Maliwawa Figures—A Previously Undescribed Arnhem L and Rock Art Style

AUSTRALIAN ARCHAEOLOGY https://doi.org/10.1080/03122417.2020.1818361

ARTICLE Maliwawa figures—a previously undescribed Arnhem L and rock art style

Paul S. C. Tac¸ona , Sally K. Maya , Ronald Lamilamib, Fiona McKeaguea, Iain G. Johnstonc , Andrea Jalandonia , Daryl Wesleyd , Ines Domingo Sanze , Liam M. Bradyd , Duncan Wrightf and Joakim Goldhahng aPlace, Evolution and Rock Art Heritage Unit, Griffith Centre for Social and Cultural Research, Gold Coast Campus, Griffith University, QLD, Australia; bNamunidjbuk Aboriginal Corporation, Darwin, NT, Australia; cAustralian Institute of Aboriginal and Torres Strait Islander Studies, Acton, ACT, Australia; dCollege of Humanities, Arts and Social Sciences, Flinders University, Adelaide, SA, Australia; eICREA, Universitat de Barcelona, Barcelona, Spain; fSchool of Archaeology and Anthropology, Australian National University, Canberra, ACT, Australia; gCentre for Rock Art Research þ Management, School of Social Sciences, University of Western Australia, Perth, WA, Australia

ABSTRACT ARTICLE HISTORY In this paper, a previously undescribed rock art style consisting of large human figures and Received 5 June 2020 animals with stroke-line infill is introduced. These depictions have been named Maliwawa Accepted 30 August 2020 Figures. They are primarily found in northwest Arnhem Land and to date have been KEYWORDS recorded at 87 sites from Awunbarna (Mount Borradaile area) to the Namunidjbuk clan state Rock paintings; Arnhem of the Wellington Range. There are solitary figures and others arranged in compositions or Land; Maliwawa; scenes. We describe the features of this style, its distribution, subject matter and probable chronology; scenes; age. The results of a detailed analysis of all sites are discussed and a new, refined Arnhem Namunidjbuk; Awunbarna Land rock art chronology is presented. It is concluded that Maliwawa Figures are most likely to date between 6,000 to 9,400 years of age and to be contemporaneous with Northern Running Figures and Yam Figures found at sites to the south.

Abbreviations: Awum: Awunbarna; Nam: Namunidjbuk; m: metre

Introduction Arnhem Land rock art, spanning tens of millenia, have also been proposed (e.g. Brandl 1973; Western Arnhem Land, including Kakadu National Chaloupka 1993; Chippindale and Taçon 1993, Park, has a remarkable range and number of rock art sites with some sites known to the outside world 1998; David et al. 2013; Gunn 2016, 2018; Jones since the 1800s. Focussed rock art research since the et al. 2017; Lewis 1988; Taçon 1987; Taçon and late 1940s has seen increasingly refined chronologies Brockwell 1995; Taçon and Chippindale 2008; vari- across a temporal span of tens of thousands of ous papers in David et al. 2017). Although there are years, with regional differences defined (see David debates about terms to be used and the age of west et al. 2017 for a review). Several thousand sites have Arnhem Land rock art styles, essentially the now been documented and each year new discov- sequence from oldest to most recent consists of eries are made by various teams of scholars working large naturalistic animals, Dynamic Figures, a mix closely with Aboriginal communities. May et al. of forms and styles including Yam Figures, Simple (2015), and May and Taçon (2014) have recently Figures, Northern Running Figures and early X-ray summarised the nature of this rock art based on paintings, and then detailed X-ray art, a range of detailed studies such as those of Brandl (1973), stick figures, and designs made of beeswax. Stencils Chaloupka (1984a, 1993), Jelinek (1989), Lewis run through the entire sequence. Despite profound (1988) and Taçon (1989). Adding to this knowledge change brought about by the arrival of people from base, a number of recent PhD theses have explored Asia and Europe since at least the 1600s (Taçon different aspects of the rock art, focussing on spe- et al. 2010), rock paintings continued to be made cific areas within the region (e.g. Gunn 2016; regularly through the contact period up to the mid- Hayward 2016; Johnston 2018; Jones 2017; Marshall 1980s (e.g. Taçon 1992) and have occasionally been under review; Wesley 2014). made up to the present. Both recent and older rock After undertaking vast amounts of fieldwork, art remains important for Kakadu-Arnhem Land various forms, styles and periods of art have been communities today (e.g. Brady et al. 2020; May defined. Varying stylistic chronologies for western et al. 2019, 2020a, 2020b; Taçon 2019).

CONTACT Paul S. C. Tac¸on [email protected] PERAHU, Griffith Centre for Social and Cultural Research, Gold Coast campus, Griffith University, QLD, 4222, Australia ß 2020 Australian Archaeological Association 2 P. S. C. TAÇON ET AL.

Figure 1. Map of Kakadu/Arnhem Land showing the general location of the Awunbarna and Namunidjbuk areas (Produced by A. Jalandoni; base map by Stamen Design [OpenStreetMap]).

In this paper, we introduce a previously unrecog- were very large, even life-size, and composed in nised style of rock painting that is primarily found scenes. They differed from most large human figures at sites in northwest Arnhem Land and report on that occur in other parts of Kakadu-Arnhem Land. some of this rock art from Awunbarna (Mount During the 2016–2018 field seasons numerous exam- Borradaile area) and the Namunidjbuk clan estate of ples of these figures emerged, including on previously the Wellington Range (Figure 1). Large- to medium- undocumented large panels with a range of subject sized human figures and animals dominate this style matter. Unusual scenes, compositions and activities in terms of subject matter. Besides describing them were noted and it soon became apparent that the in detail we also place the style in an updated and style was an important missing link from the middle refined Arnhem Land rock art chronology. period of the known Arnhem Land rock art chronology (see below). In 2018–2019, nine sites in the Awunbarna area, over 100 kilometres southwest of Background the Wellington Range, were also documented with Between 2008 and 2011, we recorded 146 rock art sites figures in this recently identified style. in the Namunidjbuk Estate of the Wellington Range in At the end of the 2016 field season, senior northwest Arnhem Land with Aboriginal Traditional Traditional Owner Ronald Lamilami decided to Owners. This was part of the Picturing Change and the give the style and figures a local Aboriginal name Baijini, Macassans, Balanda and Bininj research projects in his Mawng language. He chose to name the that explored very recent rock art, especially from the style after an area within his traditional clan estate, Asian-European contact period (e.g. May et al. 2010; ‘Maliwawa’ (also spelt ‘Maliwarl’), that has a range Tac¸on et al. 2010, 2012;Wesley2014). In 2016–2018, as of important rock art sites with imagery from vari- part of the History Places research project, a further 206 ous periods (May et al. 2010) and some highly sig- rock art sites were located and recorded during exten- nificant panels of these figures. We have used the sive surveys with Aboriginal research partners, bringing term Maliwawa Figures since and introduce the the total to 352. term and style here while also acknowledging that In 2008, we began to notice that at some this style also appears in other clan lands. In this Wellington Range rock art sites there were unusual paper, we describe these distinctive figures in detail groups of human and animal figures in a style unlike after reviewing the evidence and chronology of those previously described for western Arnhem Land large human and animal figures from elsewhere in or elsewhere in Australia. Often the human figures Kakadu-Arnhem Land. AUSTRALIAN ARCHAEOLOGY 3

Kakadu-Arnhem land large naturalistic humans Land (Chaloupka 1993:104, Figure 100). However, and animals Lewis (1988:72) concludes that Chaloupka’s large naturalistic period of animals and humans is poorly ‘Large Naturalistic Designs’ were first described in described and that most of these paintings are much detail by Chaloupka (1977:248) who argued that more recent than Chaloupka believed. ‘An alterna- ‘The oldest paintings at the Mt. Gilruth sites are of tive, and in my opinion, more likely possibility is large anthropomorphs and animal designs … exe- that most of the ‘large naturalistic style paintings’ of cuted as line paintings, in which the whole of the the type Chaloupka has published post-date ‘early’ figure is filled with lines which follow the outline of Mimi art and predate early X-ray art’ the body, its appendages and contours’. Brandl (Lewis 1988:72). (1973:183) had earlier proposed the oldest Arnhem Haskovec (1992) came to similar conclusions Land rock art phase as ‘red crudely painted animals after revisiting Mt. Gilruth sites critical for estab- and anthropomorphic figures’; however, he does not lishing Chaloupka’s sequence. After detailed ana- describe this phase in any more detail. Chaloupka lysis, he found no support for the contention that (1977:248, 1985:274) noted that many of the large large humans and animals were made early in the naturalistic animals were speared with three- sequence and states ‘It is my opinion that the pronged multibarbed spears and sometimes were Large Naturalistic style evolved from or immedi- associated with schematised human figures that ately followed the original Yam style’ have three-pronged spears, single pronged multi- (Haskovec 1992:70). ‘ ’ barbed spears and what he calls a fighting pick . About the same time in the early 1990s, ‘ Curiously, Chaloupka (1985:274) remarks The fight- Chippindale and Taçon (1993) undertook a detailed ing pick appears again in the second last style of analysis of two large rock art panels in Kakadu this period, whilst the three pronged spear came National Park and showed that large naturalistic into use again during the following estuarine animals do occur under most earlier forms of rock ’ period . It is unusual for distinctive material culture art, as Chaloupka contended, but not large natural- to disappear from use/depiction and then reappear istic humans. After documenting and analysing a thousands of years later, so this hints at a problem few hundred sites in the 1990s, they reconfirmed ’ with Chaloupka s chronology. this finding time and again, resulting in their pub- In relation to a large naturalistic female anthro- lished chronology only having large naturalistic ani- ‘ pomorph he stated The shape of its head is similar mals early in the sequence without humans ‘ ’ ’ to that of the later yam representations (Chippindale and Taçon 1998:106; Taçon and (Chaloupka 1977:248). This also suggests these fig- Chippindale 1994:215, Table 1). Instead, they argue ures are more recent in the Arnhem Land chron- Chaloupka’s Dynamic Figures are the first depic- ology. Chaloupka also noted that for large tions of the human form (see also May et al. 2018). naturalistic human figures ‘Noticeable attention was In 1992, Taçon and Chippindale recorded a pre- given to faces shown in profile with recognisable viously undocumented panel with two life-size features, and also to hands and feet’ (Chaloupka human figures above Jim Jim Creek near the 1985:274). For Chaloupka (1993:102): Kakadu—Arnhem Land border (see Taçon and A typical figure is depicted with the head in profile, Chippindale 2008:91 and their Figure 5.9). The fig- detailing its facial features and the shape of the ures have yam-like heads and appear to float hori- cranium. The arms are held open and outstretched, zontally across the panel. Because of these features or, in some horizontal figures, hanging down at and the weapons they hold, Taçon and right angles to the body. Although the upper part Chippindale (2008) place them in the of the body is depicted frontally, the lower part is – slightly turned aside, showing a slightly swelling 4,000 6,000 year period that includes Yam Figures abdomen and penis, and, on the opposite side, the and Simple Figures: buttocks. The legs are bent at the knees, with the … Some of these human-like figures hold the three- feet protruding down Most of the figures have prong spears, spearthrowers and other artefacts an unearthly aura about them, appearing as if characteristic of this transitional period (Figure floating in space. So far, they have not been found 5.9). Occasionally, animals (mostly macropods) are engaged in any activities or associated with shown with a three-pronged spear piercing hind, weapons or implements. back or stomach (Figure 5.10). Some of the human Chaloupka (1993:102) states that most of these and macropod figures have tassels, bobbles or other designs characteristic of classic Yam figures but figures are male, depictions of females are rare and used sparingly. These features, along with the ‘ the majority of human figures are individual repre- artefacts, help place the figures. In some ways they sentations’. He also notes that some have detailed appear transitional between Yam Figures/Simple X-ray features (1993:102) and illustrates this with an Figures and the solid infill/X-ray paintings of the example from the Liverpool River area of Arnhem most recent phase. 4 P. S. C. TAÇON ET AL.

In summary, Chaloupka (1977, 1985, 1993) con- tends that large naturalistic humans and animals were made before Dynamic Figures and could have a Pleistocene age. Lewis (1988) and Haskovec (1992) argue that large naturalistic humans and animals postdate Dynamic Figures and were made in the mid-Holocene. Tac¸on and Chippindale (2008; and see Chippindale and Tac¸on 1993, 1998), on the other hand, conclude that large paintings of animals both postdate and predate Dynamic Figures but large naturalistic humans were only made after Dynamic Figures. Haskovec (1992:70) also argues that large naturalistic humans and animals ‘evolved from or immediately followed the original Yam style’. With new data from northwest Arnhem Land, we set out to test these propositions and to update the Arnhem Land chronology accordingly.

Maliwawa figure description General features Maliwawa Figures comprise mainly naturalistic human and animal forms with stroke-line infill throughout but occasionally they are in outline with only a few strokes of infill. The stroke infill typically runs parallel to the limbs and to the length of the figure. Maliwawa Figures are often large (over 50 cm Figure 2. Large male Maliwawa human figures from an Awunbarna site. The largest male is 1.15 metres wide by high), sometimes life-size (Figure 2), although there 1.95 metres high (Photo: P. Tac¸on). are also some small Maliwawa figures 20–50 cm in height. Generally, they are larger than Dynamic related recent paintings. A few Maliwawa females Figures. Maliwawa Figures are primarily found in are also shown with an extended abdomen. various shades of red to mulberry. At two sites, ani- The arms of human figures are often bent with mal and human figures have some faded dark yel- the forearm and hands shown at a right angle or low with the red, suggesting many Maliwawas may lowered at an obtuse angle in relation to the body, once have been bichrome paintings. Almost all mirroring the feet. Sometimes hands/fingers are also Maliwawa Figures were painted, but there is one drooping downwards. Defined fingers and toes drawing of a Maliwawa human figure. (although not always five of each) and defined Unlike Simple Figures, Maliwawa human figures muscles in the legs (especially calves) are common. have fully formed musculature akin to X-ray paint- Indeed, there is much definition of the legs, showing ings. Maliwawas are in less animated poses than musculature and realistic foot details. The angle of Dynamic Figures but more animated than Simple, the foot between the foot and the lower leg tends Yam or X-ray figures (e.g. Johnston 2017). towards obtuse, giving the impression of ‘floating’, However, some are static or appear to be ‘floating’. ‘dream-like’, ungrounded figures. The head of Often Maliwawa human bodies are somewhat human figures is more likely to be orientated or twisted with different aspects of perspective shown. tilted upwards/backwards, rather than down. Human figures were classified as male if male geni- Maliwawa human figures appear animated and are talia were depicted, female if they had breasts and often depicted doing things but not in vigorous no male genitalia, and indeterminate if they had nei- motion the way that Dynamic Figures are usually ther. Some male figures and others that lack sex- portrayed (see Johnston 2018:224). specific characteristics have hair or headdresses Generally, Maliwawa paintings do not have X-ray shown (see below). Maliwawa males are usually in features but a few depictions of animals have rudi- profile and often have a bulging stomach above a mentary body cavities depicted, suggestive of a first penis. Penises are rarely depicted in Dynamic Figure step towards early X-ray paintings in which body paintings (Chaloupka 1984b; Johnston 2018; May cavities feature. Maliwawa human figures are almost et al. 2018) but are more common for X-ray and as detailed in terms of body features as Dynamic AUSTRALIAN ARCHAEOLOGY 5

Figure 4. A male Maliwawa human figure with cone and feather headdress (0.80 metres wide by 1.44 metres high) reaches out to a large bird with one hand and a large macropod with the other (possibly an emu and kangaroo), Namunidjbuk. Between the macropod and the human figure is a theriantrope (Photo: I. Domingo Sanz).

Figure 3. Back-to-back Maliwawa macropods in the ‘piano player’ pose, Namuidjbuk (Photo: P. Tac¸on).

Figures but have much less material culture depicted. As is typical of early Arnhem Land animal depictions, Maliwawa animals are usually in profile. Some macropods are shown in a human-like sitting pose with front paws in front, resembling a sitting person playing a piano (Figure 3). Tracks and geo- Figure 5. Digital tracing of Figure 4. The therianthrope has metric designs are rare. Only one animal-headed a macropod-like head and human body. It is positioned being has been recorded, a therianthrope with a between the macropod and the male Maliwawa figure giving the suggestion of transformation from animal to animal- human-like body and a macropod-like head. The human to human (Digital tracing: Fiona Brady). macropod-like head is similar to some Dynamic Figure theriantropes but the legs and arms are Maliwawa-like (Figures 4 and 5). Therianthropes are Land at a total of 87 sites, at several locations within more common in Dynamic Figure rock paintings, Awunbarna in the west to sites at the eastern end of with 14 recorded (4.7% of all Dynamic Figures) in the Namunidjbuk clan estate of the Wellington the Jabiluka lease area (Johnston 2018; May et al. Range, which are about 130 kilometres northeast 2018:73) and 45 others from the rest of Kakadu- of the westernmost Awunbarna site. There may be Arnhem Land analysed by Tac¸on and Chippindale others at sites between these areas but a review of (2001:190–196). Chaloupka (1993:112) concluded site records made by Chaloupka did not indicate they comprise 3% of Dynamic Figure paintings. any. At Awunbarna, Maliwawas are found at nine of 39 sites (23.08%) we recorded while at Namunidjbuk they are found at 78 of 352 sites Distribution (22.16%). Roberts and Parker (2003:37) were the Most Maliwawa Figures are in accessible/visible pla- first to publish a photograph of a panel of what we ces at low landscape elevations rather than hidden are calling Maliwawa Figures, from one of the away or at shelters high in the landscape. Maliwawa Awunbarna sites we studied, although they did not Figures were recorded in northwestern Arnhem explore their significance. 6 P. S. C. TAÇON ET AL.

Further south and west in Kakadu-Arnhem Land, Namunidjbuk has a greater proportion of macro- Maliwawa Figures are much scarcer. A handful of pods, snakes, flying foxes and longtom fish than definitive sites with Maliwawa Figures are found on Awunbarna. In contrast, Awunbarna has a higher the Arnhem Land plateau, the best of which is on percentage of birds, turtles, thylacines and ‘bilbies’ Upper Jim Jim Creek near the Kakadu-Arnhem than was found at Namunidjbuk sites. Namunidjbuk Land border (see Tac¸on and Chippindale 2008:91, has five sets of back-to-back macropods (Figure 3), Figure 5.9 and 2008:78, as well as quote above) and something more common in the recent rock art of an Upper Twin Falls Creek site with many contem- western Arnhem Land. Awunbarna has two back-to- ‘ ’ porary Yam Figures (e.g. see Chaloupka back bilbies (Figure 6). There are seven depictions 1993:140–141; Figure 140). A probable Maliwawa of animals long extinct in the Arnhem Land region scene is located at a site near the East Alligator (2.19% of faunal subject matter), consisting of four River and, in 1993, two large Maliwawa-like males were recorded in central Arnhem Land (Tac¸on Table 2. Maliwawa faunal subject matter. – 1993:53 54, Figures 24 and 25). Also, Brandl Sites/animals Nam Per cent Awun Per cent Total Per cent (1973:95, Figure 221) recorded one female Macropod 136 48.57% 11 28.21% 147 46.08% Maliwawa-like figure at the Cadell River. It is Bird 28 10.00% 10 25.64% 38 11.91% Snake 32 11.43% 3 7.69% 35 10.97% important to note, however, that we are convinced Longtom fish 18 6.43% 1 2.56% 19 5.96% many more examples of this style exist and have not Indeterminate fish 14 5.00% 1 2.56% 15 4.70% Flying fox 13 4.64% 1 2.56% 14 4.39% yet been identified, including in the work of other Goanna 9 3.21% 1 2.56% 10 3.13% researchers. Thylacine 6 2.14% 2 5.13% 8 2.51% Turtle 4 1.43% 3 7.69% 7 2.19% Brush or ring 5 1.79% 1 2.56% 6 1.88% tail possum Maliwawa subject matter Eel-tail catfish 3 1.07% 1 2.56% 4 1.25% Echidna 4 1.43% 0 0.00% 4 1.25% A detailed analysis of the subject matter of all 87 Bilby 0 0.00% 3 7.69% 3 0.94% sites was undertaken. In total, we recorded 572 Northern quoll 1 0.36% 0 0.00% 1 0.31% Bandicoot 1 0.36% 0 0.00% 1 0.31% Maliwawa motifs consisting of 319 animals Sugar glider 1 0.36% 0 0.00% 1 0.31% (55.77%), 240 humans (41.96%), 8 geometric designs Dugong 1 0.36% 0 0.00% 1 0.31% Unidentified 4 1.43% 1 2.56% 5 1.57% (1.40%), 4 macropod tracks (0.70%), and an animal- Total 280 100.01% 39 99.97% 319 99.97% headed being (Table 1). However, the Awunbarna sample is about two-thirds human (62.39%) and one-third animal (35.78%) while Namunidjbuk sites Table 3. Broad Maliwawa faunal categories. have more animal depictions (60.48%) than Sites/animals Nam Per cent Awun Per cent Total Per cent Mammals 168 60.00% 18 46.15% 186 58.31% human (37.15%). Reptiles 45 16.07% 7 17.95% 52 16.30% Birds 28 10.00% 10 25.64% 38 11.91% Fauna Fish 35 12.50% 3 7.69% 38 11.91% Unidentified 4 1.43% 1 2.56% 5 1.57% Animals are mostly in profile but occasionally are Total 280 100.00% 39 99.99% 319 100.00% depicted in split/dual perspectives. There are 18 faunal categories including ‘unidentified’. Macropods, birds and snakes are the most frequent subjects, in that order, for Namunidjbuk, Awunbarna and the entire sample (Table 2). Macropods, birds, snakes and longtom fish make up 74.92% of the total. More generally, mammals are most common (58.31%), followed by reptiles (16.30%), birds (11.91%) and fish (11.91%) (Table 3).

Table 1. Eighty-seven Maliwawa sites with 572 figures from the Namunidjbuk clan estate (Nam) and Awunbarna (Awun) were analysed in detail. Nam Percent Awun Percent Total Percent Human 172 37.15% 68 62.39% 240 41.96% Figure 6. Two Maliwawa style depictions of back-to-back Animal 280 60.48% 39 35.78% 319 55.77% Geometric 7 1.51% 1 0.92% 8 1.40% animals more similar to bilbies than any known Arnhem Macropod track 3 0.65% 1 0.92% 4 0.70% Land creature, Awunbarna. The figure on the left is 0.50 Animal-headed being 1 0.22% 0 0.00% 1 0.17% metres wide by 0.49 metres high while the right hand figure Total 463 100.01% 109 100.01% 572 100.00% is 0.52 metres wide by 0.53 metres high (Photo: P. Tac¸on). AUSTRALIAN ARCHAEOLOGY 7 thylacines and three ‘bilbies’. At one Namunidjbuk (4.82%) and a boomerang (1.20%). There are 10 site there is a rare depiction of a dugong. headdress forms. The most common is a ball headdress (33; 45.83% overall, 45.83% of Namunidjbuk Human figures headdresses and 61.54% of Awunbarna headdresses), A third of human depictions were classified as male followed by oval (10; 13.89%), cone (7; 9.72%), fea- because they have male genitalia. Females, identified ther (6; 8.33%) and rayed line (6; 8.33%). There are because breasts were shown, are rare, comprising three cone þ feather headdresses. A greater range of only 5.00% of human depictions. Almost 59% of headdress types was found at Naminidjbuk sites human figures could not be determined to be either compared to Awunbarna, as well as in association male or female because they lack sex-specific charac- with indeterminate human figures compared to teristics and thus are labelled ‘indeterminate human males. At Awunbarna, three males have a bag figure’ (Table 4). Most human figures face right, 99 shown suspended from the neck and also have (41.25%), 86 (35.83%) face left and 50 (20.83%) are headdresses, while an indeterminate figure carries a obverse (although five of these may be shown from bag in its hand. Barbed spears are held by one male behind). For males, 58 (66.66%) face right, 25 at one Namunidjbuk site, three males at a second (28.74%) face left, 4 (4.60%) are obverse. This com- site, and two indeterminate figures at a third pares to 40 (28.37%) indeterminate human figures Namunidjbuk site. At one Namunidjbuk site, a facing right (including three sitting), 58 (41.13%) barbed spear was depicted through a macropod’s facing left and 40 (28.37%) obverse. One lone drawing is of an indeterminate figure facing left and shown leaning over (Table 5). Half (6) of all females are obverse (Figure 7), three face left and only one faces right. One is shown in obverse and facing left in pronounced twisted perspective. Another female is upside down next to a male with their genitals in proximity to each other. Several males and indeterminates in both Awunbarna and Namunidjbuk are depicted in various stages of bending over. Two indeterminate human figures at Awunbarna sites, and one at Namunidjbuk, are shown sitting. Another at Awunbarna is on its knees while one is lying on its back at Namunidjbuk and three appear to be crawling. Human figures generally have round-shaped or oval-shaped heads; some have lines on the head suggestive of hair; others have a range of headdresses. One unique indeterminate human figure has hair- like lines all over its body (Figure 8). A total of 83 (34.58%) of 240 human figures are shown with material culture, consisting of 72 headdresses (86.75%), six barbed spears (7.23%), four bags

Table 4. Maliwawa human figures analysed. Site Nam Per cent Awun Per cent Total Per cent Male 62 36.05% 25 36.76% 87 36.25% Female 11 6.40% 1 1.47% 12 5.00% Figure 7. Obverse female Maliwawa with arms up in the air, Indeterminate 99 57.56% 42 61.76% 141 58.75% a small head and large breasts like later Simple Figure Total 172 100.01% 68 99.99% 240 100.00% females, Namunidjbuk (Photo: P. Tac¸on).

Table 5. Human Figure orientation. Nam Awun Site total Nam orientation total Awun orientation Total Total Range Male 62 42r, 18l, 2o 25 16r, 7l, 2o 87 58r (66.66%), 25l (28.74%), 4o (4.60%) Female 11 1r, 2l, 6o, 1lo, 1oud 1 1l 12 1r, 3l, 6o, 1lo, 1oud Indeterminate 99 24r, 39l, 35o, 1sr 42 16r, 19l, 5o, 2sr 141 40r (28.37%), 58l (41.13%), 40o (28.37%), 2sr, 1sr Total 172 67r, 60l, 43o, 1lo, 1oud 68 31r, 26l, 7o, 2sr 240 99r (41.25%), 86l (35.83%), 50o (20.83%), 2sr, 1sr, 1lo, 1oud l: left; r: right; o: obverse/reverse; s: sitting; ud: upside down. 8 P. S. C. TAÇON ET AL.

Scenes/compositions/themes As with earlier Dynamic Figure rock art (e.g. see Chaloupka 1984b; Johnston 2018; May et al. 2018), some Maliwawa Figures are arranged in complex scenes of activity. Importantly, human figures are frequently depicted with animals, especially macropods, and these animal–human relationships appear to be central to the artists’ message. Another key theme at Awunbarna sites is a male or indeterminate human figure holding an animal, another human figure or an object. For instance, at one site an obverse male holds a turtle. At another site, an obverse male holds a large bird. At a third location, an indeterminate human figure is depicted bending over holding a snake by the tail (Figures 9 and 10). Some human figures reach down to others or hold hands. For example, two indeterminate human figures facing opposite directions hold hands (Figures 9 and 10). At another Awunbarna site, a male Maliwawa reaches down and holds the head of a shorter male facing him and a female stands behind the taller male. At one site, two males standing side- by-side are facing right with the slightly taller forward male reaching back so that his arm is around the rear figure’s shoulders and he is looking back to him. At another site, an indeterminate human figure Figure 8. Indeterminate Maliwawa human with lines sug- holds a bag. gestive of hair all over its body, Awunbarna (Photo: Occasionally there are rows of male and indeter- P. Tac¸on). minant human figures. In one row, three males wear a bag strung from the neck. There are 11 indeterminate human figures or males in various stages of bending over, including the one holding a snake by the tail mentioned above. Another indeterminate human figure was portrayed as if on its knees. At one site, there is a sitting male with possible faeces below its bottom. All of these themes can also be found at Namunidjbuk sites. For instance, holding something, especially a snake, is common. At one site, a male is holding a snake with both hands and the blunt end of a spear. At another, back-to-back indeterminate figures with ball headdresses are bending down and one holds a snake by the tail. At a third site, a male is shown holding an enormous snake with both hands around the snake’s lower body. In a fourth Figure 9. Awunbarna panel of Maliwawa figures including rock shelter, a male and an indeterminate figure are two indeterminate humans holding hands, a large male with shown next to each other with hands on/next to cone and feather headdress (0.65 metres wide by 1.45 metres high) and an indeterminate human bending forward their hips and the indeterminate figure is standing to hold a large snake by the tail (Photo: P. Tac¸on). on the tail of a snake. At two other sites a male holds a snake. At another locality, a snake encircles three ovals. Elsewhere, a male is shown reaching down hold- neck but it is not in relation to a human figure. All ing what appears to be a dead bird. An unusual barbed spears are single-sided. Only one figure, composition consists of a male Maliwawa human from Namunidjbuk, holds a boomerang. No female figure with cone plus feather headdress (0.80 metres figures have headdresses, bags or spears (Table 6). wide by 1.44 metres high) reaching out to a large AUSTRALIAN ARCHAEOLOGY 9

Figure 10. Digital tracing of Figure 9. The three figures at the left are lighter because they are in a lighter shade of red on the shelter wall (Digital tracing: Fiona Brady).

Table 6. Human Figures with material culture Site Nam Total Nam Range Awun Total Awun Range Total Total Range Male 36 (32h or 36 ¼ 14hb (43.75% of 11 (8h or 72.73%) 5hb (62.50% of h), 47 (40h or 47 ¼ 19hb (47.50% of 88.89%) h), 4ho, 4hc, 3hf, 2hf, 1hc, 3b 85.11%) h), 5hf, 5hc, 4ho, 2hrl, 3hbrl, 1hbf, 2hrl, 3hbrl, 1hbf, 1hcf, 4s 1hcf, 3b, 4s Female 0 0 0 0 0 0 Indeterminate 30 (27 h or 30 ¼ 11hb (40.74% of 6 (5h or 83.33%) 3hb (60.00% of h), 36 (32 h or 36 ¼ 14hb (43.75% of 90%) h), 5ho, 4hrl, 2hc, 1 ho, 1hdl, 1b 88.89%) h), 6ho, 4hrl, 2hc, 2hcf, 1hf, 1hl 1hbf, 2hcf, 1hdl, 1hf, 2s, 1bo 1hl, 1hbf, 1b, 2s, 1bo Total 66 (59h or 66 ¼ 25hb (42.37% of 17 (13h or 76.47%) 8hb (61.54% of h), 83 (72 h or 83 ¼ 33hb (45.83% of 89.39%) h), 9ho, 6hc, 6hrl, 2hf, 1hc, 1ho, 86.75%) h), 10ho, 7hc, 6hf, 4hf, 3hcf, 3hbrl, 1hdl, 4b 6hrl, 3hcf, 2hbf, 2hbf, 1hl, 6s, 1bo 3hbrl, 1hl, 1hdl, 4b, 6s, 1bo h: headdress; hb: ball headdress; ho: oval headdress; hc: cone headdress; hf: feather-shaped headdress; hrl: rayed line headdress; hcf: cone and feather headdress; hbf: ball and feather headdress; hbrl: ball and rayed line headdress; hl: line headdress; hdl: double line headdress; b: bag; s: spear; bo: boomerang. bird with one hand and a large macropod with the stands behind the male on the left. A vertical other (possibly an emu and kangaroo). Between the macropod is behind the female. A second male fac- macropod and the human figure is a therianthrope. ing the first and the indeterminate human figure in The therianthrope has a macropod-like head and the middle also reaches down towards the head of human body. It is positioned between the macropod the indeterminate figure. Both males have ball head- and the male Maliwawa figure giving the suggestion dresses. It is somewhat similar to another of transformation from animal to animal-human to Namunidjbuk scene and one at Awunbarna in terms human (Figures 4 and 5). At other sites, an indeter- of the number and arrangement of the figures. minate holds a bird, a bird holds an oval with its There are two sets of back-to-back indeterminate small wings, a large indeterminate figure bending human figures, including the set mentioned above, forward to the left holds a boomerang, and a male and two sets of back-to-back macropods at the same holds a fish. At one site, a thylacine appears to be site as one of the sets of back-to-back humans. climbing/holding something below a male Three other sites have back-to-back macropods Maliwawa figure. while at another site two macropods face each other. On one rock shelter wall, two males with ball Some sites have rows of humans or rows of macro- headdresses face each other and reach down to a pods. At one location, there is a row of six males shorter obverse indeterminate human figure. On and an indeterminate figure on a boulder across another wall, a male Maliwawa reaches down to from a row of three macropods. Three males hold hold the head of a shorter indeterminate human fig- barbed spears. At another site there is a row of ani- ure facing away from him (Figure 11). A female mals and humans including a male with a snake 10 P. S. C. TAÇON ET AL.

Figure 13. Maliwawa macropod over 3MFC hand stencil, Namunidjbuk (Photo: P. Tac¸on).

and holding their hands close to their heads (Figure 12). Elsewhere at this site a small indeterminate human facing left next to a male facing left has similar lines coming from its bottom. At another site, two indeterminate figures with their backs to us are side by side and something is coming out of one’s bottom. At one site, a male figure has dots coming out of, and around, its penis. Three sites have scenes suggestive of sexual intercourse. The first has a male and female shown with the female upside down near the male and with Figure 11. Scene of two male Maliwawas with ball head- their genitals in close proximity to each other. At a dresses reaching down to a shorter indeterminate human second site, where the male with dots coming out of figure with a snake behind the male on the right and its penis is located, a male and an indeterminate behind the left male a female and a macropod, human figure appear to be engaged in intercourse. Namunidjbuk. An indeterminate human figure with a cone and feather headdress is above (Photo: P. Tac¸on). At a third site, an indeterminate human figure is standing next to a male that is bending over as if the two are engaged in intercourse. Interestingly, there is only one scene that could be related to hunting in this style, unlike with Dynamic Figures, X-ray paintings and recent stick figure compositions. However, as mentioned above, there is also one Maliwawa macropod that has a barbed spear through its neck.

Chronology and possible age Maliwawa figures were noted to be superimposed under Simple Figures at two sites and over Dynamic Figures at four locations. At three sites a Maliwawa Figure 12. Three Maliwawa figures bending over with lines human or animal figure is over 3 Middle Finger coming from their faces suggestive of vomiting, Closed (3MF) hand stencils (e.g. Figure 13) and at Namunidjbuk (Photo: P. Tac¸on). one site a Maliwawa macropod is over a 2 Middle Finger Closed (2MF) hand stencil. These types of partly wrapped around his torso. One Namunidjbuk stencils are often associated with Dynamic Figures site has a life size male 1.8 m high in a ‘flying/float- (Chaloupka 1984b; Johnston 2018) but some appear ing’ stance with a crawling indeterminate human under Dynamic Figures at Namunidjbuk sites. figure reaching up to the male’s waist. At various Maliwawa animals are typical of a freshwater envir- sites, human figures are shown bending over. This onment rather than estuarine (e.g. freshwater fish includes a scene of three indeterminate human fig- such as longtom; no barramundi or fork-tail catfish; ures with lines coming from their faces leaning over macropods and snakes common). Some Maliwawa AUSTRALIAN ARCHAEOLOGY 11

Table 7. New refined chronology for Arnhem Land rock art highlighting inclusion of Maliwawa style (shaded). Main Rock Art Styles Estimated Years BP and Techniques Regionalism Environment Unknown but pre-18,000 Various types of hand stencils and Pan Arnhem Land Variable animal stencils; object and hand prints 18,000 15,000 Large Naturalistic Animals (including Pan Arnhem Land Arid and cool various extinct animals); hand stencils 15,000 13,000 Large Naturalistic Animals (including Pan Arnhem Land Arid and rising temperatures; some extinct animals); sea level 120–140 m hand stencils below present 13,000 10,000 Dynamic Figures (mostly human but Pan Arnhem Land with Arid; sea levels rise some animals, including the now regionalism developing rapidly; flooding extinct thylacine; very small to life size) and associated stencils, especially material culture; grass prints 10,000 6,000 Maliwawa Figures, Northern Running Highly regionalised Sea levels rise more slowly; Figures, and Yam Figures; some increasing rainfall; flooding large animals; rare thylacine depictions; Simple Figures start to appear; stencils 6,000 4,000 Multiple new styles and subjects (e.g. Simple Figures are Pan Arnhem Sea level stabilised; maximum dingo) appearing including Land but there are several effective precipitation moderate to small sized Simple forms that vary in Figures and Bininj Kunburrk concentrations regionally; Figures; stencils Kunburrk Figures are in one region only. 4,000 2,000 Early X-ray and associated paintings Pan Arnhem Land with ENSO begins to dominate; of humans and animals (moderate regionalism developing increasing climatic size), stencils and prints that variability; sharply develop into the Complete decreasing rainfall Figure Style 2,000 – 500 Detailed X-ray and other aspects of Regionalisation Increasing climatic variability; the Complete Figure Style, decreased rainfall including large humans and animals, stencils and prints 500 - present X-ray and other aspects of the Regionalisation Contact period and changing Complete Figure Style plus cultural and climatic introduced subject matter, environments including large humans and animals, stencils and prints human figures are associated with Maliwawa thyla- setting provides context to our chronology, we note cines. These observations, along with data from that the environment is not the key stylistic deter- recent literature (e.g. papers in David et al. 2017; minate player in rock art change. We argue that Gunn 2018; Hayward 2016; Johnston 2018; Jones rock art stylistic change is influenced by both envir- 2017; Jones et al. 2017; Wesley 2014), and the past onmental and cultural factors, such as contact with decade of our fieldwork in various parts of western other groups of people, population increase and, in Arnhem Land, has led to a new refined rock art other parts of the world, the domestication of ani- chronology that fleshes out the middle of the mals and plants. sequence and indicates that Maliwawa, Yam and However, there were shifting forms of interplay Northern Running Figures were regional traditions between environmental and cultural influences on from the same time period with some overlap. They rock art change in the past 20,000 or more years in were also transitional styles between Dynamic Australia. As June Ross (2013:165) notes in her pan- Figures and later Simple Figures, X-ray humans and Australian study of regionally distinct rock-art animals and other recent forms of rock art styles, ‘alterations noted in the archaeological record (Table 7). such as the introduction of a new art style are This detailed chronology (primarily developed by viewed as adaptive strategies aimed at ameliorating PT, SM and IJ) is based on past proposed chronolo- changing ecological conditions’. Ross (2013:165) also gies by various researchers including our own (sum- suggests there is more to this story than just adapta- marised in David et al. 2017), and data collected tion to environmental change. She argues that: ‘each from several locations in Kakadu National Park and of the varied art assemblages were introduced at a in northwest Arnhem Land between 2008–2019, time when the relationship between people and recent rock art research elsewhere across Kakadu- place was under pressure, whether the pressure Arnhem Land by others (see above), and the latest came from outside intruders, rising sea levels, popu- environmental records. Although the environmental lation increase or the introduction of more intense 12 P. S. C. TAÇON ET AL. interaction in the form of trade networks, or a com- NRF art style occurs during the bination of these factors’. Pleistocene–Holocene transition, a period which It should also be noted that pecked/pounded palaeoenvironmental research and archaeological modelling suggest was one of major transformation cupules were made at several locations and probably for Indigenous Australian societies. during different time periods; drawings are rare and mostly confined to the past few hundred years However, Jones et al. (2017) obtained radiocar- except for a pair of large drawn thylacines that are bon age constraints by dating oxalate crusts. This probably at least 5,000 years of age, two drawn technique is controversial so although the age con- Dynamic/Post-Dynamic Figures (Gunn and Whear straints they obtained make sense, future confirma- 2007) and one Maliwawa Figure; engraved animal tions are needed. What is more secure is that (bird and macropod) tracks are found at only a few Northern Running Figures appear after Dynamic locations and could be of considerable age given Figures in the rock art chronology and that they are patinas and crusts over them; engraved figures are regionally constrained to near the East Alligator rare, but there is one site with some small fish pet- River (Jones and May 2017). Further south, it is roglyphs and another with pecked human feet/foot- Yam Figures that appear in the rock art sequence prints. Hand and material culture stencils run after Dynamic Figures. To the north, we have found through the entire western Arnhem Land rock art Maliwawa Figures occur in the rock art sequence sequence, but some forms of hand stencils, such as after Dynamic Figures and there are no instances of three or two middle fingers closed variants, are very either Northern Running Figures or Yam Figures at early in the sequence while open hand stencils are Awunbarna or Namunidjbuk. Some Maliwawa, found in every period. Northern Running and Yam Figures are reminiscent of Dynamic Figures in varying ways, for instance, the Dynamic nature of the bodies of Northern Constraining the age of Maliwawa figures Running Figures, the Maliwawa fauna and range of Taçon and Brockwell (1995:689) used various lines headdress types and some Yam Figure heads/head- of evidence to argue that Dynamic Figures were dresses resemble those of Dynamic Figures. most likely to have been made between 12,000 to The Maliwawa Figures are younger than 13,000 years ago, no younger than 10,000 years and Dynamic Figures but older than X-ray paintings and no older than 15,000 to 16,000 years. Various Simple Figures, thus less than 12,000 years of age researchers (e.g. Barry 1997; Barry and White 2004; and more than 6,000. If they are the regional coun- Lewis 1988, 1997; Taçon 1998) have argued that terparts of Northern Running Figures, they would Dynamic Figures and Gwions of the Kimberley of probably have a maximum age of about 9,400 years. Western Australia should be the same age given There is a similar range of Maliwawa faunal subject their stylistic similarity and where they sit in matter as in Dynamic Figure style paintings (see regional rock art chronologies. We still do not have Taçon and Chippindale 2008:77; May et al. direct scientific dates for Dynamic Figures, but 2018:72). This suggests a similar ‘freshwater’ envir- Finch et al. (2020) recently used AMS C14 assays to onment as the range of fauna depicted in subse- date mud wasp nests over and under Gwion figures. quent styles includes new species, such as The results are exactly as predicted for Dynamic barramundi, fork-tail catfish and saltwater crocodiles Figures by Taçon and Brockwell (1995), with most that are typical of the ‘saltwater’ environment that dated Gwions created about 12,000 years ago and dominated Arnhem Land after the sea rose to its one possibly as much as 16,000 years old (Finch current level about 6,000 years ago. Although et al. 2020:7). Maliwawa material culture was rarely depicted, Jones et al. (2017) dated Northern Running other than headdresses, the boomerang, single-sided Figures and other ‘Middle Period’ paintings to barbed spears and bags are more similar to 6,000 cal. BP 9,400 cal. BP, and early X-ray begin- Dynamic Figure material culture than that of any ning at 6,000 cal. BP. As Jones et al. (2017:88) note: subsequent styles. Therefore, various lines of evidence suggest Maliwawas were most likely to have The early Holocene age determinations of this art style are significant for a number of reasons. Firstly, been created between 6,000 and 9,400 years ago. The they provide a minimum age for not only the NRF fact that thylacines, and what appear to be other art style, but also other associated art styles extinct animals, were depicted further supports this occurring concurrently within the ‘Middle Period’. age estimation. Secondly, the age determinations provide a Excavations at sites within the Bald Rock/ ‘ ’ minimum age for the Early Period art styles of Maliwawa area of Namunidjbuk revealed spikes of western Arnhem Land. Subsequently, these results confirm Pleistocene antiquity for ‘Early Period’ art activity at various times. For instance, both stone styles in the relative chronology, until now a highly artefact and ochre discard is variable throughout the likely but unsubstantiated hypothesis. Finally, the 25,000 year occupation of the Bald Rock 1 site. For AUSTRALIAN ARCHAEOLOGY 13 stone, high artefact abundance occurs at 20,000, age. Furthermore, we have observed that different 18,000, 15,000 and 9,000 years ago, and finally in the sorts of large and small naturalistic animals both last 1,000 years (Wesley et al. 2018:120). ‘Ochre dis- postdate and predate Dynamic Figures, as card from Bald Rock 1 mirrors the stone artefact Chippindale and Taçon (1993) argued, with discard trends starting with a small peak at Maliwawa naturalistic animals postdating Dynamic 18,000 years ago, then a major peak at 15,000 years animal figures. Although Haskovec (1992:70) sug- BP and another peak after 9,000 years BP’ (Wesley gested large naturalistic humans and animals post- et al. 2018:124). The peak after 9,000 years sits well date Yam Figures, we argue Maliwawa variants with what we contend would be the beginning of were contemporary with Yam Figures. the Maliwawa rock painting period and the Maliwawa animal paintings are of fauna that would Rare animal subject matter and extinct species have been in Arnhem Land at that time. It also reflects an Australia-wide pattern of increasing Thylacines are an indicator of age as until now they population as detailed by Williams et al. (2018). have only been associated with some of the older They conclude that ‘The resulting population den- rock art styles, such as Large Naturalistic Animals sities, along with improving climatic conditions and Dynamic Figures, and thylacines on the between 9 and 6 ka, may be associated with the Australian mainland became extinct between 3,500 appearance of increasingly large and complex to 4,000 years ago (see Taçon et al. 2011, especially Aboriginal societies observed in the Holocene arch- for a Dynamic Figure thylacine over a Large aeological record’ (Williams et al. 2018:151). Naturalistic one at Namunidjbuk). The discovery of As noted above, when superimpositioning what appear to be depictions of ‘bilbies’ (Figure 6) occurred, Maliwawa Figures consistently overlay at an Awunbarna site is surprising as bilbies are Dynamic Figures and old hand stencil types but associated with arid and semi-arid environments far were under Simple Figures. Thus, all lines of evi- to the south and Arnhem Land has not been within dence suggest human Maliwawa Figures lie between their range in historic times (see map in Warburton these better-known Arnhem Land styles and stencil and Travouillon 2016:165). Two of these animals forms, while Maliwawa animals transition from are back-to-back and almost identical in size. The Dynamic Figure style animals to early X-ray fauna. third bilby-like depiction is to the right of them and appears to have been made at a different time, and perhaps by a different artist, as it is larger, has a Discussion and conclusions longer snout, has more line infill, and is in a lighter Despite many decades of rock art research across shade of red (Figure 14). northern Australia, new discoveries of sites and In the Northern Territory during historical times unique imagery are made each year. New styles bilbies were as far north as about half way between that were not recognised by previous chronologies Alice Springs and Tennant Creek. However, given are being defined (e.g. Brady et al. 2020; Gunn the other fauna depicted, typical of an arid environ- et al. 2019) and known styles are slowly being bet- ment, and supporting the likely age of Maliwawa ter assigned an age (e.g. Finch et al. 2020; Jones Figures closer to 9,000 years ago than 6,000 years et al. 2017). New relationships between styles pre- ago, it is conceivable that at that time bilbies existed viously thought to be discreet and separated in in Arnhem Land. Alternatively, a Maliwawa artist time are also being detailed (e.g. Travers and Ross may have seen bilbies on a trip south and later 2016) and a much more complex picture of rock painted them in the Awunbarna shelter to tell of the art change over time is emerging. For instance, strange creatures seen. There is also the possibility Jones et al. (in press) advocate a complete rethink- that the depictions are of Agile Wallabies, Northern ing of the age and unity of large naturalistic ani- Nailtail Wallabies or Short-eared Rock-wallabies, all mals in the rock art of western Arnhem Land and widespread across Kakadu-Arnhem Land today, but conclude that Chaloupka’s(1977, 1985, 1993) def- all of these species have much shorter ears and inition of the Large Naturalistic Style and its snouts than extant bilbies and the creatures depicted chronological placement in the relative rock art at Awunbarna. sequence is not supported. The results of our The solitary painting of a dugong in association research in northwest Arnhem Land support Taçon with a macropod (Figure 15) seems out of place as and Chippindale’s(2008; Chippindale and Taçon well. Depictions of dugongs in more recent styles 1998) conclusion that large naturalistic human fig- are found at some Namunidjbuk sites and elsewhere ures postdate rather than predate Dynamic Figures in northern Australia but not in styles argued to be and, in keeping with Lewis (1988) and Haskovec mid-Holocene or older (Taçon 2017). Indeed, this (1992), that they are early- to mid- Holocene in painting is the oldest depiction of a dugong yet 14 P. S. C. TAÇON ET AL.

Figure 14. Digital tracing of panel of three bilby-like animals, Awunbarna (Digital tracing: Fiona Brady).

ethnographic times (Chaloupka 1981) and the people who made the earlier Dynamic Figures left their paintings in rock shelters from Namunidjbuk to the southern end of the Arnhem Land Plateau, some 180 kilometres distant, as well as over a 200 kilo- metre area east to west (May et al. 2018:68). The dugong painting indicates a Maliwawa artist visited the coast but the lack of other saltwater fauna may suggest this was not a frequent occurrence. At some sites there are two large macropods shown back-to-back with a small space between them (Figure 3). There are also some back-to-back human figures and the back-to-back ‘bilbies’ (Figure 4). Back-to-back macropods occur in more recent art at Namunidjbuk, elsewhere in Arnhem Land and in Kakadu National Park. They are often very large or life-size. During the contact period at one site, Djarrng, life-size back-to-back paintings of buffaloes were made (Taçon et al. in press). Back-to-back figures were painted on sheets of bark in the 1900s (e.g. see Mountford 1956:198–200). They are still painted by western Arnhem Land artists today on paper and have specific stories (Benson Nagurrgurrba pers. comm. 2020). Back-to-back figures are not found in Dynamic Figure rock art. Thus, given the age of between 6,000 to 9,400 years argued above for Maliwawa Figures, the Maliwawa Figure 15. Maliwawa dugong above and to the left of a back-to-back figures are the oldest known for west- small Maliwawa macropod, Namunidjbuk (Photo: P. Taçon). ern Arnhem Land and it appears this painting con- vention began with the Maliwawa style. found. Today it is located about 15 kilometres south Cultural stories and talented artists of the Arafura Sea but 6,000–9,400 years ago the coast would have been further north. The sea level Maliwawa Figures are on average larger than would only have been one metre lower than now Dynamic Figures, in more accessible places and at 6,000 years ago but 9,000 years ago it would have lower elevations in the landscape and in rock shel- been about 20 metres lower. About 7,000 years ago ters, suggesting they were meant to be seen, possibly the coast would only have been about 10 kilometres from some distance. Dynamic Figures are more sub- further north (Taçon and Brockwell 1995:679). Even tle in their size and placement and usually are found 9,400 years ago the coast would have been within at higher elevations in the landscape or at individual easy reach as it was only about 90 kilometres further sites (see Johnston 2018:149–151) and in difficult to north (see Williams et al. 2018:148, Figure 2). reach locations (May et al. 2018:79). Often, Aboriginal people walked such distances in Maliwawa Figures dominate shelter walls with rows AUSTRALIAN ARCHAEOLOGY 15 of figures in various arrangements including those described above. An important aspect of Maliwawa Figures is the level of detail depicted in individual figures and between characters in scenes. Maliwawa Figures and scenes are not simple depictions of everyday life. The artists are clearly communicating aspects of their cultural beliefs, with an emphasis on important animals and interactions between humans and other humans or animals. Indeed, animals are much more common than in the Dynamic Figure style rock art in terms of percentage of subject matter, as most Dynamic Figures are human, almost 89% (May et al. 2018:71; Taçon and Chippindale 2008:77), whereas only about 42% of Maliwawa Figures are human. Animals also feature in scenes with humans more than is the case with Dynamic Figures and in some instances almost appear to be participating in or watching some human activity. This occurrence, and the frequency and variability of headdresses, suggests a ritual context for some of the production of Maliwawa rock art, as has been argued for Dynamic Figures (Johnston 2018). For Dynamic depictions of the Jabiluka study area, 18 headdress types were identi- Figure 16. Infographic summarising some of the main fea- fied (Johnston 2018:204), while 10 have been identi- tures of Maliwawa rock art (Infographic: P. Taçon; digital fied for Maliwawa Figures. Of the 209 complete tracing: Fiona Brady). Jabiluka Dynamic human figures, 71% have headdresses (Johnston 2018:199) while only 30% of 240 period of time? We know from the work of recent Maliwawa Figures have headdresses (72 human fig- rock art artists, such as Nayombolmi, that individual ures). The frequency and range of material culture rock art artists can produce a remarkable number of illustrated is much less for Maliwawa Figures. paintings in their lifetime. Nayombolmi himself is Jabiluka Dynamic Figures were depicted with 15 dif- known to have produced about 650 paintings at ferent types of material culture (Johnston 2018:197) around 50 sites across a vast landscape and includ- and 93% of Dynamic Figures have material culture, ing in Country for which he was not a Traditional while 34.58% of Maliwawa Figures have material Owner (e.g. see Haskovec and Sullivan 1989; May culture and are shown with only four items (head- et al. 2019; Taçon 1989). With 87 Maliwawa sites dresses, spears, bags and a boomerang). These pat- documented to date, we cannot rule out the possi- terns indicate a shift in iconographic emphasis for bility that Maliwawa rock paintings were produced communicating with rock art away from material by a small number of artists (see Johnston et al. culture in favour of animals. The diversity of fauna 2017). It is even possible only a couple of artists and scenes suggests Maliwawa Figures were used to made most of the paintings, with one responsible communicate many different sorts of information. for the more outline forms with minimal infill and In summary, 572 Maliwawa paintings were another creating much of the fuller stroke-line infill recorded at 87 northwest Arnhem Land rock art examples. For instance, outline bird depictions with sites. Large naturalistic humans and macropods are minimal stroke infill at an Awunbarna site are common subjects but animals were more often almost identical to others at Djulirri in the depicted than human figures. They are painted in Namunidjbuk estate, about 100 kilometres away various shades of red with stroke-line infill or out- (Figures 17 and 18). At the same time, much art line forms with a few red strokes as infill. Maliwawa produced after the Maliwawa style demonstrates a Figures are depicted with little material culture other remarkable consistency in the manner of depiction than various forms of headdresses. Maliwawas are and a significant increase in the standardisation of depicted as solitary figures and as part of group some subject matter (i.e. X-ray fish). So, perhaps scenes showing various activities and some may what we are observing is increasing standardisation have a ceremonial context (Figure 16). But was the in the manner of depiction after the period in which Maliwawa rock art sporadic and made during a Dynamic Figures were made. This has implications short time period or did it continue over a long for rock art research everywhere in which a style or 16 P. S. C. TAÇON ET AL.

University, ICREA and the University of Barcelona are thanked for supporting this research.

Disclosure statement No potential conflict of interest was reported by the author(s).

Funding This research has resulted from a series of Australian Research Council grants ([DP0877463, LP0882985, DP160101832 and FL160100123]). The Northern Figure 17. Maliwawa bird paintings from an Awunbarna site Territory Government Department of Lands Planning and with distinctive stylistic features (Photo: P. Tac¸on). the Environment and Dr John Kearney also provided some funding between ARC grants.

ORCID Paul S. C. Taçon http://orcid.org/0000-0002-0280-4366 Sally K. May https://orcid.org/0000-0003-2805-023X Iain G. Johnston http://orcid.org/0000-0003-3722-2128 Andrea Jalandoni http://orcid.org/0000-0002- 4821-7183 Daryl Wesley http://orcid.org/0000-0002-3537-5668 Ines Domingo Sanz http://orcid.org/0000-0003- 4707-8094 Liam M. Brady http://orcid.org/0000-0002-3997-3470 Duncan Wright https://orcid.org/0000-0003-2546-0559 Joakim Goldhahn http://orcid.org/0000-0003- Figure 18. Similar looking Maliwawa bird paintings from 4640-8784 Djulirri, Namunidjbuk about 100 kilometres northeast (Photo: P. Taçon). References Barry, M. 1997 I Would Sooner Not Call Them Bradshaws. Unpublished BA(Hons) Thesis, University manner of depiction is suggested to have been made of Sydney, Sydney. over hundreds of years or millennia. Barry, M. and P.J. White 2004 ‘Exotic Bradshaws’ or Australian ‘Gwion’: An archaeological test. Australian Aboriginal Studies 2004(1):37–44. Acknowledgements Brady, L.M., J.J. Bradley, A. Kearney and D. Wesley 2020 A rare miniature and small-scale stencil assemblage The authors are indebted to the Lamilami family for per- from northern Australia’s Gulf of Carpentaria region: mission and assistance with field research that underpins Replication and insights into meaning and symbolism. this paper, especially Patrick Lamilami (also known as Antiquity 94(375):780–796. Cecil Namunidjbuk) and Leonard Lamilami. Many volun- Brady, L.M., S.K. May, J. Goldhahn, P.S.C. Taçon and P. teers, students and colleagues assisted with this research Lamilami 2020 What painting? Encountering and inter- over a ten-year period, too many to list individually but preting the archaeological record in western Arnhem their contributions are greatly appreciated. Janet and Phil Land, northern Australia. Archaeology in Oceania Davill deserve special recognition as they volunteered a 55(2):106–117. great deal of time and energy throughout this period, Brandl, E.J. 1973 Australian Aboriginal Paintings in along with Wayne Brennan, Michelle Langley and Irina Western and Central Arnhem Land, Temporal Ponomareva. For Awunbarna research we are indebted to Sequences and Elements of Style in Cadell River and Traditional Owner Charlie Mungulda. The authors also Deaf Adder Creek Art. Canberra: Australian Institute of thank Lee and Dianne Davidson, Ray Curry, Bridget Aboriginal Studies. Adams (nee Thomson), Scott Adams, the rest of Chaloupka, G. 1977 Aspects of the chronology and sche- Davidson’s Arnhemland Safaris’ staff and research student matisation of two prehistoric sites on the Arnhem Roxanne Tsang. Thanks to Injalak Arts and Benson Land Plateau. In P.J. Ucko (ed.), Form in Indigenous Nagurrgurrba for discussing the cultural significance of Art: Schematisation in the Art of Aboriginal Australia the back-to-back figures with us. and Prehistoric Europe, pp.243–259. Canberra: The authors thank the Northern Land Council for Australian Institute of Aboriginal Studies. access and research permits. Griffith University, Linnaeus Chaloupka, G. 1981 The traditional movement of a band University, Monash University, The Australian National of Aboriginals in Kakadu. In T. Stokes (ed.), Kakadu AUSTRALIAN ARCHAEOLOGY 17

National Park Education Resources, pp.162–171. Haskovec, I.P. and H. Sullivan 1989 Reflections and rejec- Canberra: Australian National Parks and Wildlife tions of an Aboriginal artist. In Morphy, H. (ed.), Service. Animals into Art, pp.57–74. London: Unwin Hyman. Chaloupka, G. 1984a From Palaeoart to Casual Paintings. Hayward, J. A. 2016. Reading the Signs: Depictions of Monograph Series 1. Darwin: Northern Territory People and Things in the Rock Art of Mirarr Country, Museum of Arts and Sciences. Northern Territory, Australia. Unpublished PhD Chaloupka, G. 1984b Rock Art of the Arnhem Land Thesis. Australian National University, Canberra. Plateau: Paintings of the Dynamic Figures Style. Jelinek, J. 1989 The Great Art of the Early Australians: Unpublished report. Darwin: Northern Territory The Study of the Evolution and Role of Rock Art in the Museum of Arts and Sciences. Society of Australian Hunters and Gatherers. Brno: Chaloupka, G. 1985 Chronological sequence of Arnhem Moravian Museum-Anthropos Institute. Land Plateau rock art. In R. Jones (ed.), Archaeological Johnston, I.G. 2017 Body ornamentation in dynamic Research in Kakadu National Park, pp.269–280. scenes: Ritual and headdresses in the Dynamic Figure Canberra: Australian National Parks and Wildlife rock art of western Arnhem Land, Australia. The – Service (Special Publication 13). Artefact 38:67 80. Chaloupka, G. 1993 Journey in Time: The World’s Longest Johnston, I.G. 2018 The Dynamic Figure Art of Jabiluka: Continuing Art Tradition: The 50,000-Year Story of the A Study of Ritual in Early Australian Rock Art. Australian Aboriginal Rock Art of Arnhem Land. Unpublished PhD Thesis, Australian National Chatswood: Reed. University, Canberra. Chippindale, C. and P.S.C. Taçon 1993 Two old painted Johnston, I.G., J. Goldhahn and S.K. May 2017 Dynamic ’ panels from Kakadu: Variations and sequence in Figures of Jabiluka: Chaloupka s 4-phase theory and ‘ ’ Arnhem Land rock art. In J. Steinbring, A. Watchman, the question of variability within a rock art style .In P. Faulstich and P.S.C. Taçon (eds), Time and Space: B. David, P.S.C. Taçon, J.-M. Geneste and J.-J. Delannoy (eds), The Archaeology of Rock Art in Dating and Spatial Considerations in Rock Art – Research, pp.32–56. Occasional AURA Publication 8. Western Arnhem Land, pp.109 127. Terra Australis 47. Melbourne: Archaeological Publications. Canberra: ANU Press. Chippindale, C. and P.S.C. Taçon 1998 The many ways of Jones, T. 2017 Disentangling the Styles, Sequences and dating Arnhem Land rock-art, north Australia. In. C. Antiquity of the Early Rock Art of Western Arnhem Chippindale, and P.S.C. Taçon (eds), The Archaeology Land. Unpublished PhD Thesis, Australian National University, Canberra. of Rock-Art, pp.90–111. Cambridge: Cambridge Jones, T. and S.K. May 2017 Rock art and ritual function: University Press. The Northern Running Figures of western Arnhem David, B., B. Barker, F. Petchey, J.-J. Delannoy, J.-M. Land. The Artefact 38:53–66. Geneste, C. Rowe, M. Eccleston, L. Lamb and R. Jones, T., V.A. Levchenko, P.L. King, U. Troitzsch, D. Whear 2013 A 28,000 year old excavated painted rock Wesley, A. Williams and A. Nayingull 2017 from Nawarla Gabarnmang, northern Australia. Journal Radiocarbon age constraints for a of Archaeological Science 40(5):2493–2501. Pleistocene–Holocene transition rock art style: The David, B., P.S.C. Taçon, R. Gunn, J.-J. Delannoy and J.- Northern Running Figures of the East Alligator River M. Geneste 2017 The archaeology of western Arnhem ’ region, western Arnhem Land, Australia. Journal of Land s rock art. In B. David, P.S.C. Taçon, J.M. Archaeological Science: Reports 11:80–89. Geneste and J.-J. Delannoy (eds), The Archaeology of – Jones, T., D. Wesley, S.K. May, I.G. Johnston, C. Rock Art in Western Arnhem Land, pp.1 17. Terra McFaffen and P.S.C. Taçon in press Rethinking the age Australis 47. Canberra: ANU Press. and unity of large naturalistic animal forms in early Finch, D., A. Gleadow, J. Hergt, V.A. Levchenko, P. western Arnhem Land rock art, Australia. Australian Heaney, P. Veth, S. Harper, S. Ouzman, C. Myers, and Archaeology. H. Green 2020 12,000-year-old Aboriginal rock art Lewis, D. 1988 The Rock Paintings of Arnhem Land, from the Kimberley region, Western Australia. Science Australia: Social, Ecological and Material Culture Advances 6(6):eaay3922. Change in the Post-Glacial Period. Oxford: British Gunn, R.G. 2016 Art of the Ancestors. Spatial and Archaeological Reports International Series 415. Temporal Patterning in the Rock Art of Nawarla Lewis, D. 1997 Bradshaws: The view from Arnhem Land. Gabarnmang, a Major Jawoyn Cultural Site on the Australian Archaeology 44(1):1–16. Arnhem Land Plateau. Unpublished PhD Thesis, Marshall, M. Under review Rock Art Conservation and Monash University, Melbourne. Management: 21st Century Perspectives from Northern Gunn, R.G. 2018 Art of the Ancestors: Spatial and Australia. Unpublished PhD Thesis, Australian Temporal Patterning in the Ceiling Rock Art of Nawarla National University, Canberra. Gabarnmang, Arnhem Land, Australia. Oxford: May, S.K. and P.S.C. Taçon 2014 Kakadu National Park: Archaeopress. Rock art. In C. Smith (ed.), Encyclopedia of Global Gunn, R.G. and R.L. Whear 2007 Dynamic sketches Archaeology, pp.4235–4240. New York: Springer. 6000þ year old dry – pigment drawings from Arnhem May, S.K., P.S.C. Taçon, D. Wesley and M. Travers 2010 Land. Archaeology in Oceania 42(1):22–28. Painting history: Indigenous observations and depic- Gunn, R.G., B. David, L. Douglas, J.-J. Delannoy, S. tions of the ‘other’ in northwestern Arnhem Land, Harper, P. Heaney, S. Ouzman, and P. Veth 2019 Australia. Australian Archaeology 71(1):57–65. Kimberley Stout Figures: A new rock art style for May, S.K., P.S.C. Taçon, D. Wright and M. Marshall 2015 Kimberley rock art, North-Western Australia. The rock art of Kakadu: Past, present and future Australian Archaeology 85(2):151–169. research, conservation and management. In S. Haskovec, I.P. 1992 Mt. Gilruth revisited. Archaeology in Winderlich (ed.), Kakadu National Park Symposia Oceania 27(2):61–74. Series. Symposium 6: Walk the Talk: Cultural Heritage 18 P. S. C. TAÇON ET AL.

Management in Kakadu National Park, 19–20 May Taçon, P.S.C. and C. Chippindale 1994 Australia’s ancient 2011, Jabiru Youth Centre, pp.36–44. Internal Report warriors: Changing depictions of fighting in the rock 625. Darwin: Australian Government, Supervising art of Arnhem Land, N.T. Cambridge Archaeological Scientist. Journal 4(2):211–248. May, S.K., I.G. Johnston, P.S.C. Taçon, I. D. Sanz and J. Taçon, P.S.C. and C. Chippindale 2008 Changing places: Goldhahn 2018 Early Australian anthropomorphs: Ten thousand years of north Australian rock-art trans- Jabiluka’s Dynamic Figure rock paintings. Cambridge formation. In D. Papagianni, H. Maschner and R. Archaeological Journal 28(1):67–83. Layton (eds), Time and Change: Archaeological and May, S.K., J. Huntley, M. Marshall, E. Miller, J. Hayward, Anthropological Perspectives on the Long-Term in A. Jalandoni, J. Goldhahn, I.G. Johnston, J. Lee, G. Hunter-Gatherer Societies, pp.73–94. Oxford: Oxbow O’Loughlin, K. May, I. Domingo Sanz and P.S.C. Books. Taçon 2020a New insights into the rock art of Taçon, P.S.C. and C. Chippindale 2001 Transformation Anbangbang Gallery, Kakadu National Park. Journal of and depictions of the First People: Animal-headed Field Archaeology 45(2):120–134. beings of Arnhem Land, N.T., Australia. In K. Helskog May, S.K., L. Tayler, C. Frieman, P.S.C. Taçon, D. (ed.), Theoretical Perspectives in Rock Art Research, – Wesley, T. Jones, J. Goldhahn, and C. Mungulda 2020b pp.175 210. Oslo: Novus forlag. Survival, social cohesion and rock art: The painted Taçon, P.S.C., W. Brennan and R. Lamilami 2011 Rare hands of western Arnhem Land, Australia. Cambridge and curious thylacine depictions from Wollemi Archaeological Journal 30(3):491–510. National Park, New South Wales and Arnhem Land, May, S.K., J. Maralngurra, I.G. Johnston, J. Goldhahn, J. Northern Territory, Australia. In J. Specht and R. Lee, G. G. O’Loughlin, K. May, C. Nabobbob, M. Torrence (eds), Changing Perspectives in Australian ‘ ’ Archaeology. Technical Reports of the Australian Garde and P.S.C. Taçon 2019 This is my father s – painting’: A first-hand account of the creation of the Museum Online 23(11):165 174. Taçon, P.S.C., A. Paterson, J. Ross and S.K. May 2012 most famous rock art in Kakadu National Park. Rock Picturing change and changing pictures: Contact period Art Research 36(2):199–213. rock art of Australia. In J. McDonald and P. Veth Mountford, C.P. 1956 Records of the American-Australian (eds), A Companion to Rock Art, pp.420–436. Scientific Expedition to Arnhem Land, 1, Art, Myth and Chichester: Blackwell Publishing Ltd. Symbolism. Melbourne: Melbourne University Press. Taçon, P.S.C., S.K. May, S.J. Fallon, M. Travers, D. Roberts, D.A. and A. Parker 2003 Ancient Ochres: The Wesley and R. Lamilami 2010 A minimum age for Aboriginal Rock Paintings of Mount Borradaile. early depictions of Southeast Asian praus in the rock Marleston: J.B. Books. art of Arnhem Land. Australian Archaeology 71(1): Ross, J. 2013 A continent of nations: The emergence of 1–10. regionally distinct rock art styles across Australia. – Taçon, P.S.C., S.K. May, D. Wesley, A. Jalandoni, R. Quaternary International 285:171 198. Tsang and K. Mangiru in press History disappearing: Taçon, P.S.C. 1987 Internal-external: A re-evaluation of ’ ‘ ’ The rapid loss of Australia s contact period rock art the X-ray concept in western Arnhem Land rock art. with a case study from the Djarrng site of Arnhem – Rock Art Research 4(1):36 50. Land, Northern Territory. Journal of Field Archaeology ‘ ’ Taçon, P.S.C. 1989 From Rainbow Snakes to X-ray Fish: Travers, M. and J. Ross 2016 Continuity and change in The Nature of the Recent Rock Painting Tradition of the anthropomorphic figures of Australia’s northwest Western Arnhem Land. Australia. Unpublished PhD Kimberley. Australian Archaeology 82(2):148–167. Thesis, Australian National University, Canberra. Warburton, N.M. and K.J. Travouillon 2016 The biology Taçon, P.S.C. 1992 The last rock painters of Kakadu. and palaeontology of the Peramelemorphia: A review – Australian Natural History 23 (11):866 873. of current knowledge and future research directions. Taçon, P.S.C. 1993 An Asessment of Rock Art in the Australian Journal of Zoology 64(3):151–181. Mann River Region, Arnhem Land, N.T. A Report to Wesley, D. 2014. Bayini, Macassans, Balanda, and Bininj: the Bawinanga Aboriginal Corporation and the Djomi Defining the Indigenous Past of Arnhem Land Museum. Sydney: Australian Museum. Through Culture Contact. Unpublished PhD Thesis, Taçon, P.S.C. 1998 Magical paintings of the Kimberley. Australian National University, Canberra. Nature Australia 26(3):40–47. Wesley, D., M. Litster, S. O’Connor, E. Grono, J. Theys, Taçon, P.S.C. 2017 Whales, dolphins and dugongs in the A. Higgins, T. Jones, S.K. May and P.S.C. Taçon 2018 rock art of Australia. In S. Lee (ed.), Whale on the The archaeology of Maliwawa: 25,000 years of occupa- Rock, pp.71–84. Ulsan: Ulsan Petroglyph Museum. tion in the Wellington Range, Arnhem Land. Taçon, P.S.C. 2019 Connecting to the ancestors: Why Australian Archaeology 84(2):108–128. rock art is important for Indigenous Australians and Williams, A.N., S. Ulm, T. Sapienza, S. Lewis and C.S.M. their well-being. Rock Art Research 36(1):5–14. Turney 2018 Sea-level change and demography during Taçon, P.S.C. and S. Brockwell 1995 Arnhem Land pre- the last glacial termination and early Holocene across history in landscape, stone and paint. Antiquity the Australian continent. Quaternary Science Reviews 69(265):676–695. 182:144–154.