Cannibalism in the Neolithic

Cannibalism in the Neolithic Author(s): Paola Villa, Claude Bouville, Jean Courtin, Daniel Helmer, Eric Mahieu, Pat Shipman, Giorgio Belluomini and Marilí Branca Source: Science, New Series, Vol. 233, No. 4762 (Jul. 25, 1986), pp. 431-437 Published by: American Association for the Advancement of Science Stable URL: http://www.jstor.org/stable/1697806 Accessed: 01-04-2015 18:44 UTC REFERENCES Linked references are available on JSTOR for this article: http://www.jstor.org/stable/1697806?seq=1&cid=pdf-reference#references_tab_contents You may need to log in to JSTOR to access the linked references. Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org. American Association for the Advancement of Science is collaborating with JSTOR to digitize, preserve and extend access to Science. http://www.jstor.org This content downloaded from 132.174.254.159 on Wed, 01 Apr 2015 18:44:30 UTC All use subject to JSTOR Terms and Conditions 19. The actualtime of importanceis the total recyclingtime of the cold atomicgas. This includesthe actuallifetimeof the molecularcloud plus the time it takesa cloud'sgaseousremnantsto reacha low enoughtemperature to allowrecollection by anothershockwave. 20. P. E. Seiden,L. S. Schulman,B. G. Elmegreen,Astrophys. J. 282, 95 (1984). 21. D. H. Rogstadand G. S. Shostak,ibid.176, 315 (1972). 22. M. A. GordonandW. B. Burton,ibid.208, 346 (1976). 23. J. S. YoungandN. Scoville,ibid.258, 467 (1982). 24. P. M. Solomonet al., ibid.266, L103 (1983). 25. Thethinwispy,almostcircular,featuresin thesimulatedgalaxiesaredueto the fact that only purely circularrotations are used in the simulations.The random noncircular componentof velocitiesin a realgalaxysmearsout thesefeatures. 26. P. C. van der Kruitand L. Searle,Astron. 95, 105 (1981). Astrophys. 27. H. Gerola,P. E. Seiden,L. S. Schulman,Astrophys. J. 242, 517 (1980). 28. L. Searle,W. L. Sargent,W. G. Bagnuolo,ibid. 179, 427 (1973). 29. A. CliffandP. Haggett,Sci.Am. 250, 138 (May 1984). 30. A. Sandage,TheHubbleAtlas of Galaxies(CarnegieInstitutionof Washington, Washington,DC, 1961). I Cannibalism in the Neolithic PAOLA VILLA, CLAUDE BOUVILLE, JEAN COURTIN, DANIEL HELMER, ERIC MAHIEU, PAT SHIPMAN, GIORGIO BELLUOMINI, MARILI BRANCA Cannibalism is a provocative interpretation put forth repeatedly for practices at various prehistoric sites, yet it has been so poorly supported by objective evidence that later, more critical reviews almost invariably reject the proposal. The basic data essential to a rigorous assessment of a cannibalism hypothesis include precise contextual information, analysis ofpostcranial and cranial remains of humans and animals, and detailed bone modification studies. Such data are available from the Neolithic levels of the Fontbregoua Cave (southeastern France) where several clusters of human and animal bones have been excavated. The analysis of these bones strongly suggests that humans were butchered, processed, and probably eaten in a manner that closely parallels the treatment of wild and domestic animals at Fontbregoua. ESPITE ABUNDANT LITERATUREON THE SUBJECT [SEE bibliographiesin (1) and (2)], the occurrenceof human cannibalismin Old World prehistoryremainsan open question.We areconcernedherewith dietarycannibalism-theuse of humans by humans as food-evidence for which is found in patternsof bone modificationand discard.The key featuresof dietarycannibalisminvolve close, detailedsimilaritiesin the treatmentof animalandhumanremains.If it is acceptedthatthe animal remainsin questionwere processedas food items, then it can be suggestedby analogythatthe humanremains,subjectedto identical processing,were also eaten. Evidenceof deliberatediscard,cut marks,and bone breakageto extractmarroware criteriaused to deduce that animalbones at archeologicalsites were food refuse;these same criteriahave been used to interpretisolated and scatteredhuman bones at various prehistoricsites as evidenceof cannibalism(3). However,in many casessuch an interpretationis weakenedby doubts about whether humanscausedthe observeddamageand by lackof precisecontextual evidence.Poorly recordedexcavationdata, insufficientdocumentationandanalysisof damageanddiscardpatterns,andthe high frequencyof pre- and postdepositionaldisturbancesby nonhuman agentsat archeologicalsites havefueledthese doubts.Thesearethe main reasons why explanations of cannibalism are often ignored or rejected (4-6). It has been suggested that human bones with cut marks are not the remains of cannibal meals but the traces of funerary rites involving the handling of corpses without consumption of human tissues (2, 7). Secondary burial may mimic cannibalism if it includes active dismemberment and defleshing of the body; however, the absence of bone breakagefor marrow and the mode of bone disposal will set it apart from dietary cannibalism (8). A hypothesis of dietary cannibalism must be based on four types of evidence: (i) Similar butchering techniques in human and animal remains. Thus frequency, location, and type of verified cut marks and chop markson human and animal bones must be similar, but we should allow for anatomical differences between humans and animals; (ii) similar patterns of long bone breakagethat might facilitate marrow extraction; (iii) identical patterns of postprocessing discard of human and animal remains; (iv) evidence of cooking; if present, such evidence should indicate comparable treatment of human and animal remains. We studied recently excavated materials from a Neolithic cave site in southeastern France. A combination of excellent bone preservation, primary depositional context, and fine excavation techniques allows us to present evidence of cannibalism at the site. The Site and Bone Occurrences The Fontbregoua Cave (9) is divided into three spatially discrete areas: the porch, the main room, and the lower room (Fig. 1). All areas have yielded skeletal and cultural materials: pottery, stone tools, remains of domestic and wild faunas, carbonized seeds of domestic wheat and barley, and human remains. Stratigraphic and cultural evidence suggest that during the 5th and 4th millennia B.C. the cave was repeatedly used as a temporary P. Villa,Departmentof Anthropology,Universityof Colorado,Boulder,CO 803090233. C. Bouville,Laboratoire FaculteNord,Universitede Provence, d'Anthropologie, 13326 Marseille,France.J. Courtin,D. Helmer,E. Mahieu,U.R.A. 36, Centrede RecherchesArcheologiques, C.N.R.S., SophiaAntipolis,06565 Valbonne,France.P. Shipman,Departmentof CellBiologyandAnatomy,.Johns HopkinsUniversity,School of Medicine,Baltimore,MD 21205. G. BelluominiandM. Branca,Centrodi Studio e la Geochimica delle Formazioni Geocronologia per Recenti,CNR, UniversitaLa Sapienza,Rome, Italy. 25 JULY1986 ARTICLES This content downloaded from 132.174.254.159 on Wed, 01 Apr 2015 18:44:30 UTC All use subject to JSTOR Terms and Conditions 431 Table 1. Location,age, and relativedepthof features. demarcated. (v) Rodent or carnivore tooth marks, suggesting a nonhuman agent of collection or damage, are not present. Fresh bone surfaces with sharp fracture edges and intact anatomical segments found in six of the features (Table 3) provide further evidence of an undisturbed context. Features Loer Main room room 7 6 8 5 H3, 9 10 Age of level (yearsB.C.)ample Porch 3150 ? 110 3100 ? 120; 2930 ? 110 End of 4th millennium About 3700 3740 + 190; 3740 ? 130 4300 to 3700t Late5th millennium 4750 ? 100 GSY2432 GSY2101, 2433 Clusters of Animal Bones (Features 1-10) Four features contained the remains of severalwild animals, either wild boars (features 1, 9, and 10) or animals of several different H2 H1 species (feature 3). Features 4, 5, 6, and 7 each contained a partial 4 skeleton of a domestic sheep (Ovis aries). All analyzed features are 3 GSY2990 1, 2 judged to have resulted from single episodes of butchering and I i . i . ... . i *Gif-sur-Yvettelaboratory sample number; uncalibrated14C dates on char- discard. coal. tThe age andrelativepositionof the two disturbedclusters,HI andH2, are Three features with animal bones (features 2, 3, and 8) have been approximate. excluded from detailed analysis of the body parts representedin each feature. Feature 8 contained a cluster of sheep bones found under a heavy stone; most bones were very fragmented, some pulverized. residentialcamp (10). In the Early Neolithic, hunting and sheep and Feature 3, which contained the skulls and some shoulder blades of goat herding were of comparable importance, whereas in the Middle six red deer, one roe deer, five marten, two badgers, one fox, and Neolithic hunting played a minor role (11). one wolf, is at the edge of an unexcavated area, at the base of the Preserved habitation features include 13 clusters of bones, which Neolithic sequence; only half of the feature has been uncovered. occur in shallow, probably man-made, hollows of relatively small Feature 2 differs from others because it is not a discard cluster size (20 to 100 cm wide and 8 to 35 cm deep). Ten of these clusters containing several bones. It consists of a circle of stones (diameter, (features 1-10) preserve the butchered remains of wild or domestic 75 cm) surrounding a single left frontal bone and horn of a domestic animals; three clusters [features H1 through H3; (12)] contain only ox with skinning marks above the orbit. This is the only feature that human remains. The location, chronology, and relative depth of may be qualified as "ritual." Cut marks and location data from these features are given in Fig. 1 and Table 1. All clusters are judged features 2, 3, and 8 have been studied. to be intact, with the exception of two of the human clusters (H1 and H2). We verified the integrity of the features according to five criteria. (i) Bone fragments that could be conjoined were found Clusters of Human Bones (Features H1-H3) within each feature; numerous refitting links extend across the depth of feature. Very few links with pieces outside each feature were Feature H3 in the main room is a shallow depression (80 by 40 found (Table 2). (ii) Bones within a cluster can be rearticulatedto cm wide and 15 cm deep) containing 134 fragments of postcranial show they were derived from a single individual. (iii) The vertical bones that lack most of the articularends. These bones are from a and horizontal distribution of bones within each feature is restricted. minimum of six individuals: three adults, two children, and one (iv) Horizontal boundaries of features were sharp and clearly individual of indeterminate age. Also in the feature were eight stone GSY2757, 2756 19 18 17 16 16 15 15 ?14 ^^ ? 13 12 ' 14 ^\ III e 13 H2 .. 12 11 10 9 8 7 6 5 4 3 Fig. 1. Planof FontbregouaCavewith features.Thetwo unnumberedfeatures nearH3 arestoragepits. 2 0! .J A 2 ml . Lower room R S P 1o1iRS 1o1 SCIENCE, VOL. 233 432 This content downloaded from 132.174.254.159 on Wed, 01 Apr 2015 18:44:30 UTC All use subject to JSTOR Terms and Conditions bracelet fragments that conjoin to form two round bracelets. In addition, H3 contained one broken, small polished ax, with a chopping edge of 1.1 cm, which was probably used for butchering the axial skeleton (13). Refitting links combined with vertical plots of elements (Fig. 2 and Table 2) show that feature H3, like the animal bone clusters, represents a single event. The bones of the six individuals were processed and discarded at the same time. Clusters Hi and H2 were disturbed. We define a disturbed cluster as a group of bones that were originally deposited together but were later displaced vertically and horizontally by other agents. The former existence of a cluster is indicated by pieces that can be refitted and by a higher density of pieces within a restricted area, as shown by horizontal and vertical plots of their observed positions (Fig. 3). The Hi cluster in the lower room contains mostly cranial bones (five incomplete crania, isolated fragments of two others, and six mandibles) and 34 postcranial elements. The minimum number of individuals (MNI) is seven, that is, three adults and four children. One H1 bone shows rodent tooth marks and another shows carnivore tooth marks similar to those produced by wolves or dogs. The H1 bones were found in a zone 8 m long and 2.5 m wide, parallel to the cave wall. The maximum vertical distance between pieces that refit is 70 cm; the longest horizontal link is 4.6 m. Four observations suggest that most of the bones deposited in H1 have been recovered in the present excavation. (i) The densest patch Table2. Link frequenciesin features.Genusand minimumnumberof individualsin eachfeaturearein parentheses.Abbreviations:NA, not applicable; NISP, number of identified specimens, after refitting. Features Linkdescriptors HI H3 1 9 10 4 5 6 (6 Homo) 7 (7 Homo) (3 Sus) (4 Sus) (2 Sus) (1 Ovis) (1 Ovis) (1 Ovis) (1 Ovis) 18 116 62.7 20 59 33.9 38 102 32.8 61 154 65.3 36 116 60.4 9 35 36.8 2 16 72.7 13 65 45.1 4 36 46.1 1 3 5.5 240 4 2 2.1 149 1.0 87 0 0 0 57 0 0 0 49 Conjoinedgroups(n) Conjoinedpieces (n) Conjoinedpieces(%)* Outsidelinks (n)t Horizontal Vertical Piecesin outsidelinks(%) NISP NA NA NA 84 0 0 0 134 0 0 0 81 *Numberof conjoinedfragmentsdividedby the totalnumberof bone fragments,excludingunidentifiedsplinters. depositedin the featureor were displaced.Verticallinksindicatedisplacement. 0 0 0 8 tOutsidehorizontallinksarewith piecesthatwerenever Table 3. Percentage of element representation. Abbreviations: MNE, minimum number of elements; CUT (%), number of bones with cut marks divided by the number of identified specimens, excluding teeth; AU, anatomical units found intact in situ; total number of pieces in anatomical units are in parentheses. Features Element Cranium Mandibles Cervicalvertebrae Thoracicvertebrae Lumbarvertebrae Sacrum Caudalvertebrae Clavicle Ribs Scapula Humerus Ulna Radius Pelvis Femur Patella Tibia Fibula Carpals Tarsals Metacarpals H1 (7 Homo) H3 (6 Homo) 100.0 85.7 8.2 0 0 0 05 21.4 0 14.3 21.4 0 14.3 0 21.4 0 14.3 25.0 0 2.0 0 0 0 1.4t 16.7 0? 25.0 6.2 50.0 50.0 37.511 8.0 33.0 0 75.0 25.0 0 0 0 1 (3 Sus) 6 (1 Ovis) 7 (1 Ovis) 0* 100.0 0 0 0 0 43.7 0 0 0 0 0 0 0 00t 100.0 0 76.9 100.0 100.0 14.3 50.0 0 69.2 42.9 0 0 25.0 16.7 66.7 33.3 33.3 0 66.7 0 16.7 33.3 37.5 30.9 0 0 0 62.5 50.0 0 50.0 25.0 37.5 12.5 40.6 3.6 35.7 50.0 50.0 50.0 50.0 25.0 25.0 0 0 0 0 3.6 18.7 57.7 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0? 100.0 70.0 100.0 0 0 0 0 0 50.0 50.0 50.0 50.0 0? 0 50.0 0 53.8 0 50.0 50.0 50.0 100.0 100.0 50.0 100.0 0? 60.0 62.5 0 57.7 0 100.0 100.0 100.0 100.0 50.0 0 50.0 50.0? 50.0 20.0 0 32.3** 0 3.6 1.0 1.2 0 48.6** 45 45.6 0 55 30.3 0 216 15.4 12 (37) arepresent. 5 (1 Ovis) 50.0 50.0 42.9 53.6 50.0 50.0 15.0 0 *Themaxillaeand 4 (1 Ovis) 0 0 0 0 0 0 0 Hand phalanges Foot phalanges premaxillae -pe-ntg u- manr-ad. ot umae and radii. 2ILateral percentage 10 (2 Sus) 100.0 66.7 4.8 38.1 83.3 0 8.3 Metatarsals TotalMNE CUT (%) AU 9 (4 Sus) 144 11.4 3 (6) 6.2 100.0 0 0 0 8.3** 100.0 100.0 0 0 0 0 0 0 76 57.5 6 (15) 81 23.5 6 (25) 8 37.5 2 (5) 57 50.9 0 48 49.0 5 (37) tTotal percentageof cervical,thoracic,andlumbarvertebrae. ?Coccyx. tfTherightioccipital condyleis present. -I percenages r meaapl fo han an fotpaags *Tta pecnae an mettasas. ora Laera malleolus. maueous. #Total percentages for metacarpalsand metatarsals. I Total **Total percentages for hand and foot phalanges. 25 JULY 1986 ARTTCT,F x %-LlrI3 Iq rlxl\ This content downloaded from 132.174.254.159 on Wed, 01 Apr 2015 18:44:30 UTC All use subject to JSTOR Terms and Conditions A2 ltj: of bones is 3 m away from the edge of older excavations, so it is unlikely that any H1 bones were previously excavated. (ii) All bone fragments from every level in the lower room have been thoroughly sorted in search of additional human material. (iii) Although some bones may have been destroyed by mechanical attrition or carnivore damage, visible damage on the preserved human bones is rare (1.5% of the specimens, excluding teeth). (iv) Animal bones in the same deposits also show limited carnivore damage (2%). Finally, the high proportion of refitting links (Table 2) and their spatial pattern suggest that most H1 bones were originally in association. Therefore, we include the HI material in the analysis, except in the study of fracture patterns, since breakage might have occurred during postdepositional disturbances. The H2 pieces were scattered along the cave wall in a strip 1.7 m long and 10 cm wide. The vertical spread was 74 cm. Of 20 fragments, 15 have been conjoined in five groups. No carnivore or rodent marks appear on the bones; of ten bones, three have cut marks. The number of postcranial bones (n = 2) is too small to be informative, and we cannot be sure that all the bones have been recovered. Thus, only cut marks and location data have been considered. Analyses of Fontbregoua (10) suggest that there were initially more discard clusters than found during this excavation; presumably many such clusters were disturbed by various agents, including the inhabitants' own digging activities. It is notable that there are no graves at the site. The mode of burial in Provence for this time period was individual inhumation; however, documentation of this practice is not extensive (14). Location and Mode of Discard Human and animal clusters are found in all parts of the cave; there is no special area reserved for features with the human bones (Fig. 1 and Table 1). This is especially evident in feature H3, which is in the same level as features 9 and 10 and spatially close to them. The absence of animal bone clusters in the porch is not significant because deposits are very disturbed. In size and shape, H3 (80 by 40 by 15 cm) is similar to other clusters, especially feature 1 (70 by 40 by 7 cm), which contained the partial skeletons of three wild boars. Table 3 shows the frequencies of body parts present in each cluster. Each figure is obtained by dividing the observed minimum number of a skeletal element by the expected number of the same element, based on the MNI; the ratio is expressed as a percentage (15). This statistic is often used to express patterns of differential survival. Here, since these clusters (with the exception of HI) are intact and have undergone no postdepositional destruction, the percentage of element representation reflects discard patterns. Data in Table 3 suggest the following observations: (i) In all clusters animals or humans are represented by selected anatomical parts; other parts are missing or are present in lower than expected frequencies. For example, in feature 4 the left foreleg and the right hindleg are missing, yet all four limb extremities (metapodials and phalanges) are present and intact. Crania and limb extremities are missing from features 5, 6, 7, and H3 (16). Feature 9 contains the manus and pes of four wild boars plus some leg bones; all other body parts are missing. In H3 only six scapulae and six humeri are present out of the 12 expected for each; in feature 1 there are only four humeri out of the six expected. (ii) Sometimes only a small portion of an anatomical segment is present. Thus in feature 6 the braincase is missing, but the muzzle bones are present; the right foreleg from scapula to phalanges is missing, but the right carpals are present. In feature 7 the cranium and the neck are missing, but 0 20 I 500 - 40 I 60 I I o 510a) E C a) 520- 520 , 0 I CCU - Bones o Fragments ?* Stone ax of bracelets Fig. 2. FeatureH3: verticalprojectionsand refittinglinks (short links omitted). the right occipital condyle is present. In H3 the sacrum and pelvis are represented by only small fragments of one element. In features 1 and 4 sacra are absent, but caudal vertebrae are present. The pattern that emerges from all human and animal clusters shows discard of selectively butchered parts. Two facts are intriguing. First, missing anatomical segments are represented by isolated elements or scraps of little food value, for example, carpals,occipital condyle, minute bits of sacrum, or, as in feature 4, intact lower leg parts. Second, these isolated elements are near or at points of disjointing and segmentation. We conclude that the missing anatomical segments have been culled from essentially complete carcasses at the cave itself. After disarticulation, selected body parts were set aside for separate processing and consumption; thus they are missing from the features. If segmentation had taken place outside the cave, it is unlikely that scraps from the culled units would have been collected and transported inside the cave for the purpose of discarding them. Two observations support this view of butchering in the cave. (i) Sheep were penned at the site (17); thus we infer that they were killed and butchered at the site. (ii) All types of bones from wild boar and human skeletons are present in the cave bone assemblage, including parts of low utility such as heads, necks, caudal vertebrae, and phalanges (10). It is possible that filleting (defleshing) and marrow fracturing were done on a skin; the residue was then discarded in a single pile. The use of a skin would explain why the bones are tightly packed in well-defined clusters and why so many fragments can be conjoined. It would also explain the presence in each cluster of many small unidentifiable splinters resulting from the operations of marrow fracturing (18) and the presence of bits of culled units. In sum, it is clear that human and animal carcasseswere processed and discarded according to the same pattern of selective butchering (19). Segmentation and selection of parts for differential use or distribution are normally practiced when butchering animals (20); their occurrence in the processing of human carcassesis significant. Although domestic sheep were butchered one at a time, wild animals were captured and butchered in groups. Interestingly, two of the three clusters of human bones correspond to the wild animal pattern of butchering. Cut Marks Cut marks on human bones have been compared to marks on homologous animal bones. Of 223 bones bearing 246 cut marks (21), we verified a sample of 27 bones with 31 cut marks by scanning electron microscope (SEM) studies, using procedures described in (22). The verified sample includes 29% of the observed cut marks on the human bones (25 of 85) and 4% of the marks on SCIENCE, VOL. 233 434 This content downloaded from 132.174.254.159 on Wed, 01 Apr 2015 18:44:30 UTC All use subject to JSTOR Terms and Conditions Fig. 3. DisturbedfeatureHI: verticalprojections and refittinglinks (some peripherallinks omitted). InsetshowscaveareawhereHI boneswere found. The A-B line is the axis of maximum dispersal. A B 500 - 550 - a 600 - 0 D () E L c 0 * Skull fragments o Postcranial bones 650 - the animal bones (6 of 161). All putative cut marks replicated for microscopic study (23) were confirmed as cut marks. Nearly all types of human bones bearing observed cut marks have at least one verified cut mark. The fine-grained substratum (24) and undisturbed context of the features and the placement and patterns of cut marks (25) are further evidence that these are purposive toolmarks and are not due to carnivores or trampling (26). The interpretation of activity is based on descriptions by Binford (5) and on our experimental butchering of a sheep and a goat with flint blades and a stone ax. All cut marks, regardless of the taxonomic identity of the bones, show features suggesting that they were made shortly after death (immediate processing), rather than a year or more after death (delayed processing). This assessment of the timing of processing is based on SEM comparisons of the Fontbregoua material and experimentally altered bones (27). Immediate processing is consistent with an interpretation that both animals and humans were processed for use as food. There are strong similarities in frequencies of marked bones and types of cut marks (Tables 3 and 4). Especially significant is the abundance of filleting marks on both human and animal bones, indicating that meat was routinely removed from the bones. Frequencies of filleting versus dismembering marks on long bones are: 80.0% in H3, 70.6% in feature 1, 75.0% in 4 and 6, 54.5% in 7, 80.0% in 9, and 75.0% in 10 (28). Meat may have been filleted from still-articulated units, as is suggested by some of the anatomical segments found intact in situ. These units include: distal tibia, tarsals or lateral malleolus or both (features 4, 5, and 7); distal radius, ulna, and sometimes carpals (features 7 and 10); distal femur and proximal tibia (feature 4); tarsals (features 1 and 5); vertebral segments (features 1, 4, 7, and 10); and phalanges or metapodials or both (features 1, 4, and 9). Articulated units were not observed in features 6 and H3, which contained only highly fragmented bones. With respect to cut mark location and morphology, a remarkable degree of concordance can be observed between animal and human bones. Of 33 cut mark varieties on human cranial and postcranial bones, 23 can be matched with similar markson homologous animal bones (25). Differences in cut mark location between animal and human remains are important for two elements, the scapula and the cranium. The greater variety of dismembering cut marks on human scapulae (Table 4) can be attributed to the greater complexity of the shoulder joint in humans, who possess a clavicle, unlike suids and ruminants. Although the treatment of human crania closely parallels that of animal craniawith respect to sagittal skinning marks (29), the human material bears cut marks in locations that are undamaged on animal bones. Thus, for example, human craniashow cut marksnear the insertion of the sternocleidomastoideus muscle on the mastoid process, on the vault bones in areas normally covered by the temporalis muscle, and on the facial bones overlaid by musculature. These marks are interpreted as defleshing marks. In contrast, the only defleshing marks observed on animal skulls in the features, and in a larger sample from the Early Neolithic deposits, are associated with removal of the tongue. These cut marks occur on the hyoid and on the internal face of the mandibular corpus. It is possible that human crania were more extensively defleshed because they were kept as trophies or ritual objects, as is documented in later periods in the same region (30). However, in all other ways the frequencies and types of marks on the Fontbregoua bones are consistent with a conclusion that human and animal carcasses were treated similarly. Marrow Fracturing All marrow bones in the features and all bones in the H3 cluster are broken, each in several fragments. Although some damage is Table 4. Frequenciesof bones with cut marksfrom all features.Only homologousbones presentin both animaland humansamplesare listed. Small sampleswith combined N less than 20 (radii, vertebrae)are not included.Abbreviations:N, numberof specimensafterrefitting;for crania we have used the MNE to avoid problemsrelatedto the high degree of CUT, percentagesof bones with cut marks;F, Sk, and D, fragmentation; percentagesof bones with filleting, skinning, or dismemberingmarks, one bone may havetwo types of marks. respectively; Bone sBoe sample N CUT F Sk D Postcranial Human Animal 12 20 Human Animal 13 29 Human Animal 25 13 Human Animal 16 9 Human Animal Cranial 38 88 Human Animal 9 14 Human Animal 8 13 Humerus 41.7 41.7 40.0 40.0 Femur 38.5 23.1 41.4 27.6 Tibiaandfibula 32.8 28.0 38.5 38.5 Scapula 50.0 18.7 44.4 22.2 Ribs 26.3* 21.0 59.1 32.9 Mandibles 88.9 78.6 Cranium 100.0 84.6 0 10.0 38.5 24.1 4.0 7.6 50.0 22.2 13.2* 36.4 66.7 57.1 55.6 54.5 75.0 76.9 25.0 30.8 differentfrom the correspondingvaluein the animalgroup (X2test on *Significantly rawfrequencies, P < 0.05). The low frequencyof cutmarksand,morespecifically, of D markson human ribs is due to a scarcityof proximalfragments.No significant differences arefoundin othergroups,accordingto X2or Fisher'sexactprobabilitytests. bones;possibledefleshingmarks Skinningmarksarenot foundon the listedpostcranial on skullsarediscussedin the text. 25 JULY 1986 ARTICLES This content downloaded from 132.174.254.159 on Wed, 01 Apr 2015 18:44:30 UTC All use subject to JSTOR Terms and Conditions 435 likely due to postdepositional alteration and sediment pressure, the high degree of fragmentation of the long bones is primarilyattributed to deliberate breakage for marrow. In H3 most of the long bone fragments (88 of 107) are thin, elongate shaft splinters, and many can be identified only by refitting them into larger pieces. Their mean length is 9.4 cm with a range of 2 to 28 cm (and a standard deviation of 4.5). Some attributes indicate fresh bone breakage: fracture edges are smooth in 73% of the cases; 71% have acute or obtuse angles (31). Perhaps the most significant criteria of dynamic loading (by a blow) are wide impact scars with radiating fissure lines. They are present in 20.7% of the long bone fragments in H3; half of these are characterizedby broad, thin spalls still attached to the bone, with platforms bounded by arcuatefissure lines behind the point of impact (5, 32). The frequency of impact scars on human long bones compares well with values observed on animal bones from features 1, 4, 7, and 10 (23.4, 12.5, 13.3, and 15.0%, respectively) and with ethnographic observations (5). Nevertheless, neither fracture morphology nor impact scars with splintered margins are exclusively associated with human marrow fracturing, and they may be produced by carnivores (31, 33). The absence of gnaw marks on the bones from all features (with the exception of H1) will not hold as a valid argument against carnivore damage since the analyst's ability to identify such marks may be doubted. Evidence against carnivore damage and for the human origin of bone breakageis provided, instead, by the repetitive spatial patterns of the bone clusters: sharp horizontal boundaries and localized densities of homogeneous items, abundant refitting links within each feature, and few or no outside links. These patterns provide evidence that the clusters are intact and man-made. Evidence of Cooking Two indicators of cooking that might be found on archeological bones are changes in collagen chromatographs (34) and changes in the microscopic morphology of bone surfaces (35). Both were absent from the Fontbregoua bones. Amino acid analyses of bone collagen in nine samples from Hi, H3, feature 1, and the main room deposits show that these bones were not exposed to temperatures greater than 150?C; SEM inspection of various bone samples did not reveal changes in microscopic morphology known to occur at 185?C. However, temperatures achieved by meat-covered bones during boiling or roasting are lower than these thresholds, as experimental studies confirm (35, 36). Additional evidence that casts doubt on the idea of cooking is provided by the abundant filleting marks and intact anatomical units, both features that one would not expect to find in roasted or boiled remains. Clearly, there is no good evidence showing that cooking of meat-on-bone occurred. However, the treatment of animal and human remains does not differ in this regard; in both cases uncooked bones were discarded after filleting and marrow fracturing. Conclusions Our inference that animal and human meat was eaten is based on the evidence of ordinary butchering practices and unceremonial patterns of discard in a domestic setting. Similarities in the treatment of animal and human remains are striking. The evidence of breakage to extract marrow and the mode of discard contrast strongly with known secondary burial practices (8). Elements of rituals seem to be present in the treatment of human skulls, but they are consistent with an interpretation of exocannibalism. Feature 2 suggests that Bos skulls could also be an object of special consideration. We believe that cannibalism is the only satisfactory explanation for the evidence found at Fontbregoua Cave. Taphonomic studies of human bones at additional Stone Age French sites should help to establish whether our findings represent isolated events or institutionalized practices (37). REFERENCES AND NOTES 1. M. K. Roper, Curr. Anthropol. 10, 427 (1969). 2. F. Le Mort, thesis, Universite de Paris VI (1981). 3. F. Weidenreich, The Skull of Sinanthropus pekinensis (Paleontologia Sinica, New Series D, no. 10, Peking, 1943), pp. 184-190; H. V. Vallois, La Grotte de Fontechevade,deuxiemepartie, anthropologie(Archives Institut Paleontologie Humaine 29, Paris, 1958), pp. 17-84; H. de Lumley et al., in La Grottemoustrienne de l'Hortus,H. de Lumley, Ed. (Etudes Quaternaires 1, Universite d'Aix-Marseille I, 1972), pp. 527-623. 4. W. Arens, The Man-Eating Myth (Oxford Univ. Press, New York, 1979). 5. L. R. Binford, Bones:Ancient Men and ModernMyths (Academic Press, New York, 1981). 6. E. Trinkaus,J. Hum. Evol. 14, 203 (1985). 7. E. Cartailhac,La Franceprehistoriqued'apreslessepultureset lesmonumentshistoriques (Alcan, Paris, 1889), pp. 91-121; K. Branigan,Nature (London)299, 201 (1982). 8. D. H. Ubelaker, Reconstructionof DemographicProfilesfrom OssuarySkeletalSamples (Smithsonian Contribution to Anthropology, Washington, DC, 1974); W. M. Bass and T. W. Phenice, in The Sonota Complexand AssociatedSites on the Northern GreatPlains, R. W. Neumann, Ed. (Nebraska State Historical Society Publication in Anthropology, no. 6, Lincoln, 1975); D. Ferembach and M. Lechevallier, Palsorient 1, 223 (1973). 9. J. Courtin, Sites neolithiqueset protohistoriques de la region de Nice (Livret-Guide de l'Excursion B2, 9th International Union of Prehistoric and Protohistoric Sciences Congress, Nice, 1976), pp. 21-27; P. Villa and J. Courtin,J. Archaeol.Sci. 10, 267 (1983). 10. P. Villa, D. Helmer, J. Courtin, Bull. Soc. Prehist. Franf., in press. 11. D. Helmer, thesis, Universite de Montpellier II (1979). 12. The letter H stands for human. 13. Fontbregoua's Neolithic levels have yielded 15 polished axes with edge widths of 1.1 to 4.8 cm. We believe one of the uses of these small axes was butchering. Chop marks, made with an ax, are present on a human rib and a vertebral fragment in feature H3; others are found on wild boar vertebrae in features 1 and 10. Experimental butchering with a 2.5-cm-wide polished stone blade set in a wooden handle has produced similar marks on vertebrae and pelvis of a sheep and goat. 14. J. Courtin, in La PrehistoireFranfaise,J. Guilaine, Ed. (Editions du Centre National de la Recherche Scientifique, Paris, 1976), vol. 2, p. 259; J. Courtin, Gallia Prehistoire25, 536 (1982). 15. For example, in feature H1 there is a minimum number of three humeri; the expected number of humeri is 14 since the minimum number of individuals is seven. The percentage of representation is (3/14) x 100 = 21.4. See C. K. Brain, in Human Origins, G. L. Isaac and E. R. McCown, Eds. (Benjamin, Menlo Park, , The Hunters or the Hunted? (Univ. of Chicago CA, 1976), pp. 97-116; Press, Chicago, 1981), p. 21; D. P. Gifford-Gonzalez, in Proceedingsof the First International Conferenceon BoneModification, R. Bonnichsen, Ed. (Center for the Study of Early Man, Orono, ME, in press). 16. The H1 cluster may contain the skulls that are missing from the H3 cluster; however, we can neither prove nor refute this idea. No refitting links have been found between the two clusters; the postcranial bones are too fragmented to be matched for size and age with some degree of confidence. The two clusters are in deposits of broadly equivalent age, but the gap left by the older excavations forbids any assessment of stratigraphic continuity between the two areas. 17. The Fontbregoua's deposits contain two diagnostic traces of cave herding. The first is abnormally high frequencies of ovicaprine milk teeth with maximum degree of wear and totally resorbed roots. These teeth were lost naturally, and their abundance suggests that the animals were kept in pens inside the cave [D. Helmer, in Animals and Archaeology, J. Clutton-Brock and C. Grigson, Ed. (British Archaeological Reports, International Series 204, London, 1984), vol. 3, pp. 3945]. The second is large quantities of calcite spherulites, representing the mineral residue of ovicaprine dung. Similar traces are found in other Neolithic caves [J. Brochier, Bull. Soc. Prehist. Franf. 80, 143 (1983)]. 18. For example, feature H3 contained 154 indeterminate bone fragments >2 cm and 133 g of small bone chips recovered through water sieving; feature 10 had 61 indeterminate bone fragments >2 cm and 470 g of smaller ones. 19. Selective processing of different body parts may be due to patterns of delayed consumption or to sharing with other members of the group who were not living at the cave. 20. L. R. Binford, Nunamiut Ethnoarchaeology(Academic Press, New York, 1978). 21. Several marks at the same location are counted as one. 22. P. Shipman, in The ResearchPotential ofAnthropologicalMuseum Collections,A. M. Cantwell, J. B. Griffin, N. Rotschild, Eds.,Ann. N.Y.Acad. Sci. 376, 357 (1981); P. Shipman and J. Rose,J. Anthropol.Archaeol.2, 57 (1983); J. Rose, Am.J. Phys. Anthropol.62, 255 (1983). 23. Replicas of marked surfaces are used to avoid transporting of and damage to the originals. 24. Predominantly fine sand and silt; J. Brochier, in preparation. 25. Photos, drawings, and lists of cut marks are provided in (10) and in P. Villa et al., Gallia Prihistoire,in press. SCIENCE, VOL. 233 436 This content downloaded from 132.174.254.159 on Wed, 01 Apr 2015 18:44:30 UTC All use subject to JSTOR Terms and Conditions 25. Photos,drawings,andlistsof cut marksareprovidedin (10) andin P. Villaet al., in press. GalliaPrehistoire, K. D. Gordon,G. T. Yanagi,Nature (London)319, 768 26. A. K. Behrensmeyer, (1986). 27. M. D. Russell,P. Shipman,P. Villa,Am.J. Phys.Anthropol. 66, 223 (1985). markson long bonesfromfeaturesH3, 28. Totalcountsof filletinganddismembering 1, 4, 6, 7, 9, and 10 are: 35, 17, 4, 8, 11, 10, and 8, respectively.See (21) for countingprocedures. 29. Five humanand sevenanimalcraniahavelong sagittalmarksalong the midline, frontalto occipital. andliteraryevidenceindicatesthatCeltictribeslivingin Provenceat 30. Archeological the endof the firstmillenniumB.C.keptskulltrophiesin theirshrinesandhouses. Ed. Excavations in Europe,R. Bruce-Mitford, F. Benoit, in RecentArchaeological (Routledgeand Kegan, London, 1975), pp. 227-259; B. Cunliffe,The Celtic New York,1979), pp. 82-83. World(McGraw-Hill, 31. G. Haynes,Am. Antiq.48, 112 (1983). 32. H. Martin,Bull.Soc.Prehist.Frann.7, 299 (1910); H. T. Bunn,Nature(London) 10, 338 (1984), figure4h. 291, 576 (1981); C. Fisher,Paleobiology R. B. Potts, thesis,HarvardUniversity(1982). Romana19, 171 (1980). G. Belluominiand P. Bacchin,Geologia P. Shipman,G. Foster,M. Schoeninger, J. Archaeol.Sci. 11, 323 (1984). Aminoacidanalysesof two modernsamples(a sheeppelvisboiledfor4 hoursanda sheephumerusfroma shoulderroastcookeduntilwell-doneon an open firefor 1 identicalto thoseof moder unheated hourand15 minutes)showchromatographs achievedby meat bones and to those of the archeologicalbones. Temperatures than are less 100?C [J. Child,L. Bertholle,S. Beck,Masteringthe duringroasting Art ofFrenchCooking(Knopf,New York,1968), p. 379]. 37. Bone fragmentsfrom featureH3 have been'dated by the Lyon laboratoryto 14Cdateon bone; Ly 3748). 3930 + 130 B.C. '(uncalibrated 38. Supportedby grantsfromthe WennerGrenFoundation,the AmericanCouncilof LearnedSocieties,and the LeakeyFoundationto P.V. The Fontbregouaexcavations arefundedby FrenchMinistryof Culturegrantsto J. C. 33. 34. 35. 36. ?_~ of the H-2 Biology Molecular Complex Histocompatibility RICHARD A. FLAVELL,HAMISH ALLEN, LINDA C. BURiLY, DAVID H. SHERMAN, GERALD L. WANECK, GEORG WIDERA The H-2 histocompatibility complex of the mouse is a multigene family, some members of which are essential for the immune response to foreign antigens. The structure and organization of these genes have been established by molecular cloning, and their regulation and function is being defined by expression of the cloned genes. HE MAJOR HISTOCOMPATIBILITYCOMPLEX (MHC) OF mammals is a multigene family whose members encode cell surface glycoproteins involved in the recognition and immune response to foreign antigens. The MHC has been conserved throughout vertebrate evolution, and the MHC's of mouse (H-2) and human (HLA) have been studied extensively. The H-2 complex, located on mouse chromosome 17, has been divided into class I and class II genes on the basis of structuraland functional similarities (1- 5). The class I genes are located at four genetic loci defined by serologic analyses of recombinant inbred mice: H-2K, H-2D/H-2L, Qa-2,3, and Tla (Fig. 1). These genes encode heavy chains of a molecular size of approximately 45,000 (45 kD) that are noncovalently associated as heterodimers with a 32-microglobulin(32m), a 12-kD polypeptide encoded by a gene on mouse chromosome 2 (6). The 45-kD polypeptide has three extracellulardomains (here called al, ta2, and a3) anchored in the membrane by a short transmembrane segment, and a cytoplasmic peptide of some 35 amino acids (Fig. 2a). The K, D, and L molecules are highly polymorphic (7), are expressed on the surface of virtually all cells, and appearto direct the recognition of virus-infected and neoplastic cells by cytotoxic T lymphocytes (CTL) (8, 9). The antigen-specific receptors of CTL recognize viral glycoproteins only when they are associated with these class I molecules on the cell surface. In contrast, products of the Qa-2,3 region (Qa-2,3) and the Tla region (TL) are less polymorphicand their expressionis limitedto certaintissues (1013). The Qa-2,3 and TL moleculesare not involvedin associative recognitionby CTL, and their functionis unknown. The classII genesarelocatedat two geneticloci (I-A andI-E) that map between H-2K and H-2D/H-2L (Fig. 1). The I-A region containsthe Ap, AW,and Ep genes and the I-E region containsthe Engene.Thesegenesencodeheterodimers(Ia molecules)consisting of a 35-kDr( chainnoncovalentlyassociatedwith a 29-kD p chain domains,a (14). Both ac and 3 chainsconsist of two extracellular and a transmembrane segment, cytoplasmicregion (Fig. 2a). The Ia moleculesarehighlypolymorphicandareexpressedprimarilyon the surfaceof B lymphocytes,macrophages,dendriticcells, and certain epithelialcells.The antigen-specificreceptorsof helperT cells that arerequiredfor the generationof CTLandfor antibodyproduction by B cellsrecognizeforeignantigenonly when it is associatedwith Ia molecules(15, 16). The domain organizationof class I and class II molecules is reflectedby the exon-intronorganizationof the corresponding genes. The 0t3 domain of class I molecules and the (x2 and 32 domainsof class II moleculeshave strong sequencehomology to domainsof immunoglobulin-constant regions and thus belong to the immunoglobulinsupergenefamily(17). Organization of Class I Genes The organizationof class I genes of the BALB/c (H-2d) and C57BL/10,or B10 (H-2b), haplotypesis known in detail,and the R. A. Flavellis principalresearchofficerof the BiogenGroupandpresidentof Biogen ResearchCorporation,Cambridge,MA 02142. H. Allen, L. C. Burkly,and G. L. Waneckarescientistsat BiogenResearchCorporation,Cambridge,MA 02142. D. H. Shermanis a postdoctoralfellow at the Centerfor CancerResearch,Massachusetts Instituteof Technology,Cambridge,MA 02139. G. Widerais an assistantmemberat the ResearchInstituteof the ScrippsClinic,La Jolla,CA 92037. ARTICLES 437 25 JULY 1986 This content downloaded from 132.174.254.159 on Wed, 01 Apr 2015 18:44:30 UTC All use subject to JSTOR Terms and Conditions