an early case of color symbolism: ochre use by modern

33
An Early Case of Color Symbolism: Ochre Use by Modern Humans in Qafzeh Cave Citation Hovers, Erella, Shimon Ilani, Ofer Bar#Yosef, and Bernard Vandermeersch. 2003. “An Early Case of Color Symbolism: Ochre Use by Modern Humans in Qafzeh Cave.” Current Anthropology 44 (4) (August): 491–522. Published Version doi:10.1086/375869 Permanent link http://nrs.harvard.edu/urn-3:HUL.InstRepos:12211574 Terms of Use This article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Other Posted Material, as set forth at http:// nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of-use#LAA Share Your Story The Harvard community has made this article openly available. Please share how this access benefits you. Submit a story . Accessibility

Upload: others

Post on 28-Apr-2022

5 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: An Early Case of Color Symbolism: Ochre Use by Modern

An Early Case of Color Symbolism: Ochre Use by Modern Humans in Qafzeh Cave

CitationHovers, Erella, Shimon Ilani, Ofer Bar#Yosef, and Bernard Vandermeersch. 2003. “An Early Case of Color Symbolism: Ochre Use by Modern Humans in Qafzeh Cave.” Current Anthropology 44 (4) (August): 491–522.

Published Versiondoi:10.1086/375869

Permanent linkhttp://nrs.harvard.edu/urn-3:HUL.InstRepos:12211574

Terms of UseThis article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Other Posted Material, as set forth at http://nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of-use#LAA

Share Your StoryThe Harvard community has made this article openly available.Please share how this access benefits you. Submit a story .

Accessibility

Page 2: An Early Case of Color Symbolism: Ochre Use by Modern

491

C u r r e n t A n t h ro p o l o g y Volume 44, Number 4, August–October 2003� 2003 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved 0011-3204/2003/4404-0003$3.50

An Early Case ofColor Symbolism

Ochre Use by Modern Humansin Qafzeh Cave1

by Erella Hovers, Shimon Ilani,Ofer Bar-Yosef, andBernard Vandermeersch

Prehistoric archaeology provides the temporal depth necessaryfor understanding the evolution of the unique human ability toconstruct and use complex symbol systems. The long-standingfocus on language, a symbol system that does not leave directevidence in the material record, has led to interpretations basedon material proxies of this abstract behavior. The ambiguities re-sulting from this situation may be reduced by focusing on sys-tems that use material objects as the carriers of their symboliccontents, such as color symbolism. Given the universality ofsome aspects of color symbolism in extant human societies, thisarticle focuses on the 92,000-year-old ochre record from QafzehCave terrace to examine whether the human capacity for sym-bolic behavior could have led to normative systems of symbolicculture as early as Middle Paleolithic times. Geochemical andpetrographic analyses are used to test the hypothesis that ochrewas selected and mined specifically for its color. Ochre is foundto occur through time in association with other finds unrelatedto mundane tasks. It is suggested that such associations fulfillthe hierarchical relationships that are the essence of a symbolicreferential framework and are consistent with the existence ofsymbolic culture. The implications of these findings for under-standing the evolution of symbolic culture in the contexts of theAfrican and Levantine prehistoric records are explored.

e r e l l a h o v e r s is a lecturer in the Institute of Archaeology,The Hebrew University of Jerusalem (Mt. Scopus, Jerusalem91905, Israel [[email protected]]). Her fields of interest in-clude Middle Paleolithic lithic technology and human ecology,

1. Excavations of the Mousterian deposits at Qafzeh Cave in theyears 1965–73 and 1977–79 were sponsored by the French Ministryof Foreign Affairs. The petrographic, chemical, and mineralogicalanalyses reported in this study were carried out in the geochemicallaboratories of the Geological Survey of Israel with the help of Y.Deutsch and O. Yoffee. Geological fieldwork was conducted withthe help of E. Zilberman. The geological map of the Qafzeh area(in the electronic edition of this issue) and figure 9 were drawn byN. Sharagai and S. Hoyland. J. Skidel-Rhymer made the drawingsof lithic artifacts, and the photographs were taken by G. Laron atthe Institute of Archaeology. We thank A. Belfer-Cohen and J. Speth,as well as six diligent reviewers, for their comments and usefulsuggestions on previous drafts of this paper. [Supplementary ma-terials appear in the electronic edition of this issue on the journal’sweb page (http://www.journals.uchicago/edu/CA/home.html).]

the evolution of symbolic behavior and of human cognition andconsciousness, and the archaeology of the Late Pliocene andearly Pleistocene.

s h i m o n i l a n i is a researcher in the Geological Survey of Is-rael. His main interests are petrography, mineralogy, andgeochemistry.

o f e r b a r - y o s e f is a professor in the Department of Anthro-pology at Harvard University. He is interested in the cultural se-quence of the Middle Paleolithic, the demise of the Neandertals,the emergence of Upper Paleolithic entities, and the Neolithicrevolution in the Near East.

b e r n a r d v a n d e r m e e r s c h , now retired, was a professor ofphysical anthropology at the University of Bordeaux. He is inter-ested in human evolution, particularly in the issues of Neander-tals and modern humans and their cultural and environmentalcontext.

The present paper was submitted 1 ix 02 and accepted 9 x 02.

For many researchers the ability to create arbitrary re-lationships between ideas and their referents—that is, toconstruct and use complex symbolic systems—is the de-fining characteristic of Homo sapiens. Biologists, cog-nitive scientists, and philosophers address questionsabout the origins and functions of these higher cognitiveabilities. In this context, it is prehistoric archaeology thatprovides information about the time frames and tem-poral depth involved in these evolutionary processes.Thus archaeologists are constantly searching for earlymanifestations of symbolic behavior in the prehistoricrecord.

Language, considered the most complex symbol sys-tem used by H. sapiens, is often the focus of discussionin the various disciplines which aim to understand hu-man symbolism. Unfortunately, the prehistoric record ofnonliterate societies by definition cannot contain directevidence for the existence of language, an abstract entity.In the absence of such evidence, archaeologists constructever-broadening tiers of interpretation, attempting torecognize material proxies for language in the archaeo-logical record and to infer from them the cognitive fac-ulties that underlie them. Indeed, many of the debatesin the recent archaeological literature on the evolutionof modern behavior and symbolism stem from disagree-ments among researchers about such inferences.

There is a solution to this conundrum. Ethnographicevidence suggests that in many societies there exist sym-bolic networks in which material objects are the sym-bols. Among these, symbolic color systems are wide-spread and shared by many societies. If similar systemsexisted in the prehistoric past, it would make sense forarchaeologists to identify and study them directlythrough their preserved material manifestations ratherthan speculating about the meanings of material objectsfor understanding the evolution of language.

A rich prehistoric record of pigment use indicates thatred and black pigments are relatively ubiquitous in Pa-leolithic habitation and quarry sites, from the Plio/Pleis-tocene to Upper Paleolithic (Late Stone Age) times (Bar-ham 1998; Bednarik 1992b; Bordes 1952; d’Errico andSoressi 2002; Henshilwood et al. 2001, 2002; Marshack

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 3: An Early Case of Color Symbolism: Ochre Use by Modern

492 F current anthropology Volume 44, Number 4, August–October 2003

1981, 1989; McBrearty and Brooks 2000; Solecki 1982;Watts 1999; Wreschner 1983). There is an increase overtime in the frequency of pigment occurrence in the pre-historic record, but although it occurs at different timesin different regions, in general it can be placed in thecontext of the Middle Stone Age/Middle Paleolithic inAfrica and in Europe, respectively. It is therefore withinthis time period that we look for indications of the useof pigments in a symbolic rather than a utilitarian con-text. In this paper we examine the ochre finds from theMiddle Paleolithic deposits of Qafzeh Cave, Israel, andevaluate the possibility that their occurrence was relatedto the operation of a symbolic cultural system.

Humans today routinely engage in symbolic behavior.It is often a premeditated activity with symbols at itscore (e.g., the playing of national anthems at formal re-ceptions for heads of states). In more mundane contexts,encoded information about socioeconomic status maybe emitted purposefully by one’s fashion statements orunconsciously by one’s table manners, whereas theposter of a rock star symbolizes her to people who havenever met her. And yet, although symbolic behaviortakes place all around us, understanding it is not simplebecause symbols themselves are not simple (Deacon1997). In the examples just mentioned symbolic behavioroccurs in a number of referential frameworks. To thedegree that symbol systems reflect the logic of thoughtprocesses in the modern human mind (Peirce 1897 and1903, as discussed by Deacon 1997), they incorporatethree fundamental forms of referential associations: (1)Icons point to their referents by resembling them (as inthe case of the rock star poster). (2) Indices are mediatedby some redundant physical or temporal association be-tween sign and object (the tag of an expensive designeron one’s clothes is indexical of the amount of moneyone can afford to invest in clothing). (3) Symbols aremediated by arbitrary, formal, or agreed-upon links, ir-respective of the physical characteristics of either thesignifier or the signified (see also Chase 1991), as is thecase with flags, national anthems, and team colors.

The following discussion revolves around symbols, themost complex of these referential associations, but it isimportant to recognize that they invariably rest on afoundation of icons and indices. Iconic reference is thedefault, basic, and irreducible referential form. At theother end of the scale, symbolic reference is based onthe recognition that the relationship of a sign to an objectis more than just a function of their co-occurrence (whichwould count only as an indexical relationship). Symbolsrefer to things in the world indirectly and by virtue ofreferring to other symbols. We recognize that there is anindexical relationship between the tag of an expensivedesigner and the amount of money invested in buyingclothes made by him, and we make the additional ref-erence that this relationship is itself indexical of socio-logical status and economic ability; the tag becomes asymbol of socioeconomic success. Symbolic referencestems from combinatorial possibilities and impossibil-ities. The referential powers of symbols are derived fromtheir positions in an organized system of other symbols.

These allow the recognition of higher-order regularities,which in turn enable symbolic predictions. This processfacilitates the construction (learning) of symbols andtheir deconstruction and use (interpreting them and thuscommunicating through them).

Of all the symbolic behaviors in which H. sapiens en-gages, language is judged the most complex and consid-ered a unique trait of the species (e.g., Deacon 1997).Because it is an adaptation that confers enormous ad-vantages on its bearers, its antiquity and the paleobiol-ogical and archaeological evidence that may indicate itsexistence have been the focus of intense research. Theorigins of language have been placed as early as 2 millionyears ago (Deacon 1989, 1990, 1997) and as late as theLate Pleistocene (e.g., Klein 1995, 2000; Lieberman 1991;Mellars 1996; Mithen 1996; Noble and Davidson 1996).

Language is the most powerful but not the only symbolsystem known to us. This observation is pertinent toarchaeologists’ efforts to recognize symbolism. Too oftendiscussion of the origins and antiquity of language in thearchaeological literature has been conflated with dis-cussion of the symbolic meanings of objects. Much ofthe debate about language has been propelled over theyears by the obvious fact that language leaves no materialremains and its existence in Paleolithic cultures has tobe inferred from material manifestations which may beindexical of it. Art, decorations and ornaments, and in-tentional burials are among the consensual although notunanimously agreed-upon (e.g., Humphrey 1998, Nobleand Davidson 1996) such proxies in later prehistory. Theethnographic record indicates, however, that some sym-bol systems use material elements (tokens) to commu-nicate complex social and cosmological messages. It issuch symbol systems that should be taken as the appro-priate models for the investigation of prehistoric sym-bolic behavior. Because much of the material lives oftokens is dictated by human decision making, symbolictransactions may be largely predictable (Robb 1998 andreferences therein). The attraction for archaeologists ofthis approach is evident. Here, material evidence neednot be considered only as a proxy for language by virtueof its assumed relation to it and does not necessitatesecond-tier inferences. Rather, it can be studied and un-derstood in its own right. Indeed, the pros and cons ofstyle, standardization, and imposition of arbitrary formsin lithics as indications of symbolic behaviors havesometimes been discussed in such contexts (Ambrose1998, Chase 1991, d’Errico and Nowell 2000, Goring-Morris and Belfer-Cohen 2001, Marshack 1989, Sackett1983, White 1989, Wynn 1996).

Color symbolism is one of the symbolic frameworksused extensively by contemporary societies to conveyinformation and abstract messages through material ob-jects. Ethnographic data document the worldwide use ofcarefully chosen colors and patterns in body decorationin ritual and in practical and social contexts. Turner’s(1966) work among the Ndembu of Zambia revealed thecomplex symbolism of the basic color triad—black,white, and red—in this society’s life. Here three colorsare symbolic, through a complex chain of associations,

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 4: An Early Case of Color Symbolism: Ochre Use by Modern

hovers et al . An Early Case of Color Symbolism F 493

of human organic experiences and social relations andprovide a kind of primordial classification of reality(Turner 1966, 1970). Sagona’s (1994:10–26 and referencestherein) survey of ethnographic case studies underlinesthe fact that the Ndembu are not unique in their use ofthe basic color triad, the symbolism of the individualcolors, or the significant role of the color symbol systemin their cosmology. Red, in particular, has a symbolicsignificance that crosscuts cultural boundaries (often be-ing associated with life, success, and victory in African,Australian, and native North American societies).

Cross-cultural linguistic studies support this hy-pothesis. Most languages have been found to catego-rize colors according to a single classificatory systemwhich corresponds to a seven-stage evolutionaryscheme of color terminology (Kay and McDaniel 1997and references therein). Monochromatic “black” and“white” are the basic color terms used in any humanlanguage. Where more than two color terms exist,these two are universally followed by the term for“red,” with terms for other colors following in a con-sistent sequence across languages (Berlin and Kay1969, Kay and McDaniel 1997, MacLaury 1992). Theproperties structuring the universality of color cate-gorization have been traced to and correlated with theneuro-optical processes involved in human trichro-matic color vision (e.g., types of retinal receptors fordifferent wavelengths of light, the neural machinerythat measures the relativity of photon capture in thedifferent classes of receptors), and the dimensions ofhuman color perception (lightness, hue, and satura-tion) (Mollon 1997 and references therein). Trichro-matism is an inherent property of human visionshared only with Old World monkeys and one genusof New World monkeys, the howlers (Dominy and Lu-cas 2001). The characteristic neural and physiologicalpathway of primate color vision and the psychophys-iological organization of primate color space likelyemerged as an adaptive response to regularities in thephysical aspects of terrestrial environments (Shepard1997). Frugivory (Mollon 1997) and, more recently, theconsumption of leaves (a critical resource when fruitis scarce) have been implicated as having a uniquevalue in maintaining trichromatism in Old Worldmonkeys (Dominy and Lucas 2001).

The evolution of color terms in human languagesthus appears to recapitulate physical and neuropsy-chological processes of color sensation and perceptionin the human visual cortex (Mollon 1997, Shepard1997, Sun 1983, von Wattenwyl and Zollinger 1979).2

By the same token, the worldwide emphasis in symbolsystems on the basic color triad goes hand in handwith both the observed linguistic patterns and theneuropsychological primacy of the perception of three

2. The alternative, Sapir-Whorf hypothesis argues that linguisticstructures affect and modify patterns of perception, thought, andcultural interactions. Despite a range of suggestive data, anthro-pologists and linguists disagree about the extent of the influenceof language on culture.

colors in the human brain (see Goldstone and Barsalou1998). It would seem that the neuro-optical infrastruc-ture underlying extant symbolic color systems has ex-isted from the early days of the hominins (i.e., humansand their direct ancestors). While this does not nec-essarily mean that high cognitive faculties actuallyexisted, it renders legitimate the search for symboliccolor systems in the record of early prehistory throughtheir direct manifestations in material objects.

The record of pigment use in the course of prehistoryis consistent with the neurological and symbolic regu-larities discussed above. Black and red pigments werethe earliest to occur in prehistory and are relatively abun-dant in Paleolithic sites. Of the two colors, it is red thatdominates the Paleolithic color palette (Bahn and Vertut1997:115), usually in the form of ochre, “an earthy, pul-verulent, red [hematite], yellow, or brown [limonite, goe-thite] iron oxide that is used as a pigment” (Bates andJackson 1980). Clearly, the ochre found in archaeologicalsites was not necessarily acquired and/or used for itscolor and in symbolic contexts. The inclusion of ironoxides in an archaeological deposit may be the result ofnatural depositional processes or of its use (e.g., as pre-servatives or medicines) because of physical or chemicalproperties to which color was incidental and irrelevant.Indeed, 15 years ago acceptance of archaeological findsas indications of Middle Paleolithic symbolism—in-cluding color symbolism—was a risky, not to say naıve,proposition (Bar-Yosef 1988, Chase and Dibble 1987).Since then, however, new discoveries and new analyseshave been advanced as support for the claim that themental and cognitive capacities for symbolic behaviorwere already in place by the Middle Paleolithic (d’Erricoand Nowell 2000; Hayden 1993; Hovers et al. 1995; Hov-ers, Kimbel, and Rak 2000; Marshack 1989, 1996; Sche-partz 1993).

When asking questions about the very earliest symbolsystems, the employment of positivist tools may be themost reliable path open to us. We will apply these toolsto the ochre record of Qafzeh Cave. To argue that thisrecord results from the operation of a symbolic colorsystem, we will have to show that other explanationsfor the occurrence and the characteristics of the ochreassemblage are less parsimonious than a symbolic one.In the following discussion we will demonstrate thatochre was selected and mined for its color rather thanfor any other property. We will review the contexts ofochre use throughout the stratigraphic sequence, includ-ing recurring associations with other finds, and showthat they are consistent and specific through time. Fi-nally, we will show that such associations fulfill the hi-erarchical relationships that are the essence of a sym-bolic referential framework. Our approach to theproblem is an inductive one in the sense that we buildup from the archaeological data to contextual patterningand relationships. It is this focus on context that enablesus to evaluate the existence of symbol systems on theirown terms.

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 5: An Early Case of Color Symbolism: Ochre Use by Modern

494 F current anthropology Volume 44, Number 4, August–October 2003

The Site

Qafzeh Cave is known for the occurrence of skeletal re-mains of anatomically modern humans in the older ar-chaeological layers (XXIV–XVII, henceforth the lower lay-ers) of the cave’s terrace (Tillier 1999, Vandermeersch1981), at least some of which represent intentional burials(Belfer-Cohen and Hovers 1992, Tillier 1990, Vander-meersch 1981). Thermoluminescence (TL) and electronspin resonance (ESR) age estimates for these layers suggestoccupation at 100,000–90,000 years ago, with an averageTL date of 92,000 � 5,000 years ago (Schwarcz et al. 1988,Valladas et al. 1988). These layers contain hearths, lithicartifacts and bones of large mammals, and abundant mi-crofauna (Tchernov 1988). Worth noting are numerousburnt artifacts and thermic flint artifact debris (Hovers1997). Human remains and microfauna are absent fromthe later Middle Paleolithic stratigraphic units of the ter-race (layers XV–III, henceforth the upper layers), whichare as yet undated, and evidence for fire in these layers isscanty. In contrast with this stratigraphic dichotomy,there is no clear stratigraphic patterning in the densitiesof lithic and megafaunal bone residues (Hovers 1997, Ra-binovich and Tchernov 1995). Lithic technology is rela-tively homogeneous throughout the sequence. At thesame time, the typological characteristics of the lithic as-semblages are highly variable in the upper part of the se-quence, whereas in the lower layers notches and denti-culates predominate (Hovers and Raveh 2000). Such atypological composition may sometimes result fromtrampling (McBrearty et al. 1998), but in the current casethe absence of typical diagnostic features of trampling im-plies that the typological difference is probablyanthropogenic.

Other finds from the Mousterian layers of Qafzeh Caveterrace (hereafter “Qafzeh Cave”) include a few unpub-lished Glycymeris sp. shells and lumps of ochre, both ofwhich occur only in the lower layers of the terrace. Nonewere found in the Mousterian layers in the interior of thecave, which are correlated stratigraphically with the up-permost layers of the terrace sequence (Vandermeersch1981). A single ochre lump with clear signs of scrapingwas reported (Vandermeersch 1969), and this publicationhas often been referred to in the debate about Middle Pa-leolithic symbolism (e.g., Bar-Yosef 1988, Bar-Yosef andVandermeersch 1993, Bednarik 1992a, Chase and Dibble1987, d’Errico and Nowell 2000). However, the full recordof ochre from the site contains numerous other finds.

The Sample

Seventy-one pieces of ochre were available for detailedstudy (table 1). The original field documentation indi-cated that additional pieces had been recovered duringthe excavations. Given that not a single case of mis-identification was encountered during our study of theavailable samples, we assumed that these missing pieces

had also been identified correctly as ochre. Thus, theminimum number of ochre pieces was 84.

For the existing pieces, the available most precise pro-venience was recorded. The provenience of 68 pieces,derived from the whole stratigraphic column, is knownto the closest meter. Of these 38 (56%) were found intwo adjacent squares (C10 and C11 [fig. 1]). On the whole,50 pieces (73.5%) originated from 4 m 2 in the westernpart of the excavation. The spatial clustering occursthroughout the stratigraphic sequence of the ochre-bear-ing (lower) layers; the upper layers are devoid of bothochre pieces and artifacts with ochre stains.

Methods

We first present the full inventory, including mineralog-ical and chemical characteristics, of ochre lumps foundin the assemblages and elucidate the temporal and spa-tial trends of ochre distribution. We go on to identify thesources of ochre, using petrographic, chemical, and min-eralogical analyses to determine the characteristics ofthe archaeological ochres and compare them with thoseof such potential sources, and to assess possible criteriaof selection. The technological organization of ochreutilization is studied by looking at patterns of modifi-cation of ochre lumps and indications of tool use in ochreprocessing. We then move to a discussion of the temporaland spatial associations of ochre with lithics, fauna, andhuman skeletal remains. Finally, we discuss the impli-cations of our findings for the notion that ochre was partof a Middle Paleolithic symbolic culture.

The Munsell color chart (Munsell 1962) was used tostandardize color description of all the ochre pieces. Inaddition, each lump was described as to its shape, hard-ness, and signs of modification. For the purpose of initialsorting and definition of the ochre types, all the lumpswere examined under a binocular microscope (#50 mag-nification), and their preliminary lithology (sand, clay,or silt) and mineralogical classification (hematite, cal-cite, dolomite, quartz) were recorded.

Nine pieces (#1–5 and 7–10) from the three layersrichest in ochre were selected for chemical analysis. Thenumber of pieces selected from each layer reflected itsrelative wealth of ochre pieces, but selection from thelayer’s assemblage was random. The chemical compo-sition of these pieces was determined by inductively cou-pled plasma-atomic emission spectroscopy (ICP-AES) us-ing a Jobin Yvon (JY-48) polychromator for barium,cobalt, chromium, copper, nickel, manganese, and zinc.The element content of arsenic, antimony, cadmium,selenium, lead, and uranium was detected using a PerkinElmer Sciex Elan-6000 inductively coupled plasma-massspectrometer (ICP-MS). After the petrographic study hadbeen completed, 14 additional pieces were submitted toanalysis by ICP-AES, bringing the frequency of chemi-cally studied samples to 32.4% of the total assemblage.Selection of this second series of samples was based inpart on outstanding features observed under the micro-

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 6: An Early Case of Color Symbolism: Ochre Use by Modern

hovers et al . An Early Case of Color Symbolism F 495

table 1Ochre from the Mousterian Layers of Qafzeh Cave Terrace

LayerandSample

Squareand

NumberX

(cm)Y

(cm)

DepthbelowDatum

(cm) NMaximalSize (mm)

Weight(g)

Color(Munsell) Description

Petrographicand Lithologic

Description

XVII4 (Q79) B10 – – – 1 31 8.94 7.5R 3/8 Fragment. Fine-grained, hema-

titic limestone.Sand present innegligible amounts.Some ferruginousoolites.b, c

5 (Q78) B11-128 – – 518 1 38 18.82 7.5R 4/8 Fragment. Fine-grained hema-titic dolomite withsome ferruginousoolites.b, c

6 B16 1 56 56.12 7.5R 3/6 Fragment with scrap-ing marks on oneface and edge (firstreported by Vander-meersch 1969).Grooves in variousdirections.

Fine-grained hema-titic siltstone con-taining some ferru-ginous oolites ! 0.1mm. Grooves on ar-tifact surface cov-ered with secondarycalcite.

7 (Q79) C10-288 97 18 534 1 49 41.72 7.5R 3/6 Fragment. Very hardmaterial, with a lotof quartz.

Hematitic sandstone.Quartz grains sub-rounded at 0.25mm. Ferruginousoolites in the sizeorder of 0.2 mm.b, c

8 (Q79) C11-590 – – 524 1 45 19.27 7.5R 4/8 Fragment. Fine-grained hema-titic sandstone. Fer-ruginous oolites !

0.1 mm.b

9 A13 13 31 491 2 43 28.55 7.5R 3/8 Two pieces of thesame lump, onechipped off theother.a One surfaceof the larger pieceis worked.

Fine-grained (clay-sized), hematiticsiltstone containingferruginous oolites! 0.1 mm.b, c

13 A15 – – – 4 20 9.59 7.5R 4/8 Several small lumps,one covered withdark brown clay-siltwith many whitespots (as in #12).

Maghemitic dolo-mite.b, c

30 (Q78) B11-119 55 55 501 2 49 19.13 5.0R 4/10 Red pigment, colorseasily, coated inashes (?).

Hematitic dolomiticsandstone. Abun-dant ferruginous oo-lites ! 0.1 mm.b, c

33 (Q79) C10-276 97 87 531 1 27 5.42 5.0R 3/10 Very hard lump ofdark red material.

Hematitic siltstone.Few ferruginous oo-litic ! 0.05 mm.

51 (Q79) C10-339 – – – 1 39 33.83 5.0R 5/10 Pebble-like lump ofdark red material.

Hematitic dolomite.Calcite in vugs.

XVIII43 (Q79) C11-498 34 72 527 1 31 7.82 5.0R 5/8 Dark red material,

coated in silty ash.Hematitic carbonatic

sandstone; no ferru-ginous oolites.

XVIIIa25 A12-446 – – – 1 39 35.66 7.5R 3/6 Lump of bright red,

relatively hardmaterial.

Hematitic carbonaticsandstone; no ferru-ginous oolites.

XIX10 (Q79) C10-383 70 20 601 1 58 62.67 7.5R 5/8 A fragment with

scraping marks.Hematitic siltstone

with a dolomiticmatrix.b

15 (Q79) C11-607 – – 546 4 36 14.69 7.5R 3/6 Two large lumps andtwo very small frag-ments of friable redmaterial.

Maghemitic lime-stone. b, c

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 7: An Early Case of Color Symbolism: Ochre Use by Modern

496 F current anthropology Volume 44, Number 4, August–October 2003

table 1(Continued)

LayerandSample

Squareand

NumberX

(cm)Y

(cm)

DepthbelowDatum

(cm) NMaximalSize (mm)

Weight(g)

Color(Munsell) Description

Petrographicand Lithologic

Description

16 (Q75) C13-58 50 50 501 1 33 25.26 7.5R 4/8 A lump of hard lightred material.

Hematitic limestoneand dolomite; noferruginous oo-lites.b, c

17 (Q75) C13-84 74 60 – 1 29 5.16 10R 4/6 A small lump of harddark red material.

Siltstone containingabundant ferrugi-nous oolites, ! 0.2mm.

18 (Q75) C13-75 37 40 532 1 – 3.87 7.5R 3/6 Very friable, almostpowdery dark redmaterial.

Hematitic siltstonecontaining abun-dant ferruginous oo-lites ! 0.05 mm.

19 (Q75) C13-86 68 32 – 2 22 4.92 7.5R 4/8 Small lump of harddark red material.

Hematitic siltstonecontaining someferruginous oolites! 0.1 mm.

34 (Q79) C10-368 – – 598 1 15 1.16 5.0R 3/8 A small fragment ofsoft dark redmaterial.

Hematitic siltstonecontaining ferrugi-nous oolites ! 0.05mm.

35 (Q79) C10-369 – – 595 1 14 1.69 5.0R 3/10 A small fragment ofsoft dark redmaterial.

Hematitic siltstonecontaining ferrugi-nous oolites ! 0.05mm.

36 (Q79) C10-370 56 53 598 4 18 5.03 5.0R 3/8 Three small frag-ments of soft darkred material � oneof harder, stonelikematerial.

Hematitic siltstonecontaining ferrugi-nous oolites ! 0.05mm. Many calcitevugs.

37 (Q79) C11-780 11 11 589 1 13 0.73 5.0R 3/10 Fragment. Hematitic siltstonecontaining ferrugi-nous oolites ! 0.05mm.

46 (Q79) C10-382 – – 601 1 28 8.41 7.5R 3/4 “Flake” of very harddark red material,one face worked toa smooth surface.

Hematitic siltstone.Few ferruginous oo-lites ! 0.05 mm.

54 (Q79) C10-374 20 25 500 1 51 86.22 5.0R 5/8 Large, broken concre-tion with concen-tric rings of variouscolors (dark on theoutside, lighter onthe inside). Outerlayer scratches iron.

Concretion of goethi-tic siltstone, includ-ing shiny (pyrite?)crystals situatedwithin depressionsin the sandstone,the inner part ofwhich is coatedwith calcite. Outerlayer crystalline sil-ica.b, c

55 C11-504 89 0 534 1 13 0.69 5.0R 5/8 Very small fragment. Hematitic siltstone.XXI

1 (Q73) B12 – – – 1 54 67.71 7.5R 3/8 Large piece, one sur-face worked andsmoothed.

Hematitic siltstonecontaining few fer-ruginous oolites !

0.05 mm.b

3 (Q79) C11-761 60 30 561 1 34 10.75 7.5R 3/6 Fragment. Fine-grained hema-titic limestone anddolomite. Oolithicsize 0.2 mm. Con-tains maghemite.b, c

12 (Q78) – – – – 1 43 42.55 5.0R 4/8 A lump covered withdark brown clay-siltwith many whitespots in it (ashes?).

Hematitic siltstonecontaining few fer-ruginous oolites.

23 (Q75) C13-72 – – 531 1 57 39.61 7.5R 4/6 Lump of friable red-yellow material.

Hematitic lime-stone.b, c

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 8: An Early Case of Color Symbolism: Ochre Use by Modern

hovers et al . An Early Case of Color Symbolism F 497

table 1(Continued)

LayerandSample

Squareand

NumberX

(cm)Y

(cm)

DepthbelowDatum

(cm) NMaximalSize (mm)

Weight(g)

Color(Munsell) Description

Petrographicand Lithologic

Description

26 B14-193 – – – 1 44 56.41 7.5R 3/6 Lump of very harddark red material.

Silty, hematitic lime-stone. Calcite(showing as whitespots) fills thevugs.b, c

27 B14-188 – – – 1 57 57.78 10R 4/6 Lump of very harddark red materialwith one smoothed(abrasion?) surface.

Silty hematiticsandstone.

28 C13-78 – – – 1 – 5.70 7.5YR 5/8 Very small fragments. Goethitic and hema-titic (carbonatic)sandstone. Abun-dant ferruginous oo-lites ! 0.05 mm.

29 (Q75) C12-62 – – 549 1 41 23.82 – Lump of hard darkred material.

Hematitic siltstonecontaining ferrugi-nous oolites ! 0.5mm.

32 (Q79) C10-386 – – 598 1 17 2.08 5.0R 4/10 Soft red piece. Hematitic limestoneand dolomite withfew ferruginous oo-lites. Containsmaghemite.b, c

38 (Q79) C10-394 – – – 1 28 7.21 5.0R 3/10 Fragment. Hematitic siltstone.39 (Q79) C10-406 – – – 3 19 3.78 5.0R 3/8 Fragment. Hematitic sandstone

with abundant fer-ruginous oolites !

0.1 mm.40 (Q79) C11-531 40 94 568 1 – 0.18 5.0R 5/8 Very small fragment. Hematitic siltstone

with few ferrugi-nous oolites.

41 (Q79) C11-537 58 83 572 1 31 9.27 5.0R 5/10 Very hard lump ofdark red material.

Hematitic limestoneand dolomite withfew ferruginous oo-lites.b, c

42 (Q79) C11-536 88 90 571 2 22 3.89 5.0R 5/8 Fragment. Hematitic siltstonecontaining few fer-ruginous oolites.b, c

47 (Q79) C11-790 66 9 613 1 22 8.07 7.5R 3/6 Small lump of redmaterial, coatedwith white ash (?).

Hematitic sandstonecontaining someferruginous oolites! 0.05 mm. Calcitein vugs.

49 (Q79) C11-548 – – 580 1 19 2.18 7.5R 3/8 Fragment, dark red. Hematitic siltstone inphysical contactwith dolomite (?).

50 (Q79) C11 – – – 1 21 4.55 5.0R 5/8 Fragment. Hematitic siltstone.52 (Q79) C11-543 – – – 1 53 37.40 5.0R 5/8 Fragment. Fine-grained hema-

titic sandstone.XXI/XXII

22 (Q75) C13-93 65 80 533 1 32 16.09 7.5R 3/4 Lump of dark red ma-terial covered withremains of ashes.

Hematitic limestonecontaining few fer-ruginous oolites.Manganese coat-ing.b, c

XXII2 (Q73) A12-588 – – – 1 33 10.15 7.5R 3/6 A “flake” of very hard

dark red material.Hematitic sandstone

containing ferrugi-nous oolites ! 0.2mm.b, c

14 (Q73) B12 – – – 1 31 6.6 7.5R 3/6 A small lump of hardred material.

Hematitic limestone.Contains dolomite,manganese, andmaghemite.b, c

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 9: An Early Case of Color Symbolism: Ochre Use by Modern

498 F current anthropology Volume 44, Number 4, August–October 2003

table 1(Continued)

LayerandSample

Squareand

NumberX

(cm)Y

(cm)

DepthbelowDatum

(cm) NMaximalSize (mm)

Weight(g)

Color(Munsell) Description

Petrographicand Lithologic

Description

20 (Q75) C12-103 24 48 574 1 42 23.45 7.5R 3/5 A lump of hard darkred material.

Hematitic carbonaticsandstone; no ferru-ginous oolites.

24 B12-663 – – – 1 33 9.93 7.5R 3/8 Lump of very harddark red material.

Hematitic limestone;no ferruginous oo-lites. Containsmaghemite.b, c

55 (Q69) A13 – – – 1 12 5.20 5.0R 4/10 Very hard pink-redlump.

XXIII11 (Q79) C10-410 84 44 644 1 39 14.81 7.5R 3/8 Yellow and red

fragment.Hematitic siltstone

containing someferruginous oolites.

44 (Q79) C11-792 – – 624 1 12 0.78 7.5R 4/8 Fragment. Hematitic siltstonewith few ferrugi-nous oolites.

48 (Q79) C10-408 – – 630 1 24 5.19 7.5R 3/6 Pebble-like lump ofdark red material.

Hematitic siltstonecontaining abun-dant ferruginous oo-lites ! 0.05 mm.Calcite in vugs.

XXIV21 (Q75) C12-113 – – 611 1 35 31.77 7.5R 4/8 Lump of very hard

dark red material.Hematitic siltstone.

31 (Q79) C11-798 65 30 653 1 27 11.19 5.0R 4/10 Very hard pink-redlump, possibly aburnt stone(?).

Hematitic dolomite.b, c

45 (Q79) C10-420 – – 664 1 39 49.43 7.5R 3/4 Pebble-like, brokenlump of very harddark red material,one tip and edgerounded, one sur-face concave.

Hematitic sandstonecontaining abun-dant ferruginous oo-lites ! 0.05 mm.

53 C12-114 – – 613 1 65 69.58 – Fragment of very hard(stonelike) red-pinkmaterial with greyincrustations andadhered ashes.

Hematitic limestoneand dolomite. Metalparticles in hema-titic part probablyderived from toolsused in the excava-tion during theearly 1970s.b, c

Total 71 1,142.48

aWhen a sample consists of more than a single piece, measurements are given for the largest of the two.bExamined chemically.cExamined by X-ray diffraction.

scope, as this second analysis was intended to describethe range of chemical variation rather than the norm.

The mineralogical compositions of 21 samples (30% ofthe assemblage) were determined by X-ray diffraction, us-ing a Philips PW1730 X-ray generator and PW1710 dif-fractometer control, with Cu K a irradiation and Ni filter.

In order to identify the geological sources of the ochre,we compiled data from geological maps and survey anddrilling reports in the region and conducted a limited fieldsurvey. Seven rock samples collected from a locality con-sidered (on the basis of the petrographic analyses) to havebeen a potential source of the archaeological ochres were

submitted to ICP-AES and X-ray diffraction. Similar datafor other, more remote potential sources near Qiryat She-mona, in the north of Israel, were obtained by samplingand from the available literature and unpublished data.

Results

characteristics of the ochre assemblage

The relative frequencies of pieces of ochre within thetotal lithic assemblage (knapped lithics and ochre lumps

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 10: An Early Case of Color Symbolism: Ochre Use by Modern

hovers et al . An Early Case of Color Symbolism F 499

Fig. 1. Map of excavation in Qafzeh Cave, showing lateral distribution of ochre lumps and of hominid re-mains (regardless of stratigraphic distribution). Triangles, burials; rectangles, skeletal remains of adults; circles,remains of infants (after Tillier 1999 and Vandermeersch 1981). Numbers indicate number of pieces in square.

but not lithic chips and chunks) do not show a cleardirectional change through time (fig. 2). The observedcolor range is between red and pink, usually 5.0R 5/8 to10.0R 4/6, with few (ca. 5%) yellows and browns. Thelargest numbers of ochre pieces were found in layers XIXand XXI (table 2). Combined, they constituted 60% ofthe total number of ochre pieces in the sequence andwere followed by the assemblage from layer XVII. Thesame pattern held when the missing ochre pieces wereadded to the calculation (layer XIX, N p 28; layer XXI,N p 23; layer XVII, N p 18). The absolute frequenciesof ochre lumps vary with the excavated volume of thestratigraphic layers (see Hovers 1997:table II.2 for pri-mary data). Ochre mass had a different distribution. Thehighest cumulative weight per layer occurred in layerXXI, followed by layers XVII (in which over 50% of thetotal weight of ochre occurred [table 3, fig. 2]) and XIX.

Because the sample sizes are small, statistical com-parisons are not meaningful, but some general tenden-cies can be observed. Each of the ochre assemblages con-sisted of both relatively large and small pieces. Of the

three assemblages with ten or more pieces (layers XVII,XIX, and XXI), that of layer XVII stood out for its ten-dency toward the occurrence of larger and heavier pieces(figs. 3 and 4, table 4). Absolute frequencies of ochre inlayers XIX and XXI were indeed higher than in layer XVII,but the descriptive statistics demonstrated that, in con-trast to the situation in layer XVII, these were often verysmall crumbs, as was also the case for layer XXIII.

petrographic, chemical, and mineralogicalproperties

While the sizes and shapes of the 71 ochre pieces werehighly variable, their lithological properties were rela-tively homogeneous. Ochre is an iron oxide that is usu-ally impure—that contains a matrix. Petrographic andmineralogical examinations revealed that the sedimen-tary types of iron oxides in Qafzeh, in order of decreasingfrequency, were siltstone, quartzic sandstone, limestone,dolomitic limestone, and dolomite. The chemical anal-yses showed that the archaeological ochres were char-

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 11: An Early Case of Color Symbolism: Ochre Use by Modern

500 F current anthropology Volume 44, Number 4, August–October 2003

Fig. 2. Distribution of ochre by layer. Diagonal striping, percentage of total weight; white, percentage of totalnumber of pieces; black, percentage of total lithic number of lithics and ochre.

acterized by their relatively high iron content (20–50%),with a few exceptions (notably, #5, #16, #42, #31, and#53, which contained 0.7–9% Fe2O3). The common ironmineral was hematite (a�Fe2O3), as was suggested by thered-to-pink color that characterized most of the samples.Only a few pieces contained goethite (a-FeOOH) (e.g.,#54 and #26) or maghemite (g�Fe2O3) (#13, #15, #32, #14,and #24). Small ( ! 0.2 mm) ferruginous oolites wereobserved in the vast majority of the pieces, regardless ofthe specific iron mineral.

archaeological evidence for ochreprocessing

Evidence for the utilization of ochre came from two typesof finds: ochre lumps bearing signs of utilization andochre-stained artifacts. The most obvious signs of usagewere found on a lump from layer XVII (#6) which wasscraped on both faces (Vandermeersch 1969). One surface(fig. 5) bore a deep, wide groove accompanied by shal-lower striations subparallel to one another. A second faceof the same piece (fig. 6), perpendicular to the first, ap-peared to have been heavily scraped and exhibited a con-cave surface. This face bore several distinct types ofmarks. On the elevated part of the surface there weredeep grooves that intersected one another and created apattern reminiscent of the ones reported from BlombosCave (Henshilwood et al. 2002), albeit on a smaller scale.On the concave lower area of the same surface there wereshallow striations subparallel to one another.

A single piece (#10) also bore remnants of shallow stri-ations on one of its edges.

One facet of piece #27 was very smooth and shiny red,with very fine striations that could be seen under mag-nification; all the other facets of this piece were rough,with only patches of dull red color. Another piece (#45)exhibited two perpendicular smoothed facets, one con-vex and the other concave. The geometry of this pieceresulted from manipulation of the ore. On the rounded

contact zone formed artificially between the two facetsthere were very fine striations.

Two additional pieces were “flakes” of ochre. Piece #2exhibited a bulb of percussion and convex-concave pro-file identical to those of flint flakes. Piece #46 was bro-ken at its “proximal” end; its profile was also similar tothat of a flake. The shapes of these two pieces were in-consistent with natural breakage patterns. It is thus pos-sible that the Qafzeh hominids practiced flaking ochrelumps as a technique for splitting the original chunksand obtaining pieces of the desired size.

Several lithic artifacts bearing ochre stains were re-covered from layers XXII–XVII, the majority originatingfrom layer XVII (table 5). This group of artifacts consistedmainly of unretouched flakes and a few retouched tools.On some of these ochre appeared along the edges,whereas in other cases the tips were stained with ochre(e.g., the notch on a broken flake from layer XXII; a core-trimming element from layer XVII [fig. 7, 1]). Micro-scopic analysis revealed the presence of traces of ochreon the engraved face of a Levallois core from layer XVII(Hovers, Vandermeersch, and Bar-Yosef 1997, Nowell,d’Errico, and Hovers 2001).

An exceptional artifact was a large Levallois core fromlayer XIX, flaked by the centripetal recurrent mode ofLevallois flaking (Boeda, Geneste, and Meignen 1990),that bore extensive ochre stains on its flaking surface(fig. 8). The residues of red pigment were thickest anddensest in the deepest part of the large negative scar ofthe last Levallois flake removed from the core. Outsidethe contours of this scar, ochre was rare and, when pre-sent, occurred as small, thin, discontinuous stains.

selection and acquisition of ochre

Qafzeh Cave is situated on the slope of Har Qedumim(Jebel Qafzeh) at an elevation of 250 m above sea levelon the eastern bank of a narrow creek (Arabic Wadi el-Haj, Hebrew Nahal Mizra) that cuts through the steep

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 12: An Early Case of Color Symbolism: Ochre Use by Modern

hovers et al . An Early Case of Color Symbolism F 501

table 2Stratigraphic Distribution of Lithics, Ochre Pieces, and Hominid Remains

LayerExcavatedVolume

N Lithics(w/Debris)

N Lithics(w/o Debris)

N OchrePieces

RelativeFrequency

Density(per m3)

N HominidRemains

XVII 3.35 2,870 912 15 .005 4.48 13XVIII 0.80 346 103 2 – 2.50 –XIX 3.10 1,738 543 20 .012 6.45 1XXI 2.25 1,059 329 21 .020 9.33 –XXII 1.08 616 287 6 .008 4.63 1XXIII 1.20 12 12 3 .250 2.50 –XXIV 0.67 108 69 4 .037 5.97 –

escarpment facing the Yizrael (Esdraelon) Valley (Israelgrid 178.05/231.75). At present it is about 110 m abovethe valley floor. The exposed geological section in thevicinity of the cave consists mainly of sedimentary rocksfrom Albian (Lower Cretaceous) to Eocene age, whichform a ca.-1,000-m-thick column (Weiler 1968). The ma-jor part of the lithological column in the immediate vi-cinity of the site consists of dolomite and limestonewithout any siltstone, sandstone, or ochre occurrences.Specifically, the terrace sediments contain angular rub-ble derived from the dolomite bedrock (Farrand 1979:377). The predominance of siltstone and sandstone inthe ochre matrices strongly undermines the possibilitythat ochre accumulated on site as natural debris. Addi-tionally, since excavation volumes do not differ signifi-cantly between the lower and upper parts of the sequence(Hovers 1997:table II.2), the absence of ochre and ofochre-stained artifacts from the upper part of the se-quence cannot be explained as stemming from differ-ences in excavation volumes. The available geologicaland archaeological data thus suggest that the ochre wasactively collected by the site’s occupants, who trans-ported it to the cave when layers XXIV–XVII were de-posited some 100,000–90,000 years ago.

Possible sources of hematitic ochre occur in threeforms within 8 km of the cave. Veins of iron oxide, com-posed mainly of goethite and some hematite, are onepotential source located 2.5 km west of the cave. Theseare found as joint fillings in the dolomitic rocks of Ce-momanian age. Another possible source of availableochre is ferruginous concretions 1 to 10 cm in size, com-posed mainly of goethite, with some occurrences of he-matite. These are present in limestone, chalk, and marlof the Lower Cretaceous at Mt. Devora, 4 km east ofQafzeh Cave, and in Paleocene and Eocene sediments 3to 4 km northeast of the cave. Finally, blocks of yellow-brown-red quartzic sandy limestone containing ferrugi-nous oolites, of Lower Cretaceous age (Weiler 1968), arefound on the lower southern slopes of Mt. Devora as wellas in the Mt. Tavor area some 8 km east of the cave.

The petrographic and mineralogical analyses show thetotal absence of ferruginous concretions and of fragmentsof iron veins in the archaeological sample, indicatingthat the majority of the potential geological sources werenot tapped for ochre. At the same time, the abundancein the archaeological ochre sample of clastic siltstone

and of sandstone enriched with ferruginous oolites sug-gests that the ochre source is in quartzitic sandstonestrata intercalated with beds containing ferruginous oo-lites. Rocks of such lithology are typical of the LowerCretaceous sequence and are exposed in the Mt. Devoralocality and on the western slopes of Mt. Tavor. It thusappears that Qafzeh Cave’s inhabitants selected ochreexclusively from these localities.

Alongside some broad similarities in the color and gen-eral petrographic characteristics of the archaeologicalmaterial and the geological samples and their minera-logical properties there were a number of significant dis-similarities. First, the ochre from the cave came mainlyfrom siltier sediments than the ochre from Mt. Devoraand Mt. Tavor. Second, as we have seen, the ochre lumpsfrom the cave were relatively rich in iron, compared withonly 0.7–14% in the geological samples, and as a resultthere were also anomalously high values of some traceelements (Chao and Theobald 1976). Third, the archae-ological material was characterized by the presence ofhematite, whereas in the geological samples goethite wasrelatively abundant. Finally, in more than 50% of thecave samples, the abundance of ferruginous oolites wasfar greater than in the sandstone exposures of Mt. Devoraand Mt. Tavor.

Comparative fieldwork and mineralogical and petro-graphic studies made it possible to correlate the Mt. De-vora stratigraphic section with the well-known exposureof the Lower Creataceous Hidra Formation of RamimRidge, near Qiryat Shemona. This correlation indicatedthat the part of the Hidra Formation in the Mt. Devoraarea that contained the siltstones rich in ferruginous oo-lites was covered by subrecent (post–Middle Paleolithic)talus debris up to 12 m thick in the nearby drillholeDevora-1. The implication of this geomorphic feature isthat while the particular lithological bed from which theQafzeh hominids obtained their ochre likely existed inthe Mt. Devora area, our geological samples did not de-rive from it. This situation thus explains the incompat-ibility between the geological and the archaeologicalmaterials.

Apart from the two sites at Mt. Devora and Mt. Tavor,there are three exposures of Lower Cretaceous ferrugi-nous oolite quartzic beds in northern Israel (fig. 9). Inorder of increasing distance from the site, these are ‘Einel Assad, about 30 km northeast (D. Levite, Internal Re-

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 13: An Early Case of Color Symbolism: Ochre Use by Modern

502 F current anthropology Volume 44, Number 4, August–October 2003

table 3Descriptive Statistics for Weight of Ochre Pieces (g)

Layer Count Total Weight Mean S.D. Minimum Maximum Median

XVII 10 241.39 24.14 16.04 5.42 56.12 19.20XVIII 2 43.48 21.74 19.69 7.82 35.66 21.74XIX 13 220.51 16.96 26.84 0.69 86.22 5.03XXI 18 382.93 21.27 22.64 0.18 67.71 8.67XXII 6 66.22 11.90 6.80 5.20 23.45 10.15XXIII 3 20.91 6.92 4.14 0.78 14.81 5.19XXIV 4 161.96 40.49 24.90 11.19 69.58 40.60

port to the Geological Survey of Israel), Wadi Malikh,about 40 km southeast (Mimran 1972), and the LowerCretaceous Hatira and Hidra Formations exposed atRamim Ridge near Qiryat Shemona, about 60 km north-east (Kafri 1991). The mineralogical and chemical com-positions of the first and last of these exposures (Zack-heim 1997) differ from those of the archaeologicalmaterial found in Qafzeh Cave mainly in the concen-trations of trace elements and in some cases also in thecommon form of iron oxide (goethite rather than he-matite [Zackheim 1997: tables 12, 14]).

Discussion

the operational sequence of ochre use

Selection criteria. The ochre lumps from Qafzeh Cavewere mostly silty and clayey, with their hues signifi-cantly skewed toward the red. Red ochre can either becollected from natural sources or produced, primarily byheating, from other iron minerals (see Weinstein-Evronand Ilani 1994, Wreschner 1980, Zackheim 1997). Theevidence for the existence of hearths in the layers ofQafzeh Cave which contained the ochre (Hovers 1997,Vandermeersch 1981) raises the possibility that yellowgoethite was transformed by heating, sometimes acci-dentally, into red hematite (e.g., Weinstein-Evron andIlani 1994). We were unable to use Pomies, Menu, andVignaud’s (1999) mathematical manipulation of X-raydiffraction data to distinguish between natural hematiticochre and red ochre which originated from the heatingof goethite. On the basis of two lines of reasoning, how-ever, we assume that the ochre assemblage from Qafzehrepresents the color preferences of the hominids inde-pendent of the use of heating.

First, if goethite had been transformed into red he-matitic ochre by accidental, random heating, lumps ofyellow iron oxide should have occurred in the archaeo-logical ochre assemblages. These, however, are ex-tremely rare (table 1). The chemical properties of goethitedo not change when the mineral is heated. Therefore, analternative hypothesis—that fire was used intentionallyto change the yellow color of goethite to red—wouldimply that the characteristic red hues were the specificgoal of the heating. This hypothesis is weakened by thefact that during the Middle Paleolithic goethite with the

sandy matrix characteristic of the ochre found at QafzehCave would have been mined from the same geologicalunits of the Lower Cretaceous sequence that containedred hematitic ochre. Hominids with access to the formercould have acquired the latter with the same amount ofeffort and would not have needed an additional processto obtain the very same end product.

Lithological units containing ores similar to the ar-chaeological ochres are not ubiquitous among the iden-tified probable sources of ochre or in the more distantLower Cretaceous deposits. Still, Qafzeh’s occupants didnot tap the more readily available ferruginous concre-tions and veins of iron oxides. The combined geologicaland archaeological data therefore point to active searchand sampling in order to meet well-defined selection cri-teria in the course of ochre acquisition.

Identifying these criteria is not an easy task, since thearchaeological ochres shared a number of chemical andvisual properties any one of which could have been sucha criterion. One possibility is that it was a combinationof characteristics rather than any single one that satisfiedthe selection criteria. Alternatively, Middle Paleolithichominids might have selected for a single desired prop-erty that was accompanied by other characteristics asby-products or side effects. The red hues (and, some-times, the visible ferruginous oolites) likely advertisedto the Qafzeh hominids the presence of ochre of the de-sired properties—but was the red color the main criterionfor selection?

The utilization of ochre through grinding and scrapingmight explain the selection for iron oxides derived fromrelatively soft materials (e.g., clays and silts) which couldbe pulverized easily. It does not explain the overrepre-sentation of red hues in the archaeological assemblages.Yellow goethite is ubiquitous in all the Lower Creta-ceous exposures considered above as sources and occursin matrices of similar hardness and chemical properties.Accepting that the archaeological ochres had not under-gone heat-induced chemical changes, the absence of goe-thite from the Qafzeh sample suggests that the red hueof the hematitic ochre was a crucial selection criterion.Given the limited geographic distribution of exposurescontaining the particular type of ochre and the paucityof hematitic ochre within the relevant beds of the LowerCretaceous strata, it seems that considerations of cost-effectiveness did not play a central role in decision mak-

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 14: An Early Case of Color Symbolism: Ochre Use by Modern

hovers et al . An Early Case of Color Symbolism F 503

Fig. 3. Frequency distribution (counts) of maximal dimensions (mm) of ochre fragments. Circles, layer XVII;squares, layer XIX; triangles, layer XXI.

ing related to ochre acquisition. Moreover, the data sug-gest that the same selection criterion and similarprocesses of decision making were employed repeatedlythroughout the time span of the deposition of the lowerlayers.

Techniques of ochre processing. The modificationsseen on the pieces of ochre in Qafzeh fall within therestricted range of utilization techniques typical of ochreexploitation in a variety of temporal and geographicalcontexts (d’Errico and Nowell 2000). The appearance ofdeep intersecting grooves on lumps of ochre (e.g., #6) ischaracteristic of the use of the pointed tips of lithic toolsin the process of scraping. Subparallel shallow striationsare normally associated with the use of the lateral edgeof a blank or of a retouched tool (d’Errico and Nowell2000:160). The apparent marks on three of the ochrelumps thus suggest that both modes of scraping wereknown to the Qafzeh hominids. In the case of lump #6,the shallow marks appear to postdate the deep grooves.This implies either that the utilization of the piece tookplace during two or more sequential events of scrapingor that tools were switched and the manner of ochreexploitation changed in the course of its utilization.

Another technique of extracting powder from the im-ported natural lumps was grinding (pieces #27 and #45).In contrast to the evidence from the African MiddleStone Age, in which ochre-stained groundstone tools arereported from a number of sites (McBrearty and Brooks2000:528 and references therein), and to some Europeanexamples (e.g., Bordes 1952), the lower layers of QafzehCave produced no well-executed groundstone artifacts.However, the placement and concentration of ochre onthe Levallois core from layer XIX fit the definition ofUpper Paleolithic ochre receptacles as exhibiting “tracesof pigments in their concavity [which] imply that theyserved in storage and preparation of pigment” (de Beaune1993:177). It is possible that this core was recycled into

an ochre receptacle (fig. 8). The shape and the topographyof the large negative scar of the core on which the ochreis found resemble those of a small cup-hole (see Kraybill1977:495 for examples of pre–Upper Paleolithic grindingstones with ochre).

None of the scraped pieces was reduced to the formof a thin tablet (see Henshilwood et al. 2001:fig. 8; Watts1999). Similarly, grinding never proceeded to the pointthat it resulted in the distinct morphology of a “crayon,”in which the ground facets converge into a point (e.g.,Henshilwood et al. 2001:433; Watts 1999:figs. 7.2–7.4).It is also noteworthy that, while relatively few lumps(only 8.5% of the total number) were modified in anyway, these tended to be among the largest and heaviestpieces (43% of pieces weighing over 50 g showed evi-dence of modification; in two cases, #6 and #27, theywere the heaviest pieces in their respective layers). Nosigns of grinding and/or scraping were observed on anyof the light pieces ( ! 10 g). Since there was no correlationbetween the size and the hardness of the ochre pieces,it is reasonable that the larger and heavier pieces werethe sources from which pigment was extracted and thesmall pieces constituted processing debris. The distri-butions of size and mass of the ochres are consistent withthis interpretation (figs. 3 and 4) in that each occupationlayer contained a few large and many small pieces. Thesedata also suggest that ochre lumps were not exhaustedbefore discard. Combined, these sets of data are com-patible with short transport distances, which did not im-pose heavy costs on the use of the natural ores.

Ochre pieces throughout the sequence of the lowerlayers showed a distinct spatial clustering, but this wasnot the case with the few ochre-stained artifacts. Theweak spatial association between the stained lithics andthe ochre lumps undermines suggestions that the pres-ence of the former resulted from unintentional, randomstaining due to physical proximity to the pieces of ochre.

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 15: An Early Case of Color Symbolism: Ochre Use by Modern

504 F current anthropology Volume 44, Number 4, August–October 2003

table 4Descriptive Statistics for Maximal Dimension ofOchre Pieces (mm)

Layer Count Mean S.D. Minimum Maximum Median

XVII 10 39.70 11.09 20.00 56.00 41.00XVIII 2 35.00 5.66 31.00 39.00 35.00XIX 12 27.50 14.98 13.00 58.00 25.00XXI 16 35.13 14.74 17.00 57.00 32.50XXII 6 30.50 9.89 12.00 42.00 32.50XXIII 3 25.00 13.53 12.00 39.00 24.00XXIV 4 41.50 16.44 27.00 65.00 37.00

Fig. 4. Frequency distribution (counts) of ochre by weight (g). Circles, layer XVII; squares, layer XIX; triangles,layer XXI.

By the same token, the locations of the ochre on someartifacts (table 5) in fact conform with predictions de-rived from the two methods of scraping observed on theochre pieces—using the tip of a tool or its lateral edges.

explaining the spatio-temporal patterningof ochre finds

The ochre record of Qafzeh Cave is extraordinary in itstemporal disjunction. Frequencies of ochre pieces andochre-stained lithics did not simply change gradually orstochastically throughout the sequence; instead, themodification and use of ochre seem to have stopped afterthe deposition of layer XVII. Where present, however,the archaeological ochres were remarkably homoge-neous in their chemical and mineralogical properties andmode of extraction. Plausible explanations for the pres-ence and significance of ochre must account for thesetwo seemingly unrelated phenomena.

Although the authenticity of ochre finds in MiddlePaleolithic contexts has been questioned in the past, theoccurrence of ochre in these contexts and its use by hom-inids of this period are now uncontested (Chase and Dib-ble 1987, McBrearty and Brooks 2000, Mellars 1996 andreferences therein). What is still controversial is the be-haviors that these remains represent. Two types of ex-planations that have been offered agree that ochre and/or the products of its processing played a role in theadaptive strategies of a site’s occupants but differ in thefunctions they assign to it. One type emphasizes the

practical applications of ochre and/or of its processingproducts. If such explanations were substantiated by ar-chaeological data and accepted, it would be impossibleto argue for the symbolic use of ochre on the basis of thematerial evidence alone. The second type of explanationis that the use of ochre and ochre products evolved andlater occurred repeatedly in symbolic contexts. To sup-port this argument, one would need to demonstrate thatochre was positioned within an organized system ofother symbols and thus had referential powers and thecapacity to generate symbolic predictions among the par-ticipants in the symbolic network. This proposition can-not be tested in most of the pre–Upper Paleolithic sitescontaining ochre. Fortunately, the context of the ochrein Qafzeh Cave provides such an opportunity.

Practical uses. Unless shown otherwise, practical uses

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 16: An Early Case of Color Symbolism: Ochre Use by Modern

hovers et al . An Early Case of Color Symbolism F 505

Fig. 5. Lump of ochre (#6) showing a deep, widegroove and shallower striations subparallel to oneanother.

Fig. 6. Another face of the same piece, perpendicularto the first, showing concave surface apparently pro-duced by heavy scraping.

provide the more parsimonious of the two types of ex-planation for the presence of ochre in early sites. Variouswriters have suggested the use of ochre as a technologicalaid in hide tanning (e.g., Bueller 1988), in hafting (Cour-aud 1988:28; Inizian 1976), as a painting and sealing sub-stance in domestic contexts (Mellars 1996:370), and formedicinal purposes (see Erlandson, Robertson, and Des-cantes 1999, Sagona 1994 for review and ethnographicexamples, Velo 1984).

Hafting as an explanation for the ochre record ofQafzeh Cave runs into difficulties. First, the locations ofochre residues on the surfaces of the artifacts are notcongruent with the use of ochre in the process of toolhafting. Even where the colored areas occur on the prox-imal ends and lateral edges of the flakes, their borderson the dorsal and ventral faces do not mirror one anotheras would be the case where lithic artifacts were hafted(Beyries 1983, Inizian 1976). Also, for the hypothesis ofhafting to survive scrutiny, it would have to explain theabsence of ochre in the upper layers. One possible suchexplanation is that the observed interassemblage typo-logical variability reflects differences in frequencies andprocesses of hafting. This explanation, however, is noteasily reconciled with the record of the site. Althoughthe lower layers are relatively rich in notches/denticu-lates, the frequencies of these tools among the pieceswith visible signs of ochre are low (table 5). Additionaltypological differences between the lithic assemblages ofthe lower and the upper layers are much subtler. These

minor differences, combined with technological homo-geneity (Hovers 1997), do not furnish a satisfactory ex-planation for the alleged presence of hafting in one butnot the other cluster of lithic assemblages. Finally, it isnow evident that hafting in the Middle Paleolithic wasnot restricted to specific types of lithic artifacts or, in-deed, even to retouched pieces (Boeda, Connan, and Mu-hesen 1998; Friedman et al. 1994–95; Shea 1989, 1997).Given that, the existence of typological differences perse is hardly a sufficient explanation of the dichotomousdistribution of ochre seen in Qafzeh Cave.

Hide tanning, using ochre as a preservative, is oftenmentioned as a possible utilitarian use of ochre. Severalcharacteristics of the ochre record in Qafzeh Cave areinconsistent with this hypothesis. First, the preservativeproperties of ochre are not color-dependent (Watts 1999:121), and therefore the tanning hypothesis does not offerany explanation for the observed repeated, focused se-lection for red hues of ochre.

Another difficulty with the tanning hypothesis is itsinherent emphasis on the exploitation of faunal re-sources. Studies of lithic use-wear and typo-technologysuggest that exploitation of faunal resources may nothave been a prominent activity throughout the durationof human occupation of Qafzeh Cave (Hovers 1997, Shea1991). Taking the frequencies of mammal bones as aproxy for the relative extent of animal exploitationwithin the sequence, the upper layers, which showhigher absolute frequencies of large mammal bones (Ra-binovich and Tchernov 1995), would be expected to have

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 17: An Early Case of Color Symbolism: Ochre Use by Modern

506 F current anthropology Volume 44, Number 4, August–October 2003

table 5Ochre-Stained Artifacts in Qafzeh Cave Terrace

Layer andNumber Description

Location ofOchre Stain

XXII, B15-77 Kombewa flake On one of twobulbs ofpercussion

XXII, C14-60 Notch on a brokenLevallois flake

On dorsal face,distal

XX, C12-658 Eclat debordant On ventral face,proximal rightedge

XIX Levallois core On flaking surface,especially in thenegative of a largeflake

XVIIB12-575 A double tool—side

scraper andmassive denticulate

On dorsal face

C15-58 End scraper On dorsal faceC11-672 Core-trimming

elementOn distal tip of the

flake, previouslythe core’s strik-ing platform

A11-336 Flake On dorsal faceB12-545 Levallois flake On butt and along

one edge56 Flake On ventral faceC10-280 Primary flake Whole flake

smeared withochre

Fig. 7. Ochre-stained lithic artifacts (stained areasshown as dotted). 1 and 2, from layer XVII; 3, fromlayer XXI.

more ochre in them if ochre were associated with hidetanning and/or other hide-working-related activities(e.g., Philibert 1994). The record from Qafzeh is in agree-ment with other studies that have questioned the use ofochre in tanning activities on grounds of social theoryand ethnographic and biological evidence (Watts 1999:121 and references therein).

Symbolic use? From the formal positivist standpoint,hypotheses focusing on symbolic behaviors are notori-ously difficult to test and refute. Symbols are abstract,society-specific constructs. It is only within the socialnetworks that constructed them in the first place thatsymbols can be deconstructed into communicated in-formation (e.g., Chase 1991, Conkey 1978, Deacon 1997,Gage 1999, Wobst 1977). As a result, ethnographers andethnologists, working with extant societies, rely on thehelp of informants to understand discrete symbolic be-haviors (and their material expressions) and their placeas components within whole symbolic systems (e.g., Ba-likci 1970, Barley 1983, Spencer and Gillen 1968 [1898]).Archaeologists are obviously at a disadvantage given thelimitations of their subject materials. In the particularcase of ochre there is also the possibility that its symbolicuse may have entailed activities that are not observablein the archaeological record, such as mimicry and bodypainting (e.g., Bordes 1952, Clark 1988, Knight, Power,and Watts 1995, Power and Aiello 1997, Watts 1999).

The combined outcome of such constraints is that thenotion of ochre as part of a symbolic system is fraught

with analytical difficulties and involves evaluationrather than direct testing of specific hypotheses. Thecomplex, hierarchical nature of symbol systems renderssuch an evaluation a matter of analyzing patterns of as-sociations and dissociations (Foster 1990). An importantcaveat is that such an endeavor is acceptable only afterother, testable explanations have failed to fit the ar-

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 18: An Early Case of Color Symbolism: Ochre Use by Modern

hovers et al . An Early Case of Color Symbolism F 507

Fig. 8. Levallois core heavily stained with ochre.

chaeological evidence. Having shown that practical hy-potheses fail to account for the spatio-temporal patternsof the ochre record at Qafzeh Cave, we now explore thehypothesis that the ochre occurred in the context of asymbolic color system.

The data at hand indicate that at Qafzeh Cave a singleoperational sequence of ochre selection and acquisitionpersisted through time. The analysis documents repeti-tive targeting of particular geological localities on thelandscape as a matter of informed choice based on un-wavering selection criteria. Specific hues of red appearto have been the sought-after property of the ochre ob-tained in the particular localities. Lumps of ochre werebrought into the cave and processed there. The acqui-sition and transport behaviors are inconsistent withstraightforward least-cost economic principles.

The available radiometric dates for the ochre-bearinglayers XXIV–XVII suggest a time span of ca. 10,000 yearsfor their accumulation (Valladas et al. 1988). This maybe an overestimate, as sedimentological evidence sug-gests rapid sediment accumulation (Farrand 1979). Still,the stretch of time encapsulated by the sequence ofochre-bearing layers likely encompasses some thousandsof years (e.g., Bar-Yosef 1998). The overall pattern seenat Qafzeh Cave is that of a robust and redundant systemof decision making guided by nonpractical criteria anddirected toward nonpractical goals. The redundancy ofthe selection parameters, combined with the absence ofindications of the practical use of ochre at the site, sug-gests that the phenomenon of ochre is to be interpretedas a persistent tradition handed down through the gen-

erations of the use of the color red as an index for objects,ideas, or events.

Ochre and burials. Ochre and human burials (as wellas other human remains that are not burials) occur onlyin the lower layers of Qafzeh Cave, sometimes associatedwith marine mollusks. The majority of ochre lumps tendto be clustered in a particular part of the excavated areaof the terrace (fig. 1). The occurrence of large lumps andsmall debris of ochre (figs. 2–4), a pattern present in allthe ochre-bearing layers, hints that this may have beena work area where ochre was processed, perhaps to beemployed in activities related to burials. Interestingly,in layer XVII, which yielded the largest number of in-tentional burials,3 the frequencies of large, heavy piecesof ochre are higher than in any of the other layers. Alsoin this layer there is an intriguing spatial associationamong the burial of hominid 8, an intensively scrapedpiece of ochre (#6; Vandermeersch 1969), and an engravedlithic artifact (Hovers, Vandermeersch, and Bar-Yosef1997), all found less than a meter from one another insquare B16.

In addition to the presence of ochre and human re-mains, the lower layers of Qafzeh Cave are distinguishedfrom the upper ones by the occurrence of hearths (Van-dermeersch 1981, Bar-Yosef and Vandermeersch 1993,Hovers 1997). Human occupations in the upper layerswere as intensive as the occupations in the lower layers

3. Five burials were identified in this layer. One of these, Qafzeh13, was initially reported as originating in layer XVa, but reex-amination of stratigraphic sections and elevations in 1997 indicatedthat its more probable source was layer XVII.

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 19: An Early Case of Color Symbolism: Ochre Use by Modern

508 F current anthropology Volume 44, Number 4, August–October 2003

Fig. 9. Location of ochre sources mentioned in text.1, Qafzeh Cave; 2, Mt. Devora; 3, Mt. Tavor; 4, ‘Ein elAssad; 5, Qiryat Shemona; 6, Wadi Malikh.

and often more so (Hovers 1997). Still, evidence for theuse of fire is considerably weaker than in the lower ones;visibly burnt sediments and burnt flints (either inten-tional or by-products of the use of fire) are rare. It maybe that at the time of the upper layers (VIII–III) the terracewas only peripheral to the main occupation inside thecave, but it should be noted that the sediments haveundergone in situ erosion. Because lithics are less vul-nerable to the vagaries of diagenetic processes than ashysediments or bones, the difference in their relative fre-quencies between the two stratigraphic blocks cannot beattributed solely to the effects of differential taphonomicprocesses. Whatever the reason, in the excavated areasof the cave’s entrance the dichotomy between the upperand lower layers reflects a real behavioral difference.

The co-occurrence of burials (and hominid remains ingeneral), intensive use of fire, ochre, and some inediblemarine mollusks and their temporal clustering are strik-ing. None of these types of finds were documented inthe upper layers. Ochre processing did not fully overlapwith the stratigraphic and spatial occurrence of the bur-ials (the distribution of ochre was more clustered thanthat of the human remains in the excavated areas, andsome ochre was found in layers that did not bear hominidremains [e.g., layers XXIV, XX]). Still, ochre utilization

appears to have been part of a structured ensemble ofbehaviors. But what types of behaviors were they?

The answer to this question is more ambiguous in thecases of the mollusks and the hearths. For the former,in the absence of a detailed publication little discussionis possible. It is clear, however, that their occurrence wasnot associated with dietary habits, as they are of inediblespecies and occur in small numbers (Bar-Yosef and Van-dermeersch 1993). By the same token, the possibility thatthe hearths found in lower layers were used in everydaypractical contexts cannot be ruled out. In this case, thedifference in the magnitude of the phenomenon betweenthe two stratigraphic blocks of the Qafzeh sequence iscircumstantially suggestive of fire use in contexts thatwere not strictly mundane during the time of depositionof the lower layers. For these two types of finds, thereare no compelling indications that the phenomena assuch were symbolic in nature.

The presence of human burials exclusively in the samestratigraphic block as the ochre is more telling. Despitearguments to the contrary (Gargett 1999 and referencestherein), a growing number of researchers now subscribeto the view that intentional burial existed in the MiddlePaleolithic (Belfer-Cohen and Hovers 1992, Hovers, Kim-bel, and Rak 2000, Hovers et al. 1995, Mellars 1996, Riel-Salvatore and Clark 2001, Tillier 1990). If so, such burialsare, because of their intentionality, symbolic. Mellars(1996:381, emphasis added) has suggested that “at thevery least we must assume that the act of deliberateburial implies the existence of some kind of strong socialor emotional bonds [within Neandertal societies], whichdictated that the remains of relatives or other close kinshould be carefully protected and perhaps preserved insome way after death.” Even in this cautious, minimalistview a symbolic component is implied, for the bones ofdead kin are at least iconic of the living person in thatthey point to their referent by physical resemblance. Inthe particular case of Qafzeh Cave, the presence of aburial gift in one instance (the deer antlers with Homo11 [Vandermeersch 1970]) and a double burial of an adultfemale and an infant (Homo 9 and 10 [Vandermeersch1981]) suggest the existence of referential associations ofa higher order. We interpret the occurrence of ochre withburials as a symbolic system in that we observe indexicalrelations between two sets of indices, with perceptuallybased representations being juxtaposed to create new ab-stract concepts (Goldstone and Barsalou 1998:167).

The process of learning to use a complex abstract sys-tem (symbols) instead of concrete, iconic associations asa normative referential framework is cognitively a dif-ficult one. Deacon (1997) has hypothesized that master-ing the ability to use symbols as a communications sys-tem required a two-stage process: hominids “were forcedto learn [construct] a set of associations between signsand objects, repeat them over and over, and eventuallyunlearn [deconstruct] the concrete association in favorof a more abstract one” (p. 402). Ritual played a majorrole in this process because by definition it entails re-petitive associations between particular sets of actionsand particular sets of objects and thus aids the transition

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 20: An Early Case of Color Symbolism: Ochre Use by Modern

hovers et al . An Early Case of Color Symbolism F 509

from concrete to abstract associations (Deacon 1997:401–10). The redundant co-occurrence through time ofochre and burials in Qafzeh Cave is consistent with thepatterns predicted where symbolic referential relationsare being shaped.4

Possibly, the intensive use of hearths in the lower lay-ers of Qafzeh Cave is to be understood in this context.While there is no evidence that fire per se had symbolicmeaning, the redundant presence of intensive heatingwith the ochre-burial association may be understood inthis broader context of symbolic behavior. The dichot-omous occurrence of this “package” of behaviors in thestratigraphic sequence may reflect changes through timein site function (Bar-Yosef and Vandermeersch 1993,Hovers 1997). In this sense, the record of Qafzeh Cavedepicts the shift from a structured symbolic system inthe making to one that is established and requires lessintentional investment (Hovers 1997).

implications for the evolution of symbolicsystems

The human cognitive capacity for creating symbols maygo back a long time. In his discussion of the most com-plex and variable symbolic behavior, Deacon (1997) ar-gues that the processes that have shaped modern lan-guage have a long history and involve multiple levels ofcausality and constraints—genetic, neurological, oral-vo-cal, social, and semiotic. According to this view, changesin brain architecture (specifically, the expansion of theprefrontal cortex) at the transition from Australopithe-cus to Homo forms the core adaptation for language. Theco-occurrence of large brains, stone tools, reduction indentition, and changes in hand morphology alreadymarks a socio-ecological environment that required sym-bolic solutions (Deacon 1997:348). Over time, the humanbrain and the capacity for symbolism evolved in tandem.

In accordance with the claim for early beginnings ofthe gradual coevolution of the brain and symbolism, thearchaeological record of the Lower and Middle Paleo-lithic contains potential manifestations, albeit meager,of symbolic capacities. Red and black pigments are rel-atively abundant among these rare manifestations, andtheir presence is often mentioned in support of the viewthat symbolism is expressed in the early Paleolithic rec-ord (Bednarik 1992b; Bordes 1952; d’Errico and Soressi2002; Marshack 1981, 1989; McBrearty and Brooks 2000;Wreschner 1983). Given the neuropsychological princi-ples underlying human color perception and categori-zation and their postulated antiquity, evidence for theuse of red and black pigments, let alone their mere oc-currence in early prehistoric sites, may indicate the pres-ence of cognitive capacities for categorization and sym-bolic thought. However, it does not in itself constitute

4. Deacon discusses early hominids, presumably earlier than thoseof Qafzeh Cave. This seems to be irrelevant to the point that weare making here. The process of unlearning associations betweensigns and discrete objects in favor of abstract ones would have hadto be repeated whenever a new symbolic framework was beingstructured.

evidence of symbol use in the context of a cultural net-work. The question of interest for this study has there-fore been not whether the Middle Paleolithic occupantsof Qafzeh Cave were capable of symbolic behavior butwhether the ochre finds from the site indicate that suchcapacities had become a system of symbolic cul-ture—that is, normative social constructs. Having es-tablished that such was the case, we now explore brieflythe implications of this finding for the explanation of theemergence of such systems. The issue is of particularinterest because it has recently been suggested that redochre was at the core of the emergence of the earliestappearance of symbolic culture as defined here (Knight,Power, and Watts 1995, Power and Aiello 1997, Watts1999).

On the basis of the abundant ochre record from Africa,whose beginnings can be traced to the late Middle Pleis-tocene, Knight, Power, and Watts (1995) and Watts (1999)have postulated that symbolic culture as a systematicbehavior emerged within an African Middle Stone Agepopulation of modern humans. Their line of reasoningis characteristic of the “symbols as tokens” approach inthat it views symbols as information carriers and placestheir origins in rational behavior designed to satisfy ma-terial needs (Robb 1998). According to the suggestedmodel, the use of ochre as the first cultural symbolicconstruct emerged as a sociobiological response to thereproductive stress experienced by females during thephase of encephalization associated with archaic H. sap-iens. Menstruation was taken to indicate fertility, andochre was used to fake menstruation in nonmenstruatingfemales (“sham menstruation”), thus ensuring males’ at-traction to and support for a large number of females.For the fraud to be effective, the use of ochre had to beanchored in elaborate social dynamics (Knight, Power,and Watts 1995, Power and Aiello 1997). Put succinctly,Knight, Power and Watts (1995) see the first symbolicculture as the set of all such deceptive signals. Ochrewas recognized as the signal for menstruation and thuscame to stand for a biological phenomenon to which ithad no relation in real life. They argue that the signal ofmenstruation, appropriated from an individual by a col-lective, communicated for the first time a “symbolic”construct, and because the symbol as such came intobeing only within a social context they view the use ofochre as the beginning of symbolic culture. Knight,Power, and Watts (1995) and Watts (1999) see clear in-dications in the African archaeological record that thebeginnings of the postulated symbolic culture long an-tedate the production of representational imagery on in-animate surfaces (i.e., Upper Paleolithic rock painting).Watts (1999) accepts that the Middle Stone Age does notshare the elaborate symbolic culture evident in the LateStone Age (and the Upper Paleolithic in Europe) butmaintains that all the essential elements appear to havebeen in place by, or shortly before, the Last Intergla-cial—within the Middle Stone Age 2 and approximatelycoincident with the emergence of anatomically modernhumans. Significantly, he notes that the frequency and

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 21: An Early Case of Color Symbolism: Ochre Use by Modern

510 F current anthropology Volume 44, Number 4, August–October 2003

intensity of the use of ochre increase gradually over timefrom the Middle Stone Age onwards.

A recent review (McBrearty and Brooks 2000) providesample evidence to support the argument for early begin-nings of ochre use, and new finds (e.g., Henshilwood etal. 2002) furnish evidence that it was used symbolically.However, the archaeological record as reviewed byMcBrearty and Brooks poorly supports the explanatorymodel. Corroboration for the model is recruited fromKhoisan ethnography, where menarchal rituals were doc-umented as the context in which red pigments were al-most invariably used (Knight, Power and Watts 1995;Watts 1999:134). This is problematic on several levels.Apart from the circular reasoning involved in suchclaims (and, in fact, unavoidable in ethnographic anal-ogies), the worldwide survey of color symbolism sum-marized above indicates that the association of red withmenstruation is not necessarily shared among all extantsocieties and is not the only referential association ofred.

At Qafzeh Cave, an inductive approach to the ochrerecord shows that a symbolic explanation is the mostparsimonious one for a number of phenomena observedin the archaeological record. The patterns of ochre util-ization and the associations of ochre with other types offinds present hitherto unprecedented archaeologicalcontextual support for the existence of symbolic cultureand offer support for the identification of ochre as anancient social symbolic construct. The same evidencesuggests that the ochre at Qafzeh was used in symboliccontexts that differ from the scenario envisioned byKnight, Power, and Watts (1995) and Watts (1999). Giventhe specificity of the social construct, we cannot offerany insights into the meanings of the symbol per se, butwe can venture a suggestion as to the general context ofsymbolic use of ochre. After the Middle Paleolithic,ochre was clearly associated with burials (Chase and Dib-ble 1987:280), and its use has been accepted by archae-ologists as part of symbolic mortuary behaviors. Thestrength of the contextual evidence in Qafzeh is suchthat it permits the suggestion that it is an early (possiblythe earliest) “prototype” of such behavior.

Unfortunately, Qafzeh Cave contains the only recordof the Levantine Middle Paleolithic from which hema-titic ochre has been reported. This record does not offerready explanations of the processes by which ochre be-came part of symbolic culture, nor are such evolutionaryinsights likely to be gained from the relatively short rec-ord of a unique locality. But accepting as we do the claimfor ochre’s being a social symbolic construct from thetime of the Middle Paleolithic, the comparison of therecord from Qafzeh Cave with the patterns seen in theAfrican record reveals interesting trends.

Symbolic culture seems to have existed by ca. 100,000years ago in both regions, but the referential frameworkof ochre may not have been identical in Africa and theLevant at this early time. Color symbolism further di-verged during the Upper Pleistocene. In Africa, the useof pigments has been continuous and overall has in-creased over time (McBrearty and Brooks 2000, Watts

1999 and references therein). Analysis of selection cri-teria indicates that the brightness and hue of color werethe sought-after properties and has led to the suggestionthat ochre was utilized mainly for body painting and/orfor the decoration of perishable organic artifacts (Wattsin Henshilwood et al. 2001, Watts 1999). A general con-clusion about the use of ochre in symbolic contexts and/or to enhance stylistic traits is warranted here, althoughthe contexts of symbolic use remain obscure.

An opposite trend can be observed in the Levant. Notonly is Qafzeh Cave the only Middle Paleolithic siteknown to date that contains ochre but there are no in-dications for systematic use of ochre and/or other pig-ments during the Late Levantine Mousterian (80,000–45,000 years ago) and only a few occurrences during theUpper and early Epi-Paleolithic periods (45,000–12,500years ago). When ochre is present in the latter contextsit occurs sporadically, usually in what appear to be util-itarian context (e.g., hafting of microliths and blades[Korn 2000]). Quantities of ochre increase only in thelate Epi-Paleolithic period, in the Natufian and the Har-ifian entities, dating to 12,700–10,500 years ago and10,500–9,500 years ago (uncalibrated dates), respectively(Korn 2000, Weinstein-Evron 1998, Zackheim 1997 andreferences therein). For the Natufian, ochre has beenfound in sites that contained burials. A relevant obser-vation is that from the Middle Paleolithic on the pre-Natufian record of the Levant was depauperate of sym-bolic manifestations of any type. It has been suggestedthat the stability and availability of resources in the Le-vant necessitated smaller territories and caused socialgatherings to be briefer and perhaps smaller in scale than,for example, in Europe. This in turn reduced the needfor elaborate symbolic expressions (Bar-Yosef 1997 andreferences therein, Hovers 1990). If the use of ochre in-deed started out as a symbolic construct in the MiddlePaleolithic, its near absence from the record is congruentwith this interpretation, suggesting that the social con-texts in which it was used were rare in the Levant formost of the Upper Pleistocene. By that time, symbolicframeworks in the Levant may have diversified and fo-cused on other, possibly nonmaterial signs (in contrastto the Middle Paleolithic, for which language is debat-able, its use cannot be easily rejected for the majority ofthe Upper Pleistocene). The social and demographic im-pact of the shift toward reduced mobility could have beena driving force in the return to color symbolism in thelate Epi-Palaeolithic.

The symbolic contexts of ochre use in the Levant andAfrica seem to have differed already during the MiddlePaleolithic/Middle Stone Age. Such diversification of thesymbolic color system supports the argument for its an-tiquity although not necessarily the sociobiological ex-planation suggested for it. It is the trajectories of theochre record in the two regions during post–Middle Pa-leolithic/Middle Stone Age time which speak of a rapiddiversification of the content of the symbols employedin cultural systems, breaking the evolutionary con-straints which may have driven its initial appearance.This later, independent development supports the iden-

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 22: An Early Case of Color Symbolism: Ochre Use by Modern

hovers et al . An Early Case of Color Symbolism F 511

tification of ochre utilization as an already deeply rootedsymbolic construct attaining different meanings and dif-ferent significance for the human groups that have usedit in varied circumstances, as is the case with symboliccolor systems in extant societies.

Comments

larry barhamCentre for Human Evolution Research, Department ofArchaeology, University of Bristol, Bristol BS8 1UU,U.K. ([email protected]). 18 iv 03

The search for life on Mars and the search for evidenceof symbol use before the Upper Palaeolithic have muchin common. Both arouse passionate debate, and bothphenomena would have profound evolutionary impli-cations should they exist. Where astronomers and ar-chaeologists differ is that astronomers know what tolook for and can devise unambiguous criteria for testingtheir theories. Archaeologists have a body of data—artefact styles, subsistence and burial practices, long-dis-tance exchange, modified objects, and pigment use,among others—that constitutes a common vocabularyfor discussing early symbol use, but they do not agreeon how or in what contexts symbol use might have oc-curred before the Aurignacian of western Europe. Theresult is extreme variation in the interpretation of thearchaeological record, with Klein (2000) arguing for de-velopment of symbol use 50,000 years ago and Bednarik(1992a, b) recognizing symbol-based behaviours in theLower Pleistocene.

Hovers et al. make an important contribution towardsthe development of a robust methodology for recognizingsymbol use in the absence of “art.” Associating them-selves with Deacon’s (1997) synthesis of the neurologicaland behavioural foundations of symbol use takes themto the heart of the problem—how to detect the materialsignatures of symbols when by definition symbols aresocial constructs that work in reference to other sym-bols. Fortunately for archaeologists, symbolic behaviouris embedded in the routines of daily life or habitual time(Gosden 1994), and through the repetitious structure ofrituals it may leave material traces. Chase and Dibble(1987) regard the redundancy built into symbol-basedsystems as an essential criterion for recognizing theirpresence in the past. The occasional engraved or other-wise modified object found isolated in time and space(Bednarik 1992a) would not constitute sufficient evi-dence for symbol use by their standard or that of Hoverset al., so perhaps we are moving towards agreement. Inthe case of pigment use, we should look for systematiccollection and processing of specific minerals, examinetheir properties as colorants, and consider the spatial,social, and potential functional contexts of their use be-fore deducing a symbolic role. This contextual approachhas been applied to the records of pigment use in Middle

Stone Age southern (Watts 2002) and south-central (Bar-ham 2002) Africa and is used to good effect in the caseof Qafzeh.

Watts (2002) links ethnographic accounts of pigmentuse in southern Africa with his interpretation of archae-ological pigments by referring to universal human cat-egories of colour perception deriving from our primateheritage. Hovers et al. use this same neuro-optical linkwhen analysing the selection of iron oxides by Qafzeh’soccupants. Acknowledging the various functional inter-pretations of the use of iron minerals in general and dis-missing each individually in this context, they make apersuasive case. Stripping out utilitarian options leadsto the interpretation that a colour-based symbolic sys-tem was operating 90,000 years ago at this site. There isconvincing geological evidence that red iron oxides wereselected in preference to yellow hydroxides from thesame source. Consideration of the broader context of pig-ment use supports their conclusion that these early mod-ern humans engaged in various complex behavioursbased on networks of symbols.

Research into pigment use has moved beyond cursorydescriptions of the presence of ochres to detailed geo-chemical, quantitative, and experiment-based analysesof source materials and their modification. This analyt-ical approach should become standard practice. There arecollections that would merit reexamination under mod-ern analytical techniques as in the cases of Qafzeh andPech de l’Aze (d’Errico and Soressi 2002). The detailedpresentation of the Qafzeh material is a valuable con-tribution to Palaeolithic archaeology. It may lead even-tually to methodological consensus, but I suspect thatlife will be found on Mars first.

Whilst supportive of the methodology and conclu-sions, I feel that an opportunity was missed to placeQafzeh in the context of wider debate about an Africanorigin for behavioural modernity. Reference is made tothe African Middle Stone Age in the context of conti-nuity of pigment use, but otherwise the European andAfrican records are treated as a generalized pool of Mid-dle Palaeolithic behaviours, overlooking potentially sig-nificant differences. While red and black are commonlyused colours, black is predominantly a feature of theEuropean record (Mellars 1996), and pigment use is spo-radic among Neanderthals. In contrast, systematic pig-ment use associated with processing tools (grindingstones) appears about 300,000 years ago in south-central(Barham 2002) and eastern (McBrearty 2001) Africa. Itoccurs with the earliest Middle Stone Age industries ineach region and may reflect new behaviours associatedwith the Acheulean/Middle Stone Age transition. Insouth-central Africa, hunter-gatherer pigment use con-tinued to the historic present (Barham 2002). Such con-tinuity is a real difference between the African and Eu-ropean records which deserves greater recognition.Qafzeh, as a lone example in the Near East, could beconsidered an extension of the African pattern of pig-ment use that is linked to anatomically and behaviour-ally modern humans.

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 23: An Early Case of Color Symbolism: Ochre Use by Modern

512 F current anthropology Volume 44, Number 4, August–October 2003

anna belfer-cohenInstitute of Archaeology, Hebrew University ofJerusalem, Mt. Scopus, Jerusalem 91905, Israel([email protected]). 16 iv 03

At the risk of being accused of nepotism, I would ventureto say that this is one of the best-written, most coherentand balanced papers I have read on the issue of pigmentuse by prehistoric humans. It has it all: impeccably pre-sented archaeological data, a solid background of factsand figures from other scientific domains (e.g., neurol-ogy, linguistics, ethnography), and theories concerning“symbolism.” All of these are cross-referenced and dis-cussed at length to explain the spatio-temporal pattern-ing of ochre finds in Qafzeh Cave and to explore its im-plications for the evolution of symbolic systems.

I have but one comment or, better, cautionary note:In this paper as well as in others cited therein, one ob-serves a search for a mental “achievement,” mostly todo with the “claim for early beginnings of the gradualcoevolution of the brain and symbolism.” I may bewrong, but this emphasis seems somehow tied in withthe search for evidence of becoming “clever,” as in theparticular scenario offered by Knight, Power, and Watts(1995) and Watts (1999). Eloquently presented in boththe original publications and the paper at hand, it canbe summed up as females’ discovering the art of maledeception to promote their own biological interests, thatis, cheating on the enemy.

Self-evident as it may be, I think that in dealing withscenarios of a coevolution of the brain and symbolismone should make a point of referring to the realms of thehuman “soul,” given that not only the human cortex butthe human limbic system is unique (Belfer-Cohen andHovers 2002). The authors describe an exceptional ar-tifact, a large Levallois core, with thick, dense ochre res-idues in the deepest part of the large negative scar of thelast removal (fig. 6). They consider it “possible that thiscore was recycled into an ochre receptacle” and cite ref-erences to that effect. This may be true, but there is analternative explanation for the presence of the pigmentin the core’s deepest scar. In his overview of adze pro-duction among the inhabitants of the Langda plateau,New Guinea, Dietrich Stout writes: “Some of the deeperflake scars are usually left intact and may be paintedwith red and white pigments. These markings are bothdecorative and symbolically meaningful; Petrequin andPetrequin (1993) report on informants’ ‘giving life’ to theadze by putting ‘blood’ in its wounds, while my owninterviews suggest that pigmented adzes are reserved foruse by men only” (Stout 2002:700). A Middle PaleolithicLevallois core is not a present-day Langda adze, but oneshould remember the complex relationship humans havewith their culture (spiritual as well as material), pro-duction processes, and final products (e.g., Jones andWhite 1988). This is a cautionary note against taking theself-evident for granted or ignoring it, lest we find our-selves missing out on the “soul” while gaining on the“brain.”

richard g. kle inProgram in Human Biology, Stanford University,Stanford, Calif. 94305, U.S.A. ([email protected]).28 iii 03

Hovers et al. show that the people who occupied QafzehCave 90,000–100,000 years ago collected naturally oc-curring pigment. From the reconstructed pattern of pig-ment utilization and the association of pigment withother kinds of finds, they argue that the people employedpigment mainly for a nonutilitarian purpose such as body(possibly dead-body) painting as opposed to a more mun-dane purpose such as hide tanning. Others have sug-gested broadly similar nonutilitarian pigment manipu-lation by contemporaneous or somewhat more recentMiddle Stone Age people in southern Africa (Henshil-wood et al. 2002) and by Mousterian people in France(Bordes 1952, d’Errico 2003). Pigment manipulation maythus be added to other advanced behaviors, such as asophisticated ability to flake stone, control over fire, bur-ial of the dead, and the recurrent hunting of large mam-mals, that Mousterian and Middle Stone Age populationsshared with each other and with succeeding Upper Pa-leolithic and Later Stone Age people after 50,000 yearsago. Arguably Mousterian and Middle Stone Age peopleinherited most of these behaviors, including pigmentuse, from a shared Late Acheulean ancestor.

These researchers and others believe that nonutilitar-ian pigment manipulation implies “symbolic culture,”and this raises the question why Mousterian and MiddleStone Age people left little or no unambiguous evidencefor representational art, jewelry, burial rituals, or othersymbolic indicators that mark the Upper Paleolithic andthe Later Stone Age. One possibility is that only UpperPaleolithic and Later Stone Age people possessed the bi-ological (neural) capacity for full symbolic culture. An-other is that Mousterian and Middle Stone Age peoplehad the capacity but routinely expressed it only afterthey had crossed some as yet undetected technological,social, or demographic threshold about 50,000 years ago.

Archeology alone cannot eliminate either alternative,but human fossils help. These show that EuropeanMousterians were Neanderthals, Middle Stone Age Af-ricans were anatomically modern or near modern, andUpper Paleolithic/Later Stone Age people were the de-scendants of Middle Stone Age people who expandedfrom Africa to Eurasia beginning about 50,000 years ago.Their expansion appears to have been grounded in thedevelopment of fully modern (“symbolic”) culture, andits appearance only about 50,000 years ago would explainwhy anatomically modern or near-modern Africansfailed to spread to Eurasia before then. Archeology, hu-man fossils, and the genes of living people suggest thatthe spread occurred without significant cultural and ge-netic exchange between modern human invaders andNeanderthals or other nonmodern Eurasians. In this re-spect, it differed significantly from subsequent well-doc-umented prehistoric and historic population expansions,in which invaders and indigenes usually engaged in sub-stantial cultural and genetic exchange. A plausible ex-

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 24: An Early Case of Color Symbolism: Ochre Use by Modern

hovers et al . An Early Case of Color Symbolism F 513

planation for the difference is that the Neanderthals andtheir nonmodern Eurasian contemporaries were biolog-ically (neurologically) incapable of adopting the more so-phisticated culture of the invaders, while the invaderssaw no advantage in the more limited culture of non-modern people.

Recent research suggests that it may be possible toisolate genes that bear on modern cognition or com-munication (Enard et al. 2002), and future research mayshow that a particularly crucial gene achieved its presentform about the time that modern humans spread fromAfrica. In this event, it would be reasonable to infer thatgenetic change underlay the development of full sym-bolic culture. Nonutilitarian pigment manipulation byearlier people would then mean only that they were cog-nitively advanced in the direction of modern humans,not that they were cognitively the same.

chris knight, camilla power, and ianwattsDepartment of Anthropology, University of EastLondon, Barking Campus, Longbridge Rd., Dagenham,Essex RM8 2AS, England ([email protected]) (Knight,Power)/University of Cape Town, Cape Town, SouthAfrica (Watts). 22 iv 03

Hovers et al. convincingly argue that the Qafzeh ochrerecord belongs to an organized system of symbols. Thiscorroborates a growing body of research results (mostrecently Barham 2002, Watts 2002) implicating habitualred ochre use in the emergence of symbolic culture iden-tified with African modern humans between 100,000 and200,000 years ago rather than with the Eurasian Middle/Upper Palaeolithic transition. Their broad theoretical ap-proach is refreshing. We welcome their engagement withour explanation for the ochre record (Knight, Power, andWatts 1995) but query the reductionism apparent inmuch of their approach. Can the “basic colour triad”described by Turner among the Ndembu plausibly bemapped to ancestral primate trichromacy? The Ndembutriad includes black and white, which do not involvemechanisms of colour perception at all (De Valois andDe Valois 1993, Abramov 1997), and certain cultureshave only two terms—not “black” versus “white,” asBerlin and Kay (1969) originally proposed, but composite“light/warm” versus “dark/cool” (D’Andrade 1995, Kayet al. 1997, Heider 1972). Had Berlin and Kay’s originalformulation been correct, we might have expected black(and white) pigments to predate red ones. Colour refer-ence, however, has evolved in adaptation to both thehuman nervous system (with its biases) and the prag-matic constraints of habitual use (Deacon 1997, Watts2002).

Hovers et al. acknowledge ritual’s role in installingsymbolic representations in human minds. Key tohunter-gatherer symbolic inheritance is initiation. Thisis always in some sense a “death” followed by “rebirth”(Van Gennep 1960 [1908]), making it misleading to assignmortuary symbolism to a separate and unrelated cate-

gory. For females, initiation coincides roughly with firstmenstruation, but regardless of sex, artificially amplified“blood” signals (e.g., circumcision, subincision, cicatri-zation, nose-bleeding) recurrently connote concepts var-iously translated as “temporary death,” “resurrection,”“fertility,” “ritual inviolability,” etc. (Knight 1991).

Hovers et al. proceed as if language as an abstract en-tity could be detached from the study of materially em-bodied symbols. But Darwinian theory tells us that forsignals to be effective they must be reliable and to bereliable they must be costly (Zahavi and Zahavi 1997).Words are cheap, hence ineffective in signalling com-mitment. An additional framework is required to estab-lish the reliability of verbal signals, and we argue thatritual—in proving commitment to in-group honest com-munication—serves this purpose (Knight 1998, 1999,2000, 2002; Power 1998, 2000). Hovers et al. demonstratethat the procurement of the Qafzeh pigments entailedgreater investment in time and energy than would bepredicted on strict grounds of cost-efficiency. Blood iscostly, and so is high-quality ochre. It is precisely thiscostliness that proves that ritual participants “mean it.”

Archaeology necessarily draws on ethnographic anal-ogy. The critical point is how tightly constrained this is.Our point of departure was not ethnographic observationof pigment use. We developed a purely theoretical modelthat led us to specific predictions testable against thefossil, archaeological, and ethnographic records. Ourmodel predicted certain cross-culturally invari-ant—“time-resistant”—syntactical features of magico-religious tradition. We chose to test it against Khoisancosmology because of the geographical focus of our re-search, the time depth of Khoisan genetic lineages, andthe cultural continuities between recent Khoisan andLater Stone Age traditions. It was on Darwinian groundsthat we focused on menstruation, but ethnography pro-vides cross-cultural evidence of ritually constructed ta-boos on symbolically “menstruating” initiates. We ar-gued that the symbolic domain was born as suchprohibitions and rituals became regularly established.Hovers et al. observe that menstruation “is not the onlyreferential association of red.” Obviously not. But wenever suggested that ochre must necessarily “refer” to“menstrual blood.” Ritual is not referential but perfor-mative—it constitutes its own “truth” (Rappaport 1999).In our model, ritually displayed red cosmetics invoke andsustain collective representations translatable not nar-rowly as “menstruation” but more broadly as “fertility,”“supernatural potency,” etc. We made predictions aboutthe form of such metarepresentations (see esp. Power andWatts 1999) and about when in the archaeological recordwe should first expect pigment use, when we might seea shift from irregular usage, and which colours shouldbe selected.

A manganese/burial co-association at Qafzeh wouldhave falsified our model. Again, with the hypothesized“sham menstruation” strategy being driven by the re-productive costs of encephalization between 500,000 and150,000 years b.p., the model would be falsified if Homoerectus/ergaster used pigment or, alternatively, if such a

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 25: An Early Case of Color Symbolism: Ochre Use by Modern

514 F current anthropology Volume 44, Number 4, August–October 2003

ritual tradition arose only after brain volumes in modernhumans had already stabilized. There is no compellingevidence for pigment use predating the middle of theMiddle Pleistocene (Watts 1999, Barham 2002), and thereare no reports of black pigment predating the Late Pleis-tocene. Watts (1999) reported a dramatic increase in redochre use in the southern African Middle Stone Agerather than a “gradual” increase “from the Middle StoneAge onwards.” This increase was tentatively placed inthe early Late Pleistocene; revised dates for the BorderCave sequence (Grun and Beaumont 2001) may push thisback to the terminal Middle Pleistocene, while Barham’s(2002) and McBrearty’s (1999) research suggests that reg-ular use may be earlier still in the African Tropics.Claims of ubiquity, particularly those made by Bednarik(1992a), need to be carefully evaluated.

The temporal contrasts that Hovers et al. draw be-tween the Levantine and African ochre records are in-teresting, but we don’t know how significant a presencemodern humans had in the Levant during the Late Mous-terian. Furthermore, an absence of ochre need not implya lack of interest in red pigments—in the Kalahari, scar-city obliged Khoisan to employ plant substitutes (Watts1999:133), and henna is widely used in the Near East.Hovers et al. argue that the symbolic contexts of ochreuse in the Levant and Africa differed as early as the Mid-dle Palaeolithic/Middle Stone Age, but all that we canbe sure about is that, in contrast to their Levantine coun-terparts, Middle Stone Age people did not bury their deadin caves.

Finally, where symbolic culture exists, local symbolicmeanings will vary. What strikes us is that, regardlessof variability on this level, structural features of ritualshow extreme conservatism, red pigments being used togenerate multiple meanings. It is on this syntactical levelthat our model can account for such recurrent featuresas red-ochre burials in the global archaeological recordof modern humans in the Late Pleistocene.

sally mc breartyDepartment of Anthropology, University ofConnecticut, Storrs, Conn. 06269, U.S.A. ([email protected]). 17 iv 03

Much of the debate concerning the origin of modern hu-man behavior stems from disagreements about what con-stitutes symbolic behavior and what standards of proofare required to establish it. I am pleased to see a paperthat situates this discussion firmly in the material world,independent of unverifiable speculation about languageor the nature of the symbolic content itself. The methodemployed here, treating archaeological remains as to-kens, may ultimately prove to be as important to un-derstanding Paleolithic symbolic behavior as it has beento clarifying the origins of writing (Schmandt-Besserat1996).

Hovers and her coauthors provide solid empirical ev-idence for the collection and manipulation of ochre inthe Levantine Middle Paleolithic. The archaeological,

lithological, chemical, and mineralogical evidence pro-vided here supports the conclusion that the ochre wasselected for its color properties, that it was transportedto the site, that coloring matter was obtained by scrapingit with stone tools, and, intriguingly, that it was perhapsfurther processed in the concavity of a Levallois core. Ithink that the case for ochre, fire, and burials functioningtogether as a symbolic system is less firmly established.The materials do not seem to be in direct archaeologicalassociation, and the debris in which they are found isestimated to have accumulated over a period of perhaps10,000 years.

It would be very interesting to compare the behaviorreflected in the lower, ochre-bearing levels at Qafzeh,presumably the product of anatomically modern Homosapiens, with that in the upper layers, apparently accu-mulated by Neanderthals. Lieberman and Shea (1994)have outlined a model for the Middle Paleolithic of theLevant which contrasts the foraging behavior and rang-ing strategies of modern humans with those of Nean-derthals. Does the Qafzeh evidence support or refutetheir ideas?

The linguistic evidence for the universality of red inhuman color classifications (Berlin and Kay 1969) isstrong circumstantial evidence for the very great antiqu-ity of the color red as a symbolic category. While I happento believe that the use of ochre is in itself evidence forsymbolic behavior, I doubt that this paper will convincethose who do not share my view, and no doubt utilitarianuses for the ochre will be invoked.

One of the most difficult tasks for the prehistorian isto identify the emergence of novel behaviors. The chal-lenge is similar to that encountered in recognizing theorigin of biological species. The earliest members of anew species will be few and will probably resemble notonly their ancestral but also their sibling species. Traitsin a new species may be hypervariable until they arecanalized by adaptation, sexual selection, or other weed-ing-out processes. Similarly, it may be problematic torecognize innovations in the archaeological record priorto their becoming stereotypical or “normal” behavior forthe society.

Populations of both Neanderthals and early H. sapienswere geographically widespread, and one would not ex-pect the behavior of either of them to have been uniformthroughout their range. Specific case studies such as theone provided here allow a richer view of the behavior ofH. sapiens in the Levant of 90,000 years ago, and it willbe interesting to compare it with cases of pigment useelsewhere at a similar time depth.

alexander marshackPeabody Museum of Archaeology and Ethnology,Harvard University, Cambridge, Mass. 02138, U.S.A.([email protected]). 16 iv 03

Hovers et al. provide important archaeological and ge-ological evidence for the use of ochre during the lateMiddle Paleolithic at Qafzeh Cave. Their paper is a sig-

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 26: An Early Case of Color Symbolism: Ochre Use by Modern

hovers et al . An Early Case of Color Symbolism F 515

Fig. 1. Pure white chalk plaquette (11.5 # 8 cm),abraded to produce a white powder and then used asa crayon, probably for ritual marking of a body or ahide. Magdalenian IV, Bedeilhac (Ariege).

nificant addition to the recent discussion of the MiddlePaleolithic use of ochre at Blombos Cave (Henshilwoodet al. 2001, 2002). The far earlier Middle Paleolithic caveof Becov in Middle Europe, ca. 250,000 b.p., produced apiece of intentionally striated ochre, an abraded quartziterubbing stone, and a huge quantity of ochre powder dis-persed in a circle around the stone on which the abraderhad sat (Marshack 1981). The feet of the abrader wereoutlined on the ground as ochre-free, negative footprints.Particularly important was the fact that a piece of ochre,a rubbing stone, and apparently a container had beenbrought from different places to this particular cave at aparticular time for a particular purpose—processes sim-ilar to those inferrable for the use of ochre at Qafzeh andBlombos. The production of ochre at Becov, while ar-cheologically and quantitatively unique for the period(see Chase and Dibble 1987, Chase 1991), was clearly alearned, cultural and social behavior under frontal-lobemediation and probably therefore, at some level, “sym-bolic.”

There was no evidence that ochre had been producedelsewhere in Becov or anywhere else in Middle Europeduring this period, and Chase and Dibble would considerthis quantitative singularity a statistical argumentagainst Middle Paleolithic “culture” or the cultural useof ochre. There may, however, be a need for caution indrawing such an inference. The Middle Paleolithichunter-gatherers at Becov were seasonally mobile. So-cial, cultural, and ritual events would therefore havebeen periodic and place-and-time-specific, as in a re-gional aggregation, or aperiodic, as in the death of anelder, a birth, the failure of a resource, or a sudden epi-demic. The lower levels at Qafzeh represented ephemeralseasonal camps (Bar-Yosef 1994:43). The production ofochre in these early cultures may thus always have beentime-, context-, and place-specific and so not alwaysavailable archeologically. Besides, the major rituals orburials of mobile hunter-gatherers are often conductednot within the habitation cave or on its terrace but in aculturally specified ritual place (Marshack 2001). It islikely that early ochre was often prepared and used else-where and that its absence from a particular cave wouldnot denote its absence from the culture.

There is another problem. Hovers et al. note that oneof the trichromatic colors is white. White is an importantcolor in body painting and other modes of surface dec-oration. In the Franco-Cantabrian caves, white was ef-fectively used within images by leaving areas of lime-stone wall blank or by scraping a coating of surface clayor dust to create an area of white. The bichrome “Chi-nese horse” in Lascaux has an unpainted white under-belly that is indexical of the dark creamy tan of its sum-mer coat. Could white have also been used as an appliedcolor, say, in body painting, during the Levantine Pale-olithic or the European Upper Paleolithic and not be ev-ident archeologically?

In the Middle Magdalenian decorated cave of Bedeilhac(Ariege) a large number of flat, irregularly shaped clayplaquettes had been peeled and lifted from the groundand incised with images. One hand-sized pure-white clay

or “chalk” plaquette (11.5 # 8 cm (Field Museum #212709) differed from the others in that while the rearretained its original sharp edge of breakage the front halfhad been lightly abraded and thinned to produce a whitepowder, shaped to a point, and, as the edge polish re-vealed, used as a pencil or crayon to mark a smoothsurface—probably a skin garment or a human body in aritual (fig. 1). After use for its color it had been symbol-ically incised (see Henshilwood et al. 2002 for theovermarking of ochre at Blombos Cave).

These rare data pose a number of interpretive prob-lems. Red and red ochre have profound affective and sym-bolic resonance, but white, through less common ar-cheologically, has its own powerful semantic: in bones,skulls, and skeletons. White is common as body paintin shamanic and popular ritual and, like black, is sym-bolic in many cultures. However, white paints, whenderived from a mixture of water and clay or chalk, wouldprobably not have survived archeologically as well asochre, manganese, or charcoal. White would surely havebeen effective in human perception and the functionalcolor palette. Have archeologists investigated the sourcesand possible uses of Middle and Late Paleolithic claysand chalk as a possible pigment?

Hovers et al.’s paper has made it possible to address anumber of issues. Reference to human neurology in anattempt to understand the human productive, cultural,and symboling capacity and its relation to the archeo-logical record has taken many routes, but, as far as Iknow, this is the first paper to discuss trichromatic hu-

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 27: An Early Case of Color Symbolism: Ochre Use by Modern

516 F current anthropology Volume 44, Number 4, August–October 2003

man perception as an aspect of early, pre-Upper Paleo-lithic symboling behavior.

antonio sagonaSchool of Fine Arts, Classical Studies andArchaeology, University of Melbourne, Melbourne,Victoria 3010, Australia ([email protected]).16 iv 03

Hovers et al. are to be congratulated for advancing ourunderstanding of prehistoric colour systems in a majorway. Their detailed analysis of ochre procurement andusage by the Middle Palaeolithic occupants of QafzehCave is as persuasive as it is thought-provoking. Theyargue that layers XXIV–XVII, the ochre-bearing depositsat Qafzeh, represent a unique episode in the history ofthe site, specific to the conditions of the time, in whichthe hominids found themselves constructing new socialbehaviours that in part revolved around colour symbol-ism. Moreover, for reasons presumably associated withdifferent modes of symbolic expression, ochre usage allbut disappeared in later layers at the site.

It seems that the Qafzeh hominids most likely had anumber of desirable criteria in mind when they targetedthe nearby ochre sources (ferruginous oolites embeddedin sandy limestone) at Mt. Devora and Mt. Tavor. Hoverset al. quite correctly suggest that the hominids probablyvalued a range of characteristics in the ochre, which mayhave included texture as much as colour. Certain modernsocieties also value the lustre that can be obtained fromochre with specific impurities. While the silty quality ofthe archaeological samples found in the cave at Qafzehmay have been more desirable than matrix from thesources that were not exploited (veins of iron oxide indolomitic rocks and ferruginous concretions), it wouldalso be useful to know how they compared in terms ofcolour (i.e., content of iron oxide). If these unexploitedsources are indeed redder, it could support the idea thatalthough redness was the preeminent attraction, the in-tensity of red was not necessarily the sole criterion.There is good evidence that certain Australian Aborigi-nal communities, for instance, bypassed ochre sourcesthat were very red and embedded in undesirable matricesin favour of less colour-saturated sources that had a moredesirable plasticity.

While it seems, judging from the absence of randomsamples of yellow ochre, that goethite was not heated atQafzeh, I am not sure I would agree that mining andsubsequently heating goethite would have required agreater investment of time and energy. Goethite can betransformed into red haematite quite readily—it wouldhave simply required “roasting” of the lumps of ochrein an open fire. If a source of ochre is interleaved orspeckled from yellow through red, collecting ochre as itcame, irrespective of colour, and then heating it mightin fact have been less time-consuming than followinghaematite veins.

Hovers et al. distinguish between practical and sym-bolic uses of ochre. This distinction concurs with tra-

ditional views that often separate subsistence strategiesand related activities from ritual and ceremony, whichare conventionally viewed as epiphenomenal. Increas-ingly, however, the argument is that the boundaries be-tween the cultural, natural, and economic worlds of thepast were blurred and were rich in symbolic meaning(Bradley 1998, 2000; Gosden 1994). Therefore, whileochre may not have been used for tanning hides (a “prac-tical” activity) at Qafzeh, for instance, we should notassume that such an activity was necessarily devoid ofsymbolism.

I find the idea that the hearths in the ochre-bearinglayers at Qafzeh may have had symbolic meaning a tan-talizing one even though the authors choose to err onthe side of caution. For later prehistoric complex soci-eties, especially the highland communities of Anatolia,for example, there is a growing body of contextual evi-dence that the hearth was not only the focus of thehousehold but, through an association of deliberatelyplaced objects, had a multiplicity of purposes rangingfrom utilitarian to ritual (Sagona 1998). Similarly, theremay well have been a package of symbolic elements ofthe Middle Palaeolithic including the hearth as the au-thors suggest.

Finally, the Qafzeh data make one reflect on the ex-traordinary longevity of the symbolism of red, which haspersisted into modern times. Although ubiquitous, ochreusage was nonetheless historically and culturally con-tingent. In terms of colour symbolism what is so intrigu-ing is that certain societies—early Archaic states in theNear East, for instance—appear to have abandoned,wholly or in part, their attachment to red in favour ofblue (Sagona 1996). This change in symbolism appearsto have embraced a new set of values and social con-structs. While we shall never know what the coloursactually meant to the people who used them, it is worthmaking a suggestion. If at Qafzeh we see the beginningsof red as a symbolic colour for life, vigour, blood, fertility,and so on, the appearance of blue may represent the needfor an apotropaic colour at a time when there was a con-spicuous increase in personal wealth and prestige.

Reply

erella hovers , ofer bar-yosef , shimon il -ani , and bernard vandermeerschJerusalem, Israel. 30 v 03

We thank the editor of current anthropology, BenOrlove, for insisting on our turning a lengthy technicalreport on the ochre of Qafzeh Cave into a broader (andeven lengthier) article. We thank the commentators fortheir thoughtful and illuminating responses. It pleasesus that all of them see much merit in the presentationof the data and in the analytical methods we employedin our study and that, to varying degrees, they acceptour conclusions. The commentaries are diverse and

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 28: An Early Case of Color Symbolism: Ochre Use by Modern

hovers et al . An Early Case of Color Symbolism F 517

rarely focus on the same issues, and this dictates brevityin our response to each of the points raised.

Some of the comments address specific properties ofthe ochre assemblage of Qafzeh Cave. Sagona questionsthe robusticity of the selection criterion based on thecolor red. Analyses of ochre sources in the area of QafzehCave (mentioned but not presented in the article) indi-cate that the color of ochre there is more often yellow(goethite) than red. However, ochre from these sourcesis sometimes harder than the common ochre at Qafzeh.We do not doubt that texture was a criterion that helpedform decisions as to the particular exposure to exploit,since the softer silty material would have been easier toscratch and grind into powder. But if red was not thepreferred color, more goethite from the same sourcewould be expected.

Sagona also wonders whether the heating of goethitecollected at the same source would have increased thecost of the utilization of ochre. The answer is yes. Atthe time of writing we had submitted several pieces ofochre from Qafzeh Cave to luminescence analysis butdid not yet have the data to address the issue directlyand relied on reasoning to make the argument that ochrewas not heated. The recent publication of preliminaryresults of the analysis (Godfrey-Smith and Ilani 2003)shows that some goethite may have been heated. How-ever, red becomes black if the temperature rises above450�, far below the 1,000� reached in a hardwood hearth.The fact that a few lumps of goethite turned red but noneturned black suggests that temperatures were closelycontrolled and the heating was not accidental. Appar-ently, turning yellow goethite to red ochre was not sim-ply “roasting” it in the fire but a complex and time-consuming procedure. As is emphasized by Knight et al.in their comment, high-quality red ochre is costly. Com-bined with the overall paucity of yellow goethite in theassemblage, this is another indication that red was a de-sired quality.

Belfer-Cohen, McBrearty, and Sagona reflect on themeaning of ochre in this particular case study and in thearchaeological record in general. While their specificpoints differ, their comments go beyond the immediatecase to issues of hypothesis building, on the one hand,and epistemology, on the other. In recognition of theelusiveness of symbolic behavior and the difficulty oftesting and refuting hypotheses about symbolism, wepresented an inductive study that incorporated attemptsto refute a number of working hypotheses about the non-symbolic nature of the ochre assemblage. McBrearty, aself-professed believer in the symbolic use of ochre,points out that utilitarian uses for the Qafzeh ochre willdoubtless be invoked. Probably so—we likely did notaddress all the potential utilitarian explanations for thepresence of ochre—but we believe that we presented astrong enough case for the symbolic use of ochre thatnotions about additional utilitarian uses are no longernecessarily the most parsimonious explanation. Quite tothe contrary, Belfer-Cohen cautions against a too ma-terialistic, “brainy” as opposed to “soulful” reading ofthe evidence, while Sagona notes that “utilitarian” (for

want of a better term) uses of ochre may in fact havebeen symbolic as well. Both points are well taken, andsome archaeological cases (e.g., Inizian 1976) have indeedbeen interpreted in this way. We of course agree that thesoul was involved in symbolic activity at any given timein the human past. In the specific case of Qafzeh, oneshould bear in mind that we were trying to evaluate thecase for symbolism in a period for which its existenceis not a trivial notion. Our policy was to formulate work-ing hypotheses that went against the argument for sym-bolism, test them with the archaeological evidence, andmake an interpretation based on parsimony. Here weconcur with Speth (n.d.) that “parsimony—giving pri-ority to the simplest explanation—is not a fact of theway nature works but a reflection of our inability to graspand deal with the complexity of the real world” (see alsoKlein 2000:28). Beyond that, as observers of rather thanparticipants in the postulated symbolic system of QafzehCave we chose to err on the side of caution in our in-terpretations of the record and possible symbolic con-tents of artifacts. For this reason we discussed only ingeneral terms the idea of the diversification of early colorsymbol systems. The clarification by Knight, Power, andWatts that they expect such variation in the symboliccontents of local systems broadens the scope of theiroriginal model significantly and is welcome in itself.This expected evolutionary pattern might not interferewith drawing high-relevance analogies from present so-cieties to the prehistoric past. Still, they have not en-gaged in any treatment of dissimilarities between a widerange of sources and their subject—a criterion that isrecommended for evaluating the strength of a relationalanalogy (Wylie 1985)—and it remains unclear how ahighly relevant ethnographic analogy (Watts 2002) canbe used both as a building block of a model and as a testof that model.

A number of comments are concerned with the mean-ing of the Qafzeh ochre record in the context of the or-igins of behavioral modernity. Qafzeh Cave is wellknown as an early occurrence of anatomically modernhuman remains that are far better preserved than thosein African Middle Stone Age sites (Klein 2000:26; Kleinand Blake 2002:224–27). There are no Neandertal re-mains anywhere in the sequence of the site and no com-pelling indication of a change in the lithic assemblagesthat might suggest a change in population (Hovers andRaveh 2000), to the degree that such a distinction is pos-sible. The disjunction in the characteristics of the ochrecannot be related to the existence of two different pop-ulations. For the Levant in general, the behavioral dif-ferences suggested to have existed between Neandertalsand moderns (Lieberman and Shea 1994) have not with-stood testing in terms of methodology and biological andarchaeological evidence (Hovers 1997; Lieberman 1998;Speth and Tchernov 2001, n.d.; Stutz 2002). From abroader geographical perspective, Barham, McBrearty,and Knight, Power, and Watts see a clear dichotomy be-tween the African Middle Stone Age and the EuropeanMiddle Paleolithic in behavioral modernity. Explicitly orimplicitly, they view the ochre record from Qafzeh as

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 29: An Early Case of Color Symbolism: Ochre Use by Modern

518 F current anthropology Volume 44, Number 4, August–October 2003

vindicating the idea that the Levant should be perceivedas a geographical extension of the African Middle StoneAge (see Klein 1995, 1999, 2000; Klein and Blake 2002)and is therefore expected to be different from the MiddlePaleolithic record of pigment use in by Neandertals inEurope. How much of the observed difference is the re-sult of research history and/or bias is to some extent anopen question. Before the publication of this work, forexample, Barham was likely to include Qafzeh Cave inthe European “color-depauperate” realm on the basis ofthe single lump of ochre published from this sequence(Vandermeersch 1969). As he points out, there are oldcollections that merit reanalysis by modern techniques.We would suggest that this reexamination be expandedto include white and black ones, if any, to which (asMarshack points out) little attention has been paid inthe past. This would certainly be in line with linguisticstudies of color terms and of the neural, psychological,and physiological infrastructure of trichromatic visionin humans. We should also bear in mind that nonutili-tarian artifacts found in Middle Paleolithic sites havesometimes been considered intrusions from Upper Pa-leolithic horizons, and their cultural and chronologicalaffinities have not always been examined seriously (e.g.,Bar-Yosef 1988, d’Errico et al. 1998, Zilhao and d’Errico1999). The fact is that when pigments and/or other typesof nonutilitarian objects occur in Eurasian Middle Pa-leolithic sites, there is often a co-occurrence of severaltypes of such finds (Hovers, Vandermeersch, and Bar-Yosef 1997 and references therein) which may suggestmore than just random coincidence and possibly meritsin-depth contextual study. Finally, we draw attention tothe fact that burial has so far been encountered in MiddlePaleolithic Europe and the Near East but not in AfricanMiddle Stone Age sites. As Knight, Power, and Wattssuggest, this may be due to the fact that mortuary be-havior was different in the two regions—but then thesame logic can be applied, on the basis of current evi-dence, to the use of pigments in general and ochre inparticular in Neandertal Europe.

Klein emphasizes a temporal dichotomy, with homi-nids in both Africa and Europe having essentially thesame nearly but not fully modern behaviors until50,000–40,000 years ago, when neural changes kicked into create “us” and a late “Out of Africa” dispersion tookplace. While the growing body of genetic evidence pointsto the separation of H. neanderthalensis from H. sapiens,age estimates for this speciation event remain variable(e.g., Ingman, Paabo, and Gyllensten 2000, Knight 2003,Krings et al. 1999 and references in these works). Thedating of the FOXP2 gene, which may be implicated inthe ability to make the mouth and facial movementsessential to speech, to 200,000 at the earliest (Enard etal. 2002) is controversial (Balter 2002). Indeed, this studyraises the hope that future research may underpin a clearrelationship between genetic changes and dispersalevents out of Africa—but not just yet. Moreover, theFOXP2 gene may have been selected for “precisely be-cause it improved vocal communication once languagehad already evolved” (Paabo, in Balter 2002, emphasis

added). Its evolution into its known modern form mayin fact point to the earlier existence of language. Ar-chaeologically, the Late Out of Africa model may withsome difficulty explain the European Upper Pleistocenerecord but not that of the Levant prior to 50,000 yearsago. The Qafzeh Cave ochre record clearly satisfies thecriterion established by Chase and Dibble (1987) and byKlein, namely, that credible claims for modern humanbehavioral markers before 50,000 years ago “must in-volve relatively large numbers of highly patterned ob-jects from deeply stratified, sealed contexts” (Klein 2000:28). A number of African sites possibly attest to similarpatterning from roughly the same and even earlier time(Barham 2002; Brooks et al. 1995; Deino and McBrearty2002; Henshilwood et al. 2001, 2002; Yellen 1996, 1998;see discussion in Klein 2000). These occurrences can(with difficulty, we believe) be explained as the productsof humans who were “cognitively advanced in the di-rection of modern humans,” as Klein suggests. But whatis to be made of the Levantine Upper Paleolithic recordpost–50,000 years ago, in which little of the traditionalpackage of “behavioral modernity” is to be found beyondthe proliferation of blade technologies (Belfer-Cohen1988; Hovers 1992, 1997)? The overall picture wouldseem to suggest that both H. neanderthalensis and H.sapiens had the capacity for symbolic behavior and thatthe difference in the expressions of this capacity may bedue to demographic and social circumstances more thanto biological differences. The sporadic and erratic ex-pressions of modern behavior in the Eurasian Middle Pa-leolithic (and, indeed, the Upper Paleolithic as well) mayreflect the instability of mechanisms of long-term com-munal memory and failure to retain and inherit socialknowledge, possibly due to the instability of demo-graphic systems. This would also account for the differ-ent trajectories of establishment and development ofmodern human behavior in various geographical regions(Belfer-Cohen and Hovers 2002, Hovers and Belfer-Cohenn.d.).

References Citeda b r a m o v, i . 1997. “Physiological mechanisms of color vi-

sion,” in Color categories in thought and language. Edited byC. Hardin, C. Maffi, and L. Maffi, pp. 89–117. Cambridge:Cambridge University Press. [ck, cp, iw]

a m b ro s e , s . h . 1998. Chronology of the Later Stone Age andfood procurement in East Africa. Journal of Archaeological Sci-ence 25:377–92.

b a h n , p . g . , a n d j . v e r t u t . 1997. Journey through theIce Age. London: Weidenfeld and Nicolson.

b a l i k c i , a . 1970. The Netsilik Eskimo. Garden City: NaturalHistory Press.

b a l t e r , m . 2002. “Speech gene” tied to modern humans. Sci-ence 297:1105.

b a r h a m , l . s . 1998. Possible early pigment use in South-Cen-tral Africa. current anthropology 39:703–10.

———. 2002. Systematic pigment use in the Middle Pleistoceneof south-central Africa. current anthropology 43:181–90.[lsb, ck, cp, iw]

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 30: An Early Case of Color Symbolism: Ochre Use by Modern

hovers et al . An Early Case of Color Symbolism F 519

b a r l e y, n . 1983. The innocent anthropologist: Notes from amud hut. London: Penguin Books.

b a r - y o s e f , o . 1988. “Evidence for Middle Palaeolithic sym-bolic behavior: A cautionary note,” in L’homme de Neander-tal, vol. 5, La pensee. Edited by M. Otte, pp. 11–16. Liege:Universite de Liege.

———. 1994. “The contributions of Southwest Asia to the studyof the origins of modern humans,” in Origins of anatomicallymodern humans. Edited by M. H. Nitecki and D. V. Nitecki,pp. 23–66. New York: Plenum Press. [am]

———. 1997. “Symbolic expressions in later prehistory of the Le-vant: Why are they so few?” in Beyond art: Pleistocene imageand symbol. Edited by M. Conkey, O. Soffer, D. Stratman, andN. G. Jablonski, pp. 161–87. Memoirs of the California Acad-emy of Sciences 23.

———. 1998. “The chronology of the Middle Paleolithic in theLevant,” in Neandertals and modern humans in the Levant.Edited by T. Akazawa, K. Aoki, and O. Bar-Yosef, pp. 39–56.New York: Plenum Press.

b a r - y o s e f , o . , a n d b . v a n d e r m e e r s c h . 1993. Modernhumans in the Levant. Scientific American 268:94–99.

b a t e s , r . l . , a n d j . a . j a c k s o n . Editors. 1980. 2d edi-tion. Glossary of geology. Falls Church, Virginia: American Ge-ological Institute.

b e d n a r i k , r . g . 1992a. Palaeoart and archaeological myths.Cambridge Archaeological Journal 2:27–43.

———. 1992b. The stuff legends in archaeology are made of: Areply to critics. Cambridge Archaeological Journal 2:262–65.

b e l f e r - c o h e n , a . 1988. “The appearance of symbolic ex-pression in the Upper Pleistocene of the Levant as compared toWestern Europe,” in L’homme de Neandertal, vol. 5, La pen-see. Edited by M. Otte, pp. 25–29. Liege: Universite de Liege.

b e l f e r - c o h e n , a . , a n d e . h o v e r s . 1992. In the eye ofthe beholder: Mousterian and Natufian burials in the Levant.current anthropology 33:463–71.

———. 2002. The way we were: Reflections on human behaviorin the Middle Paleolithic. Paper presented at the session “TheMiddle Paleolithic: A long uphill climb, or going nowherefast?” 67th annual meeting of the Society for American Arche-ology, Denver, Colo. [ab]

b e r l i n , b . , a n d p . k a y. 1969. Basic color terms: Their uni-versality and evolution. Berkeley and Los Angeles: Universityof California Press.

b e y r i e s , s . 1983. “Fonction et mode d’utilization d’un serie delames ocrees capsiennes,” in Traces d’utilization sur les outilsneolithiques de Proche Orient. Edited by M-C. Cauvin, pp.135–42. Lyon: Maison de l’Orient.

b o e d a , e . , j . c o n n a n , a n d s . m u h e s e n . 1998. “Bitu-men as hafting material on Middle Paleolithic artifacts formthe el-Kowm Basin, Syria,” in Neandertals and modern hu-mans in Asia: Recontres Internationales d’Archeologie etd’Histoire d’Antibe. Edited by T. Akazawa, K. Aoki, and O.Bar-Yosef, pp. 181–204. New York: Plenum Press.

b o e d a , e . , j . m . g e n e s t e , a n d l . m e i g n e n . 1990.Identification de chaınes operatoires lithiques du Paleolithiqueancien et moyen. Paleo 2:43–79.

b o r d e s , f . 1952. Sur l’usage probable de la peinture corporelledans certains tribus mousteriennes. Bulletin de la Societe Pre-historique Francaise 49:169–71.

b r a d l e y, r . 1998. The significance of monuments: On shap-ing of human experience in Neolithic and Bronze Age Europe.London: Routledge. [as]

———. 2000. An archaeology of natural places. London: Rout-ledge. [as]

b ro o k s , a . s . , d . m . h e l g r e m , j . s . c r a m e r , a .f r a n k l i n , w. h o r n y a k , j . m . k e a t i n g , r . g .k l e i n , w. j . r i n k , h . p . s c h w a r c z , j . n . s m i t h ,k . s t e w a r t , n . e . t o d d , i . v e r n i e r s , a n d j . e .y e l l e n . 1995. Dating and context of three Middle Stone Agesites with bone points in the Upper Semliki Valley, Zaire. Sci-ence 268:548–53.

b u e l l e r , j . 1988. “Handling, hafting, and ochre stains,” in In-dustries lithiques: Traceologie et technologie. Edited by S. Bey-

ries, pp. 5–32. British Archaeological Reports International Se-ries 411.

c h a o , t . t . , a n d p . k . t h e o b a l d j r . 1976. The signifi-cance of secondary iron and manganese oxides in geochemicalexplanation. Economic Geology 71:1560–69.

c h a s e , p . g . 1991. Symbols and Paleolithic artifacts: Style,standardization, and the imposition of arbitrary form. Journalof Anthropological Archaeology 10:193–214.

c h a s e , p . g . , a n d h . l . d i b b l e . 1987. Middle Paleolithicsymbolism: A review of current evidence and interpretations.Journal of Anthropological Archaeology 6:263–96.

c l a r k , d . j . 1988. The Middle Stone Age of East Africa andthe beginnings of regional identity. Journal of World Prehistory2:235–305.

c o n k e y, m . 1978. “Style and information in cultural evolu-tion: Toward a predictive model for the Paleolithic,” in Socialarchaeology: Beyond subsistence and dating. Edited by C. L.Redman, M. J. Berman, E. V. Curtin, W. J. Longhorne Jr., N. M.Versaggi, and C. Wansen, pp. 61–84. New York: AcademicPress.

c o u r a u d , c . 1988. Pigments utilises en prehistoire: Prove-nance, preparation, mode d’utilisation. L’Anthropologie 92:17–28.

d ’ a n d r a d e , r . 1995. The development of cognitive anthro-pology. Cambridge: Cambridge University Press. [ck, cp, iw]

d e a c o n , t . w. 1989. The neural circuitry underlying primatecalls and human language. Human Evolution 4:367–401.

———. 1990. “Brain-language coevolution.” The evolution of hu-man languages: Proceedings of the workshop on the evolutionof human languages held August 1989 in Santa Fe, New Mex-ico, New York, 1990, pp. 49–83.

———. 1997. The symbolic species: The co-evolution of languageand the brain. New York: Norton.

d e b e a u n e , s . a . 1993. “Nonflint stone tools of the EarlyUpper Paleolithic,” in Before Lascaux: The complex record ofthe Early Upper Paleolithic. Edited by H. Knecht, A. Pike-Tay,and R. White, pp. 163–91. Boca Raton: CRC Press.

d e i n o , a . l . , a n d s . m c b r e a r t y. 2002. 40Ar/39Ar datingof the Kapthurin Formation, Baringo, Kenya. Journal of HumanEvolution 42:185–210.

d ’ e r r i c o , f . 2003. The invisible frontier: A multiple speciesmodel for the origin of “behavioral modernity.” EvolutionaryAnthropology 12. In press. [rgk]

d ’ e r r i c o , f . , a n d a . n o w e l l . 2000. A new look at theBerekhat Ram figurine: Implications for the origins of symbol-ism. Cambridge Archaeological Journal 10:123–67.

d ’ e r r i c o , f . , a n d m . s o r e s s i . 2002. Systematic use ofmanganese pigment by Pech de-l’Aze Neandertals: Implica-tions for the origin of behavioral modernity. Journal of HumanEvolution 42:A13.

d ’ e r r i c o , f . , j . z i l h a o , m . j u l i e n , d . b a f f i e r ,a n d j . p e l e g r i n . 1998. Neanderthal acculturation inWestern Europe? A critical review of the evidence and its in-terpretation. current anthropology 39:S1–S44.

d e v a l o i s , r . l . , a n d k . k . d e v a l o i s . 1993. A multi-stage color model. Vision Research 33:1053–65. [ck, cp, iw]

d o m i n y, n . j . , a n d p . w. l u c a s . 2001. Ecological impor-tance of trichromatic vision to primates. Nature 410:363–66.

e n a r d , w. , m . p r z e w o r s k i , s . e . fi s h e r , c . s . l .l a i , v. w i e b e , t . k i t a n o , a . p . m o n a c o , a n d s .p a a b o . 2002. Molecular evolution of FOXP2, a gene involvedin speech and language. Nature 418:869–72. [rgk]

e r l a n d s o n , j . m . , j . d . ro b e r t s o n , a n d c . d e s -c a n t e s . 1999. Geochemical analysis of eight red ochres fromwestern North America. American Antiquity 64:517–26.

f a r r a n d , w. r . 1979. Chronology and palaeoenvironment ofLevantine prehistoric sites as seen from the sediment studies.Journal of Archaeological Science 6:369–92.

f o s t e r , m . l . 1990. “Symbolic origins and transitions in thePaleolithic,” in The emergence of modern humans: An archae-ological perspective. Edited by P. Mellars, pp. 517–39. Ithaca:Cornell University Press.

f r i e d m a n , e . , n . g o r e n - i n b a r , a . ro s e n f e l d , o .

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 31: An Early Case of Color Symbolism: Ochre Use by Modern

520 F current anthropology Volume 44, Number 4, August–October 2003

m a r d e r , a n d f . b u r i a n . 1994–95. Hafting during Mous-terian times: Further evidence. Journal of the Israel PrehistoricSociety 26:8–31.

g a g e , j . 1999. Did colours signify? Symbolism in the red.Cambridge Archaeological Journal 9:110–12.

g a r g e t t , r . h . 1999. Middle Paleolithic burial is not a deadissue: The view from Qafzeh, Saint-Cesaire, Kebara, Amud,and Dederiyeh. Journal of Human Evolution 37:27–90.

g o d f r e y - s m i t h , d . i . , a n d s . i l a n i . 2003. “Earliest ev-idence for deliberate heat treatment of archaeological ochrefrom a 100,000 year old horizon at Qafzeh Cave, Israel,” inAbstracts of the Annual Meeting of the Israel Geological Soci-ety, 25–27 March, Ein Bokek. Edited by S. Feinstein, A. Agnon,E. Farber, S. Ezra, Y. Avin, and Y. Levi. Jerusalem: Israel Geo-logical Society.

g o l d s t o n e , r . l . , a n d l . w. b a r s a l o u . 1998. “Reu-niting perception and conception,” in Similarity and symbolsin human thinking. Edited by S. A. Sloman and L. J. Rips, pp.145–76. Cambridge: MIT Press.

g o r i n g - m o r r i s , a . n . , a n d a . b e l f e r - c o h e n . 2001.“The symbolic realms of utilitarian material culture: The roleof lithics,” in Beyond tools. Edited by I. Caneva, C. Lemorini,D. Zampetti, and P. Biagi, pp. 257–71. Berlin: Ex Oriente.

g o s d e n , c . 1999. Social being and time. Oxford: Blackwell.[lsb, as]

g r u n , r . , a n d p . b e a u m o n t . 2001. Border Cave revisited:A revised ESR chronology. Journal of Human Evolution 40:467–82. [ck, cp, iw]

h a y d e n , b . 1993. The cultural capacities of Neanderthals: Areview and re-evaluation. Journal of Human Evolution 24:113–46.

h e n s h i l w o o d , c . s . , f . d ’ e r r i c o , r . y a t e s , z . j a -c o b s , c . t r i b o l o , g . a . t . d u l l e r , n . m e r c i e r ,j . c . s e a l y, h . v a l l a d a s , i . w a t t s , a n d a . g .w i n t l e . 2002. Emergence of modern human behavior: Mid-dle Stone Age engravings from South Africa. Science 295:1278–80.

h e n s h i l w o o d , c . , j . c . s e a l y, r . y a t e s , k . c ru z -u r i b e , p . g o l d b e r g , f . e . g r i n e , r . g . k l e i n , c .p o g g e n p o e l , k . v a n n i e k e r k , a n d i . w a t t s . 2001.Blombos Cave, Southern Cape, South Africa: Preliminary re-port on the 1992–1999 excavations of the Middle Stone Agelevel. Journal of Archaeological Science 28:421–48.

h e i d e r , e . r . 1972. Probabilities, sampling, and ethnographicmethod: The case of Dani colour names. Man 7:448–66. [ck,cp, iw]

h o v e r s , e . 1990. Art in the Levantine Epi-Palaeolithic: An en-graved pebble from a Kebaran site in the Lower Jordan Valley.current anthropology 31:317–22.

———. 1992. Review of: The emergence of modern humans: Anarchaeological perspective. Edited by P. Mellars (Ithaca: Cor-nell University Press, 1990). Journal of Human Evolution 22:155–57.

———. 1997. Variability of lithic assemblages and settlementpatterns in the Levantine Middle Paleolithic: Implications forthe development of human behavior. Ph.D. diss., The HebrewUniversity, Jerusalem, Israel.

h o v e r s , e . , a n d a . b e l f e r - c o h e n . n.d. “Now you seeit, now you don’t—modern human behavior in the Middle Pa-leolithic,” in Transitions before the transition. Edited by E.Hovers and S. L. Kuhn. MS.

h o v e r s , e . , w. h . k i m b e l , a n d y. r a k . 2000. Amud7—still, a burial: Response to Gargett. Journal of Human Evo-lution 39:253–60.

h o v e r s , e . , y. r a k , r . l a v i , a n d w. h . k i m b e l .1995. Hominid remains from Amud Cave in the context of theLevantine Middle Paleolithic. Paleorient 21:47–61.

h o v e r s , e . , a n d a . r a v e h . 2000. The use of a multivar-iate display technique in the analysis of inter-assemblage lithicvariability: A case study from Qafzeh Cave, Israel. Journal ofArchaeological Science 27:1023–38.

h o v e r s , e . , b . v a n d e r m e e r s c h , a n d o . b a r - y o s e f .

1997. A Middle Palaeolithic engraved artefact from QafzehCave, Israel. Rock Art Research 14:79–87.

h u m p h r e y, n . 1998. Cave art, autism, and the evolution ofthe human mind. Cambridge Archaeological Journal 8:165–91.

i n g m a n , m . , s . p a a b o , a n d u . g y l l e n s t e n . 2000.Mitochondrial genome variation and the origin of modern hu-mans. Nature 408:708–12.

i n i z i a n , m - l . 1976. Outils lithique capsien ocres.L’Anthropologie 80:39–63.

j o n e s , r . , a n d n . w h i t e . 1988. “Point blank: Stone toolsmanufacture at the Ngilipitju Quarry, Arnhem Land, 1981,” inArchaeology with ethnography: An Australian perspective. Ed-ited by B. Meehan and R. Jones, pp. 51–87. Canberra: Austra-lian National University. [ab]

k a f r i , u . 1991. Lithostratigraphy of the Judea Group in east-ern Galilee, emphasizing the Naftali Mountains. GSI/24/91. Je-rusalem: Geological Survey of Israel.

k a y, p . , b . b e r l i n , m . m a f fi , a n d w. m e r r i fi e l d .1997. “Color naming across languages,” in Color categories inthought and language. Edited by C. Hardin, C. Maffi, and L.Maffi, pp. 21–58. Cambridge: Cambridge University Press. [ck,cp, iw]

k a y, p . , a n d c . k . m c d a n i e l . 1997. “The linguistic sig-nificance of the meanings of basic color terms,” in Readings incolor, vol. 2, The science of color. Edited by A. Byrne and D.R. Hilbert, pp. 399–441. Cambridge: MIT Press.

k l e i n , r . g . 1995. Anatomy, behavior, and modern human or-igins. Journal of World Prehistory 9:167–98.

———. 1999. 2d edition. The human career: Human biologicaland cultural origins. Chicago: University of Chicago Press.

———. 2000. Archeology and the evolution of human behavior.Evolutionary Anthropology 9:17–36.

k l e i n , r . g . , a n d e . b l a k e . 2002. The dawn of humanculture. New York: Wiley.

k n i g h t , a . 2003. The phylogenetic relationship of Neandertaland modern human mitochondrial DNAs based on informativenucleotide sites. Journal of Human Evolution 44:627–32.

k n i g h t , c . 1991. Blood relations: Menstruation and the ori-gins of culture. London and New Haven: Yale University Press.[ck, cp, iw]

———. 1998. “Ritual/speech coevolution: A solution to the prob-lem of deception,” in Approaches to the evolution of language:Social and cognitive bases. Edited by J. R. Hurford, M. Stud-dert-Kennedy, and C. Knight, pp. 68–91. Cambridge: CambridgeUniversity Press. [ck, cp, iw]

———. 1999. “Sex and language as pretend-play,” in The evolu-tion of culture. Edited by R. Dunbar, C. Knight, and C. Power,pp. 228–29. Edinburgh: Edinburgh University Press. [ck, cp, iw]

———. 2000. “Play as precursor of phonology and syntax,” inThe evolutionary emergence of language: Social function andthe origins of linguistic form. Edited by C. Knight, M. Stud-dert-Kennedy, and J. R. Hurford, pp. 99–119. Cambridge: Cam-bridge University Press. [ck, cp, iw]

———. 2002. “Language and revolutionary consciousness,” inThe transition to language. Edited by A. Wray, pp. 138–60. Ox-ford: Oxford University Press. [ck, cp, iw]

k n i g h t , c . , c . p o w e r , a n d i . w a t t s . 1995. The hu-man symbolic revolution: A Darwinian account. CambridgeArchaeological Journal 5:75–114.

k o r n , n . 2000. A typology of colours: The variability and dis-tribution of red- and yellow-based pigments in the Western andCentral Negev during the Epi-Paleolithic ca. 20,000–10,000 b.p.M.A. thesis, The Hebrew University, Jerusalem, Israel.

k r a y b i l l , n . 1977. “Pre-agricultural tools for the preparationof foods in the Old World,” in Origins of agriculture. Edited byC. A. Reed, pp. 485–521. The Hague: Mouton.

k r i n g s , m . , h . g e i s e r t , r . w. s c h m i t z , h . k r a i -n i t z k i , a n d s . p a a b o . 1999. DNA sequence of the mito-chondrial hypervariable region II from the Neandertal typespecimen. Proceedings of the National Academy of Science,U.S.A. 96:5581–85.

l i e b e r m a n , d . e . 1998. “Neandertal and early modern hu-man mobility patterns: Comparing archaeological and anatomi-

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 32: An Early Case of Color Symbolism: Ochre Use by Modern

hovers et al . An Early Case of Color Symbolism F 521

cal evidence,” in Neandertals and modern humans in WesternAsia. Edited by T. Akazawa, K. Aoki, and O. Bar-Yosef, pp.263–75. New York: Plenum Press.

l i e b e r m a n , d . e . , a n d j . j . s h e a . 1994. Behavioral dif-ferences between archaic and modern humans in the LevantineMousterian. American Anthropologist 96:300–332. [sm]

l i e b e r m a n , p . 1991. Uniquely human: The evolution ofspeech, thought, and selfless behavior. Cambridge: HarvardUniversity Press.

m c b r e a r t y, s . 1999. “The archaeology of the Kapthurin For-mation,” in Late Cenozoic environments and hominid evolu-tion: A tribute to Bill Bishop. Edited by P. Andrews and P. Ban-ham, pp. 143–56. London: Geological Society. [ck, cp, iw]

———. 2001. “The Middle Pleistocene of East Africa,” in Humanroots: Africa and Asia in the Middle Pleistocene. Edited by L.Barham and K. Robson-Brown, pp. 81–98. Bristol: Western Aca-demic and Specialist Press.

m c b r e a r t y, s . , l . b i s h o p , t . p l u m m e r , r . d e w a r ,a n d n . c o n a r d . 1998. Tools underfoot: Human tramplingas an agent of lithic artifact edge modification. American An-tiquity 63:108–29.

m c b r e a r t y, s . , a n d a . s . b ro o k s . 2000. The revolu-tion that wasn’t: A new interpretation of the origin of modernhuman behavior. Journal of Human Evolution 39:453–563.

m a c l a u ry, r . e . 1992. From brightness to hue: An explana-tory model of color-category evolution. current anthropol-ogy 33:137–86.

m a r s h a c k , a . 1981. On Paleolithic ochre and the early usesof color and symbol. current anthropology 22:188–91.

———. 1989. Evolution of the human capacity: The symbolic ev-idence. Yearbook of Physical Anthropology 32:1–34.

———. 1996. A Middle Paleolithic symbolic composition fromthe Golan Heights: The earliest known depictive image. cur-rent anthropology 37:357–65.

———. 2001. Comment on: Middle and Early Paleolithic burialsand the use of chronotypology in contemporary Paleolithic re-search, by J. Riel-Salvatore and G. A. Clark. current anthro-pology 42:466–67. [am]

m e l l a r s , p . 1996. The Neanderthal legacy: An archaeologicalperspective from Western Europe. Princeton: Princeton Univer-sity Press.

m i m r a n , y. 1972. The Tayasir volcanics. Israel Geological Sur-vey Bulletin 52.

m i t h e n , s . 1996. The prehistory of the mind: A search for theorigins of art, religion, and science. London: Thames andHudson.

m o l l o n , j . d . 1997. “ ‘Tho’ she kneel’d in that place wherethey grew . . .’: The uses and origins of primate colour vision,”in Readings in color, vol. 2, The science of color. Edited by A.Byrne and D. R. Hilbert, pp. 379–96. Cambridge: MIT Press.

m u n s e l l . 1962. Pocket edition. Munsell book of colors. Vol. 1.Baltimore: Munsell Color Co., Inc.

n o b l e , w. , a n d i . d a v i d s o n . 1996. Human evolution,language, and mind: A psychological and archaeological in-quiry. Cambridge: Cambridge University Press.

n o w e l l , a . , f . d ’ e r r i c o , a n d e . h o v e r s , 2001. “Theorigin of symbolism in the Near East,” in Abstracts of the66th Annual Meeting. Baltimore: Society for AmericanArchaeology.

p h i l i b e r t , s . 1994. L’ocre et le traitement des peaux: Revi-sion d’une conception traditionelle par l’analyse fonctionnelledes grattoirs ocres de la Balma Margineda (Andorre).L’Anthropologie 98:447–53.

p o m i e s , m - p . , m . m e n u , a n d c . v i g n a u d , 1999. RedPaleolithic pigments: Natural hematite or heated goethite? Ar-chaeometry 41:275–85.

p o w e r , c . 1998. “Old wives’ tales: The gossip hypothesis andthe reliability of cheap signals,” in Approaches to the evolu-tion of language: Social and cognitive bases. Edited by J. R.Hurford, M. Studdert-Kennedy, and C. Knight, pp. 111–29.Cambridge: Cambridge University Press. [ck, cp, iw]

———. 2000. “Secret language use at female initiation: Boundinggossiping communities,” in The evolutionary emergence of

language: Social function and the origins of linguistic form.Edited by C. Knight, M. Studdert-Kennedy, and J. R. Hurford,pp. 81–98. Cambridge: Cambridge University Press. [ck, cp, iw]

p o w e r , c . , a n d l . c . a i e l l o . 1997. “Female proto-sym-bolic strategies,” in Women in human evolution. Edited by L.D. Hager, pp. 153–71. London and New York: Routledge.

p o w e r , c . , a n d i . w a t t s . 1999. “First gender, wrong sex,”in Those who play with fire: Gender, fertility, and transforma-tion. Edited by H. Moore, T. Sanders, and B. Kaare, pp. 101–32.London and New Brunswick: Athlone Press. [ck, cp, iw]

r a b i n o v i c h , r . , a n d e . t c h e r n o v. 1995. “Chronologi-cal, paleoecological, and taphonomical aspects of the MiddlePaleolithic site of Qafzeh, Israel.” Archaeozoology of the NearEast: Proceedings of the Second Symposium of the Archaeo-zoology of Southwestern Asia and Adjacent Areas. Edited byH. Buitenhuis and H.-P. Uerpmann, pp. 5–44. Leiden:Backhuys.

r a p p a p o r t , r . 1999. Ritual and religion in the making of hu-manity. Cambridge: Cambridge University Press. [ck, cp, iw]

r i e l - s a l v a t o r e , j . , a n d g . a . c l a r k . 2001. Gravemarkers: Middle and Early Upper Paleolithic burials and theuse of typochronology in contemporary Paleolithic research.current anthropology 42:449–79.

ro b b , j . e . 1998. The archaeology of symbols. Annual Reviewof Anthropology 27:327–46.

s a c k e t t , j . r . 1983. “Style and ethnicity in archaeology: Thecase for isochrestism,” in The uses of style in archaeology. Ed-ited by M. Conkey and C. Hastorf, pp. 32–43. Cambridge:Cambridge University Press.

s a g o n a , a . 1994. “The quest for the red gold,” in Bruising thered earth: Ochre mining and ritual in aboriginal Tasmania.Edited by A. Sagona, pp. 8–38. Melbourne: Melbourne Univer-sity Press.

———. 1998. Social identity and religious ritual in the Kura-Ar-axes cultural complex: Some observations from Sos Hoyuk.Mediterranean Archaeology 11:13–25. [as]

s a g o n a , c . 1996. “Red to blue: Colour symbolism and humansocieties,” in Cultural interaction in the ancient Near East.Edited by G. Bunnens, pp. 145–54. Louvain: Peeters. [as]

s c h e p a r t z , l . a . 1993. Language and modern human ori-gins. Yearbook of Physical Anthropology 36:91–124.

s c h m a n d t - b e s s e r a t , d . 1996. How writing came about.Austin: University of Texas Press. [sm]

s c h w a r c z , h . p . , r . g r u n , b . v a n d e r m e e r s c h , o .b a r - y o s e f , h . v a l l a d a s , a n d e . t c h e r n o v. 1988.ESR dates for the hominid burial site of Qafzeh in Israel. Jour-nal of Human Evolution 17:733–37.

s h e a , j . j . 1989. Tool use in the Levantine Mousterian of Ke-bara Cave, Mount Carmel. Mitekufat Ha’even: Journal of theIsrael Prehistoric Society 22:15–30.

———. 1991. The behavioral significance of Levantine Mouste-rian variability. Ph.D. diss., Harvard University, Cambridge,Mass.

———. 1997. “Middle Paleolithic spear point technology,” inProjectile technology. Edited by H. Knecht, pp. 79–106. NewYork: Plenum Press.

s h e p a r d , r . n . 1997. “The perceptual organization of colors:An adaptation to regularities of the world?” in Readings incolor, vol. 2, The science of color. Edited by A. Byrne and D.R. Hilbert, pp. 311–56. Cambridge: MIT Press.

s o l e c k i , r . s . 1982. “A ritual Middle Paleolithic deer burialat Nahr Ibrahim Cave, Lebanon,” in Archeologie au Levant:Recueil a la memoire de Roger Saidah, vol. 9, pp. 47–56. Lyon:Maison de l’Orient.

s p e n c e r , b . , a n d f . j . g i l l e n . 1968 (1898). The nativetribes of Central Australia. New York: Dover.

s p e t h , j . d . n.d. “Hunting pressure, subsistence intensifica-tion, and demographic change in the Levantine Late MiddlePaleolithic,” in Human paleoecology in the Levantine corridor.Edited by N. Goren-Inbar and J. D. Speth. Oxford: OxbowPress.

s p e t h , j . d . , a n d e . t c h e r n o v. 2001. “Neandertal hunt-ing and meat processing in the Near East: Evidence from Ke-

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions

Page 33: An Early Case of Color Symbolism: Ochre Use by Modern

522 F current anthropology Volume 44, Number 4, August–October 2003

bara Cave (Israel),” in Meat-eating and human evolution. Ed-ited by C. B. Stanford and H. T. Bunn, pp. 52–72. Oxford:Oxford University Press.

———. n.d. “Kebara Cave as a Middle Paleolithic settlement: Afaunal perspective,” in The Middle Paleolithic archaeology ofKebara Cave. Edited by O. Bar-Yosef and L. Meignen. Cam-bridge: Peabody Museum of Archaeology and Ethnology, Har-vard University.

s t o u t , d . 2002. Skill and cognition in stone tool production.current anthropology 43:693–722. [ab]

s t u t z , a . j . 2002. Polarizing microscopy identification ofchemical diagenesis in archaeological cementum. Journal ofArchaeological Science 29:1327–47.

s u n , r . k . 1983. Perceptual distances and basic color term en-coding sequence. American Anthropologist 85:387–91.

t c h e r n o v, e . 1988. “Biochronology of the Middle Palaeolithicand dispersal events of hominids in the Levant,” in L’hommede Neandertal, vol. 2, L’environnement. Edited by M. Otte, pp.153–68. Liege: Universite de Liege.

t i l l i e r , a . - m . 1990. Une controverse depassee: L’existence depratiques funeraires au Paleolithique moyen. Les Nouvelles del’Archeologie 40:22–24.

———. 1999. Les enfants mousteriens de Qafze: Interpretationphylogenetique et paleoauxologique. Paris: Editions du CNRS.

t u r n e r , v. 1966. “Color classification in Ndembu ritual: Aproblem in primitive classification,” in Anthropological ap-proaches to the study of religion. Edited by M. Barton, pp.47–84. London: Tavistock.

———. 1970. The forest of symbols: Aspects of Ndembu ritual.Ithaca: Cornell University Press.

v a l l a d a s , h . , j . l . r e y s s , j . l . j o ro n , g . v a l l a -d a s , o . b a r - y o s e f , a n d b . v a n d e r m e e r s c h . 1988.Thermoluminescence dating of Mousterian “Proto-Cro-Mag-non” remains from Israel and the origin of modern man. Na-ture 331:614–16.

v a n g e n n e p , a . 1960 (1908). The rites of passage. Translatedby M. B. Vizedom and G. L. Caffee. Chicago: University ofChicago Press. [ck, cp, iw]

v a n d e r m e e r s c h , b . 1969. Decouverte d’un object en ocreavec trace d’utilisation dans le mousterien de Qafzeh (Israel).Bulletin de la Societe Prehistorique Francaise 66:157–58.

———. 1970. Une sepulture mousterienne avec offrandes decou-verte dans la grotte de Qafzeh. Compte Rendus de l’Academiedes Sciences 268:298–301.

———. 1981. Les hommes fossiles de Qafzeh (Israel). Paris: Edi-tions du CNRS.

v e l o , j . 1984. Ochre as medicine: A suggestion for the inter-pretation of the archaeological record. current anthropology25:674.

v o n w a t t e n w y l , a . , a n d h . z o l l i n g e r . 1979. Color-

term salience and neurophysiology of color vision. AmericanAnthropologist 81:279–88.

w a t t s , i . 1999. “The origin of symbolic culture,” in The evo-lution of human culture. Edited by R. Dunbar, C. Knight, andC. Power, pp. 113–46. Edinburgh: Edinburgh University Press.

w a t t s , i . 2002. Ochre in the Middle Stone Age of southernAfrica: Ritualised display or hide preservative? South AfricanArchaeological Bulletin 57:2–16. [lsb, ck, cp, iw]

w e i l e r , y. 1968. Geology of Nazareth Hills and Mt. Tabor. Is-rael Journal of Earth Sciences 17:63–82.

w e i n s t e i n - e v ro n , m . 1998. Early Natufian el-Wad revis-ited. Liege: Universite de Liege.

w e i n s t e i n - e v ro n , m . , a n d s . i l a n i . 1994. Provenanceof ochre in the Natufian layers of El-Wad Cave, Mount Car-mel, Israel. Journal of Archaeological Science 21:461–67.

w h i t e , r . 1989. “Production complexity and standardizationin Early Aurignacian bead and pendant manufacture: Evolu-tionary implications,” in The human revolution: Behavioraland biological perspectives on the origins of modern humans.Edited by P. Mellars and C. Stringer, pp. 366–90. Edinburgh:Edinburgh University Press.

w o b s t , h . m . 1977. “Stylistic behavior and informationexchange,” in For the Director: Essays in honor of James B.Griffin. Edited by Charles E. Cleland, pp. 317–42. University ofMichigan Museum of Anthropology, Anthropological Papers61.

w r e s c h n e r , e . 1980. Red ochre and human evolution: A casefor discussion. current anthropology 21:631–34.

———. 1983. Studies in prehistoric ochre technology. Ph.D. diss.,The Hebrew University, Jerusalem, Israel.

w y l i e , e . 1985. The reaction against analogy. Advances in Ar-chaeological Method and Theory 8:63–111.

w y n n , t . g . 1996. “The evolution of tools and symbolic be-havior,” in Handbook of human symbolic evolution. Edited byC. S. Peters, pp. 263–87. Oxford: Clarendon Press.

y e l l e n , j . e . 1996. Behavioral and taphonomic patterning atKatanda 9, a Middle Stone Age site, Kivu Province, Zaire. Jour-nal of Archaeological Science 23:915–32.

———. 1998. Barbed bone points: Tradition and continuity in Sa-haran and Sub-Saharan Africa. African Archaeological Review15:173–98.

z a c k h e i m , o . 1997. Provenance of ochre from the Natufianassemblages of el-Wad, Eynan, and Hayonim (in Hebrew). M.A.thesis, University of Haifa, Haifa, Israel.

z a h a v i , a . , a n d a . z a h a v i . 1997. The handicap principle:A missing piece in Darwin’s puzzle. New York and Oxford:Oxford University Press. [ck, cp, iw]

z i l h a o , j . , a n d f . d ’ e r r i c o . 1999. The chronology andtaphonomy of the earliest Aurignacian and its implications forthe understanding of Neandertal extinction. Journal of WorldPrehistory 13:1–68.

This content downloaded from 128.103.149.52 on Tue, 20 May 2014 23:31:02 PMAll use subject to JSTOR Terms and Conditions