming osseous materials intoartifacts based on comparative studies of experimental archaeologyand the archaeological record (Averbouh, 2000; Christensen, 1999;David, 2007; Khlopachev and Girya, 2010; Knecht, 1991, 1993;Tartar and White, 2013; Tejero et al., 2012). Ideas of linear evolu-tion in osseous technology, marked by the impression that

y Place, New York, NY 10003,

el@nyu.edu (C.E. Heckel).959, Misenheimer, NC 28109,

All rights reserved.

techniques became increasingly sophisticated and efficient overtime, have been challenged by this new body of research. Shifts inapproaches to these materials over the course of the Upper Paleo-lithic have been proven to be more cyclical, and closely related toother aspects of Paleolithic technology such as available lithictechnologies (Baumann and Maury, 2013; Pétillon and Ducasse,2012).

Ivory artifacts constitute one of many developments in ap-proaches to raw materials in the Aurignacian, appearing inarchaeological assemblages from Cantabria to the Russian Plain(Álvarez-Fernández and Jöris, 2007; Vanhaeren and d’Errico, 2006).While ivory was used to produce utilitarian artifacts such as pro-jectile points, awls, and beveled tools, themost numerous andwell-known ivory artifacts of the Aurignacian are of a symbolic nature:figurines of people and animals, thousands of beads and pendants,and even musical instruments (Conard, 2003b, 2009; Floss, 2007;Hahn, 1986; Malina and Ehmann, 2009; White, 1997, 2007; Wolf,2013). Experimental ivory work has a long history (Christensen,1999; Hahn, 1986; Hahn et al., 1995; Khlopachev and Girya, 2010;Malina and Ehmann, 2009; Semenov, 1964) and some of thisresearch has focused on techniques for reducing tusks or tusk

Fig. 1. A schematic diagram of the structural features in an ivory tusk as seen intransverse view: the radial microlaminae (A), the circumferential growth rings (B), andthe Schreger Pattern (C). A diagram of the planes and axes within a tusk segment isprovided to orient the reader (D). These features are (AeC) are pictured separately forclarity, but overlie and intersect with each other in the cross-section the tusk.

C.E. Heckel, S. Wolf / Journal of Archaeological Science 42 (2014) 1e142

segments into workable blanks. This preliminary phase of artifactproduction remains more poorly understood than subsequentstages of artifact fabrication. This article explores the evidence forivory debitage by fracture in Aurignacian contexts and the rela-tionship between ivory’s complex structure and the fracture pat-terns indicative this activity.

Debitage has been defined as: “A term conventionally used todenote the intentional knapping of blocks of raw material, in orderto obtain products that will either be subsequently shaped orretouched, or directly used without further modification” (Inizanet al., 1999: 140). For most lithic materials, knapping is the pri-mary method of raw material reduction. For osseous materials, theterm “debitage” has been adapted to include a number of additionaltechniques for the reduction of a block of material into productsthat can be subsequently reshaped or directly used (Averbouh,2000; Tejero et al., 2012) These techniques have been groupedinto three primary approaches for the reduction of osseous mate-rials in the Upper Paleolithic (Averbouh, 2000; Averbouh andPétillon, 2011).

- Debitage by segmentation: a transverse operation, the reductionof a block of material into segments, often by such techniques astransverse or circumferential grooving, sawing, or chopping(Averbouh and Pétillon, 2011:41).

- Debitage by extraction: the extraction of a longitudinal segmentof predetermined size and shape from the exterior surface of thematerial (Averbouh and Pétillon, 2011:41). The technique of“groove-and- splinter” (double rainurage in the widely adoptedFrench terminology) by which parallel longitudinal grooves arecarved to extract a blank is a well-known example.

- Debitage by fracture: fracture of a block of material by knappingin order to produce flakes (Averbouh and Pétillon, 2011: 41). Theprocess of splitting and wedging, also common in the Aurigna-cian (Knecht, 1991) is another type of debitage by fracture.

These approaches are not mutually exclusive, but it has beennoted that for certain periods of the Upper Paleolithic, one is oftenpredominant in osseous assemblages while others are largely orentirely absent. Debitage by extraction, for example, is widelyknown in Gravettian contexts (Goutas, 2009), while segmentationand fracture aremore commonmethods in the Aurignacian (Liolios,1999; Tartar, 2009; Tejero et al., 2012; White, 1997).

Distinctive tool traces are frequently evident in cases of debitageby segmentation and debitage by extraction, but the products andbyproducts of debitage by fracture can be difficult to detect inosseous materials, especially when the features diagnostic of thistechnological process are not clearly defined. It has been demon-strated that the morphology of fracture planes on purposely frac-tured osseous materials can be used to identify osseous-materialexploitation in the archaeological record, and even to identifyspecific techniques of fracture (Averbouh, 2000; Averbouh andPétillon, 2011; Baumann and Maury, 2013; Tejero et al., 2012).When these diagnostic features are familiar to the analyst, flakemorphology can indicate osseous material processing even in theabsence of more commonly recognized tool traces.

Debitage by fracture has been most extensively discussed inreference to antler-working in the Aurignacian (Tejero et al., 2012)and Late Upper Paleolithic (Averbouh and Pétillon, 2011; Baumannand Maury, 2013; Pétillon and Ducasse, 2012). In these cases,experimental and archaeological research have improved therecognition of antler flakes produced by debitage by fracture in thearchaeological record, and contributed substantially to currentunderstandings of technological processes in the Upper Paleolithic.With this article, we hope to add to these growing discussionsthrough experimental and archaeological examples of ivory

debitage by fracture in the Aurignacian. An understanding of thestructure and mechanical behavior of proboscidean ivory pairedwith experimental debitage by fracture aids in the identificationand interpretation of ivory flakes in archaeological contexts. Theaim of this article is therefore three-fold: 1) to demonstrate thediagnostic features produced by experimental debitage by fractureof ivory; 2) to contextualize these features in terms of the complexinternal structure of ivory; and 3) to present archaeological evi-dence for ivory debitage by fracture in early Aurignacian levels fromHohle Fels Cave (Swabian Jura, Germany) and Abri Castanet (Dor-dogne, France).

2. Characteristics of ivory and ivory flakes

2.1. Structural features of ivory

In extant proboscideans and their extinct relatives, “the per-manent tusks are composed of a highly modified dentinecompletely unique in structure and which alone is properly calledivory” (Saunders, 1979: 56). This “modified dentine” is unique bothchemically and structurally. All osseous raw materials are rigidbiological composites composed of a network of collagen fibersembedded in a mineral matrix of hydroxyapatite. The mineralmatrix in proboscidean ivory is not true hydroxyapatite, but amaterial very similar to hydroxyapatite in which there is a tenpercent substitution of magnesium for calcium within the apatitecrystals. These crystals are smaller than those that make up themineral matrix of antler and bone, a fact that contributes to therenowned fineness of ivory (Su and Cui, 1999). Compared to boneand antler, ivory is a highly homogenous material. Except for a thinlayer of enamel at the tip of the tusk (which often wears off in thefirst several years of the animal’s life) and a thin layer of cementumcovering the surface of the tusk, the tusk presents a solid mass ofmodified dentin. The apparent homogeneity of ivory, however,masks a remarkable structural complexity (Locke, 2008) whosehierarchical arrangement makes ivory a truly unique material

C.E. Heckel, S. Wolf / Journal of Archaeological Science 42 (2014) 1e14 3

(Banerjee and Eckmann, 2011; Saunders, 1979; Su and Cui, 1997,1999; Zhang et al., 1993).

The nature of the different structural components and of theirrelationships to one another are still not fully understood, and arethe subject of ongoing research in a variety of fields, includingbiology and materials science (Jakubinek et al., 2006; Locke, 2008;Virág, 2012).With regard to ivory’s properties as a rawmaterial andits response to percussive force, there are three basic structuralfeatures that are of interest: the radial microlaminae, the circum-ferential growth rings, and the Schreger Pattern. These three fea-tures operate on three interlocking planes (radial, circumferential,longitudinal), reinforcing the strength of the material andcontributing to the patterns that develop on fracture planes whenthe material is broken. The radial microlaminae (Fig. 1A) aremicroscopic sheets extending from the tusk axis to the cementume

dentine junction (CDJ) that house the dentinal tubules, which arefiner andmore closely packed in ivory than in other dental material(Saunders, 1979). The growth rings (Fig. 1B) form regularlythroughout the animal’s lifetime, as new dentine is deposited at thesurface of the CDJ. In cross-section these appear as rings, thoughthey are in fact a series of stacked cones, as the tusk tapers at thedistal end. The interface between some growth rings is macro-scopically visible. As ivory ages, and especially when it is subjectedto fluctuations in heat and humidity, it begins to separate along theinterfaces between both the microlaminae and the growth rings(Fig. 7A).

The Schreger Pattern (Fig. 1C) is the product of a complexstructural feature that is the subject of some debate. Oftendescribed as a series of intersecting arcs of darker and lighter ma-terial, this pattern has more recently been observed microscopi-cally and described as a series of adjacent rhomboid andrectangular sections (Virág, 2012). This research suggests that thedentinal tubules in ivory are actual sinusoidal, creating a wave-likepatternwhose phase-shifts contribute to the contrasting colors thatcompose the Schreger Pattern. Much remains to be learned aboutthe origins and mechanical function of the Schreger Pattern, butVirág (2012) observes that the structures related to the SchregerPattern are visible as a series of longitudinally oriented ridges anddepressions on naturally broken ivory surfaces. These ridges aresmall but macroscopically visible, and will be familiar to specialistswho have examined archaeological ivories.

Fig. 1 shows a schematic cross-section of a tusk and illustratesthe radial microlaminae (Fig. 1A), the circumferential growth rings(Fig. 1B), and the Schreger Pattern (Fig. 1C). In the figure, these

Fig. 2. An illustration of a regular ivory flake, presenting many features defined in theterminology of lithic analysis. This flake, (also pictured in Fig. 15D) from the Auri-gnacian layer III at Hohle Fels Cave, presents clear ripples of percussion, as well as anegative of removal on the dorsal surface and a striking platform. (Illustration: R.Ehmann). The terms “basal end” and “terminal end” have been selected over theanalogous “proximal end” and “distal end” so as to avoid confusion with theanatomical proximal and distal ends of the tusk.

features have been separated so as to be clearly visible, but onemust imagine them superimposed upon one another, creating acomplex network of structural features that serve to reinforce thestrength of the tusk. When percussive force is applied to ivory, theforce breaks through and across these structural features, creatingcomplex fracture planes that are diagnostic of debitage by fracture.

While there are slight differences in the structure of mammothand elephant ivory (Espinoza and Mann, 1992), the overall struc-tural features such as radially arranged microlaminae, circumfer-ential growth rings, and the Schreger pattern are present in bothmaterials. The goal of these experiments was not to compare theresponse of elephant and mammoth ivory to debitage, but todemonstrate the range of features that occur on proboscidean ivorywhen it is fractured by percussion. The angles created by theSchreger Pattern in elephant ivory are more acute than those inmammoth ivory. This might alter the detailed appearance of thefracture features in each respectivematerial, but it does not seem toaffect their occurrence in our samples. As will be shown, no sub-stantial differences in these features were observed between themammoth ivory sample and the elephant ivory sample (see Figs. 9e11).

2.2. Ivory flake morphology and fracture features

In the context of stone-knapping, Inizan et al. (1999:141) define‘flake’ as “A general term for a fragment of hard stone that is

Fig. 3. A schematic diagram of a terraced surface in cross section (A) and a photographof an experimental mammoth ivory flake that presents an extensive terraced surface.The stacked layers that form the terraced surface are created as force travels across thegrowth rings in a stepping-down pattern, and occur parallel to the longitudinal axis ofthe tusk.

Fig. 4. A schematic diagram of a sawtooth fracture feature (A) and a photograph ofseveral areas on an experimental mammoth ivory flake where the pattern occurs.These features always occur transverse to the longitudinal axis of the tusk.

C.E. Heckel, S. Wolf / Journal of Archaeological Science 42 (2014) 1e144

removed. .. [that] does not imply a particular morphology, a specificuse, or particular dimensions.” Following their definition, we usethe term ‘flake’ to describe any fragment removed from a piece ofivory by percussion, regardless of the morphology, eventual use, ordimensions. Some of these flakes have a more classic, or regular,morphology familiar from knapped stone (Fig. 2), while otherspresent a more irregular morphology that little resembles classiclithic flakes. For regular ivory flakes, terms from the analysis oflithic artifacts can be applied (see Fig. 2). Such “regular” flakes wereproduced in the experimental series, and are present in thearchaeological record (see Sections 4 and 5.2 of this paper). Morefrequently, though, the morphology of ivory flakes is unpredictable

Fig. 5. Two variations on tongued fractures, dihedral (A) and “W”-shaped (B), shownin diagrams on the left, and on experimental mammoth ivory flakes on the right.

and varies widely, as examples from the experimental seriesdescribed in Section 4 illustrate. The fracture features present on allof the experimentally produced flakes can be explained in terms ofthe internal organization of ivory discussed in Section 2.1.

Previous research (Heckel, 2009b) has defined a number offeatures present on experimentally fractured ivory flakes. The ex-periments described in this article have added to that list, whichnow includes the following features: bulbs of percussion, ripples ofpercussion, smooth surfaces, terraced surfaces, saw-tooth fractures,tongued fractures, and impact scars. Bulbs and ripples of percussionare terms borrowed from the terminology of lithic technology, andhave the same meaning and a similar appearance when applied toivory flakes (see Fig. 2). Smooth surfaces, most often occurring atthe interface between two natural surfaces (such as adjacentgrowth rings) are present in isolated areas on the surface of flakes,among areas presenting other fracture features such as terracedsurfaces. Smooth surfaces dominate on fragments of ivory that havedelaminated and fractured over time due to natural processes.

Terraced surfaces are unique to fractured ivory, and reflect themorphology of the proboscidean tusk. The terraced surface resultswhen the fracture force traverses layers of material in a stepping-down pattern, creating visible overlying layers, or steps. Fig. 3 il-lustrates a terraced surface in a schematic drawing as well as aphotograph of an experimentally produced flake. As will be dis-cussed in Section 5.1, terraced surfaces and saw-tooth features areparticularly indicative of fracture by percussion, as they indicatethat force has propagated by traversing structural features presentin the tusk.

Saw-tooth features (Fig. 4) are thin fracture planes that present arepeating, angular, “toothed” profile, much like the teeth of a saw.Similar features have been observed on intentionally fracturedantler by Averbouh (2000: 79), who employed the analogousFrench term “dent de scie.” Saw-tooth features have been previouslyidentified on worked ivory (Heckel, 2009b) and antler (Averbouh,2000; Tartar and White, 2013). On ivory, these features are muchfiner and more regularly repeating than on bone and antler. Theycan occur at the termination of the flake or anywhere on theventral, dorsal or lateral surfaces of the flake, depending upon itsmorphology and orientation; saw-tooth features always occurtransverse to the longitudinal axis of the tusk. Two or more rows ofsaw-tooth features can be superimposed in successive layers. Fig. 4presents a schematic drawing of a saw-tooth feature as well as saw-tooth features apparent at several positions on the ventral surfaceof an experimentally produced ivory flake.

Tongued fractures are commonly known in osseous technology,and often occur on bone and antler fractured by bending(Averbouh, 2000; Tartar and White, 2013). The tongued fracturesurface results from the tearing away of a section of material fromthe adjoining body of material, creating a pointed or jagged

Fig. 6. Impact scars on the dorsal surface of a mammoth ivory flake from the firstexperimental series (A) and on the platform of an elephant ivory flake from the secondexperimental series (B).

Fig. 7. Materials for the first experimental series. 7A shows the transverse surface of the mammoth tusk before removal of the experimental sample. In image 7B, the tusk segmentis shown in profile, with the removed sample (b). Negatives of previous removals are visible (a), and the direction of removal (c) is indicated. The red arrows bound the negativefrom the removal of the experimental sample that is pictured in detail in 7C. 7D shows the ivory sample, a quartzite hammerstone, and the limestone anvil used in the experiment.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

C.E. Heckel, S. Wolf / Journal of Archaeological Science 42 (2014) 1e14 5

protrusion on one segment, and the negative of this removal on theother (Fig. 5). As initially observed by (1985), these features can bedihedral or occur in a “W”shape. Bone and antler “tongued pieces”(French “pièces à languette” or “langues de chat”) are common by-products of the Aurignacian production of split-base points(Knecht, 1993). Tongued features occur in a variety of shapes andsizes on ivory flakes, most often at the termination of the flake.

Impact scars are defined as visible points of impact between theivory surface and the percussor. These scars most often take theform of localized points of crushing, as seen in Fig. 6. Impact scarsare not visible on all of the experimentally produced flakes, and theuse of rounded hammerstones and pebbles probably partlyresponsible for the low number of visible impact scars (Galán et al.,2009). Ivory is difficult to fracture and requires repeated forcefulblows. A number of experimental specimens therefore bear impactmarks at points where no fracture occurred. Conversely, many ofthe samples analyzed bear no macroscopically visible impact scars.

Some of these features (such as bulbs of percussion) can standalone as indications of ivory debitage by fracturewhile others (suchas smooth surfaces) cannot. As we shall demonstrate, however,these features do not occur in isolation on ivory that has beenfractured by percussion, but rather co-occur in various combina-tions. Ivory fracture does occur naturally as a product of animalbehavior (foraging and fighting), which can produce analogousfeatures of fracture. This is discussed in more detail in Section 5.2 ofthis paper.

Fig. 8. Diagram of the rectangular block of elephant ivory used in the second exper-imental series. The initial block measured 30 � 2.5 � 2.9 cm. It was sawed into twosegments of 15 cm, one of which was frozen and knapped. The other 15-cm segmentwas knapped at the outdoor temperature of 2� Celcius at one end. A segment 7 cm inlength was then sawed off and heated at 30� Celcius for 50 min before being knapped.The photograph in the upper right shows the knapping process for the frozen segmentwith the block (A) and the first removed flake (B). The materials used for knapping theblocks are labeled with lowercase letters a-e (a ¼ quartzite hammerstone,b ¼ quartzite pebble, c ¼ antler billet, d ¼ antler punch, e ¼ sandstone slab used as ananvil).

3. Experimental ivory debitage

3.1. Goals of the experiments

Two distinct experimental series were carried out, one by eachof the authors. The focus of the first experimental series (CH) wasto examine the morphology of ivory fragments produced byrepeated non-preferential blows to a sample of mammoth ivorywith quartzite hammerstones. This approach was designed toproduce a variety of flake morphologies and fracture features by

Fig. 9. Five mammoth ivory flakes produced by the first experimental series. Each flake presents a variety of features diagnostic of debitage by fracture. The numbers in the imageindicate the sample number from Table 1.

Fig. 10. Four elephant ivory flakes produced from the frozen sample in the second experimental series. The numbers in the image indicate the sample number from Table 3.

C.E. Heckel, S. Wolf / Journal of Archaeological Science 42 (2014) 1e146

the application of direct percussion from several orientations, inorder to establish the full range of diagnostic features likely tooccur with percussive fracture. Such signs of percussive fracture onfragments recovered archaeologically would indicate intentionaldebitage of ivory rather than ivory breakage through taphonomicand other natural processes. A secondary goal was to observe thekind of impact marks left by repeated direct percussion with arounded hammerstone, in order to test whether such impactmarks consistently occur with this approach to debitage byfracture.

Fig. 11. Four elephant ivory flakes produced from the untreated sample in the second expproduced in the first series. The numbers in the image indicate the sample number in Tabl

The second experimental series (SW) took an approach ofpreferentially oriented, controlled blows to remove flakes in a di-rection parallel to the longitudinal axis of the tusk with a variety ofpercussors. This technique more closely resembles approaches tocontrolled flint-knapping. In this series, pre-formed blocks ofelephant ivory were used. One goal of this series was to establishthe range of features visible on flakes produced by controlledknapping of ivory blocks. Another aimwas to examine the effects oftemperature on the process and products of ivory-knapping. To thisend, one sample was heated and another frozen in order to

erimental series. These flakes are thinner and more elongated in general than thosee 2. Number 5 is a transverse flake, while all others are longitudinal.

Fig. 12. Four small elephant ivory flakes produced from the heated sample in the second experimental series. The largest of these flakes measured 1.7 cm in length, and the majoritywere under 1.5 cm. These flakes were not analyzed for fracture features.

Table 1Occurrence of each fracture feature on the mammoth ivory flakes produced in the first experimental series. “1” indicates the presence of the feature, “0” indicates its absence.

Specimen Sawtooth Bulb Terraced Impact Smooth Tongued Ripple Total

1 1 0 0 0 1 0 0 22 1 0 1 0 1 0 0 33 1 0 1 0 1 0 0 34 1 0 1 0 1 0 0 35 1 0 1 0 1 1 0 46 1 0 1 1 1 0 0 47 1 0 1 0 1 1 0 48 1 0 1 0 1 0 1 49 1 0 0 1 1 1 0 410 1 0 0 1 1 1 0 411 1 0 1 0 1 1 0 412 1 0 1 1 1 0 0 413 1 0 1 0 1 1 0 414 1 0 1 0 1 1 0 415 1 0 1 0 1 1 0 416 1 0 1 1 1 1 0 517 1 0 1 1 1 1 0 518 1 0 1 1 1 1 0 519 1 0 1 1 1 1 0 520 1 0 1 1 1 1 0 521 1 0 1 1 1 1 0 522 1 0 1 1 1 1 0 523 1 1 1 0 1 1 0 524 1 1 1 1 1 0 1 625 1 0 1 1 1 1 1 626 1 0 1 1 1 1 1 6Total: 26 2 23 14 26 18 4 Avg:Percent: 100 8 88 54 100 69 15 4.35

C.E. Heckel, S. Wolf / Journal of Archaeological Science 42 (2014) 1e14 7

examine the effects on the outcome of the knapping process, incomparison to an unheated control sample knapped at an ambienttemperature of 2� Celsius.

3.2. Materials and methods

3.2.1. Experimental series 1: mammoth ivoryThe ivory sample for the first experimental series was removed

from a segment of mammoth tusk de-accessioned from theAmerican Museum of Natural History to the care of Dr. RandallWhite at New York University. The segment is a medial section ofmammoth tusk sawed cleanly at both ends before its accession bythe Museum, forming a large cylinder approximately 15 cm inmaximum diameter and 30 cm in length (see Fig. 7A and B). Themuseum’s accession notes indicate that it was recovered in the late1920s from a permafrost context in Alaska and has an estimated ageof 28,000 BP. As is commonwith ivory as it is exposed to changes in

temperature and humidity, the tusk has begun to separate alongthe circumferential and radial interfaces (Fig. 7A). In spite of this,the tusk is in very good condition, and successful removal of theexperimental sample required several forceful blows with a ham-merstone. Specifically, direct percussion with a rounded quartzitehammerstone was applied parallel to the longitudinal axis of thetusk segment, resulting in the removal of awedge-shaped fragmentapproximately 9 cm long, 3.5 cm wide, and 3 cm at its thickestpoint, near the platform of removal (Fig. 7B and D). This componentof the experiment was performed in the Paleolithic ArchaeologyResearch Facility at New York University.

The subsequent stages of the experiment were conducted dur-ing the 2011 excavation season at Abri Castanet/Abri Blanchard(Sergeac, France). The process of experimental fracture was con-ducted on a three-meter by four-meter area covered with blackplastic sheeting to aid in the recovery of fragments, and theexperimental process was filmed for later analysis. Thematerials, as

Fig. 13. The blocks of elephant ivory at the end of the experimental procedures. Flakenegatives are clearly visible and arrows indicate the direction of flake removal.Following knapping, each piece was placed on the ground and struck with the ham-merstone in order to fracture it completely. The frozen (A) and heated (C) samples splitlongitudinally, while the untreated sample (B) broke transversely.

C.E. Heckel, S. Wolf / Journal of Archaeological Science 42 (2014) 1e148

arranged for the experiment, are pictured in Fig. 7D. Quartzitecobbles collected from the surrounding area were used as ham-merstones. These cobbles are deposited in the area by fluvialtransport and their presence, with impact marks from use in per-cussion, has been observed in Aurignacian context at Abri Castanet(White, 2008). A block of local limestone (40 � 40 � 15 cm) wasused as an anvil, based on the utility of a stone anvil in previousivory fracture experiments (Heckel, 2009a,b). Because one aim ofthe experiment was to observe eventual impact marks from thehammerstone, a layer of thin plastic foam was positioned betweenthe limestone anvil and the ivory fragment. This prevented anypotential confusion of impact marks from the anvil with those fromthe hammerstone. With the ivory sample placed on the limestoneanvil, repeated blows with the quartzite hammerstones wereapplied to remove flakes and fragments from the sample. Theorientation of the ivory sample was regularly changed to facilitatethe removal of fragments. Flakes of quartzite, limestone, and ivorywere all byproducts of the process and their spatial distribution inthe experimental area was documented. The fragments resultingfrom the experiment were collected and separated by material.

3.2.2. Experimental series 2: elephant ivory at varyingtemperatures

The elephant ivory for this experimental series was confiscatedby the Agency for Nature Conservation (Amt für Naturschutz) inBonn, Germany. The ivory appeared to be in an excellent state ofpreservation, but there is no record of how long or under whatconditions it was stored before seizure by the authorities. Thesample was a longitudinal tusk segment rectangular in section(30 2.9 � 2.5 cm), cut from an elephant tusk by agents of the Amtfür Naturschutz before being transferred to the author (SW). It wassawed transversely into equal halves, and one of these halves wasagain divided, yielding three samples: one to be heated, one to befrozen, and one “control,” which was knapped at the ambienttemperature (Fig. 8). The half to be frozen was subjected to atemperature of �18� Celsius for 26 days. This is the maximumtemperature at which Khlopachev and Girya (2010) observed thatlarger segments of ivory were easily fractured. One portion of thesubdivided unfrozen half was heated for 50min at a temperature of30� Celsius, and the other untreated.

This experimental series focused on blank production throughmore controlled knapping techniques (involving surface prepara-tion, preferential orientation of the blank, direct hard percussion,direct soft percussion, and indirect soft percussion) by an experi-enced flint-knapper who has also worked with osseous raw ma-terials. The tools utilized in knapping were a quartzitehammerstone, two quartzite pebbles, an antler billet, and an antlerpunch (Fig. 8aee). Most flakes were removed longitudinally,though the knapper did attempt transverse removals in some cases(for example, flake number 5 in Fig. 10). After a number of flakeshad been removed from each specimen by controlled knapping,each block was placed on the floor and subjected to several non-preferential blows with the quartzite hammerstone, which resul-ted in splitting of the block in every case.

3.3. Analysis of fracture features

Though the experiments were performed separately, the sameprotocol was used for analysis of the fracture features on the flakesresulting from both series. Flakes measuring 1.5 cm and greaterwere analyzed, and no flakes meeting this size requirement wereexcluded. Each fracture feature defined above was scored on apresence/absence basis. The authors examined the flakes together,counting as “present” only those features that were agreed to bepresent.

Additional observations made during the experimental pro-cesses are reported in Section 4.

4. Results

In reporting the results of these experiments, we first describeboth the processes of debitage and the immediate products of theseprocesses for all four components of the experiment (mammothivory, frozen elephant ivory, untreated elephant ivory, heatedelephant ivory). The morphology of the ivory flakes and the various

Table 2Occurrence of each fracture feature on the elephant ivory flakes produced on the frozen ivory sample. “1” indicates the presence of the feature, “0” indicates its absence.

Specimen Sawtooth Bulb Terraced Impact Smooth Tongued Ripple Total

1 0 1 1 0 0 0 0 22 0 0 1 0 1 0 0 23 0 1 1 0 0 0 1 34 0 1 0 1 0 0 1 35 0 1 1 0 1 0 0 36 0 1 1 1 0 0 0 37 1 0 1 1 0 0 1 48 0 1 0 1 1 0 1 49 1 0 1 0 1 0 1 410 0 1 0 0 1 1 1 411 0 1 1 0 1 1 0 412 1 0 1 0 1 0 1 413 0 1 1 1 1 1 0 514 0 1 1 1 1 0 1 515 1 1 1 0 1 0 1 516 0 1 1 0 1 1 1 517 0 0 1 1 1 1 1 518 1 1 1 1 1 0 1 619 1 1 1 0 1 1 1 620 1 1 1 1 0 1 1 621 1 1 1 1 1 1 1 7Total: 8 16 18 10 15 8 15 Avg:Percent: 38 76 86 48 71 38 71 4.33

C.E. Heckel, S. Wolf / Journal of Archaeological Science 42 (2014) 1e14 9

fracture features observed on the flakes are described. Data on theoccurrence of each fracture feature on each individual flake ispresented.

4.1. Mammoth ivory

In spite of the age of the mammoth ivory block (ca. 28,000 BP),the material was well-preserved and difficult to fracture. Theamount of percussive force required to fracture the ivory led to thecomplete fracture of the limestone anvil into two pieces, as well asthe fracture of two of the quartzite hammerstones, each of whichwas immediately exchanged for an un-fractured hammerstone ofsimilar size and composition. In several instances, direct blowswiththe hammerstone produced cracks in the material that did notimmediately result in the removal of a flake. Substantial cracks orfissures that do not immediately result in catastrophic failure arecharacteristic of ivory fractured by percussion (Heckel, 2009b). Bythe end of the experiment, the initial sample had been reduced tosixty-four smaller fragments, which were collected from theexperimental area following the session. Of these, 38 (59 percent)were under 1.5 cm in greatest dimension. The remaining 28 spec-imens were analyzed for fracture features as described in Section2.2. Some of the flakes produced are pictured in Fig. 9.

The occurrence of fracture features on the mammoth ivoryflakes is presented in Table 1. Only one specimen showed less thanthree of the features analyzed, while three of the specimensshowed six of the seven features. The average number of featureson a flake was 4.35. All twenty-six analyzed flakes (100%) showedsmooth surfaces and sawtooth fractures. Twenty-three of thetwenty-six (88%) showed terraced surfaces. Only four showed rip-ples of percussion, and only two showed clear bulbs of percussion.Eighteen (69%) showed tongued fractures. Roughly half (54%) of theflakes showed macroscopically visible impact scars.

4.2. Elephant ivory

On the frozen sample of elephant ivory, direct percussion withthe quartzite hammerstone resulted in the propagation of cracks,but not in flake-removal. Of fifty-one flakes produced, twenty-one(41%) measured over 1.5 cm and were included in the analysis

described here. The quartzite pebble was then used to prepare thestriking surface with repeated light blows, as is commonly prac-ticed in flint-knapping. Subsequent removals were attempted withthe antler billet and the antler punch. These blows resulted in small,longitudinal flakes, whose detachment was again only possiblewith the application of considerable force. The resulting flakes wereof a variety of shapes and sizes, of which four examples are picturedin Fig. 10. Continued attempts to remove flakes with the quartzitepebble were unsuccessful when the knapper’s thigh was used tostabilize the block. Once the samplewas placed on a sandstone slab,the removal of more flakes was possible. This is in accordance withobservations from the first experimental series and previous ex-periments (Heckel, 2009a, 2009b). An attempt was made to pre-pare a striking platform through the removal of material from theplatform, but this attempt was largely unsuccessful; only smallsplinters resulted. The piece was then placed on the ground andsubjected to heavy non-preferential direct percussion, resulting inthe longitudinal splitting of the core (Fig. 13A).

The occurrence of fracture features on the twenty-one flakesproduced on frozen ivory are reported in Table 2. Only two speci-mens showed less than three fracture features, and one showed allseven features analyzed. The average number of features found onthese flakes was 4.33. In contrast to the non-preferential debitage,controlled knapping produced a high frequency of bulbs of per-cussion (76%) and ripples of percussion (71%). Sawtooth fractures(38%) and tongued fractures (38%) were more rare. In contrast tothe flakes from the first experimental series, smooth surfaces onlyoccurred on 71% of the flakes. As in the first series, macroscopicallyvisible impact marks occurred on roughly half (48%) of the flakes.

On the untreated sample, the first blow with the quartzitehammerstone produced a flake. Successful flake removal with theantler billet was also achieved, but after six blows. Direct percus-sion with the quartzite pebble also resulted in successful flake-removal. Several flakes produced from this sample are pictured inFig. 11. Like the frozen sample, this sample was then placed on theground and subjected to non-preferential direct percussion, thistime resulting in transverse fracture (Fig. 13B). The occurrence offracture features on the flakes produced on the untreated elephantivory appear in Table 3. The sample size for this group is consid-erably smaller than the previous two, with only six flakes over

Table 3Occurrence of each fracture feature on the elephant ivory flakes produced on the untreated ivory. “1” indicates the presence of the feature, “0” indicates its absence.

Specimen Sawtooth Bulb Terraced Impact Smooth Tongued Ripple Total

1 0 0 1 0 1 1 1 42 0 1 1 0 0 1 1 43 0 1 1 0 0 1 1 44 0 1 1 1 0 0 1 45 0 1 1 1 0 0 1 46 1 1 1 0 1 1 1 6Total: 1 5 6 2 2 3 5 Avg:Percent: 16 83 100 33 33 50 83 4.33

Fig. 14. At left, an illustration of longitudinal (A) and transverse (B) samples in Bonfield and Li’s (1965) tensile tests (left). The longitudinal samples form the jagged, fibrous-typefailures described by the authors, while the transverse samples break cleanly at the interface between growth rings. At right, the stress/strain curves reported by Bonfield and Li(1965) for longitudinal (A) and transverse (B) specimens. It is clear that the longitudinal specimens require nearly three times more stress to fail and experience more plasticdeformation before failure than the transverse specimens. (Reproduced by C. Heckel.)

C.E. Heckel, S. Wolf / Journal of Archaeological Science 42 (2014) 1e1410

1.5 cm. All of these flakes show at least four of the seven features,and one of the flakes showed six of the seven features(average ¼ 4.33). Sawtooth features are rare, occurring on only onespecimen. Terracing (100%), bulbs of percussion, and ripples ofpercussion (83%) are common. Tongued features occur on half ofthe specimens (50%), while smooth surfaces and impact scars occuron only one-third of the examples (33%).

Attempts to fracture the heated sample were largely unsuc-cessful in comparison to the other two samples. Though a crackformed along one-third of the total length of the piece duringheating, this crack did not prove to facilitate further fracture. Directpercussion with the quartzite pebble only resulted in tiny splintersand flakes (Fig. 12). The largest of these flakes measured 1.7 cm andthe majority measured less than 1.5 cm. It was decided that heatingivory (at least with themethods used for this experiment) is of littleutility for knapping. Flake removal was once more followed by theplacement of the block on the ground and forceful direct percus-sion, which split the piece longitudinally (Fig. 13C). Because veryfew flakes were produced and those produced were very small,data on fracture features was not collected for the heated elephantivory sample.

5. Discussion

The experiments described above demonstrate that the fracturefeatures introduced in Section 2.2 occur with regularity on ivoryflakes produced by debitage by fracture. Of the fifty-three flakesproduced in the two experimental series, 98% (n ¼ 52) showed atleast three of these features in combination, with an average of4.33e4.35 features per flake in all three of the sample groups. As

could be expected, bulbs and ripples of percussion occur morefrequentlywith the application of controlled, preferentially orientedblows.While the features on the elephant ivory are finer than thosethat appear on themammoth ivory, theyare identical in appearance.With regard to temperature change, flakes could be produced onboth the frozen sample and the sample that was knapped at theoutdoor temperature of 2� Celsius. Heating greatly reduced the sizeof flakes and made them more difficult to produce. The heating ofthe ivory apparently creates changes that make it more brittle andtough to fracture, but identifying the underlying causes of thischangewould requiremore detailed chemical and physical analysesof ivory heated at different temperatures.Water loss, changes to themineralmatrix, and/or changes to the organicmatrix could all occurwhen ivory is heated and alter the properties of the material.

5.1. The mechanics of ivory fracture

It has been demonstrated that the fracture features presented inSection 2.2 consistently occur on the products of ivory debitage byfracture. In order to clarify why these features occur and that theyare necessarily indicative of the application of force, we return tothe structure of ivory as presented in Section 2.1. Little recentliterature exists on the mechanical properties of ivory, specificallyon its response to compressive force and percussion. Data fromstudies of ivory fracture in the 1960s, though, helps to clarify howand why knapped ivory develops the fracture features that it does,and why ivory is notably more difficult to fracture experimentallythan bone (Heckel, 2009b).

When ivory breaks apart naturally, it does so primarily along thegrowth rings and the radial microlaminae (see Fig. 7A). These

Fig. 15. Ivory flakes from Aurignacian assemblages at Hohle Fels Cave (AeD) and Abri Castanet (E, F), and an ivory core from Abri Castanet (G). The fracture features that occur oneach flake are listed in Table 5, as are the find numbers of the individual flakes. Table 4 gives the archaeological context (layer) for each piece as well as the most recent availabledates for those layers.

C.E. Heckel, S. Wolf / Journal of Archaeological Science 42 (2014) 1e14 11

fracture planes are relatively smooth and regular. Percussive forceapplied to ivory meets with a variety of interfaces and obstaclesalong or across which it must travel to produce catastrophic failure.The result of this process is flakes with unpredictable morphologyand complex fracture features related to the three structural fea-tures discussed in Section 2.1. These observations are supported bytensile tests performed on ivory samples taken in longitudinal andtransverse orientations (Bonfield and Li, 1965). Bonfield and Lifound that when subjected to tensile stress, transverse samples ofelephant ivory fractured relatively easily, and with little plasticdeformation, primarily along the interfaces presented by thegrowth rings. Longitudinal samples, however, experienced agreater extent of plastic deformation and required nearly threetimes as much tensile force to fracture (see Fig. 14B). The tensileforce in the longitudinal samples was not able to travel uninter-rupted through the material (Fig. 14Ab). It was redirected by otherstructural features, which created a jagged fracture plane(Fig. 14Aa).

Bonfield and Li (1965: 3184) explain this phenomenon as fol-lows: “The fracture behavior of ivory can be qualitatively correlateddirectly with the interfacial regions. It is likely that fracture willcommence in the areas of relative weakness provided by the in-terfaces. In the case of the transverse specimen, the crack can thenpropagate relatively easily through the interfacial region resultingin a brittle cleavage type failure. However, in the longitudinalspecimen, a similar crack in an interfacial region cannot propagatealong the interface as it is now a parallel direction to the tensionaxis. Hence, for the crack to continue to move, it must then shearacross the adjacent ring into the next interfacial region and as thereis no suitable plane of weakness, this will occur in a direction at 45�

to the tension axis to conform with the maximum resolved shearstress on the specimen. Thus the crack will propagate in a discon-tinuous manner and form 45� fibrous type failure.” These obser-vations made under controlled testing of preferentially oriented

samples can be extended to the interpretation of ivory flakes pro-duced with debitage by fracture. The blocks of ivory used in thearchaeological experiments were much larger than those used byBonfield and Li (1 inch in length, cross-section 0.18 � 0.06 inches)and the application of force took a different form (percussion ratherthan tensile stress). Even so, it can be observed that nearly all of theexperimentally produced flakes exhibit fibrous type failures, spe-cifically in the form of sawtooth, terraced, and tongued features.Tensile tests confirm that creating these features requires a rela-tively great amount of force, and support the assertion that thesefeatures are indicative of fracture by force, rather than taphonomicprocesses. In the experiments presented in this paper, it ispercussive force that must “conform to the maximum resolvedshear stress on the specimen” by propagating in a discontinuousmanner through the material.

That the interfaces between the radial microlaminae and be-tween the growth rings create opportunities for force to travelthrough the material is evident in the morphology of the flakesproduced and the relative ease of creating longitudinal flakes. Mostof the flakes have an elongated shape, as the fracture force traveledparallel to the longitudinal axis of the ivory. This is even true of themammoth ivory sample, which was not knapped from a prefer-ential orientation. Even the successful transverse removal (flakenumber 5 in Fig. 11) is far wider than it is long, as the force thatremoved it traveled once more along the longitudinal axis of thetusk. In doing so, the fracture force also moved through barriers tofracture propagation, creating a terraced surface on the transverseflake.

5.2. The problem of equifinality: animal behavior vs. human action

This article has presented features indicative of intentionalfracture of ivory in the context of human exploitation of the ma-terial, but this is not the only case inwhich forcible fracture of ivory

Table 4Layers and associated dates for archaeological examples pictured in Fig. 15. Thereferences listed offer additional details on the dating of these layers.

Flake(Fig. 15)

Layer Dates(uncalibrated BP)

Reference

A Hohle Fels IId/IIe 29,600e30,600 Conard and Bolus, 2003B Hohle Fels Va 31,800e34,600 Conard and Bolus, 2003C, D Hohle Fels III 29,700e31,100 Conard and Bolus, 2003E, F, G Castanet Aurign. Mean 32,400 White et al., 2012

C.E. Heckel, S. Wolf / Journal of Archaeological Science 42 (2014) 1e1412

can occur. During their lifetimes, proboscideans employ their tusksin a variety of activities that can result in fracture, includingforaging and fighting. Haynes (Conybeare and Haynes, 1984;Haynes, 1988, 1993) has documented fracture caused by animalbehavior and the morphology and features of the resulting flakes inseveral contexts. In the case of natural ivory breakage, fracture ismost often due to bending force, and not percussive force, yet manyof the features presented in Section 2.2 occur on the flakes pro-duced by animal behavior. Such fracture also most commonly oc-curs at the outer surface and/or distal end of the tusk (Haynes,1988: 151e152).

The similarity of some naturally fractured ivory flakes to thoseproduced by human action presents a potential problem of equi-finality. In contexts inwhich there is no other evidence of anthropicivory exploitation, the presence of fracture features cannot beattributed a priori to human action. Even in contexts where humanexploitation of ivory is clearly evidenced, caution must be exercisedwhen attributing products of fracture to human action: ivoryfragments created by animal behavior could be collected from thelandscape by hominids. A direct comparison of flakes produced byanimal behavior and flakes produced by human actionwould makea useful contribution to this field of study. For the time being,caution should be exercised in the identification of anthropic ivoryflaking at archaeological sites, with reference to literature on bothnatural and anthropic fracture products.

In the early Aurignacian at Hohle Fels Cave and Abri Castanetthere are three factors that make the attribution of ivory flakes tohuman action reasonably secure. 1) Extensive ivory exploitation isevidenced in the form of all stages of production, from larger tuskfragments to production stages to final products. The presence ofivory cores (Fig. 15G) further supports the hypothesis of on-siteivory reduction. 2) Debitage by fracture of a variety of otherosseous materials (bone and antler) is widely evidenced in theearly Aurignacian and is recognized as a significant component ofearly Aurignacian technological repertoires. 3) Each site has yiel-ded larger segments of tusks that are highly unlikely to havebroken away from the tusk of a living animal and have beenfurther reduced on-site. Furthermore, there is little evidence thatflakes come exclusively from the exterior of the tusk, as would beexpected in the case of fracture caused by animal behavior. Theangle of the Schreger Pattern in the tusk is much more acute to-ward the center of the tusk, a factor which can be used to situateflakes, cores, and artifacts within the tusk. The ivory core from Abri

Table 5Occurrence of fracture features on each of the archaeological examples pictured in Fig. 1

Specimen Figure Sawtooth Bulb Terraced

Va 25_976 IId 15B 1 1 156_3762.1 IIIa 15A 1 0 188_973 IIIa 15C 1 1 198/1204 15D 0 0 1H13D-219 15E 1 0 1H12C12-439 15F 1 0 1

Total 5 2 6Percent 83 33 100

Castanet pictured in Fig. 15G, for example, displays very acuteangles in the Schreger Pattern (33�) that indicate its origin near thecenter of the tusk. While it is not outside the realm of possibilitythat ivory flakes were gathered from the landscape, the quantityand nature of the ivory present at both sites as well as theextensive evidence of ivory, bone, and antler exploitation bypercussive fracture in the period indicate that human action is themost parsimonious explanation for the presence of ivory flakes atAbri Castanet and Hohle Fels Cave.

5.3. Aurignacian evidence for ivory debitage by fracture

Ivory is recovered from Aurignacian contexts in a variety ofconditions, and post-depositional alterations can hinder attemptsat analysis. The fracture features created by debitage by fractureand the force required to produce themmakes products of debitageby fracture relatively easy to identify on pieces that have not un-dergone extensive post-depositional alteration. By way of example,we present flakes from the sites of Hohle Fels Cave (Swabian Jura,Germany) and Abri Castanet (Dordogne, France). Both sites haveyielded assemblages of artifacts attributed to the aurignacien ancien(also “early Aurignacian” or “Aurignacian I”), including objectsmanufactured in mammoth ivory. Hohle Fels Cave contains de-posits ranging from the Mousterian through the Neolithic, and itsAurignacian deposits are especially famous for the numerous ivoryartifacts they have yielded, including several figurines of animalsand the recently discovered female figurine known as the “Venus ofHohle Fels” (Conard, 2003a, 2003b, 2009; Conard and Bolus, 2008;Floss, 2007). Hohle Fels Cave has an excavation history of over 140years (Conard, 2002). Current excavations have been undertakenannually since 1997 under the direction of Nicholas J. Conard(Universität Tubingen). Overall, the Aurignacian layers (see Table 4)have yielded approximately 8000 ivory pieces, of which 1500 havebeen identified as worked (Wolf, 2013).

The southern sector of Abri Castanet has only a single archae-ological layer: material from the aurignacien ancien that has yieldedan average age of 32,400 BP (uncalibrated) (Mensan et al., 2012;White et al., 2012). Ivory artifacts are limited to ornaments andbyproducts of their production, but are numerous. Excavation ofthe northern sector of the site was carried out by Peyrony in theearly 20th Century (Peyrony, 1935). The southern sector of the site(several meters south of Peyrony’s excavation) was excavated from1995 to 1999 under the direction of Randall White and JacquesPelegrin, and from 2005 to 2010 under the direction of RandallWhite. Because the assemblages from the early excavations at AbriCastanet were dispersed to numerous institutions in France andabroad, a conclusive count of ivory artifacts is not currently avail-able. Such an inventory is the subject of ongoing research by one ofthe authors (CH).

We have selected seven examples from the Aurignacian depositsat Hohle Fels Cave (n ¼ 4) and Abri Castanet (n ¼ 3) that show clearevidence of debitage by fracture (Fig. 15). Six of them are flakes, and

5. Presence is indicated by a “1” and absence is indicated by a “0.”

Impact Smooth Tongued Ripple Total

1 1 1 1 70 1 0 1 40 0 0 1 50 1 0 1 40 1 0 0 30 1 1 0 41 5 2 4 Avg:

17 83 33 67 4.5

C.E. Heckel, S. Wolf / Journal of Archaeological Science 42 (2014) 1e14 13

one is a core. These artifacts are listed by site and layer with relativedate ranges in Table 4. Each of the samples displays a combinationof several of the features identified on the experimental samples:bulbs of percussion, ripples of percussion, saw-tooth fractures,smooth fracture planes, and terraced fracture planes (Table 5).Table 5 also provides the artifact numbers of the flakes pictured inFig. 15.

The selected flakes were removed parallel to the longitudinalaxis of the tusk, and their regular morphology and the consistentpresence of bulbs of percussion suggest controlled knapping ratherthan brute percussive force. Several of the flakes (Fig. 15 B, D, E, F)show evidence of additional flake removal on the dorsal and/orventral surfaces, indicating that they were stages in a longer chaîneopératoire of flake production. Unfortunately, there is insufficientrefitting of flakes at both sites to determine a more exact sequencein this chaîne opératoire. Two of the flakes (Fig. 15 B, D) show evi-dence of use subsequent to their production, in the form ofcrushing localized at the terminal end (Fig. 15B) or extending overmuch more of the basal end perimeter than would result from aknapping blow (Fig. 15C). This crushing covers parts of the flakethat would not have been exposed prior to knapping, and extendsover the basal end to partially encroach on the lateral sides (Wolf,2013).

In addition to flakes, ivory “cores” bearing negatives of removaland other signs of debitage by fracture exist in Aurignacian as-semblages. Fig. 15G is a picture of one such ivory core found at AbriCastanet. Clear negatives of removal are present (direction ofremoval indicated by the white arrows), and flakes appear to havebeen struck parallel and transverse to the longitudinal axis of thetusk. Two adjacent flake removals have been enlarged in the imageto better show the detail of the terraced patterns on their surfaces.The archaeological core in Fig. 15 is similar in appearance to theexperimental ivory cores pictured in Fig. 13. The acute angle of theSchreger Pattern visible at the proximal and distal ends of thearchaeological core reveal that it came from the center of the tusk,rather than the outer layers. Its abandonment at Abri Castanet maybe due to the difficulty of working the dense material at the centerof the tusk. Such ivory cores also exist in Aurignacian context atHohle Fels Cave (Wolf, 2013).

6. Conclusions

Recognition of the products and byproducts of debitage byfracture of osseous materials is rapidly improving thanks to anumber of recent explorations of the evidence for these activ-ities. It has been demonstrated that debitage by fracture was animportant component of Aurignacian osseous technology(Liolios, 1999; Tartar, 2009; Tejero et al., 2012; Wolf, 2013). Theexperiments described here have shown that, under a variety ofconditions, proboscidean ivory produces a set of complex frac-ture features diagnostic of debitage by fracture. These featuresresult from fracture force traversing barriers to and conduits forits propagation, in response to the unique internal architectureof proboscidean tusks. Continued experimentation with samplesof ivory of varying size and in varying states of preservation willhelp to provide a more complete picture of how ivory respondsto different technological treatments and processes. Examplesof flakes that show features diagnostic of debitage by percussionhave been recovered from Aurignacian layers known to haveyielded large assemblages of worked ivory. Improved recogni-tion of ivory debitage by fracture in other Aurignacian assem-blages will serve to foster an improved understanding of theexploitation of this material within the broader technologicalcontext of early Upper Paleolithic approaches to osseous rawmaterials.

Acknowledgments

The authors wish to thank those who assisted with various as-pects of the two experimental series: Jens A. Frick, Sarah Ranlett,Matthew Sisk, and Marek Thomanek. Special thanks to Nicholas J.Conard and Randall White for advisement on this and related en-deavors and for access to archaeological collections of Hohle FelsCave and Abri Castanet, respectively. Many thanks as well toChristian Tryon for helpful comments on previous versions of thispaper. Research conducted by C. Heckel has been supported by: theGraduate School of Arts and Sciences at New York University, theChateaubriand Foundation, and the Georges Lurcy Foundation.Research conducted by C. Heckel has also been supported by theFranco- American collaborative exchange entitled “AurignacianGenius: Art, daily life and social identity of the first modern humansof Europe”, UMI 3199-CNRS-NYU & UMR 5608-TRACES, U. of Tou-louse 2-Mirail, funded by a three-year grant from the PartnerUniversity Fund and the Andrew Mellon Foundation. Researchconducted by S. Wolf has been supported by: the Alb-Donau-Kreisand the Landesamt für Denkmalpflege Baden-Württemberg. Thanksalso to the Bundesamt für Naturschutz Bonn for providing theelephant ivory used in the second experimental series.

References

Alvarez-Fernandez, E., Jöris, O., 2007. Personal ornaments in the EarlyUpper Paleolithic of Western Eurasia: an evaluation of the record. Eur. Prehist.5, 31e44.

Averbouh, A., 2000. Technologie de la Matière Osseuse Travaillée et ImplicationsPalethnologiques: L’exemple des chaînes d’exploitation du bois de cervidé chezles Magdaléniens des Pyrénées. Unpublished Doctoral thesis. Université de ParisI, Paris.

Averbouh, A., Pétillon, J.-M., 2011. Identification of “debitage by fracturation” onreindeer antler: case study of the Badegoulian levels at the Cuzoul de Vers (Lot,France). In: Baron, J., Kufel-Diakowska, B. (Eds.), Written in Bones: Studies onTechnological and Social Contexts of Past Faunal Skeletal Remains. InstytutArcheologii Wroclaw, Poland, pp. 41e51.

Banerjee, A., Eckmann, C., 2011. Elfenbein und Archäologie. INCENTIVS-Tagungs-beiträge 2004e2007. Römisch-Germanisches Zentralmuseum, Mainz.

Baumann, M., Maury, S., 2013. Ideas no longer written in antler. J. Archaeol. Sci. 40,601e614.

Bonfield, W., Li, C.H., 1965. Deformation and fracture of ivory. J. Appl. Phys. 36,3181e3184.

Christensen, M., 1999. Technologie de l’ivoire au Paléolithique supérieur: caractér-isation physico-chimique du matériau et analyse foctionnelle des outils detransformation. In: BAR International Series, vol. 751. Hadrian Books, Oxford.

Conard, N.J., 2002. Der Stand der altsteinzeitlichen Forschung im Achtal derSchwäbischen Alb. Mit. Ges. Urgeshichte 11, 67e77.

Conard, N.J., 2003a. Paleolithic ivory sculptures from southwestern Germany andthe origins of figurative art. Nature 426, 830e832.

Conard, N.J., 2003b. Eiszeitlicher Schmuck auf der Schwäbischen Alb. In: Kölbl, S.,Conard, N. (Eds.), Eiszeitschmuck: Status und Schönheit. UrgeschichtlichesMuseum, Blaubeuren, Germany, pp. 15e49.

Conard, N.J., 2009. A female figurine from the basal Aurignacian of Hohle Fels Caveof southwestern Germany. Nature 459, 248e252.

Conard, N., Bolus, M., 2003. Radiocarbon dating the appearance of modern humansand timing of cultural innovations in Europe: new results and new challenges. J.Hum. Evol. 44, 331e371.

Conard, N.J., Bolus, M., 2008. Radiocarbon dating the last Middle Palaeolithic andthe Aurignacian of the Swabian Jura. J. Hum. Evol. 55, 886e889.

Conybeare, A., Haynes, G., 1984. Observations on elephant mortality and bones inwater holes. Quat. Res. 22, 189e200.

David, E., 2007. Technology of bone and antler industries: a relevant methodologyfor characterizing early Post-Glacial societies (9the8th millennium BC). In:Gates St-Pierre, C., Walker, R.B. (Eds.), Bones as Tools: Current Methods andInterpretations in Worked Bone Studies. Archaeopress, Oxford, UK, pp. 35e50.

Espinoza, E.O., Mann, M.J., 1992. Identification Guide for Ivory and Ivory Substitutes.World Wildlife Fund Publications.

Floss, H., 2007. L’art mobilier aurignacien du Jura souabe et sa place dans l’artpaléolithique e Die Kleinkunst des Aurignacien auf der Schwäbischen Alb undihre Stellung in der paläolithischen Kunst. In: Floss, H., Rouquerol, N. (Eds.), Leschemins de l’Art aurignacien en Europe e Das Aurignacien und die Anfängeder Kunst in Europa. Éditions Musée-forum Aurignac, Aurignac, France,pp. 295e316.

Galán, A., et al., 2009. A new experimental study on percussion marks and notchesand their bearing on the interpretation of hammerstone-broken faunal as-semblages. J. Archaeol. Sci. 36, 776e784.

C.E. Heckel, S. Wolf / Journal of Archaeological Science 42 (2014) 1e1414

Goutas, N., 2009. Réflexions sur une innovation technique gravettienne importante:le double rainurage longitudinal. Bull. Soc. préhist. fr. 106, 437e456.

Hahn, J., 1986. Kraft und Aggression: die Botschaft der Eiszeitkunst im AurignacienSüddeutschlands? Archaeologica Venatoria, Tübingen.

Hahn, J., et al., 1995. Le Travail et l’Usage de l’Ivoire au Paléolithique Supérieur.Istituto Poligrafico e Zecca dello Stato, Rome.

Haynes, G., 1988. Longitudinal studies of African elephant death and bone deposits.J. Archaeol. Sci. 15, 131e157.

Haynes, G., 1993. Mammoths, Mastodonts, and Elephants: Biology, Behavior and theFossil Record. Cambridge University Press.

Heckel, C., 2009a. Physical characteristics of mammoth ivory and their implicationsfor ivory work in the Upper Paleolithic. Mitteilungen der Gesellschaft fürUrgeschichte 18, 71e91.

Heckel, C., 2009b. Widening the Lens: Physical Science Research and Microscopy inthe Archaeological Study of Mammoth Ivory. Unpublished Master’s thesis.Department of Anthropology, New York University, New York, NY.

Inizan, M.-L., Ballinger, M., Roche, H., Tixier, J., 1999. Technology and Terminology ofKnapped Stone. Préhistoire de la Pierre Taillé, vol. 5. CREP, Meduon, France.

Jakubinek, M.B., et al., 2006. Elephant ivory: a low thermal conductivity, highstrength nanocomposite. J. Mater. Res. 21, 287e292.

Khlopachev, G.A., Girya, E.Y., 2010. Secrets of Ancient Carvers of Eastern Europe andSiberia: Treatment Techniques of Ivory and Reindeer Antler in the Stone Age.NAUKA, Saint Petersburg.

Knecht, H., 1991. Technological Innovation and Design during the Early UpperPaleolithic: a Study of Organic Projectiles. Unpublished doctoral thesis.Department of Anthropology, New York University, New York, NY.

Knecht, H., 1993. Splits and wedges: the techniques and technology of early Auri-gnacian antler working. In: Knecht, H., Pike-Tay, A., White, R. (Eds.), BeforeLascaux: the Complex Record of the Early Upper Paleolithic. CRC Press, Florida,pp. 137e162.

Liolios, D., 1999. Variabilité et caracteristiques du travail des matières osseuses audebut de l’Aurignacien: Approche technologique et economique. Unpublisheddoctoral thesis. Université de Paris X, Paris.

Locke, M., 2008. Structure of ivory. J. Morphol. 269, 423e450.Malina, M., Ehmann, R., 2009. Ivory Splitting in the Aurignacian: reconstructing the

manufacturing technique of the ivory flute from Geißenklösterle. Mitteilungender Gesellschaft für Urgeschichte 18, 93e108.

Mensan, R., et al., 2012. Une nouvelle découverte d’art pariétal aurignacien in situ àl’abri Castanet (Dordogne, France): contexte et datation. Paléo 23, 171e188.

Pétillon, J.-M., Ducasse, S., 2012. From flakes to grooves: a technical shift in ant-lerworking during the last glacial maximum in southwest France. J. Hum. Evol.62, 435e465.

Peyrony, D., 1935. Le Gisement Castanet, vallon de Castelmerle (commune de Ser-geac), Aurignacien I et II. Bull. Soc. préhist. fr. 32, 418e443.

Saunders, J., 1979. A close look at ivory. Living Mus. 41, 56e59.Semenov, S.A., 1964. Prehistoric Technology; an Experimental Study of the Oldest

Tools and Artefacts From Traces of Manufacture and Wear. Cory, Adams &Mackay, London.

Su, X., Cui, F., 1997. Direct observations on apatite crystals in ivory. J. Mater. Sci. Lett.16, 1198e1200.

Su, X., Cui, F., 1999. Hierarchical structure of ivory: from nanometer to centimeter.Mater. Sci. Eng. C 7, 19e29.

Tartar, E., White, R., 2013. The manufacture of Aurignacian split-based points: anexperimental challenge. J. Archaeol. Sci. 40 (6), 2723e2745.

Tartar, É., 2009. De l’os à l’outil: caractérisation technique, économique et sociale del’utilisation de l’os à l’aurignacien ancien. Unpublished doctoral thesis. In: Étudede trois sites: l’Abri Castanet (secteurs nord et sud); Brassempouy (Grotte desHyènes et Abri Dubalen) et Gatzarria. Université de Paris I, Paris.

Tejero, J.-M., et al., 2012. Red deer antler technology and early modern humans inSoutheast Europe: an experimental study. J. Archaeol. Sci. 39, 332e346.

Vanhaeren, M., d’Errico, F., 2006. Aurignacian ethno-linguistic geography of Europerevealed by personal ornaments. J. Archaeol. Sci. 33, 1105e1128.

Virág, A., 2012. Histogenesis of the unique morphology of proboscidean ivory.J. Morphol. 273, 1406e1423.

White, R., 1997. Substantial Acts: from Materials to Meaning in Upper PaleolithicRepresentation, beyond Art: Pleistocene Image and Symbol. California Academyof Sciences, San Francisco, pp. 93e121.

White, R., 2007. Systems of personal ornamentation in the Early Upper Paleo-lithic: methodological challenges and new observations. In: Mellars, P.,Boyle, K., Bar-Yosef, O., Stringer, C. (Eds.), Rethinking the Human Revolution:New Behavioural and Biological Perspectives on the Origin and Dispersal ofModern Humans. McDonald Institute for Archaeological Research, Oxford,pp. 287e302.

White, R., 2008. Abri Castanet, Secteurs Sud et Nord (Commune de Sergeac, Dor-dogne) Rapport de fouille programmée Années 2006 à 2008, Service régionalde l’archéologie. Ministére de la culture, France, Bordeaux.

White, R., et al., 2012. Context and dating of Aurignacian vulvar representationsfrom Abri Castanet, France. Proc. Natl. Acad. Sci. 109, 8450e8455.

Wolf, S., 2013. Schmuckstücke e Die Elfenbeinbearbeitung im SchwäbischenAurignacien. Unpublished doctoral thesis. Eberhard-Karls Universität Tübingen,Tübingen, Germany.

Zhang, H., et al., 1993. Characterizing hierarchical structures of natural ivory. In:Materials Research Society Symposium Proceedings. Cambridge UniversityPress, p. 151.