South African Archaeological Society

Ochre in the of Southern Africa: Ritualised Display Hide Preservative? Author(s): Ian Watts Reviewed work(s): Source: The South African Archaeological Bulletin, Vol. 57, No. 175 (Jun., 2002), pp. 1-14 Published by: South African Archaeological Society Stable URL: http://www.jstor.org/stable/3889102 . Accessed: 03/01/2013 17:48

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

.

South African Archaeological Society is collaborating with JSTOR to digitize, preserve and extend access to The South African Archaeological Bulletin.

http://www.jstor.org

This content downloaded on Thu, 3 Jan 2013 17:48:58 PM All use subject to JSTOR Terms and Conditions South African ArchaeologicalBulletin 57 (175): 1-14, 2002 l

RESEARCHARTICLES IN THE MIDDLE STONE AGE OF SOUTHERN AFRICA: RITUALISED DISPLAY OR HIDE PRESERVATIVE? IAN WATTS

Department of Archaeology, University of , Rondebosch, 7700. Email: [email protected] (Received April 2000. Accepted February 2002.)

ABSTRACT (Riel-Salvatore& Clark2001) but, following colonizationof Europeby Homo sapiens, the majorityof burialsdating to the Symbolic and utilitarian interpretations have been early and middle Upper Palaeolithic also contained ochre proposedfor ochre use in the African Middle Stone Age, (Aldhouse-& Pettitt 1998; Mussi 2001; Riel-Salvatore but these have rarely been This developed. . paper reviews & Clark2001). This suggeststhat the ethnographicpattern of the hypotheses, recasts them in more explicit form and habitualritual use is of great antiquityand may be species- addresses the need basic data for for quantifying and specific. If we accept that modem humans have a recent ochre describing assemblages andfor synthesizingobserva- Africanorigin and only began to migratebeyond Africa with- tions across a range of sites. Percentages utilized of materi- in the last 100 kya (Hedges2000), then parsimonywould pre- al, by geological form and Late streak,from Pleistocene dict use of red ochrein ritualcontexts as an establishedpart of shelter sequences in southernAfrica are used to investigate early modem humanbehaviour prior to initialmigrations. past selective preferences. Materials with saturated red streaks are disproportionatelyrepresented among utilized However,in Africa,nearly all archaeologicalassociations and pieces, particularly among crayons. Thefindings are most contexts that make a symbolic interpretationof ochre-use consistent with use as in a costly signalling strate- fairly compelling (e.g. ochred ostrich beads, rock- gy involving ritualized display. Theoreticaland substantive painting,ochred ) are Later Stone Age (LSA), post- grounds are givenfor inferringthat the contextfor such dis- datingc. 30 kya (but see Henshilwoodet al. 2002). The issue play was probably collective ritual. is furthercomplicated because ground pieces of red ochrehave much greaterantiquity. In Europe,India and Africa occasional Introduction pieces dateback about250 kya (Knightet al. 1995) while con- The presenceof red ochreor haematitehas been notedat most siderablequantities of ochrehave recentlybeen reportedfrom Middle Stone Age (MSA) shelter excavationsbut, with few severalslightly earlier MSA contextsin south-centraland east- exceptions(e.g. Barham2000), site reportsfail to presentdata ernAfrica (Barham 1998, 2000; McBrearty2001). Most MSA on raw materialvariability, proportions utilized, temporal vari- archaeologistshave inferredthat utilized pieces were groundto ation. In some cases, they fail to presentany quantitativedata producea pigmentpowder for use primarilyas body-paintand whatsoever(Thackeray 1989; Avery et al. 1997).This is unfor- possiblythe decorationof otherorganic surfaces (Tobias 1949; tunate,as understandingochre use has become importantin Mason 1962;Volman 1984; Walker 1987; Clark 1988; Deacon currentre-evaluations of the MSA (Barham1998; Watts 1999; 1995). This interpretationhas rarelybeen robustlypresented McBrearty & Brooks 2000; Deacon 2001) and in wider and more utilitarianroles have been proposed-the most fre- debates concerningthe evolution of symbolic culture(Klein quently raised alternativebeing hide-preservation(Wadley 1995, 1999; Knight et al. 1995; Mithen 1996; Stringer& 1993, 2001; Klein 1995, 1999;Mithen 1999). McKie 1996; Deacon 1997). Evaluation of these contending interpretationshas been Ochreis a generalterm for any ferruginousearth, clay, miner- hamperednot only by inadequatedata but also by under- al or rock containingsufficient haematite (an oxide) or theorization of ochre-as-pigment hypotheses, insufficient ironhydroxide (e.g. )to produce,respectively, either a evaluationof the bases for utilitarianhypotheses and a fail- red or streak(Bateman 1950; Jercheret al. 1998). It is ure to specify divergent archaeologicalimplications. Here, typically a weatheringproduct where residualconcentration these interpretationsare recast in more explicit form so they and oxidationof iron(with or withouthydration) has occurred may direct observationand evaluationof the archaeological in complex mixtures with other minerals (predominantly record. The body of the paper concerns a descriptive study quartz,clay and ). of southern African MSA ochre assemblages. Low-level observations are presented on assemblages from eleven Ethnographically,red ochre is the most widely used earthpig- sheltersites (treatingKlasies River Mouth as a single depos- ment, applied to human bodies and culturalartefacts in the itory) with Late Pleistocene (beginning 128 kya) sequences course of symbolic practice, especially rituals. predatingc. 17 kya. General attributesof the assemblages Archaeologically,red ochre is reported from two of the are outlined, drawingattention to regional variability,selec- world's earliest modem human burials, at Qafzeh in Israel tive preferences (based on rates of utilization among geo- (Vandermeersch1969, 1981) at c. 100 kya (Valladaset al. logical forms and colour categories) and extremes of high- 1988) and the earliestburial in Australiaat c. 62 kya (Thome quality on the one hand and questionableones on et al. 1999). Ochre is not reported from Neanderthalburials the other.Many of the observationswere subjectively made

This content downloaded on Thu, 3 Jan 2013 17:48:58 PM All use subject to JSTOR Terms and Conditions 2 SouthAfrican ArchaeologicalBulletin 57 (175): 1-14, 2002 but I am reasonablyconfident that any biases are fairly con- Body-painting and cosmetics are among the simplest of sistent and that a more standardizedclassification would such transformationaltechniques. provide similar overall results to those reportedhere. These generalpropositions accord with source-sideobserva- Ochre-as-pigment hypotheses tions. The most pervasivedomain of Khoisanpigment use was in ritualbehaviour (Rudner 1982; Watts 1999). Redness and Lacking direct evidence as to how ochre was used in the brillianceare consistentlyassociated with constructsof super- MSA, proponents of ochre-as-pigment have relied on naturalpotency (Watts 1999). A girl's menarchealinitiation ethnography (Tobias 1949; Boshier & Beaumont 1972; was the only ritualwhere red pigments were almostinvariably Deacon 1995), presumed selection for visual salience used and occasionedtheir most socially inclusivereported use (Boshier 1969; Dart & Beaumont 1969; Deacon 1995) and (Knight et al. 1995; Power & Watts 1996; Watts 1998). the crayon- or pencil-like morphology of some abraded Menarchealrituals are also thoughtto providea templatefor pieces (Deacon 1995). Perhaps because the interpretation other rites of passage in Khoisan hunter-gatherersocieties has seemed intuitively obvious, these approacheshave been (Lewis-Williams1981; Knight et al. 1995; Power & Watts poorly substantiated. 1997; 1999). Ethnographicprecedents do not constitute an analogical A Khoisan-basedanalogical argument, with collectiveritual as argument (Lewis-Williams 1991); relations of relevance the connectingprinciple, would predict that early collective (Wylie 1985, 1988;Stahl 1993) linkingsource-side and subject- ritual(unlike speech) should leave a loud archaeologicalsig- side observations are required. In the present context, a nal the habitualuse of red ochre. However, as ochre use backgroundconnecting principle is the genetically inferred remainsthe principleline of evidence bearingupon possible temporal depth of San populations (Vigilent et al. 1991; symbolismin the MSA, the analogyis in dangerof assuming Soodyall & Jenkins 1992, 1998; Ingman et al. 2000), sug- what needs to be demonstrated,namely that MSA ochre use gesting that ancestralSan were responsible for most, possi- occurredwithin a symbolic context. bly all, Late Pleistocene archaeologicalremains in southern Visualsalience provides a way out of this circularity,a linking Africa. A higher-level relation of relevance, alluded to by principlefor an analogicalargument partially independent of Deacon (1995, 2001), concerns the role widely assigned to symbolic cultureand a point of entryfor neo-Darwinianevo- collective ritual (initiationritual in particular)in generating lutionarytheory, specifically, costly signallingtheory (Zahavi and sustaining the symbolic domain (Van Gennep 1960; 1987;Grafen 1990; Zahavi & Zahavi 1997). The linkingprin- Durkheim 1965; Turner 1967; Gellner 1992; Maynard- ciple concernsneurological mechanisms and perceptualbias- Smith & Szathmatry1995; Deacon 1997). Collective ritual es presumedto be commonto bothmodern and ancestral pop- provides the necessary compulsive constraintto ensure that ulations. From the continuumof wavelengths in the visible collective representations constructs that have no reality spectrum,we discriminate(find salient)those corresponding in the perceptibleworld-are faithfully transmitted. to four uniquehues. This is achievedby pitting signals from differentphotoreceptors against one anotherto derivea differ- Ritualization is a strategy of power (Knight 1991; Bell ence signal, such that red is opposed to green and yellow to 1992), whereby status functions (Searle 1996) are collec- (De Valois& De Valois 1993;Abramov 1997; Wooten & imposed on agents, objects and events. Social rela- tively Miller 1997). Otherexamples of perceptualbiases arethat we tions of power are demarcated, identifying groups with respondmore rapidlyto variationin luminancethan chromat- interests and setting them in opposition to other common ic stimuli(Boynton 1979:308)and thatred is the only unique 1954; Bell 1992). To make credible the groups (Leach that remains saturatedin peripheralvision (Gordon & nature of symbolic representations (e.g. a intangible Abramov1977). promise), it is necessary to bring to the fore the intrinsical- ly credible substructureof symbolic social relationships Perceptualbiases constrainthe pragmaticuse of perceptsin (Rappaport 1979; Searle 1996; Deacon 1997; Knight signallingbehaviours, including, in the humancase, symbolic 1999). In this substructure, reference depends not on use and labelling of such percepts (Deacon 1997; Morphy symbolism but on the cognition of relationshipsof associa- 1989). For a visual signal to be effective it needs to be eye- tion (indexicality) or identity (iconicity) and is therefore catching and salient (Guilford& Dawkins 1991; Zahavi & rooted in the perceptible world. Collective ritual typically Zahavi 1997). As boundaryconditions for perception,light- achieves this through regular, formal performances using ness and darknessare necessarily the most salientvisual expe- multimedia effects of gesture, song and dance in invariant riences. However, red is the most common colour signal sequences (Rappaport 1979). The signals employed are deployed in nature, partly because it is highly contrastive characteristicallyeye-catching (Morphy 1989; Clark 1986), (Humphrey1976). Humansfind red andbrilliant percepts par- costly (Sperber 1975), amplified, stereotypedand prone to ticularly eye-catching (Humphrey 1976; Ratliff 1976; massive redundancy (Rappaport 1979). Rituals are essen- Bradshaw& Rogers 1993). Culturesemploy between two and tially deceptive and manipulative (Leach 1954; Lattas eleven basic colour terms (Berlin & Kay 1969; Kay et al. 1989; Bell 1992); to the extent that they convey honest 1997). Binary classificationscomprise composite light/warm information,it is an ostentatiousdisplay of their very cost- and dark/coolterms. Despite red being a colour of relatively liness (Power 1999). Critically, ritualization is also an low brightness(Solso 1994), it is a focus of the light/warm embodied experience (Bell 1992), with transformationsof term, along with and yellow (Heider 1972; Kay et al. body surfaces playing a key role (Turner1980; Gell 1993). 1997). In the classic studyof such a system (Heider1972), the

This content downloaded on Thu, 3 Jan 2013 17:48:58 PM All use subject to JSTOR Terms and Conditions South African ArchaeologicalBulletin 57 (175): 1-14, 2002 3 most commonly selected focal point (best example) of the symbolic culture,see Knightet al. 1995; Knight 1998, 1999; light/warmterm was not white, as Berlin and Kay (1969) had Power & Aiello 1997; Power 1998, 1999). However, where proposed,but a saturateddark red, followed by a saturated pigmentuse can be shown to be habitualand ubiquitous,we pale . As culturesacquire additional colour terms, red is are probablycloser to collective ritualthan individualritual- invariablythe firstchromatic colour to be named,followed by ized display,with iconic and/orindexical reference now sub- either yellow or green and then blue (Kay et al. 1997). sumed within an over-archingweb of symbolic meaning. If Neurologicalbiases can accountfor the namingof uniquehues utilitarianhypotheses prove inadequatein accountingfor reg- before otherbasic colours (e.g. brownor )and explain ularuse (see below), then suchuse may meet a relaxedversion why maximally saturatedcolour chips are selected as best of one of Chase& Dibble's (1987, 1992) archaeologicalcrite- exemplarsof basic colours. They may also explain why, in ria for identifyingsymbolism, namely, repetition of form. comparisonto blue and green(short-wavelength colours), red So-called crayons or pencils are widely reportedfrom MSA and yellow (long-wavelengthcolours) elicit high levels of sheltersites (Watts1998). If the morphologyof such pieces is cross-culturalagreement regarding a narrow range of best not simply the resultof intensivepowder reduction and rel- examples (D'Andrade 1995). However, such biases cannot if, ative to the backgroundpopulation, they have colour proper- accountfor variationin the numberof basic terms, nor why ties consistentwith preferentialselection, then the use red is invariablythe firstunique colour to be named.The point implied by their designation-creating colour patterns-becomes I wish to stressis thatcolour reference has evolved in adapta- moreplausible. This would meet anotherof Chase& Dibble's tion to boththe humannervous system (with its biases)and the (1992) proposedcriteria: grounds for inferringintent to pragmaticconstraints of human use (Deacon 1997). Such pro- duce a visual design. With crayonssometimes from pragmaticconstraints are thought,in the first instance,to con- reported LSA contexts (e.g. Leslie-Brooker 1987; Kaplan 1990; cern ritualized performance(Turner 1967; Morphy 1989; Mitchell 1993) and scanty historical O'Hanlon 1989; Knight 1999). observations(Rudner 1982), the category may also meet their proposed test for To inferthat red ochrewas collectedfor its visual saliencepre- inferring symbolism, that there be analogues from known supposesa signallingcontext. Habitual procurement, process- symbolic contexts. ing and use of red ochre in the MSA would therebyimply a Ochre as hide preservative stereotyped,costly, volitional signalling behaviour. According to Zahavi's handicap principle (Zahavi & Zahavi 1997), The hide-preservationhypothesis, provoked by the occasional stereotyped,costly signals typically occur in contexts of ritu- discovery of traces of ochre on endscrapersand blades, has alized display.In the animalworld, the characteristicfeatures been exploredprimarily in relationto late Upper Palaeolithic of such displayare similarto thosejust summarizedfor human and Mesolithic contexts in Europeand Israel (Keeley 1978, ritual,except that they are ego- ratherthan socio-centricand 1980; Audouin & Plisson 1982; Moss 1983; Vaughan1985; reference remains exclusively indexical. Ritualized signals Dumont 1988; Bueller 1993; Phillibert 1994). Withoutany mustbe amplifiedand costly in orderto be effective,signallers similarengagement with the archaeologicalrecord, it has been andreceivers having different interests. Signallers, advertising raised severaltimes as a possible alternativeinterpretation of individualquality, exploit perceptualbiases. Receiversevalu- MSA ochre use (Wadley 1993, 2001; Klein 1995, 1999; ate the credibilityof the signal indexically,in terms of the Mithen 1999). The bases of the hypothesis (experimental, hard-to-fakecurrency of high costs (the handicap).Stereo-typ- ethnographicand archaeological)have been reviewed else- ing and formalizingof the signal are positively selected by where (Knightet al. 1995; Power & Watts1996; Watts1998, receiversto facilitatecomparison between signallers. 1999) and are only summarizedhere.

Durkheimianand neo-Darwinianparadigms, while method- While laboratoryexperiments indicate that, at sufficientcon- ologically opposed, convergein theirrespective characteriza- centration,most metal ions have an inhibitingeffect on bacte- tions of ritualizedsignalling. If it can be shown thatred ochre rial productionof collagenase,a collagen-destroyingenzyme was used as pigment, we have compelling, independently (Mandl 1961), field experimentsfailed to demonstratethat derivedtheoretical grounds for inferringthat its use evolved in ochre had any preservativeeffect (Audouin& Plisson 1982) contexts of ritualizeddisplay. With a robust causal mecha- and taxidermistsdoubt its efficacy (Phillibert1994). Claimed nism, an ethnographically-basedanalogical argument for ritu- ethnographicsupport for the hypothesis(references in Keeley al use of pigmentsas body-paintsor cosmetics can be placed 1980;Audouin & Plisson 1982;Moss 1983;Vaughan 1985) is on a firmerfooting. Combinedwith the findingsof cognitive ambiguous.Several cited accounts do not concernhide-working anthropologistsand visual scientists regardinguniversals of (e.g. Roth 1890; Bonwick 1898; Sollas 1924 cited in Keeley colour-labellingand perceptualbiases, an early and enduring 1980). Wherea functionalrole is implied,this seems to con- interestin red and a preferencefor saturatedreds would also cern the use of fats and greases,which are known to have an be predicted. emulsifyingeffect (Audouin& Plisson 1982) but which also often served as a medium for applying red ochre. The What indexical salience did redness originally have? What involvementof ochre in hide-workingis almost invariablyat conflicts of interestprovided the selection pressurefor such the finishingstage as a decorativeinclusion, a generalization costly displays?How did a ritualizedsignal in the humancase amplysupported by Khoisandata (Rudner 1982; Watts1998). become the property of a coalition?These interestingques- In Eurasianlater Upper Palaeolithic and Mesolithicmanufac- tions lie beyondthe scope of the presentpaper (for discussion ture of prestige goods like buckskin (Hayden 1990, 1998), of the sham menstruationhypothesis concerning the origin of similardecorative inclusions could have precededfinal tailor-

This content downloaded on Thu, 3 Jan 2013 17:48:58 PM All use subject to JSTOR Terms and Conditions 4 South AfricanArchaeological Bulletin 57 (175): 1-14, 2002

ing, which may account for Bueller's (1993) and Dumont's pigmentsis given in the Appendix. (1988) micro-wear determinations regarding ochre polish on The need for a large sample and the limited time available blades. Ochred endscrapers at several sites had a specialized meant that most observationswere made subjectivelyrather morphology (Keeley 1978; Phillibert 1994). Replicative than based on standardizedmeasures. Although not recom- experiments led Phillibert (1994) to interpret these scrapers as mended, this approachhad the merit of approximatingthe having been used with an ochre abrasive to raise the nap of kind of common sense discriminations(Atran 1990) or folk already tanned buckskin. Production of prestige goods is taxonomies(D'Andrade 1995) thatpeople employ in classify- unlikely to be informative about much earlier contexts of ing salientattributes of the naturalworld, such as colour. ochre use. Methods Mandl's (1961) experiments imply that the hide-preservation hypothesis should be treated as null with respect to colour Geologicalform selection; materials rich in manganese or goethite should be as Whereverpossible, the form of both typical and unusualspec- effective as haematite. With little difference in iron content imens was briefly discussed with either the excavatoror the between goethite and haematite (61% and 70% respectively analyst. However, I frequentlyfound it impossible to assign [Taylor et al. 1988]) and both minerals being common weath- pieces (particularlysmall debris)confidently to any geological ering products, we might expect both to have been widely form, so ochre is necessarilya defaultas well as a generic cat- exploited if use was primarily utilitarian. Similar considera- egory. The defaultassessments were: tions apply regarding proposals that ochre served as a sun- * fine- to medium-grainedtexture (fine equivalentto shale block or barrier (Bahn & Vertut 1988; Wadley 1993). and mudstone, medium equivalent to siltstone or fine For such to stand proposals as alternative general accounts for sandstone) ochre it use would need to be shown that other clay or mud * fairly soft (an approximaterange of 2-4 on Mohs hardness with applications, or without grease, failed to provide similar scale) protection. * sufficientlyferruginous to leave a red or yellow streak,but The study sample not such as to be unusually dense (making a haematite designationmore likely). South of the Limpopo River, archaeologists widely employ Volman's (1981, 1984) informal techno-typological frame- This defaultcategory inevitably contains enormous variability work of MSA sub-stages. I previously employed this frame- (Watts 1998). Twenty-eightother categorieswere also used, work to investigate change in the relative frequency of ochre some conveyingthe continuumof change in beddingand tex- across a large number of shelter sites (Watts 1998, 1999). The ture from mudstoneand shale to fine sandstone,rather than 11 sites reported here represent the majority of excavated deep indicating any geological species. Mudstone also contains sequences in the region, biased towards earlier (MSA 2) sub- considerablevariation, partly because it embraces massive, stages. For present purposes, the examined material is treated fine-grainedforms with light-brownstreaks. Specularite is a as an aggregated sample, the vast bulk representing sub-stages glittery, micaceous expression of haematite. In Table 2 the falling within the Late Pleistocene, bracketed by a few termi- originaldesignations are groupedinto 11 categorieswith their nal Middle Pleistocene MSA 2a assemblages, e.g. Border respectivesum weights. Cave (GrUn & Beaumont 2001) and some MSA/LSA transi- tional and early (>17 kya) LSA assemblages. The sub-stages will only briefly be touched on here. However, given the inter-

pretative significance placed on habitual ochre use, it must be OLIEBOOWOORTi M U'S stressed that probable Middle Pleistocene MSA assemblages CANIW in the region, whether designated MSA 1 or MSA 2a, have , .ts,, BUSBAGN produced either no ochre or only isolated pieces (Watts 1998). ROCKSHELTER 26? By contrast, younger MSA 2a assemblages, such as the basal APOLLO -__ * LBS member at Klasies River Mouth, dating to early in the ? 1 by BORDER / Late Pleistocene at c. 110 kya, show regular use of ochre, as J COTTAGE CAVE / . _{tt do all subsequent sub-stages (Watts 1999). -- 0

SEHONGHON X The derivation of a sample of 4038 examined potential pig-

ments is presented in Table 1, along with total and mean X HOLLOW RAOICK weights by site. Site locations are shown in Fig. 1. sm~~uim ~ Z.ANA7 Numerically, by far the largest contributions are from BOOAVLAAS Umhlatuzana and Hollow Rock Shelter. By mass,

Olieboompoort accounts for half the sample. The finely- .A, ~~~~~~~~~~~~~~~~34o\- KLASIES RIVER sieved Rose Cottage Cave Gail assemblage was not included in the original database (Watts 1999). It is included in Tables 18? 260 1 and 5 for comparativepurposes but is excludedfrom colunmn totals. A breakdownof the sample by excavationunits, with techno-typologicalsub-stage designations, counts of potential Fig. 1. Map of southernAfrica showing the locationof sites pigments and of the combinedtotal of definite and probable providingthe examinedassemblages.

This content downloaded on Thu, 3 Jan 2013 17:48:58 PM All use subject to JSTOR Terms and Conditions South African Archaeological Bulletin 57 (175): 1-14, 2002 5

Table 1. Derivationof examined potential pigment assemblages. 'Validn' refers to the numberof weighed pieces.

Mesh (mm) Potential Valid n Sum (g) Mean mass (g) s.d. Reference pigment (n) SHG+ 1.5 99 99 66.2 0.7 1.1 Mitchell 1995 RCC (Gail) * 2.0 296 295 243.8 0.8 2.0 Clark 1997 BC 1987 2.0 105 105 292.1 2.8 6.5 Unreported APlI 2.5 & 4.0 105 105 889.4 8.5 15.0 Vogelsang 1998 BP 3.0 134 133 1340.0 10.1 21.2 Deaconetal. 1983 BC 1970+ 3.0 6 6 39.5 6.6 4.4 Beaumont 1978 HRS 3.0 1143 1123 1345.3 1.2 4.0 Evans 1994 BRS 3.0 58 41 907.7 22.1 40.5 Unreported UMH 3.0? 1721 1675 3436.2 2.1 3.7 Kaplan 1990 KRM1A (H-P) 6.5 179 163 1268.5 7.8 11.0 Singer & Wymer 1982 RCC not known 112 112 1335.1 11.9 19.9 Wadley & Harper 1989 MC not known 13 13 477.8 34.5 27.5 Tobias 1949 OBP not known 304 304 11953.1 39.3 65.5 Mason 1962 KRM1A (non-H-P) 13.0 19 19 293.0 15.4 12.8 Singer & Wymer 1982 KRM1 13.0 37 32 1419.0 44.34** 75.4 Singer & Wymer 1982 KRM1B 13.0 3 3 100.5 33.5 6.3 Singer & Wymer 1982 Total 4038 3933 25163.4

+SHG = Sehongohong;RCC = Rose Cottage Cave; BC = BorderCave; API I Apollo 11; BP Boomplaas;HRS = Hollow Rock Shelter; BRS = Bushman Rock Shelter;UMH = Umhlatuzana;KRM = Klasies River; MC = Mwulu's Cave; OBP = Olieboompoort. * Rose Cottage 'Gail'(Transitional MSA/LSA) could not be included in the main databaseand is excluded from column totals. ** Klasies Cave 1 has an outlier value of 437.2g, if this is removed, the mean falls to 31.7g with an s.d. of 23.8g.

Colour Othervariables The colour of a potential earth pigment may be determined by Hardnesswas assessedin the courseof streaking,with subjec- its streak-the colour of its powder, usually taken by rubbing a tive designationsoccasionally comparedto Mohs hardness sample on unglazed porcelain (hardness 7 according to Bates & scale.No quantitativemeasures of textureor ironcontent were Jackson [1984]). In this study, white vein quartz was found to employed. Subjective assessments of iron content, based give satisfactory results and was more convenient. As far as largely on relativedensity, were useful for highlightingvari- possible, observations were made under bright natural light. abilitywithin gross geological forms.For example,the dens- Although streaking requires very little material, it is a destruc- est haematiteswere classified as ore-gradematerial, while tive procedure and testing of utilized surfaces must be avoided. haematisedshales were describedas highly ferruginousto dis- tinguishthem Mineralogists describe streak adjectivally; haematite, for from the defaultassessment of ferruginous. example, has a red- streak (Lurie 1977). However, in a The four originalassessments regarding the presenceof use- study of potential earth pigments, the streak of most ferrugi- wear traces(definite, probable, possible and absent)are treat- nous forms will be brownish, with haematite redder than most ed here as two pairs.Use-wear traces were themselvesdivid- ochre, hence the emphasis on the redness of streak here. ed into two broadcategories, abrasive (grinding, rubbing, pol- ish) and fixed-positionutilization (scraped, Forty-seven adjectives or combinations of adjectives were scored, notched). Grindingwas loosely defined as flat facets used in the original classification (Watts 1998), here grouped bearingmultiple, fine, parallelstriations. Intensively ground into 10 categories (Table 3). The category strong- pre- pieces with multi- ple facets convergingto a point were categorized dominantly comprises streaks I perceived as poppy, -red as crayons. Rubbedor polishedwere termsapplied where and dark-red. Two additional categories of various reds pieces appeared to have artefactualsurfaces but where striae (n=1904) and various colours (n=57) relate almost exclusive- were only very faint or not detectable.Both terms need ly to collectively described unmodified pieces from HRS and refining but, unlike Couraud's(1991) study, where such Umhlatuzana, accounting for 1902 of the 'various reds' and 42 use-wear traces were imputedto of the 'various colours'. reflect differentmodes of use relativeto grinding, here they are viewed as primarilyreflecting hardness proper- Ochre studies have employed a variety of colour notational ties of the pigments.Scraped or scoredpieces hadwider, deep- systems (Jercher et al. 1998; Hughes & Solomon 2000; er striae,which were less evenly alignedand less densely dis- Henshilwood et al. 2001), any of which would have provided tributedthan grinding striae. Inferences about the likelihoodof greater standardization and precision than adjectival descrip- pieces havingbeen collectedas pigmentswere made at sever- tion. However, comprehensive matching would have greatly al levels of confidence.The most robustwere based on obser- reduced the sample size. Regardless of method, some consis- vations of use-wear in combinationwith streak and texture tency problems would remain (e.g. variable lighting condi- (coarse-grainedforms, having poor pulverulence[cf. Brabers tions). While adjectival description of streaks is indicative of 1976] will generallymake poor pigments).Probable and pos- hue, it is less useful for brightness and saturation, the English sible designations(largely relating to unmodifiedpieces) were language having sacrificed brightness-based discriminations arrivedat more contextually,where the assessmentdrew upon in favour of hue-orientated ones (Casson 1997). streak,texture and the frequencyof use-wearacross geological

This content downloaded on Thu, 3 Jan 2013 17:48:58 PM All use subject to JSTOR Terms and Conditions 6 SouthAfrican Archaeological Bulletin 57 (175): 1-14, 2002 forms.The groundsfor less confidentor negativeassessments The predominantcategory in the Easternand WesternCape will be illustratedas the overallsample is characterized. sites (Klasies River Mouth, Boomplaas and Hollow Rock Shelter) is shale (57.3%, 56%, 45% respectively),much of Results which might be better described as well-bedded siltstone Raw materialvariability (Henshilwoodet al. 2001). Commercialpigment quarrying in Table6 tabulatesgeological formassessments by site. Despite SouthAfrica (De Villiers 1959) has primarilyfocussed on red the loose parametersof default ochre, most assemblagesare and yellow associated with weathered Bokkeveld clearlyheterogeneous, consistent with archaeometricobserva- shales, which are widely distributedthrough the southernand tions elsewhere(Smith et al. 1998;Hughes & Solomon2000). southwesterncoastal regions (Visser 1937). Two commentson the groupedcategories are needed before discussing general patterns.Although shale is predominant Default ochres predominate at the remaining sites. The (39.7%) at BushmanRock Shelter,the original designation absence of magnetiteand rarityof manganeserules out any was shaley-mudstone almost exclusively restrictedto this significantutilitarian role for these materials,suggesting that site. On a range of grounds,most notably the forms of use- manganeseshould be treatedas a minorclass of earthpigment. wear,most of this materialwas judged to have servedprima- In this and several otherrespects (Power & Watts 1996), the rily as hammers,burnishers or rubbersrather than as pigments MSA record of metal oxide use differs from its Mousterian (Watts1998). The groupedsandstone category (n=77) comes counterpart,suggesting that the two shouldnot be elided (con- almost exclusively from two sites: at Klasies River Mouth, tra Klein 1995). 92% of this material(n=50) was originallydesignated shaley- Mesh size and utilizedproportions sandstone or mudstone-sandstone,explicitly finer textured Before the 1970s, most MSA assemblages were coarsely thanthe ferruginoussand (n=21) and sandstone(n=3) at Rose sieved (mesh >4 mm). Quarteror half inch mesh can be CottageCave. inferredfor the Not known values in Tables 1 and 2. This Table2. Proportions of grouped geological categories, affects the mean weight of pieces (Table 1) and the represen- numericallyand by weight. 'Validn' refers to numberof tationof utilizedpieces (row marginalpercentages in Table5). weighed pieces where geological form was determined The most finely sieved assemblages(mesh <3 mm) have low utilized percentages(<6%). Of the coarsely sieved (>4 mm) GroupedGeol. % total n Weight sum (g) % total mass assemblages, with the exception of Olieboompoort (see Category (valid n) below), between 29% and 70% of materialbore signs of uti- Default ochre 56.5 2217 5002.4 19.9 lization.Among the 3 mm-sieved assemblages,utilized per- Mudstone 6.0 239 598.7 2.4 centagesrange from 8%to 39%.Much of this variationseems Sand/stone 1.9 76 1645.8 6.5 to reflecthuman behaviour rather than taphonomic or excava- 12.6 Shale 19.2 753 3160.3 tion biases. At Hollow Rock Shelter,for example, while the 0.4 16 132.2 0.5 Grit overall mean weight is comparablewith values from the Haematite 4.1 161 1475.2 5.9 finest-sievedassemblages, the mean for utilizedpieces is sta- Specularite 7.7 307 12180.2 48.4 from counterpartsat most other 3 Manganese 0.4 16 35.1 0.1 tistically indistinguishable because Other 1.5 54 747.4 3.0 mm- sieved sites (5% level of Watts[1998[). Perhaps M2 Carbonate 0.3 14 41.0 0.2 of the very small size of this shelter(c. 27 ) and the conse- Not known 2.0 78 141.4 0.6 quent redundantuse of space, we are simply seeing much n or total g 4002 3931 25118.7 more small processingdebris. The small areawould have pro- Missing 36 107 hibitedhide-processing.

The distributionof the more distinctive forms agrees with Table3. Frequenciesof groupedstreak values. the broad outlines of regional geology, ethno-historical Missing values = 42. accounts of pigment exploitation (Beaumont 1973; Watts 1998) and commercial pigment quarrying(Brabers 1976). In the highly metalliferousrocks of Limpopo and Mpuma- GroupedStreak n o langa Provinces (Olieboompoort,Mwulu's Cave, Bushman 16 0.4 Rock Shelter), haematite and specularite predominate. 22 0.6 125 3.1 Haematitealso figures prominentlyat Rose Cottage Cave in Orange & yellowish- browns 199 5.0 the eastern Free State and in KwaZulu-Natal. Light Mid & darkbrowns 55 1.4 The Border Cave haematite (several pieces designated as Light reds 663 16.6 describedas [Watts1998]) may ore-grade,a couple specular Strong reds 720 18.0 be of particularsignificance; the nearest source of such Reddish-browns 176 4.4 materials is thought to be Ngwenya Ridge and the MSA 12 0.3 specular-haematitemine of Cavern, 120 km to the Greys 8 0.2 northwest (Beaumont 1978; Watts 1998; see also Barham Various reds 1904 47.6 1989 re Sibebe). Elsewhere, haematiteis never more than a Various 57 1.4 minor component. Specularite, except for two pieces at No streak 39 1.0- Apollo 11, is restrictedto sites northof the Vaal River. Total 3996 t00.0

This content downloaded on Thu, 3 Jan 2013 17:48:58 PM All use subject to JSTOR Terms and Conditions South AfricanArchaeological Bulletin 57 (175): 1-14, 2002 7

Utilizationand geologicalform effect of Umhlatuzana(where mean weights are comparative- The within-site frequencyat which differentmaterials were ly low and much of the ochre was judged to be of fairly low utilized can help identify preferentialselection as well as quality).Exclusion of this site increasesthe percentageutilized unlikely pigments.Table 5 provides frequenciesof use-wear to 15.6%.The very largeHollow Rock Shelterassemblage has within raw materialcategories at each site (arrangedby mesh a similarlydepressing effect on the proportionof utilizedshale. size) andthe percentagethis representsof the respectivemate- Exclusionwould see an increasefrom 14%to 25%,intermedi- rial at the site. Withonly two specimensof calciumcarbonate ate between the range of values for the majorityof materials or manganesedefinitely utilized, these formsare here grouped and haematite.Exclusion is warrantedfor indicativepurposes into the Othercategory. on the groundsthat shale contributesdisproportionately to the Severalsite-specific effects stronglyinfluence the overallrates small debrisat Hollow Rock Shelter(Watts 1998). The large of utilizationwithin geological forms (column total percent- variationin rates of utilizationamong the three reasonably ages in Table 5). The overall frequencyof utilizationamong sized mudstone assemblages is largely attributableto site- defaultochre (8.4%) is severely depressedby the samplesize specific factorsconcerning colour (see below).

Table 4. Percentagesof groupedgeological forms by site, groupedby geographicregion. See Table 1 for full names of sites.

Geological Limpopo& Mpumalanga Western& EasternCape FreeState Lesotho KwaZulu-Natal Namibia Form OBP MC BRS HRS BP KRM RCC SHG BC UMH AP Defaultochre 1.0 15.4 12.1 30.9 38.1 8.8 56.4 88.9 40.5 93.2 35.2 Mudstone - - - 13.4 - 1.3 0.9 - 0.9 3.9 14.3 Sand/stone - - - 0.1 - 22.0 21.8 - - - 1.9 Shale - - 39.7 45.2 56.0 57.3 - 2.0 - 0.1 25.7 Grit - - - 0.5 - 3.5 0.9 - - 0.1 - Haematite 1.3 38.5 29.3 3.3 - 4.0 18.2 6.1 30.6 1.7 1.0 Specularite 97.7 46.2 3.4 ------1.9 Manganese - - - 1.2 - - - - 0.9 - 1.0 Other - - 15.5 0.1 5.2 2.2 0.9 3.0 27.0 0.2 1.0 Carbonate ------13.3 Not - known - - 5.2 0.7 0.9 0.9 - - 0.7 4.8 n= 304 13 58 1123 134 227 110 99 111 1718 105

Table 5. Modification across grouped geological forms by site, arrangedin orderof increasingmesh size. Mod' = number of modifiedpieces, % M = modificationas percentageof respectiveraw materialcategory at the site, derivedfrom Table4, column total percentagesare percentagesmodified for all sites (excluding Rose Cottage level Gail). See Table I for full names of sites.

Site Mesh GroupedGeological Form All mod' St of site size (mm) Dflt'Ochre Mudstone Sandstone Shale Haematite Specularite OtherA Not Known Mod' %M Mod' %M Mod' '4 M Mod' %M Mod' %M Mod' %M Mod' %M Mod' '} M (n) Total SHG - 1.5 - 0.0 - - I 16.7 - - - I 1.0 RCC 2.0 - - 10 6.3 2 3.8 4 9.1 - - - 16 5.4 (Gail)* BC 2.0 1 2.2 1 100 - - 5 14.7 - 2 6.7 - 9 8.1 3.0 HRS 3.0 41 11.8 5 3.3 - 42 8.3 3 8.1 - 2 13.3 1 1.7 94 8.4 UMH 3.0? 86 5.4 34 50.7 - - 17 56.7 - 3 75.0 2 16.7 142 8.1 BP - - 3.0 13 25.5 12 16.0 - - - I 14.3 - 0.0 26 19.4 BRS 3.0 - - 3 42.9 5 22.7 10 58.8 - 4 44.4 - 22 38.6 APIl 2.5 - 12 32.4 7 46.7 8 29.6 - 0.0 2 100.0 2 12.5 - 31 29.5 4.0 KRM - 6.5 2 40.0 1.3 44.8 30 26.3 3 37.5 - - - 48 28.6 (H-P) RCC - n.k. 20 32.3 2 8.3 - I1 55.0 - - - 33 30.0 MC ii.k. - - - 1 50.0 2 40.0 4 66.7 - - 7 53.8 OBP - n.k. 3 100.0 - - 2 50.0 35 11.8 - - 40 13.2 KRM (non 13.0 7 58.3 2 100.0 17 81.0 10 62.5 1 100.0 - 2 100.0 - 39 69.6 H-P) All sites 189 8.4 49 20.5 32 34.8 107 14.0 55 34.1 41 13.6 16 17.6 3 3.8 492 12.3

* Rose Cottage level 'Gail'is excluded from column totals of modificationand the percentagesmodified A for all sites. Manganese and Carbonate are here grouped into 'Other'.

This content downloaded on Thu, 3 Jan 2013 17:48:58 PM All use subject to JSTOR Terms and Conditions 8 SouthAfrican Archaeological Bulletin 57 (175): 1-14, 2002

Given that coarsermaterials tend to make poor pigments, a However, the presence of crayons raises a possible qualifi- surprisingly large proportion of sandstone was utilized. cationto such an inference.As originallycoded (Watts1998), However,29 of the 32 utilizedpieces are fine sandstonefrom these intensivelyground pieces (n=48) accountfor 10%of the Klasies River Mouth,the high incidenceof utilizationacross utilizedsample. Eleven of these were deemedprobable rather all forms at this site being attributableto extremely coarse thandefinite examples but an additionalfour cases were iden- sieving. Of the remainingsample (n=42), only three pieces tified aftercoding (Watts1998). While disproportionatelyrep- were utilized. resented among coarsely sieved assemblages,crayons were encounteredin most sheltersand across all sampledMSA sub- Once the exceptionally high sandstone value has been stages (Watts1998). All majorgeological formsare represent- accountedfor, a higher proportionof haematitewas utilized ed (Watts1998) but Defaultochre and shale predominate;the (34.1%)than any othercategory. The low utilizedpercentage morphologycannot, therefore, be explainedby the grindingof at Hollow Rock Shelter,below thatof severalother categories generallyharder forms such as haematite.Equally, the mor- at the site, is depressedby 20 small nodules, 15 of which had phology of most examples, along with examples of bevelled orange-brownstreaks and would betterhave been classifiedas edges (Watts1998), is unlikelyto be accountedfor simply in (grouped into Other).Despite this, MSA people evi- terms of the requirementsof powder production(where we dentlyregarded haematite as havingespecially attractive prop- could expect maximizationof the surface areas of ground erties. The high rate of utilization,particularly at sites where facets). Althoughreplicative experiments are required,these haematiteis not common,also suggestsit may oftenhave been observationssuggest that at least some pieces were directly of non-localorigin. abradedon surfaces to be coloured and that there was a requirementfor defined areas or lines of colour, consistent That relatively little speculariteexhibited use-wear (13.6%) with design. Colourselection amongthe crayons(see below) can be attributedto the influence of Olieboompoort,which furthersupports such an inference. accountsfor 96.4% of the sample.The sheermass of material (c. 12 kg) at this site indicatesthat the source outcropwas Scrapingor scoringwas only identifiedon 14 specimens,fre- nearby.The sheltermay have serveda special role in the pro- quentlyin combinationwith grinding(Watts 1998). A tighter curementof this brilliantpigment. The low utilized percent- definition(Henshilwood et al. 2001) would probablyincrease age, compared to the overall haematite sample and other the representationof this form of use-wear.The functionof coarsely-sievedassemblages, suggests that procurementand scrapingremains unexplored; one possibilityis thatthe scrap- caching may have been more importantthan processing.Of ing implement,loaded with ochre powder,might have been the remaining 10 pieces of specularite,from three sites, six used to applypigment directly. At least one repeatedlyscored were utilized, indicatingmore generally,that specularitewas piece from Klasies River Mouth (MSA 2a) can also be at least as likely to be utilizedas haematite. describedas deliberatelyengraved, with the connotationof design. (In Knightet al. [1995] the captionto Fig. 5 incorrect- the and only 3.8% of the Not Known None of grits (n=16) ly assignsthis piece to layer 33; this shouldbe layer38 [Watts were utilized. If utilized percentagesare taken pieces (n-79) 1998]). Subsequentresearch at Blombos (Henshilwoodet al. of the esteem in which differentmaterials as proxy indicators 2002) has provided examples of unequivocal geometric then the resultingranking is very similarto region- were held, engravings on ochre from Still Bay layers. Fragmentsof valuationof differentearth al ethno-historicallydocumented ostricheggshell paintedwith red ochre were recoveredfrom Beaumont 1973; Rudner 1982; Watts pigments (How 1962; the MSA 3 layer at Apollo 11 (Vogelsang 1998), several of the assemblagebiases just discussedinto 1998, 1999). Taking which indicatedeliberate design (Watts1998). account, speculariteand haematiteare most likely to be uti- lized, probablyfollowed by shale, otherochres (including fine In summary, most utilized pieces were probablyground to sandstone)and otherpigments. producepigment powder. However, the crayonssuggest that some pieces may have been used to executedesigns, with rare Modeof use examplesof geometricallyengraved ochre andpainted ostrich utilized Of the 480 pieces deemed definitely or probably eggshell directlytestifying to abstractdesign. (Watts 1998, excluding open sites), 80.4% were definitely ground (29 in combinationwith other use-wear traces) and Colourselection 10.2%were probablyground (three in combinationwith other Reds account for 86.6% of the total streakvalues; yellows, traces).Traces of utilizationwere sometimesharder to identi- orange-browns and yellowish-browns account for 3.7%; most fy on the few very soft materials(most notablyat Apollo 11), other browns (excluding reddish-browns)account for consistentwith Armstrong's(1931) observationsat Bambata of the remainder(Table 6). Although streak should not be Cave. Most of the rubbedpieces were soft, while polish was used in isolation to distinguish between iron oxides and indica- largelyrestricted to hard,crystalline haematitic forms. Ground hydroxides, streak assessments can serve as proxy pieces were presumablyabraded on stone.Ochred grindstones tors of their relative contributionin the absence of minera- arepresent at KiasiesRiver Mouth, Apollo 11, Olieboompoort logical analysis and given the crudenature of the classifica- and BushmanRock Shelter(Watts 1998), with furtherexam- tion of geological form.On this basis, iron hydroxides(yel- ples at Die Kelders (Avery et al. 1997) and Blombos (pers. lowish streaks)were rarelycollected. This is unlikely to be obs.).Although the evidenceis infrequent,grindstones suggest entirely attributableto encounter rates. Yellow ochres are that ochre was generallyground to producea powder rather widely encountered where Bokkeveld shales have been thanabraded in the course of directapplication. deeply weathered(Visser 1937) but are rarein archaeological

This content downloaded on Thu, 3 Jan 2013 17:48:58 PM All use subject to JSTOR Terms and Conditions South AfricanArchaeological Bulletin 57 (175): 1-14, 2002 9

sites of the Westernand , while red ochreous attributableto the non-pigmentcategory of shaley-mudstone shale or siltstone of similar origin is common. at BushmanRock Shelter(Watts 1998) and a small sub-group of high-quality micaceous mudstone pigments from Table6. Colour proportionsfor entire and sample utilized Umhlatuzana,many of which were describedas having rich pieces (having grouped all reds). values = Missing 42. streaks.It is also possible that in collectively describingthe No streak = 3 streaksof much of the unmodifiedochre from Hollow Rock % of total % Shelter and Umhlatuzanaas various reds, I inadvertently Whites 0.4 0.4 included some pieces that should more appropriatelyhave Yellows 0.6 0.8 been placedin eitherof these perceptuallyadjacent categories. Orange & yellowish-browns 3.2 4.5 Nevertheless, among the larger colour groupings (where Light browns 5.0 3.1 n=>100), the most striking difference between overall per- Reds 87.5 87.7 centagecontribution and contribution to the utilizedtotal is the Mid- & dark-browns 1.4 3.3 under-representationof light-brownsamong the latter.Most 0.3 0.2 unmodifiedmaterial in this category is mudstone(n=98) or 0.2 - shale (n=45) fromHollow Rock Shelter,originally designated Various colours 1.4 - as having a streakand interpretedas less ferruginous Total n 3957 489 expressionsof materialsintroduced as pigments,i.e. pigment Three-quartersof the haematitehad strong-redstreaks (Watts waste (Watts1998). Eight of the 15 utilized light-brownshad 1998). The appearanceof speculariteis dependenton mica an originalstreak designation of pinkish-brown.A tighterdef- abundanceand the mannerof processing;when highly mica- initionof the groupedcategory would thereforehave produced ceous speculariteis pounded,the productis glitteryblack but an even lower rateof utilization. when it is ground the streak is generally a sub-metallic Light-browns greyish-red (Watts 1998). The range of streaks among the aside,the broadcomparability between the con- tribution principalargillaceous and arenaceousforms is much broader of groupedstreak categories to the overall and uti- but, with the exception of mudstone,the distributionsremain lized totals is consistentwith colour-basedselection across a diverse focussed on reds. The anomalous position of mudstone is range of raw materials.Utilitarian hypotheses requir- mainly attributableto the large contributionfrom Hollow ing a high metal ion content are inconsistentwith the occa- sional utilization Rock Shelter, where much of the mudstone had a light- of pieces with pale streaks (e.g. pinkish- brown streak(see below). browns) and with the use of calcium carbonateat Apollo 11 (calcium being one of the few metals having no inhibitory Table 7. Percentagesof colour categories for all potential effect on collagenase[Mandl 1961]). pigments and for utilized pigments, excluding Hollow While the collection of red ochre Rock, Umhlatuzana,and non-pigment 'shaley-mudstone' was clearly the overriding from BRS. No streak = 5. concern, the crudely aggregated data do not indicate any preferentialutilization of reds. More interestingpatterning % of No utilized % utilized % of total emerges if we exclude the non-pigmentcategory of shaley- total utilized mudstone from Bushman Rock Shelter and all of the Whites 1.4 2 12.5 0.8 Hollow Rock Shelter and Umhlatuzana material, where Yellows 1.9 4 18.2 1.6 collective description led to considerable information loss Orange & yel- 3.7 8 18.6 3.2 lowish-browns (Table 7). More than two fifths of the remaining sample Light browns 2.4 4 14.3 1.6 (n=1 151) had light-red streaks and a third had strong-red Light reds 44.8 92 17.8 36.7 streaks,but less than one fifth of the light-redswas utilized, Strong reds 34.1 133 33.8 53.0 comparedto a third of the strong-reds-more than half the Various reds 0.2 - - - utilized total (n=25 1). This preference cuts across the main Reddish-browns 6.9 8 10.1 3.2 raw materialcategories (Watts 1998) so it cannot be attrib- Mid- & dark-browns - - - 2.2 uted to the abundance of unmodified specularite nodules Blacks 0.3 - - - from Olieboompoort (mostly coded as light-red streaks). Greys 0.7 - - - Site-specific sub-groups where strong-reds Various 1.3 - - predominate, such as haematised shale at Klasies River Mouth Total n 1151 251 251 (n= 16) and ore-grade haematite at Bushman Rock Shelter (n-=7), show exceptionally high rates of utilization (75% and 57% Collapsingall reds in orderto accommodatethe unmodified respectively). The same preference is particularly pro- material, collectively describedas variousreds, from Hollow nounced among intensively utilized pieces, with 60% of Rock Shelterand Umhlatuzana.Table 6 comparesthe propor- crayons having strong-red streaks (Watts 1998). Not only tional contributionof colour groupingsto the entire sample were MSA people preferentially collecting red pigments with their representation among utilized pieces. Most cate- but there was also preferentialselection of pieces to utilize, gories were utilized in proportionsapproximating or slightly with more intensive use of darker, more saturated reds. above theirpercentage contribution to the total. That orange- Similar, more accurately described selection has recently & yellowish-browns and mid- & dark-brownsare betterrep- been demonstratedfor the MSA 2b and Still Bay sequence resented amongutilized pieces thanone mightexpect is largely at (Henshilwood et al. 2001).

This content downloaded on Thu, 3 Jan 2013 17:48:58 PM All use subject to JSTOR Terms and Conditions 10 South AfricanArchaeological Bulletin 57 (175): 1-14, 2002

MSA miningof specular-haematiteat Lion Cavern(Ngwenya No compellingethnographic or archaeologicalsupport and lit- Ridge, Swaziland)and the large quantitiesof speculariteat tle experimentalsupport could be found for suggesting that Olieboompoortsuggest that redness was not the sole criterion ochre played any hide-preservationrole. Some observations by which potential pigments were assessed. At Ngwenya were inconsistentwith this or similar functionalroles: these Ridge, plain haematitewas more accessible than the mica- includedthe rarityof yellow ochre and manganese,the occa- ceous expressionencountered in the cliff-face location of the sionaluse of palerochres and calciumcarbonate, the selection aditmine (Dart& Beaumont1969). The targetingof specular- of specular-over plain-haematiteat Ngwenya Ridge and the over plain-haematitewould not be predicted by the hide- copious use of ochre at Hollow Rock Shelter-a sheltertoo preservationhypothesis. This said, brilliance in the MSA small to permithide-dressing. The vast majorityof the sam- seems to overlap with redness rather than being an ples producedhues within an extendedrange of red; among autonomouscriterion: specularite, while frequentlyblack in utilized pieces, the largest hue sub-group were those I appearancestill gives a red streakand saturatedreds arethem- describedas strong-red,approximating a saturatedfocal range. selves brilliant(Morphy 1989: D'Andrade1995). The use of Takingcrayons as exemplarsof the most intensivelyutilized materialswith a black or white brilliantappearance, such as pieces, these show particularlypronounced selection for talc, mica, or ilmenite is almost exclusively restrictedto the strong-reds.An overlappingrather than autonomous selective LSA (Barham 1989; Robbins et al. 1993; Mitchell 1993, preferencewas for materialswith brilliantappearance, most 1994). To my knowledge,the only instancewhere such mate- notably specularite.Late PleistoceneMSA people were pre- rialshave been reportedfrom an MSA contextis Mitchelland paredto go to considerableeffort to obtainthese materialsas Steinberg's(1992) account of ilmenite from the MSA 3 at is graphicallytestified at Lion Cavern.Cognitive anthropol- NtloanaTsoana, Lesotho. ogy and vision researchsuggests that redness and brilliance are likely to have been perceptualqualities exploited from an I judged 90% of the total sample (excludingopen sites) to be early stage in ritualdisplay. Both perceptsare deeply impli- definite or probablepigments, 8.2% to be possible pigments cated in Khoisan constructsof supernaturalpotency, but the and 1.8%to be doubtfulor non-pigments(Watts 1998). Of the only Khoisan ritualwhere use of red pigment was virtually = definiteand probablepigments (n 3634) only 4.1 % provid- ubiquitouswas menarchealinitiation. Initiation ritual is wide- ed non-redstreaks (Watts 1998). Most of these were orange- ly thoughtto have playeda criticalrole in the establishmentof and yellowish-brownsfrom the large Hollow Rock Shelter symbolicreference. I concludethat the vast majorityof mate- and Umhlatuzana assemblages, that among some rials was collected for visually salient propertiesof redness Khoisangroups are embracedwithin an extendedrange of red and brillianceand used accordinglyas pigments, in the first (Wilhelm 1954). Of the remainder,the small amountof black instancein 'skin-changing'collective ritual performances. This manganesewas mostly from Still Bay layers at Hollow Rock places previous,similar interpretations on a more securefoot- Shelterand most of the white and yellow pigmentscame from ing thanhereto. Ochre may well have served additionalroles Apollo 11, largelyfrom the HowiesonsPoort layer where they (whetherutilitarian or as a pigmentin contextsother than col- predominateover red ochre (Watts 1998). Klasies River lective ritual)but these cannotprovide an alternativegeneral Mouthalso sees some use of yellow ochre in the Howiesons accountfor MSA or laterochre use. Poort(Watts 1998). Thereis no comparablebroadening of the range of pigment hues until well into the LSA, with late ApplyingChase and Dibble's (1987, 1992) archaeologicalcri- Robbergand Oakhurst/Albanyassemblages (Beaumont 1981; teria for identifyingsymbolism, the evidence presentedsug- Mitchell 1995; Watts1998). gests that symbolic use of red ochre had been establishedby earlyin the LatePleistocene. Alternative (functional) accounts Conclusions have been discounted.The use of redpigment was regularand ubiquitous.Similar behaviours (including 'crayon'manufac- The descriptive methods employed in this study require ture) are documentedfrom latercontexts where symbolismis improvementand supplementingwith archaeometricobser- not in doubt.The rareexamples of geometricengravings pro- vations,but they permitan evaluationof the two principlecon- vide directevidence for graphicrepresentations, while crayons tendinginterpretations of MSA ochreuse. providemore wide-spread,indirect evidence arguablyconsis- with Forthe vast majority,perhaps all, of the Late A principalreason that hide-preservation and similarhypothe- tent design. in southernAfrica, red ochre was used within an ses have been invoked is scepticismas to whetherany sym- Pleistocene web of symbolic meanings, a uniquely human bolic behaviour predates the LSA. The problem has been overarching culturalcontext that the behaviouritself had helped establish. compoundedbecause of a tacit presuppositionon both sides North of the River, symbolic use of red ochre may of the debate that pigment use implies symbolism. It is this Limpopo have still greaterantiquity. presuppositionthat underliesthe questioningof the pigment status of earlier ochre occurrences.In this paper,I invoked Acknowledgements Zahavi's handicap principle to show that the evolution of pigment-usecould be addressedas a case of costly ritualized Financial support from the University of London Central display without assuming symbolism. However, the conver- Research Fund, the British Academy, the South African gence of features with a Durkheimiancharacterization of National Research Foundation and the L.S.B. Leakey humancollective ritualsuggests that where the archaeologi- Foundationis gratefullyacknowledged. For access to archaeo- cal record indicates habitual pigment use we are probably logical collectionsI thankAron Mazel (formerlyof the Natal dealing with coalitions of signallers. Museum), Lyn Wadley (Department of Archaeology,

This content downloaded on Thu, 3 Jan 2013 17:48:58 PM All use subject to JSTOR Terms and Conditions South AfricanArchaeological Bulletin 57 (175): 1-14, 2002 11

University of Witwatersrand),Peter Beaumont (formerlyof Oxford:Oxford University Press. the McGregorMuseum, Kimberley), GrahamAvery (Iziko Berlin, B. & Kay, P. 1969. Basic color terms:their universalityand South African Museum), John Kinahan (formerly of the evolution.Berkeley, California: University of CaliforniaPress. State Museum, Windhoek),Eric Wendt(personal capacity); Bonwick, J. 1898. Daily life and origin of the Tasmanians. H.J. Deacon (formerly Department of Archaeology, Second edition.London: Sampson, Low, Marston& Co. Universityof Stellenbosch),Royden Yates and UrsulaEvans Boynton, R. 1979. Human color vision. New York: Holt, (formerly of the Departmentof Archaeology,University of Rinehart& Winston. Cape Town) and Peter Mitchell (Pitt Rivers Museum, Boshier,A. 1969.. genesis. Mining Survey64: 21-28 Oxford). For comments on earlier drafts, thanks are due to Boshier, A. & Beaumont,P. 1972. Mining in southernAfrica and Judith Sealy, Royden Yates, Chris Knight, Anne Thackeray the emergenceof modernman. Optima22: 2-12. and two anonymousreferees. Brabers,A.J.M. 1976. Mineral pigments. In: Coetzee, C.B. (ed.) References Mineral resources of the Republic of South Africa: 391-393. Fifth edition. Geological Survey of South Africa Handbook7. Abramov, I. 1997. Physiological mechanisms of color vision. In: Pretoria:Government Printer. Hardin, C. & Maffi, L. (eds) Color categories in thought and Bradshaw, J. & Rogers, L., 1993. The evolution of lateral language: 89-117. Cambridge: Cambridge University Press asymmetries, language, tool use and intellect. San Diego: Aldhouse-Green, S. & Pettit, P. 1998. Paviland Cave: contextualising AcademicPress. the 'Red Lady'. 72: Antiquity 756-772. Bueller, J. 1993. The methodology to establish the types of Armstrong, A.L. 1931. Rhodesian archaeological expedition (1929): microwearon sampledblades from Urkane-Rub Ila and Netiv excavations in Bambata Cave and researches on prehistoric sites Hagdud.Journal of the IsraelPrehistoric Society 25: 163-188. in southern Rhodesia. Journal of the Royal Anthropological Casson, R.W. 1997. Color shift: evolution of English color terms Institute 61: 239-276. from brightnessto hue. In: Hardin,C. & Maffi L. (eds) Color Atran, S. 1990. Cognitive foundations of natural history. Towards an categories in thought and language: 224-239. Cambridge: anthropology of science. Cambridge: Cambridge University CambridgeUniversity Press. Press. Chase P.G. & Dibble, H. 1987. Middle Palaeolithicsymbolism: a Audouin, F. & Plisson, H. 1982. Les ocres et leurs temoins au review of current evidence and interpretations.Journal of Palholithique en France: enquete et experiences sur leur validite AnthropologicalArchaeology 6: 263-96. archeologique. Cahiers du Centre de Recherches Chase, P.G. & Dibble, H. 1992. Scientific archaeology and the Prehistoriques 8: 33-80. origins of symbolism: a reply to Bednarik. Cambridge Avery, G., Cruz-Uribe, K., Goldberg, P., Grine, F., Klein, R., ArchaeologicalJournal 2: 43-51. Lenardi, M., Marean, C., Rink J., Schwarcz, H., Thackeray, A. & Clark,A. 1997. The MSAJLSAtransition in SouthernAfrica: new Wilson, M. 1997. The 1992-1993 excavations at the Die Kelders technological evidence from Rose Cottage Cave. South Middle and Later Stone Age cave site, South Africa. Journal of AfricanJournal of Science 52: 113-121. Field Archaeology 24: 263-291. Clark, G. 1986. Symbols of excellence: precious materials as Bahn, P. & Vertut, J. 1988. Images of the Ice Age. London: expressionsof status.Cambridge: Cambridge University Press. Windward. Clark, J.D. 1988. The Middle Stone Age of East Africa and the Barham, L.S. 1989. The Later Stone Age of Swaziland. Unpubl. beginnings of regional identity. Journal of World Prehistory Ph.D. thesis: University of Pennsylvania. 2: 237-305. Barham, L.S. 1998. Possible early pigment use in south-central Couraud, C. 1991. Les pigments des grottes D'Arcy-sur-Cure Africa. Current Anthropology 39: 703-1 0. (Yonne).Gallia Prehistoire 33: 17-52. Barham, L.S. 2000. The Middle Stone Age of Zambia, south D'Andrade,R. 1995. The developmentof cognitive anthropology. central Africa. Bristol: Western Academic & Specialist Press. Cambridge:Cambridge University Press. Bateman, A.M. 1950. Economic mineral deposits. New York: Dart, R. & Beaumont, P. 1969. Evidence of iron ore mining in John Wiley. southern Africa in the Middle Stone Age. Current Bates, R. & Jackson, J. (eds) 1984. Dictionary of geological Anthropology10: 127-128. terms. Third edition. New York: Anchor Press/Doubleday. Deacon, H.J. 1995. Two Late Pleistocene-Holocenearchaeological Beaumont, P.B. 1973. The ancient pigment mines of southern depositories from the southern Cape, South Africa. South Africa. South African Journal of Science 69: 140-146. AfricanArchaeological Bulletin 50: 12-131. Beaumont, P.B. 1978. Border Cave. Unpubl. M.A. thesis: Deacon, H.J. 2001. Modern human emergence: an African . archaeological perspective. In: Tobias, P.V., Raath, Beaumont, P.B. 1981. The Heuningsneskrans shelter. In: Voigt, M.A., Moggi-Cecchi, J. & Doyle, G.A. (eds) Humanityfrom E. (ed.) Guide to archaeological sites in the northern and African naissance to coming millennia: 213-222. Firenze: eastern Transvaal: 132-145. Pretoria: Transvaal Museum. Firenze University Press; : Witwatersrand UniversityPress. Beaumont, P. 1990a. Nooitgedacht 2 and Roseberry Plain 1. In: Beaumont, P. & Morris, D. (eds) Guide to archaeological sites Deacon, H., Deacon, J., Geleijnse,V.B., Thackeray,F., Scholtz,A., in the : 4-6. Kimberley: McGregor Museum. Brink, J.S. 1983. Late Quaternaryenvironment and culture relationships in the southern Cape. Final Report to the Beaumont, P. 1990b. . In: Beaumont, P & Morris, D. (eds) National Monuments Council on excavations in the Guide to archaeological sites in the northern Cape: 75-100. Oudtshoorn, Heidelberg, Willowmore Kimberley: McGregor Museum. and Districts. Unpublished report. South African Heritage Bell, C. 1992. Ritual theory, ritual practice. New York and ResourcesAgency: Cape Town

This content downloaded on Thu, 3 Jan 2013 17:48:58 PM All use subject to JSTOR Terms and Conditions 12 SouthAfrican Archaeological Bulletin 57 (175): 1-14, 2002

Deacon, T. 1997. The symbolic species: the co-evolution of Jercher, M., Pring, A., Jones, P.G. & Raven, M.D. 1998. Rietveld language and the human brain. London: Penguin Books. X-ray diffraction and X-ray fluorescence analysis of ochres. Archaeometry 40: 383-401. De Valois, R.L., & De Valois, K.K. 1993. A multi-stage color Australian Aboriginal model. Vision Research 33: 1053-1065. Kaplan, J. 1990. The Umhlatuzana Rock Shelter sequence: of Stone Age history. Natal Museum Journal of De Villiers, J. 1959. Mineral Pigment. In: Vermeulen, P. (ed.) 100,000 years 1-94. The mineral resources of the Union of South Africa: 515-519. Humanities 2: Pretoria: The Government Printer. Kay, P., Berlin, B., Maffi, M. & Merrifield, W. 1997. Color languages. In: Hardin, C. & Maffi, L. (eds) Dumont, J. 1988. A microwear analysis of selected artifact types naming across categories in thought and language: 21-58. Cambridge: from the Mesolithic sites of Star Carr and Mount Sandel. Color University Press. Oxford: British Archaeological Reports, British Series 187. Cambridge Keeley, L. 1978. Preliminary microwear analysis of the Meer Durkheim, E. 1965 (1912). The elementary forms of religious assemblages. In: Van Noten, F. (ed.) Les chasseurs de Meer: life. New York: Free Press. 73-86. Brugge: De Tempel. Evans, U. 1994. Hollow Rock Shelter: a Middle Stone Age site in Keeley, L. 1980. Experimental determination of stone tool uses: a the Cederberg. Southern African Field Archaeology 3: 63-73. microwear analysis. Chicago: University of Chicago Press. Gell, A. 1993. Wrapping in images: tattooing in Polynesia. Klein, R.G. 1995. Anatomy, behaviour and modern human Oxford: Clarendon Press. origins. Journal of World Prehistory 9: 167-198. Gellner, E. 1992. Reason and culture. Oxford: Blackwell. Klein, R.G.. 1999. The human career: human biological and peripheral Gordon, J. & Abramov, I. 1977. Color vision in the cultural origins. Second edition. Chicago: University of of the Optical Society of retina. II. Hue and saturation. Journal Chicago Press. America 67: 202-207. Klima, B. 1987. Mladopaleolitichy trojhrob z Dolnieh Vstoniv. the Guilford, T. & Dawkins, M.S. 1991. Receiver psychology and Archeologick6 Rozhledy 39: 241-254. evolution of animal signals. Animal Behaviour 42: 1-14. Knight, C.D. 1991. Blood relations. Menstruation and the origins of Grafen, A. 1990. Biological signals as handicaps. Journal of culture. New Haven and London: Yale University Press. Theoretical Biology 144: 517-546. Knight, C.D. 1998. Ritual/speech coevolution: a solution to the GrUn, R. & Beaumont, P. 2001. Border Cave revisited: a revised problem of deception. In: Hurfurd, J., Studdert-Kennedy, M. ESR chronology. Journal of Human Evolution 40: 467-482. & Knight, C. (eds) Approaches to the evolution of language: Hayden, B. 1990. The right rub: hide working in high ranking social and cognitive bases: 68-91. Cambridge: Cambridge households. In Knutson, K. (ed.) Proceedings of the ninth University Press. conference on use-wear, Uppsala, Aun 14: 89-102. Upp-sala: Knight, C.D. 1999. Sex and language as pretend-play. In: Dunbar, R., Societas Archaeologica Upsaliensis. Knight, C. & Power, C. (eds) The evolution of culture: Hayden, B. 1998. Practical and prestige technologies: The 228-247. Edinburgh: Edinburgh University Press. evolution of material systems. Journal of Archaeological Knight, C., Power, C. & Watts, I. 1995. The human symbolic Method and Theory 5: 1-55. revolution: a Darwinian account. Cambridge Archaeological Hedges, S.B. 2000. A start for population genomics. Nature Journal 5: 75-114. 408: 652-653. Lattas, A. 1989. Trickery and sacrifice: Tambarans and the Heider, E. R. 1972. Probabilities, sampling and ethnographic appropriation of female reproductive powers in male method: the case of Dani colour names. Man 7: 448-466. initiation ceremonies in west New Britain. Man (n.s.) 24: 451-69. Henshilwood, C., Sealy, J., Yates, R., Cruz-Uribe, K., Goldberg, P., Leach, E. 1954. Political systems of Highland Burma. London: Grine, F., Klein, R., Poggenpoel, C., Van Niekerk, K., Watts, I., Bell. 2001. Blombos Cave, southern Cape, South Africa: Leslie-Brooker, M. 1987. An archaeological study of the preliminary report on the 1992-1999 excavations of the Middle Uniondale Rockshelter, Albany District, Eastern Cape. Stone Age levels. Journal of Archaeological Science Unpubl. M.A. thesis: University of Stellenbosch. 28:421-448. Lewis-Williams, D. 1981. Believing and seeing. Symbolic Henshilwood, C., d'Errico, F., Yates, R., Jacobs. Z., Tribolo, C., meanings in Southern San rock paintings. London: Academic Duller, G., Mercier, N., Sealy, J., Valladas, H., Watts, I., & Press. Wintle, A. 2002. Emergence of modern human behaviour: Lewis-Williams, D. 1991. Wrestling with analogy: a methodo- Middle Stone Age engravings from South Africa. Science logical dilemma in Upper Palaeolithic art research. 295: 1278-1280. Proceedings of the Prehistoric Society 57: 149-62. How, M.W. 1962. The mountain Bushmen of Basutoland. Lurie, J. 1977. South African geology for mining, metallurgical, Pretoria: J. L. van Schaik. hydrological and civil engineering. Johannesburg: McGraw- Hughes, J. & Solomon, A. 2000. A preliminary study of ochres Hill. and materials from KwaZulu-Natal, South pigmentaceous Mandl, I. 1961. Collagenases and elastases. Advances in Africa: towards an understanding of San pigment and Enzymology 23: 164-264. use. Natal Museum Journal of Humanities 12: 15-31. Mason, R.J. 1962. Prehistory of the Transvaal. Johannesburg: nature.In: Porter,T., Humphrey,N. 1976. The colour currencyof Universityof the WitwatersrandPress. & Mikellides, B. (eds) Color for architects:95-98. New York: E. 1995. The major transitions VanNostrand. Maynard-Smith,J., & Szathmairy, in evolution.Oxford: W. H. Freeman. Ingman, M., Kaessmann,H., Pitibo, S. & Gyllensten, U. 2000. of east Africa. In: Mitochondrial genome variation and the origin of modern McBrearty,5. 2001. The Middle Pleistocene Human roots: Africa humans.Nature 408: 708-713. Barham, L. & Robson-Brown,K. (eds)

This content downloaded on Thu, 3 Jan 2013 17:48:58 PM All use subject to JSTOR Terms and Conditions South AfricanArchaeological Bulletin 57 (175): 1-14, 2002 13

and Asia in the Middle Pleistocene: 81-98. Bristol: Western 101-132. London& New Brunswick(N.J.): The Athlone Press. Academicand SpecialistPress. Rappaport,R.A. 1979. Ecology, meaning and religion. Berkeley, McBrearty,S. & Brooks, A. 2000. The revolution that wasn't: a California:North Atlantic Books. new interpretationof the origin of modem human behavior. Ratliff, F. 1976. On the psychophysiologicalbases of universal Journalof HumanEvolution 39: 453-563. color terms. Proceedings of the American Philosophical Mitchell, P.J. 1993. Archaeologicalinvestigations at two Lesotho Society 120: 311-330. rock-shelters: terminal Pleistocene/early Holocene Riel-Salvatore, J. & Clark, G. 2001. Grave markers. Current assemblages from Ha Makotoko and Ntloana Tsoana. Anthropology42: 449-460. Proceedingsof the PrehistoricSociety 59: 39-60. Robbins, L.H., Campbell,A., Murphy,M.L. & Ferone, T. 1993. Mitchell, P.J. 1994. Understandingthe MSA/LSA transition:the Prehistoric mining and new discoveries of at the pre-20,000 BP assemblages from new excavations at TsodiloHills. Nyame Akuma40: 2-5. Sehonghong rock-shelter, Lesotho. Southern African Field Roth, H.L. 1890. The Aborigines of Tasmania.London: Archaeology3: 15-25. Kegan Paul. Mitchell, P.J. 1995. Revisiting the Robberg: new results and a Rudner, I. 1982. Khoisan pigments and and their revision of old ideas at Sehonghong Rock Shelter, Lesotho. relationshipto rock paintings. Annals of the South African SouthAfrican Bulletin50: Archaeological 28-38. Museum87: 1-280. Mitchell, P.J. & Steinberg,J.M. 1992. Ntloana Tsoana:a Middle Searle, J. 1996. The construction of social reality. Stone Age sequence from western Lesotho. South African Harmond-sworth:Penguin Books. ArchaeologicalBulletin 47: 26-33. Singer, R. & J. Wymer 1982. The Middle Stone Age at Klasies Mithen, S. 1996. The prehistoryof the mind. A search for the River Mouth River Mouth in South Africa. Chicago: originsof art, and religion science. London:Thames & Hudson. Universityof ChicagoPress. Mithen, S. 1999. Symbolism and the supernatural.In: Dunbar,R., Smith, M.A., Fankhausser,B., & Jercher,M. 1998. The changing Knight, C. & Power, C. The (eds) evolution of culture: provenance of red ochre at PuritjarraRock Shelter, central 147-169. Edinburgh:Edinburgh Press. University Australia: Late Pleistocene to present. Proceedings of the Morphy, H. 1989. From dull to brilliant: the aesthetics of PrehistoricSociety 64: 276-292. spiritualpower amongthe Yolngu.Man 24: 21-40. Sollas, W.J. 1924. Ancient huntersand their modernrepresentatives. Moss, E. 1983. The functional analysis of flint implements. London:Macmillan. Pincevent and Point d'Amlon: two case studies from the Solso, R.L. 1994. Cognition and the visual arts. Cambridge, French Final Palaeolithic. Oxford: British Archaeological Massachusetts:MIT Press. Reports,International Series 177. Soodyall, H. & Jenkins, T. 1992. Mitochondrial DNA Mussi, M. 2001. Comment on 'Grave markers' by J. Riel- polymorphismsin Khoisan populations from southernAfrica. Salvatore& G. Clark.Current Anthropology 42: 467-468. Annalsof HumanGenetics 56: 315-324. O'Hanlon, M. 1989. Reading the skin: adornment,display and Soodyall, H. & Jenkins, T. 1998. Khoisan prehistory: the society among the Wahgi. Bathurst, Australia: Crawford evidence of the genes. In: Bank, A. (ed.) Proceedings of the House Press. Khoisan Identities and Cultural Heritage Conference, South Phillibert, S. 1994. L'ocre et le traitementdes peaux: revision African Museum, Cape Town, July 1997: 374-382. Cape d'une conception traditionelle par l'analyse fonctionelle des Town: University of the WesternCape, Institutefor Historical grattoirs ocrds de la Balma Margineda (Andorre). Research. L'Anthropologie98: 447-453. Sperber, D. 1975. Rethinking symbolism. Cambridge: Power, C. 1998. Old wives' tales: the gossip hypothesis and the CambridgeUniversity Press. reliabilityof cheap signals. In: Hurford,J., Studdert-Kennedy,M. Stahl, A. 1993. Concepts of time and approachesto analogical & Knight, C. (eds) Approachesto the evolution of language: reasoning in historical perspective. American Antiquity social and cognitive bases: 111-129. Cambridge:Cambridge 58: 235-260. UniversityPress. Stringer,C. & McKie, R. 1996. African exodus: the origins of Power, C. 1999. 'Beauty magic': the origins of art. In: Dunbar,R., modernhumanity. London: Random House. Knight, C. & Power, C. (eds) The evolution of culture:92-112. Taylor,D.J.C., Page, D.C. & Geldenhuys,P. 1988. Iron and steel Edinburgh: EdinburghUniversity Press. in South Africa. Journal of the South African Institute of Power, C. & Aiello, L. 1997. Female proto-symbolicstrategies. Miningand Metallurgy88: 73-95. In: Hager, L. D. (ed.) Women in human evolution: 153-171. Thackeray,A.I. 1989. Changing fashions in the Middle Stone New York& London:Routledge. Age: the stone artefactsequence from Klasies River main site, Power, C. & I. Watts, 1996. Female strategies and collective SouthAfrica. The AfricanArchaeological Review 7: 33-57. behaviour:the archaeology of earliest Homo sapiens sapiens. Thorne, A., GrUn, R., Mortimer,G., Spooner, N., Simpson, J., In: Steele, J. & Shennan, S. (eds) The archaeologyof human McCulloch, M., Taylor, L. & Curnoe, D. 1999. Australia's ancestry: power, sex and tradition: 306-330. London: oldest human remains: age of the Lake Mungo 3 skeleton. Routledge. Journalof HumanEvolution 36: 591--12. Power, C. & Wafts, I. 1997. The woman with the zebra'spenis: Tobias, P.V. 1949. The excavationof Mwulu's Cave, Potgietersrust gender, mutability and performance. Journal of the Royal District.South African Archaeological Bulletin 4: 2-13. AnthropologicalInstitute 3: 1-24. Turner,T.S. 1980. The social skin. In: Cherfas, J. & Lewin R. Power, C. & Wafts,I. 1999. Firstgender, wrong sex. In: Moore, H., (eds) Not work alone: a cross cultural view of activities Sanders, T. & Kaare, B. (eds) Those who play with fire: superfluousto survival:112-140. London:Temple Smith. gender, fertility and transformation in east and southernAfrica: Turner, V. 1967. The forest of symbols: aspects of Ndembu

This content downloaded on Thu, 3 Jan 2013 17:48:58 PM All use subject to JSTOR Terms and Conditions 14 SouthAfrican Archaeological Bulletin 57 (175): 1-14, 2002

ritual. Ithaca & London: Cornell University Press. Appendix. Derivation of the sample by excavation unit. nI Valladas, H., Reyss, J.L., Joron, J.L., Valladas, G., Bar-Yosef, 0. = total potential pigments, n2 = total Definite & Probable & Vandermeersch, B. 1988. Thermoluminiscence dating of pigments. Site acronyms as in Table 1. For divergences Mousterian 'Proto-Cro-Magnon' remains from Israel and the from Volman's (1984) sub-stage designations (Mwulu's origin of modem man. Nature 331: 614-616. Cave) and des-ignationsmade for unpublishedassemblages Vandermeersch, B. 1969. Decouverte d'un objet en ocre avec (BRS & BC 1987), see Watts 1998. traces d'utilisation dans le Mousterien de Qafzeh (Israel). Bulletin de la Societ6 Prehistorique Frangaise 66: 157-158. Units Sub-stage nl n2 Vandermeersch, B. 1981. Les hommes fossiles de Qafzeh (Israel). APllLyr.H MSA2a 20 10 Paris: Centre National de la Recherche Scientifiques. AP11 Lyr. G MSA2b 48 43 Van Gennep, A. 1960 (1909). The rites of passage. Translated by AP11 Lyr. F Howiesons Poort 23 17 Vizedom, M.B. & Caffee, G.L. Chicago: University of AP1lLyr.E MSA3 14 12 Chicago Press. OBP Bed 2 MSA 2b 304 304 tools. Vaughan, P.C. 1985. Use-wear analysis of flaked stone BRS Lyrs. 53-32 MSA 2a? 5 3 Tucson: University of Arizona Press. BRS Lyrs. 31-19 MSA 2b ? 41 22 Vigilant, L., Stoneking, M., Harpending, H., Hawkes, K. & BRS Lyrs. 18-16 MSA 3 12 4 Wilson, A. 1991. African populations and the evolution of MC Bed I MSA 2a 1 I mitochondrial DNA. Science 253: 1503-1507. MC Beds II-III MSA 2b 12 12 Visser, D. 1937. The ochre deposits of the Riversdale District, SHG Lyrs. RFS-OS TransitionalMSA/LSA 99 99 . Department of Mines, Geological Series RCC spits 242"-213' MSA 2b 48 46 Bulletin 9. Pretoria: Government Printer. RCC spits 213"-204" Mixed Assemblage 8 7 Vogelsang, R. 1998. Middle-Stone-Age-Fundstellen im Stidwest RCC spits 204"-144" Howiesons Poort 44 27 Namibia. Africa Praehistorica 11. K6ln: Heinrich Barth RCC spits 144"-81" MSA 3 12 11 Institute. BP LPC3 ELSA 77 57 38 33 Volman, T. 1981. The Middle Stone Age in the southern Cape. BP BPK MSA3 19 Unpubl. Ph.D. thesis: University of Chicago. BP OCH (inclusive) Howiesons Poort 19 HRS Sand IA-IIB Still Bay 1071 875 Volman, T. 1984. Early prehistory of southern Africa. In: HRS Sand IIIA-IIIB MSA 2b 72 48 Klein, R.G. (ed.) Southern African prehistory and UMHLyr. 28 MSA2b 7 7 paleoenvironments: 169-220. Rotterdam: Balkema. UMH Lyr. 27 Mixed Assemblage 19 19 Wadley, L. 1993. The Pleistocene Later Stone Age south of the UMH Lyrs. 26-22 Howiesons Poort 405 404 Limpopo River. Journal of World Prehistory 7: 243-96. UMH Lyrs. 21-19 MSA 3 359 359 Wadley, L. 2001. What is cultural modernity? A general view and UMH Lyrs. 18-14 TransitionalMSA/LSA 723 723 a South African perspective from Rose Cottage Cave. Cambridge UMH Lyrs. 13-10 ELSA 208 203 Archaeological Journal 11: 201-221. BC 1987 6BS & 5WA MSA2a 4 4 Wadley, L. & Harper, P. 1989. Rose Cottage Cave revisited: BC 1987 5BS-4BS MSA 2b 101 92 Malan's Middle Stone Age collection. South African BC 1970+ BACO D MSA 2a 1 1 Archaeological Bulletin 44: 23-32. BC 1970+ RGBS A Howiesons Poort 5 5 Walker, N.J. 1987. The dating of Zimbabwean rock art. Rock Art KRM1 Lys. 40-37 MSA 2a 11 8 Research 4: 137-149. KRM1 Lys. 17-14 MSA 2b 26 24 179 113 Watts, I. 1998. The origin of symbolic culture: the Middle Stone KRM1A Lyrs. 21-10 Howiesons Poort Age of southern Africa and Khoisan ethnography. Unpubl. Ph.D. KRM1A Lys. 9-1 MSA 3 19 19 thesis: University of London. KRM1B Lys. 12-1 MSA 2b 3 3 4038 3634 Watts, I. 1999. The origin of symbolic culture. In: Dunbar, R., Total Knight, C. & Power, C. (eds) The evolution of culture: 113-146. Edinburgh: Edinburgh University Press. Wilhelm, J.H. 1954. Die !Kung-Buschleute. Jahrbuch des Museums fir V6lkerkunde, Leipzig. 12: 91-188. Wooten, B. & Miller, D. 1997. The psychophysics of color. In: Hardin, C. & Mafti, L. (eds) Color categories in thought and language: 59-87. Cambridge: Cambridge University Press. Wylie, A. 1985. The reaction against analogy. Advances in Archaeological Method and Theory 8: 63-111. Wylie, A. 1988. 'Simple' analogy and the role of relevance assumptions: implications of archaeological practice. International Studies in the Philosophy of Science 2: 134-150. Zahavi, A. 1987. The theory of signal selection and some of its implications. In: Delfino, V.P. (ed.) Internationalsymposium of biologicalevolution: 305-327. Bani:Adriatic Editrice. Zahavi, A, & Zahavi, A. 1997. The handicapprinciple: a missing piece of Darwin's puzzle. New York & Oxford: Oxford UniversityPress. T

This content downloaded on Thu, 3 Jan 2013 17:48:58 PM All use subject to JSTOR Terms and Conditions