Christopher Chippindale

This is one of the web files where I keep published or unpublished writing, so they are available in electronic form. The purpose of putting this on the Internet is so that people can have access to it. Any fair use of it is welcome, including its use as much as you want in learning and teaching, and printing-out or using items from it in reasonable quantities. It is not fair to pass it off as your own work, or to use material in a way that generates income for you.

I am always glad to hear from people who find anything in my work useful, or interesting. Feedback is most welcome.

The texts are taken from whatever is to hand. For published work, that is usually the word-processor file originally submitted to the editor of the publication. It will be essentially what was published, but without revisions, corrections and proof-reading fixes: for a word- and letter-perfect version you must look at the publication. They also lack, usually, all the pictures, for which again you must look at the publication.

Any footnotes in the printed text are not here.

To send an e-mail to me, click here, to return to my home-page click here.

Christopher Chippindale currently at

Cambridge University Museum of Archaeology & Anthropology, Downing Street, Cambridge CB2 3DZ, England
cc43@cam.ac.uk

This page is written by Christopher Chippindale. Last update: 12/2/2000. © Christopher Chippindale and/or its other author(s).

The many ways of dating Arnhem Land rock-art, north Australia

Christopher Chippindale

Cambridge University Museum of Archaeology & Anthropology, Downing Street, Cambridge CB2 3DZ, England
cc43@cam.ac.uk

&

Paul S.C. Taçon

Australian Museum, 6-8 College Street, Sydney NSW 2010, Australia

Chippindale, Christopher & Paul S.C. Taçon. 1998. The many ways of dating Arnhem Land rock-art, north Australia, in Christopher Chippindale & Paul S.C. Taçon (ed.), The archaeology of rock-art: 90—111. Cambridge: Cambridge University Press.

Fortunate circumstances – and good work by the first generation of field researchers – have provided an unusually strong framework for dating the long and varied repertoire of rock-art in Arnhem Land, north Australia: it is now one of the very few regional rock-art sequences with a clear relative and absolute chronology. With a range of evidence pertinent to date, it now becomes possible to explore – rather than to suppose – whether conventional guiding assumptions are true, whether a distinctive 'style' is in fact a good marker of date. And a clear chronology guides the varied approaches from informed knowledge and from formal methods which bind together in an integrated knowledge of a complex and highly informative rock-art region.

Arnhem Land (Berndt and Berndt 1954) is a large region on the central north coast of Australia, a block of land some 400 km east—west by 250 km north—south. It is bounded on the north by the Arafura Sea separating Australia from New Guinea, and on the east by the Gulf of Carpentaria. To the west is the modern city of Darwin, and the rest of the 'Top End' of the present-day Northern Territory. To the south, Arnhem Land runs into the semi-desert country as it approaches arid central Australia. The climate is tropical, with an annual nine months of dry season, and a flooding three months of monsoon wet.

Parts of Arnhem Land, especially the region of the three Alligator rivers in western Arnhem Land, now partly in Kakadu National Park, have large exposures, with crags and cliffs, of ancient pre-Cambrian sandstone. This 'western Arnhem Land' or 'Alligator Rivers' region (Fig. 6.1) is celebrated for the quantity, range and artistic quality of its rock-art (Brandl 1973; Chaloupka 1993). Several hundred sites are known with many thousands of figures –so many not even a reliable estimate of the tally exists; since not all the area has been covered by intensive survey (for example Chaloupka et al. 1985; Chaloupka 1992), much likely remains to be re-discovered, and very few sites have been recorded in detail (see Chippindale & Taçon 1993). The bulk of the art is painted. There are also some engravings (Sullivan 1989), figures etched into the rock by application of some corrosive fluid (Chaloupka 1993: 235—236), and figures made in the rare medium of moulded beeswax applied to rock-surfaces and making low-relief figures (Brandl 1968).

Western Arnhem Land, specifically the drainages of the Alligator rivers, has also benefitted from palaeoecological and archaeological research of an intensity rare in Australia. There is a well-studied framework of ecological change, in relation to changing sea-levels and river regimes (Woodroffe et al. 1986); the consequent changes in flora and fauna can be sketched from that, and in turn the subsistence opportunities seen for hunter—gatherers that depend on resources and their seasonal availability. The archaeological deposits give well-dated sequences that relate to the palaeo-ecological record (for example, Schrire 1982); the Arnhem Land soils, acid sands that are soaked each wet season, are hostile to the continuing survival of any materials other than stone. The long-term archaeological record, accordingly, is overwhelmingly a story of lithics which is not easy to interpret in human terms.

It is not unfair to say that in Arnhem Land, the archaeology tends to offer 'chronology without information'; once below the upper levels where the shells in middens shield materials from the acid soil, there are commonly only the lithics. Below a certain depth, even the charcoal and finely divided carbon perishes, so the basal dates for the sequences are by luminescent techniques applied to the sand matrix (for example Roberts et al. 1994). The rock-art, with its intricate record of humans and their activities, shows their kit of hunting and gathering gear and of ceremony. Nearly all the objects are in organic materials that perish in the archaeological record; at one time, it promised the complement – 'information without chronology'.

How a developed chronology for the rock-art has been established is the subject of this essay: partly by working with features of the rock-art itself, partly by 'bridging' across to the palaeo-ecological and archaeological records, we now have information and chronology in the rock-art.

Unusually, painting on the rocks has continued in Arnhem Land up to the present. There are many rock-paintings from this century. The iconography, pigments, and design-rules of the later rock-art continue in the celebrated bark-paintings of the region and – within the last few years – in paintings on fine art-paper. Traditional knowledge of the paintings is perpetuated along with the images themselves. By drawing on a variety of evidence, as detailed in this paper, a long sequence is documented in which both the subjects and the manners of depiction change. Links between the pictures and the dated palaeo-ecological—archaeological sequence for western Arnhem Land permits this relative sequence to be tied, for its later portions, to an absolute chronology. And there are some radiocarbon dates directly from or pertinent to the art itself.

The rare good fortune of Arnhem Land painting having continued into present times allows opportunity of direct insight into the images. For a very few pictures, one can today ask the artist directly what they depict, and why they were painted! A continuing knowledge and memory of hunter-gatherer ways, and of the stories of the country, enables many images to be understood, sometimes by reference to knowledge of the ancient Dreaming-time when the world was made and ordered, sometimes by everyday considerations: one type of fish, a mullet, is painted with its head snapped back – because you do well to break its neck when you catch one, lest it cheekily flaps and flips its way from the bank back into the water. Distinctive figures with distinctive attributes can be traced back from their appearance in the modern stories far through the art sequence.

While a great many elements in the recent rock-art are known in modern times or ethnographically, elements in the older art are different. The didgeridu, the musical instrument of everyday life and of ceremony today, seems absent from the ancient art, and so is the modern spear-thrower; the boomerang, nowadays used only as clap-sticks in music-making, is seen in old art as a hunting and a fighting weapon. The intricate conventions of the X-ray manner of depiction, commonplace in recent art, are known from contemporary bark- and paper-painting. The older paintings use different conventions, as well as depicting other subjects; more formal methods are there called for.

With exceptional, even unique opportunities for insight from contemporary knowledge, with a good dating sequence, and with older paintings full of intricate well-drawn detail, Arnhem Land researchers have also worked in a framework of analogy (e.g. Lewis 1988: 68—70).

Alison Wylie (1989) has usefully distinguished two habits of archaeological reasoning.

Commonly, we work by chains of logic: from observation x of the evidence we develop proposition y, and from that there follows deduction z. Each is a link in a chain of reasoned deduction. Neither archaeological observation nor archaeological deduction is usually secure with any real certainty: a lengthening chain of reasoning accumulates the weaknesses in its numerous links. Many a link is qualified with a 'probably' or 'in all likelihood', and many an unqualified statement – 'this is a midden', 'these are the attributes of a state-level society' – in reality carries large uncertainty. Suppose a link from x to y, so qualified or not, has a real certainty of 0.7; that is, it is true in 70% of cases and un- or uncertainly true in 30%. Suppose the next link, y to z, has the same 0.7 certainty. Then the overall certainty, from x to y to z, is 0.7 times 0.7 = 0.49 – 49%, and less than 50%. In just two steps of chained reasoning, we have reached an outcome which is more likely to be wrong than right! These rules, which apply to all reasoning by a sequence of deduction, are particularly telling for archaeology, where there is so often a weak coupling between what we are able to observe - 'what are the features of these chipped-stone pieces?' - and what we hope to study – 'did people move seasonally about the landscape, or was it divided into the territories of distinct groups?'

Against chains of reasoning, Wylie commends another approach, of 'cabling'. As a strong cable is made by combining together many threads, each one individually weak, so can a variety of forms of evidence, and individually weak deductions drawn from them, be tied and be pulled together, collectively to build areasonably secure knowledge. This is the established habit of Arnhem Land rock-art research.

Certainly, cabling is the right spirit in which to approach the dating of Arnhem Land art. Relevant to date are many distinct sources of information, or aspects to sources of information, and distinct ways of working with them. Some apply to very few figures; for only a handful is there known to be a direct memory of their making. Some apply to figures of an infrequent class, such as those that depend on certain characteristic subjects. Some that apply to many figures are weak, such as those that depend on the relative weathering of different colours of pigment. Yet, taken together, these each-fallible methods build a chronology for Arnhem Land art which is unusually detailed and unusually secure, by the standards of rock-art dating; they permit detailed studies of change over time, of a kind rarely possible with the materials of rock-art. Arnhem Land researchers now have grounds for sufficient confidence in rock-art chronology, that we see anomalies and contradictions as fruitful opportunities to advance knowledge by exploring how the anomaly comes about, rather than challenges that upset the dating framework.

Dating has been a central issue since considered study of Arnhem Land rock-art began (Spencer & Gillen 1928: 823—4; McCarthy 1958; 1960: 297—414). Brandl (1973: 171—178), working with the distinctions amongst the paintings made by Aboriginal people in their country, recognized an earlier period of 'Mimi' art (divided into early and late), and a transitional period leading into the X-ray conventions that are seen both on the painted rocks and on bark-paintings, on which he developed a chronology. Lewis (1988) drew on changing technology as depicted in the pictures for a finer chronology. Chaloupka (1977; 1984; 1985) made further divisions into distinct successive styles of depiction, as well as changing subject-matter, and linked changes in the art to environmental shifts, thereby tying the relative sequence towards absolute dates. More recently, Harris matrices have been used to analyse sequence in detail (Chippindale & Taçon 1993), and the first radiocarbon dates directly applicable to beeswax art have become available (Nelson et al. 1995).

Most potential sources of information for dating Arnhem Land art were noticed a generation ago by Brandl (1973: 171—178). Although there has been some disputation between researchers (see, for example, Lewis 1988: 6—13; Haskovec 1992a), behind the differences there is a good consistency in the patterns discerned by different researchers emphasizing each a certain aspect.

Even the casual tourist is made aware of the lively present-day meaning of Arnhem Land rock-art (for example, by Breeden and Wright 1989: 27ff.). Contemporary knowledge extends to an Aboriginal account of the history of Arnhem Land pict the land befothis survey with that indigenous understanding, of the pictures that refer to the Mimi people (Carroll 1977). Arnhem Landers know the story of their rock-painting this way: Before us in the land were the Mimi people, and it was the Mimi people who taught us how to paint. You can see paintings on high ceilings that no human can reach; the Mimi must have pulled the roof down to reach the surface or flown up so they could paint up there. The Mimi have gone as people now, but they are still there as spirits, thin as they can slide into cracks in the rocks. Clever people sometimes see them.

Arnhem Landers, naturally enough, have no fixed definition of what constitutes the Mimi art, as Brandl (1973: 165, 167, 172—178) notes. They may recognize it by a faded red colour, by strange habits of depiction, by obsolete weapons and implements, by ceremonial dress unknown today – that is, by features which set it apart from the imagery of present-day knowledge.

In our archaeological view of the paintings we see a matching distinction, recognizing a noticeable step in a long sequence between the archaic paintings – ones Arnhem Landers identify with the Mimi – and the more recent ones with conventions that Arnhem Landers recognize as their own. But this is not an abrupt or total break: there are continuities as well. (How different is different, how much the same is the same, when one is judging change through a long-term sequence?)

Recent art on bark and on paper from Gunbalanya (Oenpelli), a painting centre in the heart of the western Arnhem Land rock-art region, often depicts the Mimis (for example, paintings by GarryCC check spelling Djorlomorle, Peter Nabarlambarl, Djawida Nadjong in Dyer 1994: plates 30, 36, 38). In the contemporary way of painting them, Mimis are indeed narrow and thin like spirits that can slip into the cracks; but other present-day conventions of painting Mimi, in the treatment of the head and in the kind of spear-throwers they carry, come from the later traditions of the rock-art.

Since rock-art in Arnhem Land continues into recent times, and since its conventions are perpetuated in contemporary paintings on bark and on paper, there is memory and historical knowledge of paintings and painting conventions.

For now, the practice of rock painting has largely come to an end in western Arnhem Land. A number of pictures have been painted in the last 20 or so years, although often – it has been noted – in circumstances of non-Aboriginal people being present (or even at their prompting), and in that sense not painted in the traditional way. Still, it remains a remarkable fact that here one can go to rock-art with the painter, and hear directly and on the spot when and how a picture was made (Taçon 1992a; Taçon & Garde 1995).

Beyond these few are the paintings, still not numerous, of artists now passed away, but of which there is some record, particularly due to the work of Chaloupka, and thereby of known date. A beeswax composition of water-buffalo and buffalo-hunter with rifle and skinning knife in lower Deaf Adder Creek (Chaloupka 1984: figure 25) is known to have been made by Najombolmi in the early 1950s (George Chaloupka, pers. com.); at Nourlangie (Nawulandja) is a group of paintings, now the most celebrated in the region, by the same master artist, painted in 1964 (Chaloupka 1982: 22-25).

The Najombolmi paintings at Nourlangie have some distinctive conventions, within the elaborate X-ray manner of recent polychrome paintings, that one can see elsewhere. The bark painters have their own individual styles, of course, and with practice one can recognize the work of the individual artist. Bardayal Nadjamerrek paints his human figures always with a very large first toe (for example, West 1995: plates 7—8), and the rarrk (fine-line cross-hatching) of a painting follows an individual artist's habit. (One of us remembers feeling his eyes had begun to understand Arnhem Land painting, when he recognised an individual artist's rarrk on a piece glimpsed in a souvenir shop window and knew immediately who had painted it.) Since a distinct style is evident in the rock paintings known to be by Najombolmi, one can hope to recognise his own personal style in other figures, and this has been done (Haskovec & Sullivan 1989). But Chaloupka (1993: 238; pers. com.) and Lewis (pers. comm.) fear the style identified in that study with Najombolmi is not of a single individual but common also to his companions Djimongurr and Djorlom and/or others. Whatever the case, one can have confidence that figures with these distinctive traits are broadly of the period when this artist was active, since the others were his contemporaries.

European people and European things have been known in Arnhem Land since settlements were planted on the Cobourg Peninsula about 160 years ago. Depictions of European people and things (Fig. 6.2) (Chaloupka 1993: 193—205) are datable to that period onwards.

Guns are seen in some paintings (Fig. 6.3), and some are painted so exactly they can be identified as the Martini-Henry rifle, British military-surplus and the favourite buffalo-hunting weapon for much of this century (Chaloupka 1993: plate 226). At Cannon Hill, there are stencils of European steel axes, and detailed paintings of aircraft (the Qantas airline's 'Kangaroo Route', 1940s: Chaloupka 1993: plate 230) and of ships, one with the name of a lugger that traded into Oenpelli Mission. At a remote shelter on the plateau is a picture of the whole wharf at Darwin with its buildings and boats, well painted with much detail although some hundreds of kilometres away (Chaloupka 1993). The Europeans themselves are painted, with their hats and pipes, and without hands (in their trouser pockets!); and a human figure near the East Alligator River crossing is painted with a gun and long pigtails down his back in the Chinese manner (Chaloupka 1993: plate 222); he is surely one of the Chinese labourers brought in the late 19th century to Darwin, who worked the mines at Pine Creek to the south.

Domestic animals brought to the Cobourg settlements quickly wandered away into the bush. When Leichardt came through western Arnhem Land on his famous walk across Australia in 1845, western Arnhem Landers' knowledge of white men and of their animals was his first notice that he was close to the goal of the Port Essington settlement. So paintings of cattle, horses and water buffalo (Chaloupka 1993: plates 223—225) will post-date the Port Essington settlement of 1838—1849 or the earlier failed plantation at Raffles Bay (1827—1829), whose pigs, ponies, buffaloes and red cattle wandered the woods (Spillett 1972: 81, 91). Near the west bank of the East Alligator River is painted a line of horses, with riders and bells around their necks, who must be Europeans on their horses; Chaloupka (1979) identifies these as pictures of the McKinlay Expedition which lost itself in that region in 1865; or, they are others among the buffalo-shooters, miners and others who went on horseback through the country (Lewis prso. comm.).

No natural blue pigments are used in Aboriginal art but the artificial pigment of laundry blue was adopted for some rock-paintings (Chaloupka 1993: plates 74, 75) when it became available at some point in the 19th century.

For some centuries before the European settlement (Macknight 1976; Clarke 1994), Macassan fishermen camped on Arnhem Land shores to cure the trepang they had caught. Images of their boats and huts (Chaloupka 1993: plates 214, 215), are dated without exactness in that way.

Images of these historically dated subjects are not frequent in the art, but they are useful beyond their own interest. One can be certain the materials, conventions and manners of depiction applying here were active within the last century and a half. The 'McKinlay paintings' of the ridden horses are in red and somewhat faded; they are a useful proof that red paintings can be recent, and that they can blur and fade in a matter of about a century.

The European and Macassan subjects aside, there are many subjects and activities depicted which are clearly those of Arnhem Land hunter-gatherers from the time of contact onwards. It is reasonably supposed that many of these subjects were in use for many centuries before. Some can be specifically associated with landscape features of limited age, such as the freshwater swamps where magpie geese now flock and make a major food-source. Paintings of the distinctive fans made from the goose wings will be of the freshwater era, and the distinctive spear-thrower of Arnhem Land today (Lewis 1988: 53—55) will indicate that a painting may be 'new'.

Palaeo-ecological evidence (below) shows the freshwater swamps are some hundreds of years old (rather than decades or thousands). Images specifically linked to the freshwater swamps and life-ways will fall within that age-range.

Modern Aboriginal knowledge and continuity in painting – nowadays on bark and on paper more often than on rock – identifies the many and distinctive conventions in picture-making of recent times. Again, these show a rock-painting may be new without ruling out that it is old. Among these are the distinctive and elaborate X-ray conventions where the interior features of an animal subject are shown within the body. It can be hazarded that the 'simple X-ray' variant of the convention, in which an unelaborated empty space stands for the body cavity, or a single simple line stands for the gut, is a precursor of the elaborate convention – hence its other name, 'early X-ray' (Brandl 1973). Stratigraphic study of sequences in the paintings confirms that antiquity; and a radiocarbon date for a beeswax figure using the convention (below) now places a calendar age on it.

The several aspects to dating noted so far depend on historical knowledge of aspects of the pictures. It follows that aspects not known historically are likely not to be recent and have the potential to be old.

Equally, distinctive subjects seen in the paintings that are not seen in contact-era paintings, and are not used in Arnhem Land life-ways today, will be old.

A frequent subject is boomerangs, sometimes stencilled, sometimes painted, and in use as a hunting or a fighting weapon (Fig. 6.4). Yet Arnhem Landers today use boomerangs only as clap-sticks in music-making. As well as the spear-throwers of the common modern type, there is a simpler form, painted as a simple hooked stick (Fig. 6.4): this will be archaic – if it can separated from the variant spear-throwers known in modern times nad called by Cundy (189: 104) the 'North Australian cylindrical spearthrower'.

Many subjects may be of recent or of ancient date – dilly-bags, for example – and are therefore not evidence diagnostically.

An obvious group among these subjects is the several animals that are extinct in the region. Chief among these is the thylacine (Tasmanian tiger: Thylacinus), commonly seen in older art styles painted with striped flanks and a distinctive tail (Brandl 1972; Lewis 1988). The thylacine is supposed to have been extinguished in mainland Australia consequent on its being out-competed by dingoes; the coming of the dingo into Australia is uncertainly dated about 3500 b.p. (Corbett 1995: 14—17). Identified with less certainty are pictures of the long-nosed echidna, because it closely resembles the common, and commonly-painted, echidna; it is known today only in New Guinea, and its date of extinction – if it were in Arnhem Land – is unknown. The local disappearance of other sometimes-depicted creatures such as numbats is unknown. A bone of Tasmanian devil (Sarcophilus harrisii) dated after 3100 b.p. from the Paribari shelter (Schrire 1982: 53) is a reminder that extinctions may not be very ancient events. A clear and archaic painting of a strangecreatu re in Upper Deaf Adder Creek is identified without certainty as depicting Palorchestes, a species of the Australian Pleistocene megafauna (Murray and Chaloupka 1984), again with unknown date of regional extinction.

The same goes for conventions of picture-making and the manners in which things are depicted. Conventions that are not seen in paintings identifiable as of recent date by other characteristics will be archaic.

By their several common features, Arnhem Land researchers can establish paintings that together make a distinctive group (called by Chaloupka 'styles') – Dynamic figures, Northern Running Figures, and so on. Often – as befits a refined art tradition of classical characteristics, particular traits are so distinctive one comes to recognize them with confidence even from a worn scrap on a ruined surface; the S-curve in the body of a Northern Running Figure, or the bobbles, trailers and tassels of Yam Figures are unmistakable. The energy of the Dynamic Figures, well named by Chaloupka, is such that even a static coiled snake, painted in that manner, appears dynamic to the eye.

The first means of establishing the sequence of these styles is provided by those many surfaces in which paintings lie one over the other, often so densely the rock surface itself is hidden and little patches of covered-over paintings are spied under and between those that lie on top (e.g. Yuwengayay: Walsh 1988: plate 314). We have some, but not complete, confidence that one can often discern on careful examination which line lies over which; therefore the order in which the paintings were placed can be deduced. It is harder to be sure when the more transparent yellow pigment is involved, when one line is thick and sharp and the other broad and indistinct, when the lines are old or faded, or when the paint of two lines is of similar hue. We are encouraged by the cautious confidence of workers in Baja California, where figures are in faded red ochre on sandstone as the older Arnhem land figures are, when they attempt similar observations (Stanley Price pers. comm.). It is valuable when multiple observations can be made at a single panel, and when a pattern of sequence is repeated at several panels, so an erroneous judgement as to any one superposition is not fatal.

From his observations at two sites on Mount Gilruth, Chaloupka (1977) was able to discern a stratigraphic sequence of styles (observations there since contested by Haskovec 1992a). A detailed study of a panel at Mount Brockman (137 figures) and another at Kungurrul (172 figures), which have many paintings with distinctive traits in some reasonably secure visible superposition, has confirmed essentials of the Gilruth sequence (Chippindale & Taçon 1993). The Kungurrul site, in particular, is valuable in including more styles, and therefore providing a fuller stratigraphy which may now have superseded those seen at Mount Gilruth.

One odd class of figures is the marks of 'thrown objects' seen generally on high ceilings in irregular lines and splotches of colour, and thought to have been made by throwing objects such as pieces of bush twine soaked in ochre. These, in high places all on their own, are not in useful superimposition in relation to other figures; they do not find a place in the sequence as observed stratigraphically.

Rarely, a more certain kind of sequence is provided by a rock-fall which covers or hides a painted surface; it may at the same time create a new surface which can be painted. Then the figures on the 'lost' surface must be later than those on the 'made' surface, while both may be co-eval with pictures on the unaffected surfaces adjacent.

The palaeo-ecology and archaeology of Arnhem Land provides well-dated sequences relevant to the art sequence in the same region. We have no direct links between them. The ecological changes have transformed the Arnhem Land environment without, so far as is known, making surfaces newly available for painting. Although many of the Arnhem Land archaeological sites are in shelters with paintings, the archaeological deposits do not include or cover painted surfaces – as if the water in the acid sands carries the paint off the wall once it is buried. Instead, one is obliged and able to 'bridge' across from aspects of the art to aspects of the palaeo-ecology and archaeology of Arnhem Land that share common features, together to make a history of Arnhem Land in landscape, stone and paint. Taçon & Brockwell (1995), a detailed study in this approach for the period 15,000—7,000 BP, could be extended to a fuller time-scan.

The region's environmental history divides into three distinct periods. The post-glacial rising sea flooded across the broad and flat Arafura Plain to the north, converting it into the shallow Arafura sea, and reaching about the position of the present shore-line about 6000 BP (Chappell & Grindrod 1983: 67-9, 87-8); this is the pre-estuarine period. There followed an era when the lowland ecology was dominated by the marine presence, and it is likely there were intensely saline flats, as nowadays persist in some lower reaches of the East Alligator River (Woodroffe et al. 1986; 1987; Woodroffe 1988); this is the estuarine period. During the estuarine, sand-bars and mangrove thickets grew up in the river systems, to the point that they blocked tidal access by sea-water. The salt water was gradually washed out by the wet-season flow of storm-water, and the upper portions of the estuarine systems were converted into fresh-water swamps, an enormously rich and productive resource: this is the freshwater period. This transformation is evident in the archaeological sequences from the sites near rivers; after a period without shell (pre-estuarine), marine shells appear at a date corresponding to the arrival of the sea (estuarine), and give way to freshwater species at the time of freshwater conversion (freshwater) (e.g. Schrire 1982 in respect of sites near the lower East Alligator River; Allen & Barton 1989: 101—102).

Images in the rock art reflect the same changes: the dominant animals, and likely animal-food resource, in the old styles are the several varieties of macropod. There are fish, but it is noticeable that the salmon-tailed catfish, tolerant of difficult conditions, is frequent. In the later paintings, fish dominate over macropods, and the dominant fish are the species of the big swamps and of the freshwater rivers – commonly fork-tail catfish, barramundi and saratoga (Taçon 1988). The change in rock-art subjects, following the ecological sequence, is most evident in the northern sites that are close to the modern wetlands, such as Cannon Hill and Ubirr. It is less apparent or not visible at all in the upland sites on the southern plateau, a region above the escarpment not directly affected by the transformation of the low landscape downstream.

Accordingly, Chaloupka's scheme divides the art into four phases:

• pre-estuarine to about 6000 BP;

• estuarine about 6000—1200 BP;

• freshwater about 1200—100 BP; and a recent

• contact period from the mid-late 19th century up to the present.

The contact period is distinguished by depictions of European people, subjects and introduced creatures and by other markers of a modern date (but ecologically part of the freshwater).

Just as one can bridge from the art to the environmental sequence, so one can bridge from the art to the archaeological sequence. Difficult soil conditions mean that the archaeology at most sites consists only of stone ar tifacts until the last several hundred years (e.g. Ngarradj Warde Djobkeng: Allen & Barton 1989). And stone artifacts are not frequent in the art – a plain reminder of how small a proportion of an elaborate material culture survives in the defining archaeological record. The recognizable stone artifact commonly seen in the paintings is the ground stone axe (e.g. Taçon & Brockwell 1995: figures 8, 9), hafted with a withy passed round the stone head and bound together into a handle with gum/resin and bush-twine just as it always has been in Arnhem Land in recent times, and as it is so depicted in contemporary bark and paper paintings (e.g. paintings by Ralph Nganjmirra, Danny Djorlom and Garry Djorlom in Dyer 1994: plates 16, 27, 30). But the stone axe has a long history in Arnhem Land; its occurrences in sites near the East Alligator was many years ago shown to extend back to some 20,000 years (White 1967), and there are rotted fragments of igneous rock likely to be used for axe-making in old strata at Nauwalabila (Jones & Johnson 1985). So bridging across from axes in the pictures to axes in the archaeology – which can be reliably done for the hafted axe as a distinctive subject – gives only slight indication of date (Lewis 1988: 46).

Recently recognized from a single painted example in the art is another stone artifact, the hafted chisel, introduced about 4000 BP, and therefore providing a sharper marker of date (Taçon & Brockwell 1995: 680).

A certain number of Arnhem Land sites have both an art and an archaeological component. Many art sites have no archaeological component because they are well above the ground surface, or because the rock at their base is washed clean by wet-season storms. Where paintings extend down towards the modern ground surface, they appear to vanish as they reach the earth; it is likely the soil conditions do not permit the pigments to persist on the rock. Nor are engravings known in the region in a direct stratigraphic relation to archaeological deposits. So we have no direct stratigraphical links between art and archaeology.

A repeated feature of deep Arnhem Land archaeological sites is the occurrence of ochre, consistently from the earliest levels for which these are not small and stray pieces, but of good quality and – at Nauwalabila – shaped with distinct facets from a level dating to about 12,000 BP (Taçon & Brockwell 1995: 687). Does ochre mean painting, and therefore the rock art goes back in Arnhem Land to the very beginning, with two sites now dated by luminescence as having sequences that begin some 50,000 to 60,000 years ago? Is Arnhem Land art therefore 'the world's longest continuing art tradition' (Chaloupka 1993)? As well as rock-art, ochre has many uses in modern Aboriginal ceremony, and is in repeated association with burial from the later Palaeolithic elsewhere in the world.

In Arnhem Land, there is no certainty either that ochre was used for painting from the beginning; or that painting with ochre was on rock surfaces (rather than on perishable subjects); or that the first paintings on rock surfaces are amongst the ones that survive. It is equally clear that a demonstrated potential for rock art from that first settlement is shown by the presence of ochre.

At sites towards the Vistoria River art province, beyond the southwest limits of Arnhem Land, excavated sequences show ochre appearing abruptly and in large quantities in strata dating from around 1400 years ago onwards (David et al. 1994.). The ochre is there reasonably linked with the large ochre figures painted on the walls of the same shelters, without its being directly demonstrated that the ochre in the ground is the self-same stuff as the material on the walls. In Arnhem Land itself, no excavated site has provided or hinted at such a direct link, because nowhere is there a grand mass of ochre in the strata that can confidently be linkedto a matching mass of ochre on the rock-surface above; analysis of the sequences does show 'pulses' in ochre presence in the deposits, and these might correlate with new styles (Taçon & Brockwell 1995).

Principles of continuity

In resolving the sequence of different distinctive 'styles' – whether defined by subject, by manner of depiction, or by both aspects in combination, one can fairly expect that a particular feature will have been depicted over some continuous time-span – rather than coming into the pictures, disappearing, and then recurring; therefore those styles with common features should fall adjacent to each other in the chronological sequence.

A subject illustrating this is the boomerang, depicted in the distinct painting styles defined by Chaloupka as 'Dynamic Figure', 'Post-Dynamic Figure', 'Simple Figure', and 'Northern Running Figure'. All of these, it is likely, will form a coherent group occupying a distinct block of time, rather than the boomerang's having entered, left and re-entered the repertoire of subjects (see Jones & Johnson 1985: 218, for a smiliar argument in respect of axes). A priori, one could expect also that the stencils of boomerangs will be of broadly the same era as the painted group.

In distinctive aspects to manners of depiction, the same expectation of continuity will apply as it does to what is depicted. In the depiction of how boomerangs are held in the hand, for example, one sees two distinct conventions. In the one, found in paintings of the 'Dynamic' manner, the boomerangs – commonly two or three – are commonly drawn held together and crossing at the hand; they are commonly of much the same length. In the other, found in figures of the 'Post-Dynamic' and 'Simple' manners, the two or three boomerangs are commonly drawn concentric, as if nested one inside the other and with the inner ones shorter. It is reasonable to expect these last two manners to be adjacent (or contemporary) within the sequence.

A puzzle in current knowledge of the Arnhem Land sequence is an apparent contradiction concerning archaic pictures of large animals, commonly macropods; these are the distinctive and defining element of Chaloupka's 'Large Naturalistic' style, which he sets early in the sequence. Our field observations of superposition – since Lewis's (1988), Haskovec's (1992a) and Chippindale & Taçon's (1993) notice of the anomaly – lead us to conclude that closely similar subjects and manners of depiction exist in both that early phase and in the later 'Yam Figure'/'Simple Figure' phase; between these, in our current understanding, is the Dynamic Figure style, where a different and distinctive manner of depicting animals is evident. Here, it appears, the continuity principle does not hold.

Chemical analysis of paint from recent figures found its organic component to be nil (Clarke & North 1991). Painters of early modern times, it is remembered, used to use the juice of an orchid root as a binder, but one does not know if it is preserved in the paint – apparently not, if those studies are relied on (and there is no reason to question them).

Experience in direct dating of organics from painted surfaces in Australia and elsewhere has been mixed. Sometimes plausible dates have been obtained, sometimes dates are erratic for the same figures; usually, it is not proven that the organic carbon being dated in fact relates to the painting of the image.

It remains to be demonstrated, therefore, whether Arnhem Land paintings can directly be dated by the radiocarbon method. Indeed, with a well-dated sequence established by other methods, it may be that Arnhem Land will have a special value as a region to check novel radiocarbon studies by reference to paintings of known age.

A small component in the Arnhem Land repertoire, especially in the region of the lower East Alligator River, is figures made by moulding wax from the nests of the native Trigona bee. The soft wax is kneaded into pellets or strips or sheets and applied to the rock in varied designs and images. Analytical study proves what one would suppose, that beeswax is a good material for carbon dating, made by the bees in a matter of not many months, collected and used for art soon after. Chemically, it is complex and stable, and a pilot programme of carbon dating has shown that the beeswax figures are up to the remarkable age of some 4000 years (Nelson et al. 1995). The majority, however, are at most a few hundred years old; radiocarbon dates for these, with their large error factors and unhelpful calibration curves, give a reliable but inexact date.

Where dated beeswax figures are stratified over or under paint, a date for the beeswax figures gives a minimum or a maximum age for the painted figure.

For the most part, the repertoire of beeswax forms is special to the medium rather than shared with paint; the lines and arrays of dots that are the most common motif are vary rare in paint. Where characteristics are shared, the beeswax date is pertinent: the 4000-year-old beeswax figure is of a turtle drawn in the simple x-ray convention common in paint. It gives a certain date when that convention was in use.

Inspection of single figures and of larger surfaces shows that weathering of the art surfaces is very variable, both between surfaces and within any one surface;. Often one can see how: paint is worn away, where, for instance, water runs down otherwise protected surfaces. Sometimes one can see no such cause, and one remembers that water-runs will change their course as a shelter erodes.

In an approximate way then, one can expect a weak rule that crisp and distinct images will be newer than worn and indistinct paint. Along with that pattern goes the different behaviour of the three common colours – red, yellow and white – in the art.

The red ochre, haematite, appears enduring; it is well said that, closely examined, the colour of ancient paintings is not on the rock but in the rock. Even paintings annually exposed to wash and water survive crisply when 'in the rock; some surfaces that support early painting in red – like the celebrated emu-hunter panel at Mount Brockman (Chippindale & Taçon 1993) – are not well protected from sun and rain. A distinctive feature of some of the most archaic red paintings is the turning of the pigment to a bluer or pinker hue, aptly called 'mulberry' by Walsh (1994) who finds the same tone in archaic red-ochre paintings of north Western Australia.

The yellow ochre, limonite, is more often seen washing down a rock surface. Over the long term of many hundreds of years it is not chemically stable, turning slowly to haematite (Davidson19??). A few of the Dynamic Figures, certain to be ancient, are in yellow while most are in red; most of these few yellow Dynamics are not in the clear 'chrome yellow' of recent figures, but are turning a more butterscotch tone.

The white, generally a kaolin-type clay (Clarke & North 1991: 84), makes a thicker pigment, which one can see sits very much out on the rock rather than passing into the surface; it does not seem to endure. A telling illustration is a solid picture of a wallaby painted by Bill Miyarki at Koongarra in 1972 (illustrated in Edwards 1979: 130) on an unprotected surface and over ancient red paint. Not a trace of the white pipe-clay from 1972 is seen on the surface today, but the ancient red figure it was painted over remains.

In general, it can be said that red is enduring, yellow transient and white fugitive as pigments on the surfaces. In Northern Running Figures (Haskovec 1992b), otherwise complete pictures are missing portions of the whole subject: for instance, a figure with well-drawn arms will have no hands. In these cases, it seems the red only has endured, and it is conjectured that the white or yellow component of the full bi- or poly-chrome composition is lost.

The figures made of beeswax (below) show a distinctive weathering sequence, at first dark and shiny, they crack, fissure become paler, as they are reduced to a white brittle skin which then disappears. Exposure to ultra-violet light, which weakens the chemical bonding of the wax, is the likely cause; when a single figure spreads over surfaces with differing exposure to light, its state of preservation varies accordingly.

The rare engravings in western Arnhem Land (Chaloupka 1993: 234—237) are most found heavily patinated, sometimes with a red mineral skin, sometimes with an uncoloured or dark skin. They are regarded as ancient.

Commonly, Arnhem Land paintings are on patination or mineral skins covering the exposed rock-surface; sometimes skins cover paintings (Hughes & Watchman 1983; Watchman 1985; 1990). Preliminary radiocarbon studies of deposited salts in stratigraphic relation to rock-art gave determinations of up to 8000 years for a multi-layered crust on a surface with paintings (Watchman 1987). The chemistry and carbon-dating of rock skins and crusts, a difficult and a developing field, has yet decisively to contribute to dating in Arnhem Land.

Many painted shelters are frequented by mud-wasps, whose nests become robustly fixed on to the surface. There are nests over paint, and nests under paint. Luminescent studies of the mineral component as ameans of dating mudwasp nests, and thereby to date art in a stratigraphic relationship to it, has been reported from the Kimberley region of Western Australia (Roberts et al. 1997); it has potential for Arnhem Land.

We have reported on these approaches to dating Arnhem Land rock-art:

• Aboriginal knowledge of the antiquity of Arnhem Land paintings.

• Dating by reference to historical knowledge:

• Memory and historical record of artists and their styles;

• Subjects depicted, and pigments used, of known historical date;

• Distinctive subjects in relation to contact-era and recent Arnhem Land hunter-gatherer life-ways;

• Distinctive manners of depiction in relation to contact-era and recent Arnhem Land painting.

• Dating by reference to lack of historical knowledge:

• Distinctive subjects not known in contact-era and recent Arnhem Land hunter-gatherer life-ways;

• Distinctive manners of depiction not known in contact-era and recent Arnhem Land painting.

• Stratigraphic sequences on the painted surfaces.

• Bridging to dated palaeo-ecological and archaeological evidence:

• Bridging across to the palaeo-ecological record;

• Bridging across to the archaeological record;

• Art materials within archaeological deposits.

• Principles of continuity.

• Direct dating of images by radiocarbon (and implications therefrom by superposition, by subject depicted, and by manner of depiction).

• Weathering.

• Direct dating of skins and other natural features in stratigraphic relation to art.

Some of the sub-divisions of these categories may run into each other: memory and historical record of artists and their styles runs into the distinctive manners of depiction in relation to contact-era and recent Arnhem Land painting; the distinctive subjects of known historical date shade into those of contact-era and recent hunter-gatherer life-ways; and these in turn, run into the sources of the palaeo-ecological. We think it convenient to keep them a little distinct, as each category of evidence makes a different contribution.

Other lines of reasoning can be brought to bear. Chaloupka (1993: 89) places the paint imprints of thrown objects (Chaloupka 1993: plate 79), grass-prints (thought to be made by striking paint-loaded grass stems against the surface) (Chaloupka 1984: figure 4), and hand-prints (Chaloupka 1993: plate 80) early in the sequence, in the expectation that prints – as direct images taken by imprinting from the object itself – may be earlier than the more developed artistry of painting.

It was acknowledged above that the sources for dating vary in their applicability and in their strength. Their independence compensates here, when it is consistently found that the several lines of evidence are coherent with each other.

Chronology for the animated images of the Dynamic Figures, for example, depends on the following lines of evidence:

• Aboriginal knowledge they are Mimi paintings;

• Distinctive subjects not known in contact-era and recent Arnhem Land hunter-gatherer life-ways;

• Distinctive manners of depiction not known in contact-era and recent Arnhem Land painting.

• Stratigraphic sequences on the painted surfaces.

• Bridging across to the archaeological record.

• Principles of continuity.

• Weathering.

Chronology for the beeswax figures depends on the following:

• Aboriginal knowledge of the figures.

• Memory and historical record of artists and their styles.

• Subjects depicted of known historical date;

• Distinctive manners of depiction in relation to contact-era and recent Arnhem Land painting.

• Stratigraphic sequences of beeswax figures in relation to other beeswax figures, and in relation to paint.

• Principles of continuity.

• Direct dating of images by radiocarbon.

• Weathering.

In short, the dating of Arnhem Land rock-art illustrates Wylie's 'cabling', and the way in which varied lines of independent evidence – 7 for one of these groups, 8 for the other – can be brought together into a single coherent and strong strand (Fig. 6.5).

It is consistently found that the deductions made from each line of evidence are consistent with each other. It follows that when they are not consistent, one can have confidence that a real anomaly exists, rather than there being gross fault in the dating evidence. One example of this has been noted above, the discrepancy in the archaic large depictions of animal subjects, which leads us to think that there were two distinct periods when these subjects were portrayed: they occur as older 'Large Naturalistic' and as 'Yam/Single' figures. By using clues as to which of the collected group may be late and which early, it may be possible to recognize distinctive features of subject or of manner of depiction peculiar to each, and thereby to divide the group into its two components. Stratigraphic study has shown another anomaly: 'Yam' style figures are stratified over the 'Simple' style figures (regarded as an earlier style) – but 'Simple' style figures are also stratified over 'Yam' style figures (both successions are clearly seen on a key panel on Upper Twin Falls Creek: unpublished field observations). Accordingly, there is an overlap between the styles, corresponding to a length of time when both manners of depiction were in use; this is consistent with other evidence the repertoire of spears and early spear-throwers found with figures of each 'style' appears to be the same). In considering the two styles, we are encouraged to explore a relationship between them other than a chronological sequence: we find that Simple figures depict activities that appear to belong to the human domain (e.g. scenes of humans fighting), while Yam figures depict activities in a spiritual domain (e.g. images of Rainbow Serpents).

Again, this seems to us to be a strength, rather than a weakness of the research approach to Arnhem Land art as it has developed. Unstated in much of the Arnhem Land literature is the fair starting premiss that at any one period, a coherent range of subjects would be depicted with a coherent manner of depiction – the entities combined in the idea of a 'style', and the expectations that the styles will form a chronological series. This is the common starting-point for rock-art studies everywhere. Nevertheless, recent Arnhem Land rock-art demonstrates a more complex reality: as well as the intricate X-ray depictions in polychrome colours for which modern Arnhem Land art is celebrated, there are the stick-figures of the 'Energetic' style and many other elements which taken altogether make up Taçon's 'Complete figure style' (1992: 204—205). Unpublished preliminary results from the programme of radiocarbon dating for beeswax figures show the same finding: the variability evident in that repertoire of forms is not a simple function of changing time. If dating were precarious, or depended on a single line of evidence, this would be disconcerting. With a robust framework for dating and multiple lines of evidence, one can instead proceed with the working premiss that variability is primarily a function of time, and the expectation of a sequence of distinctive styles. When that premiss is contradicted, as it is for the large archaic animal images, for the Yam/Simple figures, and for the beeswax figures, then one can use that anomaly to move from the simplifying premiss towards knowledge of a more complex reality.

Figure 5, reproduced from Taçon & Chippindale 1994: table 1, sets out our then understanding of the major elements in Arnhem Land rock-art chronology. Our view of it has moved on a little since 1994, and other workers in the fast-moving field each have their own slightly different schemes (e.g. Lewis 1988; Chaloupka 1993; Haskovec 1992a). The fundamentals of the main features are largely held in common.

Arnhem Land is unusual among rock-art regions in Australia in our detailed grasp of its chronology. A consistency beyond the immediate region is provided by neighbouring regions of the Northern Territory. In the Katherine region to the south, at Dead Man Pocket near the west coast of the Northern Territory, in the Victoria River District, and at Keep River on the border with Western Australia a similar fundamental sequence is found (Lewis 1984): an archaic class of figures in red (sometimes mulberry), and a recent strand of polychrome figures well known to Aboriginal people of the regions today. On a larger scale, the same broad pattern seems to apply to the Kimberley region of Western Australia and to northern Queensland, united with the 'Top End' of the Northern Territory in a grand province (Morwood & Hobbs 1995b) of similarity across the north of the Continent. Morwood & Hobbs (1995a), a study of first importance from north Queensland, bridges across between the archaeological and the art records to develop an integrated picture, following the model of Arnhem Land studies.

The fair chronology that now exists for Arnhem Land enables varied study of change over time: Chaloupka (1993) is a good broad survey of changing subjects depicted and of changing manners of depiction (1993). There are a few, and could be many, specialist studies to explore specific themes: as well as regional variants, in aspects specific to a period (e.g. Taçon 1989 on recent rock-art), in changing technology (e.g. Lewis 1988 on spear-throwers), in changing social relations (e.g. Taçon & Chippindale 1994 on fighting), in the changing sense of images (e.g. Wilson et al. in press on Rainbow Serpents).

What these have in common is a capacity to study change and variation over time, that defining purpose of archaeology which is – unhappily – often elusive in rock-art research. Instead of a whole repertoire of images being treated as a single group without regard to change over time (because there is slight evidence for chronology), or some grasp of chronology being the goal of study, it becomes possible to track change. Importantly in Arnhem Land, this means one can deduce – on a basis of fairly secure dating – an understanding of archaic aspects not accessible by the informed knowledge of present-day and recent times. Only in that way is it possible to work well and fairly with that varied range of information offered by formal and by informed methods. And it means one can look beyond the changes visible in the imagery to other transformations – in landscape and in settlement – to which changing imagery may relate.

One can contrast that with the present state of research in southern Africa (Dowson & Lewis-Williams 1994), where ethnohistorical sources provide informed information, formal methods are available, but an effective chronology remains lacking, despite varied studies (e.g. in that collection Deacon 1994; Hall 1994; Loubser & Laurens 1994; Morris & Beaumont 1994; Walker 1994; Whitley & Annegarn 1994; Yates et al. 1994; see Thackeray 1983 for an earlier survey). A late date for some panels is proved by contact-era figures with European subjects (Campbell 1987), and a late date has long been thought likely for those many panels which seem similar in style and date and preservation (Lewis-Williams 1981: 24). The possibility of a very early date is given by the charcoal, ochre and white images from the Apollo 11 cave, Namibia, dated to over 26,000 years (Wendt 1976). The ethnohistorical records – though they are accounts from San people who did not themselves paint – have clear relevance to recent paintings. But without a developed chronology in between, interpretation is 'flat', lacking time-depth, or the means to study how picture-making has changed over time. One believes, or does not believe, that recent San knowledge informs all the paintings, rather than having the means to develop a secure knowledge.

We work in Arnhem Land with the consent of its Aboriginal people, who we thank for that permission and for their welcome into their country, as expressed through the Gagudju and Jawoyn Associations. We thank in particular those individuals concerned with granting that permission, and those who went with us as companions in the field. We are grateful to the Australian Nature and Conservation Agency (ANCA/ANPWS) and the Northern Land Council for permits. We thank colleagues, friends, acquaintances and strangers for assistance in and after the field, and for energetic comments on a draft of this paper

Allen, H. & G. Barton. 1989. Ngarradj Warde Djobkeng: White Cockatoo Dreaming and the prehistory of Kakadu. Sydney (NSW): Oceania Publications. Oceania Monograph 37.

Berndt, R. M. & C. H. Berndt. 1954. Arnhem Land: its history and people. Melbourne (Vic): F.W. Cheshire.

Brandl, E.J. 1968. Aboriginal rock designs in beeswax and description of cave painting sites in western Arnhem Land, Archaeology and Physical Anthropology in Oceania 3(1): 19–29.

Brandl, E.J. 1972. Thylacine designs in Arnhem Land rock paintings, Archaeology and Physical Anthropology in Oceania 7(1): 24—30.

Brandl, E.J. 1973. Australian Aboriginal paintings in western and central Arnhem Land: temporal sequences and elements of style in Cadell River and Deaf Adder Creek art. Canberra (ACT): Australian Institute of Aboriginal Studies. Australian Aboriginal Studies 52, Prehistory and Material Culture series 9.

Breeden, S. & B. Wright. 1989. Kakadu: looking after the country – the Gagudju way. East Roseville (NSW): Simon & Schuster.

Campbell, C. 1987. Art in crisis: contact period rock art in the south-eastern mountains of southern Africa. Master's dissertation, University of Witwatersrand.

Carroll, P. J. 1977. Mimi from western Arnhem Land, in P. Ucko (ed.), Form in indigenous art: schematisation in the art of Aboriginal Australia and prehistoric Europe: 119—130. Canberra (ACT): Australian Institute of Aboriginal Studies.

Chaloupka, G. 1977. Aspects of the chronology and schematisation of the prehistoric sites on the Arnhem Land plateau, in P. Ucko (ed.), Form in indigenous art: schematisation in the art of Aboriginal Australia and prehistoric Europe: 243–59. Canberra (ACT): Australian Institute of Aboriginal Studies.

1979. Pack-bells on the rock face: Aboriginal paintings of European contact in north-western Arnhem Land, Aboriginal History 3(2): 92–95.

1982. Burrunguy, Nourlangie Rock. Darwin (NT): Northart.

1984. From palaeoart to casual paintings. Darwin (NT): Northern Territory Museum of Arts and Sciences. Monograph 1.

1985. Chronological sequence of Arnhem Land plateau rock art, in Rhys Jones (ed.), Archaeological research in Kakadu National Park: 269–80. Canberra (ACT): Australia National Parks and Wildlife Service. Special Publication 13.

1989. Groote Eylandt archipelago rock art survey 1988. Report to the Heritage Branch, Conservation Commission of the Northern Territory. Darwin (NT): Northern Territory Museum. 'Limited edition manuscript'. [ref. from Chaloupka 1993]

1992. Rock art survey: Arnhem Land plateau – central region. Report to the Heritage Branch, Conservation Commission of the Northern Territory. Darwin (NT): Northern Territory Museum. 'Limited edition manuscript'.

1993. Journey in time: the world's longest continuing art tradition. Chatswood (NSW): Reed.

Chaloupka, G., N. Kapirigi, B. Nayidji & G. Namingum. 1985. A cultural survey of Balawurru, Deaf Adder Creek, Amarrkananga, Cannon Hill and the Northern Corridor. Darwin: Museum and Art Galleries Board of the Northern Territory: report to Australian National Parks and Wildlife Service. Unpublished report. Darwin (NT): Northern Territory Museum of Arts and Sciences.

Chappell, J., & A. Grindrod. 1983. 7+2KA 'spike', in J. Chappell & A. Grindrod (ed.), CLIMANZ: 87–8. Canberra (ACT): Department of Biogeography and Geomorphology, Research School of Pacific Studies, Australian National University.

Chippindale, C., & P.S.C. Taçon. 1993. Two old painted panels from Kakadu: variation and sequence in Arnhem Land rock art, in J. Steinbring et al. (ed.), Time and space: dating and spatial considerations in rock art research (papers of Symposia F and E, AURA Congress Cairns 1992): 32—56. Melbourne (Vic): Australian Rock Art Research Association. Occasional AURA Publication 8.

Clarke, A. 1994. Winds of change: an archaeology of contact in the Groote Eylandt archipelago, Northern Territory. Ph.D thesis, Australian National University, Canberra.

Clarke, J. & N. North. 1991. Pigment composition of post-estuarine rock art in Kakadu National Park, in C. Pearson & B.K. Swartz, Jr (ed.), Rock art and posterity: conserving, managing and recording rock art: 80–87. Melbourne (Vic): Australian Rock Art Research Association. Occasional Publication 4.

Corbett, L. 1995. The dingo in Australia and Asia, University of New South Wales Press, Sydney.

Cundy, B.

David, B., I. McNiven, V. Attenbrow & J. Flood. 1994. Of Lightning Brothers and White Cockatoos: dating the antiquity of signifying systems in the Northern Territory, Australia, Antiquity 68: 241—251.

Davidson, I. 19??. details at proof

Deacon, J. 1994. Rock engravings and the folklore of Bleek & Lloyd's /Xam San informants, in Thomas A. Dowson & David Lewis-Williams (ed.), Contested images: diversity in southern African rock art research: 237—256. Johannesburg: Witwatersrand University Press.

Dowson, T. A. & D. Lewis-Williams (ed.). 1994. Contested images: diversity in southern African rock art research. Johannesburg: Witwatersrand University Press.

Dyer, C. A. (ed.). 1994. Kunwinjku art from Injalak 1991—1992: the John W. Kluge Commission. North Adelaide (SA): Museum Art International.

Edwards, R. 1979. Australian Aboriginal art: the art of the Alligator Rivers region, Northern Territory. Canberra (ACT): Australian Institute of Aboriginal Studies.

Hall, S. 1994. Images of interaction: rock art and sequence in the eastern Cape, in T. A. Dowson & D. Lewis-Williams (ed.), Contested images: diversity in southern African rock art research: 61—82. Johannesburg: Witwatersrand University Press.

Haskovec, I.P. 1992a. Mt Gilruth revisited, Archaeology in Oceania 27: 61–74.

1992b. Northern Running Figures of Kakadu National Park: a study of a regional style, in J. McDonald & I. P. Haskovec (ed.), State of the art: regional rock art studies in Australia and Melanesia: 148—158. Melbourne (Vic.): Australian Rock Art Research Association. Occasional Publication 6.

Haskovec, I. & H. Sullivan. 1989. Najombolmi: reflections and rejections of an Aboriginal artist, in H. Morphy (ed.), Animals into art: 57–74. London: Unwin Hyman.

Hughes, P.J. & A. Watchman. 1983. The deterioration, conservation and management of rock art sites in the Kakadu National Park, in D. Gillespie (ed.). 1983. The rock art sites of Kakadu National Park: 37–82. Canberra (ACT): Australia National Parks and Wildlife Service. Special Publication 10.

Jones, R., & I. Johnson. 1985. Deaf Adder Gorge: Lindner Site, Nauwalabila I, in R. Jones (ed.), Archaeological research in Kakadu National Park: 165—227. Canberra (ACT): Australia National Parks and Wildlife Service. Special Publication 13.

Lewis, D. J. 1984. Mimi on Bradhaw. Australian Aboriginal Studies 2: 58—61.

The rock paintings of Arnhem Land, Australia: social, ecological and material culture change in the Post-Glacial period. Oxford: British Archaeological Reports. International Series S415.

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

Loubser, J. & G. Laurens. 1994. Depictions of domestic ungulates and shields: hunter/gatherers and agro-pastoralists in the Caledon River, in T. A. Dowson & D. Lewis-Williams (ed.), Contested images: diversity in southern African rock art research: 83—118. Johannesburg: Witwatersrand University Press.

Macknight, C.C. 1976. The voyage to Marege: Macassan trepangers in northern Australia. Melbourne (Vic): Melbourne University Press.

McCarthy, F.D. 1958. Australian Aboriginal rock art. Sydney: Australian Museum.

1960. The cave paintings of Groote Eylandt and Chasm Island, in C.P. Mountford (ed.) , Records of the American–Australian scientific expedition to Arnhem Land 2: Anthropology and nutrition: 297–414. Melbourne (Vic): Melbourne University Press.

Morris, D. & P. Beaumont. 1994. Portable rock engravings at Springbokoog and the archaeological contexts of rock art of the Upper Karoo, in T. A. Dowson & D. Lewis-Williams (ed.), Contested images: diversity in southern African rock art research: 11—28. Johannesburg: Witwatersrand University Press.

Morwood, M.J. & D.R. Hobbs. 1995a. (ed.). Quinkan prehistory: the archaeology of Aboriginal art in SE Cape York Peninsula, Australia. Brisbane: Anthropology Museum, University of Queensland. Tempus 3.

1995b. Themes in the prehistory of tropical Australia, in Jim Allen & James F. O'Connell (ed.), Transitions: Pleistocene to Holocene in Australia & Papua New Guinea: 747—768. Antiquity 69 (Special number 265).

Murray, P. & G. Chaloupka. 1984. The Dreamtime animals: extnct megafauna iun Arnhem Land rock art, Archaeology in Oceania 19: 105—116.

Nelson, D. E., G. Chaloupka, C. Chippindale, M. S. Alderson & J. Southon. 1995. Radiocarbon dates for beeswax figures in the prehistoric rock art of northern Australia, Archaeometry 37(1): 151156.

Roberts, R. G. 1996. Preliminary investigations using optically-stimulated luminescence to date fossil mud-wasp nests associated with rock pictures in the Kimberley. Unpublished paper presented at the First Workshop on Australian Rock Picture Dating, Lucas Heights (Sydney), February.

Roberts, R.G., R. Jones, N.A. Spooner, M.J. Head, A.S. Murray & M.A. Smith. 1994. The human colonisation of Australia: optical dates of 53,000 and 60,000 years bracket human arrival at Deaf Adder Gorge, Northern Territory, Quaternary Geochronology, Quaternary Science Reviews 13: 575—583.

Roberts, R.G., et al. 1997. details at proof - in press Nature ??.

Schrire, C. 1982. The Alligator Rivers: prehistory and ecology in western Arnhem Land. Canberra (ACT): Department of Prehistory, Research School of Pacific Studies, Australian National University. Terra Australis 7.

Spencer, W. Baldwin, & F.J. Gillen. 1928. Wanderings in wild Australia. London: Macmillan.

Spillett, P.G. 1972. Forsaken settlement: an illustrated history of the settlement of Victoria, Port Essington, North Australia 1838–1849. Dee Why West (NSW): Lansdowne.

Sullivan, H. 1988. Rock engravings in Kakadu National Park. Unpublished paper, First AURA Congress, Darwin, August.

Taçon, P. S. C. 1988. Identifying fish species in the recent rock paintings of western Arnhem Land, Rock Art Research 5(1): 3—15.

1989. From Rainbow Snakes to 'x-ray' fish: the nature of the recent rock painting tradition in western Arnhem Land, Australia. Unpublished Ph.D thesis, Australian National University, Canberra.

1992a. The last rock painters of Kakadu, Australian Natural History 23(11): 866–73.

1992b. Somewhere over the rainbow: an ethnographic and archaeological analysis of recent rock paintings of western Arnhem Land, Australia, in J. McDonald & I. P. Haskovec (ed.), State of the art: regional rock art studies in Australia and Melanesia: 202—215. Melbourne (Vic.): Australian Rock Art Research Association. Occasional Publication 6.

Taçon, P. S. C. & S. Brockwell. 1995. Arnhem Land prehistory in landscape, stone and paint, in J. Allen & J. F. O'Connell (ed.), Transitions: Pleistocene to Holocene in Australia & Papua New Guinea: 676—695. Antiquity special number 265.

Taçon, P. S. C. & C. Chippindale. 1994. Australia's ancient warriors: changing depictions of fighting in the rock art of Arnhem Land, N.T., Cambridge Archaeological Journal 4(2): 211—248.

Taçon, P. S. C. & M. Garde. 1995. Kun-wardde bim, rock art from western and central Arnhem Land, in M. West, Rainbow sugarbag and moon: two artists of the stone country: Bardayal Nadjamerrek and Mick Kubarkku: 30—36. Darwin (NT): Museum & Art Gallery of the Northern Territory.

Taçon, P. S. C., M. Wilson & C. Chippindale. 1996. Birth of the Rainbow Serpent in Arnhem Land rock-art and oral history, Archaeology in Oceania 31:103—124.

Thackeray, A. I. 1983. Dating the rock art of southern Africa, in New approaches to southern African rock art: 21—26. Cape Town: South African Archaeological Society. Goodwin Series 4.

Walker, N. 1994. Painting and ceremonial activity in the Later Stone Age of the Matopos, Zimbabwe, in T. A. Dowson & D. Lewis-Williams (ed.), Contested images: diversity in southern African rock art research: 119—130. Johannesburg: Witwatersrand University Press.

Walsh, G. L. 1988. Australia's greatest rock art. Bathurst (NSW): E.J. Brill – Robert Brown.

1994. Bradshaws: ancient rock paintings of north-west Australia. Carouge-Geneva: Edition Limitée.

Watchman, A. 1979. Summary and interpretation of petrological, geochemical and mineralogical studies, in P.J. Hughes, A. Watchman & D. Gillespie, The deterioration, conservation and management of rock art sites in the Kakadu National Park, NT. Report to Australian National Parks and Wildlife Service.

1985. Mineralogical analysis of silica skins covering rock art, in Rhys Jones (ed.). 1985. Archaeological research in Kakadu National Park.: 281–289. Canberra (ACT): Australia National Parks and Wildlife Service. Special Publication 13.

1987. Preliminary determinations of the age and composition of mineral salts on rock art surfaces in the Kakadu National Park, in W. Ambrose & J. Mummery (ed.). 1987. Archaeometry: further Australasian studies: 36–42. Canberra (ACT): Australian National University, Department of Prehistory, Research School of Pacific Studies.

1990. A summary of occurrences of oxalate-rich crusts in Australia, Rock Art Research 7(1): 44–50.

Wendt, W. E. 1976. 'Art mobilier' from the Apollo 11 Cave, South West Africa: Africa's oldest dated works of art, South African Archaeological Bulletin 31: 5—11.

West, M. 1995. Rainbow sugarbag and moon: two artists of the stone country: Bardayal Nadjamerrek and Mick Kubarkku. Darwin (NT): Museum & Art Gallery of the Northern Territory.

White, C. 1967. Early stone axes in Arnhem Land, Antiquity 41: 147–52.

Whitley, D. S. & H. J. Annegarn. 1994. Cation-ratio dating of rock engravings from Klipfontein, northern Cape, in T. A. Dowson & D. Lewis-Williams (ed.), Contested images: diversity in southern African rock art research: 189—197. Johannesburg: Witwatersrand University Press.

Woodroffe, C. D. 1988. Changing mangrove and wetland habitats over the last 8000 years, northern Australia and southeast Asia, in Deborah Wade-Marshall& Peter Loveday (ed.), Northern Australia: progress and prospects 2: Floodplains research: 1–33. Darwin (NT): North Australia Research Unit, Australian National University

Woodroffe, C. D., J. Chappell, B. G. Thom & E. Wallensky. 1986. Geomorphological dynamics and evolution of the South Alligator tidal river and plains, Northern Territory. Canberra (ACT): Australian National University, North Australian Research Unit. Monograph 3.

Woodroffe, C. D., B. G. Thom, J. Chappell, E. Wallensky, J. Grindrod & J. Head. 1987. Relative sea level in the South Alligator River region, north Australia, during the Holocene, Search 18: 92–94.

Wylie, A. 1989. Archaeological cables and tacking: the implications of practice for Bernstein's `Options beyond objectivism and relativism', Philosophy of the Social Sciences 19: 1—18.

Yates, R., A. Manhire & J. Parkington. 1994. Rock painting and history on the south-western Cape, in T. A. Dowson & D. Lewis-Williams (ed.), Contested images: diversity in southern African rock art research: 29—60. Johannesburg: Witwatersrand University Press.

Its north-west limit is Van Diemen Gulf, the bay within the Arafura Sea into which flow the three Alligator Rivers. Along their courses are the present-day areas of wetlands.

To the south and east is the 'stone country', often edged by a sheer escarpment.

Within the area of woodland between low wetlands and high stone country are scattered rocky islands and outliers.

Rock-art may be found anywhere offering suitable protected surfaces on the rock. There are concentrations where outliers, such as Nawulandja, Mount Brockman and Ubirr, are close to the resources of the wetlands.

The polychrome paint, and X-ray technique in the manner of depiction, are characteristic of the Post-Estuarine, The large fishes are typical of the Freshwater period. The human figure, centre lower, with tobacco-pipe, rifle and Chinese hair-style places that image into the contact era.

The steel axe, subject of a white stencil and then painted in red, is diagnostically of the contact period; the boat with its name, can be identified to the years she was plying the coast and its rivers. Hand- and arm-stencils, as a subject, exist in earlier and in later periods; here, the fugitive white pigment, shows they may be presumed recent.

These Dynamic figures are characteristic in the 'mulberry' colour of pigment which sets 'in' rather than 'on' the rock surface (and reproduces in monichrome as a mid-grey), in the fine technique of 'brush-painting', in the Dynamic manner of depiction, and in aspects of the subject-matter (notice the boomerangs held by the human figure below and by the animal-headed being above).

These figures, in Upper Deaf Adder Creek, were unusually (and valuably for study of chronology) amended by the addition in a thicker line and different, darker-coloured pigment of the simple spear-thrower – and object and art-subject not present until a later period.

We anticipate work in progress will lead to revision.

From Taçon & Chippindale 1994.