The Antiquity of Man

   Home   |   About   |   What is Evolution   |   World Archaeology & Palaeoanthropology   |   North African Archaeology   |   Pseudoscience   |   Recommended Readings   |   Reviews   |   Links   |   Search   |   Contact

Anne Solomon. 1999. Meanings, models and minds: a reply to Lewis-Williams. South African Archaeological Bulletin

'Meaning’ in San rock art is contested terrain, with different researchers bringing different models of meaning and mind to bear on the issue. Models of meaning, and the assumptions they incorporate about rock art as an avenue to understanding prehistoric ‘mind’, are discussed. On the one hand, what is held to be ‘meaning’ is a product of particular theoretical premises, methods and epistemological foundations, but, on the other hand, the ‘meaning(s) of the art’ are not merely a question of perspective, with free play accorded to the researcher’s imagination. This distinction is illustrated via discussion of Qing’s commentary to Orpen, and interpretations thereof.


Rock art researchers agree that the (archaeological) meaning of San art is that which it had for people in the past, but there is less agreement on what meaning ‘is’, how it is constituted, and the methods used to ‘access’ that meaning. Indeed, the complex problem of meaning and its operations (semiosis) has preoccupied scores of eminent scholars this century. Lewis-Williams (e.g. 1981; 1998) has offered a detailed exposition of his theoretical premises and the model of meaning that he favours. Other researchers, including Skotnes (1994) and myself (e.g.1989, 1992,1995, 1997a), have proceeded from different premises, which do not correspond with Lewis-Williams’ structuralist-semiotic model. Different interpretations and accounts of meaning put forward in San rock art research have been addressed by Lewis-Williams (1998), crystallising some of the problems and contested positions that characterise contemporary research. However, ‘meaning’, in Lewis-Williams’ account, is variously conflated with ‘interpretation’, ‘context’, ‘intention’ and ‘truth’, and depends on a particular model of ‘mind’. From a different theoretical standpoint, this account of meaning is deeply problematic. Theoretical, methodological, interpretive and epistemological issues related to the issue of meaning in San art are discussed here, as well as interpretive problems associated with nineteenth century San testimonies on rock paintings.

What is meaning?

Meaning as context

Lewis-Williams (1998) has evaluated the work of researchers who have considered gender (e.g. Solomon 1989, 1992, 1994, 1995; Parkington 1989, 1996, Parkington and Manhire 1997) and visual/formal considerations (Skotnes 1990, 1994), concluding that these are merely secondary meanings, “penumbral, yet vibrant” (Lewis-Williams 1998:96), while shamanism is the primary and central meaning of the art. But are shamanism, gender or form “meanings” that can be ranked in order of importance in this way? The notion that this is possible depends on a conflation of ‘meaning’ with ‘context’, and a particular view of what ‘context’ comprises. For Lewis-Williams, the context of production is a “shamanistic cosmology”. It is claimed that “all images are shamanistic in that they are part of a shamanistic cosmology and are situated on a surface that had meaning within that cosmology” (Lewis-Williams 1998:89). In other words, context (shamanism) is equivalent to meaning (shamanistic); meaning and context are interchangeable. However, this strategy isolates one aspect of the “original context of production”, namely ‘belief context’, which is spuriously split off from other realities. The prioritisation of belief over social relations, economy and other material aspects of life cannot easily be sustained. In San ethnographies, the gendering of rain, sun and moon (for example) must be understood in terms of social relations, and not cosmology alone, and social and ‘belief’ contexts cannot be divorced as Lewis-Williams advocates. Both are aspects of the original context of production; the privileging of one over another is an analytical choice, rather than an imperative, and furthermore, a choice that is unmotivated in the shamanistic model.

Separating contexts in this way allows Lewis-Williams to ‘assess’ the relative importance of each for understanding the art, and to proclaim shamanism as the primary meaning. This is inevitable, since ‘belief’ is set up a priori as primary, more determining and more ‘meaningful’ in his model of ranked meanings. It ignores the fact that neither Parkington, Skotnes nor I have proclaimed our analytical foci as “the meaning of the art”. Rather, these qualitatively different dimensions are each relations of meaning, intertwined and simultaneously present in the original production context. Neither belief, nor gender, nor form/artistic praxis constitute pure ‘contexts’ in themselves. It is rather their inter-relation which is of analytical interest. The privileging of belief context derives from the idealism of Lewis-Williams’ analysis (see below).

To say that meaning is always contextual is a different proposition from saying that context equals meaning, and the notion of “contextual meaning” - irrespective of which context is prioritised – is problematic. Hodder’s “archaeology of contextual meanings” (1987) has been discussed by Davis (1996), who notes the ultimate redundancy of the notion: ‘Contextual meaning’ “is not a special variety of definition, the meanings of a term exist in a social and historical context, namely the context of just those people who assign sense (Sinn) to a reference (Bedeutung)...In the same way as ‘contextual adaptation’ would just be adaptation as such, contextual meaning is just meaning as such”. The limitations run deeper: “to recognise ‘context’ is not in itself to recognise the variability and open-endedness of meaning...It remains to be shown for any given case that the context of meaning - of people assigning sense to a reference - is characterised by variability rather than invariance, or openness rather than closure” (Davis 1996:118-9).

The shamanists’ argument for the relative invariance of meaning relies on prioritisation of belief-context-as-meaning, and appeal to the conservatism of San religious belief. Indeed, it is difficult to see the remarkable similarities in the forms of San beliefs in different times and places in other terms, and Lewis-Williams’ emphasis on some kind of pan-San cosmology (e.g 1984b) has been valuable. Nevertheless, problems arise. Forms may remain relatively constant, while ‘meaning’ (‘content’) changes. Similarly, the same form may connote differently to different people at one time (see further below). The notion of a primary belief context depends on an abstraction, a statement about putative similarities in San thought, irrespective of time and space differences, and separated from the activities of people in any specific “social and historical context”. This reified ‘belief context’, which, since it is already divorced from socio-economic processes, is timeless and homogeneous, becomes the rationale for a view of meaning as principally static and largely closed, and hence for shamanism as the ‘essential’ meaning of the art in all times and places.

Within this notion of an essential meaning, Lewis-Williams (1998:88-9) acknowledges only limited polysemy, which he discusses in relation to the eland “symbol”. However, since ‘the meaning of the art’ has already been defined as ‘shamanistic’, his analysis is really concerned with shades of (shamanistic) meaning, rather than multiple meanings. This view of the consensual nature of symbolic meanings is part of the standard semiotic account of the conventional meaning of the symbol, and, by extension, depends on the notion of a homogeneous, consensual and conflict-free social community in which meanings are unproblematically agreed upon. “Contextual meaning” is again invoked. Lewis-Williams (1998:88) argues that “Key symbols are central in that they lie at the heart of a belief system and as such have a semantic spectrum rather than a meaning”. Here Lewis-Williams departs from his earlier analysis (1981), which he now sees as flawed. It is no longer the iconographical associations of a figure (e.g. an eland) that he sees as significant. Instead he distinguishes between “the eland antelope” (the ‘real thing’), “a rock painting of (what appears to be) an eland antelope” as “a concrete item of material culture” and “the eland symbol”. The last-mentioned is described as “an abstract concept that (if they think about it at all) exists in people’s minds” (1998:89). A painted eland is said to be “one contextualised manifestation of the eland symbol, and as such will have a restricted rather than a diffuse semantic focus” (ibid.). The symbol is thus the sum of its potential ‘contextual meanings’. This model of meaning is strongly contested, principally on the basis of its structuralist assumptions.

The structuralist account of meaning, and idealist foundations

The success of structuralist analyses of rock art in the 1960s was carried over into studies of San art by Vinnicombe (e.g. 1976) and Lewis-Williams (e.g. 1972, 1974) in the 1970s. Lewis-Williams moved from an initial focus (1972, 1974) on classical structural analysis (based on structural linguistics, where signs are studied as if structured like a language) to semiotic analysis (1981), (a development of structural analysis which also includes non-verbal signs, also (1980) invoking structural-marxist analysis, in an effort to incorporate the social dimensions of art-making. Despite these shifts, and claims of eschewal, Lewis-Williams’ analysis remains irrevocably structuralist, with its attendant problems. The boundaries of semiotic studies, which have been enormously productive, are, according to Hawkes (1977:124), “coterminous with structuralism”, and part of a larger field which might be called “communication”. Lewis-Williams uses a simplified version of Pierce-Morris semiotics, focussing on the broad sign categories icon, index and symbol. Since Pierce’s work on ‘the science of signs’ early this century, many writers, including, for example, Eco (e.g. 1976), have addressed themselves to the manifold problems that have subsequently emerged. Much recent writing on semiotic issues has specifically addressed problems of structuralist tenets in semiotic analysis.

Lewis-Williams describes structuralism as “the proposition that human thought proceeds in terms of binary oppositions”. This minimal definition inevitably oversimplifies matters. Whilst Levi-Strauss did maintain that humans characteristically order and classify experience in terms of binary oppositions, this does not mean that every time one encounters contrasting dualistic terms that one is dealing with a structuralist analysis! The founding metaphor of structuralism is rather the depth : surface dualism, where depth refers to “mental structures” or a “cognitive template” (cf. Ingold 1993, Thomas 1996), and “surface” to cultural ‘expressions’ as the realisation of those ‘deep structures’.

This appears unmistakably in the shamanistic model, and explicitly in Lewis-Williams’ analysis of the eland symbol. It is said to exist “in people’s minds” (depth) before being expressed as a “contextualised manifestation” in the concrete, material form of the painted image (surface). A similar notion of depth and surface appears in the idea that the neurophysiological system of all anatomically modern humans (depth) generated forms which were construed and expressed (surface) by San artists in the medium of paint (e.g. Lewis-Williams and Dowson 1988). The ‘eland concept’ is, in effect, a structure of, and in, the mind. This begs the question of where and how? Lewis-Williams suggests that it was not necessary for people to have “thought about it at all”. How then does the eland concept exist in the mind unless it is thought about? How can a concept exist independently of thought? How did it come into being in the first place? How does it change? Structuralist analyses are notoriously inadequate in this regard, being synchronic and unable to accommodate history and change. This may be understood in relation to the idealism of the structuralist project in general, and the shamanistic model in particular.

In the shamanistic model, with its structuralist foundations, ideas - belief, “cognition” and what is optimistically called “neuropsychology” - are accorded considerable importance, even determining force. Material factors have been considered only as a secondary relation of meaning (for example in the idea that shamanism had material consequences [*]). The decision that (a shamanistic) ‘belief context’ is more fundamental and determining than social context exemplifies this prioritisation of ideas. Lewis-Williams sees socio-economic contexts as something to be “factored in” post hoc to the shamanistic model (1998:90). The idealism of the model is not ameliorated by such “add-gender-and-stir” strategies (cf. Moore 1988). However, the problem is not resolved simply by replacing an idealist model with a materialist one, although this is technically possible: for example (to replicate, but invert, Lewis-Williams’ logic), it is as legitimate to propose that shamanistic beliefs were formed within an already gendered hunter-gatherer economy, with the social context more determining; in which case, gendered social relations can be seen as the context and meaning of the art. This seems to be the way in which Lewis-Williams has understood my arguments. But gender is no more ‘the meaning of the art’ than shamanism is, and the inverted scenario is equally problematic. Rather, in contemporary research in many disciplines, efforts are made to transcend such bald separations of the material and ideal domains.

After structuralism and semiotics

Theoretical perspectives and analytical strategies of (at least) the last two decades have been formulated as critiques of and responses to both structuralism and “vulgar marxism”. These various responses are commonly lumped together as “post-modern” by those unfamiliar with the terrain and the variety of positions such critiques espouse. The following discussion elucidates some of these developments and the problems they have sought to address, with reference to the ways in which they have both permeated my research and contributed to my critique of the shamanistic model.

My consideration of the gender(ed) dimension of San art depends on a model of meaning that diverges from Lewis-Williams’. The focus on gender demanded that the material conditions of the production of art receive more attention, in such a way as to avoid the simple polarisation of ideal and material (above) and the anti-historicity of the structuralist project. Practice theory (Bourdieu 1977; Moore 1986), structuration theory (Giddens 1979) and feminist post-structuralist theory (e.g. the materialist and post-structuralist perspectives of Delphy (1984) and Cixous (1985 [1975]) respectively) were amongst the theoretical perspectives explored as an advance on the structuralist-semiotic model (e.g. Solomon 1989, 1992). Lewis-Williams’ claim that gender studies are irredeemably structuralist (because gender focusses on “the male : female opposition” [Lewis-Williams 1998:91-2]) misses the point, which was that gender is a mutable social construct, not a product of deep structures.

Following Delphy (1984), a distinction was drawn between male/female and masculine/ feminine, with the former dualism referring to biological givens, but the latter as relatively autonomous social constructs, not derived from or caused by biology (if that were the case, the ‘content’ of gender would be invariant in time and space; this is too obviously not the case to warrant further discussion). As such, my model rejected the idea that gendered binary oppositions are products of deep structure (cf. Cixous 1985 [1975]). Instead, following Moore’s influential study (1986), I considered gendered oppositions in San texts in terms of gender stereotyping, where such stereotypes operate as organising principles (not structures) in processes of meaning formation that occur only in social practice, and commonly in the contestation of power. The emphasis on practice offered a way of analysing ideas and material conditions in conjunction, rather than seeing one as simply determined by the other. It also addressed structuralism’s structure, and the idea that practices are enactments of underlying rules or structures. Rather, in Giddens’ terms, structure is only instantiated in practice, and is not otherwise ‘in’ the mind (Giddens 1979). The approach also addressed the untenable notion that gender relations are necessarily consensual, co-operative and complementary, as Lewis-Williams assumed in an early paper (1982).

In retrospect, these post-structuralist revisions seem to analyse the issue of multiple meanings in a limited way, viz. the “slippage” (Davis 1996;125n) between reference (what an image denotes) and sense (its potential connotations, “symbolic meanings”), and I have subsequently explored other theoretical options (see below). Nevertheless, the aforegoing still seems to me to be an improvement on Lewis-Williams’ structuralist model in various ways. In particular, by integrating social and economic factors, and via the notion of contested meaning(s), the possibility of recognising the potential open-endedness of meaning, and hence historical change, was introduced. This model of meaning entails a different epistemological position, but this is not relativistic, although it does stress relationality, and the contingency of ‘meaning’. This stress on semantic contingency contrasts with Lewis-Williams’ position, in which one of the meanings of ‘meaning’ is ‘truth’.

Meaning as truth

The argument that the context, and hence the primary meaning, of San art is shamanistic functions as a claim to truth; shamanism is proclaimed as all-encompassing, and no other ‘meaning’ is permitted. As such, the model is essentially monosemic, although Lewis-Williams allows for constrained polysemy of a lesser order at the level of the symbol. Lewis-Williams (1998:88-9) further distinguishes between polysemy and multivocality, with the latter idiosyncratically described as the potential use of different “categories of image” by different social groups. However, in an account of meaning which emphasises process rather than structure, polysemy may in fact be a product of multivocality and the negotiation of meaning(s). (This need not - and did not, in my research - concern different groups using different images.) The epistemological dimension of Lewis-Williams’ model of meaning is illustrated by further reference to the eland symbol.

The question was posed above: How did Lewis-Williams’ “abstract eland symbol” come into being? According to my premises, the meanings of the eland arise in practice and practical negotiation - including contestation - of meaning. Moore (1986) gives an example of the meaning of ash in Marakwet society, where the perceived meaning of ash depends on the practical context and the gender of the person ‘enacting’ that meaning. By smearing herself with ash outside the conventional context of its use, a woman challenged the conventional meaning of ‘ash’ (associated with male dominance) and created a new, and contrary, meaning. Madonna’s use of the Christian cross in her videos is another example of creative subversion of conventional symbolic meanings. As such, the notion of a single meaning shared by all members of a group is problematic. Multivocality, in my analysis, referred to the same term (image or ‘symbol’) potentially having qualitatively different meanings for different social groups (particularly men and women). This derives from a view of meaning as fundamentally unstable, as continually being remade in practice, and as such, subject to change. This disrupts the stasis of the notion of meaning as structure, replacing it with a view of meaning as an ongoing process of negotiation of semantic possibilities. As such, it offers both an account of how meanings come into being, as well as how they might change, which shamanistic explanations do not and cannot. The shamanistic model’s ‘truth’ depends on an assumption of unitary, fixed meaning. Lewis-Williams confuses my analysis of meaning as potentially unstable (and hence subject to change) with a relativistic epistemology. This is possible because the model conflates meaning with ‘truth’.

In this regard, Lewis-Williams (1998:89) misconstrues my argument (Solomon 1989:161), viz. that one particular interpretation is not objectively (original italicisation) better than another. One interpretation may eclipse another on other grounds, but, short of time travel, the correspondence of one or another to the Truth in the past cannot be objectively or empirically proven (although it might be the case that one interpretation is a closer approximation of the truth than the other). Lewis-Williams’ criterion (e.g. 1985) that an adequate hypothesis must be characterised by “verifiability” is misplaced in this regard. Assessment must take place according to other criteria. Since interpretations are shaped by their theoretical tenets, it is necessary to debate the premises of conflicting interpretations, not only the interpretations themselves, since results are shaped by premises. Lewis-Williams’ structuralist premises cannot be empirically established as ‘truer’ than my alternative premises; the aptness of premises can only be motivated and argued, not proven. In these terms, the quality of argumentation assumes crucial importance when it comes to adjudicating between interpretations.

For example, (cf. Lewis-Williams 1983, 1984a, 1985) the argument must be internally consistent (including ‘consistent with its premises’). In this sense, if my argument is as internally consistent as Lewis-Williams’, then both may be equally rigorous. Nevertheless, rigour does not equal ‘truth’, although it does allow discrimination amongst competing interpretations. Lewis-Williams’ other oft-cited criteria include “quantity of data explained” and “heuristic potential”. However, the alleged “explanatory power” of the model does not validate the model or its premises retrospectively; such claims depend on a logically unsound, teleological or consequentialist argument (Sparkes 1991).

To acknowledge the relationality of an interpretation to its premises and the impossibility of empirical verification of the approximation-to-truth of one or another model does not imply a relativistic epistemology, nor that all interpretations are equal, nor does it licence the free play of a researcher’s imagination (nor, for that matter, does it equate with Stalinism or “a lack of respect for data” [Lewis-Williams 1998:90]). My epistemology and model of meaning do not coincide with Lewis-Williams’; I contend, however, not that they are equal, but that the structuralist-semiotic model, and the particulars of the resulting interpretation, are highly problematic. To illustrate this further, an iconographical/ interpretive problem - the question of what therianthropic figures represent - is discussed in some detail in the second part of this paper.

Meaning as intention; or Humpty Dumpty revisited

For Lewis Carroll’s Humpty-Dumpty, words mean what the speaker means them to mean. Lewis-Williams’ view of meaning and intention is not dissimilar, insofar as it contains the idea that an ‘expression’ (e.g. a painted image) was a faithful copy of what was in the artist’s mind, and what s/he wished to convey. Davis (1996:95-127) has astutely discussed meaning and its relation to intentions. He relates theories of “copying” to “an entrenched view of human consciousness and the way it is directed at the world...According to the entrenched view, a metaphysics of consciousness propounded by many Western philosophies and harking back to Platonic themes, thoughts precede their particular material representation, expression or embodiment. The natural embodiment is merely a stuff or medium of some kind supplementary and relaying thought itself” (Davis 1996:98). In other words, images are un-mediated reflections of thought. Rock paintings, for Lewis-Williams, are precisely thoughts converted directly into “concrete items of material culture”, remaining as solid evidence of past intentions, thoughts and minds.

However, this depends on an ahistorical notion of ‘mind’, and received assumptions about consciousness which derive from the Enlightenment view of the rational subject, “characterised by self-possession, immediacy, control and self-awareness” (Davis 1996:101). Davis, who has favoured deconstructionist and historical materialist critiques, discusses the dichotomies which arise from this view of the mind, where the first term describes the pure, original, rational intention, and the second its ‘realisation”: “Idea/Expression, Intention/Artifact, Form/Matter, Concept/Statement, Rule/Practice...” (1996:103). To escape this structuralist depth/surface dichotomy, Davis proposes an approach similar to (but more radical than) the model of meaning I previously employed, namely that it does not exist as a thing in the mind, but is instantiated only in practice: “In the deconstructive revision of the entrenched account, value systems, strategies of life or ideology are “written”. They are constituted only in the activity of thinking them out, or more accurately, thinking them down in some medium or other...Thus the logoi are not “in” the heads of the makers, and then, like a Word made Flesh, somehow emptied into matter. Rather, they are made, written, by the makers neither before nor after, but only always in the ongoing, temporally and spatially ramified structure of making” (Davis 1996:98-9). Skotnes’ emphasis on artistic praxis and the “visual as a site of meaning” (1994) variously echoes this.

The importance of Davis’ analysis of consciousness and intentionality is that, whilst acknowledging that people “intend” to do things, it does not assume that the product is an accurate reflection of that intention. There’s many a slip ‘twixt cup and lip, and intention, product and meaning are not identical. It is in this sense that meaning concerns disjunction. For Davis, disjunction is not merely the slippage between sense and reference, but is, at least, also the gap between ‘pure’ intention and mediated product (Davis 1996:125, 78n). The contrast between this and Lewis-Williams’ notions of the painted eland as a concrete manifestation of the abstract concept and of the rock art as an unmediated copy of perceptions, intentions or concepts (Davis 1996:127) is obvious. In short, rock paintings are not fossilised ideas. Lewis-Williams’ model of meaning assumes that the image or painting = thought, and that thought = meaning: that is, the meaning that the artist had “in mind”. This position cannot be sustained, and from this perspective, models of meaning in San rock art research are in urgent need of further attention.

Theory, method and “mythic women”

The “ethnographic method”, pioneered by Stow (1905) and considerably extended by Vinnicombe (e.g. 1976) and Lewis-Williams (e.g. 1981), has been of enormous importance in rock art research, but it is heavily logocentric. As Preziosi (1989:37) has observed, interpretation as “contextual reintegration” often boils down to no more than “mechanical practices of text-matching”. It focusses on the referent of an image (rather than the image itself), translating the visual into verbal terms, with the verbal then masquerading as “meaning”. Partly in response to critiques of iconography-centred approaches (cf. Skotnes 1994), I considered the “mythic women” figures, which are - unusually - depicted in frontal view, in order to incorporate questions about their form rather, than their referents as such (Solomon 1995, 1998). I drew also on Ingold’s (1993) work on temporality, as an advance on structuralism and post-structuralism. Ingold’s study, which utilises phenomenological theory, offers the potential to move beyond the analysis of static form, via his argument that form is generated in movement, and that western thought has tended to prioritise form over process. The focus on temporality seemed important in relation to Schapiro’s view of visual arts as portraying “an order of time in an order of space” (1985:215). Phenomenological accounts of consciousness (which share aspects of the deconstructionist approach, but are elsewhere in tension) were used as a way of relating the art to ongoing “lived experience”, rather than mental structures, and to (gendered) San lifeways and material conditions, rather than a split-off universe of ‘beliefs’.

In his critique, Lewis-Williams fails to engage with the themes of my research, and the problems outlined and addressed therein, implying that such studies do not offer an account of how “the making of gender statements articulated with the rock art images” (Lewis-Williams 1998:91). That research was not, however, concerned with “the making of gender statements”, neither did I suggest that “San people painted in order to negotiate gender” (Lewis-Williams 1998:91). Rather, I analysed San art, its themes and forms, in relation to the rhythms and processes of specifically hunter-gatherer societies, in which gender is unquestionably a pre-eminent socio-cultural division. These rhythms and processes, I argued, are the context within which “shamanism” operated and developed its form(s) (Solomon 1995). Contra Lewis-Williams’claim, the entire dissertation was concerned with the ways in which gender and rock art were “articulated”. Exploration of similar strategies for addressing Lewis-Williams’ timeless shamanistic cosmology and relentless structuralism continue in my current research, with the important difference that I now believe that the art cannot correctly be described as “shamanistic” (cf. Katz 19*).

The approach adopted was designed to address several problems, not least of which was the notion that the forms of San art were derived from hallucinatory experience. Lewis-Williams (1998:91) mistakes my linkage of a range of feminine (not female) figures to a wider complex of gendered beliefs as “intentional ambiguity”. Ambiguous, certainly; but not “intentional”, since “intentionality’ was a major part of the problem at hand (see above). The issue of frequencies is another component of the analysis that Lewis-Williams misconstrues, arguing that, because of the small percentage of “mythic women”, my study cannot constitute more than “a limited addition to our understanding”(Lewis-Williams 1998:91). This is ironic in terms of his dismissal (1998:95) of “naive percentages”, and ignores the fact that I discussed frequencies, relative to the question of whether a rare depiction is less “meaningful” than a virtually ubiquitous motif such as the eland. In short, Lewis-Williams’ critique proceeds from a position which already denies my different premises, and hence my different findings. The failure to understand and engage with those premises means that his criticisms remain insubstantial. The same applies to his critique of art historical approaches.

Art history and “westernism”

Art historical approaches are not necessarily antithetical to those of archaeology or anthropology. Theorisation in art history - of agency/practice, consciousness, iconography, “meaning” and so on – constiitutes a parallel critique of the limitations of structuralism. In the art/historical work of Skotnes, Davis and Preziosi (cited above) the model of meaning that grounds the shamanistic model is explicitly rejected. These studies cannot so simply be dismissed as inappropriate “Western formalist art history” (Lewis-Williams 1998:94). For one thing, to consider the form (rather than the referent) of an image, panel or site is not the same as conducting a “formalist” analysis. For another, the way in which Lewis-Williams mobilises the notion of ‘the West(ern) is uneven at best.

Lewis-Williams criticises “western” art history’s (allegedly) universalising tendency (this charge made with reference to Skotnes’ work on form and artistic practice). Yet a far deeper universalism is at work in the claim that all anatomically modern humans ‘see’ the same hallucinatory forms in ASCs - a biologistic statement about forms and universals, which permits the shamanists to extend their analogical arguments virtually all people (irrespective of time, place or identity) and their arts. A similar universalism is evident in Eliade’s (1964) notion of shamanism as “primitive religion”, although Eliade is more circumspect in what he includes in his definition of shamanism (Solomon in prep.).

Elsewhere, Lewis-Williams charges critics with western, “etic” perspectives, thus invoking the very difference of the San. This appears in his response to Skotnes’ (1994) analysis of a western Cape site, in terms of the artist’s engagement with space and ground;my discussion of lateral versus frontal view in paintings of human figures at a KwaZulu-Natal site (1995, 1997b); and Parkington and Manhire’s (1997) study of “directedness”. It is claimed that the way that the San ‘viewed’ paintings was “almost certainly...very different from ours” (Lewis-Williams 1998:95). This is yet another conflation, this time of “looking at” with “meaning”. Of course San people would have “looked at” the art with different eyes (in the sense that the art was an integral part of their cosmology and environment), in a way which researchers cannot. For Lewis-Williams, “to view” is virtually synonymous with “to perceive its meaning”. Nether Skotnes or I were concerned with this; nor did either of us deny that the significance of the art for prehistoric communities was, inevitably, “different”. But, in the physical sense of “look at”, there can be no difference between the San and the contemporary viewer - unless one is to resort to the highly problematic notion that San vision was fundamentally different to our own. On such claim about the Other apartheid was built. Moreover, since Lewis-Williams admits that “we do not know how the San viewed paintings” (1998:95), and he does not give even hypothetical examples of different ways of viewing, the claim that Skotnes, Parkington and my research universalises viewing practices remains inconsequential.

In contrast, my suggestion that the frontal view in which the “mythic women” were depicted was a bridge between original and contemporary viewers was theoretically grounded. A way of linking past and present via the ‘experience’ of form is offered by Ingold (1993). Although the argument needs to be read in its entirety for its complexity to be appreciated, Ingold’s epistemology is based on the idea that “the native dweller and the archaeologist... - insofar as they both seek the past in the landscape - ...are engaged in projects of fundamentally the same kind”, though their resulting narratives may diverge. Ingold’s epistemology bridges past and present via the argument that “the practice of archaeology is...a form of dwelling”, and that “the knowledge born of this practice is thus on a par with that which comes from the practical activity of the native dweller, and which the anthropologist, through participation, seeks to learn and understand” (Ingold 1993:152). This acknowledges the difference between ‘original inhabitants’ and researchers, while also specifying how the contemporary researcher can nevertheless come to ‘know’ anything about past realities. Importantly, this is achieved without recourse to universals of biology or neurology, and without the binary opposition of western: ‘non-western’ (cf. Said 1978). The alterity or identi(cali)ty of the San and “us” appears in Lewis-Williams’ arguments not as a consistent view, but shifting in relation to the point under discussion. One may well ask: Cui bono?

‘Art history’ is constructed in a particular way in Lewis-Williams’ argument, not only as a universalising western perspective. For example, he equates it with “an approach to the forms or ‘compositions’ of San rock art” (1998:94). But form and composition are not the same thing (although composition may be an element of ‘form’). The critique contains another example of the problem of meaning, sense and reference. Lewis-Williams charges that “influenced by Western art, researchers divide a painted rock shelter into ‘panels’ - areas of concentrated images that are separated from similar areas by blank spaces” (ibid.). What is the difference between a “panel” and Lewis-Williams’ “area of concentrated images”? When Lewis-Williams prepares an illustration for publication, how does he define the boundaries of a set of paintings, except by the notion of ‘panel’, by any other name? However, labelled, the referent remains the same. That problems in defining the boundaries of ‘panels’ are abundant is well recognised; that the conventions of San art are not the same as Renaissance or modern art, likewise. And to which rock art researchers and studies does Lewis-Williams refer?

A contrary view is that the insights of art history, and of research dedicated to understanding the visual, are invaluable, and warrant archaeological attention. Recently there has been a convergence of archaeological and art historical interests, and a productive interplay of disciplinary perspectives. Many art historians would reject the semiotic model, wherein words and images are seen as ‘signs’, in favour of a more rigorous analysis of visual imagery and its particular operations (cf. Davis 1985, 1989, 1996). In South Africa, Skotnes is one art(ist) historian’ who has valuably introduced such a view, by active engagement with archaeology and anthropology. Interdisciplinary studies contribute to the vitality of rock art research, rather than merely challenging existing ‘truths”. One need only consider changing views of “the meaning of the art” this century to realise that the exposition of meaning is always processual, not to be conflated with ‘truth’ - although researchers seek interpretations which most probably corresponds with the reality of the past. This does not mean that meaning is only perspectival, as the following discussion, of Orpen (1874) and readings of nineteenth century San commentaries, illustrates.

San testimonies: evaluating readings

On the assumption that the truth of the original context of the art has been discovered (rather than modelled or hypothesised), Lewis-Williams (1998:86) believes that dissent centres on the “naive” question: “how much of the art can be explained in terms of San shamanistic beliefs, rituals and experiences?”. My question, however, is: Can San rock art be called shamanistic at all? After all, Katz, author of the definitive work on !Kung curing, explicitly stated that they have “no shamanistic tradition” (Katz 1982:231). A detailed argument against seeing the art as shamanistic (extending my previous (1997a) study) is in preparation, and need not be addressed here, where the focus is on theory and epistemology. I concentrate here on evaluating Lewis-Williams’ reading of nineteenth century San testimonies (Lewis-Williams 1980, 1998) relative to my own interpretation, with special reference to Orpen (1874).

Iinterpretation of Qing’s testimony to Joseph Orpen, is a cornerstone of the shamanistic model. Lewis-Williams uses it, along with /Xam and !Kung texts, to argue that San religion was shamanistic and that the art must be understood in terms of ritual, trance/ hallucinatory experience and an over-arching “shamanistic” cosmology. But most of the references which Lewis-Williams has cited as evidence for San shamanism refer to dead people and mythological/spirit beings, as does the testimony of both Qing and Dia!kwain on paintings of therianthropic figures (Solomon 1997a; and cf. Lee and Woodhouse (e.g. 1970); Pager (e.g. 1975,1984). Responding to my interpretation, Lewis-Williams has reiterated his earlier position, viz. that Qing’s comments were cobbled together subsequently by Orpen, who was unaware that Qing was speaking in trance metaphors. As evidence, Lewis-Williams cites Orpen’s statement: “I shall string together Qing’s fragmentary stories as nearly as I can as he told them to me. I noted them down from him then and since; I only make them consecutive...” (Orpen 1874:3).

However, there can be no question of Orpen “fabricating” Qing’s comments in this way. It is clear from Orpen’s text that it was the series of stories narrated by Qing that Orpen made consecutive. The comments on the paintings were made first: “I asked …[Qing] what the pictures of men with rhebok’s heads meant. He said “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.” I asked him when were the elands spoilt and how. He began to explain...” (Orpen 1874:2; original italicisation). Thereupon he related the stories which Orpen made “consecutive”; there is no indication that his initial comments on the therianthropes are anything less than a verbatim account (albeit translated). Orpen’s use of direct quotation marks, and his statement that he had made notes at the time of their conversations (and after) all suggest that Orpen faithfully recorded Qing’s explanations.

The question remains: What did Orpen mean by ‘consecutive’? The description of the conversation indicates that Qing first related the story of the ‘spoiling’ of the eland, in response to Orpen’s specific question about this event and time. Cagn’s name, and his role as creator of the world, only came up subsequently, in the course of the telling of the eland creation story, which, logically, took place after Cagn’s initial creation of the world. Therefore, when Orpen said he had made the stories consecutive, he meant only that he had ordered them according to the trajectory of San mythology itself, starting with Cagn making the world and its contents, followed by stories which feature later in the San mythological time line. There is thus no question of Orpen having re-combined or misunderstood Qing’s comments on the rhebok-headed men. Lewis-Williams (1998:92) suggests that my disagreement on this point is “not crucial”; on the contrary, my reading obviates the necessity for interpreting Qing’s comments as a confused string of statements created by a bemused Orpen, and the need for “trance-lation” disappears. Seen in relation to San mythology, it is not “anfractuous” [circuitous, indirect] (Lewis-Williams (1980:473), nor couched in trance metaphors (as “death”, “underwater” and “spoil” are claimed to be), but entirely straightforward.

Various San groups, including the /Xam and Kua, recount stories about a powerful being associated with death and underwater, who corresponds broadly to the “lesser god”, associated with death and disease. Qing himself related a regrettably brief story about an underwater being who caught people by the foot and held them there (Orpen 1874:9). It is unnecessary to assume that death ‘means trance’, since a dangerous being associated with underwater is a feature of San mythology and belief more widely. There is therefore no need to suggest that Qing’s first statements - about “men who had died and now lived in rivers” – refer to anything other than people believed to have actually died. To argue this, Lewis-Williams must make recourse to /Xam texts, where ‘death’ is also assumed to refer to trance, not mortality. However, the majority of the /Xam texts cited also refer to physically dead !gi:ten (Solomon 1997a). As Lewis-Williams himself has pointed out (*), the singular, !gi:xa, translates as “full of magical power”. Lucy Lloyd and Dorothea Bleek translated it as “sorceror”[sic], whereas Lewis-Williams’ has substituted “shaman”. The /Xam term does not, however, warrant so specific a translation, and the more literal translation may usefully be reinstated, to cover both living and dead beings with magical abilities (Solomon 1997a; in prep.). Instead, the existence of a dangerous, underwater being (with magical powers) is ignored in favour of a trance explanation which depends on diverse analogies with !Kung and /Xam. In short, the reference to dead men in rivers in entirely in accordance with San cosmology and religious thought, and does not require elaborate analogies to decode.

The third part of Qing’s statement concerns the spoiling of the eland, progeny of Cagn’s wife, its killing by Cagn’s sons, and its reincarnation, no longer as part-human, but as fully animal prey. The day the herds of eland were re-created from Cagn’s eland’s remains was “the day the elands were spoilt and became wild”. This story - and its occurrence at a particular time - are easily understood in relation to the trajectory which characterises the mythology of almost all recorded San. In myth, the world was created by a superior being (southern San: Cagn, /Kaggen), and in the early days the earth was populated by beings part-human and part animal. Animals could speak, while the first San were human-like, but uncultured and lacking in manners and customs. Later, humans and animals were separated. The event marking the separation is different in plot and characterisation from group to group, but the effects are identical: humans become human, and acquire proper behaviour (but lose immortality), whereas animals lose their capacity for speech, becoming mere animals and prey. The widely told story of the Origin of Death, which describes the loss of immortality and the necessity of death, also relates to this transition to modernity.

The spoiling of the eland story relates to this transition from primal time to the world of civilised, modern San. The part-human eland of the old order is replaced in the new by herds of ‘real’ eland. When Qing said that this was when they “were spoilt and became wild”, he was referring to this transformation of the old order after the transition. “Spoilt” here means “transformed” (see further below). In these terms, the therianthropes were indeed dead people living in rivers, vanquished at the same time as Cagn’s eland, at the end of the old order. This was confirmed in *Dia!kwain’s independent testimony, when he stated that painted therianthropes belonged to the prior “ancient race” of Bushmen, who were believed to kill people. In the /Xam texts, and in other parts of Dia!kwain’s narrations to Bleek and Lloyd, the people who existed in primal time are referred to as “the people of the early race”; it was apparently to these whom Dia!kwain was referring when describing the therianthropic figures. The testimony of both San men thus refers to mythology and mythical beings. This recognition allows us to Qing’s statement as entirely coherent: the therianthropes are beings who died at the time when the current order was being instituted, and who are associated with the death realm.

Only Qing’s fourth statement refers to the activities of the living. He said that the dead men were also “ the dances of which you have seen paintings”. Lewis-Williams (1980, 1998) insists that “spoil” here is a metaphor for “enter trance” (by analogy with the spatially and temporally distant, and linguistically divergent, !Kung). But the word “spoil” is used in several different ways within Qing’s account, with general connotations of “transformed, with negative connotations” (Vinnicombe (1976:320, 38n) suggested it should be translated as “harmed”). The “spoiling” of the eland refers to their separation from humans and reincarnation as prey. In another occurrence (Orpen 1874:4), Cagn rebuked his sons for slaying his eland child, saying “for you have spoilt the elands when I was making them fit for use”; this refers to the phase when creation was still in progress, and to the fact that Cagn had not finished making and rearing his creation when his sons killed it. “Spoilt” refers here to the interruption of the process of creation, as outlined in mythology, and there can be no question whatsoever of this usage being a coded reference to entering a trance state. Nor can the reference to men “spoilt’ by the dances necessarily be interpreted in trance terms. Dia!kwain’s testimony indicates that the therianthropic ‘dead men’ were believed to kill people even though they themselves had died, while Qing stated that they still lived in rivers, even though they had died. This is entirely in accordance with /Xam accounts as well as those of the !Kung, Basarwa and others, which indicate that the spirits of dead people were greatly feared. Therefore, when Qing said that the men who had died “were the dances of which you have seen paintings”, he was referring to the effects of the dance (practiced by living San) on the denizens of the death realm (spirits of the dead/mythological beings), and the capacity of the dance to control (harm, damage, spoil) their potentially maleficent powers. Qing’s four statements are consistent with each other in this light.

Although this could be consonant with !Kung trance performance (which does involve control of the spirits of the dead who shoot illness into people), it does not prove that Qing was referring to trance. Nor does Qing’s later account of the ‘dance of blood’ reveal anything about the state of consciousness of the dancers, since falling about and bleeding from the nose are products of exertion-induced,*high blood pressure, not ASCs per se). The assumption that the same (or similar) forms necessarily indicate the same ‘content’ cannot hold. However, for the sake of argument only, it may be assumed that it was indeed a trance dance comparable to that of the !Kung. Nevertheless, Qing and Dia!kwain’s comments still identify therianthropes as mythical beings/spirits of the dead, and not as live shamans. The iconographical implications (Solomon in prep.) are manifold. The present task, however, is to examine the ways in which different readings may be evaluated.

The reading I propose has various advantages over the shamanistic reading, summarised as follows:

(1) All four clauses of Qing’s commentary appear as a coherent whole. Assumptions about the limitations of Orpen and Bleek’s competence are unnecessary, as is the assumption that Qing was speaking in trance codes. (2) All four components of Qing’s comments are addressed. (3) Ethnographic analogy, except by reference to the broad forms of San mythology, is kept to a minimum; analogy does not depend on the specific sense of particular references; and similar forms are not necessarily assumed to have similar ‘contents’ among spatially, historically and linguistically separated groups. In this way the reading does not reproduce the homogeneity and anti-historicity of the shamanistic model. (4) Methodological problems - such as the claim that certain terms (e.g. “spoilt’) always mean the same thing (viz. trance) are avoided, and the meaning of terms is seen as (contextually) mutable, rather than fixed and invariant.

Lewis-Williams’ reading, on the other hand, relies heavily on all manner of analogies with both /Xam and !Kung “shamanism, and all but ignores the mythology which Qing offered in explanation of the paintings. For example, he explains the reference to the ‘spoiling of the elands’ by reference to /Xam “shamans transformed by animal potency” (Lewis-Williams 1998:93), but the references cited refer to spirit possession and the control of animals by dead /Xam, not live shamans (Solomon in prep.). Lewis-Williams claims that when he spoke about the spoiling of the eland Qing “probably meant that the medicine men in trance exploited the eland’s power as they danced”. This, I suggest, is “vague, rather than subtle” (Lewis-Williams 1998:91), as well as methodologically convoluted, and it fails utterly to explain why the eland were spoilt at a particular time. The same applies to the analysis of references to death. Sickness and ‘real’ death are the fundamental problems dealt with in the curing dance and !Kung trance ritual is a strategy for dealing with the problem of mortality. To argue that the !Kung do not distinguish between real and trance death merely introduces more analogies. If trance is seen as a form of death, it is more probably because actual physical expiry is the real problem with which the ritual deals (as does the mythology also). The fact that in San thought the dead are believed to ‘live’ on after death also contextualises this issue. Analysis in terms of (trance) metaphors reintroduces a variety of the depth : surface distinction, where key words (death/trance/spoilt) are seen as having a superficial label, but another, deeper meaning.

In short, I contend that the reading I have put forward is methodologically more satisfactory than Lewis-Williams’, more internally consistent and more rigorous, addressing and integrating all components of Qing and Dia!kwain’s commentaries. Lewis-Williams (1998:93) claims that the test of the reading must be the art itself, and that the art does not support my reading, but this depends on a fallacious argument. Qing’s comments, as interpreted by Lewis-Williams, are fundamental to the notion that the art is shamanistic. It cannot then be argued that therianthropes are “too frequently depicted in shamanistic contexts and with a variety of indisputable shamanic features” (ibid.) for my reading to be correct, since the interpretation of them as “in shamanic contexts” is already a product of Lewis-Williams’s reading of the text. In other words, he argues that Qing’s comments indicate that the art is shaman(ist)ic, and then - on the assumption that the art is shamanistic - argues backwards that the art proves the reading! This is a serious lapse of logic, and the claim that the art does not support my reading cannot stand on these grounds.

Lewis-Williams (1998:92-3) also objects to my emphasis on mythology. Although the shamanists have considered mythology, it is the way they have done so that is problematic (see note 4), with the motifs and metaphors of myth said to derive from and originate in trance experience. The fact that “some beings and creatures feature in both myths and rituals” (Lewis-Williams 1998:93) by no means constitutes a sound basis for dismissing my distinction of myth and ritual. Rather, I suggested that the relationship between them needed further attention, since it is culturally and situationally variable. In these terms, it is as likely that myth and cosmology inform trance (i.e. that the trancer experiences what s/he expects, on the basis of knowledge of this lore) as the inverse, that trance experience permeates the myths (Solomon 1997a). Certainly, the implication of Qing and Dia!kwain’s testimony is that mythology is far more important than has been acknowledged in shamanistic readings.

A related objection is that the narrativity of myth is absent in the paintings, and that “narrative must surely be an important component of an interpretation that is founded on mythology” (Lewis-Williams 1998:93). My proposition, however, was that the characters in the rock art are to be understood in relation to the trajectory of San mythology (from creation though primal time to the current order). There is no requirement for the art to be “narrative” for it to be affiliated to myth, except within the narrow confines of seeing art as illustration. Indeed, it is difficult to see how the narrativity of myths could be transposed into the art. Visual art may be seen precisely as resolving problems of representing an order of time (or narrativity) in an order of space ((Schapiro 1985:215, Solomon 1989). The discussion (above) of intentionality, and of the notion of art as “copying”, indicates that this “translation” process is far more complex. The idea that myth is a narrative form, and that if the art relates to myth it must similarly display narrative features, is misplaced, and derives from a particular view of the way in which art is said to represent thought or ‘mind’ (see above). Since it seems that health and prosperity were believed to be influenced by mythological/spirit beings, the contention that the art relates to mythology - albeit not in the sense of “illustrating” the narratives - remains valid, and has considerable potential to alter the ways in which rock art is currently understood.


The problem of meaning has been extensively debated in the literature of a range of disciplines, including philosophy, linguistics, literary studies and anthropology, to name a few. There has been little debating of semiotic issues in South African archaeology - partly, I suggest, because of a general neglect of theory in the discipline. Indeed, Lewis-Williams, despite his admirable record of having developed the theoretical basis of rock art research, seems unaware of developments since the formulation of Pierce-Morris semiotics in the 1930s. Further examination of the theoretical tenets and models of meaning which inform interpretations may bring some of the crucial debates of our times further into archaeological discourses. In particular, the structuralist/ semiotic-based model contains a number of problems (cf. also Davis 1985,1989, 1996), including: its idealism; the way in which icons and symbols are conceptualised; the persistent assumption that graphic signs function in quasi-linguistic terms; logocentric notions of art, as unmediated ideas or fossil thoughts; and more.

Recognition of the role of theoretical premises in constituting divergent interpretations may contribute to more constructive debate, founded on a proper understanding of those premises. However, other differences of opinion are less closely aligned to premises, although they do demand further attention to questions of method, evaluation and epistemology. My contention that Lewis-Williams has, quite simply, misinterpreted San texts in his construction of San “shamanism” as the context and meaning of San exemplifies this. Unfortunately, Lewis-Williams does not engage with the particularities of that interpretation, but merely reiterates his position. This neither constitutes argumentation, nor is it constructive; moreover, it speaks volumes about hegemony, and the current state of rock art research.

The implications of the position I have outlined are not limited to iconographical questions. Recognising that paintings are not simply copies of thoughts, but that they are mediated in the making, requires a more rigorous approach to the production of art. Skotnes (1994) has suggested, from a slightly different perspective, that studies of rock art need to incorporate understanding of artistic praxis. This might be extended to incorporate consideration of technology more generally (but technology as it interfaces with thought, rather than as merely a study of physical processes [cf. Dobres *]. Ironically, manufacturing processes and techniques have been studied at length in relation to stone artefacts, where the minutiae of the modifications that ultimately characterise various tool categories have been researched. Little work of this kind has been conducted in rock art research. If rock art is to be seen as a trace of ancient ‘mind’, technology and artistic praxis require further attention, alongside the iconographical studies which have largely characterised the sub-discipline.


Bleek, W.H.I. 1874. Remarks on J.M. Orpen’s “Mythology of the Maluti Bushmen”. Cape Monthly Magazine 9:1-13.

Bourdieu, P. 1977. Outline of a Theory of Practice. Cambridge: Cambridge University Press.

Cixous, H.1975. Sorties. In: La jeune née. Union Générale d’Editions 10/18. Reprinted in E. Marks and I. de Courtivron (eds.) 1985, New French Feminisms: An Anthology. Brighton: Harvester Press.

Davis, W. 1985. Present and future directions in the study of rock art. South African Archaeological Bulletin 141:5-10.

Davis, W. 1989. Style and history in art history. In: M. Conkey and C. Hastorf (eds), The uses of style in archaeology, pgs. 18-31. Cambridge: Cambridge University Press.

Davis, W. 1996. Replications: archaeology, art history and psychoanalysis. Pennsylvania: Pennsylvania State University Press.

Delphy, C. 1984. Close to Home - a materialist analysis of women's oppression. Edited by D. Leonard. London: Hutchinson, in association with the Explorations in Feminism Collective.

*Dobres, M-A. (awaiting publication details from author)

Eco, U. 1976. A theory of semiotics. Bloomington and London: Indiana University Press.

Eliade, M. 1964. Shamanism: archaic techniques of ecstasy. Princeton: Princeton University Press.

Giddens, A. 1979. Central Problems in Social Theory: Action, Structure and Contradiction in Social Analysis. London: Macmillan.

Guenther, M.G. 1986. The Nharo Bushmen of Botswana: tradition and change. Hamburg: Helmut Buske Verlag (Quellen zur Khoisan-Forschung).

Guenther, M.G. 1989. Bushman folktales: oral tradition of the Nharo of Botswana and the /Xam of the Cape. Stuttgart: Franz Steiner Verlag Wiesbaden.

Hawkes, T. 1977. Structuralism and semiotics. Methuen New Accents Series. London: Methuen.

Hodder, I. 1987. The archaeology of contextual meanings. Cambridge: Cambridge University Press.

Ingold, T. The temporality of the landscape. World Archaeology 25(2):152-174.

Katz, R. 1982. Boiling energy: community healing among the Kalahari !Kung. Cambridge, Mass.: Harvard University Press.

Lee, D.N. and Woodhouse, H.C. 1970. Art on the rocks of southern Africa. New York: Charles Scribner and Sons.

Lewis-Williams, J. D. 1972. The syntax and function of the Giant’s Castle rock paintings. South African Archaeological Bulletin 27:49–65.

Lewis-Williams, J. D. 1974. Superpositioning in a sample of rock-paintings in the Barkly East district. South African Archaeological Bulletin 29:93–103.

Lewis-Williams, J.D. 1980. Ethnography and iconography: aspects of southern San thought and art. Man (n.s.) 15:467-482.

Lewis-Williams, J. D. 1981. Believing and seeing: symbolic meanings in southern San rock paintings. London: Academic Press.

Lewis-Williams, J.D. 1982. The economic and social context of southern San rock art. Current Anthropology 23:429-449.

Lewis-Williams, J.D. 1983. Science and rock art: introductory essay. South African Archaeological Society Goodwin Series 4:3-13.

Lewis-Williams, J.D. 1984a. The empiricist impasse in southern African rock art studies. South African Archaeological Bulletin 39:58-66.

Lewis-Williams, J.D 1984b. Ideological continuities in prehistoric southern Africa: the evidence of rock art. In: C. Schrire (ed.) Past and present in hunter gatherer studies. New York: Academic Press.

Lewis-Williams, J.D. 1985. Testing the trance explanation of southern African rock art: depictions of felines. Bollettino del Centro Camuno di Studi Preistorici 22:47–62.

Lewis-Williams, J.D. 1998. Quanto? The issue of many meanings in southern African San rock art. South African Archaeological Bulletin 168:86-97.

Lewis-Williams, J.D. and Dowson, T.A. 1988. Signs of all times: entoptic phenomena in Upper Palaeolithic art. Current Anthropology 29:201-245.

Lewis-Williams, J. D. & Dowson, T. A. 1989. Images of power: understanding Bushman rock art. Johannesburg: Southern Book Publishers.

Lewis-Williams, J. D. & Loubser, J. H. N. 1986. Deceptive appearances: a critique of southern African rock art studies. In: Wendorf, F. & Close, A. E. (eds), Advances in World Archaeology Vol. 5: 253–89. New York: Academic Press.

Moore, H.1986. Space, Text and Gender: An Anthropological Study of the Marakwet of Kenya. Cambridge: Cambridge University Press.

Moore, H. 1988. Feminism and Anthropology. Cambridge: Polity Press.

Orpen, J.M. 1874. A glimpse into the mythology of the Maluti Bushmen. Cape Monthly Magazine 9:1-13.

Pager, H. 1975. Stone Age myth and magic. Graz: Akademische Druck.

Parkington, J. E. 1989. Interpreting paintings without a commentary. Antiquity 63:13–26.

Parkington, J. 1996. What is an eland? N!ao and the politics of age and sex in the paintings of the Western Cape. In: Skotnes, P. (ed.) Miscast: negotiating the presence of the Bushmen: 281–289. Cape Town: University of Cape Town Press.

Parkington, J. E. & Manhire, A. 1997. Processions and groups: human figures, ritual occasions and social categories in the rock paintings of the Western Cape, South Africa. In: Conkey, M. W., Soffer, O., Stratmann, D., & Jablonski, N. G. (eds) Beyond art: Pleistocene image and symbol. San Francisco: Memoirs of the California Academy of Sciences, No. 23:301–320.

Preziosi, D. 1989. Rethinking art history: meditations on a coy science. New Haven and London: Yale University Press.

Said, E. Orientalism. London and New York:

Seymour-Smith, C. 1986. Macmillan dictionary of anthropology. London and Basingstoke: Macmillan.

Schapiro, M. 1985 [1969]. On some problems in the semiotics of visual art: field and vehicle in image signs. In: R.E. Innis (ed.), Semiotics: an introductory anthology. Bloomington: Indiana University Press.

Skotnes, P. 1990. Is there life after trance? De Arte 44:16-24.

Skotnes, P. 1994. The visual as a site of meaning: San parietal painting and the experience of modern art. In: Dowson, T. A. & Lewis-Williams, J. D. (eds) Contested images: diversity in southern African rock art research: 315–329. Johannesburg: Witwatersrand University Press.

Solomon, A. 1989. Division of the earth: gender, symbolism and the archaeology of the southern San. Unpublished MA dissertation, University of Cape Town.

Solomon, A. 1992. Gender, representation and power in San ethnography and rock art. Journal of Anthropological Archaeology 11:291-329.

Solomon, A. 1994. “Mythic women” a study in variability in San art. In: T.A. Dowson and D. Lewis-Williams (eds.), Contested images: diversity in southern African rock art research: 331-371. Johannesburg: Witwatersrand University Press.

Solomon, A. 1995. Rock art incorporated: an archaeological and interdisciplinary study of certain human figures in San art. Unpublished Ph.D. dissertation, University of Cape Town. .

Solomon, A. 1997a. The myth of ritual origins? Ethnography, mythology and interpretation of San rock art. South African Archaeological Bulletin 52:3–13.

Solomon, 1997b. Landscape, form and process: some implications for San rock art research. Natal Museum Journal of the Humanities 9:57–73.

Solomon, A. 1998. Ethnography and method in southern African rock art research. In: C. Chippindale and P.S.C. Taçon (eds), The archaeology of rock art. Cambridge: Cambridge University Press.

Solomon, A. 1999. The quest for mind via San rock art. Unpublished paper presented at the 4th World Archaeological Congress, Cape Town, 1999.

Solomon, A. in prep. Magic, myth and San art (working title).

Stow, G. W. 1905. The native races of South Africa. London: Swan Sonnenschein

Sparkes, A.W. Talking philosophy: a wordbook. London and New York: Routledge.

Thomas, J. 1996. Time, culture and identity: an interpretive archaeology. London, New York: Routledge.

Vinnicombe, P. 1976. People of the eland: rock paintings of the Drakensberg Bushmen as a reflection of their life and thought. Pietermaritzburg: University of Natal Press

back to top of page