Sauvet Et Al CAJ2009

8/13/2019 Sauvet Et Al CAJ2009

1/18

1

Thinking with Animals in Upper Palaeolithic Rock Art

Georges Sauvet, Robert Layton, Tilman Lenssen-Erz,Paul Taon & Andr Wlodarczyk


that all sites in his survey are essentially similar anddid not use ethnographic case studies of motif distri-

bution as a baseline. Hartleys (1992) analysis of motiffrequencies in the rock art of the Colorado Plateau wasmore successful. Hartley found that there was a greaterdiversity of art motifs on isolated boulders than in rockshelters. He argued (drawing on local and compara-tive ethnography) that the dierence probably arosethrough prolonged use of rock shelters by the same orrelated groups, while isolated boulders were probablyvisited by members of more diverse local groups. Asimilar method is used here to locate Upper PalaeolithicEurope in a eld of ethnographically dened culturalvariation. Like Hartley (1992, 64), we are not seeking toconstrue the meaning of individual motifs, but ratherthe cognitive structures of which the use of motifs issymptomatic. By cultural context we therefore meanthe relationship between aspects of behaviour, mate-rial culture and the system of ideas characteristic of aparticular community. A high proportion of the iden-tiable gures in the Upper Palaeolithic art of Franceand Spain are animals. Following Lvi-Strausss (1962,89) dictum that animals are good to think with, weanalyse the cultural context of animal gures in vari-ous recent rock-art traditions and ask what light thiscan throw on the cultural context of the animal art ofthe Upper Palaeolithic.

Almost as soon as archaeologists realized thatthe Upper Palaeolithic rock art of southwest France

This article develops a novel method for assessing the cultural context of rock art, andapplies it to the rock art of the Upper Palaeolithic of France and Spain. The article relieson a generative approach, assuming that artists have the potential to choose which motifsto select from the repertoire or vocabulary of their artistic system, but that appropriatechoices at any place are guided by the location of that site within the culturally-mediatedgeography of the region. Ethnographic studies of rock art depicting animal species producedin the contexts of totemism, shamanism and everyday life are used as reference points in

an analytical framework, which is then applied to a number of ancient traditions.

Background to the article

IntroductionIn a review article published in Cambridge Archaeologi-cal Journal(2000), Layton argued that interpretationsof Upper Palaeolithic rock art based on shamanismhad come to dominate the eld, displacing viable

alternative hypotheses. Layton proposed that the artof shamanism was merely one way of appropriatingmotifs from the artistic vocabulary of a cultural systemand outlined a method for discriminating betweenthe distinctive ways in which totemic, shamanic andsecular rock art draw on such a vocabulary throughthe choice of which motifs to represent, both at indi-vidual sites and across a wider region. The presentarticle tests this method using more extensive data andimproved statistical methods. The results of the testare presented, and the validity of the method subjectedto critical assessment.

Few existing archaeological studies of rock

art develop a method where a range of alternativehypotheses can be given equal weight; most tend toact as advocates for one preferred interpretation, butwe believe that the method described here is capableof further renement and wider application. A similarmethod to ours was used by Tilley (1991) to comparethe statistical frequency of motifs on panels at theprehistoric Namforsen site complex and other Scandi-navian sites of the same period. Tilley, however, found

Cambridge Archaeological Journal19:3, ?????? 2009 McDonald Institute for Archaeological Researchdoi:10.1017/S0959774309000??? Received 18 Dec 2008; Accepted 16 Feb 2009; Revised 18 May 2009


2/18

2

Georges Sauvet et al.

had been created by prehistoric hunter-gatherers, theybegan to search for a key to its cultural signicancein the art of modern hunter-gatherers. As early as1903 Reinach argued that living hunter-sher socie-ties could provide beer parallels with the societies of

the Age of the Reindeer than could medieval Frenchculture. Two possible parallels have dominated theliterature: Reinachs preferred parallel with totemismand the now more familiar parallel with shamanism.

TotemismThe word totemcomes from the Ojibwa, a native NorthAmerican people, while the word shamancomes fromthe Tungus of central Siberia. Their use cross-cultur-ally, to refer to types of religion (i.e. shaman ismandtotemism), is an artefact of anthropology. The validityof totemism as a cross-culturally valid category hasbeen vigorously debated in anthropology (see Lvi-

Strauss 1962 for an authoritative review). It is gener-ally agreed to refer to the use of animals or plants asemblems or guardians of socialgroupscelebrated inritual. The rationale of totemism is that each group isidentied with a dierent species; the signicance ofeach species derives from its place in the cognitivestructure. Group A is kangaroo because it is not emu orpython (see Lvi-Strauss 1964, ch. 5). Hartleys (1902)analysis of Colorado rock art relied on ethnographicevidence for clan totemism in the American Southwestto predict the consequences of the long-term accumu-lation of motifs at dierent sites.

Reinach (1903) rejected the initial Art for ArtsSake hypothesis that Upper Palaeolithic cave art was amere leisure-time amusement on the grounds that onlycertain species (food animals but not dangerous ones)were represented in the caves of southwest France,and because the paintings were placed far from thecave entrance, in places that were dicult to reach. Hefound the central Australian ethnography of Spencer& Gillen (1899) particularly enlightening. The UpperPalaeolithic is dominated by paintings and engrav-ings of animal species. Central Australian rituals aredesigned to increase numbers of each clans totemicspecies, both through rock paintings representing

the totem and through dances imitating the animalsbehaviour. Reinach assumed that only desirable foodanimals would be the object of increase rites. Most ofthe central Australian painted sites are, moreover, inlocations that women, children and uninitiated menare forbidden from entering, suggesting a parallelwith the location of surviving French cave art. Reinach(1903, 2612) appreciated that although all have thesole goal of increasing the number of the animals orplants aer which their totem is named, nonetheless

each [clan] totem has a special ceremony, and no twoceremonies are alike. He did not, however, follow thisobservation to its logical conclusion by asking whethereach French cave was characterized by paintings of aparticular species, the putative totem of a local clan.

ShamanismShamanism refers to the use of spirits as guardiansand helpers of individuals, contacted through trancetoeect healing. The cross-cultural validity of the termshamanism has been considered by Eliade (1972),Hultkrantz (1989), Lewis (1971) and Vitebsky (1995).These authors do not agree on whether all knowncases of shamanism have a common origin, or whetherdierent traditions of healing during spirit possessionarose independently. Beaune (1998) notes that shaman-ism has been proposed more than once as a contextfor the production of Upper Palaeolithic art. In their

monograph on Altamira, Cartailhac & Breuil (1906)compared some engravings to the animal masks wornby Inuit shamans. Beaune argues that shamanismnonetheless made lile contribution to the repertoireof ethnographic parallels during the era dominatedby Breuil, because most of the ethnographic literaturewas published in Russian. Shamanism assumed moreimportance during the 1950s60s, particularly in thework of Kirchner (1952) and Kuhn (1956). Parallelswere drawn between the antler-bearing sorcerer ofLes Trois Frresand the costumes of Siberian shamans,while Kuhn (1956, 378), citing Kirchner, added a sec-ond example, comparing the apparently bird-headedhuman gure in the Lascaux Pit Scene to a Siberianshaman falling into trance.

Lewis-Williamss revival of the shamanic inter-pretation of Upper Palaeolithic art (Lewis-Williams& Dowson 1988; Cloes & Lewis-Williams 1996) wasinspired by his previous work on the rock art of theDrakensberg Mountains in South Africa. In his earlywork, Lewis-Williams (1981) used the nineteenth- andtwentieth-century ethnography of hunter-gatherers insouthern Africa to interpret the art of the Drakensbergas an expression of local Maluti San religion. Lewis-Williamss interpretations were convincing because

he was able to match rituals recorded in other Sanethnographies with detailed scenes in the rock art,and because the special, spiritual relationship betweenhumans and eland suggested in the fragmentaryethnography of the Drakensberg provided by Qing,survivor of a massacre, corresponded to the highfrequency of eland (60 per cent of animal gures) inthe Drakensberg rock art recorded by Vinnicombe(1975, 364). Lewis-Williams did not use the term sha-man in this work, but he did devote one chapter to


3/18

3


The medicine man and the eland, where he arguedthat when medicine men went into trance duringcuring rituals they identied themselves with theeland. Lewis-Williams cites Bleeks well-regardednineteenth century ethnography of the /Xam San of

the Strandberg-Katkop region, which described menwearing gemsbok scalps when performing curingrituals (Bleek 1935; 1936; cited in Lewis-Williams 1981;see Barnard 1992, 79, 82 for an assessment of Bleekswork; the /Xam are now, in Barnards words culturallyand linguistically extinct(1992, 12)). Lewis-Williams(1981) interpreted representations of composite eland-human gures in Drakensberg art as depictions oftrance experience.

Parallel research was carried out by Whitley(1994) on the rock art of the last 1500 years in theCoso Ranges in eastern California. Fiy-one per centof engravings in the Coso Ranges depict mountain

sheep, followed by 29 per cent made up of variousgeometric paerns (Whitley 1994, 360). Whitley rejectsearlier archaeologists hunting magic explanation onthe grounds that fewer mountain sheep were huntedduring this period than earlier ones, and there is noethnographic record of hunting magic. Rather, theethnography associates mountain sheep with sha-manic rituals.

Although the relative frequency of culturallysignicant animal species played an important evi-dentiary role in both these studies, Lewis-Williams& Dowson (1988) proposed an alternative line ofevidence for shamanism in the Upper Palaeolithic,namely the similarity in shape between some of thePalaeolithic signs and forms that neuropsychologistshad identied as characteristically perceived duringaltered states of consciousness. There is no ethno-graphic evidence from southern Africa that geometricgures depict shapes seen during trance (Lewis-Williams & Dowson 1988, 205). The evidence advancedwas of formal similarity between the geometric formsseen in the rock art of the Drakensberg and Coso,psychological studies of drug-induced states of alteredconsciousness, and Reichel-Dolmatos (1978) eth-nographic study of Tukano art, in which the Tukano

themselves identify forms seen during the rst phaseof drug-induced trance. Numerous criticisms havebeen made of this argument, of which the most force-ful is Helvenston & Bahn (2005).

We therefore advocate a closer look at paerningin the distribution of the animal motifs that have beenappealed to repeatedly as evidence for the culturalcontext of Upper Palaeolithic art, and which play animportant role in all the ethnographic cases cited aspossible parallels.

The theoretical frameworkThe approach taken in this article is based on a genera-tive model. In any situation, artists have the potentialto choose which motifs to select from the repertoireor vocabulary of their artistic system. The appropri-

ate choice will be dictated by the context, in this case,the location of the site within the culturally mediatedgeography of the region. Painters and engravers aremotivated by considerations of appropriate subjectand context. If these motivations remain constant forlong enough, a visible, archaeologically-measurablepaern will be generated.

Ethnographic cases frequently demonstrate thatmore than one cultural theme is expressed througha communitys rock art, but these are typically dif-ferentiated by the size or style of gures. The analysiscarried out below is concerned only with the distribu-tion of animal motifs. In the nal section of the article,

we consider the potential eect that the co-presenceof more than one cultural theme (e.g. totemism andsorcery) may have on the distribution of animalmotifs. We do not analyse the distribution of anthro-pomorphs, which are very common in some traditionssuch as the rock art of the Australian Kimberleys, butrare in the art of the European Upper Palaeolithic and,for example, the recent art of Western Arnhem Land.The discernable motifs of the Upper Palaeolithic arepredominantly of two classes: animals and signs.Signs are quite possibly non-representational, andmay express a dierent aspect of the culture, but arather large number of animal species can be distin-guished by body shape. A number of recent hunter-gatherer rock-art traditions are also characterized bya large number of recognizable animal species (oenbetween 15 and 25 species), whereas the iconographyof anthropomorphs is more opaque and dependent oncultural convention. The distribution paerns seen inthe location of animal motifs therefore provides a use-ful means for comparative analysis that is particularlypertinent to the Upper Palaeolithic.

In order to test the method, we take the threehypotheses for the context of Upper Palaeolithicart outlined above (Art for Arts Sake, Totemism

and Shamanism) as our starting point, and explorehow closely the distribution of animal species in theUpper Palaeolithic conforms to the paern seen inethnographic cases where data on species frequencywithin and between sites are available. Although theArt for Arts sake interpretation was quickly rejectedby Upper Palaeolithic archaeologists, we considerthat it should not be dismissed out of hand. Thereare several ethnographic cases of secular art fromAustralia. In an unpublished report on coastal Cape


4/18

4


York, North Queensland, John and Leslie Havilandreport that they were told by local Aboriginal peoplethat the great majority of rock paintings they saw werethe result of lengthy stays in the caves during the

rainy season a sort of rainy day pastime (quoted inLayton 1995, 222). David Turner spoke to Balrumba, aliving artist who had painted in shelters on BickertonIsland in the Gulf of Carpentaria. Balrumba identiedpaintings as records of successful hunts, the visit of aEuropean ship from which men had bought tobaccoand so forth (Turner 1973). Men in Eastern ArnhemLand described paintings in wet-season rock sheltersas records of foraging expeditions to make the cavelook good, or to show others there was plenty offood in the area (Layton 1992, 74). Art for Arts Sakecorresponds to what we call secular or quotidian artin the following paragraphs.

Sympathetic magic is not a category of the sameorder as those we consider. Magic seemed a plausiblemotive for prehistoric rock art to early research-ers because it was identied as the rst step in thethree-phase theory of human cognitive evolutionproposed by J.G. Frazer in The Golden Bough (rstpublished 1890), where magic precedes religion,which is followed by science. There are three compel-ling reasons for abandoning this interpretation. Manyethnographic cases of sympathetic magic, such asAboriginal increase rites, or shamans dreaming ofhunting success (York et al. 1993, 7071, 136), areembedded within the local religion. Contrary to Fraz-ers understanding, sympathetic magic is oen not

believed to take eect automatically, but to rely on theintervention of spiritual agencies. Second, much of theevidence for supposedly wounded animals in UpperPalaeolithic art is ambiguous (see, for example, Bahn1991). Thirdly, it is impossible, without ethnographicinformation, to tell whether a painting or engravingof a speared animal depicts a wished-for outcome, ora successful hunt. As the Aboriginal artist Namandi-jaugwa explained to David Turner (1973), Nawanna

killed it [a porpoise] and I saw it and painted it. KenMulvaney gives another example (Mulvaney 1996,11). We therefore exclude sympathetic magic, as acategory, from our analysis.

Layton (2000) represented the three alternativehypotheses as a simple four-cell matrix, based ontwo axes: the frequency of particular motifs withinthe assemblage at individual sites, and their relativefrequency across the region occupied by a hunter-gatherer community (Fig. 1). In regions where theanimal art is an aspect of totemism the number ofspecies represented is high, but most are depictedat a small proportion of sites, since each belongs to aparticular clan. Consequently, the frequency of eachmotif in the total sample is low, and the number of sitesat which it appears is small. On the other hand whererock art is associated with shamanism, if the cases

of Lewis-Williams and Whitley are representative,a small number of animal species will predominateboth at individual sites and across the whole region.Secular art represents a kind of null hypothesis thatno species has any special signicance for anyone.All sites in the region have the same characteristics,such that all motifs are depicted with roughly equalfrequency (as in totemism), but every one occurs atall or most sites (as in shamanism). These dierentsituations are represented in the matrix. Note thatthe empty cell represents a permutation for whichwe lacked empirical evidence. It also seemed unlikelythat certain motifs could occur at least twice as oen

as others in the region yet only appear at a smallproportion of sites.

Testing the model

A preliminary surveyLayton (2000) compared a sample from the UpperPalaeolithic with three ethnographic cases repre-senting the three principal interpretations based onethnographic analogy advanced for the Palaeolithic.

Distribution

FrequencyEach motif only appears in

certain sitesThe same motifs are present in

nearly all sites

Certain motifs are present at leasttwice as frequently as the meanfor other motifs

Shamanic [C]

All motifs have low orsimilar frequencies

Totemic [A] Secular [B]

Figure 1.Dierentiation of rock-art traditions by motif frequency.


5/18

5


The Western Kimberley of Australia was taken asrepresentative of the totemic scenario, as there isboth good ethnography and good documentation onsite-by-site motif frequencies (Fig. ??), based partlyon Laytons own eldwork but augmented by data inSchulz (1956). As no site-by-site breakdown for the fre-quency of motifs in the Drakensberg was available, theshamanic scenario was represented by Deacons 1994

data on the Northern Cape of South Africa, relativelyclose to the home of the /Xam. The secular case wasrepresented by the art of Laura in North Queensland,for which Trezise (1971) has published good site dataand which the limited ethnography (Layton 1995)suggests was predominantly secular.

Each data point in Figure 2 represents a particu-lar animal species, coded according to the assemblageto which it belongs. The gure shows that the methoddiscriminates clearly between the three recent cases,each of which illustrates one of the permutations inthe four-cell model. In the ethnographic case whereanimals function as clan totems (the Western Kimber-leys) the curve is conned to the le side of the graph,owing to the fact that each species is present at a lowproportion of sites. The most widely distributed motif,the generic category snake, occurs at 44 per cent ofsites. At the other extreme, the curve for the probableshamanic case, the Northern Cape of South Africa,falls entirely within the right half of the graph, sinceall species depicted occur at a high proportion ofsites. The ve most frequent motifs occur at all sites.The curve for motifs from the probable secular art of

the Laura district of North Queensland is situatedbetween the other two recent cases, since a largenumber of species is depicted and many occur at alarge proportion of sites. All three curves follow thepredictions of the model. The curve for the UpperPalaeolithic sample lies between those for the WesternKimberley and Laura, but is relatively far from thetrajectory of shamanic art.

A larger surveyEncouraged by these results, we collected more com-parative cases and a much fuller data set on the UpperPalaeolithic. The rst of two further ethnographicstudies is from Western Arnhem Land, Australia (dataprovided by Taon). The recent rock art of WesternArnhem Land is, as noted above, another exampleof secular art. Figures depicted with so-called X-ray detail, i.e. realistic rendering of internal organsand tissues (Fig. ??), are identied by local artistsas records of foraging, or just meat (Taylor 1996,1720). Taons Aboriginal instructors showed himhow a dozen dierent species of sh are carefullydepicted with numerous anatomical details (Taon1988), along with other species including lizards andturtles (Fig. ??). A parallel tradition of representinganimals as transformations of totemic ancestors ispractised on bark painting, in which the internalorgans are replaced by geometric cross-hatching(Taylor 1996, 118). The second ethnographic case isthe Stein River, British Columbia. This small area onthe Columbia Plateau has the benet of a long period

0 20 40 60 80 100

30

25

20

15

10

5

0

% of sites at which motif appears

Frequencyofmotif(%

)

Upper Palaeolithic

Laura

Northern Cape

Kimberley

Figure 2.Initial comparison of motif frequency in whole samples and between sites within samples.


6/18

6


Figure ??.Brolgas (Grus rubicunda) or Australian cranes, Hann River, Kimberley, Australia. (Photograph: MichaelRainsbury.)

Figure ??. Barramundi (Lates calcarifer), Arnhem Land, Australia, X-ray period. (Photograph: Paul Taon.)


7/18

7


Figure ??.Springbok (Antidorcas marsupialis), Daureb/Brandberg, Namibia. (Photograph: Tilman Lenssen-Erz.)

Figure ??. Turtles, Laura, North Queensland, Australia. (Photograph: Robert Layton.)


8/18

8


Figure ??.Macropod, Arnhem Land, Australia, Dynamic art period. (Photograph: Paul Taon.)

Figure ??.Cale, Ennedi, Chad. (Photograph: Tilman Lenssen-Erz.)


9/18

9


of ethnographic documentation, starting with JamesTeit (1918), who advised and guided Boass JessupExpedition, and extending to information providedby Annie York between 1989 and 1991 (York et al.1993), which show that rock art was associated bothwith the adolescent vision quest, and the activities ofprofessional shamans.

The Upper Palaeolithic data have been greatlyincreased using data collected by Sauvet and Wlo-darczyk. These data have been divided into regionalsub-sets that allow a spatial and diachronic analysis.Sauvet & Wlodarczyk (200001) explain why theyconsider it safe to distinguish only two time periods;

an archaic period extending from the Aurignacian tothe early Magdalenian, and a later period covering theMiddle and Upper Magdalenian. Available C14 dateslocate this temporal division at around 15,000 . Thetwo periods correspond approximately to the phasesthat Leroi-Gourhan (1965) called Style II andStyle IIIon one hand, and Style IVon the other.

Additional archaeological cases were providedby Ndedema Gorge, a small valley in the north of theDrakensberg Mountains of South Africa fully recorded

by Pager (1971), the High Brandberg in Namibia(Fig. ??), also exhaustively recorded by Pager (1989to 2000), the ancient Dynamic rock art of ArnhemLand (Fig. ??), and the art of Ennedi, Chad (Fig. ??).Lenssen-Erz (2001) and Richter (1991) consider that theart of the High Brandberg (Daureb in the indigenousKhoekhoegowab language), Namibia, emerged from alargely secular context where springbok, the dominantmotif, may have been emblematic of abundance inand aer the rainy season. Pagers detailed recordsallowed us to distinguish six very similar sub regionsin the Daureb/High Brandberg. The Dynamic Art ofArnhem land preceded the so-called X-ray tradition.

The absence of marine and coastal subjects, and thepresence of boomerangs, which are now only usedto the south of Arnhem Land, have been taken byLewis (1988) and other archaeologists as evidencethat Dynamic art was painted during a period ofaridity, when Arnhem Land was far from the northcoast of Australia, probably during the last glaciation.Both Lewis and Taon & Chippindale (2008, 81), linkthe origin of X-ray art to the formation of extensivecoastal wetlands over the past 1500 years. The rock art

Figure ??.Bison, Portel, France, Upper Palaeolithic period. (Photograph: Georges Sauvet.)


10/18

10


of Ennedi, a mountainous region in the northeast ofChad, is our only case not created by hunter-gatherers.The rock art is characterized by a very strong focuson domestic cale, which are present at practicallyall sites. It is likely that the art is an expression ofcale-centred behaviour and ideology analogous tothe cale complex dened by Herskovits (1926),where possession of cale is of far more than simpleeconomic value. This complex has been extensivelydocumented by ethnographers from Evans-Pritchard(1940) onwards, among the Nilotic cale herders ofEast Africa. The art of Ennedi may also have a strongaesthetic component, as Coote (1992) showed in hisstudy of the aesthetics of cale coat colour amongNilotic peoples today.

Unfortunately, we found it impossible to plotthe entire data set for each case on a single graph.

We experimented by ploing only the most frequentmotif, but found its characteristics could be skewedby a single, atypical site and eventually decided tocompromise by ploing each case with a point repre-senting the mean distribution of the two most frequentspecies (Fig. 3).

Figure 3 shows that the case of X-ray art fromArnhem Land sits close to the other ethnographic caseof secular art, Laura. The plot for Western Kimberleytotemic art also lies low on the graph, owing to the

fact that even the most common animal motifs havea low frequency. While the two commonest animalspecies occur at a lower proportion of sites in theWestern Kimberley than at Laura (corresponding tothe models predictions), the two commonest animalspecies are no more widely distributed in WesternArnhem Land than they are in the Kimberley (againstthe models predictions).

The distribution paern for animal motifs duringthe Upper Palaeolithic varied in both time and place,

but the eld of variation represented by the largerUpper Palaeolithic samples sits tantalizingly betweenmost of the ethnographic cases. During the archaicphase of Palaeolithic art the Rhne-Languedoc andQuercy regions approach situations A and B (i.e. nosingle motif dominates, as in totemic or secularart), whereas the Iberian Peninsula (Andalucia and

Cantabria) and Prigord tend rather to situation C(shamanic art), due to the heavy predominance of asingle motif. The dominant species in Spain is red deer(especially female), in Prigord the horse. At the startof the Middle Magdalenian there seems to have beena signicant cultural convergence between Cantabriaand the Pyrenees, in which the structure of the rockart conforms increasingly closely to situation C. Nowthey share the same pre-eminent motif: bison (Fig. ??).Prigord appears at this time to be marginalized, but

0 20 40 60 80 100

30

20

10

0

% of sites at which motif appears(mean of 1st and 2nd most common motifs)

Mea

nfrequencyof1stand2ndmostcomm

onmotifs(%)

40

50

Arnhem X-ray

(Australia)

Kimberley

(Australia)

Laura

(Australia)

Drakensberg

(South Africa)

Northern Cape

(South Africa)

Ndedema

(South Africa)Stein River

(Canada)

Arnhem Dynamic

(Australia)

Mp

Mg

ArAmis

H

P5

P3USh

Prigord Recent

Quercy Recent

Southern Spain Archaic

Cantabrian Archaic

Rhne-Languedoc ArchaicQuercy Archaic

Pyrenees Recent

Cantabrian RecentK

Brandberg

(Namibia)

Ennedi

(Chad)

Upper Palaeolithic

(France & Spain)

Figure 3.Enlarged survey of ethnographic and archaeological case studies.

(What do Mp, Ar, Mg, Sh, U, P5, Amis, K, P3 and H stand for?) - Dora


11/18

11


this is due to an increase in the number of mammothat a few sites the Rougnac eect that we discussbelow. Were it not for the huge numbers of mammothat these sites, horse and bison would be the dominantmotifs as they are in the other regions, and the posi-

tion of Prigord in Figure 3 would lie closer to thoseof Cantabria and the Pyrenees. The only direct hit, soto speak, from the ethnography is the case of SteinRiver. Interesting as this is, we caution that the SteinRiver covers too small an area to guarantee a region-ally representative sample.

The other three archaeological case studies(Arnhem Land Dynamic art, the Daureb/HighBrandberg, and the pastoralist art of Ennedi) all lie inthe upper half of the graph, beyond the range of theUpper Palaeolithic cases and even further from theethnographic ones.

The overall outcome of this comparative exercise

is to show that Upper Palaeolithic rock-art traditionshave no close modern parallel, other than perhaps,the Stein River Valley, and generally display uniquecongurations in the distribution of animal motifswithin and between sites. On the other hand, an inter-esting trend emerges within the Upper Palaeolithic inwhich at least two, possibly three, of the sub-regionsapparently converge on a common paradigm. Whilethis paradigm creates an archaeological signatureapproaching that of South African shamanism, wenote below that this is not necessarily diagnostic. Thearchaeological cases from Africa and Australia furtherextend the range of possible scenarios and highlightthe danger of projecting known ethnographic culturesonto those of the distant past.

A critique of the method

There are a number of methodological and interpre-tive issues raised by our approach.

Coding the dataStatistics based on the frequency of motifs makingup an artistic vocabulary will depend directly uponthe number of motifs recognized in the analysis, yet

because they lack sucient visual information somemotifs are either deliberately ambiguous or appear soto the uninformed observer. Published records of rockart rely, to some extent, on general categories such assnake, birdor shwhen more specic identicationsare impossible. The frequency-distribution graphsobtained if a single generic sh motif is recognizedin X-ray art (Fig. 4a), or if individual species aredistinguished (Fig. 4b) are very dierent, and wouldradically alter the location of the region in Figure 3,

had Taon (1988) not been instructed in the diagnosticiconography of each species. It is possible that the

artists of the Dynamic Period in Arnhem Land rockart distinguished several species of macropod, butwe cannot now discriminate between them and cantherefore only group them as generic category.

A second diculty is the denition of a site ora region. Is every rock shelter a separate site or cana group of adjacent rock shelters or platforms form asingle site? The denition of a site is not always clearin the literature. The scale of the geographical entitytreated as a distinct region may also cause problems. If

0 20 40 60 80 100

60

40

20

0


Frequencyofmotifs(%)

80

100

Arnhem Land recent X-ray paintingsFish divided into 12 species

(10 motifs, 304 sites, 2622 figures)

0 20 40 60 80 100

60

40

20

0


Frequencyofmotifs(%)

80

100

Fish

Macropod

Freshwater

tortoise

Arnhem Land recent X-ray paintingsAll species of fish combined

Figure 4.Distribution paerns for X-Ray art, where(a) all sh grouped as sh, (b) individual species ploedseparately. In Fig 4a why are some dots labelled andothers not? - Dora


12/18

12


the region is dened too broadly, its homogeneity maybe questionable (e.g. Western Arnhem Land), but if thescale is too small there is a risk that the corpus will notbe representative of variability within the region. Inthe Northern Cape region of South Africa each sitecontains an average of 230 gures (Deacon 1994), but

the Stein River sample from British Columbia consistsof a mere dozen sites containing fewer than 150 guresin total. The homogeneity of Northern Cape sites may

be overrated, while the Stein River may not reect thefrequency of motifs at a regional level.

Our method depends on gures accumulatingfor long enough to reveal the generative principlesexpressive of cultural context. If the motive for paintingis the same at each site, and if all local groups sharethe iconography, one can predict that the diversity ofmotifs present will increase in relation to the numberproduced at the site (Kintigh 1989). We tested this

hypothesis by measuring the diversity of motifs presentat a site (expressed in terms of the proportion of thetotal vocabulary of motifs that are represented in theregion) as a function of the size of the site (expressedas a percentage of the total number of gures in thesample that are present at that site). This test was car-ried out for the Western Kimberley region (totemic), theLaura region (secular) and for two regions within theUpper Palaeolithic tradition (Archaic period Quercyand Recent period Pyrenees) (Fig. 5).

In the Western Kimberley, the diversity of motifswithin a site remains low and increases lile withthe number of gures at the site, since each site isassociated with a particular totemic species. On theother hand, in both the Laura region and the UpperPalaeolithic, the diversity of motifs increases substan-tially with site size. This analysis distinguishes neatlybetween an art based on totemism and one in whichall local groups depict the same motifs, but it doesnot allow us, in the laer case, to distinguish betweenalternative possible functions. One of the two regionsof the Upper Palaeolithic Archaic Quercy is situ-ated in the graph Figure 3 quite close to the Kimberley,

but Figure 5 shows this is misleading. As the numberof gures at sites increases its iconographic signaturestarts to conform to that of the other Franco-Cantab-rian regions. This complementary analysis is thus ameans to verify whether the dierent sub-assemblages

within a single region are homogeneous or not.The problem of multiple themes in a regions rock artOur analysis has only examined distribution paernsamong animal species represented in the art. Whileour results tend to conrm that these paerns varyaccording to the cultural theme expressed throughanimal art, we cannot disregard the fact that manyethnographic rock-art traditions express more thanone theme in the culture.

0 10 20 30 40 50

60

40

20

0

Proportion of total corpus of motifs in the region at site (%)

Pro

portionoffiguresinregionfoundatsit

e(%)

80

100

Laura

Pyrenees Recent

Quercy Archaic

Kimberley

Figure 5.The eect of increased sample size on heterogeneity of motifs present.


13/18

13


The rock art of Laura, North Queensland,includes a number of sites where human head-dressedgures are depicted. Rosenfelds (1982) archaeologicalresearch showed that there were two types of site atLaura. One is lled with animal paintings of numerous

species, oen superimposed in many layers. Thesetend to be found in rock shelters with substantialhabitation debris. The head-dressed humans arealmost always painted on otherwise empty walls insites with lile habitation debris, suggesting a ritualtheme at sites to which access was restricted. In thiscase, it seems that the ceremonial dimension of therock art is clearly segregated from the animal com-ponent, but the paerning of the animal componentdoes not imply that the Aboriginal people of Lauralacked ceremonies.

The rock art of both Laura and the WesternKimberleys contains sorcery paintings. In the Kim-

berleys these consist of small, distorted human guresportraying the desired state of the victim. They arepainted close to the ancestral totemic gures in thehope that the ancestors will punish the victim forbreaking the code of social behaviour (Layton 1992,856). They do not have their own paern of animalassociations. In the Laura area, sorcery paintings doimpinge on the frequency of animal gures. Sorcerypaintings are the only type of rock art on which Trezisewas given information based on the personal experi-ence of his Aboriginal guides. He was told that snakeor catsh were sometimes painted next to the paintingof a sorcery victim in the hope the animal would harmthe victim (Trezise 1969, 108). Although catsh is, bya small margin, the most common species depictedin rock shelters (at 14 per cent), catsh and snakeare only two of the 18 species categories identied inLaura rock art. Sorcery therefore has lile eect on theoverall distribution of animal motifs.

The inuence of the religious signicance ofanimal species may be more pervasive in the secularart of Western Arnhem Land. Many of the speciesrepresented with X-ray details also appear in thesacred art represented on bark paintings. Initiationintroduces young people to the sacred signicance

of some features or qualities of animals, particularlyin species such as crocodiles and carnivorous sh.The location of recent Arnhem Land rock art maybe related, therefore, to the fact that much of it isa combination of secular and totemic (see Taon &Chippindale 2008).

The art of the Stein River valley includes sitesassociated with both the adolescent vision quest andthe restricted activities of shamans. The valuable earlyethnography of Teit and Harlan Smith is based on

interviews and site visits at a time when rock art wasprobably still being made, and is supported by thetestimonies of Chinook people living on the Colum-bia River a hundred years later (Layton 2006). Boashad a high opinion of Teits skill as an ethnographer

(Joanitis 1991, 186). While the vision quest is a questfor a personal spirit guardian during a heightenedform of consciousness, it is not a form of shamanism.Our Chinook instructors on the Columbia River werealso adamant that entoptic forms would interfere withthe clarity of vision sought, in which you see in yourminds eye the power in everything, even in a bladeof grass (Layton 2006, 93n). Only a few particularlyskilled individuals proceed to the specialized trainingrequired of shamans. Their rock-art sites are privateand may be characterized by a dierent frequency ofmotifs (see York et al. 1993, 114; Layton 2006, 89), butaccess to, and knowledge of, such sites is restricted.

Shamanism and totemism co-exist on the NorthwestCoast. Daly (in York et al. 1993, xvii) reports that as oneapproaches the coast the possession of a special bodyof knowledge by each family becomes more formal-ized. We would welcome a statistical comparison ofthe distribution of animal motifs in rock art across theColumbia Plateau and on the Northwest Coast if suchdata are available.

The extent to which South African rock art isexclusively concerned with trance experience has longbeen debated among South African researchers. Itseems likely that at least some of the art relates to otheraspects of San culture (Guenther 1994; Parkington &Manhire 1997). Dowson (2007) has recently advocatedcalling a truce in this debate. He argues that the Carte-sian distinction Western researchers make between themundane action of hunting and the trance activities ofcurers obscures the sense that hunter-gatherers have ofhunting and gathering as a dialogue between sentienthumans and sentient non-human beings, includinganimals. Povinalli (1993) has made the same pointwith reference to Australian hunter-gatherers. Impor-tant as this insight is, many hunter-gatherer cultureshave techniques, such as the use of particular styles,locations or occasions, for foregrounding particular

contexts such as the ritual versus the everyday. Amongthe Kwakiutl, for example, shamanic experiences wereassociated with the winter secret societies (Boas 1996),while in Australia totemic ceremonies are performedat restricted locations.

How predictive is our model?

This article began by exploring the archaeological sig-natures of totemic, shamanic and secular art because


14/18

14


these were, historically, the three most favouredhypotheses for the cultural context of Upper Palaeo-lithic cave art. However, while they are exemplars of thethree possible congurations A, B and C identied inFigure 1, they may not be the only contexts that would

generate the three observed distribution paerns.Hunter-gatherers typically rely on a wide range

of animal and plant foods and one might reasonablypredict that, at least within a single ecological zone,a similarly wide range of animals will gure in thesecular or quotidian art of neighbouring bands whereno species evokes particular social signicance, but itis unlikely that any rock, or other art tradition locatedin cell B will be devoid of symbolism (see Layton 2000,170). The diagnostic quality is that no image has beenappropriated by a particular local group, nor has anyspecies been singled out for overriding aention. Totake an example far removed from rock art, Chinese

embroidery draws on a national tradition in whicha variety of owers express wishes for good luck,prosperity and many children. All these motifs areavailable to any needlewoman and can be seen in vil-lages across many parts of China (Williams 1976).

Shamanism is clearly not the only religiouscontext in which a few animal species might befavoured by all artists, and which would therefore bepositioned in cell C. To researchers whose principleethnographic contact with religion is Christianity, thecase of the lamb will come to mind. If any researcherhas data on a rock-art tradition associated with amonotheistic religion, we would welcome the resultsof an analysis of motif frequency and distribution. Thepre-eminence of bison in the Magdalenian art of thePyrenees is clearly compatible with any religion inwhich the bison played a pre-eminent role. Nor doesthere seem to be a necessary reason why art associatedwith shamanism should fall into cell C. Other casescan be found in the ethnography where shamanismis associated with a pre-eminent species, such as thejaguar in lowland South America (Campbell 1989), orthe bear in Siberia (Jordan 2003). Marshall (1969), onthe other hand, documented rapid changes in the fash-ionable trance rituals of the Kalahari, each associated

with a dierent species, as does Tsuru (1988) amongthe Baka of Central Africa. The predominance of asingle species is not a necessarycondition for shamanicrock art, especially if a long enough period is underconsideration. Moreover, the case of the herder rockart of the Ennedi demonstrates that the all-embracingesteem of a certain species, without implying trance asa ritual medium, may generate paerns of art produc-tion that t into cell C and thus supercially compareclosely with presumable shamanic art.

Among more complex societies there may be avariety of motives for identifying dierent sites in aculturally-constructed geography with dierent ani-mal emblems. In the type of small-scale, egalitarian,clan-based societies discussed in this article, however,

the association of a type A distribution paern withtotemism seems more secure. It follows, almost bydenition, that the association of dierent specieswith dierent local groups is a form of totemism.Two case studies illustrate the potential of rock artto elucidate the origin of totemic traditions in suchsocieties. Olsen (1989) used the evidence of the spatialdistribution of art motifs to trace the origin of the typeof clan totemism practised today by Hopi and Zuniclans, to their predecessors in the Anasazi farmingculture that emerged around 500. Portable Anasaziobjects tended to carry geometric designs, while xedlocations such as rock walls, boulders and kivawalls

bore animal and human representations. Thirty-eightper cent of the Anasazi vocabulary of animal motifswere identical in form to modern clan emblems. Sincethey were non-randomly distributed in the landscape,Anasazi art provides evidence for the antiquity of theclan system and its expression in rock art.

In Australia, the development of the smalltool tradition appears to be associated with a recon-guration of relationships between people and thelandscape. The ancient (pre-5000 ) core tool andscraper stone tool tradition can be characterized as anexpedient stone technology. Gould (1969) and Hayden(1979), for instance, observed choppers similar toancient core tools in use in central Australia. Theynoted that such choppers are sometimes made whenneeded to fell trees or butcher kangaroo meat, andthen le at the site. However, the more recently addedsmall tool tradition is a curated technology. Suchtools are mounted on wooden shas for future use.Excavation of camp sites associated with the small tooltradition shows that stone tools were manufacturedin camp and worn out or broken tools were discardedon return to camp (see review in Flood 1995, 221.).During the same period (c.60004000 ), the ancientdynamic style of rock painting in western Arnhem

Land is replaced (or supplemented) by a new stylewhose themes include composite Rainbow Serpentgures (Lewis 1988, 1517). Lewis makes the interest-ing suggestion that composite Rainbow Serpent paint-ings are symptomatic of new, regional religious cultsstimulated by rising post-glacial sea-levels (Lewis1988, 39, 109; cf. Taon et al.1996). Indirect evidencesuggests the totemic, Wandjina style of rock art in theKimberleys is at least 3000 years old, since the earlierstyles do not depict the use of stone-tipped spears


15/18

15


and Kimberley stone spear points, characteristic ofthe small tool tradition, are thought to appear at about3000 (see Welch 1990, 123; Crawford 1968, 8990;Crawford 1977, 358). It is therefore possible that clantotemism crystallized in parts of Australia during the

reconguration of social relationships associated withthe small tool tradition.

Conclusion

Although we have used the three dominant hypothesesfor the cultural context that generated the productionof Upper Palaeolithic rock art to set the parameters forour analysis, once the categorical distinctions of Figure1 are replaced by the numerical axes of Figures 2 and3, it becomes possible to plot the location of dierentrock-art traditions in a two dimensional space whereethnographic and archaeological case studies are put

on the same footing. Each, as a possible cultural world,can occupy a unique position in that space. This is par-ticularly demonstrated by the cases of Ennedi and theDaureb/High Brandberg, which lie far from any of theethnographic cases. One of the unexpected outcomesof ploing Figure 3 is the range of variation in timeand space revealed within the Upper Palaeolithic,which might otherwise have been obscured by theremarkable uniformity in the style of the animal artduring the 20,000 years from Chauvet to Niaux. Whilethe failure of the ethnographic cases which we haveavailable, other than perhaps the Columbia Plateau, tomatch the Upper Palaeolithic closely is disappointing,it is not entirely surprising to nd that the culturalcontext of Upper Palaeolithic animal art has lile orno exact modern parallel. This is a useful correctiveto the slavish application of ethnographic models,

but it does suggest distribution paerns in the art ofthe Columbia Plateau (within which the Stein Riveris situated) deserves further analysis.

While our analysis allows us to identify totem-ism as the least probable of cultural contexts for theanimal art of the Upper Palaeolithic, we have notedOlsens (1989) and Hartleys (1992) productive use ofsimilar methods and shown how the method can be

used in Australia and the American Southwest to datethe probable emergence of recent totemic systemsfrom less well understood precursors. We hope otherresearchers will nd the method suciently interest-ing to apply it to other recent and ancient bodies ofrock art and have noted two cases where we believethe collection and analysis of further data wouldparticularly help to take our approach further.

In summary, the structural parameters express-ing the frequency of motifs in a region and the

distribution of these motifs between sites within theregion provide important lines of evidence point-ing to the cultural context of rock art in small-scalesocieties. If the scheme that emerges from the abovediscussion is more complex than that originally

suggested by Layton (2000), it still enables us to testalternative hypotheses. Even if we abandon the ideathat there is a direct and unambiguous relationshipbetween the frequency and distribution of motifsand totemic organization or shamanic practices, thethree distribution paerns identied by our cells A,B and C are likely to be generated by the use of rockart in dierent contexts. The analysis of rock art inthese structural terms reveals similarities betweenhistorically unrelated rock art traditions and helps usto plot continuity or change in cultural traditions overhundreds or thousands of years.

Acknowledgements

This article is dedicated to the memory of Peter Ucko andAndre Rosenfeld: two inspirational leaders in rock-artresearch. The research was partly conducted within theframework of the Emergence et fonction des systmes smi-ologiques dans les groupes humains du Palolithique suprieurwhich was funded by the CNRS programme Origine delHomme, du Langage et des Langues.

Note

The data from which Figure 2 is constructed are published

in Layton 2000. The data from which Figure 3 is constructedare published in Sauvetet al. 2006.

Georges SauvetCentre de Recherche et dEtude de lArt Prhistorique

UTAH (UMR 5608)Post code??Toulouse

FranceEmail: ???????????????

Robert LaytonDepartment of Anthropology

Durham University

Dawson BuildingSouth RoadDurham

DH1 3LEUK

Email: [email protected]:

Tilman Lenssen-ErzAfrican Research Unit

Institute of Prehistoric Archaeology


16/18

16


University of CologneJennerstr. 8

50823 CologneGermany

Email: [email protected]

Paul TaonSchool of Humanities (Gold Coast)

Building G30Arts and Education 1

Gold Coast campusGrith University

Parklands DriveSouthport, Queensland 4222

AustraliaEmail: ???????????????

Address correct?

Andr WlodarczykCentre de Linguistique thorique et appliquMaison de la Recherche (CELTA - EA 3553)

Universit de Paris-Sorbonne (Paris 4)28, rue Serpente

75006 ParisFrance

Email: ???????????????

References

Bahn, P., 1991. Wheres the beef? The myth of hunting magicin Palaeolithic art, in Rock Art and Prehistory, eds. P.

Bahn & A. Rosenfeld.Oxford: Oxbow, page nos??.Barnard, A., 1992. Hunters and Herders of Southern Africa.

Cambridge: Cambridge University Press.Beaune, S. de, 1998. Chamanisme et prhistoire: un feuilleton

pisodes. LHomme147, 20319.Bleek, D.F., 1935. Beliefs and customs of the /Xam bushmen,

part VII: Sorcerors. Bantu Studies9, 147.Bleek, D.F., 1936. Beliefs and customs of the /Xam bush-

men, part VIII: More about sorcerors. Bantu Studies10, 13162.

Campbell, A., 1989. To Square with Genesis: Causal Statementsand Shamanic Ideas in Wayapi.Edinburgh: EdinburghUniversity Press.

Cartailhac, E. & H. Breuil, 1906. La Caverne dAltamira San-

tillana, prs Santander (Espagne). Monaco: Imprimeriede Monaco.

Cloes, J. & J.D. Lewis-Williams, 1996. Les Chamanes de laPrhistoire: Transe et Magie dans les Groes Ornes. Paris:ditions Seuil.

Coote, J., 1992. Marvels of everyday vision: the anthropol-ogy of aesthetics and the cale-keeping Nilotes, in

Anthropology, Art and Aesthetics, eds. J. Coote & A.Shelton.Oxford: Oxford University Press, 24573.

Crawford, I., 1968. The Art of the Wandjina.Oxford: OxfordUniversity Press.

Crawford, I., 1977. The relationship of Bradshaw and Wand-jina art in northwest Kimberley, in Form in IndigenousArt, ed. P.J. Ucko. Canberra: Australian Institute ofAboriginal Studies, 35769.

Deacon, J., 1994. Rock engravings and the folklore of Bleekand Lloyds /Xam San informants, in Contested Images:Diversity in Southern African Rock Art Research, eds. T.Dowson & D. Lewis-Williams. Johannesburg: Witwa-tersrand University Press, 23756.

Dowson, T., 2007. Debating shamanism in Southern Africanrock art: time to move on. South African ArchaeologicalBulletin62, 4961.

Eliade, M., 1972. Shamanism: Archaic Techniques of Ecstasy.London: Routledge.

Evans-Pritchard, E.E., 1940. The Nuer. Oxford: OxfordUniversity Press.

Flood, J., 1995.Archaeology of the Dream Time. Revised edition.Glasgow: Collins.

Gould, R.A., 1969. Yiwara: Foragers of the Australian Desert.Sydney: Harper-Collins.

Guenther, M., 1994. The relationship of bushman art toritual and folklore, in Contested Images: Diversity inSouthern African Rock Art Research, eds. T. Dowson& D. Lewis-Williams. Johannesburg: WitwatersrandUniversity Press, 25773.

Hartley, R.J., 1992. Rock Art on the Northern Colorado Plateau.Aldershot: Avebury.

Hayden, B., 1979. Palaeolithic Reections: Lithic Technologyand Ethnographic Excavations among the Australian

Aborigines.Canberra: Australian Instituteof Aborigi-nal Studies.

Helvenston, P. & P. Bahn, 2005. Waking the Trance Fixed.Louisville (KY): Wasteland Press.

Herskovits, M.J., 1926. The cale complex in East Africa.

American Anthropologist28, 23072.Hultkrantz, A., 1989. The place of shamanism in the his-

tory of religions, in Shamanism: Past and Present, eds.M. Hoppl & O.J. von Sadovszky. Los Angeles (LA):Fullerton, 4351.

Joanitis, A., 1991. From the Land of the Totem Poles.New York(NY): American Museum of Natural History.

Jordan, P., 2003. Material Culture and Sacred Landscape: theAnthropology of the Siberian Khanty. Walnut Creek(CA): Altamira.

Kintigh, K., 1989. Sample size, signicance, and measures ofdiversity, inQuantifying Diversity in Archaeology, eds.R.D. Leonard & G.T. Jones. Cambridge: CambridgeUniversity Press, 2536.

Kirchner, H., 1952. Ein archologischer Beitrag zur Urges-chichte des Schamanismus.Anthropos47, 24486.Kuhn, H., 1956. The Rock Pictures of Europe. London: Sidgwick

and Jackson.Layton, R., 1992Australian Rock Art: a New Synthesis. Cam-

bridge: Cambridge University Press.Layton, R., 1995. Rereading rock art: text and discourse, in

Perceiving Rock Art: Social and Political Aspects, eds. K.Helskog & B. Olsen. Oslo: Novus forlag, 21727.

Layton, R., 2000. Shamanism, totemism and rock art: Leschamanes de la prhistoire in the context of rock art


17/18

17


research. Cambridge Archaeological Journal 10(1),16986.

Layton, R., 2006. Habitus and narratives of rock art, in Talk-ing with the Past: the Ethnography of Rock Art, eds. J.D.Keyser, G. Poetschat & M. Taylor. Portland (OR): TheOregon Archaeological Society, 7399.

Lenssen-Erz, T., 2001. Gemeinscha Gleichheit Mobilitt.Felsbilder im Brandberg, Namibia und ihre Bedeutung .Cologne: Heinrich-Barth-Institut.

Leroi-Gourhan, A., 1965. Prhistoire de lArt Occidental. 1stedition. Mazenod: Paris.

Lvi-Strauss, C., 1962. Totemism, trans. R. Needham. London:Merlin.

Lvi-Strauss, C., 1964. The Savage Mind. London: Weidenfeldand Nicolson.

Lewis, D., 1988. The Rock Paintings of Arnhem Land: Social,Ecological and Material Culture Change in the Post-GlacialPeriod.(British Archaeological Reports 145.)Oxford:BAR.

Lewis, I.M., 1971. Ecstatic Religion: an Anthropological Studyof Spirit Possession and Shamanism. Harmondsworth:Penguin.

Lewis-Williams, D., 1981. Believing and Seeing: SymbolicMeanings in Southern San Rock Paintings. London:Academic Press.

Lewis-Williams, J.D. & T.A. Dowson, 1988. The signs of alltimes: entoptic phenomena in Upper Paleolithic art.Current Anthropology29, 20145.

Marshall, L., 1969. The medicine dance of the !Kung bush-men.Africa 39, 34781.

Mulvaney, K., 1996. What to do on a rainy day: reminis-cences of Mirriuwung and Gadjerong artists. Rock ArtResearch13, 320.

Olsen, N., 1989. Social roles of animal iconography: impli-

cations for archaeology from Hopi and Zuni ethno-graphic sources, inAnimals into Art, ed. H. Morphy .London: Unwin, 41739.

Pager, H., 1971. Ndedema Gorge: a Documentation of the RockPaintings of the Ndedema Gorge. (2 vols.) Graz: Akad-emische Druck- und Verlagsanstalt.

Pager, H., 1989. The Rock Paintings of the Upper Brandberg,part 1:Amis Gorge. (2 vols.) Cologne: Heinrich-Barth-Institut.

Pager, H., 1993. The Rock Paintings of the Upper Brandberg,part 2:Hungorob Gorge. (2 vols.) Cologne: Heinrich-Barth-Institut.

Pager, H., 1995.The Rock Paintings of the Upper Brandberg,part 3:The Southern Gorges. (2 vols.) Cologne: Hein-

rich-Barth-Institut.Pager, H., 1998. The Rock Paintings of the Upper Brandberg,part 4:Umuab and Karoab Gorges. (2 vols.) Cologne:Heinrich-Barth-Institut.

Pager, H., 2000. The Rock Paintings of the Upper Brandberg,part5: Naib Gorge (A) and the Northwest.(2 vols.) Cologne:Heinrich-Barth-Institut.

Parkington, J. & A. Manhire, 1997. Processions and groups:human gures, ritual occasions and social categoriesin the rock paintings of the Western Cape, SouthAfrica, in Beyond Art: Pleistocene Image and Symbol, eds.

M. Conkey, O. Soer, D. Stratmann & N. Jablonski. SanFranciso (CA): Memoirs of the California Academy ofSciences 23, 30120.

Povinalli, E.A., 1993. Might be something: the languageof indeterminacy in Australian Aboriginal land use.

Man28, 679704.Reichel-Dolmato, G., 1978. Drug-induced optical sensa-

tions and their relationship to applied art among someColombian Indians, inArt in Society, eds. M. Green-halgh & V. Megaw. London: Duckworth, 289304.

Reinach, S., 1903. LArt et la magie propos des peintureset des gravures de lAge du Renne.Anthropologie14,25766.

Richter, J., 1934. Studien zur Urgeschichte Namibias. Cologne:Heinrich-Barth-Institut.

Rosenfeld, A., 1982. Style and meaning in Laura art.Mankind13, 199217.

Sauvet, G., R. Layton, T. Lenssen-Erz, P. Taon & A. Wlo-darczyk, 2006. La structure iconographique dun artrupestre est-elle une clef pour son interprtation?Zephyrus59, 195208.

Sauvet, G. & A. Wlodarczyk, 20002001. El arte parietal,espejo de las sociedades paleolticas. Zephyrus5354,21538.

Schulz, A., 1956. North-west Australian rock paintings.Memoirs of the National Museum of Victoria 20, 757and plates.

Spencer, B. & F. Gillen, 1899. The Native Tribes of CentralAustralia.London: Macmillan.

Taon, P.S.C., 1988. Identifying sh species in the recentrock paintings of western Arnhem Land. Rock ArtResearch5, 315.

Taon, P.S.C. & C. Chippindale, 2008. Changing places:ten thousand years of north Australian rock-art

transformation, in Time and Change: Archaeological andAnthropological Perspectives in the Long-Term in Hunter-Gatherer Societies, eds. D. Papagianni & R. Layton.Oxbow Press: Oxford, 7394.

Taon, P.S.C., M. Wilson & C. Chippindale, 1996. Birth of theRainbow Serpent in Arnhem land rock art and oralhistory.Archaeology in Oceania 31, 10324.

Taylor, L., 1996. Seeing the Inside.Oxford: Oxford UniversityPress.

Teit, J., 1918. Notes on Rock Paintings in General, 1918.Unpublished manuscript, Glenbow Alberta InstituteArchives.

Tilley, C., 1991.Material Culture and Text: the Art of Ambiguity.London: Routledge.

Trezise, P., 1969. Quinkan Country.Sydney: Reed.Trezise, P., 1971. Rock Art of Southeast Cape York. Canberra:Australian Institute of Aboriginal Studies.

Tsuru, D., 1998. Diversity of ritual spirit performancesamong the Baka pygmies in southeastern Cameroon.

African Study Monographs, Supplement25, 4784.Turner, D., 1973. The rock art of Bickerton Island. Oceania

43, 286325.Vinnicombe, P., 1976. People of the Eland. Pietermaritzburg:

University of Natal Press.Vitebsky, P., 1995 The Shaman: Voyages of the Soul.London:


18/18

18


Macmillan.Welch, D., 1990. The bichrome art period in the Kimberley,

Australia. Rock Art Research7, 11024.Whitley, D., 1994. By the hunter, for the gatherer: art, social

relations and subsistence change in the prehistoricGreat Basin. World Archaeology25, 35673.

Williams, C.A.S., 1976. Outlines of Chinese Symbolism and ArtMotives. 3rd edition. New York (NY): Dover.

York, A., R. Daly & C. Arne, 1993. They Write their Dreams onthe Rock Forever. Rock Writings in the Stein River Valleyof British Columbia. Vancouver: Talonbooks.

Author biographies c. 50 words each

Documents

Sauvet Et Al CAJ2009