ABSTRACT: Mythology has long been discounted as a key to understanding San rock art, even though explanations of paintings by nineteenth century San commentators explicitly linked paintings of therianthropes to the mythological past. An interpretation which reconsiders the importance of San mythology in relation to rock paintings strongly suggests that figures with both animal and human features are not trancers or shamans, as the dominant model suggests. Rather, these and other images are better understood in relation to San myths, and to beliefs about the spirits of the dead. A model which focusses on myth, history and the spatio-temporal schemata of San cosmology and culture is outlined. This model addresses problems of the “ethnographic method”, and of interpretations which assign priority to the role of the shaman in the production of San visual imagery.
The idea that San rock art is best understood in the context of San religious beliefs and cosmology dates to the nineteenth century. In 1874, it seemed that the affiliation of art and mythology had been clearly demonstrated (Orpen 1874; WHI Bleek in Orpen 1874). However, empiricist studies in the first half of the twentieth century tended to repudiate this claim (cf. Lewis-Williams 1983, 1984a; Lewis-Williams and Loubser 1986; Lewis-Williams and Dowson 1994). Rather, a range of contexts for understanding the art was proposed. Wilman (1968 :63-4), for example, considered different categories of engraved images as variously “utilitarian” or “Art for Art’s Sake”, while accepting that some probably had religious import. The putative religious context of the art was revived when academic empirical studies (e.g. Maggs 1967; Vinnicombe 1967) directed attention to the apparent symbolic importance of paintings of eland. The religious components of the art were skilfully analysed by Vinnicombe (1972, 1975, 1976), by reference to both mythology and ritual practice. Subsequently, in the works of Lewis-Williams, and later his colleagues, links to ritual gained ascendancy, in a model which increasingly foregrounds the role of the shaman, altered states of consciousness and shamanic experience and visions (e.g. Lewis-Williams 1977, 1980, 1981, 1985, 1987, 1991, 1992, 1995; Lewis-Williams and Dowson 1988, 1989; Lewis-Williams and Loubser 1986). In that model, San mythology appears as an epiphenomenon of trance experience. Although a recursive relationship is acknowledged (Lewis-Williams and Loubser 1986), the subordination of myth has been seen as problematic (Guenther 1994). In this paper, I outline a model which retains several of the important insights afforded by the shamanistic model. It engages with key questions regarding the use of San ethnographies for understanding the rock art, as well as problems of understanding the diversity of San artistic production and belief.
Rock art, mythology and complexes of religious belief
In 1874 JM Orpen published a paper which recounted explanations of Lesotho paintings by a San man named Qing, who had acted as Orpen’s guide on colonial business. Having encountered painted shelters en route, Orpen asked Qing what the paintings of men with rhebok’s heads signified. Qing replied that they were “men who had died and now lived in rivers, and were spoilt at the same time as the elands and by the dances of which you have seen paintings” (Orpen 1874:2; original italics). Orpen enquired further about the spoiling of the eland, and Qing responded by narrating a series of myths and stories, beginning with information on Cagn, the San creator and trickster deity, and concluding with a reference to the “dance of blood” which has been astutely linked by Lewis-Williams (1980) to the trance dance, as anthropologically observed amongst Kalahari San-speaking peoples. Orpen’s copies of the Lesotho paintings were shown to the /Xam man, Dia!kwain (then resident with Wilhelm Bleek and Lucy Lloyd in Cape Town), and his interpretation recorded. On the basis of these commentaries, Bleek declared that “the fact of Bushman paintings, illustrating Bushman mythology has first been demonstrated by this paper of Mr Orpen’s” (WHI Bleek in Orpen 1874:13).
The same account was used by Lewis-Williams (1980) to establish the primacy of ritual and hallucinatory experience for understanding the art. Lewis-Williams argued that “the connexion between ...religious ideas and the art lay in a San ritual which neither Bleek nor Orpen fully understood” (1980:468) - namely, the trance dance. It was argued that Qing’s comments (cited above) constitute “a threefold metaphorical statement: the therianthropes do not depict mythical, subaquatic people on whom no further information was obtained, but medicine men who have ‘died’ or been ‘spoilt’ in trance, and whose experience is analogous to being under water” (Lewis-Williams 1980:475). On the contrary, I suggest that Qing’s comments are best interpreted not as trance metaphors, but, in the first instance, in relation to San mythology and religious beliefs.
Taking Qing’s comments in the context of Orpen’s paper it is, first, possible to argue that “spoilt” is not a trance metaphor. Asked about the spoiling of the eland, Qing commenced with some background on Cagn, before recounting his first narrative. This is the myth of the spoiling of the eland (Orpen 1874:2-5). Briefly, the story relates how Cagn’s wife, Coti, gave birth to a baby eland, which Cagn reared in a secluded kloof. But Cagn’s sons unwittingly killed it while out hunting, to Cagn’s anger. Ultimately Cagn instructed his wife to “churn” the eland’s blood together with the fat from its heart, in a pot. The mixture was then “sprinkled” and the drops turned first into eland bulls, then into eland cows: “they churned and produced multitudes of eland and the earth was covered with them” (Orpen 1874:4). The penultimate sentence concludes: “That day game was given to man to eat, and this way they were spoilt and became wild” (Orpen 1874:5).
Although Lewis-Williams has suggested that Orpen “fabricated the complex and misleading conflation I have tried to unravel”, partly by making Qing’s replies “consecutive” (1980:475), there is a logic in Orpen’s text which suggests otherwise. To understand Qing’s comments on the painted therianthropes, the testimony needs to be seen in the wider context of San mythology. This shows remarkable formal similarities over space and (to a lesser extent) time (Schapera 1965 , Guenther 1989, Schmidt 1991, Barnard 1992) - extending, according to Barnard, “even across the hunter-herder divide” (Barnard 1992:252). Virtually ubiquitous beliefs centre on what Guenther (e.g. 1986, 1989) has called “primal time” - a mythological past, populated by the predecessors of modern San, and by animals who were also human. A second creation event established the current order: animals were separated from humans, becoming merely animals, while the “first people”, “people of the early race” or (the literal /Xam translation) “first-at-sitting people”, became modern San (Bleek and Lloyd 1911; Hewitt 1986:*). In /Xam and other myths, these ancestral people were thought of as ignorant and uncivilised. Ambiguously, they were seen both as powerful (as the creators of various phenomena) and potentially dangerous. After the “second creation”, people acquired manners and customs, but also mortality. The “Origin of Death” narrative (e.g. Bleek and Lloyd 1911:56-65) which explains how people came to “die althogether” or “outright” is also apparently ubiquitous amongst San-speaking peoples (Guenther 1989).
Qing’s reference to the spoiling of the eland and the narrative summarised above fit unequivocally into this mythological schema: “spoiling” refers here to the separation of humans and animals, and to the San mythological past. The structure of Orpen’s text strongly suggests that this narrative was specifically related in response to Orpen’s request for clarification about the men who were spoilt at the same time as the elands. Independent supporting testimony came from Dia!kwain. Bleek showed him Orpen’s copies of the same paintings on which Qing had commented, before Qing’s comments had been received (Lewis-Williams 1980:469). Dia!kwain too unequivocally linked the painted therianthropes from the sites of Melikane and the Kraai River to the mythological past, stating (in Bleek’s words) that “They [the Melikane therianthropes] are said to belong to the ancient Bushmen, or to the race preceding the present Bushmen, and who it is believed killed people”...[The Kraai River therianthropes] are believed to belong to these same people” (Bleek in Orpen 1874:13). (More recently, !Kung people in Botswana commented to Wilmsen (1986) on copies of Drakensberg paintings; their interpretation of therianthropes also linked them to the time when creation was in progress). The question arises: if the therianthropic figures are indeed mythological beings or ancestral San, how can they also be San shamans in altered states of consciousness? The solution lies in understanding the significance of “death” and “underwater” in San mythology and cosmology - both interpreted by Lewis-Williams as “trance metaphors” (1980).
Examination of a range of San myths and religious accounts reveals the existence of common and widespread beliefs in dead, mythological, subaquatic or subterranean beings, who may be either malevolent or beneficent. Formal similarities, amongst diverse (Khoi)San peoples, concerning mythical personages, deities and beliefs about death are striking. Generally speaking, two deities are important, although amongst some groups aspects of their identities have become merged, or fused into a single personage (Marshall 1962). Marshall’s account of !Kung religion is described first, because of her awareness of changes through time in the details of belief.
(Khoi)San religious beliefs
The !Kung in the nineteen fifties believed in a supreme creator figure, Gao!na, who lived in the eastern sky (Marshall 1962:223) and (formerly) a death-giver, //Gauwa, associated with the west and the setting sun. Marshall (1962:234) describes how, at the time of her field work, Gao!na had become associated with the power over death, and malevolent powers formerly associated with //gauwa. Also related to //Gauwa are the //gauwasi, or spirits of the dead, who “come to earth and enter into the affairs of men”; Marshall (1962:241) also recounts peoples’ vivid belief in these spirits, and the fear with which they were regarded (see also Barnes n.d.). //Gauwa and the //gauwasi are associated with sickness and death; it is they with whom the curers intercede in the curing ritual and trance. Curers might see //Gauwa at the dance, but Marshall (1962:240-241) was also told that anyone might see him in dreams.
Silberbauer’s account of G/wi religious beliefs similarly features a supreme creator being, N!adima, and a death-giver, G//amama. The latter is associated with death and illness; and in an account virtually identical to the /Xam, he is believed to cause illness by shooting small arrows into people (especially women) (Silberbauer 1981:54-5; DF Bleek 1935:-6). A more recent account from the Central Kalahari is given by Valiente-Noailles (1993:187-201): he describes a creator being, g//ama, and an evil death-giver named !kaonxa (who sometimes appears also as a less malignant being). In both these ethnographies, the spirits of the dead are described. Silberbauer refers to a subterranean realm, occupied by these spirits (g/amadzi) and “monsters” or “angry things” (//a:xudzi) who surface and “impinge on human lives when angered by the breaking of certain taboos” (Silberbauer 1982:113). Valiente-Noailles (1993:196-99) obtained a detailed account of !kaonxa as the source of evil from a //gana man. !Kaonxa was described as “coming back” from the west; he became a big snake and was said to be still alive (indeed, immune to death). He was said to be living “in places where there is big water. And when a person comes to get water he may just splash him a lot of water and dig him into the water” [i.e. drown him]. He is the “master of all illnesses” and brings violent winds and rainstorms (Valiente-Noailles 1993:196). He is also associated with the //gamahare, or spirits of the dead. Widespread beliefs in a death-giving or evil being are described also by Schapera (1960 : 160-201) and Barnard (1992:256-60). Amongst the Christianised Nharo, /gauwa is equated with Satan (Guenther 1989). In Angola, the !Xu held similar beliefs about a “god” associated with an underworld: Esterman (1976  and Viegas Guerreiro (1968) both recorded that the belief “that tilling the soil is regarded as contrary to the world order established by //Gawa” (Barnard 1992:56). This set of beliefs about the creator and the death-giver - which varies somewhat in terms of both form and content amongst southern African Khoisan speakers - provides a way of understanding southern San texts and beliefs about death and underwater.
Schapera (1965:194), in his review of Khoisan beliefs in deities and supernatural beings, commented on the similarity of the /Xam rain being, !Khwa, to the //gaua (Schapera’s 1965  orthography) figures of other groups. In /Xam lore, !Khwa is the embodiment of the rain and of the water in the water hole, his home; he is particularly linked to violent and dangerous rain storms. Like the Nharo /gauwa, he is attracted to women (Guenther 1989:117) and is particularly linked to initiation (amongst the /Xam, female initiation). Most of the !Khwa stories pivot on his attraction to female initiates in seclusion. One narrator, Dia!kwain, described drowned female initiates as “the water’s wives” (Bleek and Lloyd 1911:395). !Khwa appears in herbivorous form, including the eland, or, generically as a rain animal or rain bull. (Bleek and Lloyd 1911; DF Bleek 1933a; Hewitt 1986; Solomon 1989, 1992a)
“The girl’s story; the frog’s story” (Bleek and Lloyd 1911:198-205) exemplifies the tenor of the menarcheal stories. A female initiate, unhappy in menarcheal seclusion and with the associated food rationing and taboos, sneaked off to the waterhole, killed a “water’s child” (described as “like a calf”*) and cooked and ate it while her relatives were out foraging. On her next attempt, an angry !Khwa enveloped her in a whirlwind and deposited her in the waterhole, where she was drowned. Frogs and reflections of stars on the surfaces of the water were said to be disobedient initiates abducted and killed by !Khwa. Her family was likewise afflicted - abducted, drowned and turned into frogs. Their possessions revert to an unworked state: mats and arrows become ?grasses and reeds respectively. In a similar narrative, skin karosses revert to being springbok (L V-20:5612-5617)
The formal arrangement of /Xam beliefs about deities and death echoes those of the northern San in the Kalahari. In /Xam texts, /Kaggen appears as the primary deity and creator, while !Khwa (cf. /gauwa and !kaonxa) is the death-giver. Both !Khwa and !kaonxa are associated with water, storms and drowning people (Bleek and Lloyd 1911; DF Bleek 1933a; Valiente-Noailles 1993). Although the mythology of the south eastern San is known exclusively from Qing’s regrettably brief account, references to Cagn, primal time and the creation of the eland show that it clearly belongs to the same suite of beliefs. References to death and underwater in the /Xam texts need to be contextualised in relation to this larger corpus of mythology.
The waterhole in /Xam myths and stories is primarily a place of death and the home of the death-giver, !Khwa (Solomon 1989, 1992a, 1994). It is this symbolism which, I propose, supports Qing’s contention that the painted rhebok-headed figures represented men who had died and now lived in rivers. In terms of /Xam mythology, death and underwater are equivalents. In Lewis-Williams’ argument, “underwater” describes the sensations of trance experience; however, a strong case may be made for the reverse, namely that trancers or curers construe their experience as a journey to the realm of sickness and death, conceptualised as a nether realm accessed through the waterhole, the home of the death-giver, !Khwa. (The curer’s task in the underworld is to wrestle with the spirits who cause illness.) In other words, a trancer experiences and describes what he or she has been socialised to expect, in terms of cosmography and models of mortality and illness; mythology provides the material for understanding the trance experience. Many of the /Xam stories were told to children by their mothers; knowledge and the capacity for imagination of the structure of the universe, the realm of the dead and the origin of illness etc. therefore long precede the experience of trance, which is the domain of adults only.
Further support for this argument comes from /Xam accounts about death, spirits and mythological beings. The only explicit /Xam account of the afterlife was given by //Kabbo (L II-6:670-671 rev.). He stated that all people (including Boers and Koranna) and dead animals go after death to a great hole in the ground, as big as from the Breakwater in Cape Town to the mountains at Mowbray. “When [?they are] dead and buried in the ground, they walk out along this path; they reach the great hole and they live there and “loep [walk] round” in it” (my parentheses). The path from the grave is described as “a Bushman’s path”, but even more significantly as “the First Bushman’s path” (original underlining) - a point to which I will return.
/Xam and other texts appear to distinguish two components of a person after death. In the /Xam texts, the heart (soul?) goes to the sky (L V-19:5506 rev, 5512 rev.; DF Bleek 1935:24) to the creator, where it is seen as a star, while the spirit, or ghost, goes to the great hole. Amongst the /Xam and !Kung, falling stars are associated with a person’s death (DF Bleek 1935:24, 29, 32; Marshall 1962:242), and the /Xam also believed certain stars to be “people of the early race” (e.g. L V-21:5660-5668). Evidence for the /Xam belief in an underworld linked to the spirits is contained in an account of a woman who communicates with the spirits of the dead, entreating them to improve her husband’s luck in hunting (DF Bleek 1935: 35-37). She communicates by banging on the ground with a stone. An almost identical account of banging on the ground with a stone is recorded from the nineteenth century !Kung (Bleek and Lloyd 1911:428).
This complex of beliefs, confused as it may sometimes appear due to temporal, regional and individual variations in belief, suggests that references to death and drowning are not primarily associated with the sensations of trance, but originate in the wider context of mythology and religious beliefs. Lewis-Williams’ (1980) argument that “death” and “underwater” should be interpreted as “trance metaphors” requires further examination in this light.
“Trance metaphors” revisited
That death, underwater and “spoiling” refer to trance was motivated by Lewis-Williams as follows: (1) The !Kung word kxwia means both “spoil” and “enter deep trance” (Biesele pers. comm. to Lewis-Williams 1980); (2) A !Kung curer described his trance experience to Biesele (1975) in terms of being underwater, and learning to cure while underwater; (3) the /Xam used the same metaphor; (4) when Qing referred to the “spoiling” of the eland, “he probably meant that the medicine men in trance exploited the eland’s power as they danced” (1980: 477); and (5) Qing’s reference to people dying from the dance of blood refers to “trance death”, as it still does in the Kalahari. The first four points may be questioned. I argued above that “spoiling” refers directly to Qing’s account of the mythological past and the separation of humans and animals. Two problems may be identified. The use of the same word does not “prove” that we are dealing with a metaphor; a more thorough linguistic analysis is required. More importantly, the fact that the !Kung use “spoil” to mean “trance” is not primarily relevant for understanding Qing’s remarks; the primary context is the Qing/Orpen text, which should take precedence. Instead of Qing’s comments being understood in the context of his wider testimony, they are subordinated to the temporally and spatially distanced Kalahari material. This is a problem within the ethnographic method; whereas Lewis-Williams has clearly demonstrated the utility of Kalahari material for understanding southern San thought and art, this works best at a general level - as I attempt to do in this paper, by invoking well documented, widespread (Khoi)San beliefs. More attention needs to be paid to questions of context, specificity and generalisation when utilising specific ethnographic accounts.
The privileging of trance experience in understanding references to death and underwater also requires further contextual analysis and consideration in terms of the relationship of myth and ritual. Again, Orpen’s paper as the immediate context of Qing’s comments is subordinated in favour of ethnographic support from twentieth century !Kung. Moreover, the prioritisation of ritual and hence shamanism depends on an a priori argument. Despite both Qing’s and Dia!kwain’s references to the mythological past, Lewis-Williams and proponents of the shamanistic model have taken as their starting point the assumption that ritual is primary. This is a view much discredited in anthropological thought since the heyday of the ritualists (e.g. Frazer, Harrison) in the early 20th century. Rather, the relationship of myth and ritual is not uniform or fixed, and needs to be examined in each instance. The most extensive coverage of the relationship of myth and ritual (and hence the art) was offered by Lewis-Williams and Loubser (1986). However, in that account it was assumed - but not argued - that mythology stands in a secondary relationship to ritual and, by extension, the art. The assumption that ritual is primary is rooted in Lewis-Williams (1980) analysis of Qing’s comments; its persistence underpins the shamanistic model.
The claim that “dying” and underwater refer to trance centres on Qing’s account of the “dance of blood”. He said “The men with rhebok’s heads, Haqwe’ and Canate’, live mostly under water; they tame elands and snakes...They are people spoilt by the [moqoma] dance, because their noses bleed. Cagn gave us the song of this dance, and told us to dance it and people would die from it, and he would give them charms to raise them again” (Orpen 1874:10). That “dying” here refers to trance “death” is irrefutable, and an insight of immense importance. However, references to death in other contexts cannot be understood as trance metaphors, but rather as references to beliefs about mortality and the afterlife. As suggested above, “underwater” is better understood in relation to the same general complex of beliefs, rather than specific shamanic experiences. As such, there is no reason to question Qing’s statement that the rhebok-headed men live mostly under water; this is entirely consistent with /Xam and San beliefs about !Khwa or //Gauwa, the underworld and the spirits of the dead.
The third strand of evidence supporting Lewis-Williams’ proposition that death/underwater=trance depends on the claim that the /Xam used the “underwater metaphor” to refer to trance. An account by Dia!kwain, entitled “More about sorcerors” was invoked (L.V. 19:5506-5530; DF Bleek 1935:31-35). It describes how a dead sorceror’s heart falls into “the waterpit which is alive” and sounds like rain: “That is why it sounds like rain, because it enters water which also lives, as does he who is a sorceror” (Bleek 1935:32). The following paragraph then mentions “the water from which sorcerors are wont to fetch water bulls”. This is a reference to the rain-making rituals described by the /Xam, which entailed medicine men of the rain capturing a rain animal by enticing it from its home in the waterhole. It would be slaughtered, and where its blood ran, rain would fall.
Lewis-Williams makes sense of these extracts by suggesting that the references to dying and underwater therein “mean” trance, and that the rain-making ritual is hallucinatory. He proposed that “both the art and the ethnography, then, confirm the view that the rain-making rituals, described by Qing and Dia!kwain, might, at least on some occasions, have been the hallucinations of trance and not an actual ritual performed with a real animal of whatever species” (Lewis-Williams 1980:473). However, if Dia!kwain’s account is read in the context of mythology and in relation to the remainder of his commentary, another interpretation is more plausible. Such a reading strongly suggests that “death” in this context refers to real death, and not primarily the experience of trance. (There is also some evidence that two Khoisan groups indeed sacrificed real animals in rain-making ceremonies (Hoernle 1922; Estermann 1976).
Testimony by Dia!kwain, /Hankasso and //Kabbo in the relevant Bleek and Lloyd notebooks, as well as in “More about sorcerors” refer to dead sorcerors, to the death-giver, !Khwa (“the waterpit which is alive”), to falling stars and to the spirits of the dead. This suite of references relates first and foremost to general beliefs about mortality, rather than the specifics of trance. Although the passage does make reference to curers, a close and contextual reading provides further evidence that it is concepts of “real” death which underpin the account interpreted by Lewis-Williams as evidence for the importance of hallucinatory experience.
The /Xam used the word !gi:xa (sing.) to refer to a “sorceror”. Translation caused difficulties. Lloyd originally translated it as “sorceror”. Dorothea Bleek later used “medicine man”, but after consultation with Maingard reverted to “sorceror” (DF Bleek 1935:1). The root, !gi, merely means “magic” or “magical power” (DF Bleek 1956:382). Analysis of the use of the term reveals a crucial and long standing problem. All references to !gi:xa or !gi:ten (pl.) have been interpreted as references to living /Xam curers or shamans. However it appears that the word applies to two categories (at least) of beings with magical powers (i.e not necessarily curers per se). Apart from the distinction made between rain “sorcerors”, game “sorcerors” and curers (“sorcerors of illness”), a further key separation must be made. While in some contexts, the term certainly applies to living /Xam “sorcerors” of one kind or another, in others it refers to spirits of dead people, who were believed to be the source of illness, although also capable of beneficence. It is not clear whether the spirit of any dead person may act like this, but it is clear that people who had been curers while alive were particularly feared; dead game and rain sorcerors appear to have been considered to be benevolent. The failure to recognise that many of the “sorcerors” referred to are spirits of the dead has many implications for the shamanistic model as it is currently conceptualised.
The key to this re-reading concerns /Xam and San notions of mortality, time and space. As //Kabbo’s comments (above) illustrate, the spirits of dead people go to live in a special place, the great hole in the ground. In that (locative) sense, people do not “die”, as understood within western temporal and conceptual schemata. On the contrary, like other San groups, the /Xam believed that the spirits of the dead were nevertheless still “present” and influential in the affairs of the living. The /Xam accounts make repeated reference to “sorcerors” as dead (i.e. as spirits, not as trancers). For example, /Hankasso said “I have seen these people, sorcerors who are now dead” (DF Bleek 1935:2). Dia!kwain stated that “When a sorceror dies, his heart comes out in the sky and becomes a star. His heart feels that he is no longer alive; therefore his body there, in which he was alive, becomes a star there, because he feels that he used to be a sorceror. Therefore his magic makes a star, in order to let his body in which he lived walk about” (DF Bleek 1935:24). //Kabbo stated that an earthquake occurs when a sorceror has died: “Then a star shoots (falls), for a sorceror who has gone about among things which are bodies has really died. His sorcery is shooting, because his spirits (?) have got bodies. Therefore they work magic” (DF Bleek 1935:26; original parentheses).
Dia!kwain, on “Stars and sorcerors”, said that “when a sorceror dies, his heart falls down from the sky, it goes into a waterpit. Our mothers said, when the star is falling approaching the waterpit into which it means to go, he takes the magic power, he shoots it back to the place where people are. For the people are those whom he wants to take away with his sorcery, for he thought of them while he was among men. Our mothers said, a sorceror does this when he dies, he takes away the people whom he has loved when he feels ill and is going to die, these are the people whom he gets, makes them follow him, that they may go with him...For a sorceror is a being who when he dies, wishes to fall heavily, taking his sorcery; for he wishes his work, which he used to do, to leave the earth on which he used to walk about...For when a sorceror dies, his magic power still goes about” (DF Bleek 1935:28-29). Specifically on the topic of the benevolent spirits of the dead, Dia!kwain said “Mother used to tell me that the spirit people were those who had been game sorcerors. When they died, their thoughts, with which they had been sorcerors and worked magic, continued, though they died and we did not see them, still their magic doings went about here. Their magic doings are like a person who always lives, they do not altogether die; thus he still lives in his sorcery” (DF Bleek 1935:35).
While these accounts affirm the existence of /Xam “shamans” (hereafter “live sorcerors” will be referred to as curers, rainmakers etc. to distinguish them from the spirits of the dead), several issues emerge. Most obviously, accounts of “sorcery” must be understood in relation to the complex of beliefs about illness, mortality and life after death, rather than as in the first instance hallucinatory. As such references to “death” must be contextualised. In some cases (e.g. Qing’s account of the dance) dying may well refer to trance (but not necessarily hallucinatory) experience. In other contexts, curers must be distinguished from spirits of the dead who influence the living. In the shamanistic model, powers and activities which have been attributed to living curers must be understood as the doings of the spirits, rather than as actual events, or as hallucinatory.
Examination of the Bleek/Lloyd Collection provides strong evidence that many of the descriptions which have been interpreted as shamanic metaphors more plausibly refer to dead people. Such an interpretation permits understanding of statements such as that made by /Hankasso on “What sorcerors eat”: “I have heard people say that some sorcerors eat people...others collect human beings for the others when a man has died. They make them eat the man, of his flesh...A man who dies they make people eat” (DF Bleek 1935:8). Are we to believe that /Xam shamans were cannibalistic? On the contrary, this account is intelligible in relation to beliefs about the spirits of the dead in the underworld.
It is clear that the rain-maker /kunn, was alive (DF Bleek 1933b:387), but !nuin-kuiten, also a rain medicine man, was dead. Dia!kwain said “My great-grandfather’s name was !nuin-kuiten; father called on him when he wanted rain to fall, although he was no longer with us, my father used to beg him for rain. For father believed that, being a rain medicine man, he would hear father when he called” (Bleek 1933b:382). The same rain-maker, described as someone who “goes by night”, is the subject of an account (DF Bleek 1936:131-134) which has been interpreted by Lewis-Williams (1985:58-9) as evidence for the activities of a live shaman, allegedly going on out-of-body travel and/or hallucinating. The story describes how “when he was on a magical expedition, he saw a Boer’s ox, while he was a lion, he killed the Boer’s ox”. After being shot, he communicated with Dia!kwain’s father: “Father told me that he...had spoken in this manner to father, when he was truly about to die”. The entire account is consistent with beliefs about the magic wrought by spirits when they travel the earth at night in animal form, as well as with the belief that some real lions were in fact spirits. As animals, the spirits are vulnerable to “real” death; hence the emphasis on “truly about to die”. (Cf. Katz’s account (1982:115) of “healers who used to travel at night”, and who were “real lions, different from normal lions, but no less real”).
Accounts such as this are more plausibly interpreted in terms of beliefs about “real death”, than as trance metaphors. Referring to rain-makers, //Kabbo said that “Dead people who come out of the ground are those of whom my parents used to say that they rode the rain...because they owned it” (DF Bleek 1933a:305). The rain medicine man, /kannu, also seems to have been dead. /Hankasso refers to “the old man, /kannu, and his friends” (DF Bleek 1933b:388-9). The /Xam term employed appears to be a form of the word for “a spirit of the dead” (DF Bleek 1956:350-351). The prefix, /nu, may refer to an elderly person, but its primary meaning relates to death and spirits of the dead; /nu, as an adjective, means “dead” (DF Bleek 1956:709). It is probably significant that the /Xam’s “people of the early race” are referred to by other San-speakers as “stories of the old people” (Biesele 1993:21). The “old woman” (/nu-tara), Tano-!khauken, the springbok sorceress, also seems to be a dead person (DF Bleek 1936:142). Significantly, sorcerors mentioned by name are almost always “old people”, and accounts of the doings of sorcerors are often prefaced by the statement that “Our mothers told us” about them, suggesting that the narrators had not themselves met them during their life times. The specifics of the linguistic evidence are, however, beyond the scope of this paper to address, and require expert attention.
Dia!kwain addresses the relationship of living and dead sorcerors. After describing the dead sorceror’s heart going into the waterpit, he expands on beliefs about illness and death:
“The [dead] sorceror goes to carry off a person whom he wants from here. It is his magic power which goes to carry off this man from here. We wonder why this person seems about to die: for we are those who do not know, we think thus. One sorceror who knows, understands that it is another sorceror whose charm is trying to carry off the person. He says to us, that we seem to think it is illness of which the person lies dying. But it is enchantment that is killing him. Then we see that the man seems dying, they do not think he will die. Then one sorceror will do this to the other sorceror who has bewitched us, he will snore him out of us. He makes the other go from the place out of which he snores him. He kills the other who has bewitched us. He strikes him dead with a stone; as he strikes him, he says “This man has been going about killing people” [followed by details of the battle]...As he is beating him away he says “may that man go to the spirits [/nu:ke] who are always killing people. He has only wanted to come here, in order to kill and carry off people”... [Then the curer] “made the other sorceror come out of the man” (DF Bleek 1935:32-33; my parentheses).
This and other references (e.g. DF Bleek 1935:20) distinguish between “sorcerors” and “other sorcerors”, illustrating the general nature of the term !gi:xa and its contextual reference to either curers or spirits of the dead who have “possessed” someone and caused illness. The failure of researchers to distinguish between them is a question of temporal conflation; but this same temporal conflation is made by the /Xam. Dead people are not only in the past, but also in the present; “the past” appears to be “locativised” (or spatially conceptualised), as the nether realm of the dead. The temporal scheme is not linear in the western sense of past-present-future. This analysis is crucial for understanding the relevance of mythology, and its bearing on the rock art.
Mythology, religion, rock art and histories
San mythology may be seen not just as a set of religious beliefs, but also as an indigenous history, incorporating culturally specific spatial and temporal schemata (cf. Solomon (1995) on phenomenological approaches to time and rock art, and the linear notions of time and history which permeate archaeological thinking). The San temporal scheme seems to centre on notions of the living and the dead; yet this scheme is also locativised, or conceptualised in spatial terms (underwater and the underworld). Mythology (rather than trance) embodies the spatio-temporal “map”. The theme of reversion in stories featuring !Khwa and a female initiate of the early race illustrates this (drowned peoples’ possessions revert to an unworked state, etc.; see above). Ideas about reversion may be related directly to beliefs about the people of the early race, who were considered “primitive”, uncultured, uncivilised and ignorant. Descent into the water hole is conceptualised both as “death” and as a passage into “the past”, the underworld of the spirits of the dead. Both Qing’s and Dia!kwain’s comments on the Lesotho therianthrope paintings suggest that the “First Bushmen” or people of the early race are also inhabitants of this underworld, or underwater; they are San “ancestors”, who are now dead, and as spirits dwell underground (or underwater), surfacing in various forms in order to enact their magic on the living. A key parallel between the spirits of dead people and the ancestral San is their capacity to be both human and animal; on the basis of Dia!kwain’s comments, the spirits of dead beings with magical powers may well be both (or either) the mythical ancestral San or people (especially curers) who have died but continue to influence the living in the present.
In terms of such a reading, the explanations of therianthropes given by Qing and Dia!kwain, and their references to the mythology, are easily comprehensible. The therianthropic figures are not necessarily shamans, as has commonly been argued; they may be either the spirits of dead people who were considered powerful while alive, or the spirits of the ancestral San (in ancestral -i.e. therianthropic - form). The references to death, underwater and the spoiling of the eland are all explicable from this perspective: the therianthropes can indeed be men who have died and now live in rivers, and were spoilt at the same time as the eland, rather than products of hallucinatory experience.
Beliefs about mortality and life after death, and their spatio-temporal forms allow Qing’s comments about the trance dance to be accommodated, in terms of the notion of temporal conflation. Trancers believe they are travelling to the mythological “past” and the realm of the dead; present and past are inseparable. When Qing said that the rhebok-headed men had been “spoilt at the same time as the elands and by the dances...” he may have been referring to the trance-battles waged by curers with the illness-causing spirits (cf. Dia!kwain’s account (above) of the /Xam curer “beating” the “other sorceror”), or to the experience of trance as temporal reversion, and a journey into the past (or both).
Even Qing’s puzzling identification of the Mangolong [Sehonghong] herbivore-like rain animal as a “snake” can be understood, in relation to these beliefs, resolving what Lewis-Williams (1980:470) considered a “serious disagreement” between the comments of the two nineteenth century San commentators. The death-giver appears in various forms: amongst the /Xam, primarily as a rain animal or eland (e.g. Bleek and Lloyd *); amongst the Kua as a snake living in the water. Either the death-giver may assume various forms, including both rain animals and snakes, or else his primary form differed between the Lesotho San and the /Xam. A third possibility relates to interactions between San and neighbours, as Lewis-Williams (1980:470) suggested. In terms of the arguments put forward by Jolly (1994, 1995, 1996), dominant cultures tend to impose their terminology; given that the Kua account refers to the death-giver with the non-San word “xonjapa” it may be that the herbivorous rain animal was the death-giver’s older form, which mutated into a snake in contact situations (although the /Xam described also snakes and other reptiles as “belonging” to !Khwa (DF Bleek 1933a:303); the association of reptiles (less advanced animals) with the rain seems to relate to the temporal themes which link underwater, reversion and primitivity.
The interpretation advanced here, which incorporates mythology and historicises and contextualises San testimonies, satisfies the criteria outlined by Lewis-Williams (1983, 1984, 1985:49-50): verifiability, compatibiity with well-established anthropological theory, internal consistency, compatibility with relevant ethnography, diversity of data explained and heuristic potential. The complex of beliefs and the model I have outlined form the basis of a new approach to understanding San art. Such an approach permits incorporation of a far wider range of San beliefs, ranging from the n!ow complex (which links ideas about birth, death, weather, gender, !Khwa, initiation and illness) to beliefs about stars, to notions of space, time and history.
Some iconographical and analytical implications
According to the shamanistic model, the rock art imagery originates in universally experienced hallucinatory forms, seen and then “construed” in culturally specific ways by trancers or shamans (e.g. Lewis-Williams and Dowson 1988; Lewis-Williams 1995). In terms of the analysis offered here, where mythology and cosmology are accorded more importance, the “trance metaphors” underpinning the notion that art has a significant hallucinatory component cannot simply be accepted. Therefore, a range of images which have been interpreted as hallucinatory or relating to shamanic experiences may be re-examined.
In particular, claims that therianthropes represent shamans are problematic. Proponents of the dominant model argue that they represent shamans “fused” in trance experience with animals of potency, or species which they “possess”, or over which they have magical control. However, a stronger case can be made for therianthropes as in the first instance, spirits of the dead (as both Woodhouse (1974) and Pager (1975) proposed long ago) or as mythical ancestors. In most cases, the notion of “possession” does not refer to live shamans’ control over game, but to the model of illness, and the deleterious effect that the spirits may have on the health of the living. This does not mean that shamans or curers were not painted; Lewis-Williams and Dowson’s interpretations of certain figures bleeding from the nose as curers who sniff or “snore” out illness is unquestionably valid. Nevertheless, the notion that the imagery arises in shamanic visions is problematic (see below).
On the same basis, underwater imagery - including fish, crabs and the like (Lewis-Williams and Dowson 1989; Ouzman 1995) - cannot simplistically be interpreted as deriving from hallucinatory experience; rather it is rooted in a fundamental (Khoi)San spatio-temporal frame, and allied forms of beliefs about mortality and life after death. By the same token, images of people and animals who seem to be dead cannot automatically be interpreted as shamans. While the possibility remains that some images depict the shaman’s trance experiences of transformation, in the light of the above analysis, this claim must be argued in each instance; it seems more likely that images of death and dying refer to mortality rather than trance. The figures which have come to be understood as “trance buck” also require re-examination as perhaps relating to the spirits or ancestors.
Because of frequency of depiction, painted and engraved eland have long received especial attention (e.g. Vinnicombe 1976; Lewis-Williams 1981). Lewis-Williams has argued that “painted eland are symbols of the potency shamans harness to enter trance” (e.g. 1987:171; and cf. 1980,1981). Whereas this may be so in a general or secondary sense, it may equally be argued that it was the eland’s mythological, rather than ritual significance that made it a favoured subject for San artists - as Vinnicombe (1976) suggested. Images of felines also require re-interpretation. Lewis-Williams (1985; 1991) interprets them in trance terms, as living curers going on out-of-body travel or in altered states. Images of lions may well represent dead curers; but again, the key word is “dead” - best understood in a literal rather than metaphoric sense (cf. the account (above) of the “man” in lion’s form who was shot after killing a white farmer’s ox).
The implications of a model which assigns greater importance to mythology and the San spirits and ancestors are multiple; a detailed analysis, in which this model is applied to a series of painted sites and panels is in preparation but beyond the scope of the present paper, which is concerned first and foremost with the ethnographic basis of San rock art interpretation and the utility of the hallucinatory emphasis in the shamanistic model.
Theoretical and methodological implications
The world-wide impact of structuralism in rock art research since the nineteen sixties has been considerable. This was also true of South African rock art research: two of the classics works on the art (viz. Vinnicombe 1976, Lewis-Williams 1981) drew heavily on structuralist theory and method. Lewis-Williams (1980, 1981) in particular introduced important theoretical innovations, including, inter alia, a shift from linguistic structuralism to semiotic analysis (Lewis-Williams 1981), the introduction of a rudimentary hermeneutic approach (Lewis-Williams 1981), structural-marxist interpretation (Lewis-Williams 1982) and a trenchant and enduring critique of empiricism (Lewis-Williams 1983, 1984). Despite the ongoing theoretical innovations introduced via the shamanistic model, aspects of it are deeply rooted in structuralist notions that are now considered problematic - including issues of historicity and synchronic analysis, and, especially, the notion of universal “deep structures”, or structures of which culture is the surface expression (Levi-Strauss 1963). This notion is variously articulated in the shamanistic model, most obviously in the idea that the production of visual imagery is, at base, a product of neurological structures common to all anatomically modern humans. This echoes Levi-Strauss emphasis on universal categories, while those aspects of the model which emphasise “neuropsychology”, the culturally specific construal of hallucinatory forms and the role of the individual invite comparison with the shift in Levi-Strauss’ later work on myth and symbolism towards examining “the free operation of the mind” (Seymour-Smith:1987:270).
It seems to me that there is a connection between notions of “deep structure” and various arguments proposed by the shamanists. The notion of universally experienced, hallucinatory “form constants’ and the like are essentially structuralist in affiliation, and have implications for the modes of argumentation employed. The notion of universals permits the generalisation of the analysis of one or a few images or panels to the rock art of southern Africa in general (and indeed the art of other continents). The notion of deep structure also relates to the central role of metaphor, where ethnographic statements (such as Qing’s) are said to represent something else, a “deeper” meaning which must be expertly excavated. I have argued above that both strategies may be questionable. Qing’s alleged “metaphors” are explicable at a “surface level” if the context of analysis is extended to incorporate mythology. And, while the death=trance metaphor seems valid in certain narrow contexts, generalising such interpretations is methodologically problematic. Hence, interpretations of shamans (or dead figures or whatever) must be argued for each case in context, since it cannot be assumed that all therianthropes (or dead figures) are trancers, regardless of their location in specific sites, panels or juxtapositions.
The art historical notion of site specificity has been invoked by Skotnes (1994) in a somewhat different vein. Skotnes argues that, since the rock face or ground is not neutral (cf. Schapiro 1969, Solomon 1989, Lewis-Williams and Dowson 1990), imagery and compositions need to be considered in relation to the specific site in which they occur. In a related argument, I propose that the iconography of San art requires site specific attention, rather than interpretation by generalisation. This relates to problems with structuralist notions of rule, rather than practice. While it seems clear that aspects of San thought and art are widespread, conventionalised and rule- (or principle-) governed, emphases on general underlying structural principles, and generalised modes of interpretation preclude understanding of diversity in the art, on scales ranging from site to continent. The centrality accorded to trance and shamanic visions tends to impose blanket explanations, and prioritise the role of the shaman-artist. Rather, the contexts of image-making may be disaggregated. If the shaman and his/her visions are not seen as the origin of artistic production, it may be posited that not only shamans painted, and that not all sites are directly related to curing rituals. By scaling down the primacy of the shaman in visualisation and image production, the likelihood of the diversity of the art arising in multiple contexts becomes a topic for investigation.
In anthropological, archaeological or sociological terms, the notion of site specificity may be used in other ways. For example, on the basis of the view that the shamanistic model subordinates gender, as an identity factor which precedes trance experience, I have considered sites with predominantly female or feminine figures, in relation to the corpus of narratives and lore concerning female initiation (e.g. Solomon 1992b; 1995). Some sites or panels in the south western Cape and elsewhere may be related to female initiation practices (rather than curing rituals). Although such an approach is complicated by the massive difficulties of dating, and of distinguishing painting episodes, I have suggested that considering single sites or panels in relation to the various foci of San thought and social life is preferable to recourse to inflexible, over-arching structuralist generalisations and quasi-universals. As such, considerations of sites as wholes is an improvement on the method of selecting “relevant” imagery from complex and sometimes heavily palimpsested art sites. I have implied above that the ethnographic method has given insufficient consideration to the contexts of oral testimony; the same applies to the images on the rocks. More attention needs to be paid to site particulars and to scales of analysis; this is crucial to understanding the regional, temporal and historical diversity in the art that the shamanistic model cannot properly accommodate.
Problems of ahistoricity are, of course, not solely determined by synchronic structural analysis, but are also a function of the problems of dating the art. Nevertheless, an approach which prioritises mythology makes some contribution to the issue. Because the shamanistic model takes little account of San mythology and art as indigenous historical and temporal “texts”, differences among northern and southern San beliefs are subordinated in favour of similarities, regardless of scales of analysis. Lewis-Williams and Dowson (1994:207) acknowledge problems with the concept of a “pan-San cosmology”. The problem, however, does not lie in the demonstrable reality of strikingly similar belief sets, but the way in which these are mobilised to understand the art. As I suggested above, it is slippage between the general and the specific that is problematic in the ethnographic method in rock art research, rather than the ethnographic method itself. “Ethnographic methods”, as they have developed since the nineteenth century, are of ongoing value in rock art research. This extends from Stow’s interpretations of the rock art in relation to myth and ritual (e.g. Stow 1905:29, 120-121) to Vinnicombe and Lewis-Williams’ erudite analyses linking Drakensberg and Kalahari materials. Nevertheless, the tendency of ethnography-based methods to homogenise “the San” and to subordinate time, process and historically situated practice requires ongoing evaluation in the light of changing knowledge in rock art research, archaeology and related disciplines.
The model I have outlined both challenges and affirms aspects of the shamanistic model. As such, the extent to which it is seen to have the capacity to affect perceptions of the art is likely to be controversial. A model which proceeds from questions of myth, history and specificity affirms the thesis that the rock art is best viewed in relation to San religious beliefs, with the proviso that this cannot simply be generalised over time and space (nor linked generally to shamans and hallucinatory experience).
Several of the criticisms I have made are equally applicable to my own previous research, and have arisen in the context of self-critique. For the purposes of the present argument, I have paid insufficient attention to the question of recursiveness, or feedback relations between myth and ritual. Yet, while this reflects the process and context of my own research, it is also based on the proposition that, even if a recursive relationship exists (as it surely does) myth, history and socialisation must ultimately be assigned causal or determining primacy over trance and altered states, rather than vice versa. The main points of departure from the shamanistic model in this analysis concern the supposed centrality of hallucinatory experience in visualisation and the production of rock art, and aspects of method and theory. In this regard it may be profitable to return temporarily to Vinnicombe’s (1976:352) position, where she ultimately declined to generalise the importance of one context of production over another, on the grounds that in San thought all domains are inter-connected. Finally, we may see the art as intrinsically historical; as a varied record of the San perceptions of their own history, whether the subject is mythological, shamanic or depicting contact with colonists and indigenous farmers.