'Exotic Bradshaws' or Australian 'Gwion': an archaeological test.
Barry, Michael ; White, J Peter
Abstract: 'Bradshaws', Aboriginal rock-art figures in the Kimberley, recently have become a focus of increased interest and publication. We discuss six recent pieces of work relevant to the suggestion that they are of exotic, rather than Australian, origin. We present robust evidence of their Aboriginality.
Background
Bradshaws are paintings on rock surfaces in the Kimberley that look like elegant, enhanced stick-figure 'humans', often with accoutrements. First reported for European eyes by the explorer Joseph Bradshaw in 1892, they are now known from an area of some 60 000 [km.sup.2] in the Kimberley of north-western Australia (Figure 1). Ngarinyin people from this country call them 'Gwion Gwion' (Ngarjno et al. 2000:11) (1).
[FIGURE 1 OMITTED]
Scholars with some knowledge of world rock-art often remark on their apparent similarity to figures in Spain or Southern Africa. These comments are seldom published, so they are difficult to reference.
Ever since Bradshaw's report, there have been two very different attitudes to them. One view is that they are so different from other figures in Australia that they must have been painted by non-Aborigines. This view has been expressed by Walsh (1994) in his sequence division of the rock-art of the Kimberley. He constructed three epochs and named them: 'Archaic', 'Erudite', and 'Aborigine'. He placed the 'Bradshaws' in the Erudite Epoch. The use of the word 'erudite' assumes that the cultures from the other epochs were inferior and less developed. The use of the word 'Aborigine' suggests that there was no Aboriginal constituent to the previous epochs, that they were new arrivals upon the scene and their heritage began, in this area, after the 'Bradshaw period' (Walsh 1994:32). That this is a commonly held view is documented by McNiven and Russell (1997), and Redmond (2002).
The other view, (e.g. Lewis 1997) is that they are Australian, and their relationship to the rest of the Aboriginal art corpus merits study.
We note that the problem we address resonates strongly with current political ideologies. We do not explore this aspect.
Although there is a considerable literature on the 'Bradshaws', six recent works are considered here. In chronological order, they are: Walsh (1994), Barry (1997), McNiven and Russell (1997), Lewis (1997), Walsh (1999) and Redmond (2002). The three dated to 1997 were not informed by each other. All have extensive bibliographies which cover various aspects of the older literature. Five of the six works have been published and are generally available, so there is no need to review them in detail. Barry (1997) is not so generally available, for which reason we concentrate on it here. His BA (Hons) thesis tackled the question of whether or not Gwion figures resemble those from overseas more or less than Australian figures. Barry's purpose was to examine the hypothesis that the Gwion resemble prehistoric figures outside Australia more than they do those within the continent.
Two other works that are contemporaries of the above papers are Welch (1993) and Tacon et al. (2003). Welch's paper is devoted to a detailed recording of the Gwion by both words and pictures, and Tacon positions like figures (some of different colours) in the Keep River region. These were considered to be aspects outside the thrust of this paper. Likewise, we were informed by Graeme Ward (written communication) that he and others have recorded 'not dissimilar purple figures along the Fitzmaurice River'.
Methods
Barry collected all the images he could find, in the literature, of apparently human figures on rock surfaces in the tropical and temperate zones of the Old World (Figure 2). Two thousand, two hundred and thirty images were collected. These include 1694 images from seventeen areas outside Australia and 536 from three areas within Australia, mostly Kimberley (Gwion figures only) and Arnhem Land. The images were analysed in three ways: visually, by a correspondence analysis of 79 attributes measured on each figure (where possible), and whether there exists a plausible explanation for any similarity.
[FIGURE 2 OMITTED]
Visual analysis
Visual analysis depends on our ability to process and categorise an enormous amount of data quickly. Its downside is that weighting of various aspects is not controllable, so the analysis is not open to rational discussion. Barry distinguished groups of images from a geographical area that displayed a visual affinity and called these groups 'genres' (Table 1).
Visual grouping of the images resulted in 118 genres, including 15 from Arnhem Land but only one from the Kimberley: the Gwion. Comparison of the genres showed that in only two cases were there any likenesses between Gwion and other genres. These were Algeria and Arnhem Land. Among the Algerian figures, three could be said to resemble the Gwion.
The Algerian-Gwion similarities (three of 206) should be compared with those of Arnhem Land--Gwion (89 of 315 figures in three genres were similar: 49 of 59, 27 of 29, and 13 of 21). The Arnhem Land similarities reside in design and inspiration: the posed stance, muscle depiction and portrayal of movement. We consider these similarities to be more significant than the Algerian ones.
The visual analysis is subjective, and an Australian scholar might more readily see similarities with Arnhem Land than with less familiar figures. We therefore turn to multivariate description and analysis.
Multivariate analysis
Seventy-nine attributes were assessed for each of the 2230 images. There were two groups of attributes: bodies and things associated with them. Thirty-six attributes of bodies included direction faced, posture, image type, eye, sexed, and so on. The 43 accoutrements included headdress, apron, shield, bow, bag, basket, and nose-bone. Technique, size and colour of image were ignored. Details are given by Barry (1997:37-43). The attributes were not weighted. They were analysed by correspondence analysis using the program MV-Nutshell (Wright 1994).
The Gwion images are isolated from all the other areas, but their closest affinities lie with Arnhem Land.
Barry's work has shown that Gwion figures, as described by the presence or absence of 79 attributes, resemble figures from Arnhem Land and differ from all but three of the 1694 figures measured from elsewhere in the world. The three exceptions could well be coincidental.
Other works
Lewis (1997) took issue with Walsh (1994), primarily on archaeological grounds. He presented a case for seeing strong similarities between Gwion and Arnhem Land paintings from the Dynamic Figure period. These similarities can be summarised as follows:
* both show small monochrome humans who bear many similar clothes and ornaments. Both are portrayed without fingers, toes or genitals;
* males dominate, females are rare, children nonexistent (sexes are identified by accoutrements);
* both groups commonly carry boomerangs and sometimes spears; and
* marsupials or artefacts are often portrayed nearby and figures are often surrounded by a cloud of dots.
Lewis also saw parallels between the subsequent periods: 'Clothes Peg Figure' (Kimberley) and 'Hooked Stick' (Arnhem Land). These include the appearance of the hooked stick / spearthrower and the regular occurrence of bichrome figures in both areas. Lewis argues that, on the basis of superposition sequence and animal content, both sets of paintings must be dated to the Late Pleistocene--Early Holocene. At this time both art areas were part of the 'Bonaparte Catchment', which, he considers, was the geographic basis of an information/trade network.
McNiven and Russell (1997) pointed out that the 'non-Aboriginal origin' argument fits within a strong discourse of colonial Australia. This began at least as early as the 1820s and drew its inspiration from the more general European view of 'savages' as completely uninventive. Such elegant and sophisticated pictures as the Gwion must thus have been painted by another race. McNiven and Russell (1997:807) linked this colonial view to current attempts to deny Aborigines 'cultural authenticity and legitimacy'.
Redmond (2002) was concerned especially with the social impact of Walsh's work (1994 and various popular press accounts). He noted that local Aboriginal peoples can link their current cultural repertoire to the Gwion figures, but that their public and published work is ignored by Walsh, who has been recording the art for the last 20 years. Redmond's discussion drew particular attention to the attraction to tourist operators and tourists of Walsh's argument that Gwion were painted by a mysterious, erudite, 'other' race.
Lewis, McNiven and Redmond take the Aboriginality of the Gwion art for granted. They accept that it was painted by ancestral Aborigines and that current communities have ownership of it. Only Lewis seeks to demonstrate the similarity of Gwion to other Aboriginal art, and he does so wholly within an Australian Aboriginal art framework. All these support Barry's findings.
In Bradshaws: ancient rock paintings of northwest Australia, Walsh (1994) displayed the most comprehensive collection of photographs and drawings of Kimberley art published at that time. '[In this] book, Walsh, a self-taught Queensland rock art consultant, brought the hypothesis of alien origins to public attention and triggered a hot international debate, raging from the Kimberley to Paris' (Dayton 1997:4).
Walsh, in Bradshaw art of the Kimberley (1999), covered the Bradshaw paintings to a greater depth than in his previous book. It is in many ways an excellent book; he expanded on his previous publication and attempted to refute opinions expressed in subsequent publications where their findings do not support his concepts. He was also critical of the subjectivity in Barry's work. Barry's statement that his collection of images was restricted to what he could find in publications was interpreted by Walsh as being 'selective'. While Barry's collection must be a sample, as it does not comprise all the images in the world, it does consist of all the 'human-like' images that could be found within the time-limit constraints. Archaeologists work with samples, because they must. It is not possible to obtain the complete population of any archaeological entity. Some examples are destroyed, some unfound, many unpublished. Walsh works with samples also.
Another of Walsh's complaints was about the attributes chosen, particularly the lack of image size as a considered variable. As Barry explained, the method used has the advantage that the attributes used can be reported, so that they can be criticised, and the experiment can be repeated with or without change to the data. Walsh attempted something of the sort in Figure 681 (1999:423); here, a carefully selected group of 13 images--3 from Arnhem Land, 4 from the Tassili area of Algeria, and 6 from the Kimberley-were assembled by Walsh to compose an imagined hunting scene. The reader is invited to attempt to group the selected hunters by 'personal assessment' (could this be visual analysis?) and thus perhaps justify the previous five pages.
Figure 14 shows a correspondence analysis of the mean centroids of the images, selected by Walsh, from the three geographical areas using the applicable attributes from Barry's thesis. This figure clearly shows that there are significant differences in the images from different areas, despite the number of images being too few for otherwise useful interpretation.
[FIGURE 14 OMITTED]
It appears that Walsh selected a small number of specific images that he thought would cluster and be associated as he wished, and expected his readers to see them in a similar light. In practice, multivariate analysis had no difficulty in untangling this attempt to construct confusion.
Discussion
Many of the images considered by Barry (and the other authors mentioned in this paper) look as though they are enhanced stick-figure pictures of 'people'. They are described as 'stick-figure' because of the common way of depicting humans as though they were made of lines or sticks. The 'enhancement' consists of varying the thickness and shape of the 'sticks' in a way that approximates with anatomical detail. As Barry (1997) has shown, this manner of drawing occurs almost universally. Its explanation should be sought in our perceptions of the shapes of humans, and the problems of depicting them on a surface, rather than in any particular culture.
A proposal for an exotic origin of the Gwion/ Bradshaw figures must meet two conditions. First, closely similar paintings need to be found outside Australia. Second, some plausible link between the two occurrences must be established.
Barry's work has demonstrated that Gwion figures do not, as suggested by the exotic origin story, resemble overseas figures more than Australian. They resemble Australian figures so strongly that an origin within the Australian continent is indicated. Finally, there is plausibility. Algeria as a source for Gwion is, in our view, about as implausible as one could imagine.
We conclude that archaeological evidence fails to establish a case for the exotic origin of Gwion images. Given the similarities that we and others have demonstrated with Arnhem Land figures, we consider that it is most likely they were the creation of ancestral Aborigines.
As a postscript, we note that, in December 2003, Justice Sundberg, in recognising the native title rights of the Ngarinyin, Wunambal and Worora peoples, singled out their rights of access to Gwion figures. Thus, the court has recognised the Aboriginality of Gwion figures. Table 1 The figures in each genre display an overall similarity Geographical unit Genres Images Outside Australia South Africa 3 94 Southern Africa 2 43 Zimbabwe 7 184 Algeria 12 206 Libya 11 106 Ethiopia 5 46 Spain 9 71 Italy 1 30 Turkey 2 12 Armenia 4 29 Saudi Arabia 3 40 United Arab Emirates 1 27 Oman 6 65 India 18 371 China 4 89 Malaysia 5 223 Indonesia 6 58 Subtotal 99 1694 Inside Australia Arnhem Land 15 315 The Kimberley 1 199 The Pilbara 2 21 The Tombs 1 1 Subtotal 19 536 Total 118 2230 Source: Barry 1997, ch. 4.
NOTE
(1.) 'Gwion Gwion' is the Ngarinyin term for the corpus of these representations; Ngarinyin spokespersons have advised that 'Gwion' is an acceptable abbreviation--Editor.
REFERENCES
Ballard, C 1988, 'Dudumahan: a rock-art site on Kai Kecil, southern Moluccas', Bulletin of the Indo-Pacific Prehistory Association 8:139-61.
Barry, M 1997, 'I would sooner not call them Bradshaws', BA (Hons) thesis, University of Sydney.
Brandl, EJ 1982, Australian Aboriginal paintings in western and central Arnhem Land: temporal sequences and elements of style in Cadell River and Deaf Adder Creek art, Australian Institute of Aboriginal Studies, Canberra.
Chaloupka, G 1977, 'Aspects of the chronology and schematisation of two prehistoric sites on the Arnhem Land plateau', in PJ Ucko (ed.), Form in Indigenous art: schematisation in the art of Aboriginal Australia and prehistoric Europe, Australian Institute of Aboriginal Studies, Canberra, pp. 243-59.
Crawford, IM 1968, The art of the Wandjina: Aboriginal cave paintings in Kimberley, Western Australia, Oxford University Press, Melbourne.
Dayton, L 1997, 'Blow-ins ruled out as the creators of ancient rock-art', Sydney Morning Herald, 27 June.
Lewis, D 1997, 'Bradshaws: the view from Arnhem Land', Australian Archaeology 44:1-16.
Lhote, H 1960 (1958), Tassilli frescoes: the story of prehistoric rock-paintings of the Sahara (trans. by AH Brodrock), Hutchinson, London.
McNiven, IJ & Russell, L 1997, '"Strange paintings" and "mystery races": Kimberley rock-art, diffusionism and colonialist constructions of Australia's Aboriginal past', Antiquity 71:801-9.
Ngarjno, Ungudman, Banggal, Nyawarra (J Doring, ed.) 2000, Gwion Gwion: secret and sacred pathways of the Ngarinyin Aboriginal people of Australia, Konemann, Koln.
Redmond, A 2002, '"Alien abductions", Kimberley Aboriginal rock-paintings, and the speculation about human origins: on some investments in cultural tourism in the northern Kimberley', Australian Aboriginal Studies 2002/2:54-64.
Roder, J 1959, Felsbilder und Vorgeschichte des MacCluer-Golfes West-Neuguinea, LC Wittich Verlag, Darmstadt.
Tacon, P, Mulvaney, K, Ouzman, S, Fullagar, R, Head, L & Carlton, P 2003, "Changing ecological concerns in rock-art subject matter of north Australia's Keep River region', Before Farming 2003/3(4), viewed 24 March 2004, <http://www.waspress.co.uk/journals/ beforefarming / >.
Walsh, GL 1994, Bradshaws: ancient rock paintings of northwest Australia, Edition Limitee, Geneva.
--1999, Bradshaw art of the Kimberley, Takarakka Nowan Kas Publications, Queensland.
Welch, D 1993, 'Early "Naturalistic" human figures in the Kimberley, Australia', Rock Art Research 10:24-37.
Wright, RVS 1994, The MV-Nutshell brochure: a concise introduction to multivariate archaeology, author, Sydney.
Michael Barry is a consultant archaeologist with extensive surveying experience.
Camperdown, Sydney, <mbarry@goanna.net.au>
Peter White is an Honorary Associate in Archaeology at the University of Sydney.
Redfern, Sydney, <peter.white@arts.usyd.edu.au>