AR09. 4 August 2004
Copyright © 2004 by Kevin Sharpe. All rights reserved.
Rock Art Research 21:1 (May 2004), pp. 57-84.







Kevin Sharpe

The Graduate College, Union Institute and University, Cincinnati, Ohio, USA
Harris Manchester College, Oxford University, Oxford, United Kingdom
10 Shirelake Close, Oxford OX1 1SN, United Kingdom


See also, ‘Responses to Kevin Sharpe, “Line Markings: Human or Animal Origin?”’

And my, ‘Rejoinder to Comments by Geoffrey D. Aslin, Robert G. Bednarik, and R. G. Gunn.’

ABSTRACT. The origin – whether human or animal – of line marks found in a cave has obvious implications for what used the cave and whether or not the prehistorian shows interest in the site. Some line marks are obviously human, some are obviously animal scratches, some are obviously geological and, from among the remainder, some could have a human or an animal origin. This paper focuses on markings in the last category and it seeks a balanced, systematic, communicable, and empirical way to determine their origin.

KEYWORDS. Animal scratches, Robert Bednarik, engraved rock incisions, Koonalda Cave, prehistoric art.

Any line marks found in caves face a question: Did or did not humans make them? For instance, lines on floor boulders in the Upper Chamber of Koonalda Cave, South Australia, provoke a spectrum of reactions:

·        Some are human. Sharpe (Sharpe and Sharpe 1976: 125; see also Sharpe 1982) describes one set as ‘some half-dozen deeply-engraved parallel lines…at a slight angle to the vertical, at [25 millimeter] distances from each other, and [12 to 20 centimeters] long. The strokes [are] vigorous and emphatic.’ Many such lines though not all of them he considers humanly made and he attempts to refute the attribution of all to animals such as bats and owls, by pointing to (Sharpe and Sharpe 1976: 126-127):

  • the presence of ‘horizontal lines and grids’;
  • vertical lines that often ‘follow the lower surface of a convex boulder to near its base’;
  • lines that ‘appear on the underside of overhanging ledges’; and
  • lines that sometimes appear ‘around and in very small holes ([25 millimeters] in diameter).’

·        Perhaps they are animal. John Mulvaney (1975: 280) portrays the lines as ‘thin parallel [lines], suggestive of animal claws.’

·        Definitely animal. Robert Bednarik (1992: 233) writes: ‘literature offers numerous examples of erroneous identifications of rock markings [including]…the description of animal scratches in caves as petroglyphs,’ and he refers to the above two papers by Sharpe as examples. Bednarik, in fact, considers the markings ‘mostly of non-trogloxenes which are unable to cope with an environment they are driven into by circumstances, such as the promise of water as indicated by increasing air humidity’ (Bednarik 1991: 40). (A trogloxene is an animal species that habitually enters caves.)

(Note that the human marks being discussed are not finger flutings, but incised with a sharp object like a flint or piece of broken or sharpened bone; the difference between flutings and scratches or incisions (in their cross-sections, for instance) is usually obvious and needs no further discussion here.)

Prior to his two papers referred to above, Bednarik identified six classes of Australian cave petroglyphs, including finger flutings, tool marks, and shallow engravings (Bednarik 1990: 64-66). The last of these he describes as ‘incised with usually single strokes of a pointed tool’ and he omits Koonalda Cave as one of the Australian sites that contain them.

The most extensive statement of Bednarik’s opinion against the human origin of the Koonalda Upper Chamber marks appears in his 1991 definitive paper on the subject of animal scratches. There he writes:

The scratch markings in Koonalda Cave commence as soon as the threshold of discernible light is passed. On the walls they generally reach to about 1.5 m in height….Many of the floor boulders are also marked by sets of incisions (Sharpe and Sharpe 1976)….The boulder markings include no configurations or groove characteristics demanding a human origin. They closely resemble marks I identify as mammalian elsewhere, and I would in fact find it remarkable if, of all the suitably endowed caves, Koonalda Cave would be the only one lacking animal scratch marks − particularly as it is such an outstanding animal trap, with its enormous sinkhole entrance. I believe that a human origin should only be postulated if an animal origin can be ruled out with confidence. In my view, this has not been proven here, nor has it been attempted to refute an attribution of the marks to animals. No researcher who is familiar with parietal wall markings as a general phenomenon (i.e., who has studied them thoroughly in more than 100 caves!) would support the proposal (Bednarik 1991: 38-39, emphasis added).

In general, Bednarik believes that

certainly more than 99.9 percent…of wall or ceiling marks [that we have seen in the 320 or so caves we have studied]…are natural marks. Most commonly they were produced by animals, with their claws, wings, or bodies, and even with their horns or antlers…. [We have established] clear criteria for identifying animal scratch marks. Although we recognize a great variety of them, the most common being those of bats, cave bears, opossums, and extinct Australian megafaunal species, we believe today that a reliable separation can be achieved by an experienced observer in nearly every instance (Bednarik 1993: 3; see also Bednarik 1994b: 65).

Bednarik contributes extensively to work on animal and natural markings (for instance, Bednarik 1991; 1994a-c; 1998) and he writes assuredly, but his above generalized statements about the markings in Koonalda Cave ought not to stand merely on rhetoric or subjective interpretation. Where are Bednarik’s ‘clear criteria’? On the other hand, Sharpe accepts that some of the lines are of animal origin; how can he tell the difference?

The question is how systematically and scientifically rather than subjectively to distinguish in general between animal and human line markings. Researchers who have seen line scratches or incisions in many caves may feel confident that they can distinguish between human and animal ones. Their subjective decision is based on experience, itself based and consolidated from successive subjective judgments. It can become comforting to ignore particular lines because they are assumed to be of animal origin; it makes the assimilation of what’s left as human an easier task. Habit can become ease, which can become bias and dogma. What though of lines whose origin is uncertain, where the experienced honestly find themselves unsure? What basis is there for making such a decision? The field needs the formalization of the experience of researchers so that decisions can be made on lines of uncertain origin, so that inexperienced researchers can learn, and so that debate and decision can be made on potentially incorrect guesses or experience-based interpretations. The field needs communicable criteria, reliable guidelines, and more objectivity. So, in an ideal world, what distinguishes markings of a human origin from those of an animal origin?

This paper extends previous publications on the subject and offers avenues to explore that might result in more objective criteria for distinguishing between animal scratches and similar human line markings. A conclusive test for origin is not yet proposed, though suggestions are gathered from which one might emerge. Bednarik himself writes: ‘Discriminating the two types of marks is the crucial precondition for any informed assessment of the rock art component and, having studied such markings in about one thousand caves, I have developed a healthy respect for the complexity of the subject’ (Bednarik 1994a: 37; see also Bednarik 1991: 28).


Several sites in Australia contain line markings similar to those on the boulders in the Upper Chamber of Koonalda Cave. Four will be used to elaborate the research question and for comparison with Koonalda.

Kintore and Cutta Cutta Caves, Northern Territory

W. P. Walsh initially discovered lines in the Kintore and Cutta Cutta Caves in 1963 (Bednarik 1985: 83; Flood 1997: 51-55; Layton 1992; Walsh 1964: 87-90). The majority of the markings occur in a characteristic grouping of from three to five sub-parallel lines. At Cutta Cutta, line width varies from 3.5 to 5.0 centimeters, length from a few centimeters to meters, and depth from 0.2 to 2.0 millimeters depending on rock hardness. These sets often occur so densely that they become inseparable. The marks occur on ceilings, walls, on floor boulders, and from the entrance to the depths of the cave. The lines in Kintore Cave resemble those in Cutta Cutta. A film of clay, which in places contains a high-water mark of vegetation debris, covers many marks in both caves (Flood 1997: 53; Layton 1992).

Geological processes did not cause the markings (Flood 1997: 54), but animals provide many potential line makers. The bent-winged bat inhabits both caves. Walsh (1964: 89-91) suggests, though, that it is ‘too small to cause the markings arranged in the observed pattern.’ The False vampire bat, which inhabited Cutta Cutta Cave until recently, would scratch lines of the correct size and spatial distance (Walsh 1964). However, it apparently never inhabited Kintore Cave and no markings appear in the section of Cutta Cutta Cave with abundant remains of the species (Flood 1997). The support for bats making the lines is weak.

Walsh (1964) suggests that, though goannas venture into caves up to 0.4 kilometers from daylight, they cannot climb up to ceilings or on over-hanging walls. Dingoes venture into the dark areas of caves as well, but could not make the high markings. This may also apply to the little northern native cat, echidna, bandicoots, and possums. Ben Gunn (n.d.: 3-4) disputes this and suggests that possums, lizards, and snakes could reach high positions, overhanging walls, and ceilings.

Can animals scratch ceilings (Jones 1925)? Bednarik affirms Gunn’s standpoint and claims that ‘animal marks commonly occur on flat cave ceilings’ (Bednarik 1987-1988: 2). Elsewhere, Bednarik (1994a: 24) illustrates a marking on the ceiling of Robertson Cave, South Australia, and adds:

Most observers would identify it as an engraving, particularly as it has been said in Australian archaeological literature that animal species do not mark cave ceilings. This marking was in fact produced by an extinct megafaunal species.

In case it appears that scratches on cave ceilings are common, it should be noted that Bednarik found only one cave bear scratch on the ceiling of Grottes des Endrevies, France (Bednarik 1991; 1994b: 62).

Perhaps several animals made the Kintore and Cutta Cutta lines: bats marking the ceilings and various animals marking the boulders and lower walls. Walsh thinks, though, that the relative similarity of the markings refutes this suggestion. In conclusion, he can find no ‘satisfactory explanation of these abundant cave markings’ (Walsh 1964: 90).

Walsh raises the possibility that humans made the lines, but rejects this because he can see no motive, no representational forms in the lines, and because the marks do not correspond to already accepted types of Aboriginal rock art. Clearly, he prefers an animal origin and he writes: ‘the most conclusive evidence would be provided by searching for bone material in all the known sites and attempting to locate the relevant common faunal element’ (Walsh 1964: 91).

McEachern’s Cave, Victoria

McEachern’s Cave lies in the southwestern corner of Victoria. A. G. Link (1967) discovered lines inside the cave and describes them as grouped in sets of four and covering the walls, the ceiling, and fallen rocks. The lines’ depth and width vary between sets, though the width of lines remains constant within sets. Lengths vary from 5 to 30 centimeters and some younger sets overlay older sets.

Link (1967: 137) favors an animal origin for the marks. J. N. Jennings (1968: 52), attributing his conclusions to the work of N. A. Wakefield (1967), suggests that wombats scratched the lines when the floor level was considerably higher than at present. He offers this interpretation for the markings in Kintore and Cutta Cutta Caves as well.

Link’s reference to the animal origin of the marks originates from E. Troughton’s work (1957: 24; see also Jones 1925). Troughton writes that the yellow-footed marsupial mouse, which nests in rock crevices and sandstone caves, can create line marks. Its ridged footpads and long claws enable it to climb about trees and vines and, he claims, to run upside-down over cave roofs. This requires more thought and observation, however; the clinging to a roof by a stationary or running animal would probably not produce lines but pricks in the rock surface.

Tantanoola Cave, South Australia

Bednarik, along with Geoffrey Aslin and Elfride Bednarik, studied the extensive markings of Tantanoola Cave, 20 kilometers southwest of Mount Gambier (Aslin, et al. 1985). The lines here generally have a V-section and occur on all exposed wall surfaces, though not on the ceiling. Most appear in sets of two to five parallel lines, each set usually deeper at one end and becoming shallower and narrower until disappearing. The larger lines occur as vertical marks and are generally deeper. Few of the lines are straight and the only geometric shapes comprise grids of separate sets of markings.

Earlier excavations by Norman Tindale (1933) in the cave revealed evidence for a variety of fauna: bats, possums, rats, water rats, bandicoots, wombats, Tasmanian devils, seals, giant kangaroos, and native cats. Gunn (n.d.) studied the markings and thinks that the floor level has not changed and, he adds, at least several of the species could reach the marked areas of the cave and have the capacity to scratch the lines. He thus concludes that the marks are of animal origin. Further, he writes, no positive evidence supports a human origin. Bednarik agrees, and adds:

None of the line arrangements differ appreciably from those studied at about 200 sites of the general region. Complexity of the ‘designs’ is in fact quite moderate in comparison to that at some other sites where a human involvement is precluded, but claw spacing is above the averages of other sites: a sample of 17 identifiable sets provided a mean line spacing of 18.8 mm (Bednarik 1991: 40).

Gunn’s logic does not convince:

·        The lack of change in floor level requires evidence.

·        The lack of discernible motifs or intentionally produced geometric shapes need not obviate a human origin. Lines that mean nothing to present-day westerners, or even present-day Aborigines, might have possessed a wealth of meaning for the local prehistoric inhabitants, and we can only grasp this hypothetically and tentatively (Bahn 1998). (Grids of separate sets of markings do not imply deliberate construction.)

·        The markings appear on the dolomite bedrock and on solidified sediment deposit. Both of these are very hard and a fingernail cannot make an impression. Are claws sufficiently strong to scratch these rocks? In addition, out of the many species listed, do any scratch rocks? Sliding could produce vertical lines, but why would these animals be sliding? How would sliding produce horizontal marks?

Aslin’s (2003) unpublished description of the improbability of human access to Tantanoola Cave until recently makes the animal origin case all the more probable.

An animal origin for the lines in the above four sites Kintore, Cutta Cutta, McEachern’s, and Tantanoola caves, based on published notes, seems probable, but not firm.

History of the Confusion

Confusion between the human versus animal origin of markings such as those in Koonalda surfaced some time ago. When Alexander Gallus studied Koonalda during the 1950s and 1960s, he fought to establish the human origin of the finger flutings at the rear of the Upper Chamber (Flood 1997). Comments Bednarik (1987-1988: 1; see also Bednarik 1993: 1):

Gallus…aroused [considerable skepticism] when he announced that wall markings in Koonalda Cave, up to three hundred meters from the entrance, were made by humans in the late Pleistocene….What made his claims even more incredible was his insistence that early Australians had engaged in systematic underground mining of [flint].

Later, Mulvaney’s book, The Prehistory of Australia (1975: 279-280; see also Mulvaney and Kamminga 1999: 365), accepts the human origin for the flutings and ‘V-sectioned linear grooves.’ Gallus no longer need fight this battle.

Markings in other caves lead to similar confusion. Frederick McCarthy includes Kintore and Cutta Cutta Caves in his survey of the Aboriginal antiquities of the Northern Territory because he thinks the lines have a human origin. Under ‘Rock Engravings,’ he writes:

In the Kintore and Cutta Cutta caves, groupings of from 3-5 sub-parallel lines, from a few centimeters to several meters long, like human or animal limbs, and some appear to be human figures, are [cut] in soft rock surfaces or in mud coatings on the walls; some very faint markings are [cut] in the hard rock or in secondary calcite (McCarthy 1970: 58).

Mulvaney (1975: 280) notes the general association of human petroglyphs with ‘thin parallel [lines], suggestive of animal claws.’

Similar markings are present in darkness in two caves near Katherine, Northern Territory, including Kintore. Sylvia Hallam recorded similar markings on the roof of Orchestra Shell Cave, near Perth. Curiously, no animal or reptile is thought capable of marking the walls or ceiling in some most inaccessible places, and there are indications that the [lines] are of some antiquity. There is a possibility, therefore, that the markings were made through human agency by using claws as an engraver [but see Bednarik (1987-1988: 2-11)]. However, W. P. Walsh informs me that in the Northern Territory, a bat species may cause the markings and he is collecting data.

Unlike McCarthy, Mulvaney appears undecided as to the human as opposed to the animal origin of the ‘thin parallel [lines].’ Other writers have decided. Walsh’s (1964: 87) original paper on these two caves underplays the Kintore and Cutta Cutta markings as human artifacts, but their exclusion from lists of human petroglyphs is now commonplace (Flood 1997: 55). Bednarik (1991; 1994a) assuredly says that animals made them. He writes that he has seen markings that only animals could make and, by comparing the marks, he concludes that the Kintore and Cutta Cutta Caves’ lines are animal scratches. After all, in Frank’s (1980: 146) words, ‘animals including birds, bats, possums, kangaroos, wombats, wallabies, goannas, and dogs can leave quite a variety of markings on a soft cave wall.’

Koonalda Cave, South Australia

Within Rockfall C in the Upper Chamber of Koonalda Cave lie smooth and rounded boulders cut with lines, as mentioned above (see Sharpe and Fawbert In prep.; Sharpe and Whitehead In prep.). They occur on the boulders, on walls meters above floor level (where the question of a change in floor level does not apply), and down to floor level on the boulders (Sharpe and Fawbert In prep.). They occur under overhangs (see also Sharpe and Sharpe 1976: 127) and in tunnels and floor holes in Rockfall C such as ‘Chadwick’s Hole’ (Sharpe and Whitehead In prep.: 17-18), where some of them are large in size.

These lines are perhaps the most controversial potential petroglyphs in Koonalda. Plates 1-3 provide examples. Were they all made by animals or were at least some of them made by humans? At least six types of prehistoric line markings may occur in the Upper Chamber:

1.      scratches that animals cut with their claws;

2.      scratch-like lines that humans cut with a severed animal paw or a comb (Marshack 1977:  310);

3.      lines that humans incise with a flake of flint;

4.      lines that humans score with a stick;

5.      lines that humans mark with a piece of bone; and

6.      flutings that humans draw with their fingers.

(Note the difference in terminology between scratches (animal claw marks), incisions (lines humans make with flint), scorings (lines humans make with a stick), bone marks (lines humans make with a bone), and flutings (lines humans draw with their fingers) (Sharpe and Van Gelder To appear). Note also that a unit is a set of (near- or sub-) parallel lines that appear to have been made by one hand or paw movement, and a cluster is a group of units that exhibit a unity, for instance because they overlay each other. To develop a more precise method for distinguishing the origin of the line marks requires a more precise nomenclature.)

This paper develops a way toward distinguishing between fine lines of Type 1, 2, and 3.


Plate 1. Rocks D2-a and D2-b in Rockfall C, Upper Chamber, Koonalda Cave, South Australia.

Plate 2. On Rock E5-a in Rockfall C, Upper Chamber, Koonalda Cave, South Australia.

Plate 3. Rock E6-a in Rockfall C, Upper Chamber, Koonalda Cave, South Australia.

Information from an initial examination of several clusters in the Upper Chamber suggests:

1.      Some lines are deep and some shallow, ranging in length from 1 to 33 centimeters.

2.      In cross-sections, the lines vary from a sharp V to a broad and shallow U, and some have a squared-off cross-section.

3.      The number of lines in the units on the boulders varies from between one and eight.

4.      The average number of lines in the units is between three and four.

5.      Units of more than one line vary from 0.2 to about 4 centimeters in width.

6.      Lines within a unit tend to wander together and apart.

7.      The length of a unit does not appear to correlate directly to the separation between the lines in the unit.

8.      Units, in general, do not appear in mirrored pairs.

9.      Horizontal units exist (see also Sharpe and Sharpe 1976: 126-127).

10.  Lines within some units branch and some cross over and back.

Other Mechanisms

Line markings can result from other actions than animal scratching and prehistoric human engraving.

Rudy Frank (1980: 145) includes geological along with animal sources in his comprehensive summary of non-human line-making agents:

Weathering of joint and bedding features especially in dolomites can etch out nice ‘incised’ grids and triangular patterns. Solutional features such as fine anastamosies and karren can form a variety of intriguing patterns.

Did geology or vegetation form the boulder lines in the Upper Chamber of Koonalda Cave (Bednarik 1994a provides a thorough overview of such processes)? Geological process can certainly change lines once made, such as those that affect moonmilk, in which many finger flutings occur (Bednarik 1994b: 51-52). Did they create the boulder lines? Are the lines solution tubes, for instance? Solution tubes abound at the entrance to the cave. However, they are larger than the boulder lines, have a different cross-section, and do not appear in sets of, say, two to five parallel marks. They also disappear into the limestone surface as tubes, a characteristic absent from the lines (Sharpe and Sharpe 1976: 126). This geological mechanism is, therefore, probably not responsible for the boulder lines (see also Bednarik 1991: 30-31). Neither do the boulder lines appear to be the Rillenkarren that Bednarik (1994a: 29; see also Bednarik 1991: 29) illustrates, or any of the other geological and vegetation mechanisms he describes in his paper.

Frank (1980: 145-146, emphasis removed) continues his survey:

Some modern activities may unintentionally leave confusing markings in caves. Hobnailed or tricounied boots can leave [line] marks on quite hard rocks. Any works that have been undertaken such as guano mining, tourist cave development, gating and digging for new caves, including the use of explosives, may also leave their mark, not to mention any casual marks as a result of routine visitation….These include [by] the blatant graffitist intent on naming his [or her] activities wherever possible, the casual doodler who has just found a nice medium to indulge in, and also the intrigued discoverer who having found some engravings proceeds in the name of science, to make his [or her] own set to work out how the originals were made.

None of these sources are likely to have produced the bulk of the markings on the Koonalda boulders. This leaves, as possible sources, animals and deliberate prehistoric human action.

Human Origin?

Do these markings themselves display any peculiarly human characteristics? Obvious geometric or otherwise recognizable motifs do not appear. Are there other signs of human use? Humans used Rockfall C. Wood (often as small twigs and often now crushed to powder) and charcoal (small, burnt stubs and large pieces of burnt wood) litter many parts of the floor and sit on some rocks. Some of the bone remains of larger animals show cut marks. A number of pieces of bone, including a set of vertebrae, lie under floor slabs, together with small, lined pieces of limestone (Sharpe and Fawbert In prep.; Sharpe and Whitehead In prep.: 10). Some floors appear cleared of stones that have been placed on the floor’s peripheries; stones set up against one lined surface produce such an enclosure (see Plate 3). A flint flake perhaps used as an engraving tool (Sharpe and Whitehead In prep.: 23; Sharpe and Fawbert In prep.) was uncovered in an area of the Upper Chamber where flint does not occur naturally. As well, line markers if human may have used the cut pieces of bone.

Hand marks and flutings obviously of human origin exist on the soft portions of the wall further into the Upper Chamber (Faulstich 1992). Present also are abraded grooves, large humanly made line markings. They and the boulder lines occur together near the Squeeze end of the Upper Chamber (Maynard and Edwards 1971: Plate 25; Sharpe and Fawbert In prep.).  (The three sorts of markings – boulder lines, abraded grooves, and flutings – resemble each other, though they differ in their tools of execution and their media.) Rockfall C with its boulder lines probably continues the same floor as at the Squeeze with its boulder lines, where the finger flutings and abraded grooves appear on the walls and ceiling (see Sharpe and Fawbert In prep.), with more recent rockfall debris covering the floor between. Further, climbing under the surface rocks between Rockfall C and the Squeeze leads to marked boulders, twigs, and charcoal similar to those in Rockfall C.

This conclusion about the cave floor answers one of Bednarik’s (1991: 38) objections to the human origin of the boulder lines. He thinks that the visible scratch-like lines

are more recent than the huge roof falls that have taken place here….They, and the wall scratches, refer to the most recent level of the talus. Since the finger flutings in the same passage (Gallus 1968) precede the most recent rockfalls, the marked clastics cannot be chronologically compatible with them.

The boulder incisions (‘the marked clastics’) if human may well be ‘chronologically compatible’ with the flinger flutings.

Two obvious features of the markings may count, however, against a human origin and for an animal origin: the lines occur in units of parallel lines that look like animal scratches, and the density of the markings is so great in places that separating out units or individual lines becomes impossible. On the other hand, abraded grooves and flutings usually also occur in units or (sub)parallel lines and their origin is human. The flutings, too, are dense in places.

Animal Origin?

Turning to discuss a potential animal origin for the boulder lines, several matters emerge from the literature and from investigations.

The literature on the fauna of the present and prehistoric Nullarbor Plain (in which Koonalda Cave lies) helps form a list of potential line makers. Bednarik (1991: 38-39) lists the animals represented by remains inside the cave: Sarcophilus, four genera of Dasyuridae, Peramelidae, and Macropodidae, including possibly Macropus rufus. Note the difference when considering the occurrence of species in the cave between the Gallus Site and the Upper Chamber sections of the cave; presence in the former does not imply presence in the latter, especially given the cliff between them. The list can omit animals of insufficient size.

An investigation of the behavior and scratchings of species on the list can be undertaken (see Sharpe In prep.). Some potential animal line-markers may have frequented the cave regularly. The Tasmanian devil probably used the Gallus Site area of the cave as a lair (Thorne 1971: 46; Walter and Pledge 1967: 43). It may have also used the Upper Chamber, though, unlike the Gallus site, the Upper Chamber lies in total darkness. Do the bone remains in Rockfall C show signs of gnawing by this animal? Does this animal mark its territory with its claws or otherwise scratch rocks? The Tasmanian wolf was evidently also active in Nullarbor caves.

Humans and animals have had access Koonalda Cave through the current entrance probably for tens of thousands of years. Explorations have turned up no other entrance to the cave or indications of one. (An entrance may have existed at times of lower sea levels from other caves via channels now flooded beneath the lakes of the current upper water table.) However, access to the Upper Chamber from the Nullarbor Plain requires the ability to traverse three substantial cliffs: the doline, the entrance incline, and the slope from the Gallus Site to the Upper Chamber, the last in darkness. Could non-climbing megafauna such as kangaroos access the chamber by themselves?

An animal able to see in limited light might sharpen its claws on or otherwise scratch the rocks, especially a trogloxene that might use the Upper Chamber as a lair. Given Rockfall C’s location in pitch-blackness, its difficulty of access, its depth within the cave, and the freely available rock on the surface of the Nullarbor Plain, however, it seems most unlikely that a non-trogloxene would go there to perform some daily activity that requires clawing. The same applies to scratches resulting from non-trogloxene animals casually wandering around. Their presence in the Upper Chamber suggests, therefore, their being carried in there, or their climbing in to escape danger, or their falling into the cave by accident and climbing to the Upper Chamber.

At least a proportion of the animals would not have escaped the Upper Chamber and would have died there. The remains of larger animals in Rockfall C and, more generally, in the Upper Chamber comprise dispersed vertebrae, crania, and cut bones (all perhaps of large kangaroos). What do these bones say about how they were deposited here? A scavenger may have removed much of the remains of any trapped and dead animals.

The animal candidates’ continuous use of the cave suggests another line of inquiry. Unlike Rockfall C, the most recent rockfall, Rockfall E, contains no markings, bones, or torches. Why did the animals whose habitat remained much the same not mark this rockfall as well as Rockfall C? Rockfalls A, B, and D are also unmarked. Only the smooth rounded boulders show the large number of markings. Perhaps the other rockfalls (A, B, D, and E) occurred after the line marker, for instance the Tasmanian devil, became extinct on the mainland of Australia. Alternatively, perhaps geological processes have removed the lines from the rockfalls, though this probably requires an unrealistic time scale (see Sharpe and Fawbert In Prep.). Further, the presence of bones within the unmarked but rough rocks of Rockfall D requires an explanation.

If an animal fell down the first or second of the access cliffs (unlike these two, the third cliff requires climbing upwards), this may imply something about its condition once inside the Upper Chamber. Perhaps frightened or injured animals made the lines in frantic attempts to escape the cave. Perhaps sliding and trapped animals made many marks in multiple attempts to scramble up rocks (Bednarik 1994a: 36-37). Is franticness apparent in the lines?

Some of the above ten points about the lines on the boulders also further the animal-related discussion:

·        Two paws could have made those units with more than five lines.

·        Units in mirrored pairs suggest that execution at the same time by the left and right paws of an animal.

·        While a multiple animal-origin (or a multiple swipe by the same animal) may explain branches and line crossings within a unit – slashes a second time creating the crossings or branches – those that appear as a single movement lend support to a human origin.

Conclusion on the Koonalda Lines

This preliminary study of the Koonalda boulder lines remains inconclusive and shows the need for further research. Some matters may count for an animal origin and some may count for a human origin. Neither ought one to assume that all the lines were made by animals or all by humans (Flood 1997: 48). Perhaps animals scratched the finer ones and humans the coarser ones. Just visually referring to Plates 1-3 perhaps Plates 1 and 3 may show human markings while Plate 2 may show animal scratches. In this case, human and animal lines would mix with each other in this area of the cave, as occurs with bear scratches and human line marks in many European caves.

Bednarik’s Method of Triangulation

What made the lines in caves like those on the Koonalda boulders: animals or humans or both? The discussion on Koonalda is inconclusive, but how could one definitively distinguish between the lines that different agents make? A definite, methodical, and objective method ought to be the aim (Clegg, et al. 1977).

Bednarik promised ‘clear criteria for identifying animal scratch marks’ (Bednarik 1993: 3).  An examination of his writing on the subject shows, rather, a lack of conclusiveness, though he moves in the right direction. His case and the critique are as follows.

A Triangulation of Coordinates

Lines that only animals could make are a key in Bednarik’s case. Princess Margaret Rose Cave, South Australia, was, he believes, inaccessible to prehistoric humans (Bednarik 1991) and the marks in it, therefore, must be animal made. Aslin provides evidence that humans could and did not enter Tantanoola Cave as well, at least until modern times. The lines there, therefore, he says, should be taken as animals scratches.

Alsin provides other criteria for distinguishing animal from human lines (alluded to as well in the writings of Walsh and Bednarik):

If…marks occur with very easily identifiable petroglyphs, i.e., circles divisioned, barred, or concentric and parallel grooves, etc., then one will need to examine the markings more closely [because they may be human]. If the markings are on the wall alone and there are no other visible signs of floor surface habitation of early people, then the marks should not be taken as human markings (Aslin 2003: 1).

In other words, if not obviously human then they are probably animal. Aslin continues:

Invariably claw marks occur near old high water levels suggesting that when an animal falls into a cave it usually goes down to the smell of water, a natural instinct for survival. It becomes disoriented and tries to claw its way out of its watery environment until it grows too weak and eventually dies in its watery grave….A few meters above its resting place would be the claw marks of the animal left in its desperate attempt to escape the water….On the first contact with the wall, the claws are splayed and will dig deeply into the surface, gradually tapering off, becoming a more shallow mark and quite often it will dig its claws in deep for one last grasp at the wall just above the water line (Aslin 2003: 1).

Find an environment where it makes sense that animals would scratch, examine those marks and use them to typify animal scratches.

The markings on the boulder, as being low down, suggests they would be animal….On  rock faces, patterns and density of claw marks are uniformly similar in caves so when a researcher enters a cave, they would expect to see, through experience and knowledge, the same repetition of marks (Aslin 2003:1).

The same principle might apply to human marks:

I have not seen any scrapings that could be attributed to an animal in places that they could not have reached either by standing upright or jumping from a ledge or rock. If it was on a sheer, vertical wall, above animal height but in reach of a human then perhaps [a person] could have made them and my interest is piqued (Aslin 2003: 1).

Not only can Bednarik tell through such means that particular lines were animal made, but sometimes he can also tell what species made them. ‘Species differentiation is often possible between markings, partly because they permit inferences concerning the substantially varying climbing abilities’ (Bednarik 1991: 38). Expanding on this, he writes (1991: 40):

Tentative species identification from claw marks is often possible, and occasionally definite identification, by a detailed assessment of factors such as: line spacing between individual sets; configuration of claws at point of commencement; height (where the former floor level can be reliably inferred); inferences regarding mobility of fore leg and shoulder; indications of body size (by spatial restrictions); and impressions of paws on soft or soot-covered surfaces.

Then we might consider the reason why an animal might scratch. Aslin (2003: 3) writes:

Scratches can be differentiated from each other and there are many variations. 1) Frantic Scratches. These are a lot deeper. In some caves I have seen where the limestone walls have been turned to powder from repeated clawing at the walls. 2) Climbing Scratches. Usually long. Shows deep impact to the wall, gradually tapering and becoming shallow. In most cases animal claw marks are deeply cut, but some can be shallow. In both cases the mark is vee-shaped at the end of it….Animals, using their paws to make marks, leave marks from all claws leaving a set of claw marks.

Scratches extend to 5.5 meters above the present floor in Princess Margaret Rose Cave. Bednarik suggests that they may relate to desperate attempts by animals to escape, or they may indicate former cave floor levels. Scratches even higher up exist in Robertson Cave and, in Gran Gran South Cave, he found a marked shaft, 18 meters deep.

Animal claw marks can be found at most levels, but they are particularly dense wherever upward progress was impeded especially in an aven about 10 m from the floor. The numerous distinct claw scratches on the perfectly flat, horizontal underside of a block measuring about 1 m across, located some 9 m above the floor, bear witness to an incredible climbing ability (Bednarik 1991: 35).

Bednarik wants to ascertain three things what he calls ‘Species,’ ‘Vehicle,’ and ‘Motivation’ about a unit or a cluster of marks:

1.      Species: the animal species that made it.

2.      Vehicle: the part of the animal’s body that made it. And

3.      Motivation: the purpose or reason why the animal made it.

He adds (Bednarik 1991: 39): ‘I submit that the status of an animal-made mark is determined by [these] three co-ordinates….The essential characteristics of each animal mark are determined only by these three factors.’ Bednarik probably intends that each coordinate of values for these three variables provides a unique set of scratches. Details for this series of points to create the triangulation of the three coordinates will, Bednarik claims, lay the ground for claiming or discounting an animal origin for particular lines. If certain marks make sense within a particular triangulation, then the animal species that the triangulation represents did scratch them. He even claims that it renders animal scratches ‘self-evident’ (Bednarik 1991: 28). Humans may have made the others.

A Matrix of Variables

This method of triangulation requires further thought and work. Certain marks suggest that a specific activity for a specific reason by a specific animal’s specific body part occurred at that place. If so, does that behavior there seem reasonable? For instance, certain marks high up in a chimney may all suggest scaling to escape. Do these same marks appear on a flat rock in the daylight portion of an easily accessed cave and, if so, do they also really suggest scaling to escape? The challenge facing Bednarik is whether he can be so consistent about scratches that Species/Vehicle/Motivation forms a self-contained and predictable triangulation, a science; Bednarik appears to think so, but he has yet to establish and publish this.

Isolating the species provides a range of possible scratches. The vehicle reduces this range. And the reason reduces it even more. But room for ambiguity still exists. Perhaps the three factors need the addition of another, such as location of the scratches, to produce a more adequate matrix of variables. For a pair of units, suppose research narrows them down to the forelimbs of Macropus titan, an extinct giant kangaroo. Bednarik’s triangulation would now provide a range of reasons for the animal to scratch the marks including, say, fear (thus trying to escape) and claw sharpening. Another factor, perhaps location of the markings, would help further narrow down the range of possible reasons. Do the scratches appear 18 meters up a pitch-dark shaft or two meters up a wall in the lighted portion of a cave that the macropus titan can easily access? The answer to this type of question offers another coordinate to form of matrix of factors that helps Bednarik’s original triangulation more adequately describe animal marks.

A fuller discussion of these four variables follows.


For a particular cave, a list of potential animal scratch makers needs to be compiled.

1.      Which animals inhabit(ed) the cave, the region, or caves in the region? (Note that many marks may be megafaunal marks (lines that large extinct species made) (Bednarik 1991).)

2.      Have excavations locally or in this cave uncovered the remains of any animals? If so, which ones? Excavations may add other information as well, for instance whether bones remain near certain lines may influence the answer as to what made them. Aslin (2003: 3) writes:

Some…bones have been identified being those of the extinct Protemnodon (giant wallaby) and Sthenurus (kangaroo species). These bones lay under claw marks on the walls, made by the animal when it became trapped. The marks are near the lower part of the walls and floors and are quite deep and spaced approx. 18 mm apart. Above these are smaller sets of claw marks noticeably different from the lower ones.

The list of species so produced needs to be narrowed down:

3.      In a stable environment, the scratching behavior by the same species would probably remain constant over time – with old and new markings visible – demonstrating a continuous use of the cave for similar if not the same reasons. Thus, marks stopping at a particular time may imply that animal’s local disappearance and this may help narrow down the list of potential line makers.

4.      Because the concern has to do with humanly made lines similar to animal scratches, many bat and bird species are too small to consider (Bednarik 1991: 31-32).

It may be possible then to roughly prioritize the remaining list by the likelihood of the species having been in the cave:

5.      Which local animals venture(d) at times into caves, and which ones into the darker regions of caves?

6.      What species have been identified from excavations in this cave?

7.      If the cave is such that animals might fall into it, what animals might do so?

8.      If an animal’s paw prints appear in (once) soft or sooty surfaces, what was the species?

In addition, the list may probably want to distinguish between relevant ages of the animal since, for instance, juveniles may produce lines of different sizes from adults.


A particular species has several vehicles, as Bednarik calls them, with which to mark the rocks, walls, and ceilings (front paws left and right, hind paws left and right, antlers, tusks, fur, and so on). These need listing.


The Species and Vehicle factors provide two coordinates for the matrix that may lead to descriptions of animal scratches. The third factor perhaps necessary for descriptions is Motivation. Bednarik writes: ‘Once we have considered why animals might mark cave walls, we will find it considerably easier to recognize the ensuing marks’ (Bednarik 1991: 28). Given a list of all possible past and present animals that do or may enter the cave, their behavior needs investigating, especially behaviors that might result in line markings. Bednarik thinks that animals would mark inside caves for many reasons, which may differ from place to place. Some of the following possibilities come from his lists (Bednarik 1991: 39-40; see also Bednarik 1991: 37-38; 1994a: 37; 1994b: 59):

1.      Usual daily climbing and walking around (by trogloxenes; in Australia, Bednarik (1991: 31-32) suggests, mostly possums).

2.      Incidentally from other everyday and normal behaviors (‘the marks of [flying] species, the polishing of rock by countless bodies rubbing against them, the scratches of the horns of ungulates in cave entrances, or the marks caused when constructing burrows’ (Bednarik 1991: 40)).

3.      Marking territory (by trogloxenes).

4.      Mating behavior or excitement.

5.      Exploring further into a cave (by trogloxenes, during which they may clear narrow passages and, in doing so, may scratch rock surfaces).

6.      In play (for instance by adolescent trogloxenes).

7.      Claw sharpening (implausible, Bednarik thinks, for most marks).

8.      Stretching (for instance by cave bears after hibernation).

9.      Attempting to escape (because of a blocked an unreachable cave entrance).

10.  Fighting (especially in a confined space).

11.  Struggling when caught in snares or trapped under rock falls (especially for deep horizontal scratches).

12.  Under stress from pursuit by a predator (perhaps withdrawing into the deeper parts of a cave and seeking a way to escape). And

13.  In pain from serious illness or injury (under which an animal may seek refuge in a cave or because it fell down a shaft).


The reason why an animal scratches inside a cave suggests where the markings might occur, as well as how the animal made them. The reverse question also applies. What does location imply about the lines that potential animal scratchers might make there? Thus, an important area for research, and one that Walsh and Aslin are quite aware of, relates to the position of the markings within the cave. These location matters allow the narrowing down of the results from the above investigations of Species and Vehicle. Location considerations include:

1.      High markings require a climbing animal or one with a long reach, or an airborne one, or higher floor levels at the time of scratching (an empirical supposition that merits investigation). High climbing markings may imply natural and agile climbers; if so, perhaps sliding and scratching should not appear. Sliding and scratching high up may imply a non-climbing species. Certain extant species that lack the ability to climb can evidently gouge symmetrical marks.

2.      Ceiling markings narrow the range of animal species.

3.      Low markings require an animal that can reach its paw down that far and at the appropriate angle. Similar lines both high up and low down narrow the potential species.

4.      The markings’ accessibility – or at least accessibility at the time of marking – also narrows the range of animal species. Marks found under overhangs or in small holes, crevices, and even tunnels may imply something about an animal line maker’s agility and size.

5.      The type and condition of the rock bearing the marks at the time of scratching may limit the number of species. Are any strong enough and do any have claws sufficiently sharp and hard to incise the rocks?

6.      Markings in darkness may lead to the question: Which animals are trogloxenes and might venture into the dark regions of a cave? Which animals can see in the dark? How much light do they need so they can see? Which animals are comfortable being in the dark? Trogloxenes may scratch a certain range of marks whereas, if a non-trogloxene not commonly found in dark places entered an unlit section of a cave, it may scratch another range of marks because it may be feeling something quite different, like fear.

7.      Marked versus unmarked surfaces: not all surfaces in a cave are marked, so why did animals only mark where they did?

Scratch Descriptions

Research following the above provides the ranges of the four variables from which to create a matrix to help understand animal scratches. Each four-valued coordinate in the matrix will produce a list of characteristics of animal scratches (Bednarik 1991: 40). For example, the claws of the front left paw of an adult macropus titan, trapped in the dark and trying to escape, might produce lines two meters up a limestone wall that look like…. To help create this list of what the scratches look like may require the results of animal studies. Several matters deserve consideration:

1.      What, if anything, distinguishes the look of frantic scratches?

2.      How do the forelimb scratches of an animal compare with their hind-leg ones?

3.      Given its physical size, agility, climbing and flying capabilities, where could the animal make lines: on ceilings, near the floor, and how far up the walls?

4.      The number of claws an animal has relate to the number of lines in a unit it makes.

5.      The configuration and relative positions of the claws on the paw that made a unit may imply something about the configuration of the marks, especially at the beginning of the unit.

6.      Claw length may correlate with the depth of lines in a unit. Short claws may not make deep lines, for example.

7.      The shape of the claw that made a line may help determine its cross-section.

8.      The potential inter-claw spacing an animal may achieve might relate to the width and spread of lines in a unit it scratches.

9.      An animal whose claws are fairly set in their relative spacing may lead to a consistent inter-line spacing in the scratch.

10.  What species scratches the marks partially determines the length of the unit. For units of lines on a vertical face, the distance that a paw can draw in any one stroke with the animal standing on the floor is relevant, as is the animal’s reach.

11.  The relationship between the length of the unit and the inter-line spacing may prove important unless the animal was sliding. Closely spaced and shallower lines in short units may require small animals.

12.  The distance between an animal’s paws restricts the distance between a pair of units that the animal’s two paws make together.

13.  The mobility of an animal’s shoulder and fore leg suggest where it may scratch and something about what its scratches would look like.

14.  An animal that initially applies its full force on the wall, gradually removing pressure on its paws as it moves them down the rock face, may produce a vertical unit that cuts deeply at the top, becoming shallower and narrower until it disappears.

Creating a matrix for any site, or a database of information that may be drawn upon for constructing a matrix for a particular site, with a photographic and descriptive archive detailing what scratches would look like corresponding to the variables in the matrix, has yet to be done.

Descriptive not Determinative

However, what if humans made lines that look remarkably like animal scratches? People could imitate the scratches or accidentally incise lines that look like them. Or perhaps they could use severed paws to mark the surfaces. Bednarik’s triangulation thus probably will not help support an animal origin for particular marks, unless:

1.      other factors such as location of the markings augment the triangulation to form a more complete matrix of variables;

2.      the matrix is self-consistent;

3.      the matrix moves from hypothesis to tested and established theory; and

4.      research can isolate and rule out human manufacture.

Even when these are addressed, the matrix would provide a description for animal scratches not a proof of their origin; the matrix is descriptive not determinative. To differentiate definitively between animal and similar human marks requires a set of separate and objective criteria.

The following two topics ask questions of a particular cluster or unit of scratch-like markings to see if humans or animals made them. Hopefully, they, building on the above discussion of Koonalda and other caves, together with the extension of Bednarik’s method of triangulation into a matrix, may help us prepare to forge better tools for the investigation of Koonalda and other sites.

Characteristics Suggesting a Human Origin

The physical dimensions, positions, and character of the markings may provide some required clues:

1.      Striations. Robert Edwards (1965: 17-18) wrote the following after his attempt to reproduce abraded grooves at Nackara Springs, South Australia, with pieces of local quartz:

By experiments with fortuitous pieces of this rock, having sharp edges, it was possible to produce markings identical with those made by the Aboriginals. The action of cutting the grooves caused little damage to the edges of the stone fragments used. A close examination of plastic negative impressions of the grooves show that the margin of the cutting tool was uneven as minute [striations] on the walls of the V-shaped cavities are evident.

Another writer says that cuts with ‘small parallel grooves within the main [line]’ suggest a jagged cutting edge, whereas a claw ‘would leave a single, flatter groove’ (Anon. 1982). (Steel cutting edges also do not usually leave striations.) If animal claws are smooth, the existence of minute striations inside a line marking may indicate its human manufacture with a piece of rock (not necessarily flaked or trimmed for the purpose). Difficulties may arise for this criterion with fine lines and a granular rock surface because it can hide evidence of secondary striations.

2.      Cross-sections. Do the cross-sections of the lines suggest the use of particular tools? A correlation between answers to this and point 12 below may suggest a human origin.

3.      Branches. A number of units on the Upper Chamber boulders in Koonalda Cave contain ‘branches’: lines that the same instrument makes (with the same cross-section) and that branch off along the length of the parallel lines in the unit. This may indicate a human origin because animals may not scratch this way.

4.      Crossings. A similar situation occurs with sub-parallel lines in a unit crossing over each other.

5.      Number of Lines. One paw of five claws may not produce a unit of six lines; probably, only humans could.

6.      Upward direction. Careful inspection of the two ends of a line may reveal the direction in which it was drawn. A human may have made any line that begins toward the bottom of a wall and moves up it. The alternative suggests an animal airborne or one tall enough to reverse its paw through that height, or one reaching downward and scratching upward.

7.      Horizontal direction. Aslin (2003: 2) writes: ‘Animals only leave marks in the vertical or near vertical position. They are unable to scrape sideways for any length of time.’ Animal clawing may not usually produce horizontal lines on a vertical face, especially in higher places (Bednarik 1994b: 62; 1991: 32).

8.      Inter-line spacing. Significant variations in inter-line spacing along a unit, the lines wandering apart and together so as to exceed possible digital flexibility, may indicate that a human cut them separately.

9.      Same instrument. The same instrument’s producing various sized and interline-spaced units in the cluster may reinforce a human-origin impression.

10.  Design and style. The presence and especially the repetition of recognizable motifs or geometric forms or composed clusters of units follows Aslin’s (quoted above) and Frank’s criterion for distinguishing between Aboriginal and other markings. Writes Frank (1980: 146): ‘Things to be taken into account are whether there is any style or design to the markings, for example, figures, patterns, grids, concentric, parallel or bifurcating lines.’ In many of the clusters that Alexander Marshack (1977; 1979) analyzes, units follow touching one another to build a meandering sequence, or a series of lines cut across one or two in a comb-like manner. Motifs, forms, or composition may imply a human origin, while the absence of incontrovertible motifs, forms, and compositions does not imply animal origin or preclude a human origin; geometric motifs lie one side of a fragment of mammoth ivory that Marshack (1979: 280-281) describes, while on its reverse lie ‘non-composed’ markings.

11.  Aesthetics. The recurrence of aesthetic elements, which use the rock shape and surface, may point to the human origin of the markings. These may include an ‘enhancement of natural hollows or protuberances’ (Frank 1980: 146) or the imitating of surface cracks (see also Marshack 1979: 305).

12.  Implement remains. Possible implements that human line-makers may have used include: a flake of flint or other rock, a sharp stick, a sharp piece of bone, a sharp piece of shell, small flakes set parallel in a resinous substance on the end of a stick, a paw of an animal set with its claws extended, and hand-held or hafted incisors. Does archaeological evidence exist from the site in question or any other appropriate site to support the presence (and perhaps use) of one or more of these implements, and does ethnographic evidence exist to document their use in relevant societies (compare with Arndt 1962; Gould 1969: 121-128; McCarthy 1967: 85, Fig. 65 (6))?

13.  Elevation remains. High human lines may have required the use of specific tools such as long sticks or platforms. The existence of the remains of such evidence may support a human origin of high marks.

14.  Human continuity. An argument from continuity might corroborate other evidence of a human origin for line markings. It centers on readily discernible types of markings found in a cave. Some markings are of indisputable human origin; for example, handprints or finger flutings drawn over a soft surface. Abraded grooves also have a probable human origin and, in Australia, a continent-wide distribution. If markings like these lie in close proximity, perhaps humans may have (but not conclusively) also marked other close-by but more debatable types of lines. The human-origin thesis may find further corroboration if the immediate area of the lines contains evidence of human use, such as charcoal, stone tools, and cut pieces of bone. The argument from continuity places some of the onus on animal-origin proponents to find evidence for their case when they accept the human origin for the other markings at the same location, particularly when additional evidence points to human use of the cave. Excavations in Kintore Cave support human presence near a marked boulder (Mulvaney 1975: 189-190). Walsh (1964: 90) could distinguish two types of markings in the cave; is one animal-made and the other human-made?

15.  Scratch imitations. A similar line of argument comes from Bednarik (1994b), though others also mention it. He notes the occurrence of flutings close to animal marks (cave bear scratchings in European caves) and suggests that the human flutings might imitate the animal markings. See Plate 4. Paul Faulstich (1986: 162) also suggests that perhaps animal claw marks provided the impetus for human flutings. In Bara Bahau Cave, France, for example, flutings lie over bear scratches (Bednarik 1985: 87; see also Gallus 1968: 45). The same scenario could apply to line incisions, that animal scratches inspired humans to create imitations. Bednarik (1986: 166) suggests that, while no objective evidence supports a cave bear cult in Paleolithic Europe, many caves contain charred and smashed osteological material; perhaps humans venerated the cave bear and replicated its claw markings with their own scratch-like incisions. Perhaps, following the same line of thought, Paleolithic Australians similarly venerated a megafaunal kangaroo and replicated its claw markings. The Upper Chamber of Koonalda Cave contained a cranium (perhaps of a kangaroo), vertebrae, and other cut bones, perhaps indicating a relationship perceived by the human cave users and the animal.

Plate 4. Human flutings in association with probably cave bear scratches. The former in places appear to imitate the latter. Rouffignac Cave, France.

16.  Severed paws. Did humans create lines with a severed animal paw that mimicked animal scratches (Hallam 1971: 101; but see Bednarik 1987-1988; 1991)? Lines that the claws of live animals scratch probably look the same as those that humans engrave with paws held in their hands. Can they be distinguished? This question requires a special discussion because now the human instruments are the same as the animals’:

a.       While it may be logically possible that humans so used paws as engraving tools, is it a reasonable assumption that they did? Aslin (2003: 2) writes: ‘Early people, in my view, would not have used animal claws to make scratch marks. Although kangaroo and wombat claws are hard enough to be used as rock engraving tools, there is no evidence of this found in the caves visited.’

b.      How might humans extend the claws from the paw of a dead animal (for example, might they pull on the tendons)? For the claws to remain extended, need the paws be somehow set rigidly (the tendons permanently extended, for instance)? The lines in a unit may, if this were the case, stay equidistant from each other whereas lines that the claws of a live animal produces may vary in separation, if that digital flexibility exists.

c.       Of assistance in distinguishing such tool use from animal clawing, Aslin (2003: 2) suggests: ‘For experimentation, a human hand holding an animal claw in it can hit the wall harder than an animal. This leaves a deeper groove with the impact also fracturing the surface. The animal would not have the power to do this.’

It is one thing to argue against an animal origin for line markings. To show clearly a human origin is another. The case must have its own empirical support and must answer potential objections.

Characteristics Suggesting an Animal Origin

For suggesting an animal origin, the following physical dimensions, positions, and character of the markings may prove important:

1.      Human non-accessibility. Sites with markings that humans could not or demonstrably did not make, but only animals could for instance, sites that offered no human access suggests important information about animal markings versus human ones. The data thus gained may prove useful for those sites where the origin remains in question. As mentioned above, Bednarik has pursued this approach with apparent success (Bednarik 1992; 1994a; 1994b).

2.      Prick marks. Prick marks at the beginning of a unit suggest that extended claws hit the surface before scratching over it (note that humans using a severed animal paw with its claws extended may produce the same prick marks). See Plate 5.

Plate 5. Probable cave bear scratches with prick marks in association with an engraving probably of a bear. Rouffignac Cave, France.

3.      Scrapings. An animal paw with claws extended and traveling over a soft rock surface may scrape off or scoop out some of that surface. This may especially show at the point of impact of the paw on the surface.

4.      Inter-line spacings. Writes Aslin (2003: 2): ‘Animals dig in with their claws, the marks becoming sharper, usually leave a fairly uniform space between them, not coming together.’ If true, this may offer a distinguishing characteristic.

5.      Depth and width. An animal origin might be suggested if most of the units comprise two to five parallel lines, each usually deeper at one end and becoming shallower and narrower until disappearing (Aslin, et al. 1985).

6.      Mirroring. Units in pairs that mirror each other and have similar physical characteristics may suggest two front paws both dragged over a surface at the same time. Such pairs of units could converge toward each other at their base. The separation between the two units must correspond with the potential paw separation of the animal in question. (Note that severed paws held in each hand of a human and used at the same time may produce this effect as well.)

7.      Animal continuity. An argument from continuity that parallels that for human origin might bolster an otherwise-established animal origin for line markings. If some markings are of indisputable animal origin, similar and close-by markings may also have an animal origin. The animal-origin thesis may find further corroboration if the immediate area of the lines contains evidence of animal use, such as cave bear hibernation pits. The argument from continuity places some of the onus on human-origin proponents to find evidence for their case when they accept the animal origin for the other markings at the same location, particularly when additional evidence points to animal use of the cave.


The above may help investigators determine whether a human or an animal created a set of lines. They could check the points against the scratches in animal-only caves:

·        Do animal scratches contain internal striations?

·        Do any animal scratches include internal branches or cross-overs?

Investigators may also wish to isolate other physical characteristics of the marks and then look at their line-maker implications. Controlled experimentation with animals could prove invaluable as well. For instance,

·        What would the result of frantic scratching behavior look like?

·        Do climbing scratches differ from stretching scratches?

·        In general, do the scratches differ that a species makes for different reasons?

Line markings that people could make in laboratory situations may provide another avenue to explore. For instance,

·        Did humans use the paws of dead animals to make lines and, if so, are there any observables that distinguish them from lines that living animals make?

Koonalda Cave is a classic site for prehistoric ‘art’ in Australia. Since the discovery of the flutings and abraded grooves within the cave, many hypotheses have tried to explain their presence (Flood 1997: 47-50). The evidence for the human use of the cave, including for those areas with markings such as Rockfall C in the Upper Chamber remains persuasive. Bednarik’s (1991) denial of the human origin for all scratch-like markings on the boulders of Rockfall C may not survive. Only further investigations, armed with a solid theoretical base, will help.

More generally, Bednarik (1991: 40) assumes that animals made the scratch-looking lines unless proven otherwise. He suggests, as a way to determine animal origin, the adoption of a three-dimensional model whose coordinates are Species, Vehicle, and Motivation, so that knowing two coordinates implies a small range for the third. A broader approach may work better, one that looks for tell-tale signs of animal and of human origin in the marks themselves, and that uses a matrix (Bednarik’s triangulation plus location of the marks) to see if it reinforces an animal origin and to help understand this behavior if animals did make them. Creating such a matrix and its resultant list of scratch characteristics remains to be attempted.

The lines and other evidence of human presence in totally dark portions of caves like Koonalda, many difficult meters from an entrance, can intrigue an investigator. Why did people go there? The line markings are important to help understand the behavior of prehistoric humans. For the flutings, grooves, and humanly made incisions, a question to consider concerns the tradition and significance of the markings. Why did people make them? A closer study of the marks and how they were made may lead to observations that increase our understanding and, in turn, produce other questions for further research (Sharpe and Lacombe 1999; Sharpe, Lacombe, and Fawbert 1998; In prep.). They also limit and may inspire hypotheses as to the original meaning of the lines. Whatever approaches are used in such a meaning investigation, they ought to move beyond pure speculation to proffering options open to consideration and assessment against data. A satisfactory answer to the meaning question remains a long way off and may never be attained.


The points about the lines in Rockfall C, the Upper Chamber of Koonalda Cave, draw on Christine Whitehead’s analysis carried out in 1976 as part of an expedition under the auspices of the National Geographic Society and the South Australian Museum. Thanks are also due to these two institutions, as well as to Sandor Gallus and the Gurney family of Koonalda Station. The paper benefited from the comments of Geoff Aslin, Robert Bednarik, Ben Gunn, W. D. L. Ride, Helen Fawbert, and Leslie Van Gelder. Others who reacted to a draft of the paper include Michael Archer, Alan Bartholomai, S. V. Dyck, and Aola Richards. The figures are by Kevin Sharpe.


ANON. 1982. Who scalped Bodo – and why? New Scientist 95 (1322): 688.

ARNDT, W. 1962. The interpretation of the Delamere lightning paintings and rock engraving. Oceania 32 (3): 167-177.

ASLIN, G. D. 2003. Comments on Kevin Sharpe’s, ‘Line markings: human or animal origin?’ Typescript.

ASLIN, G. D., E. K. BEDNARIK, and R. G. BEDNARIK. 1985. The ‘parietal markings project’: a progress report. Rock Art Research 2 (1): 71-74.

BAHN, P. 1998. The Cambridge Illustrated Prehistory of Prehistoric Art. Cambridge: Cambridge University Press.

BEDNARIK, R. G. 1985. Parietal finger markings in Australia. Bollettino del Centro Camuno di Studi Preistorici 22: 83-88.

_________. 1986. Parietal finger markings in Europe and Australia. Further Comments. Rock Art Research 3 (2): 159-170.

_________. 1987-1988. The cave art of Western Australia. The Artefact 12: 1-16.

_________. 1990. The cave petroglyphs of Australia. Australian Aboriginal Studies no. 2: 64-68.

_________. 1991. On natural cave markings. Helictite 29 (2): 27-41.

_________. 1992. Natural line markings on Palaeolithic objects. Anthropologie 30 (3): 233-240.

_________. 1993. The cave art of Australia. In W. D. Hyder, ed., American Indian Rock Art: Proceedings of the 1985 ARARA Symposium in Santa Barbara, vol. 12, pp. 81-87.

_________. 1994a. The discrimination of rock markings. Rock Art Research 11 (1): 23-44.

_________. 1994b. Wall markings of the cave bear. Studies in Speleology 9 (February): 51-70.

_________. 1994c. Malangine and Koongine Caves, South Australia. The Artefact 17: 43-60.

_________. 1997. The global evidence of early human symboling behavior. Human Evolution 12 (3): 147-168.

_________. 1998. Microscopic analysis of ‘engraved plaques’ and other objects from Devil’s Lair. Journal of the Royal Society of Western Australia 81: 23-33.

CLEGG, J., D. Bell, H. Brayshaw, M. Cooldrake, E. Godden, P. Quay, G. Summerhayes, A. Urquhart, P. Wilson, E. Gage. 1977. Science in the study of Australian prehistoric art. Memoirs of the Victorian Archaeological Survey 1: 18-41.

EDWARDS, R. 1965. Rock engravings and Aboriginal occupation at Nackara Springs in the northeast of South Australia. Records of the South Australian Museum 15 (1): 9-27.

FAULSTICH, P. 1986. Parietal finger markings in Europe and Australia: further comments. Rock Art Research 3 (2): 159-170.

_________. 1992. Massaging the earth: Pleistocene finger flutings and the archaeology of experience. The Artefact 15: 3-10.

FLOOD, J. 1997. Rock Art of the Dreamtime: Images of Ancient Australia. Sydney: Angus and Robertson.

FRANK, R. K. 1980. Archaeological features of Australian caves. Proceedings of the Australian Speleological Federation’s Twelfth Biennial Conference, pp. 141-152.

GALLUS, A. 1968. Parietal art in Koonalda Cave, Nullarbor Plain, South Australia. Helictite 6 (3): 43-49.

_________. 1977. Schematization and symboling. In P. J. Ucko (ed.), Form in Indigenous Art. Canberra: Australian Institute of Aboriginal Studies, pp. 370-386.

GOULD, R. A. 1969. Yiwara: Foragers of the Australian Desert. London: Collins.

GUNN, R. G. n.d. Wall markings in Tantanoola Cave, Tantanoola, South Australia. Typescript.

HALLAM, S. J. 1971. Roof markings in the ‘Orchestra Shell’ Cave, Wanneroo, near Perth, Western Australia. Mankind 8 (2): 90-103.

JENNINGS, J. N. 1968. Recent Australian contributions on pipes, foibe, and climatic history from speleological evidence. Helictite 6 (3): 50-56.

JONES, F. W. 1925. The Mammals of South Australia. Part III: The Monodelphia. Adelaide: British Science Guild (South Australian Branch).

LAYTON, R. 1992. Australian Rock Art: A New Synthesis. Cambridge: Cambridge University Press.

LINK, A. G. 1967. Late Pleistocene-Holocene climatic fluctuations, possible solution pipe-foibe relationships, and the evolution of limestone cave morphology. Zeitschrift für Geomorphologie 11 (2): 117-145.

MARSHACK, A. 1977. The meander as a system: the analysis and recognition of iconographic units in Upper Palaeolithic compositions. In P. J. Ucko (ed.), Form in Indigenous Art. Canberra: Australian Institute of Aboriginal Studies, pp. 286-317.

_________. 1979. Upper Palaeolithic symbol systems of the Russian Plain: cognitive and comparative analysis. Current Anthropology 20 (2): 271-311.

MAYNARD, L. and R. EDWARDS. 1971. Wall markings. In R. V. S. Wright (ed.), Archaeology of the Gallus Site, Koonalda Cave. Canberra: Australian Institute of Aboriginal Studies, pp. 59-80.

McCARTHY, F. D. 1967. Australian Aboriginal Stone Implements: Including Bone, Shell, and Teeth Implements. Sydney: Trustees of the Australian Museum.

_________. 1970. The Northern Territory and Central Australia. In F. D. McCarthy (ed.), Aboriginal Antiquities in Australia: Their Nature and Preservation. Canberra: Australian Institute of Aboriginal Studies, pp. 51-72.

MULVANEY, D. J. 1975. The Prehistory of Australia. Harmondsworth, England: Penguin Books.

MULVANEY, J. and J. KAMMINGA. 1999. Prehistory of Australia. Washington, D.C.: Smithsonian Institution Press.

SHARPE, C. E. and K. J. SHARPE. 1976. A preliminary survey of the engraved boulders in the Art Sanctuary of Koonalda Cave, South Australia. Mankind 10 (3): 125-30.

SHARPE, K. J. 1982. A working paper on the meanders in Koonalda Cave, South Australia. Typescript. See

_________. In prep. Dreaming No Man’s Land. See

SHARPE, K. and H. FAWBERT. In prep. The Upper Chamber of Koonalda Cave, South Australia: Its rockfalls, their weathering, and use. See

SHARPE, K. and M. LACOMBE. 1999. Line markings as systems of notation? In News 95: International Rock Art Congress Proceedings. Pinerolo, Italy: IFRAO – International Federation of Rock Art Organizations, 1999: 46 and NEWS 95-International Rock Art Congress Proceedings_files/sharp.

SHARPE, K., M. LACOMBE, and H. FAWBERT. 1998. An externalism in order to communicate. The Artefact 21: 95-104.

_________. In prep. An analysis of finger flutings. See

SHARPE, K. and C. E. WHITEHEAD. In prep. The Upper Chamber of Koonalda Cave, South Australia: A second report. See

SHARPE, K. and L. VAN GELDER. To appear. Trois Formes de Tracés Digitaux (ou Sevérines) en Grotte de Rouffignac, France. Préhistoire du Sud-Ouest. See

THORNE, A. G. 1971. The fauna. In R. V. S. Wright (ed.), Archaeology of the Gallus Site, Koonalda Cave. Canberra: Australian Institute of Aboriginal Studies, pp. 45-47.

TINDALE, N. B. 1933. Tantanoola caves, south-east of South Australia: geological and physiographical notes. Transactions and Proceedings of the Royal Society of South Australia 57: 130-142.

TROUGHTON, E. 1957. Furred animals of Australia. Sydney: Angus and Roberstson.

WAKEFIELD, N. A. 1967. Preliminary report on McEachern’s Cave, S. W. Victoria. Victorian Naturalist 84 (12): 363-383.

WALSH, W. P. 1964. Unexplained markings in Kintore and Cutta Cutta Caves, Northern Territory, Australia. Helictite 2: 83-91.

WALTER, M. R. and N. S. PLEDGE. 1967. Vertebrate palaeontology of the Quaternary deposits of the Nullarbor Plain. In J. R. Dunkley and T. M. L. Wigley (eds.), Caves of the Nullarbor. Sydney: The Speleological Research Council, pp. 43-46.