YolandaFernandez-JalvoThe Natural History Museum,Cromwell Road, LondonSW7 5BD, U.K., and MuseoNacional de CienciasNaturales, José GutiérrezAbascal, 2, 28006-Madrid,Spaine-mail: yfj@mnch.csic.es

J. Carlos DıezDept. Ciencias Históricas,Fac. Humanidades,Universidad Burgos, Ctra.Villadiego s/n, 09001 Burgos,Spain

Isabel CaceresArea de Prehistoria (Unidadde Investigación Asociada alCSIC), Universitat Rovira iVirgili, Pl. Imperial Tarraco,1, 43005-Tarragona, Spain

Jordi RosellArea de Prehistoria (Unidadde Investigación Asociada alCSIC), Universitat Rovira iVirgili, Pl. Imperial Tarraco,1, 43005-Tarragona, Spain

Received 1 October 1998Revision received15 December 1998and accepted 25 April 1999

Keywords: cannibalism, EarlyPleistocene, ‘‘AuroraStratum’’ Atapuerca, GranDolina, human remains,human behaviour, Homoantecessor, taphonomy.

Human cannibalism in the EarlyPleistocene of Europe (Gran Dolina,Sierra de Atapuerca, Burgos, Spain)

Human remains belonging to at least six individuals were found in anexploratory excavation made at the site of Gran Dolina (Sierra deAtapuerca, Burgos, Spain). These remains were recovered from theAurora Stratum of Unit TD6. This stratum has a thickness ofapproximately 30 cm. The area of the exploratory excavation is about7 m2. According to palaeomagnetic analyses, Unit TD6 showsreversed polarity, which is considered to belong to the Matuyamachron. This unit is immediately below TD7, where the Matuyama–Brunhes boundary has been detected, indicating an age of around780,000 years BP.

There is no specific distribution, treatment, or arrangement of thehuman remains, which were found randomly mixed with abundantfaunal remains and stone tools. Most of the faunal and human fossilbones from the Aurora Stratum have human induced damage. Stonetool cutmarks are frequent, and peeling (a type of fracture similar tobending a fresh twig between the hands) provides a specific breakagepattern together with percussion marks and chopmarks. Both non-human and human remains show similar intensive exploitation. Slightdifferences, however, have been observed between fauna and humans(e.g., peeling frequent in humans, rare in fauna), that appear relatedto different musculature, weight, and bone structure. The character-istics of this fossil assemblage suggest that it is solely the resultof consumptive activities as there is no evidence of ritual or otherintention. The possibility of distinguishing between dietary vs.survival cannibalism is discussed here.

? 1999 Academic Press

Journal of Human Evolution (1999) 37, 591–622Article No. jhev.1999.0324Available online at http://www.idealibrary.com on

Assessing cannibalism

Cannibalism has been documented inseveral different human groups and civili-zations through time, based on writtenreferences, oral tradition or remains ofthe victims. Many myths, tales and

0047–2484/99/090591+32$30.00/0

legends narrate acts of cannibalism involv-ing real or fictitious creatures. Althoughthe term cannibalism derives from theCaribbean peoples, references to cannibalpractices have been mentioned all overthe world in both prehistoric and historicperiods.

? 1999 Academic Press

592 . - ET AL.

Human cannibalism in anthropology andpalaeontology is a controversial topic thatprovokes contradictory reactions. Duringthe middle of the ninteenth century theinfluence of Darwin’s Origin of the Speciesinduced important reactions in science. Thefirst human like fossils discovered in theNeander Valley (Germany, 1856 ca. 40–50 ka) were considered from an anthropo-centric point of view—everything was madeby and for hominids. Contrary beliefs werethat the ancient humans were ‘‘barbariansavages’’ and ‘‘cannibals by definition’’. Thefirst report of cannibalism (Gorjanovic-Kramberger, 1909) was made soon after thediscovery of hominid remains at Krapina(Croatia 1895–1905, ca. 130 ka). Claims ofcannibalism were gradually linked with‘‘cults of skulls’’ in the 1930s with the dis-covery of skulls in Steinheim (Germany,1933, ca. 250 ka), Monte Circeo (Italy,1939, ca. 50 ka), and Zhoukoudian (China1928–1937, ca. 400–500 ka). These re-mains, whose cranial bases were missing,were considered to be remains of cannibal-istic feasts at which the brains had beenconsumed. However, later studies haveshown that the lack of the cranial base iscommon since this part of the skull is fragile.

Raymond Dart thought that the lack ofthe front teeth on a specimen of Australo-pithecus (Makapansgat 1948, ca. 3 m.y.a.),and broken long bones, demonstrated somemanner of violent death. Again, taphonomicstudies showed that this damage was not theresult of cannibalistic practices, but wascaused by hyaenas seeking fat-rich marrow.

Subsequent discussions of cannibalismhave been characterized by either permissivetolerance (e.g., Blanc, 1961) or extremecriticism (Arens, 1979) and disapproval ofcannibalism claims. Several authors havedemanded more scientific rigour (e.g.,Jacob, 1972; Binford, 1981; Askenasy,1994).

In his book The Man-eating Myth: Anthro-pology and Anthropophagy, Arens (1979),

presents an exhaustive analysis of claims forcannibalism in several societies at varioustimes. His main conclusion was that thereis no convincing evidence for human canni-balism (except for survival in extremeconditions of starvation). This work wasparticularly important at the time since somany uncritical publications had previouslyaccepted that cannibalism was practised bymany tribes and ancestors. However, Arens,as well as his followers, neglected or ignoredsome of the best evidence. Since 1979,taphonomic studies of bone remains havedemonstrated the validity of a number ofclaims for cannibalism. It is not our inten-tion to review the literature related tohistoric cannibalism. Discussions amongsocial anthropologists and extensive compi-lations of cannibalism claims can be foundin Binford (1981), Villa et al. (1996a,b),Villa (1992), White (1992) and Turner &Turner (1995).

Cannibalism, in spite of the origin of theword, occurs not only in humans but also inmany other species that use it as a means ofpopulation control, a source of food, or as asign of authority and strength by the domi-nant member. Cannibalism occurs amongvarious orders of mammals, insects andbirds, and there are some accounts of suchoccurrences among omnivorous primates(Bygot, 1972; Goodall, 1979), and bears(Kurt,1976). A cannibal is therefore definedas a person or animal that eats any typeof tissue of another individual of its ownkind.

Cannibalism cannot be established onthe sole basis of cutmarks. This is the casefor Bodo (Ethiopia, ca. 600 ka) andGoughs’s cave (England, ca. 12 ka). White(1985) and Cook (1986) studied thesesites, respectively, and could not reachconclusive interpretations. Remains fromboth sites bear undeniable cutmarks,indicating that the skeletons were inten-tionally defleshed, although not necessarilyeaten.

593

Figure 1. Map of the Iberian Peninsula. The black arrow points out the location of the sites, near the townof Burgos.

Some of the functional types of potentialhuman cannibalism are:( 1) Nutritional

(a) incidental: survival (periods of foodscarcity or due to catastrophes, i.e.,starvation-induced).

(b) long duration: gastronomic or dietary(humans are part of the diet of otherhumans).

( 2) Ritual, magic, funerary (in relation tobeliefs or religion).

( 3) Pathological [mental disease: parapathicdefined by Reverte (1981); for politi-cal reasons, as referred to by Zheng Yi(1997), in China].

These functional types of cannibalismhave also been sub-divided into socialdivisions that include aggressive (consumingenemies) vs. affectionate (consuming friendsor relatives), or endocannibalism (consump-tion of individuals within the group) vs.exocannibalism (consumption of outsiders).

The identification of nutritional, asopposed to ritual, cannibalism, is based on acombination of indicators, the main cri-terion being the comparison of human and

animal remains from the same archaeologi-cal context. According to Villa et al.(1986a), these indicators are:- similar butchering techniques in human

and animal remains. Thus frequency,location and type of verified cutmarksand chopmarks on human and animalbones must be similar, but allowanceshould be made for anatomical differ-ences between humans and animals;

- similar patterns of long bone breakagethat might facilitate marrow extraction;

- identical patterns of post-processing dis-carding of human and animal remains;

- when applicable, evidence of cooking; ifpresent, such evidence should indicatecomparable treatment of humans andanimal remains.

However, when human and nonhumananimal remains are found in separate con-texts, with different patterns of exploitationand distribution, ritual or some otherinterpretation should be considered as analternative cause of cannibalism (Villa et al.,1986a; Villa, 1992; White, 1992; Turner &Turner, 1995).

594 . - ET AL.

18

–450

–650

Transversal section TD6D

epth

(cm

s)

16 17

–475

–500

–525

–550

–575

–600

–625

Aurora Stratum

(a)

I

H

G

16 17 18

N

Fauna Hominids Stone tools

Wall

Trench Section

(b)

595

Sierra de Atapuerca, Gran Dolina site,TD6—‘‘Aurora Stratum’’

The site of Gran Dolina belongs to thesouthern part of the karstic site complex ofthe Sierra de Atapuerca. This is a smallmountain, 1079 m above sea level, 15 kmfrom the town of Burgos in northern centralSpain (Figure 1). The area is in the DueroBasin, bounded by the Demanda mountainrange to the east and by the Arlanzón Riverto the south. The Gran Dolina site is one ofseven sites systematically excavated in thisarea since 1980. Six of these sites are pres-ently exposed in an abandoned railwaytrench, which was opened at the beginningof the twentieth century.

Gran Dolina is an 18 m-thick cave infill-ing. Eleven sedimentary levels have beendistinguished in the sequence, many of themyielding abundant fossil fauna assemblage,as well as many stone implements, both ofwhich have provided important informationon human behaviour (Carbonell et al.,1995a). An exploratory excavation of thewhole section, from the uppermost zone ofthe stratigraphical sequence to the base ofthe infilling, has been made since 1992.Human remains have been recovered from adistinctive stratum of the unit TD6 named‘‘Aurora’’ [Figure 2(a)], after the archaeolo-gist who discovered the first human fossils atTD6, Aurora Martın Nájera. It is a 30 cm-thick layer that slopes down towards thesouthwest.

Figure 2. (a) Transversal section (E–W) of the prospective excavation area at TD6 (Gran Dolina) showingthe findings of the unit. Notice the high fossil density on top of the unit TD6 identifying the AuroraStratum. (b) Aerial plan of Aurora Stratum showing the excavation coordinates; G-H-I (from South toNorth of the excavation) and 16-17-18 (from West to East of the excavation). Note that humans, faunaand implements are randomly dispersed throughout the excavation area.

The human remains from Gran Dolina

The Gran Dolina TD6 site has recentlyyielded human remains of six individuals

found mixed together with stone tools andnonhuman fauna remains (Carbonell et al.,1995b). These humans come from the sub-unit Aurora Stratum in particular. Their ageis more than 780 ka (Parés and Pérez-González, 1995). These human fossils havebeen assigned by Bermúdez de Castro et al.(1997) to the new species Homo antecessor.The first human remains were discovered in1994, and were soon afterwards recognizedas having been cannibalized (Fernández-Jalvo et al., 1996). As the exploratory exca-vation of the Aurora Stratum has finished, itis now possible for a detailed taphonomicanalysis of the fossil of this subunit to beundertaken, as well as reconstructing theprocesses of the site formation (Díez et al.,1999). We will discuss in this paper theevidence that may allow us to specify thetype of cannibalism (nutritional vs. ritual),and whether it is possible to distinguishbetween dietary and survival cannibalism.

Results of the present study are then com-pared with sites that have also been tapho-nomically analysed and where modernmethods of excavation have been used, as inAtapuerca TD6. These study areas and sitesare Fontbregoua (France—Neolithic—Villaet al., 1986a,b), Mancos (from Colorado—AD 1100–1150—White, 1992) andthroughout the Southwest Amerindian area(Arizona) by Turner & colleagues, 1970–1999. It has to be kept in mind, however,that the ages of these sites are not compar-able to the Aurora Stratum, and, therefore,social attributes and behaviours cannotreadily be inferred or considered analogous.Furthermore, the number of human remainsfrom the Aurora Stratum (92 NISP) andthe excavated area (7 m2) are smaller than

596 . - ET AL.

Table 1 Identified human specimens from Aurora Stratum

Label Age Element Area Side Individual

ATD6-1 Juvenile Tooth Canine Lower left IATD6-2 Juvenile Tooth Incisor Left l2 IATD6-3 Juvenile Tooth Premolar Right LP3 IATD6-4 Juvenile Tooth Premolar Right LP4 IATD6-5 Juvenile Mandible Body Right side (M1-M3) IATD6-6 Juvenile Tooth Canine Right lower IATD6-7 Juvenile Tooth Premolar Right UP3 IATD6-8 Juvenile Tooth Premolar Right UP4 IATD6-9 Juvenile Tooth Premolar Left UP4 IATD6-10 Juvenile Tooth Molar Right UM1 IATD6-11 Juvenile Tooth Molar Left UM1 IATD6-12 Juvenile Tooth Molar Right UM2 IATD6-13 Juvenile Maxilla Alveolar Left IATD6-14 Inf. Maxilla Alveolar Left (dc-dm1) IIATD6-15 Juvenile Skull Frontal RightATD6-16 Juvenile Skull Temporal RightATD6-17 Adult Skull Temporal RightATD6-18 Skull Petrous-temporal LeftATD6-19 Adult Skull Zygomatic arch RightATD6-20 Skull Parietal LeftATD6-21 Juvenile Radius Diaphysis LeftATD6-22 Adult Patella Complete LeftATD6-23 Adult Carpal Distal Hamate (left)ATD6-24 Adult Carpal Complete CapitateATD6-25 Adult Metatarsal Proximal end Mtts. 2–3 leftATD6-26 Adult Metacarpal Distal condyle 2 mtcp., leftATD6-27 Adult Phalange Diaphysis Hand, 1 phal.-finger 2–3ATD6-28 Adult Phalange Complete Hand, 2 phal.ATD6-29 Adult Phalange Distal Hand, 1 phal.ATD6-30 Adult Phalange Complete Foot, 1 phal. toe 1, rightATD6-31 Adult Phalange Complete 1 phal. finger 1ATD6-32 Adult Phalange Distal Foot, 1 phal.ATD6-33 Adult Phalange Complete Foot, 2 phal. toe 2, leftATD6-34 Adult Phalange Complete Foot, 2 phal. toe 2–3ATD6-35 Adult Phalange Complete Foot, 2 phal. toe 4–5ATD6-36 Adult Phalange Distal apical tuber. Foot, 3 phal.ATD6-38 Juvenile Vertebra Body LumbarATD6-39 Adult Rib CompleteATD6-40 Juvenile Vertebra Spinous process ThoracicATD6-43 Juvenile Radius Diaphysis LeftATD6-44 Juvenile Phalange Diaphysis Hand, 2 phal.ATD6-45 Adult Vertebra Transverse process LumbarATD6-46 Adult Phalange Prox.+diaphysis Hand, 2 phalATD6-48 Juv-ad Tooth Crown Left lower incisor 2 IVATD6-49 Juvenile MaxillaATD6-50 Juvenile Clavicle Complete RightATD6-51 Adult Vertebra Complete CervicalATD6-52 Juv-ad Tooth Incisor Left lower l1 VATD6-53 Juvenile Phalange Complete Hand, 2 phal.ATD6-54 Inf. Vertebra Lamina AxisATD6-55 Inf. Clavicle Lateral LeftATD6-56 Juvenile Patella Complete RightATD6-57 Juvenile Skull TemporalATD6-58 Adult Skull Zygomatic+maxilla LeftATD6-59 Adult Metacarpal Dist.+diaphysis 2 mtcp. left

597

some of the sites with which they will becompared.

Materials and methods

Table 1 Continued

Label Age Element Area Side Individual

ATD6-60 Adult Skull Pterion LeftATD6-62 Juvenile Skull Crista galli EthmoidATD6-63 Adult Mandible Mental protuberanceATD6-64 Juvenile Clavicle Diaphysis RightATD6-66 Adult Rib Prox.+diaphysisATD6-67 Inf. Phalange Dist.+diaphysis Hand, 1 phal.ATD6-68 Juvenile Phalange Complete Foot, 3 phal.ATD6-69 Juvenile Maxilla Alveol-frontal process (L P3, M1-M3 & R I2-M1) IIIATD6-70 Adult Metatarsal Distal epiphysis 2 mtts leftATD6-71 Skull Frontal?ATD6-72 Juvenile Skull Frontal?ATD6-73 Adult Skull Fragment IndetATD6-74 Inf. Vertebra Body ThoracicATD6-75 Adult Vertebra Lamina CervicalATD6-76 Juvenile Femur Prox.+diaphysisATD6-77 Adult Skull Occipital condyleATD6-78 Juvenile Skull Frontal?ATD6-79 Adult Rib Head+diaphysisATD6-80 Adult Vertebra Lamina CervicalATD6-81 Juvenile Skull SphenoidATD6-82 Adult Phalange Dist.+diaphysis Hand, 1 phal.ATD6-84 Juvenile Skull Zygomatic archATD6-85 Adult Rib DiaphysisATD6-87 Adult Skull ParietalATD6-88 Adult Rib Head+diaphysis ii–iiiATD6-89 Adult Rib Diaphysis ix–xATD6-90 Juvenile Vertebra Complete AtlasATD6-91 Adult Skull Apophysis mast.+temp.ATD6-107 Adult Metatarsal Ep. prox.+diaph.ATD6-108 Adult Rib Diaphysis iATD6-206 Adult Rib Head+diaphysisATD6-251 Juvenile Rib DiaphysisATD6-307 Vertebra Body ThoracicATD6-308 Rib HeadATD6-308 Rib DiaphysisATD6-309 Adult Vertebra Lamina CervicalATD6-312 Inf. Tooth Incisor Left Ul2 VI

Accessory experimental work

Two of us (IC and JR) were involved inbutchering the carcass of a chimpanzee thathad recently died. It was provided by thelocal Animal Protection Association ofTarragona (Spain). We found that skinning,

dismembering and defleshing this animalhelped us to understand better some of thecuts observed on the human remains fromthe Aurora Stratum.

We experimented with flakes made fromlimestone, quartzite, Cretaceous flint andNeogene flint, the different raw materialsused at Atapuerca to make the stone toolsassociated with the Aurora Stratum fossils.Two lamb forelimbs were butchered by oneof us (YFJ), using implements made withthese four types of stone. Analyses of this

598 . - ET AL.

experiment are in progress and the resultswill be published soon.

Fossil assemblage

The human collection of TD6-AuroraStratum consists of 92 fossils that includedental, cranial and postcranial elements(Table 1). Nonhuman faunal remains fromTD6-Aurora Stratum have also been studiedfollowing identical methods of analysis.Results from this analysis have been in-cluded in a separate paper (Díez et al., 1999)to interpret site formation processes.

Spatial co-ordinates (X, Y, Z) are notedduring excavation for every fossil, stone tool,coprolite, concretion, limestone rock (biggerthan 10 cm), small mammal accumulation,or artefact, and plotted on to a map. Slope,orientation, measurements and descriptionsare noted for a given square. Animal remainshave been labelled according to the squarewhere they were found and the relatednumber of the find, whereas human fossilshave been labelled with ATD6- followed bythe number of the specimen [Figure 2(b)].

All sediment were wet screened (from5–0·5 mm mesh). Fossils recovered duringthe 1994 season were systematicallyimmersed in a preservative solution(Paraloid, a synthetic resin). The use ofpreservative may cause problems for theanalysis of cutmarks or superficial damageusing scanning electron microscope (SEM).This problem was anticipated, so fossilswere examined in the field laboratorythrough a binocular light microscope beforetreatment. This revealed that the highlymineralized condition of the TD6 fossilsmade it unnecessary to strengthen them, soimmersion in perservative was discontinued.

The fossil collection from the AuroraStratum (faunal and human) was examinedwith the aid of a Leica Wild MZ8 from 6·3to 50# binocular microscope. Some speci-mens were analysed using scanning electronmicroscopy (SEM). Two different SEMs

were used. A Philips XL20 housed at theMuseo Nacional de Ciencias Naturales(Madrid) and an ISI ABT55 SEM fittedwith an environmental chamber, operatingin the back-scattered electron emissionmode at 20 kV, which is housed at TheNatural History Museum (London). Thistype of microscope enables specimens tobe directly analysed with no necessity forcoating (Taylor, 1986), and it has beenextensively used.

High-resolution replicas were made usingEXAFLEX CG Injection type. Positivereplicas were then made using an epoxyresin (Nural-23). These replicas were coatedwith gold-palladium and analysed using thePhilips XL20 secondary electron emissionmode at a standard accelerating voltage of10 kV.

Identification of anatomical elements

Each human fossil has been identified asfollows:

- body part;- segment and portion (diaphysis, proximal

end, and distal end; complete; lateral;body; process; arch);

- age (juvenile/adult/infantile) determinedfrom dental eruption and wear, as well asepiphyseal fusion and bone texture.

The large mammal faunal composition,identified in TD6 Aurora Stratum are asfollows, H. antecessor, Mammuthus sp., Ursussp., Canidae indet, Vulpes sp., Panthera sp.,Felis sp., Muselidae indet, stenoid Equus,Stephanorhinus etruscus, Cervus elaphus,Megaloceros sp., Dama dama sp., Capreolussp., Sus scrofa, Bison sp (Garcıa and van derMade, pers. com.). Anthropologists fromthe Atapuerca research team identified thehuman remains (listed in Table 1). Theminimum number of individuals has beencalculated to be six according to detaileddental analysis (Bermúdez de Castro, pers.com.).

599

With regard to the rest of the fauna, wehave been working according to the follow-ing size classes: small (<115 kg), medium(115–350 kg), large (>350 kg) (see Díezet al., 1999 for site formation). For theanalysis performed on the faunal remains toidentify the site formation processes, thehuman collection and ‘‘possible Homo’’have been assigned to the small size class.

The relative abundances of skeletal ele-ments have been calculated by comparisonwith the expected numbers of each elementmultiplied by the minimum number ofindividuals.

Fracture- Length/width/thickness were measured on

all fossils with a micrometry calibre.- Peeling. This was described by White

(1992) and Turner & Turner (1999).Peeling is a type of fracture that occurredfrequently in the Mancos assemblage andhas also been seen in the Aurora Stratumfossil assemblage. It is defined as a rough-ened surface with parallel grooves orfibrous texture produced ‘‘when freshbone is fractured and peeled apart similarto bending a small fresh twig from a treebranch between two hands’’ (White,1992:140). Peeling was recorded aspresent/absent for each fossil.

- Percussion pits. These are pits of variablesizes and depths (Leroi-Gourhan &Brezillon, 1972; Blumenschine &Selvagio, 1988). They are considered tobe the impact point where a stone or anysolid matter struck the bone cortex andscarred the surface. Percussion pits areusually accompanied by abrasions andscratches caused by friction of the boneagainst the stone raw material thathammered it, or the anvil surface wherethe bone was resting when it was struck.Scratches may occur inside the pits aswell as the surrounding area [Figure3(a)], with all scratches having the samedirection. These pits and striae have been

named ‘‘percussion striae’’ (White,1992), ‘‘contrecoup’’ or ‘‘hammerstone/anvil scratches’’ (Turner, 1983). Thesepits and scratches were recorded aspresent/absent.

- Adhering flakes. This term refers to boneflakes that adhere to the fracture surfaceof a specimen. Curving incipient fracturelines, often hairline, which are subparallelto the fracture edge, set off these flakes.This condition was also recorded aspresent/absent.

- Conchoidal percussion scars have beendescribed and measured, following thenomenclature traditionally used in lithics(deep, marginal, cortical direct, inverse,medular flake, cortical flake, concave,convex, straight).

Tool-induced surface modificationDescription of the cut emplacement on thebone (metaphysis, epiphysis, diaphysis,articular area) and arrangement (distribu-tion: isolated marks/grouped/generalizedand orientation: oblique/transversal/longitudinal) were recorded for every cut-mark, chopmark or scrapemark, accordingto the size of the mammalian species.Lengths of striations have also beenmeasured (maximum and minimum lengthswhen sets of cuts occurred).

Cutmarks. Incisions or slicing marks havebeen analysed separately from saw cuts.Incisions or slicing marks were differenti-ated according to Schick & Toth (1993)as: incisions made with a flake edge withoutretouching, edge retouched on one face, andedge retouched on both faces [see Figures3(b) and 3(c).

Microscopic morphology of cutmarks isnot the only discriminating trait from othertypes of nonhuman induced striations.Cutmark arrangement (position andnumber of marks), placement on the ele-ment (e.g., muscle and ligament attach-ments), as well as the species affected,

600 . - ET AL.

are additional factors (Fernández-Jalvoet al., 1999; Olsen & Shipman, 1988), thatmay also indicate the objective of theprocessing activity (dismembering, deflesh-ing, skinning).

Chopmarks. These marks are the result ofstriking the bone surface with a sharp stonetool, leaving a deep, wide V-shaped scar.The action is related to cutting strongmuscle attachments or dismembering.

601

White (1992) states that the definition ofchopmarks is ambiguous and is rather simi-lar to percussion pits because both are theresult of directed blows on the bone. Whitesuggests that when percussion by a V-edgedhammer stone fails to crack a bone, aV-shaped pit may result, which is similar toa chopmark. Percussion blows are applieddirectly to the bone (the stroke is transmit-ted through the bone) with the main inten-tion being to break it, while chopmarksoccur when the bone is still covered bysoft tissue that absorbs the blow. Hence,chopmarks are probably related to dis-membering activities. Consequently, per-cussion blows leave a much rougher andless regular internal form than that seen inchopmarks.

Scraping marks. These are the result ofperiosteal and muscle removal by scrapingthe bone surface. This activity leaves a con-centrated series of parallel and superficialstriations on a broad area of the bone[Figure 3(d)]. When scraping-marks occuron long bones, they usually run parallel tothe longest axis of the bone. Scraping markshave been experimentally obtained by a vari-ation in the angle of the flake edge when atool is positioned oblique rather than per-

pendicular to the bone (Delpech & Villa,1992) but the width of the area affected ismore reduced and a single incision can berecognized [Figure 3(e)].

Figure 4 shows the % of survival (Brain,1969) represented at Aurora Stratum

%Si=(MNEi/NixMNI)#100,

where %Si=percentage of survival of ele-ment i, MNEi is the Minimum Number ofElement i found in the sample, Ni is theexpected number of element i in theskeleton, and MNI is the MinimumNumber of Individuals, which has beenestimated at six based on dental traits.

Results

Figure 3. (a) Scanning electron micrograph. ATD6-97. Detail of an impact notch or percussion markshowing scratches made during percussion. Several indications suggest this is a percussion mark (to breakthe bone already defleshed, for marrow extraction) instead of a chopmark (to dismember a bone stillcovered by meat). Scratches surround the impact mark indicating that the bone was already clean of meat.Cut-marks are interrupted by the impact mark, indicating that dismembering and filleting alreadyoccurred. Finally the impact mark appears parallel to the broken edge of the bone fragment suggesting thatthis was a failed try. (b) Scanning electron micrograph. ATD6-55 Infant clavicle and incisions made by annon-retouched flake edge. Notice the lateral irregularities have been recorded only along one side of thecut (right in this case), caused by resistance of the bone to the cut friction, and displaced bone on the sideof the striations. The lateral shoulders or ‘‘herzinian cones’’, in this case still attached to the bone(indicated by a black arrow), are directionality criteria. (c) Scanning electron micrograph. G17, n. 212fragment of long bone of unidentified species. The typical X shape is produced by a stone tool edgeretouched at both sides. The irregularity of the edge produces an X in a single motion as the angle of thetool changes during the cutting stroke (see Schick & Toth (1993)). (d) Scanning electron micrograph.H16, n. 166. Long bone fragment of a medium-sized animal showing abundant striations on the surface.The fragment was longitudinally broken, but in this case, striations are not associated to impact marks.These are scraping marks and they are associated with grease extraction or periosteum removal. (e)Scanning electron micrograph. Scraping mark obtained when a tool incises obliquely rather thanperpendicularly on the bone surface. Note the scraped area is more reduced and a single incision can berecognised.

The human sampleThe minimum number of elements, and thepercentage of survival are represented inFigure 4. Phalanges, isolated teeth, meta-podials, ribs and vertebrae are the mostcommon elements as these are the mostabundant elements in the human skeleton(56 phalanges, 32 teeth, 20 metapodials, 24ribs and 24 vertebrae). The completeness ofanatomical elements is shown in Table 1.

602 . - ET AL.

No complete cranial element (skull vault,mandible or maxilla) has been found in theAurora Stratum. Teeth are the only com-plete elements of the cranial skeleton,excepting incisor ATD6-48, which is badlybroken. There are very few complete ele-ments from the axial skeleton. One cervical

vertebra and one rib of an individual,together with the atlas and a clavicle of ajuvenile individual, are complete. Similarly,only two patellae represent complete limbelements. The skeletal parts with more com-plete elements are hands and feet (mainlyfoot phalanges).

The fragment dimensions of the AuroraStratum human fossil assemblage areshown in Table 2. Despite the differencesin natural size of these anatomical ele-ments, averages of the different fragmentsappear to us to be sufficiently similar tosuggest that there was an intense breakageactivity that led to a high degree of elementdestruction.

Ph

alan

ges

60

0

Man

dibl

e

NMI = 6

% S

urv

ival

50

40

30

20

10

Max

illa

e

Inci

sors

Mol

ars

Can

ines

Pre

mol

ars

Cla

vicl

e

Sca

pula

Hu

mer

us

Uln

a

Rad

ius

Met

acar

pal

Rib

s

Ver

tebr

ae

Sac

rum

Pel

vis

Fem

ur

Fib

ula

Tib

ia

Pat

ella

Cal

can

eum

Ast

raga

lus

Met

atar

sal

Human anatomical elements

16

200

200100

1616

22

0003

52

3431

Figure 4. Percentage of survival circles (lines) and Minimum Number of Elements (black bars) of thehuman remains recovered from Aurora Stratum.

Table 2 Fragment dimensions

CraniaRange(mm)

Mean(mm) S.D.

Length 10–76 35 17Width 8–45 20 12Thickness 4–25 10 6

AxialLength 24–256 69 63Width 10–68 24 15Thickness 5–23 15 11

Arms/legsLength 36–220 95 25Width 16–42 26 8Thickness 4–20 4 6

Hands/feetLength 11–128 30 77Width 5–33 16 12Thickness 4–26 10 5

Human modification of human fossil bonesBreakage of the human bones could not beanalysed using Villa & Mahieu’s (1991)methodology because it is based on longbones. As there are very few of these ele-ments (one fragment of femur and tworadii fragments), the resulting valuesobtained when applying Villa and Mahieu’s

603

methodology were unreliable. Our qualita-tive fracture analysis, however, considerspeeling, percussion marks, conchoidal scarsand adhered flakes (Table 3).

Crania. Heads are mainly represented byvarious skull fragments, two small fragmentsof mandible and four fragments of maxillae.The most complete specimens are a frontalfragment (ATD6-15) and a left zygomaticarch attached to a complete maxilla (ATD6-69). Nuchal skull bones are commonlyaffected by fracture (e.g. percussions,adhered flakes). The sides of the cranialvault are heavily cutmarked (e.g. temporalprocesses, occipital condyles and at pterion)corresponding to the biggest muscle attach-ments, such as sternocleidomastoid. Theother group of cranial elements affected bycutting and percussion is the face (jaws andzygomatic arches), which also has variousfirm muscle attachments. Only two smallmandible fragments were in the assemblage.Peeling and scraping marks occur on one ofthem (ATD6-63), indicating dismemberingand removal of the periosteum and overlyingtissue from the fragment.

A small temporal bone fragment (ATD6-16) shows a concentration of cutmarks run-ning along the ridge where the sternocleido-mastoid muscle attaches, joining the headand the trunk [Figure 5(a)], though it doesnot show traces of human breakage. On thecontrary, the face of a juvenile individual,specimen ATD6-69 represents a goodexample of fracture induced by humans[Figure 5(b)]. This specimen (ATD6-69)shows strong impact marks along the zygo-matic bone and the orbital margin of the leftside, and fracture edges also bear adheredflakes. Apart from that, the bone is heavilycutmarked, with long and intersecting inci-sions that affect several muscle attachments(nasalis, buccinator, levator labii superioris,levator anguli oris, and zygomaticusminor). The type of cutmarks observed onATD6-69 suggest incisions and sawing

motions, with the former extended all overthe face, probably to cut the levator muscles,and the second type (sawing), concentratedon the orbits and base of the zygomatic arch,associated with the position of origin of themasseter muscle. Most zygomatic archesfrom the Aurora Stratum are fractured, asthey are in human remains from NativeAmerican sites and Fontbregoua. White(1992) suggests that this patterned breakageis the result of either general percussion ofthe vault or a specific action to gain access tothe temporalis muscle.

Another area from the skull, which is alsoheavily cutmarked is the pterion (ATD6-60). this skull area bears several long cut-marks running obliquely all over its surface,as well as several conchoidal scars.

Peeling is also present in several skullfragments (Table 3) such as temporal, zygo-matic, mandible and occipital condyle.

Impact marks have been observed on fivedental elements from the lingual sidebetween the root and the crown (Table 3).All these teeth belong to the same individual(I). The teeth were discovered lying close toeach other in anatomical position, althoughno maxillary bone was preserved aroundthem.

Axial skeleton. The elements representedare 11 ribs, 11 vertebrae (including oneatlas and one axis) and three clavicles. Nosacra, pelves or scapulae have yet beenfound.

Articular heads with or without epiphysesor just epiphyses are the most frequentremains of the ribs (Table 3). One rib(ATD6-39) is almost complete and displaysmany marks of human processing. Theinner part of the rib has percussion marksand obliquely grouped incisions going fromtop right to bottom left, seemingly related tothe intercostal membrane and muscles. Afew scraping marks running longitudinallyalong the costal groove are possibly relatedto extraction of thoracic contents. The

604 . - ET AL.

Tab

le3

Bon

esu

rfac

em

odifi

cati

ons

onh

um

anre

mai

ns

rela

ted

toan

thro

pic

bre

akag

e

Con

choi

dal

scar

sP

ercu

ssio

nP

eelin

gA

dher

ing

flake

s

Cra

nia

l(2

5)A

TD

6-17

.T

emp

AT

D6-

17.

Tem

p9

frac

ture

dA

TD

6-19

.Z

ygom

.ar

chA

TD

6-14

.M

axill

ar,

nasa

lA

TD

6-84

.Z

ygom

atic

7to

ol-m

arke

dA

TD

6-49

.M

axill

arA

TD

6-63

.M

andi

ble

AT

D6-

58.

Mal

arA

TD

6-77

.O

ccip

ital

cond

ileA

TD

6-60

.P

teri

onA

TD

6-69

.A

lveo

-fro

ntal

AT

D6-

69A

lveo

-fro

ntal

Axi

al(2

5)A

TD

6-39

.R

ibA

TD

6-45

.L

umba

rve

rteb

raA

TD

6-44

.A

xis

5fr

actu

red

AT

D6-

75.

Cer

vica

lve

rteb

ra9

tool

-mar

ked

AT

D6-

79.

Rib

.A

TD

6-80

.C

ervi

cal

vert

ebra

AT

D6-

80.

Cer

vica

lve

rteb

raH

and

s/fe

et(2

3)A

TD

6-46

.II

Pha

l.ha

ndA

TD

6-46

.II

Pha

l.ha

nd3

frac

ture

dA

TD

6-10

7.II

Mtt

sA

TD

6-59

.II

Mtc

p.5

tool

-mar

ked

Lon

g-b

ones

(5)

AT

D6-

43.

Rad

ius

2fr

actu

red

2to

ol-m

arke

dA

TD

6-76

Fem

urA

TD

6-76

.F

emur

.

Den

titi

on(1

4)A

TD

6-1.

Lef

tL

C(l

ater

al)

5fr

actu

red

(ind

iv.

1)A

TD

6-8.

Rig

htU

P4

(lin

gual

)A

TD

6-9.

Lef

tU

P4

(lin

gual

)A

TD

6-10

.R

ight

UM

1(l

ingu

al)

AT

D6-

11.

Lef

tU

M1

(lin

gual

)A

TD

6-52

.L

eft

LI1

(occ

ulus

al)

605

Figure 5. (a) ATD6-16. Fragment of temporal, showing numerous cutmarks transversally along the bothends arrow. These cutmarks affect the mastoid crest where the sternocleidomastoid muscle is attached.Location and distribution of cut marks are suggestive of dismembering (detachment of the head) anddefleshing activities. (b) ATD6-69. Holotype of Homo antecessor. The face of this young individualshows intensive cut marking on its surface to detach meat from bone and cut all muscles associated togesture movements. Slicing and sawing marks are frequent (black arrow), together with several failedimpacts (empty arrow) to separate the face from the zygomatic processes. (c) Scanning electronmicrograph. ATD6-55 Infant clavicle. This specimen shows several parallel cutmarks and transversalfracture made when the bone was still fresh. These deep and precise cutmarks affect attachments ofdeltoid and pectoralis major muscles from the chest. The trapezius attachment (the neck muscle) from thisclavicle is also heavily affected. (d) Scanning electron micrograph. ATD6-55. Cutmark directionality (seeBromage & Boyde, 1984). Frequently cuts are unidirectional, but here it is an example of precise sawingmotion. The lateral ‘‘Hertzinian cones’’ at the right side of the striation and marked by black half trianglesand black arrows indicate opposite directionality and suggest the cut was made in at least two motionsgoing up and down.

articular end of a rib (ATD6-79), alsoalmost complete, shows peeling. Two otherrib fragments (ATD6-85 and ATD6-251)have cutmarks. ATD6-85 has cutmarks onboth outer and inner surfaces of the rib, withincisions (4·5–5 mm) forming groups alongthe diaphysis that could also be related toviscera extraction with ATD6-39.

Among the vertebrae four are cervical(one complete atlas, two laminae and onetransverse process); three are thoracic (onespinous process and two bodies); and twoare lumbar (one transverse process and onevertebral body). Three vertebrae are affectedby peeling, one at the lamina edge of acervical vertebra, and two at the transverse

606 . - ET AL.

607

processes of a lumbar and a cervical ver-tebra. Adhering flakes appear at the spine ofthe axis and the lamina of another cervicalvertebra. Two vertebrae show slicing marksthat are grouped and could be related tothe butchering of the semispinalis capitatismuscle.

Each of the three clavicles has marks thatwere made by stone tools. The completeclavicle of a juvenile (ATD6-50) has a singleincision affecting the trapezius muscleattachment. The infant half clavicle (ATD6-55) is intensively cut along the edge wherethe subclavius muscle attaches [Figure5(c)], and there are a few cutmarks on theattachment of pectoralis major. All of thesecutmarks appear to be related to removalof muscle to permit disarticulation of theclavicle. These cuts show sawing motions[Figure 5(d)] according to directionalitycriteria (Bromage & Boyde, 1984). Theinfant clavicle is broken at about mid-shaft, lacking the medial half, wherethe strong sternocleidomastoid muscleattaches. The broken edge and the type offracture is congruent with breakage duringdismembering, though no adhered flakes orpeeling can be distinguished. There is anoblique fissure that could be the result oftrauma from the breakage process duringdismembering.

Legs and arms. Apart from the two patellae(ATD6-22 and ATD6-56), a small femurfragment (ATD6-76) and two radii frag-ments (ATD6-43 and ATD6-21) are theonly representatives of the appendicularskeleton. Neither of the patellae displays

evidence of human modification. However,humans seriously damaged the radiusshaft ATD6-43. This element was foundcomplete but diagenetically broken in situ.Peeling affects the distal end of this radius[Figure 6(a)]. Incisions run obliquely fromthe top right to the bottom left, covering theanterior border of the diaphysis, with ahigher density of cutmarks towards thedistal metaphysis affecting the pronatorquadratus, as well as the attachment offlexor digitorum. Cutmarks are interruptedby the characteristic fibrosity of peeling.

Finally, the only long bone of thickdiameter recovered from the small area ofexcavation is a fragment of femur shaft(ATD6-76). This fragment has been hitheavily producing spiral fractures at bothends and multiple and successive percussionmarks on both posterior and anterior sides[Figure 6(b)]. The strong hammering actionon this piece has also produced striations(anvil abrasions according to Turner andWhite) associated with percussion scarmarks. These scar marks seem to beassociated with longitudinal breakage of theshaft, probably to extract bone marrow.Damage due to percussion has been soheavy that possible cutmarks have beenobscured.

Figure 6. (a) ATD6-43 human radius. This specimen has abundant cutmarks (empty arrow) from righttop to left down all along the length of the bone affecting the pronator quadratus, as well as the attachmentof flexor digitorum. The distal end of the radius has been broken showing peeling (black arrow). The bonewas not longitudinally opened to extract any marrow content. (b) ATD6-76 Femur fragment. This bonewas heavily hit to break it in order to open the shaft and extract the marrow. Black arrows point out someof the impacts. (c) ATD6-59 human metacarpal showing cutmarks all along the anatomical lateral (andtwo ends arrow) edge where dorsal interosseous muscle attaches. (d) H16, n.3 Impact pits on tibia ofbovid. Impact scars (some of them pointed out by black arrows) are similar to those seen in Figure 6(b)of a human femur. Marrow extraction seems to be the purpose of this heavy damage.

Hands and feet. No tool damage or inten-tional breakage has been found on the twocarpals found in the Aurora Stratum, a com-plete capitate (ATD6-24) and a distalhamate fragment (ATD6-23). There are 16phalanges and five metapodials. The humandamage observed on these elements is not

608 . - ET AL.

homogeneous, with some of the elementsheavily affected and others unaffected. Onemetacarpal has been damaged at the proxi-mal end by peeling (ATD6-59) and onemetatarsal shows conchoidal scar marks atthe distal diaphysis (ATD6-107). Only onephalanx (second hand phalanx) has beenbroken (ATD6-46), with both peeling andpercussion marks at the proximal diaphysisprobably done during dismembering. Cut-marks have been observed on ATD6-59,ATD6-107, and ATD6-46 and on two morephalanges, ATD6-53 second hand phalanxand ATD6-30 first toe phalanx. Incisionsare present all along the anatomical edge ofthe second metacarpal ATD6-59 [Figure6(c)] at the insertion of the first dorsalinterosseous muscle. Incision marks onphalanges ATD6-30 (first toe phalanx) andATD6-53 (second hand phalanx) areoblique and mainly concentrated at themetaphyses. Those cuts at the diaphysis aretransverse in orientation.

Discussion

The ages and number of hominid in-dividuals from TD6 Aurora Stratum basedon dental traits (see Bermúdez de Castroet al., 1999) are as follows: two infants of3–4 years old (individuals II and VI); twoadolescents, one of about 14 years andanother of about 11 years (individual III,the holotype of H. antecessor, Bermúdez deCastro et al., 1997); and two young adultsabout 16–18 years old (individuals IV andV). The spectrum of age amongst largemammals in the Aurora Stratum is pre-dominantly juvenile and infant individuals,and the total MNI has been estimated at22 (see Table 4 and Díez et al., 1999 fordiscussion).

Skeletal partsHuman anatomical elements are represen-tative of all major skeleton areas (heads,axial, hands/feet, arms/legs), although they

are not fully representative of the wholeskeleton, element-by-element. Some ana-tomical elements are scarce or absent. Onlyone fragment of a femur, 2 radii and 2patellae are representative of limbs. Nohumerii, tibiae, ulnae nor fibulae have beenrecovered. The presence of other limb ele-ments such as phalanges, metapodials(from both hands and feet) and radii andfemur would suggest that this lack could besampling error due to the small area ofexcavation (2·8#2·5 m) rather than to anyselection of skeletal elements made duringbutchering. Furthermore, there is great dif-ficulty in identifying those elements that arehighly fragmented and appear mixed withother taxa of similar size and fragmentationrate. As a result, there are many fragmentsthat could be human, but their identificationremains uncertain at present.

As with the human material, othermammal skeletal parts are relatively wellrepresented at Aurora Stratum. Large sizedmammals, however, show an apparent lowrepresentation of axial elements in all taxa.This has been considered by Díez et al.(1999) to be the result of anatomical partselection by hominids to facilitate the trans-port of the carcass into the site (see Díezet al., 1999, for further implications).

Damage and cutmarks on limb bonesHuman anatomical elements that have asmall diameter with little marrow contentappear almost unbroken. Radius ATD6-43is almost complete and the other shaft(ATD6-21) lacks most of the ends but it hasnot been longitudinally opened for marrowextraction. This has also happened with sixribs, three clavicles, two vertebrae (out of11), the two patellae and 13 of the 16phalanges among the human remains. Themost damaged elements are skulls, mand-ibles, all maxillae, the femur fragment, andvertebrae (plus four ribs, one metacarpaland two metatarsal that are transverselybroken).

609

Similarly, this patterning is also observedon the fossil nonhuman animal remainsfrom the Aurora Stratum. A humerusof a small mammal (H16, n. 164) isalmost complete, as are most phalanges.Large- and medium-sized mammals havefew unbroken remains with only carpal–tarsal bones remaining complete. How-ever, a bovid phalanx, of potentially lowmarrow content, is broken (see Díez et al.,1999).

The patterning of the destruction of non-human animal and human bones in theAurora Stratum is consistent with thosebones that held the most nutritional value.With regard to humans, the only femurfragment (ATD6-76) has been struck andbadly broken, providing the strongest evi-

Number of remains (NR) and minimum number of individuals (MNI)

TD6-Aurora NR MNIAge

(Inf/Jv/Ad/Sen)Total weight

(kg)

Proboscidea 2 1 1Inf 1415Stephanorhinus 7 2 1Inf/1Jv 759Bison 56 2 1Inf/1Ad 682Equus 18 3 1Inf/1Jv1Sen 706Megaloceros 8 2 1Jv/1Sen 587Indet. large size 52 —Total large size 143 10Cervus 15 2 1Inf/1Ad 206Cervidae 95 1Indet. middle size 202 —Total middle size 312 3Dama 20 2 1Jv/1Ad 138Sus 1 1 1Ad 55Capreolus 5 2 1Inf/1Jv 10Homo 92 6 2Inf/2Jv/2Ad 239Possible Homo* 103 —Indet. small size 82 —Total small size 303 11Total small size without Homo 211 5Carnivorous 11 —Indet. 287 —Total 1056 24 8Inf/8Jv/6Ad/2Sen 4797

Age estimation and weight of the individuals represented in the site. The weight ofeach animal has been calculated according to Millar (1977, 1981) formula(NM=0·045 m0·89), with NM as the weight of a neonate and m as the adult weight andGR=0·04 m0·69, with GR being the weight increment calculated in gr/day. The adultweight has been obtained from Rodriguez (1997). The age of the animals, as well as theMNI, has been estimated considering tooth eruption and born out.

Table 4

dence for marrow extraction observed in thefossil human assemblage. Similarly, inten-sive percussion pits and impact scars havealso been observed on a fragment of bovidtibia (H16, n.3), also for the marrow extrac-tion [Figure 6(d)]. Conchoidal scars arefrequent on both nonhuman animal andhuman remains in similar proportions(Figure 7).

Peeling has been observed to be mostcommon on small sized animals and humansfrom the Aurora Stratum (Figure 7),whereas percussion marks and adheredflakes are more abundant on large andmedium-sized animals. The origin of peel-ing, related to breakage and dismemberingwhen bending the bones between the twohands, suggests that this difference in

610 . - ET AL.

16.0%

0%

PeelingAdhered + flakes

14.0%

12.0%

10.0%

8.0%

6.0%

4.0%

2.0%

Conchoidalscars Percussion

marksHominids

Small

Medium

Large

Figure 7. Small-medium-large-sized of mammals and hominids are compared taking into account humaninduced damage mainly caused by fracture. Note that adhered flakes and percussion marks are inverselyabundant from large to small mammals (where humans are excluded and represented apart). Thesedifferences seem to be related to different musculature and especially to different weights. Humans likesmall-sized mammal animals, have higher abundance of peeling which can be done by bending the bonebetween both hands, while percussion marks and especially adhered flakes indicate the use of a stonehammer to smash the bone.

patterning can be related to the weight of theanimal and bone size.

Peeling is observed at the distal end of theradius ATD6-43. Peeling interrupts cut-marks related to tendon and muscle cutting.This indicates that incisions were madebefore dismembering when the wrist andprobably also the hand were still connected.Similarly, superimposition of peelingover cutmarks has also been observed atFontbregoua, Mancos, and sites in Arizona,indicating that this is a common butcheringsequence.

Phalanges from the TD6 Aurora Stratumbear cutmarks, a characteristic observed herebut absent from any of the assemblages com-pared with the Aurora Stratum (Villa et al.,1986a,b; Turner & Turner, 1990; White,1992). Two phalanges (ATD6-53, hand,and TD6-30, toe) have cutmarks as themetaphyses, which are associated with thedismembering process. Another phalanx(ATD6-46) displays peeling at the proximalend and percussion at the diaphysis, associ-ated with crushing and dismembering [Fig-ure 8(a)]. Metapodials also show cutmarks,

611

Experimental cutmarksDue to the presence of limestone stone toolsassociated with the Aurora Stratum fossils,

we carried out an experiment using toolsmade from different raw materials, includ-ing limestone on bones of a lamb. Theexperimental cutmarks using limestonestone tool showed a strong similarity tostriations on ATD6-46 [Figure 8(c)]. Inlight of this experiment and the butcheringof the chimpanzee carcass, these marks aresuggested to be the result of holding thecomplete or almost complete finger betweenthe teeth and cutting small amounts of meatwhile feeding. From the TD6 assemblage(ATD6-52) known so far, there is onehuman tooth that has oblique cutmarks likethose described by Bermúdez de Castroet al. (1988) that could be interpreted asaccidental cutting during feeding. The dis-covery of cutmarks made by limestone toolssimilar to trampling marks (Andrews andCook, 1985) or hammerstone–anvil abra-sions (Turner, 1983) is important andfurther analysis is necessary, especially atsites where limestone is used as lithic rawmaterial.

Marks similar to those experimentallymade with limestone stone tools occur onlong bones of small-sized animals from theAurora Stratum [H16, n. 62, Figure 8(d)].These are located along the edges of thefractures. The experiment of cutting lamblimb bones with limestone tools showed thattheir edges were good enough to cut a fewgrams of meat, but they soon becameblunted, making further cutting difficult.These cutmarks are not isolated, but arefound in clusters [Figure 8(e)] suggestingdifficulties in cutting, and they are widerthan cuts made with quartzite or flint.

Damage and cutmarks on craniaHuman and nonhuman skulls are broken.Cutmarks are frequent at the strongestmuscle attachments (face muscles, tem-poralis and sternocleidomastoid). While thehuman vault has almost no cutmarks, facialbones have an abundance of stone toolmarks. We interpret this abundance of

especially on the lateral diaphysis ofATD6-59 that bears several oblique slicingmarks associated with dismembering whencutting the dorsal interosseous muscle, andpeeling at the proximal end [Figure 6(c)].The metatarsus ATD6-107 shows slicingmarks also associated with the dismemberingprocess. All this evidence indicates an inten-sive dismembering process of, at least, someof the hands and feet represented at thesite. Amongst the animal bones, only a bearphalanx (I16, n. 43) shows a cutmark on itssurface [Figure 8(a)]. This is interestingbecause both bears and humans walk on themetatarsals and phalanges, and they havesimilar tendon and muscular attachmentsand, therefore, they are cut up similarly.

It is difficult to interpret a set of striationsobserved on the dorsal side of a humanphalanx ATD6-46 [Figure 8(b)]. Crushingof phalanges and metapodials has beendescribed by White (1992) at Mancos andinterpreted as a dismembering process. Ourbutchery of a complete chimpanzee showedthe great difficulty of dismembering fingersand toes by a single butcher. Assistance wasrequired for this and the difficult processyielded almost no meat or marrow. Cut-marks from ATD6-46 were clear whenobserved under the light microscope,although our SEM examination showed thatthese marks were atypical, different frommost cutmarks observed on other speci-mens. They were similar to tramplingmarks, or to hammerstone–anvil abrasiondescribed by Turner and White in thehuman assemblages from AmericanSouthwest Arizona. Phalanx ATD6-46 has apercussion mark on the palmar side of thediaphysis and peeling breakage at the proxi-mal end. These atypical marks, therefore,could have resulted from dismemberingdamage (hammerstone–anvil abrasion).

612 . - ET AL.

613

Figure 8. (a) I16, n.43 cut mark on bear phalange, the only nonhuman phalange with cutmarks. (b)Scanning electron micrograph of scratches at ATD6-46 showing transversal striations affecting the wholesurface. These scratches cut the flexor digitorium tendon attachment, apparently related to dismemberingtasks, probably while eating. The striations have a flat cross section and organized as random clusters (seetext for discussion). (c) Scanning electron micrograph of experimental cutmarks made with limestoneimplements on lamb limb bones. Marks were made when filleting. These cutmarks are not isolated, butorganized forming clusters as a result of difficulties experienced in cutting. During the experiment, theedge was not retouched to analyse the microwear traits, but in natural conditions the edge probably hadto be retouched several times to be effective. (d) Scanning electron micrograph. H16 n62 long bonefragment of a small-sized animal. This specimen has several sets of striations all along the broken edge. Asobtained experimentally, several cuts may form wide grooves with a wide diameter, formed by severalincisions. Sometimes individual cut marks (shown by white arrows) can be distinguished. Note that theseindividual incisions show irregularities at both sides of the cut, indicating that the implement wasretouched on one side (see Methods, Types of incisions). (e) Scanning electron micrograph ofexperimental cut marks made with a limestone artefact. The striation is much wider than striations madewith flint or quartzite, forming clusters of several incisions (as the groove marked by a black arrow).

cutmarks on the face, and that found on thetemporal and the nuchal areas, as evidenceof meat extraction and of the dismemberingprocesses, respectively. There is, however, asingle cutmark on the ATD6-15 frontal thatmight suggest skinning processes. Four skullfragments of small-sized animals also havecutmarks, probably related to skinning.Peeling is frequent on human skull frag-ments and it is also present on one of thetwo mandible fragments. Among nonhumananimals, peeling has been observed on skullsof small- and medium-sized herbivores, butthere is none in the skulls of large-sizedanimals.

At other sites with evidence cannibalism,there are more complete skulls than at TD6.The abundance of cutmarks on temporalbones and facial bones at TD6 has also beenobserved at Fontbregoua (Villa et al.,1986b), while White (1992) described ahigher incidence of cutmarks on the vaultthan on the facial area. Turner & Turner(1992) also found extensive facial damage atseveral sites from Arizona (Pollaca Wash,Leroux Wash, House of Tragedy, CanyonButte, and others). Villa et al. (1986a,b)found more marks on human facial bonesthan on animal faces. These differences wereinterpreted by these authors as possibleritual, also indicative of exocannibalism.Turner & Turner (1992) make a similarsuggestion regarding exocanibalism, and

based on the intensive facial damage, pro-posed violence and destructive intent ofmutiliation of a possible enemy. White(1992) suggests that the destruction of facesis also the result of gaining access to thebrain.

White (1992:207) proposes the followingprocessing technique at Mancos, ‘‘. . . thehead was heated while intact. Percussionfollowed heating and was presumablydirected toward removal of the brain tissues.The route of the easiest entry, through thefrontal and/or parietal, was followed.Percussion-related abrasion, and damage ofthe dentition, were coincident with fractureof the vault.’’ Turner has shown that craniainvolved in violence, but not cannibalism,have facial damage of various sorts anddegrees. This noncannibalistic massive facialdamage is abundantly illustrated in Turner& Turner (1999). However, White observesthat most cutmarks seen on the vault sug-gests that the scalp was removed at leastfrom some heads before burning, either toavoid the smell of burnt hair or as a trophyacquisition. Less facial damage and abun-dant intact mandibles at Mancos couldtherefore be explained by heating whichwould make face and head muscle attach-ments easier to remove.

Villa et al. (1986a,b) found that theNeolithic people from Fontbregoua did notuse fire during body processing, so that

614 . - ET AL.

damage to the faces and skulls is similar tothat observed in the Aurora Stratum.

Breakage and cutmarks found on theAurora Stratum human faces suggestdetachment of the cheeks, strongly affixed tothe bone by muscles (levators, buccinator,and nasalis). Breakage of the zygomaticarches is necessary in order to remove thetemporalis muscle so as to open the vault foraccess to the brain tissues. Cutmarks ontemporal bones indicate separation of thehead from the trunk. Our chimpanzeebutchering produced cutmarks on the faceand skull similar to those observed on theAurora Stratum hominids. Unfortunately,this animal had been autopsied (trepana-tion) so breakage of the vault or face to gainaccess to the brain could not be performedto compare with the Aurora Stratumhominids.

Cutmarks and damage on skulls and facesfrom TD6 Aurora Stratum are similar tothose from Fontbregoua. We believe thatdifferences between human and nonhumananimal treatments are due to differences inmuscle arrangement and attachment, andthe result of accessing the brain, cuttingmeat and skin off the heads, with no ritual,trophy or violence involved. A different pro-cess is observed on the Bodo skull (Ethiopia)with marks around the eye sockets (White,1985), instead of sawing and intensive cut-ting as described for the specimens fromAurora Stratum.

Apart from differences due to the use offire during processing, White (1992) alsomentions that the nuchal region has a lowfrequency of cutmarks (abundant at TD6)suggesting to him that the upper cervicalvertebrae were removed from the body alongwith the head.

White (1992) also describes tooth damageas a result of burning, but some as a resultof hammerstone–anvil abrasion. Severalhuman teeth from the Aurora Stratum(Table 3) have been found to have impactscars at the crown–root interface on the

lingual sides, and on the occlusal surface.This damage pattern could be explained asthe result of blows on top of the vault(frontal and/or parietal) while the teethrested against a hard stone surface. Thisscenario could explain the fact that severalteeth from individual I (were affected bypercussion at the lingual interface of crown–root (ATD6-8, 9, 10, 11). They were foundclose to each other, almost in anatomicalconnection, with no remains of the maxillarybone. Differential preservation of bone/teethis unlikely given that fragile infant remainshave been preserved, as well as their peri-mortem modifications [Figure 5(c)].

Damage and cutmarks on the axial skeletonOther ribs, vertebrae and clavicles representthe human axial skeleton at Aurora Stratum,since no pelves and scapulae have beenidentified. Again, the small area of excava-tion may explain the absence of missingskeletal elements (e.g. presence of femur butabsence of pelvis and tibia, or presence ofmost elements of the shoulder girdle butabsence of scapulae). The clavicle is one ofthe best represented anatomical elementsfrom the Aurora Stratum. All have signs ofhuman activity. Ribs and vertebrae are alsowell represented, with much evidence ofpeeling and/or percussion breakage, as wellas cutmarks indicating muscle cutting andtorso dismemberment, and accessing ofthe viscera. Similar processes have beenidentified on animal remains (Table 5),with abundant peeling and percussionon vertebrae and ribs of all these sizeclasses.

Turner observed that there was anabsence of vertebrae or that most of themwere crushed at the prehistoric Arizona sitesstudied by him (Turner & Turner, 1995).Turner has considered this absence ofvertebrae as a characteristic trait of cannibal-ism. He explains the low representation ofvertebrae as a result of their having first beencrushed on an anvil stone and the fragments

615

then boiled to facilitate oil extraction. Hesuggests this hypothesis based on ethno-graphic descriptions of the boiling of animalbones for marrow extraction. White (1992)has also commented on the reduction ofvertebrae from Mancos. It is interesting tosee that this absence does not occur atAtapuerca. In fact vertebrae are in similarproportion to or even higher than meta-podials or phalanges. As there is no evidenceof fire at Atapuerca in the Aurora Stratum,we would not expect vertebrae reduction.Absence of vertebrae is evident in theNeolithic assemblage from Fontbregoua,which Villa et al. (1985) consider as dueto humans having moved the discardedbones into ‘‘amas’’ (discard features). Smallelements, like vertebrae, could have beenlost during discard of butchered bones.Vertebral damage in the Aurora Stratummaterial is frequent, with specimens affectedby cutmarks, peeling, or vertical archesbroken due to percussion. This is con-sidered mainly due to dismembering,defleshing and crushing the spongy boneportions.

Table 5 Aim of the action deduced from the type of mark, cutmark organization and bone areaaffected

Butchering Large Medium Small Homo

Heads 1 mandible (F) 4 skull frags. (1F, 4S) 2 maxillae (2F)1 mandible (F)4 skull frags.

(2F, 2D, 1S)Axial 2 ribs (1F, 1E) 9 ribs (8F, 1D) 13 ribs (11F, 2E) 4 ribs (4F, 2E)

1 vertebra (D) 3 vertebrae (2F, 1D) 3 clavicles (2F, 1D)2 vertebrae (1F, 1D)

Limbs 3 femurs (3F) 2 femurs (1F, 1P) 1 femur (F) 1 femur (F, M)2 humerii (1F, 1P) 1 ulna (F) 2 tibiae (2F, 1P) 1 radius (F, D)3 tibiae (2F, 1M) 2 humerii (2F) 1 ulna (F)1 radius (F) 2 tibiae (1F, 1D) 1 long bone (F)3 long bones (2F, 1P) 11 long bones (9F, 2P) 1 scapula (F)6 metapodials

(4F, 1P, 1M)1 scapula (F) 3 flat bones (3F)1 coxal (F)2 metapodials (2F)

Extremities 2 phalanges (2D) 1 sesamoid (D) 2 metapodials (2D, 1P)1 phalanx (F) 3 phalanges (3D)

S=skinning; F=filleting; D=dismembering; M=marrow extraction; E=evisceration; P=periosteum removal.

Comparison with other sitesFinally, we have compared all tool-inducedmodifications observed on the humanremains of the Aurora Stratum with othersites as far as the data provided by differentauthors (Turner’s studies, Villa et al.,1986a,b; White, 1992) allow. Differencesregarding cutmarks on human remains havebeen discussed by White (1992:327), whocompared several sites studied by Turner(between 1% and 4·6% of cutmarkedfossils) and Fontbregoua (46·4%), with sitesanalysed by himself (Mancos 5MTUMR-2346 11·7% and Yellow Jacket 5MT-32·6%). In the Aurora Stratum, 25% of thehuman remains display cutmarks. Whitefeels the very high percentage seen inFontbregoua is because these data wereobtained after refitting, while the other siteswere recorded before refitting. Our datafrom Aurora were obtained before refitting.Cutmarks are more abundant in the AuroraStratum, probably because most anatomicalelements recovered are bones with little meatand strong attachments (such as faces, clav-icles, ribs and phalanges). This is congruent

616 . - ET AL.

18.0%

0.0%

PeelingAdhered

flakes

14.0%

12.0%

10.0%

8.0%

6.0%

4.0%

2.0%

Conchoidal scars Percussion

marksYellow jacket

Mancos

Aurora

16.0%

Figure 9. Diagram of human induced damage due to fracture in Mancos (Colorado), Yellow Jacket(another cannibalistic site from Colorado) and in TD6 Aurora Stratum. Note that the tendency observedin Figure 7 is followed for adhered flakes, which is less frequent at the Arizona sites compared with TD6site. Peeling, however, is not as common as in small-sized animals and humans in TD6, but it is still higherthan in large and medium-sized animals at TD6 (see Figure 7). Percussion marks have been marked muchmore on the bone surface of the Arizona sites than at TD6, inversely to conchoidal scars which are morefrequent at TD6. These differences seem to be related to the influence of fire at Mancos and YellowJacket, facilitating dismembering processes and reducing breakage tasks. Further, bones subject to heatingand boiling become softer and ductile (Mayne, 1990) and percussion marks are more easily recorded ontheir surface as observed at Mancos and Yellow Jacket assemblages.

with observations made by White’s (1992)analysis of element-by-element occurrenceof cutmarks (see White, 1992:328; Figure12.28) as well as the influence of fire, asdiscussed above and below.

Descriptions of the processing of differentanatomical elements is described by each ofthese authors, although data for conchoidalscars, percussion marks, peeling, and ad-hering flakes are scarce or incomplete. InFigure 9 we compare types of breakage inthe Aurora Stratum human remains withcomparable data provided by White (1992)from Mancos 5MTUMR-2346, and anothercannibalized human assemblage named

Yellow Jacket 5MT-3 from Colorado(1025-50 AD, also of the Anasazi culture).Differences between the Aurora Stratumand the Anasazi assemblages are conspicu-ous and understandable. Conchoidal scars,adhered flakes and peeling appear moreabundant in human remains from theAurora Stratum, in contrast to percussionmarks, which are more abundant on humanbones from Mancos and Yellow Jacket.This, in our opinion, indicates differenttreatment and damage due to the lack offire among the early Pleistocene hominids ofthe Aurora Stratum. The influence of fire onthe late prehistoric American Southwest

617

18

J

G

16Trench section

I

H

Heads Axial Limbs

Hands Feet

17

N

Aurora Stratum

Figure 10. Plan of human fossil bones. Heads/axial/limbs/feet and hands are represented separately. Noorganization or differential distribution of any of those skeletal elements can be differentiated. Thedistribution is random and mixed.

assemblages helped to make muscle attach-ments easier to remove, facilitating dismem-bering processes and, therefore, reducingcutting and breakage tasks associated withdismemberment. Indication of this effecthas been already discussed above withregard to skull treatment, which showed a

lower incidence of cutmarks compared toAurora. Further, osseous tissues subjectedto heating and boiling become softer andmore ductile (Mayne, 1990) and percussionmarks are more easily recorded on theirsurface as observed at Mancos and YellowJacket assemblages.

18

(a)

G

16

Y

I

H

17

Aurora Stratum-TD6-PLAN REFITTING

X

J

18

–450

–55016

Dep

th (

cms.

) –475

–500

–525

17

Aurora Stratum-TD6-TRANSVERSAL SECTION

REFITTINGS

(b)

619

Bone distribution of human and nonhumananimal remains (pattern of post-processingdiscard)The distribution of human remains fromthe Aurora Stratum seems to be random inthe area of excavation. They are mixed withthe rest of the fauna and artefacts (Figure 2).There is no clear pattern in the distributionof the different parts of the human skeleton(Figure 10), even though axial elementshave not been recovered from the northernpart of the excavation area. It is also truethat fragments of vertebrae and ribs are themost difficult anatomical elements to distin-guish from other taxa of similar body sizeespecially at a site such as the AuroraStratum, where breakage has such a highincidence. It can therefore be said that arandom arrangement characterizes the dis-tribution of human and faunal remains. Theunder-representation of axial elements insome parts of the excavation does not haveany particular taphonomic or behaviouralimplication.

Bone fragments of both nonhumananimal and human remains have beenrefitted both horizontally and transversely[Figure 11(a) and (b)], with some verticalrefitting of more than 10 cm against theslope. These refittings suggest that the sitewas not abandoned for long periods, andsupports a relatively short period of time forsedimentation.

During excavation, human remainsseemed to be in a slightly higher abundanceat the intermediate part of the AuroraStratum thickness, and this was especiallyevident at the west-central side of theexcavation (less than 0·5 m2). Apart fromthis zone, human fossil bones have beenrecovered from the whole thicknessof the Aurora Stratum. Further extensionof the excavation area would test this

observation. No differences with regard tothe distribution of other taxa have beendetected, with a rather homogeneousdistribution of all sizes classes amongsthumans, throughout the whole thickness ofthe Aurora Stratum.

Figure 11. (a) Plan section of Aurora Stratum (horizontal refitting). (b) Longitudinal section (vertical refitting).

Type of cannibalismThe Aurora Stratum, therefore, is character-ized by: (1) analogous butchering techniquesin humans and nonhuman animals such asskinning, filleting, dismembering, marrowextraction, evisceration and periosteumremoval (Table 5). However, we shouldallow for anatomical differences betweenhumans and animals. A higher frequency ofpeeling appears on small-sized animals andhominids (Figure 7), probably becausebones from these gracile groups can bebroken and bent using both hands. Large-and medium-sized animals are much morerobust and hand strength is not enough todismember and bend bones. Human faceshave been seen to have strong muscleattachments that make them likely to havemore cutmarks and modifications than otheranimals. (2) Similar breakage patterns toextract the marrow. Percussion and con-choidal fracture has been observed on large,medium- and small-sized animals andhumans, as a result of breaking the bone toextract the marrow (Figure 7). Particularly,a tibia of bovid and the human femurfragment have both been heavily struck[Figure 6(b) and (d)] in order to break themand extract the marrow. (3) Identical patternof post-processing discard of humans andanimals. Remains of human and nonhumananimals are randomly dispersed with nospecial arrangement of any one of the taxa[Figure 2(b)]. (4) Comparison between theAurora Stratum human samples and butch-ered human assemblages from other sites

620 . - ET AL.

more recent in age, where cannibalism hasbeen considered to be proven (AmericanSouthwest; Neolithic of Fontbregoua,France), show similar butchering tech-niques. There are, however, some differ-ences that have been related to the influenceof fire. The use of fire, and the fact thatboiling and roasting of bones facilitatesmuscle detachment from the bone, reducesthe amount of cutting needed to deflesh acarcass. The softer texture of both boiledmeat and bone means that impact marks areleft more easily than on bones that werenot cooked or heated. Fire also helps indismembering and breaking the bone.

In summary, butchering techniquesobserved in the Aurora Stratum were aimedat meat and marrow extraction. The humanremains recovered from the Aurora Stratumcave deposit suggest that they were thevictims of other humans who brought bodiesto the site, ate their flesh, broke their bonesand extracted the marrow, in the same wayas they were feeding on the herbivores alsopreserved in this stratum.

No ritual treatment can be suggested inthis assemblage. Nutritional purposes arepresumably the cause of this case ofcannibalism. This type of cannibalism isdivided by definition into (a) survival, wherecannibalism is incidental or a short-termmeasure, and (b) dietary or gastronomiccannibalism, which is associated with longperiods in which humans are feeding onother humans, as part of their regulardiet.

With our present state of knowledge,there are unanswered questions that make itdifficult to distinguish between survival andgastronomic cannibalism. For instance, theexact time span (number of years) repre-sented by Aurora Stratum, or the actualnumber of individuals exploiting the humanand animal remains recorded at the sitecannot as yet be rendered precisely.

Some other indications may help to pro-vide better answers. Mediterranean pollen

(Pistacea and Olea) has been found at TD6,suggesting that the climate was not severebut temperate. The mammal communitystructure suggests an holartic forest as theenvironment for TD6 (Rodríguez, 1997),cooler than suggested by the sedimentology(Aguirre & Hoyos, 1992) and pollen (GarcíaAntón, 1995), but still temperate. Thespecies diversity (see Table 4) recorded inthe Aurora Stratum is the richest found atany level from Atapuerca. Large, mediumand small-sized herbivores were butchered.At least 22 individuals, with infants,juveniles and young adults as the main agespectrum, and only two senile individuals oflarge-sized animals, have been recognized.The weight of this food supply has beenestimated at almost 5 metric tons, includingbones and meat (Table 4, see Díez et al.,1999).

If it is assumed that the Aurora Stratumrepresents a single incidental and shortevent, then the environmental conditions,the high diversity of fauna available tohumans, and the potential food supplyfound in the site apparently do not justify astarvation period that could have forcedthem to consume other humans as a survivalstrategy. This should then be consideredgastronomic cannibalism. Equally, if theAurora Stratum event represents a biologi-cally long period of time (tens or hundred ofyears), then the distribution of butcheredhominids through the whole thickness of theAurora Stratum indicates that humans wererepeatedly feeding on other humans for thisperiod of time. This also can be modelled asgastronomic cannibalism by its definition,indicating that humans were part of the dietof other humans.

Acknowledgements

We are deeply grateful to Professor P.Andrews, Professor E. Aguirre, Dr P. Villaand Dr J. M. Bermúdez de Castro for valu-able discussions on this topic during the

621

preparation process. We thank Professor P.Andrews, Professor C. Stringer, Dr T. Kingand anonymous referees for comments onthe manuscript that have greatly improvedthe final vesion. We are further grateful toDr N. Toth and Dr K. Schick, for discus-sions on tool-induced damage and handed-ness. The Sociedad Protectora de Animalesof Tarraco (Spain) provided us with an adultchimp carcass on which to practise dissec-tions and butchering. YFJ is thankful toProfessor C. G. Turner II and Dr P. Villawho kindly provided all their publicationsrelevant to this topic, to Belén Márquez forhelp and participation in the experimentalwork with different stone tool raw materials,and to Dr Begona Sánchez for assistancewith the cutting experiment. The excellentand professional work of restoration hasbeen of great value for our study. Thanks aregiven to the SEM Units and Photo Units ofthe NHM and MNCN. We thank the direc-tors of the Dolina project, J. M. Bermúdezde Castro and E. Carbonell for inviting usto participate in this monograph. Thanksare also due to Professor Leslie Aiello forplanning this special issue. This project isfunded by CICYT (PB93-066-C03-03) andJunta de Castilla y León. YFJ was alsogranted aid by the European Communities(ENV4-CT96-5043).

References

Aguirre, E. & Hoyos, M. (1992). Climate record incave deposits. The Atapuerca TD case. In (G. J.Kukla & E. Went, Eds) Start of a Glacial, NATO ASIser I, 3, pp. 127–136. Heidelberg: Springer Verlag.

Andrews, P. & Cooks, J. (1985). Natural modifica-tions to bones in a temperate setting. Man 20,675–691.

Arens, W. (1979). The Man Eating Myth: AnthropologyAnthropol. Oxford: Oxford University Press.

Askenasy, H. (1994). Cannibalism, from Sacrifice toSurvival. Amherst, NY: Prometheus Books.

Bermúdez de Castro, J. M., Bromage, T. G. &Fernández-Jalvo, Y. (1988). Buccal striations onfossil human anterior teeth: evidence of handednessin the middle and early Upper Pleistocene. J. hum.Evol. 17, 403–412.

Bermúdez de Castro, J. M., Arsuaga, J. L., Carbonell,E., Rosas, A., Martínez, I. & Mosquera, M. (1997).A hominid from the Lower Pleistocene of Atapuerca,Spain: Possible ancestor to Neandertals and modernhumans. Science 276, 1392–1395.

Binford, L. R. (1981). Bones: Ancient Men and ModernMyths. New York: Academic Press.

Blanc, A. C. (1961). Some evidences for ideologies ofearly man. In (S. L. Washburn, Ed.) The Social Life ofEarly Man, pp. 119–136. Chicago: Aldine.

Blumenschine, R. J. & Selvagio, M. M. (1988). Percus-sion marks on bone surfaces as a new diagnostic ofhominid behavior. Nature 333, 763–765.

Brain, C. K. (1969). The contribution of Namib desertHottentots to an understanding of australopithecinebone accumulations. Scientific Paper Namib DesertRes. Stn. 39, 13–22.

Bromage, T. G. & Boyde, A. (1984). Microscopiccriteria for the determination of directionality ofcutmarks on bone. Am. J. phys. Anthrop. 65, 359–366.

Bygot, J. D. (1972). Cannibalism among wild chimpan-zees. Nature 239, 410.

Carbonell, E., Giralt, S., Márquez, B., Martín, A.,Mosquera, M., Ollé, A., Rodríguez, X. P., Sala, R.,Vaquero, M., Vergès, J. M. & Zaragoza, J. (1995a).The litho-technical assemblage of the Sierra deAtapuerca in the frame of the Iberian MiddlePleistocene. In (J. M. Bermúdez de Castro, J. L.Arsuaga & E. Carbonell, Eds) Human Evolution inEurope and the Atapuerca Evidence/Evolución Humanaen Europa y los Yacimientos de la Sierra de Atapuerca,Vol. 2, pp. 445–455. Valladolid: Junta de Castilla yLeón.

Carbonell, E., Bermúdez de Castro, J. M., Arsuaga,J. L., Díez, J. C., Rosas, A., Cuenca-Bescós, G.,Sala, R., Mosquera, M. & Rodríguez, X. P. (1995b).Lower Pleistocene hominids and artefacts fromAtapuerca-TD6 (Spain). Science 269, 826–830.

Cook, J. (1986). Marked human bones from Gough’scave, Somerset. Proc. Univ. Bristol Spelaeol. Soc. 17,275–285.

Díez, J. C., Fernández-Jalvo, Y., Rosell, J. & Cáceres, I.(1999). Zooarchaeology and taphonomy of AuroraStratum (Gran Dolina, Sierra de Atapuerca, Spain).J. hum. Evol. 37, 623–652.

Fernández-Jalvo, Y., Díez, J. C., Bermúdez de Castro,J. M., Carbonell, E. & Arsuaga, J. L. (1996). Evi-dence of Early Cannibalism. Science 271, 277–278.

Fernández-Jalvo, Y., Andrews, P. & Denys, C. (1999).Cutmarks on small mammals at Olduvai GorgeBed-I. J. hum. Evol. 36, 581–589.

Flinn, L., Turner, C. G. II & Brew, A. (1976).Additional evidence for cannibalism in the South-west: the case of LA 4528. Am. Antiq. 41, 308–318.

García Antón, M. (1995). Paleovegetación del Pleisto-ceno Medio de Atapuerca a través del análisispolínico. In (J. M. Bermúdez de Castro, J. L. Arsuaga& E. Carbonell, Eds) Human Evolution in Europe andthe Atapuerca Evidence/Evolución Humana en Europa ylos Yacimientos de la Sierra de Atapuerca, Vol. 1,pp. 147–166. Valladolid: Junta de Castilla y León.

622 . - ET AL.

García, N. (1999). Carnivores from the Early Pleis-tocene hominid-bearing Trinchera Dolina 6 (Sierrade Atapuerca, Spain). J. hum. Evol. 37, 415–430.

Goodall, J. (1979). Life and death at Gombe. Nat.Geogr. Mag. 155, 592.

Jacob, T. (1972). The problem of head-hunting andbrain-eating among Pleistocene men in Indonesia.Archaeology and Physical Anthropology in Oceania 7,81–91.

Kurt, F. (1976). Large Bears, Grizmek’s Encyclopedia.Vol. 3.

Leroi-Gourhan, A. & Brezillon, M. (1972). Fouillesde Pincevent: Essai d’analyse ethnographique d’unhabitat Magdelenien. Gallia Prehistoire Supplement 7.

Mayne, P. M. (1990). Identification of cremationtrauma in cremated bone. M.A. Thesis. University ofAlberta, Canada.

Millar, J. S. (1977). Adaptative features of mammalianreproduction. Evolution 31, 370–386.

Millar, J. S. (1981). Pre-partum reproductive charac-teristics of eutherian mammals. Evolution 35 (6),1149–1163.

Olsen, S. L. & Shipman, P. (1988). Surface modifica-tion on bone: trampling versus butchery. J. Archaeol.Sci. 15, 535–553.

Parés, J. M. & Pérez-González, A. (1995). Paleo-magnetic age for hominid fossils at Atapuercaarchaeological site, Spain. Science 269, 830–832.

Parés, J. M. & Pérez-González, A. (1999). Magneto-chronology and stratigraphy at Gran Dolina section,Atapuerca (Burgos, Spain). J. hum. Evol. 37, 325–342.

Reverte, J. M. (1981). Antropología Médica I. Madrid:Editorial Rueda.

Rodríguez, J. (1997). Análisis de la estructura de lascomunidadesde mamíferos del Pleistoceno Medio dela Sierra de Atapuerca. Revisión de metodologías.Ph.D. Dissertation. University Autónoma of Madrid.

Russell, M. D. (1987). Mortuary practices at theKrapina Neanderthal site. Am. J. phys. Anthrop. 72,381–397.

Schick, K. D. & Toth, N. (1993). Making Silent StonesSpeak. Human Evolution and the Dawn of Technology.Indiana: Simon and Schuster.

Taylor, P. D. (1986). Scanning electron microscopy ofuncoated fossils. Paleontology 29, 689–690.

Turner, C. G. II (1983). Taphonomic reconstructionof human violence and cannibalism based on massburials in the American Southwest. In (G. LeMoine& S. MacEachern, Eds) Carnivores, Human Scaven-gers and Predators: A Question of Bone Technology.Proceedings of the XV Annual Conference of TheArchaeological Association of the University ofCalgary.

Turner, C. G. II (1988). Another prehistoric Southwestmass human burial suggesting violence and cannibal-ism: Marshview Hamlet, Colorado. In (G. T. Gross& A. E. Kane, compilers) Dolores Archaeological

Program: Aceramic and Late Occupations at Dolores,pp. 81–83. Denver, Colorado: Bureau of Recla-mation.

Turner, C. G. II (1993). Cannibalism in ChacoCanyon: The Charnel Pit excavated in 1926 at SmallHouse Ruin by Frank H. H. Roberts, Jr. Am. J. phys.Anthrop. 91, 421–439.

Turner, P. D. II & Morris, N. T. (1993). A massacre atHoppi. Am Antiq 35, 320–331.

Turner, C. G. II & Turner, J. A. (1990). Perimortemdamage to human skeletal remains from WupatkiNational Monument, northern Arizona. Kiva 55,187–212.

Turner, C. G. II & Turner, J. A. (1992). The firstclaim for cannibalism in the Southwest: WalterHough’s 1901 discovery at Canyon Butte Ruin 3,northeastern Arizona. Am. Antiq. 57 (4), 661–682.

Turner, C. G. II & Turner, J. A. (1995). Cannibalismin the prehistoric American Southwest: occurrence,taphonomy, explanation and suggestions for stand-ardized world definition. Anthropol. Sci. 103, 1–22.

Turner, C. G. II & Turner, J. A. (1999). Man Corn:Cannibalism and Violence in the Prehistoric AmericanSouthwest. Salt Lake City: University of Utah Press.

van der Made, J. (1999). Ungulates from AtapuercaTD6. J. hum. Evol. 37, 389–413.

Villa, P. (1992). Prehistoric cannibalism in Europe.Evolutionary Anthropology 1, 93–104.

Villa, P. & Mahieu, E. (1991). Breakage patterns ofhuman long bones. J. hum. Evol. 21, 27–48.

Villa, P., Helmer, D. & Courtin, J. (1985). Restesosseux et estructures d’habitat en grotte: l’apport desremontages dans la Baume Fontbregoua. Bulletin dela Societe Prehistorique Francaise 82 (10–12), 389–421.

Villa, P., Bouville, C., Courtin, J., Helmer, D., Mahieu,E., Shipman, P., Belluomini, G. & Branca, M.(1986a). Cannibalism in the Neolithic. Science 233,431–437.

Villa, P., Courtin, J., Helmer, D., Shipman, P.,Bouville, C. & Mahieu, E. (1986b). Un cas decannibalisme au Néolithique. Gallia Préhistoire 29,143–171.

White, T. D. (1985). Acheulian man in Ethiopia’smiddle Awash Valley: the implications of cutmarkson the Bodo cranium. Kroom Memorial Lecture, 1985,pp. 1–33. Albert Egges van Giffen Instituut voorPrae-en Protohistorie van the Universiteit vanAmsterdam, Harlem, Netherlands: Enschede enZonen.

White, T. D. (1992). Prehistoric Cannibalism atMancos 5MTUMR-2346. Princeton, NJ: PrincetonUniversity Press.

Yellen, J. E. (1991). Small mammals: !Kung Sanutilization and the production of faunal assemblages.J. Anthrop. Archaeol. 10, 1–26.

Zheng, Yi. (1997). The Scarlet Memorial, Tales ofCannibalism in Modern China. Hong Kong: WestviewPress.