Cupule engravings from Jinmium-Granilpi (northern Australia) and beyond: exploration of a widespread and enigmatic class of rock markings.

Tacon, Paul S.C. ; Fullagar, Richard ; Ouzman, Sven 等

ANTIQUITY last year reported a startlingly old series of dates from Jinmium in tropical north Australia. At Jinmium are old rock-engravings, the pecked cups or cupules that are widespread in Australia. This study of the Jinmium cupules goes beyond that immediate topic to broader issues.

Visual art, in all its manifestations, is a fundamental characteristic of fully modern humans, one of many that express cultural identity. Evidence for this key aspect of human symbolic communication figures in 'rock-art' - paintings and engravings on blocks or expanses of stone. Rock-art differs from portable art forms in that the prominent locations where rock-art was placed may provide clues about prehistoric symbolic landscape. Debate continues as to whether Homo erectus and other human ancestors, or close relatives, such as the Neanderthal, made and used some form of art, and how that art differs from that made by Homo sapiens sapiens (Anati 1993; Bednarik 1994; 1995; Chase & Dibble 1987; Davidson & Noble 1989; Lindy & Clark 1990; Marshack 1997; Mellars 1989; 1991; Noble & Davidson 1996; Pfeiffer 1982). What is clear is that cognitively modern humans have, from the beginning, had a desire and perhaps a need to mark and transform landscapes into cultural places or localities enriched with symbolic meaning (Tacon 1994). Today, we are not able to ascertain the specific meanings of the earliest cultural marks but we can study rock-art, its contexts, organizing principles and structures (Conkey 1987). When a particular form of art is found across an extensive area we can gain insight into relationships people had with large tracts of land - landscapes, territories and sometimes regions.

Alongside recent discussion about the sacred nature of landscapes for the world's indigenous peoples, it has been noted that many rock-art sites are located at unusual, special places (Ouzman 1995; in press; Tacon 1990; 1994). What makes some places, locations and natural landscapes sacred or special? Why are some more sacred or special than others? Why did people choose to mark some places, to the apparent exclusion of others, with meaningful designs? These important questions, underlying many archaeological investigations into ancient landscapes, seldom are addressed directly. And how can we, as outsiders, define what is or is not sacred or special for contemporary peoples, let alone archaeologically observed groups? We best work by applying broad theoretical constructs to specific, sustained regional research initiatives that utilise many strands of archaeological evidence (see, for example, the detailed research recently published for northern Australia, such as Chaloupka 1993; Chippindale & Tacon 1993; David et al. 1994; Haskovec 1992; Lewis 1988; Lewis & Rose 1988; McNickle 1991; 1993; Mulvaney 1996; Tacon 1991; Tacon & Brockwell 1995; Tacon & Chippindale 1994; Walsh 1994; Welch 1990; 1993).

Drawing on this regional, as well as international, archaeological research, we begin to address questions relating to human perceptions, constructions and markings of landscapes. We start by focusing on sites with large clusters of that unusual image class which comprises engraved pits or cup-shaped marks, 'cupules' (definition below). These seem the most ancient surviving evidence of symbolic activity from central Arnhem Land through to the Pilbara region of northern Australia, set most commonly on boulders or on rock-shelter walls (Chaloupka 1993; Flood 1997: 145-9; Tacon & Chippindale 1994; McNickle 1991;1993; Walsh 1996). They promise both information and insight into the country's ancient cultures. We ground our research in the Keep River region of the Northern Territory, including the Jinmium site complex (Fullagar et al. 1996) and Granilpi, a second complex near by.

Cupules, peck marks, pits and grinding hollows

A variety of rounded engraved pits, holes, hollows and cup-like forms on the walls and floors of rock shelters, boulders and large, fiat slabs of sandstone have been identified within Australian rock-art traditions; some called 'cupules' are recognized as the oldest surviving form of rock-art in northern Australia (Chaloupka 1993; Flood 1996; Flood 1997: 145-9; Tacon & Chippindale 1994; Walsh 1994; Welch 1992). True cupules are cup-shaped non-utilitarian marks (Flood 1997: 145-6). These pecked and pounded circular depressions are not to be confused with the grinding hollows, often larger and on horizontal surfaces, that resulted from food or ochre processing. Thousands of cupules were arranged to cover vertical surfaces in orderly, tightly packed rows; these could not be the by-products of non-symbolic, purely practical activities. Cupules are also not to be confused with individual peck marks that often were used to infill engraved animal designs of more recent periods, or with smaller pits less than 10 mm in diameter, sometimes found in small groups on horizontal surfaces at sites (often with starlike designs that resulted from pounding and gouging with a sharp tool). Cupules are much more circular, uniform in shape, larger, deeper and numerous. They appear to have been made with a more rounded pounding tool, although evidence of such tools is rare.

Cupules were not used to define shaped subjects, such as pictures of animals or humans. Rather, they tend to relate to the form of the surface into which they were pecked and/or abraded, often conforming to or accentuating natural cracks, crevices, joints and boundaries. Rarely a symmetrical geometric shape is defined with an orderly arrangement of a dozen or more cupules. For the most part, the cupule arrangements seem abstract in appearance and enigmatic in meaning. We classify cupules as a form of 'rock-art', in contrast to Rosenfeld (1992), Flood (1997; in press) who would argue they are a form of 'rock-mark' that includes such things as hand stencils, incised and abraded grooves, battered or pounded rims, and so forth, made under socially different circumstances. Although accepting their arguments, we do not wish to make such fine discriminations until we better understand the context of cupule production, the structure of cupule sites and landscapes and the specific nature of cupule patterning.

Comparison with grinding hollows

Grinding hollows occur across Australia on bedrock, shelter floors, boulders and small, flat stones (as in other parts of the world: e.g. Cooke 1964; Parkman 1995). Whether few or numerous at a site they have an unorganized pattern of distribution best explained as a function of their being a practical or comfortable place to work. Grinding hollows, generally much larger and deeper than cupules, were made by abrading and the grinding of various vegetable foods, fat, ochre or other substances. With this distinction in mind, we recorded a significant grinding hollow complex (KR14) on the edge of the Marralam swamp, located between Jinmium and Granilpi [ILLUSTRATION FOR FIGURE 1 OMITTED], to establish objective criteria that will assist in distinguishing horizontally placed cupules from true grinding hollows. We know there are true grinding hollows at KR14 because Aboriginal people had used the area for that purpose until quite recently, and the hollows resemble others known to be used for grinding activities in Kakadu National Park and elsewhere.

KR14 is a rock shelter, 9 m wide at the drip line, with engraved boulders in and near its entrance, some hollows on its sandstone floor, and remnants of painted rock-art on the walls. Thirty-three hollows are distributed across the shelter floor, some on the more steeply sloping areas towards the back, others on the flatter area near the front; there was no clustering as we find at the cupule sites in the area. These marks averaged just over 85 mm in diameter and almost 11 mm in depth [ILLUSTRATION FOR FIGURE 2 OMITTED].

Near the shelter entrance were 23 hollow-marked boulders, with another 4 further away to the north, the furthest 177 m north and 60 m east of the shelter. In these:

a The hollows appear less weathered and have less patina/desert varnish than cupules on boulders elsewhere.

b The boulders had 1-9 hollows each, with an average of 3-4.

c The average diameter of hollows on any one particular boulder is 77-150 mm, and for the 96 hollows just over 110 mm; about 61% are over 100 mm (compare 58 mm for cupules at Granilpi site KR20 cupules and 63-65-5 mm elsewhere: see below). Depths are 1.5-36 mm, average 12 mm. These average diameters and depths are much greater than those of wall or boulder cupules. This is in keeping with studies in North America where Parkman (1995: 1) notes that on horizontal surfaces cupules are 'depressions having a diameter of 10 cm [4 inches] or less and a depth of 4 cm [1 1/2 inches] or less'.

d The average number of hollows per boulder is smaller, 3.5 versus just over 30 cupules for Granilpi sites KR6-9 and just over 12 for KR20 (below).

In summary, grinding hollows are large, abrasion-formed rather than pecked and pounded, lack much patina or desert varnish, and are of more recent, less weathered appearance than cupules. They are usually on horizontal surfaces, sometimes on slightly sloping surfaces. Grinding hollows have unsystematic distribution patterns at particular locations, and the average number per boulder is low.

Cupules placed on boulders are on horizontal, vertical and slightly or greatly sloping surfaces, sometimes all at once. They are smaller, and appear older and more weathered. They have structured distribution patterns with high average numbers per boulder. Most were made by pecking and pounding, a few by grinding.

Antiquity and distribution of cupules

World- wide distribution

Although cupules have a world-wide distribution, they were made at different times in different places for surely different reasons (Bednarik 1993: 139; Steinbring & Lanteigne 1991 is an alternative). In Europe, cupules (distinct from but possibly related to cup-and-ring art) are a frequent motif at hundreds of sites scattered across Scandinavia (Glob 1969; Sognnes 1995; Walderhaug 1995), the United Kingdom (Bradley 1991, 1995; Morris 1970; Piggott 1973; Steinbring & Lanteigne 1991; van Hoek 1997; Walker 1970; 1977), Ireland (Jackson 1995; Johnson 1991; 1993), Spain (Costas Gorberna et al. 1993/94; van Hoek 1997) and elsewhere; Estonian rockart consists exclusively of 1500 cup-marked stones (Poikalainen 1995: 338). Some sites have also been found in France (Germond 1980; Guirand 1964; 1970), where Europe's oldest rock-art consisting of cupules can be found (Bednarik 1993: 138). Here, at La Ferrassie (Peyrony 1934), a triangular limestone rock slab containing 18 cupules, mostly arranged in pairs, was apparently placed over a burial. The slab is undated; the burial of a Neanderthal child may be over 50,000 years of age (see Bednarik 1995:610 for sketch). Pecked marks have been found at other late Mousterian sites (e.g. Leonardi 1988); again, firm dates are not available.

The north European material is much more recent, a form of 'farmer's rock art' (Sognnes 1995: 134) dating to no more than 3500 years of age. In the Sogn og Fjordane region of western Norway over 85% of rock-art sites consist of cupules, found in the sub-alpine regions of inner fjord districts (Walderhaug 1995: 17071). Mandt (1995: 265-6) proposes they are related to mountain summer farms, while Larsson (1989: 343) suggests they resulted from family and individual rituals (see also Henschen-Nyman 1982). Bradley (1991; 1995) and, later, van Hoek (1997) have shown an intimate relationship between cupule clusters and the larger landscapes of Britain and Ireland. Johnson (1993: 145) notes that cupules form 55-4% of motifs at Irish sites. For Scotland, Bradley (1995: 121) also argues that 'it was during the currency of rock art that large areas of the uplands were colonized'.

Cupule sites are scattered throughout the Americas. For North America, Parkman (1992; 1995) has recently put forward a solid argument that the oldest form of rock art consists of a 'pitted boulder tradition', common to the region's west, possibly over 9000 years old. Grant (1967: 26-7, 152) reported 'pit-and-groove' marks amongst the oldest surviving rock-art motifs both in eastern North America (1967: 140) and in Alberta (1967: 27,131). In California, cupule arrangements resembling north Australian sites are often found on large outcrops or boulders and in rock-shelters or caves; they are the most abundant type of pecked rocks (Grant 1967: 106). Keyser (1992) reports extensive but rare cupule sites on the Columbia plateau. Hedges (1993) notes an association between cupules and the remarkable 'ringing rocks' of three regions: southern California, the American Southwest and the far west of the USA. In northern California these were used by indigenous groups in rain-making and fertility ceremonies in recent times (Grant 1967: 106). Further south, cupules are reported from Mexico (Bednarik 1993), other parts of Mesoamerica (Gay 1973), Colombia (Hornell 1925), Peru (Thiermann 1977), Bolivia (Querejazu Lewis 1990), Brazil (Bednarik 1993), Argentina (Podesta et al. 1991) and Chile (Breton 1910).

In Asia cupules have primarily been reported from the Middle East (Ahlstroem 1978; Grebenart & Pierret 1966; Wreschner 1976), northern Russia (Shumkin 1991), China (Fu 1992: 369) and India (Bednarik et al. 1991: 34; Kumar 1996; Sharma et al. 1992). Bednarik (1994; 1995) argues that the oldest rock-art in the world comes from Auditorium Cave, central India, where a cupule and an adjacent meandering line are on a buried boulder lying in Acheulian deposit.

The most convincing and impressive cupule site from India, with seemingly great antiquity, was recently recorded by Kumar (1996: 38): 'With nearly 500 cupules of very archaic character, Daraki-Chattan is the richest Palaeolithic pure cupule site so far discovered in India.' As in Australia, the cupules cover walls, in this case two facing walls that make a narrow passage. Unlike north Australian cupules, they are not orderly or tightly packed; irregular distribution is a feature of early Indian cupules (Kumar 1996: 44). The cupules have no absolute dates, Kumar places the oldest in phase IA of a three-phase petroglyph chronology; the most recent stone tools from within and outside the cave are characteristic of the Middle Palaeolithic. Tentatively categorizing the cupules into six types, based on size ranges within separate clusters, he speculates they could have been made at six different times.

African cupule sites include those of Kenya reported by Odak (1992), central African sites noted by Forsbrooke (1954), a few Namibian sites (Viereck & Rudner 1957), rare Zambian (Chaplin 1963) and Zimbabwean (Clark 1958; Cooke 1964; Dart 1953; Swan 1996; Walker 1987) cupules, dots and ovals. Swan (1996) notes differences between cupules and iron-ore grinding hollows at some sites in Zimbabwe. Clark (1958) dated an archaeological deposit that partially covered the vertically positioned Chifubwa Stream rock-shelter cupules to 6310[+ or -]250 b.p. (1958: 21-2) a noteworthy parallel to Jinmium, Australia. Also in Zimbabwe, Robinson (1958: 76) has noted an association between cupules and rock gongs. South African cupules (Fock 1969; Geldmacher 1967; Schoonraad 1960; Tobias 1967) are sometimes associated with engraved human footprints. Botswana has a fascinating hill complex, Tsodilo, with numerous cupule sites, often occurring without other rockart (Campbell et al. 1995; Nick Walker pets. comm.; Rudner 1965). In Kenya, Odak interprets cup-mark sites as representations of celestial phenomena (1992: 58-9), rejecting the local indigenous idea that they might have been used in the bao game, played with pebbles, beads or large seeds; identifying four types of cup-mark arrangements, he suggests they are representations of planets, stars and galaxies.

In the Pacific region, the engraved rock art of Vanuatu is full of cupules (Garanger 1972; Hebert 1965; MacDonald 1899; Spriggs & Mumford 1992: 133), cupules are widespread in New Caledonia (Frimigacci & Monnin 1980; Monnin 1986) and there are a number of Papua New Guinea sites with pits, ovals or cups (Leask 1943; Bob McDonald pers. comm. 1996; Rosenfold 1986: 12; Williams 1931: 127; for New Ireland see Gunn 1986), some with numerous orderly rows on walls, reminiscent of north Australian sites. Cupules are also reported from Solomon Island sites (Roe 1992), Easter Island (Lee 1992) and Hawaii (Cox & Stasack 1970; Steinbring & Steinbring 1983).

Australian cupules

The most extensive and impressive Australian cupule sites lie in the far north, from Arnhem Land in the Northern Territory to the Kimberley region of Western Australia, with some sites in the Pilbara, Tanami desert and Barkly tablelands possibly related. Robert Edwards (1979: 96), early to notice the antiquity of cupules in western Arnhem Land, said 'Aborigines associated with this area have no knowledge of their origin.' George Chaloupka (1993: 235) further adds:

Pecked engravings and those made by a combination of pecked and abraded techniques seem to be of greater antiquity than those that are simply abraded, a supposition supported by their patinated appearance. By far the most common of these motifs are pecked hollows. They are usually grouped in quite large numbers on vertical walls, and occasionally also on ledges and boulders, and are found even in the hardest surfaces. The number of pecked hollows found in a particular panel, their depth and their diameter vary from site to site, but within each panel the hollows are quite uniform. The most extensive panel of these hollows is found in the Yuwunggayi shelter and is covered by several layers of painted images.

Tacon & Chippindale (1994: 215), after extensive study of superimpositioning of different art forms at over 650 sites across Kakadu and Arnhem Land, concur with Chaloupka (1993): cupules are the oldest surviving form of rock art, while pigment from deposits dated by thermoluminescence and optically stimulated luminescence (Roberts et al. 1990; 1993) may have been used for some form of painted art contemporary with or older than cupules which has since weathered away. The eastern-most examples of cupule sites, near the Mann River in central Arnhem Land, are still regarded as important to local Aboriginal peoples who relate them to Green Plum Dreaming ceremonies (Garde 1994: 4; Tacon 1993: 66), a rare instance of contemporary Aboriginal tradition relating to cupules.

To the west, David Welch (1992; 1993: 101) and Grahame Walsh (1994), working separately in the Kimberley region, have remarked on the widespread distribution and antiquity of cupules, agreeing they are the earliest in their regional sequences. Walsh, defining a 'Pecked Cupule' period, recognizes a 'pecked pit' group (pecking technique used exclusively) and a 'pebraded cup' group (pecking followed by abrading) as the oldest of all forms and styles of Kimberley rock art (1994: 33-5):

Most Kimberley cupules are now found in very hard sandstone surfaces, and assuming that this was the same at the time of their creation, each individual cupule would have required a considerable amount of time and energy. The commitment involved in some concentrations of the cupules would indeed be daunting. However, I believe the cupule creation predates the metamorphose [sic] process which altered the sandstone surfaces from their original workable composition, and this antiquity is evidenced by the often amazing skins which have since formed over them. . . .

Pecked Pit wall panels are on occasions found superimposed with paintings, but no instance shows them superimposing any form of art. Examples are occasionally found extending below floor deposit levels in art sites, which, together with their weathering and total patination, suggests considerable antiquity.

Walsh (1994: 34) also notes that the more significant cupule panels are 'most commonly found in larger shelters, with such sites frequently having prominent positions or commanding views', a pattern Bednarik (1993) finds not only in Australia but also for comparable sites overseas. They cluster, we would add, near the major river systems of north Australia. Most are no more than a few kilometres from the key drainage systems, suggesting these rivers and their tributaries may have been travel routes when the cupules were made, or that their placement was connected to water in some other, perhaps symbolic, way (see Richards 1996).

Between Arnhem Land and the Kimberley, McNickle (1991; 1993) reported cupule sites from the Keep and Victoria River Districts and comments of Keep River region sites (1993: 39):

Sites with clusters of deep pits are also fairly common in this region, but they differ from those to the east, such as at Coolibah. There, the pits are usually of a uniform size, and frequently located on vertical rear walls of shelters, whereas the pit clusters in the Keep River region are almost always found on the upper surfaces of boulders, or on rock floors in shelters. There is also a greater variety in the size of the pits and in their distribution, they often exceed a diameter of 10 cm. The configuration of the pits renders it unlikely that they are a by-product of grinding of seeds or other foods. At some sites, large oval but shallow depressions or polished patches are located alongside the circular pits. These would appear to be grinding hollows, suggesting that the pits were created for another purpose.

McNickle (1991: 44) and Walsh (1994: 33-5) see that cupules, unlike most rock-art types, are not limited to particular rock forms; they are common on hard and on soft sandstone, and even on granite. Many cupule boulders in the Pilbara region, southwest of the Kimberley, are granite (McNickle 1984; 1985: Wright 1968) and some cupules are even on metamorphosed volcanic rock (Wright pers. comm. 1996). Cupules are also found within shelters of the Tanami Desert formed by granite boulders (Graham & Mulvaney 1995).

A recent review of the earliest rock art of Australia by Josephine Flood (1996) places cupules or 'pecked pits' near the beginning of extant Australian rock art: 'Pecked pits are found in early contexts in Arnhem Land, Cape York Peninsula and the Kimberley, and occur throughout the Australian continent, including Tasmania' (1996: 29; also Flood 1987 for Cape York cupules, especially plate 4; David & Chant 1995: 485 for Chillagoe, north Queensland; Brown 1991:102-3 for Tasmania). The groups of dots and pits typical of 'Panaramitee' rock engravings (Maynard 1976; 1979; Franklin 1991) in central and southern Australia vary from northern cupule clusters, in design and arrangement; some also vary in age. And Flood (1996: 29) points out, cupules 'occur on recent rock surfaces and continue to the ethnographic present'. We have also observed this, and seen how parts of some sites have been re-marked, enhanced and re-engraved at different times; the most recent, at Jinmium (Keep River, Northern Territory), appear to have been retouched with a very sharp blade, probably a metal hatchet. We believe there is also good evidence that the oldest surviving cupules are many tens of thousands of years old.

Keep River region sites

Jinmium site complex

The Jinmium complex consists of a number of eroded sandstone outcrops lying on a forested, sandy plain near the Keep River in the Northern Territory [ILLUSTRATION FOR FIGURE 1 OMITTED]. A Dreaming Track connects it to a number of sites, geological features and quarries, including a second complex of rock-art sites at Granilpi, approximately 30 km to the north. The Jinmium sites are dominated by two shelters (KR1 and KR2) with walls covered with cupules; on some sculpted or pitted boulders, clusters or rows of cupules lead up to or accentuate natural rock features and the occasional geometric motif is made up of cupules. There are a total of 12 sites containing cupules, while other forms of engraved art, some paintings and stencils can also be found at these and other near-by locations. Jinmium was first described by Fullagar et al. (1996) as part of an investigation into Aboriginal resource management and relations to land (also Head 1994a; 1994b; Head & Fullagar 1991; in press). The site has much contemporary Aboriginal significance (1996: 754-5):

A dreaming story connects the Jinmium site to locations with important economic and ceremonial resources including swamp foods, ochre, stone and yams. Jinmium is a female ancestor-being, once pursued by Djibigun, a male ancestor-being. The site area is defined by Mr Paddy Carlton, a senior Gajerrong man and traditional custodian, and Mrs Biddy Simon, a senior Murinpatha woman and traditional custodian, in terms of proximity to the main outcrop and sandstone stacks 20m to 40m high. In the story, Jinmium turned to stone at this location where Djibigun catches her and turns himself into a small bird, associated with two near-by ochre sources. Historic and prehistoric Aboriginal campsites on the sandy ground are indicated by glass, stone and other artefacts over an area in excess of 5 hectares.

Fullagar et al. (1996) excavated one rock shelter, (C1), with a cupule-covered wall surface ([ILLUSTRATION FOR FIGURE 3 OMITTED] - KR1). The site, almost 50 m from the main outcrop, has a floor area measuring c. 24 sq. m. Some of the cupules extend below this floor, and there is a second shelter with a cupule-covered wall c. 8 m to the immediate west. KR1 faces north while its counterpart (KR2) faces south; together they present an imposing sight, a matching sculpted pair.

The cupules cover 9.5 sq. m of the shelter wall at KR1 and 7 sq. m at KR2, excluding cupules below ground level. We estimate c. 3500 cupules at KR1 and c. 3200 at KR2:KR1 has a density of 372 cupules per sq. m. while KR2 has 463-5 cupules per sq. m (estimates from counting four 50x50-cm representative squares). The size of cupules varies slightly: at KR1 13-52 mm in diameter, average 31 mm; at KR2, 10-50 mm, average 28 mm. Depths vary, up to 11 mm at KR1 (5 mm average), up to just over 9 mm at KR2 (3 mm average). Some of this variation may be linked to re-pecking; the most recent (less patinated) cupules are confined to KR1.

The cupules do not suggest particular shapes or figures; they appear to be placed so as to conform with natural changes in the rock surface. In both shelters they extend up from below the present ground level to natural cracks, ridges and other boundaries near where the shelter walls change to sloping ceilings. Only a few isolated cupules or small clusters are found above these natural dividing lines. In general, the cupules increase in height above ground level and towards the centre of each shelter, to make a semi-elliptical shape. On the side of the KR1 outcrop, at least 13 cupules were arranged to form a geometric shape [ILLUSTRATION FOR FIGURE 4 OMITTED]. At the back of KR1 a weathered macropod, engraved in the peck-infill technique, is superimposed on top of several engraved cupules [ILLUSTRATION FOR FIGURE 5 OMITTED]. Two red hand stencils are over a row of at least 15 cupules at KR1, with other hand stencils above, but not superimposed on, cupules at KR2. Eleven cupules found on the flat top of KR2 are considerably larger than their vertically placed counterparts: 58-97 mm in diameter, average 79 mm; up to 20 mm depths, average 11 mm. Although these are larger, we do not think they result from grinding; they are considerably smaller than grinding hollows at near-by food-processing sites (above) and were made by pecking rather than abrading/grinding the surface.

Amongst the eroded stacks and blocks of sandstone in the main Jinmium sacred site complex is site KR4, where cupules are arranged on either side of a natural, narrow tunnel or passageway through the rock, just large enough to squeeze through [ILLUSTRATION FOR FIGURE 6 OMITTED]. Here, most cupules were placed between the horizontal and vertical surface, on the sloping portion of the rock. On the eastern side, a double row of cupules broadens into a triple row which runs for 1.57 m toward the natural, narrow tunnel. From the west, rows of cupules appear to lead toward the tunnel for 13.2 m, with the main concentration on a sloping and vertical surface 1.65x0.20 m long, situated across from the eastern cupule branch. The two rows of cupules meet near a natural crack in the rock that runs down from the tunnel. A single cupule lies above them, next to the crack, and 5 more cupules are on a small horizontal platform on the other side of the tunnel, through the rock. These - larger, deeper and more noticeable - average 71 mm in diameter rather than the 28 mm of the cupule rows, and 13 mm in depth rather than 2-3 mm. No cupules were found in the tunnel itself, an observation consistent with comparable sites at Granilpi (below).

At KR4 the cupules were arranged in such a way that they draw the viewer towards the natural hole that leads through the sandstone outcrop. The rows of cupules lead one from either direction and, where they converge, one immediately notices the natural crack leading to the tunnel. A single cupule focuses one's eyes upward, through the tunnel to the last cluster of cupules on the other side. In this way the cupule-makers succeed in taking the visitor through the rock, physically or otherwise; it appears the site was meant to be 'entered' from the north. The KR4 cupules - arranged by natural features and boundaries - suggest conceptions, uses and experiments with marking and movement through space.

Other Jinmium sites include KR5, with abraded grooves, grinding hollows and some cupules on the floor. On the wall above the grinding hollows is a row of cupules, an area of pit-marks, a line of peck-marks and two bounded rectangles of peck-marks. There are hand stencils high up at the western end of the shelter. The row of cupules is above and parallel to a prominent crack in the rock. The crack has a very weathered appearance and almost certainly precedes the cupules; we suggest it was the crack that attracted the cupules and defined the extent of their placement.

Further sites within the vicinity of the main Jinmium complex include a third large shelter with a cupule-covered wall about 3 km away (KR21) and an unusual site with cupules on boulders, walls and horizontal surfaces that accentuate a natural tunnel (KR24), a few hundred metres from KR1. These features, consistent with the other sites, show that the patterns of cupule arrangements and relationships to natural features of landscapes we initially documented are widespread.

KR21 consists of a large overhang with a weathered south-facing, red-brown wall, the centre covered with cupules (average diameter 29 mm). There are larger cupules on the floor (average diameter 71 mm, except for one small patch with an average diameter of 30 mm) and a slab of stone with 5 large grinding hollows (average diameter 150 mm). In the middle of the shelter are the remains of 3 white hand stencils just above the uppermost cupule. A solid white turtle, on the eastern wall, is the only distinct figurative motif painted at the site. On the western side, two rows of cupules appear much more recent than the rest. Also on this side, there is a row of 8 closely spaced, elongated cupules. The main cupule-covered wall surface appears very old, with thick skins and much weathering in some places. Overall, the site gives an appearance of great antiquity. There are similarities with Jinmium sites KR1 and KR2: the curving cupule-covered wall, the height of the engraved surface extending to 1.7 m above ground, and the upper limit following a natural bend and seam in the wall.

KR24 [ILLUSTRATION FOR FIGURES 7 & 8 OMITTED] is an eroded block of sandstone resting on the sandy plain, south of the main sandstone complex. A large, natural tunnel through its base runs roughly north-south, high enough to walk through. The main support pillar, on the eastern side, is covered right the way round with small cupules (average diameter 31 mm), comparable in size to those on Jinmium wall panels. More cover the inner wall, to the west. Rows of cupules lie on a sloping and horizontal surface that runs up to and through the passageway, next to the inner eastern wall. On the sandy floor of the shelter lie two large boulders covered with sizeable cupules (average diameter 60 mm), comparable to those of the Granilpi sites 30 km distant. At least 1385 cupules, patches of peck-marks, 6 hand stencils and 5 red hand prints mark the site. At KR24 the three key, repeated elements are brought together: cupule-covered walls, sculpted boulders, rows that accentuate natural passageways through rock.

Some of these themes are repeated at other Jinmium cupules sites (KR25, KR26, KR29, KR30, KR31, KR33), especially the highlighting of natural passages.

Granilpi site complex

Apart from large wall panels, these features occur at Granilpi, a site complex dominated by cupule-covered boulders, 30 km northwest of the Jinmium site complex [ILLUSTRATION FOR FIGURE 1 OMITTED], where two vast outcrops of sandstone rise above the forested plain. A total of 14 localities with cupules are known at Granilpi, as well as many forms of painted, engraved and beeswax rockart. This rock-art at the Granilpi end of the Jinmium Dreaming Track was briefly recorded by Chaloupka prior to 1984, documented by McNickle during 1984 and 1985 (1991: 38), visited by Mulvaney (1996) with Paddy Carlton and other Aboriginal people, and documented by the present research team.

Unlike Jinmium, with its cupule-marked walls, the Granilpi site complex exhibits a marked preference for placing emphasis on horizontal shelter floors and boulders. The larger and eastern of the two heavily eroded sandstone outliers contains remarkable cupule complexes. The most elaborate is a set of 5 archaeological localities (KR6, KR7, KR8, KR9, KR13 - [ILLUSTRATION FOR FIGURE 9 OMITTED]), linked by cupule-covered boulders: three shelters that also contain paintings and, in one case, abraded grooves (KR7); a cave-like shelter (KR13); and a massive, fully carved and shaped boulder (KR6). The boulders are not isolated but are in visual proximity with at least one other; cupules occur only on some of the boulders that could have been marked. They are best interpreted as a chain, linear arrangement or 'pathway' of 35 cupulemarked boulders linking the five archaeological sites. The chain of connection is continuous, a loop or curved pathway that doubles back, connecting the five sites over a distance of almost 200 m. The average diameter of the cupules throughout the KR6-KR7-KR8-KR9-KR13 complex is 64 mm, twice the size of Jinmium wall cupules, with an average 30 cupules per boulder or piece of bedrock, and 1117 cupules in total.

The southernmost shelter, KR7, has 15 cupule-marked boulders in its immediate vicinity plus cupules on the horizontal projection of floor below the wall, at the northeast end [ILLUSTRATION FOR FIGURE 10 OMITTED]. The northernmost shelter, KRS, has 21 cupule-marked boulders in its immediate vicinity, plus cupules on the floor area at the northeast end and on a horizontal wall at the southwest end [ILLUSTRATION FOR FIGURE 11 OMITTED]. A total of 16 cupule-marked boulders or surfaces link KR7 to KR8, around the outskirts of the outlier complex, while a further 19 form the second half of the loop, via shelter KR9 and a natural hole through the back of shelter KR7.

Both KR7 and KR8 are dominated by enormous, carved boulders [ILLUSTRATION FOR FIGURE 10 & 11 OMITTED] - nodal points. The KR7 boulder, 1.9xl.56 m by 0.6 m high, with at least 211 cupules covering its horizontal; vertical and sloping surfaces (average diameter 65 mm), has more cupules on it than any other we have recorded with virtually every square centimetre utilized. The larger KR8 boulder, 2.25x1.05 m by 1.34 m high, contains only 82 cupules (average diameter 64 mm), with some patches not fully marked. Beyond the KR8 boulder, another cupule-marked boulder leads toward a patch of 20 cupules on the floor, in front of a passage that broadens into a small cave inside the rock. Between these two shelters, at locality KR6, a massive, eroded carved block, 3.2x3.25 m by 0.74 m high has 94 large cupules, some of the largest on vertical surfaces (average diameter on vertical surfaces 155 mm, on horizontal 48 mm).

KR13 is a cave-like structure, formed by two enormous blocks of sandstone resting against each other, with fiat cupule-covered boulders marking the southern and northern ends of the large passage that runs through the shelter (20 and 15 cupules; 55 mm average diameter).

The marks at the KR6-KR7-KR8-KR9-KR13 complex link 4 separate rock shelters across a 120-m distance, in a way we have never seen before; cupules on boulders and the floor of a shelter at one end of the loop lead to a natural hole that broadens into a small cave.

Other significant sites recorded in the main Granilpi complex include KR10, 142 cupules in 5 main patches on the floor: either side KR8, the entrance; in the middle of the sloping part of the floor; and at the back near natural holes that run through the rock (average diameter 66 mm; average depth 12 mm). At KR11 a small cluster of 4 pecked and pounded cupules are arranged in an arc, on the top of the uppermost ridge at Granilpi. KR12 is a flat, circular boulder 4.36x5.0 m by 1.54 m high with 3 clusters of cupules arranged in a triangular layout of 12, 12, 9, with a single cupule between two of the clusters (average cupule diameter about 58 mm). KR15 has sculpted (shaped), cupulecovered boulders (80 mm average diameter) below a vertical shaft (passageway) that extends down from the top of the outlier, rows of cupules on a near-by wall (average diameter 22 mm), and others on fiat surfaces near horizontal passageways. KR16 has cupules on a northeast projection of sandstone, at the limits of the outlier, that lead towards 2 cupule-covered boulders in line with a back wall and overhang that has further marks. KR17, the uppermost Granilpi shelter, has 29 floor cupules (average diameter 65 mm) marking its extremities.

Two cupule sites at an outlier of Granilpi are located about a kilometre west of the main complex. The first, KR19, is a small overhang formed by fallen blocks of sandstone perched on the bedrock formation; the second, KR20, is a deep, long shelter at the base of the formation [ILLUSTRATION FOR FIGURE 12 OMITTED].Each site has more recent forms of art, and KR20 has what appears to be a substantial archaeological deposit with cultural material evident.

KR19 is dominated by 4 red ochre and beeswax figures (to do with sorcery according to Paddy Carlton), exquisite peck-infill naturalistic animals (including two small birds, a possum and three macropods), and clusters of cupules. A small natural passage through its centre runs roughly east-west, with a branch to the north. The passageway is too small to crawl through, but one can easily look through it. Cupules were placed at each entrance to the tunnel: 6 cupules and some pits in front of the western entrance, 9 cupules and pits in front of the northern hole, a single cupule next to the eastern entrance. The cupules, appearing more weathered (older) than the engraved animals, are undoubtedly early at the site. Once again natural holes or tunnels through stone have been emphasized in a cultural, symbolic manner.

At KR20, a 17-5-m long rock-shelter with a deep, high overhang [ILLUSTRATION FOR FIGURE 12 OMITTED], on the back wall are paintings of a snake, goanna and emu; above them 34 hand stencils in red or white are arranged into two groups of 8 and 26. The painted snake is directly above a set of 6 small cupules on the sandstone floor. From this group, a line of 15 cupule-covered boulders runs across the front of the shelter, continuing beyond it for a total of 38.5 m. Two other boulders with cupules are further out from the front of the shelter, past the main line. There is an average of just over 12 cupules per boulder (average 58 mm diameter), and a pattern: 1 or 2 boulders with 1-6 cupules consistently lie between I or 2 boulders with 1117 cupules. In the middle of the shelter, and about a third of the way along the cupule boulder line, a particularly large block of stone was almost fully carved; 94 cupules cover its surface.

The chain of marked boulders has a certain rhythm to it, an abacabbaabababbaab, with 'c' as the boulder that has 94 cupules, 'a' as a boulder with few cupules, and 'b' as a boulder with many cupules.

Summary of cupule patterning

In the 26 sites studied, cupules were used in at least 10 different ways to mark places, define boundaries, link localities, highlight natural features and define spaces within shelters:

1 Extensively infilled shelter wall panels with natural boundaries highlighted (4 localities: KR1, KR2, KR21, KR24).

2 Carved/pitted/marked boulders (12 localities: KR2, KR6, KR7, KR8, KR9, KR12, KR15, KR16, KR20, KR23, KR24, KR25). Some large rock-shelters with deposits are dominated by a central, massive carved boulder (5 localities: KR7, KRS, KR20, KR23, KR24).

3 Cupules on high points, tops of geological features (4 localities: KR2, KR13, KR11, KR12).

4 Cupules mark/accentuate the outside edges of natural holes/tunnels/passageways through the rock but not the insides (12 localities: KR4, KR7, KR8, KR10, KR13, KR15, KR19, KR24, KR25, K_R26, KR29, KR33).

5 Cupules mark edges of shelter floors and shelter limits, inside versus outside (5 localities: KR5, KR8, KR9, KR10, KR17).

6 Cupules mark extremities of a geological complex (2 localities: KR6, KR16).

7 Cupules mark floors adjacent to walls but not in relation to other features, such as holes/tunnels or lines of marked boulders (1 locality: KR21).

8 Linear arrangement of linked localities (1 complex of 4 localities linked by 2 paths: KR6-KR7-KR8-KR9).

9 Linear arrangements run across the front of shelters, defining inside versus outside. Sometimes there is a rhythm to the arrangement of cupules from one boulder to the next (1 locality: KR20).

10 Cupules form geometric motifs and/or single rows on walls or boulders (7 localities: KR1, KR5, KR11, KR12, KR15, KR 30, KR31).

Another common feature is hand stencils placed above cupules or over their upper reaches (14 localities, 54% of sites: KR1, KR2, KR4, KR5, KR7, KR8, KR9, KR10, KR15, KR17, KR20, KR21, KR24, KR30). The hand stencils, in various shades of purple, red and white, and added long after the cupule designs were completed, tell us that subsequent visitors recognized the significance of localities marked by cupules. Presumably, visitors to the sites interacted with the cupules in other ways as well; the reticent archaeological evidence and our imperfect theoretical understanding does not reveal how; 'significance' might be the most important variable for subsequent marking activity and 'significance' may be independent of cupules.

Cupules on walls average 28-32 mm in diameter, those on boulders 58-66 mm; the grinding hollows resulting from food-processing average 110 mm, range 77-150 mm. An exception is at KR6 where four vertically oriented cupules measure 223.6 mm, 201.4 mm, 119.9 mm and 75.5 mm; on a vertical surface, these must be cupules.

Cupules were not placed inside holes or passageways but rather mark their boundaries or limits. 'Inside versus outside versus other side' is a common, repeated theme. At Granilpi, where there is extensive use of boulder surfaces, though not to the exclusion of rock-shelter walls, there appears a conformation of placement to natural features, contours and spatial indicators, as there is at Jinmium. As excavation at Granilpi sites has only just begun, their precise temporal relationship to the Jinmium complex is unknown; they may be much the same.

The age of north Australian cupules

There is general consensus (above) that some cupules are old; wherever they are found cupules are among that region's oldest surviving rockart. Yet for most of the world's cupules we have little evidence of date.

Absolute dating

Cupules are sometimes found well beneath present surface level in datable archaeological contexts. Mike Morwood (pets. comm.) has observed this for sites in northern Queensland and the Kimberley, while we have observed cupules anywhere from 30 cm to over a metre below present ground levels at several Northern Territory sites. At Jinmium, Fullagar et al. (1996) recently claimed the first minimum date for ancient examples of north Australian cupules. Excavation at KR1 recovered an engraved piece of sandstone, with at least 2 clear cupules, from almost a metre below the surface, in a level with a thermoluminescence (TL) determination of 58,000-75,000 years ago (Fullagar et al. 1996: 765; other dating by optically stimulated luminescence [OSL] and AMS radiocarbon has been initiated by the Jinmium team); the cupules are much the same size as those on the wall panel, 26-34 mm diameter. Not far from where the cupule-bearing piece of sandstone was found is a previously buried section of wall marked with a few cupules, near the same level. Initial Jinmium age determinations have been challenged and are not yet resolved.

But for the cupules, a minimum age of 58,000 years is not inconsistent with the 55,000-60,000 year TL and OSL dates of the oldest Kakadu deposits (Roberts et al. 1990; 1993), which lie within a few kilometres of cupule-covered walls and boulders remarkably similar to Jinmium area sites (see Chaloupka 1993: 235-6; figure 270; Edwards 1979: 148, plate 40). At the Kakadu sites the oldest used pieces of ochre also date to this time; but at Jinmium ochre occurs in deposits thermoluminescence-dated up to 116,000 years ago (Fullagar et al. 1996). 'This obviously controversial scenario needs to be tested further at this and other sites' (Fullagar et al. 1996: 771).

Elsewhere in northern Australia, O'Connor (1995) has recovered ochred roof fall, dated to 39,000 years ago, from deposit below Kimberley rock paintings of the type consistently found over cupules at other sites; Rosenfeld (1981) dated deposit overlying north Queensland engravings, stylistically more recent than north Australian cupules, to about 13,000 years ago. Roberts et al. (1997) have obtained OSL determinations of 17,000-26,000 years ago for mudwasp nests that overlie a variety of Kimberley rock paintings of the sort found overtop cupules at some sites (but Watchman et al. (1997) have obtained AMS determinations of up to 4000 years for paintings at similar sites, and the data suggest some Bradshaw paintings may be only about 1500 years old).

Stratigraphic sequencing/superimpositioning Rosenfeld & Smith (1997) review problems associated with the dating of rock art using single approaches and recommend a combination of radiocarbon and stylistic dating where possible. Rock art of the Keep River region is seen as a mix of western Kimberley and eastern Victoria River forms and styles, with perhaps some further influences from the south - as is consistent with its location between those two regions. Gunn (1988: 15), noting some local differences, discounts southern influences; our studies also note site - and region - specific forms. A surprising discovery was of two mulberry-coloured Bradshaw paintings at Granilpi, the easternmost examples of this form of painting yet known. Motifs particular to the GranilpiJinmium area are thin human-like figures, with curvaceous limbs, made from bright red ochre, beeswax or a combination of both. Said by Paddy Carlton to be sorcery paintings, they consistently are superimposed on top of most other imagery in the uppermost layers of heavily marked sites. Two with beeswax, sampled by Tacon & Carlton for AMS dating, give dates of 80-220 years before present (LLNL-32254: 160[+ or -]60; LLNL-32255: 130[+ or -]50).

McNickle (1991: 44-5) was the first to note cupule superimpositioning in the Keep River region, possibly at Granilpi:

A boulder at a Keep River site, over a metre high, is covered almost entirely by pits. The pits on the vertical sides are all small and regularly spaced, but on the upper surface a number of considerably larger pits have been ground, and smaller pits subsequently superimposed over these. This is the only instance of superimposition recorded within this technique of rock art.

Other notable superimpositions include: a peck infill macropod over cupules at KR1; red hand stencils over purple hand stencils (KR10); a red sorcery figure over a purple hand stencil (KR15); two red human-like figures over an upside-down purple emu which, in turn, lies over an unidentified peck infill animal, with both the painted emu and the pecked animal lying over a row of cupules (KR15); and white outline turtles over red sorcery figures (KR15 tunnel ceiling entrance).

Both east and west of Jinmium-Granilpi, sites consistently have cupules underneath all other forms of engraved, painted and beeswax rock-art. Superimpositioning at Keep River region localities (KR1, KR10, KR15, KR20) creates this rock-art sequence, from oldest to most recent:

1 Cupules.

2 Peck infill naturalistic animals.

3 Purple hand stencils, Bradshaw figures and rare animals such as an emu.

4 Red hand stencils and hand prints, humanlike figures and animals.

5 Red, white, red-and-white animals and human-like figures; red, white, yellow hand stencils.

6 Red sorcery figures with curvy limbs; beeswax figures.

7 White outline animals.

Bradshaw paintings are conjectured to be many tens of thousands of years old (Walsh (1994); preliminary dating of mud-wasp nests overlying Bradshaw paintings suggests they are more than 17,000 years of age (Roberts et al. 1997; and above). Elsewhere, in Kakadu and Arnhem Land, many forms of painting - such as Dynamic Figures and Large Naturalistic animals thought to be at least 10,000 years of age on a number of grounds (Chaloupka 1993; Tacon & Chippindale 1994) - overlie cupules.

Weathering

Most of the cupules have a very weathered appearance, even in hard granitic rock, with the exception of those recently retouched. At some sites, particularly exposed areas are extremely weathered. This is most noticeable with exposed portions of cupule-covered walls. Cupules often have post-manufacture chips, flakes and even large cracks on and through them; something less evident in other forms of rock-art.

Silica skins, oxalate crusts and desert varnish

Some cupules have a very rich patina, the same as surrounding substrate, again indicative of antiquity and comparable with some 'varnished' engravings of South Australia (Dorn& Nobbs 1992; Nobbs & Dorn 1988; 1993). Dorn dated South Australian 'varnished' engravings to over 40,000 years using cation ratio and AMS techniques, but see scepticism associated with his results (Rosenfeld 1993: 77; Harry 1995; Watchman 1989; 1992) It seems the skins are not 'sealed' and that we do not yet fully understand the perhaps several processes that can push dates up or down (Dorn 1997). Furthermore, independent resampling of some sites AMS dated by Dorn could not confirm his results (Beck et al. 1997; Dayton 1997) so we should not rely on the similar appearance of patina as an indication of a particular age.

However, some of the cupule-covered walls and boulders, and most cupule-covered floors, have oxalate crusts (a number of different coloured layers overlie individual cupules; Watchman 1997). Watchman (1991) dated similar accretions in Kakadu National Park, and concluded that some paintings lying underneath had a minimum age of 8800 years (Watchman 1987); cupules consistently are found under such paintings. In 1997 Watchman sampled KR1, KR10 and KR24, finding carbon-bearing accretions overlying cupules at each site - preliminary dates are in hand.

Contemporary knowledge

A final strand of supporting evidence for cupule antiquity is the fact that in northern Australia there is no historic ethnographic information or contemporary Aboriginal knowledge about cupules except the Green Plum Dreaming connection in central Arnhem Land and a contemporary use of similar designs in restricted men's ceremonies (which we are forbidden to reveal publicly). Aboriginal people in central and western Arnhem Land and in the Keep River region consistently state that the cupules have always been there; the designs originated in the earliest eras of the Dreaming. They suggest how cupules may have perhaps been used - such as memory aids or to mark Dreaming tracks - but have no detailed information. This is very different from what they have to say about all subsequent forms of rock-art, with a great deal known especially for paintings and engravings produced in the past few thousand years (Mulvaney 1996; Tacon 1993). But we are not saying that the sites are not important for contemporary people - in most cases they were marked by their immediate ancestors with more contemporary forms of art and some continue to feature in ceremonies (Garde 1994: 4; Tacon 1993); they lie on important Dreaming Tracks and until recently were seasonal camp sites (Fullagar et al. 1996; McNickle 1991: 44; Tacon 1993).

In central Australia, Mountford (1976) recorded the production of cupule-like hollows made during the course of an increase ceremony that involved pounding a boulder (see also Flood 1997: 146-7) and we note a widespread association between pounding, painting or marking rock surfaces and the release of spiritual, creative 'power' for Aboriginal Australians. We also note almost two dozen possible explanations as to why north Australian cupules arose but none have firm supporting evidence (Tacon et al. in preparation).

Thus with no contemporary ethnographic insight, no inside knowledge and few analogies, instead we must concentrate on formal methods of analysis (see Chippindale & Tacon in press). In this vein we must also consider that different sorts of cupules likely were made at different times, some sites may have had a number of cupule-marking events and old cupules may have been touched-up or re-marked on one or more occasions.

Regional trends

Welch (1993: 101) has counted cupule sites across the Kimberley; they occur at less than 2-5% of sites to the west of Jinmium-Granilpi:

Of the 700 sites analysed, 17 contained heavily weathered pecked and abraded hollows on vertical rock faces. These varied in size from about two and a half to six centimetres in diameter. They were found in both the largest and smallest shelters, sometimes on an outer facing wall and numbering about 400, at other times in the most awkwardly tight places at the back of low shelters and numbering only a few. At these sites they were covered in thick, smooth rock varnish and appeared to be of extreme age....

Welch notes larger cupules (6-16 cm across), some of which may be grinding hollows, on horizontal surfaces of a further 29 shelters. Some are probably cupules made for symbolic purposes, some grinding hollows that resulted from food preparation; even if all 29 are added to the sample cupule sites would make up less than 7% of Kimberley rock-art sites. (But Walsh (1996) reports dozens more sites from the same region.) They are rarer in Kakadu and Arnhem Land where the dozen or so cupule sites constitute less than 0.25% of over 5000 known rockart sites (Chaloupka 1993; Sullivan 1988). McNickle (1991) reports only 16 sites from the entire Victoria River region (plus later, a 17th site from the Coolibah area with a panel '5 m by I m, densely covered with pits of uniform size': 1993: 39). The Jinmium-Granilpi complex has so far yielded 26 cupule sites across only a 30-km stretch of land, the densest concentration to date.

But it is less the number of sites at which cupules are found than the persistent, structured and widespread distribution of the cupules which indicates they were part of a widely held cognitive framework. Subsequent bodies of rockart are much more regionally focused, indicating specific understandings within a specific region. It is reasonable to assume that the structure of cupule assemblages indicates something of a common concern for early Australian colonists, that is, if their structuring is consistent across other regions. Thus, a key concern is the organizational structure of comparable cupule sites from the Kimberley, from the Victoria River region and from Kakadu-Arnhem Land. A cursory review reveals strong formal similarity between some sites, such as shelters with cupule-covered walls or massive carved boulders.

Once cupule sites across northern Australia have been documented in the way we have recorded and analysed the Keep River sites, we can begin to assess long-range settlement and migration patterns for the earliest periods of north Australian prehistory, including some Papua New Guinea/Irian Jayan sites in this context. We may also begin to better understand how people were articulating and expressing their relationships to landscapes spread over vast distances many tens of thousand of years ago. We should be careful not to extend the results too far; as Steinbring & Lanteigne (1991: 24) note:

The universality of cupules relates firstly to their simplicity in the fundamental process of marking. The thesis entertained here is that basic cupules are the most likely form susceptible to re-invention or parallelism. The isolated forms need not be explained through diffusionistic formulae for another reason. All areas of the world provide natural prototypes. Vesicles, pits naturally formed by abrasion or impact, or both, abound throughout the world, and may be selected by humans as a basic sign, the symbolic meaning of which may vary.

Importantly, Paddy Carlton, one of the traditional custodians of Granilpi and Jinmium, recognized this. Quizzed about the origin of the cupules, on 15 June 1996, he said the vertical cupules, sculpted rocks and some of the other cupules are signs of Djibigung, the Dreaming Quail that features in the Jinmium story, and may have been made by it. Likening the colour and texture of the rock to the body colouring/marking of the Quail, he said the unique pitted and pock-marked texture of the sandstone outcrops, seen on their tops and upper reaches [ILLUSTRATION FOR FIGURE 13 OMITTED], indicates where Djibigung went. The cupule marks may be copies. This is consistent with what each of us independently arrived at in the field: that perhaps the natural geology inspired the early production of cupules. Although the cupule is undeniably cultural, it has strong 'natural' roots, indicative of an early coming to terms with and marking of the landscape. Although speculative, it is proposed here that later marks that soon followed cupules in various parts of the country also reflect a strong landscape concern, with track motifs imitating those left on the land by animals (Maynard 1979; Rosenfeld 1992) and concentric circles around cup-marks similar to marks made by rain on the surface of water holes (Tacon 1994: 118).

Conclusions

This study shows cupules were placed on rock surfaces to mark and define a human universe at and between particular localities, to accentuate or separate culturally defined areas from the natural. Thus began a long and continuous process of socializing the landscape in enduring media - a process that continued with various phases of figurative rock-art that underpins contemporary oral history and links sites together as part of an elaborate Dreaming Track with Jinmium as a focal point. But Jinmium is not a 'one-off', rather it is part of a complex of dozens of sites within a 30-km distance and many more spread across the northern portions of Western Australia and the Northern Territory. These cupule sites are archaic in the Australian sequence. It is intriguing to note that cupules are also among the oldest surviving forms of rock-art from many other parts of the world, often immediately post-dating colonization. Perhaps, among other things and in their most general sense, cupules world-wide relate to a coming to terms with landscapes (remembering, of course, that their specific meanings and functions would most definitely have varied widely from place to place and period to period).

A precise chronology for cupule art cannot yet be established and it is likely there were many periods of cupule production. We think that much cupule art is early, even as early as about 40-50,000 years ago. This idea is in keeping with current views about the presence of humans and the antiquity of rock-art in Australia and reaffirms beliefs that people engaged in typically human symbolic behaviour, in terms of modifying, marking and socializing landscapes, soon after initial colonization (Flannery 1994; Flood 1996; Tacon 1994). Cupules form a distinct rock-art tradition with a simple motif (pecked circular depression); yet the tradition's distribution, location and morphological variability suggests complex patterning related to deliberate and permanent marking of cultural places. We will never know the exact reasons why places were marked in this way but perhaps some of the same motivations were behind both early and more recent cupule production, whatever these may be. After all, cultural continuities can be remarkably enduring (Flannery 1996; Tacon et al. 1996)!

Acknowledgements. We are most grateful to Paddy Carlton, Biddy Simon, Maurice Simon and Polly Wandanga, traditional custodians of sites in the greater Jinmium-Granilpi region, for permission to undertake research. One or more of them also participated in the project on a daily basis and were valuable partners without whose assistance this paper would not have been possible. We are also thankful for the hospitality they and other residents of Marralam Community extended toward us during our stays in the Northern Territory.

Lesley Head and Jenny Atchison of the University of Wollongong are thanked for assistance in the field and for valuable comments that improved this manuscript. Peter Tozer (Waringarri Aboriginal Corporation) is thanked for logistical assistance, sound advice and hospitality in Kununarra. Sally Brockwell and Andrew McWilliam are thanked for accommodation and their usual generous hospitality in Darwin. The North Australian Research Unit and the Aboriginal Areas Protection Authority assisted with transportation in the field. Anna Murray (Sydney University) assisted with data and image analysis.

This research was primarily sponsored by the Australian Museum with funds made available by the People and Place Research Centre and a Visiting Fellowship Grant for Sven Ouzman. The University of Wollongong also provided funding for Head and Atchison. Fullagar is a Research Fellow funded by the Australian Research Council. Paddy Pallin Pry Ltd. provided some field equipment.

We also would like to thank Jim Specht and ANTIQUITY referees Andree Rosenfeld and Josephine Flood for comments and advice that contributed to the quality of this paper. All photographs are by Paul S.C. Tacon while the maps are drafted by Richard Fullagar from field sketches and assistance from Jenny Atchison and Ken Mulvaney. Previous versions of this paper were presented at The Origins and Significance of Northern Rock Art Seminar, Melbourne on 17 October 1996, the American Anthropological Association 95th Annual Meeting, San Francisco on 21 November 1996 and to the Royal Society of New South Wales on 4 April 1997.

References

AHLSTROEM, G.W. 1978. Wine presses and cup-marks of the Jenin-Megiddo survey, Bulletin of the American School of Oriental Research 231: 19-49.

ANATI, E. 1993. World rock art: the primordial language. Capo di Ponte (BS): Centro Camuno di Studi Prelstorici. Studi Cam uni 12.

BAHN, P. & A. ROSENFELD (ed.). 1991. Rock art and prehistory: papers presented to symposium G of the AURA Congress, Darwin 1988. Oxford: Oxbow. Monograph 10.

BECK, W., D. DONAHUE, G. BURR & A.J.T. JULL. 1997. AMS [.sup.14C] dating of Early Anasazi petroglyphs from the North American Southwest Desert region. Paper presented at the Sixth Australasian Archaeometry Conference, Sydney, 11 February 1997.

BEDNARIK, R. 1993. About cupules, Rock Art Research 10(2): 138-9.

1994. The Pleistocene art of Asia, Journal of World Prehistory 8(4): 351-75.

1995. Concept-mediated marking in the lower Palaeolithic, Current Anthropology 36(4): 605-34.

1996. The cupules on Chief's Rock, Auditorium Cave, Bhimbetka, Artefact 19: 63-72.

BEDNARIK, R., G. KUMAR & G.S. TYAGI. 1991. Petroglyphs from central India, Rock Art Research 8(1): 33-5.

BRADLEY, R. 1991. Rock art and the perception of landscape, Cambridge Archaeological Journal 1(1): 77-101.

1995. Rock carvings and decorated monuments in the British Isles, in Helskog & Olsen (ed.): 107-29.

BRETON, A.C. 1910. Seventeenth International Congress of Americanists, Buenos Aires, May 16th-24th, 1910, Man 10: 141-4.

BROWN, S. 1991. Art and Tasmanian prehistory: evidence for changing cultural traditions in a changing environment, in Bahn & Rosenfeld (ed.): 96-108.

CAMPBELL, A.C., L.H. ROBINS & L.C. MURPHY. 1995. Oral traditions and archaeology of the Tsodilo Hills Male Hill Cave, Botswana Notes and Records 26: 37-54.

CHALOUPKA, G. 1993. Journey in time. Sydney: Reed Books.

CHAPLIN, J.H. 1963. Rock engravings at Nyambwezu, northern Rhodesia, South African Archaeological Bulletin 19(73): 13-14.

CHASE, P. & H. DIBBLE. 1987. Middle Palaeolithic symbolism: a review of current evidence and interpretations, Journal of Anthropological Archaeology 6: 263-96.

CHIPPINDALE, C. & P.S.C. TACON. 1993. Two old painted panels from Kakadu: variation and sequence in Arnhem Land rock art, in Steinbring et al. (ed.): 32-56.

(Ed.). In press. The archaeology of rock-art. Cambridge: Cambridge University Press.

CLARK, J.D. 1958. The Chifubwa Stream rock shelter, Solwezi, northern Rhodesia, South African Archaeological Bulletin 13(49): 21-4.

CONKEY, M. 1987. New approaches in the search for meaning? A review of research in 'Palaeolithic Art', Journal of Field Achaeology 14: 413-30.

COOKE, C.K. 1964. Rock gongs and grindstones: Plumtree area, southern Rhodesia, South African Archaeological BulletIn 19(75): 70.

COSTAS GOBERNA, F.J., P. NOVOA ALVAREZ & J.M. ALBO MORAN. 1993/94. Los grabados rupestres de Pena de Chaos y Pena da Moura en San Fiz de Amarante (antas de Ulla, Lugo), Brigantium 8: 263-84.

COX, J.H. & E. STASACK. 1970. Hawaiian petroglyphs. Honolulu (HI): Bishop Museum Press.

DART, R.A. 1953. Rhodesian engravers, painters and pigment miners of the 5th millennium BC, South African Archaeological Bulletin 8(32): 91-6.

DAVID, B. & D. CHANT. 1995. Rock art and regionalism in north Queensland prehistory, Memoirs of the Queensland Museum 3(2): 357-528.

DAVID, B., I. MCNIVEN, V. ATTENBROW, J. FLOOD & J. COLLINS. 1994. Of Lightning Brothers and white Cockatoos: dating the antiquity of signifying systems in the Northern Territory, Australia, Antiquity 68: 241-51.

DAVIDSON, I. & W. NOBLE. 1989. The archaeology of perception: traces of depiction and language, Current Anthropology 30(2): 125-55.

DAYTON, L. 1997. Dating row divides rock art experts, New Scientist 2074: 10.

DORN, R. 1997. Constraining the age of the Coa valley (Portugal) engravings measured with radiocarbon dating, Antiquity 71: 105-15.

DORN, R. & M. NOBBS. 1992. Further support for the antiquity of South Australian rock engravings, Australian Aboriginal Studies 1992(1): 56-60.

EDWARDS, R. 1979. Australian Aboriginal art: the art of the Alligator Rivers region, Northern Territory. Canberra: Australian Institute of Aboriginal Studies.

FLANNERY, T. 1994. The future eaters: an ecological history of the Australasian lands and peoples. Chatswood (NSW): Reed.

1996. Defective solutions, The Australian's Review of Books 1(3): 3-4, 26.

FLOOD, J. 1987. Rock art of the Koolburra Plateau, north Queensland, Rock Art Research 4(2): 91-126.

1996. Culture in early Australia, Cambridge Archaeological Journal 6(1): 3-36.

1997. Rock art of the Dreamtime: images of ancient Australia. Sydney: Angus & Robertson.

In press. Copying the Dreamtime: anthropic marks in early Aboriginal Australia, in R. Bednarik & F. D'Errico (ed.), New approaches (Proceedings of the Turin Congress 1995).

FOCK, G.J. 1969. Non-representational rock art in the northern Cape, Annals of the Cape Provincial Museums 6(11): 10336.

FORSBROOKE, H.A. 1954. Further light on rock engravings in northern Tanganyika, Man 54(157): 101-2.

FRANKLIN, N. 1991. Explorations of the Panaramitee Style, in Bahn & Rosenfeld (ed.): 120-35.

FRIMIGACCI, D. & J. MONNIN. 1980. Un inventaire des petroglyphes de Nouvelle-Caledonie, Grand Terre et Iles, Journal de la Societe des Oceanistes 36: 17-59.

FU, C.Z. 1992. Discovery of rock art in China, in Lorblanchet (ed.): 361-71.

FULLAGAR, R., D. PRICE & L. HEAD. 1996. Early human occupation of northern Australia: archaeology and thermoluminescence dating of Jinmium rock-shelter, Northern Territory, Antiquity 70: 751-73.

GARANGER, J. 1972. Archeologie des Nouvelles-Hebrides. Paris: Musee de l'Homme. Publications de la Societe des Oceanistes 30.

GARDE, M. 1994. Mann River region rock art recording project: transcripts of field audio recordings July 1993. Bawinanga Aboriginal Corporation, Maningrida.

GAY, C. 1973. Olmec hieroglyphic writing, Archaeology 26(4): 278-88.

GELDMACHER, H. 1967. Mooiplaas: report on rediscovery of an unusual engraving site, South African Archaeological Bulletin 22: 24.

GERMOND, G. 1980. Inventaire des megalithes de la France 6: Deux-Sevres. Paris: Editions du CNRS. Gallia Prehistoire.

GLOB, P.V. 1969. Helleristninger I Danmark. Copenhagen: Gyldendalske Boghandel. Jutland Archaeological Review Publications 12.

GOREN-INBAR, N. 1986. A figure from the Acheulian site of Berekhat Ram, Mi'Tekufat Ha'Even 19: 7-12.

GRAHAM, R. & K. MULVANEY. 1995. The Granites: history, art and ethnography. Darwin (NT): Aboriginal Areas Protection Authority.

GRANT, C. 1967. Rock art of the American Indian. New York (NY): Promontory Press.

GREBENART, D. & B. PIERRET. 1966. Traits polis et cupules de l'Oued Zireg (province de Taza-Maroc), Libyca (NS) Serie Anthropologie, Prehistoire. Ethnogrophie 14: 329-35.

GUIRAUD, R. 1964. Cupules et gravures dans la commune de Combes (Herault), Cahiers Ligures de Prehistoire et d'Archeologie 13: 125-37.

1970. Nouvelles gravures et cupules dans le massif CarouxEspinouse-Saumail, Cahiers Ligures de Prehistoire et d'Archeologie 19: 39-58.

GUNN, M. 1986. Rock art on Tabar, New Ireland Province, Papua New Guinea, Anthropos 81: 455-67.

GUNN, R.G. 1988. Recording and assessment of the Nganalang rock art site, Keep River National Park. Unpublished report to the Conservation Commission of the Northern Territory and the Australian Heritage Commission.

HARRY, K.G. 1995. Cation-ratio dating of varnished artifacts: testing the assumptions, American Antiquity 60(1): 11830.

HASKOVEC, I. 1992. Mt Gilruth revisited, Archaeology in Oceania 27(2): 61-74.

HEAD, L. 1994a. Landscapes socialised by fire: post-contact changes in Aboriginal fire use in northern Australia, and implications for prehistory, Archaeology in Oceania 29: 172-81.

1994b. Aborigines and pastoralism in north-western Australia: historical and contemporary perspectives on multiple use of the rangelands, Rangeland Journal 16: 167-83.

HEAD, L. & R. FULLAGAR. 1991. 'We all la one land': pastoralist excisions and Aboriginal resource use, Australian Aboriginal Studies 1991(1): 39-52.

In press. Aboriginal landscapes of the northwest Northern Territory, Australia, in R. Layton & P. Ucko (ed.), Frontiers of landscape archaeology. London: Routledge.

HEBERT, B. 1965. Les megalithes sculptes de l'Ile Emau, Etudes Melanesiennes 18-20: 56-61.

HEDGES, K. 1993. Places to see and places to hear: rock art and features of the sacred landscape, in Steinbring et al. (ed.): 121-7.

HELSKOG, K. & B. OLSEN (ed.). 1995. Perceiving rock art: social and political perspectives. Oslo: Instituttet for sammenlignende kulturforskning.

HENSCHEN-NYMAN, O. 1982. Forhistoriska toner genom Sangelstainen. Visby: Gotlands allehana.

HORNELL, J. 1925. The archaic sculpted rocks and stone implements of Gorgona Island, South America [Columbia], Man 25: 104-7.

JACKSON, P.J. 1995. A continuing belief system? Irish passage grave art and the cup and ring engravings of the British Isles and Eire, in Helskog & Olsen (ed.): 396-406.

JOHNSON, S.A. 1991. Distributional aspects of prehistoric irish petroglyphs, in Bahn & Rosenfeld (ed.): 86-95.

1993. The utility of 'style' in the analysis of prehistoric Irish rock art, in Lorblanchet & Bahn (ed.): 143-50.

KEYSER, J.D. 1992. Indian rock art of the Columbian Plateau. Seattle (WA): University of Washington Press.

KUMAR, G. 1996. Daraki-Chattan: a Palaeolithic cupule site in India, Rock Art Research 13(1): 38-46.

LARSSON, T.B. 1989. A spatial approach to socioeconomic change in southern Sweden 500 BC - AD 500, in T.B. Larsson & H. Lundmark (ed.), Approaches to Swedish prehistory: 33551. Oxford: British Archaeological Reports. International series S500.

LEASK, M.F. 1943. Rock engravings and paintings of the Sogeri district of Papua, Mankind 3: 116-20.

LEE, G. 1992. Rock art of Easter Island: symbols of power, prayer to the gods. Los Angeles (CA): Regents of the University of California.

LEONAHDI, P. 1988. Art Paleolithique mobilier et parietal en Italie, L'An thropologie 92: 139-202.

LEWIS, D. 1988. The rock paintings of Arnhem Land: social, ecological, and material culture change in the post-glacial period. Oxford: British Archaeological Reports. International series S415.

LEWIS, D. & D. ROSE. 1988. The shape of the Dreaming: the cultural significance of Victoria River rock art. Canberra: Aboriginal Studies Press.

LINDY, J.M. & G.A. CLARK. 1990. Symbolism and modern human origins, Current Anthropology 31(3): 233-61.

LORBLANCHET, M. (ed.). 1992. Rock art in the Old World. New Delhi: Indira Gandhi National Centre for the Arts.

LORBLANCHET, M. & P. BAHN (ed.). 1993. Rock art studies: the post-stylistic era or where do we go from here? Oxford: Oxbow Books.

MACDONALD, D. 1899. Oceania: linguistic and anthropological. Melbourne: Hutchinson.

MCDONALD, J. & I.P. HASKOVEC (ed.). 1992. State of the art: regional studies in Australia and Melanesia. Melbourne: Archaeological Publications. Occasional AURA Publication 6.

McNICKLE, H.P. 1984. Variation in style and distribution of rock engravings in the Pilbara region (Western Australia), Rock Art Research 1(1): 5-24.

1985. An introduction to the Spear Hill rock art complex, northwestern Australia, Rock Art Research 2(1): 48-64.

1991. A survey of rock art in the Victoria River District, Northern Territory, Rock Art Research 8(1): 36-46.

1993. Reply to comments and an update on the Victoria River District, Rock Art Research 10(1): 38-40.

MANDT, G. 1995. Alternative analogies in rock art interpretation: the West Norwegian case, in Helskog & Olsen (ed.): 263-91.

MARSHACK, A. 1997. The Berekhat Ram figurine: a late Acheulian carving from the Middle East, Antiquity 272: 327-37.

MAYNARD, L. 1976. An archaeological approach to the study of rock art. MA Henours thesis, University of Sydney.

1979. The archaeology of Australian Aboriginal art, in S.M. Mead (ed.), Exploring the visual art of Oceania: 83-110. Honolulu (HI): University Press of Hawaii.

MELLARS, P. 1989. Major issues in the emergence of modern humans, Current Anthropology 30(3): 349-85.

1991. Cognitive changes and the emergence of modern humans, Cambridge Archaeological Journal 1(1): 63-76.

MONNIN, J. 1986. Leo petroglyphes de Nouvelle-Caledonie. Noumea: Office Culturel Scientifique et Technique Canaque.

MORRIS, R.B.W. 1970. The petroglyphs at Achnabreck, Argyll, Proceedings of the Society of Antiquaries 103: 33-56.

MOUNTFORD, C.P. 1976. Nomads of the Australian desert. Adelaide: Rigby.

MULVANEY, K. 1996. What to do on a rainy day: reminiscences of Mirriuwung and Gadjerong artists, Rock Art Research 13(1): 3-20.

NOBBS, M. & R. DORN. 1988. Age determinations for rock varnish formation within petroglyphs: cation-ratio dating of 24 motifs from the Olary region, South Australia, Rock Art Research 5(2): 108-46.

1993. New surface exposure ages for petroglyphs from the Olary Province, South Australia, Archaeology in Oceania 28(1): 18-39.

NOBLE, W. & I. DAVIDSON. 1996. Human evolution, language and mind: a psychological and archaeological inquiry. Cambridge: Cambridge University Press.

O'CONNOR, S. 1995. Carpenter's Gap Rockshelter 1:40,000 years of Aboriginal occupation in the Napier Ranges, Kimberley, WA, Australian Archaeology 40: 58-9.

ODAK, O. 1992. Cup-marks patterns as an interpretation strategy in some southern Kenyan petroglyphs, in Lorblanchet (ed.): 49-60.

OUZMAN, S. 1995. Spiritual and political uses of a rock engraving site and its imagery by San and Tswana-speakers, South African Archaeological Bulletin 50: 55-67.

In press. Toward a mindscape of landscape: rock art as expres sion of world-understanding, in Chippindale & Tacon (ed.).

PARKMAN, E.B. 1992. Towards a proto-Hokan ideology, in S. Goldsmith et al. (ed.), Ancient images, ancient thought: the archaeology of ideology: 365-70. Calgary: University of Calgary.

1995. 'California Dreamin': cupule petroglyph occurrences in the American West, in J. Steinbring (ed.), Rock art studies in the Americas: 1-12. Oxford: Oxbow. Monograph 45. PEYRONY, D. 1934. La Ferrassie, Prehistoire 3: 1-92.

PIGGOTT, S. 1973. The Dalladies long barrow, NE Scotland, Antiquity 47: 32-6.

PFEIFFER, J.E. 1982. The creative explosion: an inquiry into the origins of art and religion. Ithaca (NY): Cornell University Press.

PODESTA, M.M., M.J. HERNANDEZ LLOSAS & S.F. RENARD DE COQUET (ed.). 1991. El arte rupestre en la arqueologia contemporanea. Buenos Aires: Postgraphia.

POIKALAINEN, V. 1995. Rock art and Estonian identity, in Helskog & Olsen (ed.): 338-47.

QUEREJAZU LEWIS, R. 1990. Rock art investigations in Bolivia. Paper presented to the Canadian Archaeological Association Annual Meeting, Whitehorse (Yukon).

RICHARDS, C. 1996. Henges and water, Journal of Material Culture 1(3): 313-36.

ROBERTS, R., R. JONES & M.A. SMITH. 1990. Thermoluminescence dating of a 50,000-year old human occupation site in northern Australia, Nature 345(6271): 153-6.

1993. Optical dating at Deaf Adder Gorge, Northern Territory, indicates human occupation between 53,000 and 60,000 years ago, Australian Archaeology 37: 58.

ROBERTS, R., et al. 1997. Luminescence dating of rock art and past environments using mud-wasp nests in northern Australia, Nature 387: 696-9.

ROBINSON, K.R. 1958. Venerated rock gongs and the presence of rock slides in southern Rhodesia, South African Archaeological Bulletin 13(50): 75-7.

ROE, D. 1992. Rock art of north-west Guadalcanal, Solomon Islands, in McDonald & Haskovec (ed.): 107-27.

ROSENFELD, A. 1981. Excavations at the Early Man shelter, in A. Rasenfeld et al. (ed.), Early man in north Queensland: 5-34. Canberra: Australian National University. Terra Australis 6.

1986. Rock art in Oceania. Paper presented at Biennale, Cuba.

1992. Changing focus of symbolic expression in Late Pleistocene Australia. Unpublished paper presented at the Second AURA Congress, Cairns.

1993. A review of the evidence for the emergence of rock art in Australia, in M.A. Smith et al. (ed.), Sahul in review: 71-80. Canberra: Australian National University. Occasional Paper in Prehistory 24.

ROSENFELD, A. & C. SMITH 1997. Recent developments in radiocarbon and stylistic methods of dating rock-art, Antiquity 71: 405-11.

RUDNER, I. 1965. Archaeological report on the Tsodilo Hills, Bechuanaland, South African Archaeological Bulletin 20: 51-70.

SCHOONRAAD, M. 1960. Preliminary survey of the rock art of the Limpape Valley, South African Archaeological Bulletin 15: 10-13.

SHARMA, M.L., V. KUMAR & P.T. SHARMA 1992. New rock art sites discovered in Sahibi valley, Rajasthan, Purakala 5(12): 75.

SHUMKIN, V.J. 1991, Rock art of Russian Lapland, Rock Art Research 8(2): 109-12.

SOGNNES, K. 1995. The social context of rock art in Trondelag, Norway: rock art at a frontier, in Helskog & Olsen (ed.): 130-45.

SPRIGGS, M. & W. MUMFORD. 1992. Southern Vanuatu rock art, in McDonald & Haskovec (ed.): 128-43.

STEINBRING, J. & M. LANTEIGNE. 1991. The petroglyphs of west Yorkshire: explorations in analysis and interpretation, Rock Art Research 8(1): 13-28.

STEINBRING, J. & S. STEINBRING. 1983. Manipulative factors in Hawaiian rock art, American Indian Rock Art 9: 89-95.

STEINBRING, J. et al. (ed.). 1993. Time and space: dating and spatial considerations in rock art research. Melbourne: Archaeological Publications. AURA Occasional Paper 8.

SULLIVAN, H. 1988. Rock engravings in Kakadu National Park. Unpublished paper presented at the First AURA Congress, Darwin, 29 August.

SWAN, L.M. 1996. The use of grinding hollows for ore milling in early Zimbabwe mettalurgy, Cookeia 1(6): 71-92.

TACON, P.S.C. 1990. The power of place: cross-cultural responses to natural and cultural landscapes of stone and earth, in J.M. Vastokas (ed.), Perspectives of the Canadian landscape: native traditions: 11-43. North York: Robarts Centre for Canadian Studies, York University.

1991. The power of stone: symbolic aspects of stone use and tool development in western Arnhem Land, Australia, Antiquity 65: 192-207.

1993. An assessment of rock art in the Mann River region, Arnhem Land, NT: a report to the Bawinanga Aboriginal Corporation and the Djomi Museum. Sydney: Australian Museum (lodged at AIATSIS, Canberra).

1994. Socialising landscapes: the long-term implications of signs, symbols and marks on the land, Archaeology in Oceania 29(3): 117-29.

TACON, P.S.C. & S. BROCKWELL. 1995. Arnhem Land prehistory in landscape, stone and paint, Antiquity 69: 676-95.

TACON, P.S.C. & C. CHIPPINDALE. 1994. Australia's ancient warriors: changing depictions of fighting in the rock art of Arnhem Land, NT, Cambridge Archaeological Journal 4(2): 211-48.

TACON, 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(3): 103-24.

TACON, P.S.C., S. OUZMAN, R. FULLAGAR & L. HEAD. In preparation. The function and meaning of north Australian cupules: a preliminary assessment.

THIERMANN, U. 1977. The dots of Pantiacolla [Peru], South American Explorer 1(1): 4-5.

TOBIAS, P.V. 1967. Rock peckings with an iron Age settlement and cemetery: a brief report on the Mooiplaas site, Bronkhorstspruit district, South African Archaeological Bulletin 22: 25-6.

VAN HOEK, M. 1997. The distribution of cup-and-ring motifs along the Atlantic seaboard of Europe, Rock Art Research 14(1): 3-17.

VIERECK, A. & J. RUDNER. 1957. Twyfelfontein - a centre of prehistoric art in South West Africa, South African Archaeological Bulletin 12: 15-26.

WALDERHAUG, E.M. 1995. Rock art and society in Neolithic Sogn og Fjordane, in Helskog & Olsen (ed.): 169-80.

WALKER, M.J. 1970. An analysis of British petroglyphs, Science and Archaeology 2/3: 30-61.

1977. Schematised rock carvings as archaeological evidence, in P.J. Ucko (ed.), Form in indigenous art: schematisation in the art of Aboriginal Australia and prehistoric Europe: 452-69. Canberra: Australian Institute of Aboriginal Studies.

WALKER, N.J. 1987. The dating of Zimbabwean rock art, Rock Art Research 4(2): 137-49.

WALSH, G. 1994. Bradshaws: ancient rock paintings of Australia. Geneva: Edition Limitee.

1996. Bradshaws in the spectrum of Kimberley art. Paper presented to The Origins and Significance of Northern Rock Art Seminar, Melbourne, 17 October.

WATCHMAN, A. 1987. Preliminary determinations of the age and composition of mineral salts on rock art surfaces in Kakadu National Park, in W.R. Ambrose & J.M.J. Mummery (ed.), Archaeometry: further Australasian studies: 36-42. Canberra: Australian National University. Occasional Paper in Prehistory 14.

1989. Comment on M. Nobbs & R. Darn, 'Age determinations for rock varnish formation within petroglyphs', Rock Art Research 6(1): 65-6.

1991. Age and composition of oxalate-rich crusts in the Northern Territory, Australia, Studies in Conservation 36: 24-32.

1992. Doubtful dates for Karolta engravings, Australian Aboriginal Studies 1992(1): 51-5.

1997. Sampling of accretions, Jinmium area, Northern Territory: a progress report. Unpublished. James Cook University, Cairns.

WATCHMAN, A., G. WALSH, G., M. MORWOOD & C. TUNIZ. 1997. AMS radiocarbon age estimates for early rock paintings in the Kimberley, N.W. Australia: preliminary results, Rock Art Research 14(1): 18-26.

WELCH, D. 1990. The bichrome art period in the Kimberley, Australia, Rock Art Research 7(2): 110-24.

1992. The early rock art of the Kimberley, Australia. Paper presented to Symposium F, Second AURA Congress, Cairns.

1993. Stylistic change in the Kimberley rock art, Australia, in Lorblanchet & Bahn (ed.): 99-113.

WILLIAMS, F.E. 1931. Papuan petroglyphs, Journal of the Royal Anthropological Institute 61: 121-55.

WRESCHNER, E. 1976. The potential significance of the pebbles with incisions and cupmarks from the Yarmukian of Shaar Hagalan, Israel, Bulletin de la Societe Royale Beige d'Anthropologie et de Prehistoire 87: 157-65.

WRIGHT, B.J. 1968. Rock art of the Pibara region, north-west Australia. Canberra: Australian Institute of Aboriginal Studies. Occasional Papers in Aboriginal Studies 11, Prehistory and Material Culture Series 2.