How do we recognise rock art?

Print Friendly, PDF & Email

Aboriginal engraving

This paper was stimulated by an on-line discussion with Andrea Arcà (Case example 4: figurative and non-figurative below) about a need for archaeologists to treat figurative and non-figurative rock art differently.

by † John Clegg

TRACCE PHP-Nuke version, 2002-2011

How do we recognise rock art?


This paper was stimulated by an on-line discussion with Andrea Arcà (Case example 4: figurative and non-figurative below) about a need for archaeologists to treat figurative and non-figurative rock art differently. I have a long-term interest in the theory and methods appropriate to extract information from rock art. I am most concerned with the most difficult cases, when documentary or ethnographic information is not available to supplement formal archaeology, or when it is desirable to exploit formal and informed sources independently, perhaps as a check on each other.

The paper is quite complex, so I have separated it into distinct short chapters, sometimes illustrated with case examples. The overall argument is:-

The ability to recognise things, especially those that represent potential dangers or food, is primitive, efficient, rapid, and irrational. That makes recognition, identification, and perception difficult to analyse and deal with rationally/scientifically. It also predisposes people to think that since they can recognise depictions easily and accurately, others have and had the same capacity, and must have reached similar conclusions.

But such identification may be wrong. So archaeologists should offer substantiation for identifications. Such substantiation may be very difficult, specially if the identification of depictions is understood to refer to the original artists or users of rock art, about whom we may know very little. Particularly we may have no knowledge of what ancient rock art meant to its makers or users.

Nor, in my opinion, is it possible for us to ever fully understand ancient rock art. People who hope to gain insights into what ancient rock art meant to its makers or users may need to look outside archaeology, perhaps to the consciousness-changing techniques employed by some performances and installations.

A great deal of archaeology can avoid the hassles mentioned above, by using transparent formal methods of description and classification. It makes sense to do the testable stuff first. That means working with boring classification and typology.


Prologue: How we recognise the subject of a depiction

Chapter 1. General introduction: Graphic representation and depiction
In general, recognition is easy and irrational; it may be mistaken, and is difficult to substantiate.
Case example 1. Elvina possum/ goanna

Chapter 2. Some archaeological distinctions
An archaeological attitude: how archaeologists recognise an artefact, something made by human work (and how could we show our conclusions are correct?)

Chapter 3. How we recognise types of pictures
Classification by science and by intuition:
Case example 2: Callan point.
Case example 3: Some Archaeologists treat “iconic” or “figurative” rock art different from “non-iconic”, “non-figurative”.
Classification of rock art: Maynard’s scheme.

Chapter 4. Detailed study of form
The geometry of knobs and blobs
Comparing drawings to “identify” subjects
Case example 4: small Daras?
Case example 5: Thylacines cats, pigs and diprotodons.

Other techniques may be needed to understand and appreciate rock art. Perhaps some people, even archaeologists, believe it may be possible to understand the past, and its rock art:

Prologue: How we recognise the subject of a depiction.

It is often easy to recognize the subject of a picture, although we might get it wrong. A close examination of how we do that recognizing gets pretty complicated. For one thing, the picture (take an echidna, for example) does not resemble the real thing very much. The picture does not smell, or move, or feel spiky, or throw the right sort of shadow. It is made of stone, perhaps paint, (or metal), not made of echidna. A real live echidna is solid, not outline, it has four feet, and is not necessarily seen in profile or from above. Its head or feet are not twisted at impossible or unlikely angles.

Recording of Aboriginal engraving of an echidna from West Head, Ku-ring-gai Chase National Park (Clegg 1977:February)

Echidna on an Australian 5 cent coin

Some pictures may be of unfamiliar subjects.

Recognising the subject of a picture can be quite easy if the subject is familiar. But sometimes our knowledge or imagination is inadequate.

Tracing of an engraving about lm long at Sturt’s Meadows. There are three bird tracks (a bit like broad arrows), as well as the linear thing I want to talk about

The morning after burning off a big patch of bracken I noticed many bandicoot runs in the ash area. I was unable to map them, but they clearly ran out of holes, making loops before returning. Sturt’s Meadows has several pictures of this pattern, which could represent animals’ runs or excursions (or, as Gunn suggested in a comment on an early draft of this paper, human travels).

A subject may appear in the drawing as we are not used to seeing it.

In the rock art of the Sydney area most animals appear in profile, but animals that go about low to the ground like lizards, are shown from above, in plan. Big animals are big, small animals small, and human-sized animals are shown about human size.

‘Balloon with legs’. Aboriginal rock engravings at Basin Track, Ku-ring-gai Chase National Park (Clegg 1977:December


Two engraved figures in Ku-ring-gai Chase National Park, which looked about the same size and shape as party balloons (with two legs) were puzzling. The two examples I know of are close to engravings thought to represent penguins. I think the balloon-figures might be bird chicks, perhaps baby penguins. Young chicks are close to the ground and look spherical when fluffy. Considering the manner of drawing helps diagnose puzzling pictures.

Sometimes depictions show their subjects better than a simple photograph could.

“Twisted perspective” denotes a particular way of drawing which is widespread, well known, and much studied. It is usually explained as expressions of conflict between the subject’s appearance and the artist’s concepts. Twisted perspective consists in depicting an object from two or more different viewpoints at once, as when frontal eyes are shown in a profile face, or when profile human figures have feet showing all the toes, (some have two left feet), or when facial profiles have frontal ears. Depictions “from two or more viewpoints” conveniently includes x-ray art.

Many students have great difficulty in recognising twistedness of perspective, perhaps because their mental images are themselves twisted. Professional art schools aim to free students from any compulsion to practise twisted perspective. Twisted perspective is normal. If deliberate, it may be conventional, but it is often neither deliberate nor conscious. It is very much more common than a. renaissance or camera perspective, so it is a better norm by which to measure rock art.

The diagnosis of twisted perspective at a site argues that since some engravings at that site use twisted perspective, perhaps others do. The thought could be used as a model to try on common but inscrutable motifs. This procedure seemed to successfully demonstrate that the star-crescent-sailor motifs at Callan Point could represent a ship’s steersman concentrating on his compass.

Engraving at Callan Point

Taking a bearing (Anon. 1951: fig. 223, p. 215)

This paper is partly a search for testable, refutable ways to test the diagnosis of rock art. I wrote the passage above hoping that I would be believed, because it describes what I believe, and use as the foundation of much of my research into rock art. But I cannot think of any way to test it scientifically. It may be wrong.

Another part of this paper is a reassurance that most of what archaeology does with rock art can be done by ordinary archaeological techniques, without any need for such a test or it.

Chapter 1. Graphic representation and depiction.

This discussion is written by a person with a background, experience, and academic “qualifications” in archaeology, specifically the archaeology of rock art. It aims to set up some ideas of how archaeologists may do their job from the evidence available, which is assumed to exclude a truthful artist’s declaration “This is a picture of XXX.” It is broadly based, being aware of possible questions such as those to do with the beginnings of depiction, which may have happened before Homo Sapiens sapiens. Unfortunately my philosophy is home-made and likely to be wrong as well as out-of-date. But I hope my insights into the making of images, representations and depictions may be better founded.

It is advisable to found or base studies on observations that can be tested, verified, or substantiated – simple things like where, what shape, what materials, how made, and associations – as opposed to speculations about the contents of makers’ minds or brains. This has always been so obvious to me – even if it is not important – that I’m amazed if it is not obvious to others. As a result I have written probably too much and too long and too boringly about how one can deal with (study) rock art while avoiding too many assumptions about what the makers thought, preferring to promulgate such insights only when the reasons and arguments for them can be stated, and peer-tested.

Why identification through recognition may not be straightforward.

I do not always find it easy to communicate with other people, even those I know well, and with whom I share background and experiences. I have enough teaching experience to know that people are often unaware of things until they are pointed out to them. Richard Nisbett has shown on many books (e.g. 2003) that cultural differences of cognition are powerful and wide-ranging.

A great deal of perception and consciousness seem to be conditioned if not determined by environmental experience, powerfully mediated or reinforced through culture and language, (Rita Carter: Consciousness, Mapping the Mind). Making depictions is a further development of that subjectivity, and requires the further complications of necessary techniques and skills; reading pictures of things introduces a further stage of complexity. (Ernst Gombrich: Art and Illusion).

I imagine most of us remember the scene in Hamlet (Act III, scene 2, 391-400) where the Prince teases his elderly minder Polonius, by pointing out how a cloud can be seen to resemble a camel, then a weasel, and finally a whale. It seems clear to me that the cloud has served as a picture, moreover a picture of, these creatures. The cloud could be described as a depiction of the animals, to variable extents Figurative or Iconic or Representational or Naturalistic. There is a further complication, because the depicting happens only partly in the cloud, but it also happens (or is claimed to happen) in the mind of Hamlet, and in some interpretations, Polonius’ also.

I think that it is a mistake to believe that what you or I see now is necessarily the same as what the original artists saw. This does not make it impossible to study rock art, because rock art is artefactual.

Some people seem to have difficulty with the fact that pictures are artefacts, and open to the same techniques of analysis and study as can be productively applied to other artefacts. I think this problem arises because humans have a very strong automatic assumption to that pictures must be what they seem. Messages about pictures’ unreliability, or ability to mislead, are received with denial, ignoring, or something close to despair. Morwood (2002:158-9) wrote: Taken to its logical conclusion, this would mean that figurative art could not be used to infer anything about past things or events, … …
As my work (and Morwood’s) have shown, there are many inferences that can be made about the past from all sorts of rock art, using the traditional methods of archaeology: observing, classifying, counting, seeking associations, and all the rest. So It is not necessary to assume that what you see in a picture is necessarily the same as what anyone else sees.

I suggest there are three closely related components, which I’m presently calling object, image, and concept.

They are related so closely that it may not make sense, or even be possible to unlink them, because unlinking might change them too much.

I may have access to my own concepts, to objects they relate to, and to images of them. Although others can access the image and object, I cannot directly make my concepts available to anyone else. I can attempt to do so through various means which, though sometimes difficult, seem often to be very successful. But, as every teacher knows, communication of concepts requires the cooperation of both parties, and is easiest accomplished through many concurrent channels of communication. It can seem to be very successful. Unfortunately there is no way to be certain that concepts have been accurately and fully communicated, although various (examination and other evaluation) techniques may be applied. Attempts to communicate concepts across great gaps of culture, interest, and time, are at grave disadvantage.

The relationships between images, objects, concepts

1. images have to be images of; they can be independent of concepts, but may require a sentient audience.

images of objects can be independent of concepts, or so we must surely believe when we think of the precise detail that can be attained by footprints, shadows, and clouds or other random patterns resembling anything. Such images may be inaccurate or imprecise; they may be distorted, incomplete, or fuzzy. In shadows and footprints, reflections, and photography, the object contributes to – helps make – the image. And it’s easy to imagine the images are independent of any onlooker. But for a cloud (or pattern of damp on a wall) to be an image of anything (except itself, or its constituents, – fungi, ice crystals – doppelgangers, auras, rainbows and so on) an onlooker is required to make the connection between object and image. We call this process of connection “recognition”, and it happens (presumably, and perhaps by definition) only with the involvement of sentient beings.

A convincing example is the biological phenomenon known as mimicry, where one species mimics another, resulting in some advantage. (see Campbell, N. A. (1990): 1102) Some insects closely resemble the thorns of the plants they inhabit; many harmless snakes mimic the markings of the poisonous coral snake; some harmless insects mimic armed, stinging insects. Some butterflies and moths have frightening pictures of eyes on their wings.

These cases make it very clear that the relevant sentience is in the audience, the onlooker. I have so far resisted consideration of such an entity, because I was seeking a way to talk of and recognise depictions without needing to make assumptions about how anyone saw things in the past – I was seeking an objective way of progressing – or an acceptable approximation to objectivity. That desire seems already impractical. An ancient depiction is a depiction insofar as you or I can recognise it now.

2. Can concepts without objects have images? I will pussy-foot around “concept”, and leave for the moment the likelihood that concepts can only be expressed through objects. Animals, among them people, are very good at showing emotion, and communicating it. Perhaps emotion and other states (like hunger) is as close as we need get to “concept”. Stances of threat and submission communicate attitudes. Of course, concepts may exist without expression.

3. Can images exist without objects? Categorically no. images have to be of. The relationship “image of” is called depiction or representation. But the objects represented or depicted need not have any real counterpart. Consider the inadequate models once made of dinosaurs, or images of Gods with animals’ heads. Yes, I know “reality” may be a problem. images can be of objects or of concepts.

Summary/conclusions so far.

images are images of; they are of subjects, which may be objects or concepts. They need not involve sentient beings, for some happen through physics, like shadows and footprints. Others, including all images of concepts, require sentience, but need not require precisely human capacities. (An animal smelling another animal’s territory marker and reacting to it is good enough evidence of “concept” for me, and I hope for this discussion.) These recognitions are likely to have auras of other things they may mean, suggest, be associated with, or reactions they may generate. I don’t think I can avoid the subjectivity that, except for the purely physics ones, all representations/ depictions/ pictures of/ images of, require an owner/perceiver/audience, sort of like “environment” which is specific to individual entities.

Without a time-machine and interpreter, we cannot easily know how people saw things in the past. We may be able to reconstruct what they saw, but not how it appeared to them. Perhaps, as we know predatory birds don’t eat so many mimicking insects, we could find out something from evidence of behaviour. But better to find out how we recognise representations and their subjects, and extrapolate

Case example 1. Elvina possum/ goanna

Here’s an example I discovered recently. A year or so ago I was at Platform 0 at Elvina Track site, and I saw an Aboriginal engraving of a brush-tailed possum.

elvina possum photo

elvina possum drawing

I thought that I had I happened on it at a time when a time when the sun was just right, which partially explained why McCarthy had failed to find it when he recorded the site in 1945. Last week I looked further at McCarthy’s report, and realised that I had failed to read it well enough

McCarthy 1983 V 1, part1: p.29.

SERIES 2 (McCarthy, F D, 1954, Pl. C, Fig. 7B) is on a large rock surface on the NW side of the outcrop.

Goanna (1) 3′ long, rounded head on curved neck, no eyes, broad straight sided body, truncated legs straight out, one leg round ended, straight tail has flat foot with heel on its end, like a human foot. Groove 1/2″ wide, 3/16″ deep. 1 figure (p.158 McCarthy 1983)

My brush-tailed possum was McCarthy’s goanna with a human foot.

I think what happened was that Fred’s familiarity with engravings got in the way of his seeing objectively. He was familiar with engravings of goannas (there are 3 goannas at Elvina Track sites i.e. within 200 metres of the possum, the closest within about 50 metres.

(McCarthy 1983:vol 2: 29)

Goanna, (15) 3’6″ long, conical head, no eyes, pointed fin-like legs, concave and straight-sided body, straight tail) (McCarthy 1983: 157)

This is the classic nearby goanna engraving

Pictures of Goannas are common, but possums are rare in Sydney rock art, so he was expecting a goanna rather than a possum. In other nearby examples, in 2003 I see engravings differently from what McCarthy recorded in the 1940s and 50s, and reported later. Perhaps I have more time, and perhaps the engravings are even more shallow now, so I know they take a lot of effort to see, whereas McCarthy did a great deal in a short time, and may have found the engravings more obvious than they appear to me. This could have made him more confident of what he did see. An example is another nearby goanna engraving, where McCarthy missed the head and front legs of a goanna engraving, and identified the hind legs and tail as “Flying bird, 3’9″ long, probably a native companion or swan, long neck with round end, no eyes, slanted to its right, 2 narrow round ended wings, and a similar tail.” (1983:161)

Zoologist’s drawing of a possum (Wood Jones: 196)

Zoologist’ drawing of a goanna. (Cronin:101)

Chapter2. How do we recognise rock-art as an artefact, something made by human work? (and how might we show our conclusions are correct?)

Objects that must have been made by humans because nature does not make such things and humans do, are recognised as human artefacts. Most of the theory and practise about how to recognise artefacts was worked out in the nineteenth century with reference to knapped stone tools, and students of palaeolithic archaeology were taught about it for the first 75 years of the 20th century. The recognition of knapped stone tools depended on understanding that flint and other materials break in a conchoidal fracture, indicating where and how and in which order a piece of flint was struck. It was also necessary to understand how flints can be struck naturally, e.g. if they are held fast in a streambed and struck by passing stones.

The defining criteria “could not be natural, were made by methods available to humans” can apply to associations and assemblages, as well as to individual items. Thus “manuports” – objects transported by human activity – are recognised by association. A natural object may be moved from where it occurs naturally to somewhere else (where it does not occur naturally), associated with artefacts, perhaps to become part of a recognisable artefact. This criterion (together with an unnatural overall form) is used to recognise animals’ built nests, or ritual objects such as bower birds’ bowers. It also applies to otherwise natural shells, fossils, amber and so on found away from where they occur naturally and in association with Palaeolithic objects. (See Leroi-Gourhan, 1957, and fig 22).

At one stage stone tools were thought of as conforming to a “set and regular pattern”. That consideration meant that objects reinforce each others’ status: the more similarly-made objects found, the greater the likelihood that they are unnatural, i.e. artefacts. Unfortunately this argument works equally well for natural objects, say butterflies volcanic plugs, or sea-waves. One may be an aberration, but several become a category. This has been used by forgers to their advantage.

Close theoretical or practical examination of natural or random processes (consider brittle objects in a cement mixer) demonstrates that natural or random processes make objects conform to set and regular patterns. But they do so in a patterned way that can be understood through the statistics of chance. If a randomly-oriented six-faced object is struck twice at random, the chances of hitting the same face with both blows is given as 1/6 * 1/6 , i.e. 1 in 36. A collection of six-faced objects struck twice may be expected to show 36 objects with blows to two faces for every one object with both blows to only one face. The argument should be clear, even if my statistics or explanation are at fault. An object on its own may appear to be a convincing artefact, but be less convincing in its context. Candidates for acceptance as artefacts should be considered within their cohort of possible objects, not sorted from them.

Summary of criteria

This discussion suggests criteria for recognising artefacts: acceptable candidate objects are:-

  • made by means that are not natural
  • made by means available to humans
  • associated with other human artefacts
  • not associated with similar natural objects
  • like other human artefacts
  • distinct from similar natural objects.

To make a convincing case it may not be necessary to meet all of these criteria, but the more independent criteria the better. In accordance with Occam’s strictures, an onus of proof lies with those who are making an unlikely or particularly interesting claim.

An archaeological attitude?

Callan Point head

In 2001 (45) I claimed:
“When faced with rock art, an archaeologist may think something like this: “This rock has some unnatural marks, in the form of shallow valleys and recessions. These depressions can be read as a depiction of a human head in profile.”

Archaeologists study the marks in detail, and record them. They examine the physical details of the grooves in order to check the details of the picture, mentally flipping between picture and detail of rock surface.
A process like this may work well if the archaeologist has the necessary skills and knowledge and natural marks are easily and accurately distinguished from natural marks. In the case of Sydney rock engravings, the natural rock surface has all sorts of natural and unnatural grooves; some of the natural ones have a v-shaped cross section, whereas Aboriginal engravings are u-shaped. But there are plenty of natural grooves with much the same dimensions as the engravings; deeper knowledge of the rock and what happens to it naturally is needed.

The rock surfaced is subject to erosion, to which it has been exposed for more than two centuries since the engravings were made, so in many cases the finest details have gone. Ultimately and in practise, the groove is recognized as artificial through contrast with what occurs naturally, and can be observed round about.

The artificial groove is often diagnostic from its plan form: the groove is continuous, and after perhaps several excursions, returns to itself, forming a loop that contains an area. (It contains space.) And that loop or area or line is recognizable, essentially like other recognized pictures.

The argument that a groove is not natural because it looks like a picture can be very convincing, like many self-fulfilling processes. Such evidence is not acceptable as proof. For proof it is necessary to show that the groove is artificial in that it has characteristics that are not consistent with natural forms and processes. Those characteristics can show in the plan, in the section on in other physical criteria.

This discussion has focused on two of the criteria: rock-art things are probably artefacts if:

  • they are not natural i.e. not like what nature produces under similar conditions.

This criterion is approximated in its simplest form through understanding what forms natural processes produce on the material under prevailing (natural) conditions. Examining surrounding natural material helps.

  • they were made by means available to and practiced by the probable makers;

Evidence for this criterion is usually based on close examination, experiment, general acceptance, or information from experienced craftspersons. Things like tool-marks help enormously.

Chapter 3. How do we recognise types within pictures?

Case example 2: Classification by science and by intuition: Callan point.

It is possible to process rock art using well-established methods of formal archaeology, to describe, compare, classify, count, discover patterns of chronological and spatial distribution and association, in fact do the usual processes of formal archaeology with no recourse to interpretation, and no assumptions about rock art being pictures of things. These processes of formal archaeology can happen automatically, although if their results are to be tested, they must go through some rational processes. I tried so say something of this in 2001, where I used fictional intuitive “students” and rational “archaeologist” in an attempt to make the points.

Another Callan Point head

Archaeologists examine the physical details of the grooves in order to check the details of the picture, mentally flipping between picture and detail of rock surface. For example, that feature is NOT a nose, (or a pipe) it is a shallow wiggly valley some 100 mm long, 10 mm wide, 3-5 mm deep.

The printed picture is neither nose nor valley; but a configuration of light and shade, reflected from printed dots. Our seeing jumps automatically from print to nose, unaware of the configuration of light and shade, or the wiggly valley, (or, for that matter, the physical and chemical details of printing and photography).

No-one has difficulty (except perhaps boredom) in grouping the motifs into classes, even when they are not familiar or don’t make sense. The fictional intuitive student quickly discovers human profiles, globes, ships, fish, letters, and star-crescent combinations, but may fail to see the physical marks that make them. The archaeologist painstakingly identifies non-natural marks in the rock, and describes each in detail, then employs a computer (or does it some otherhow) to produce a multivariate classification and finds that the motifs group and subgroup, into levels arranged hierarchically into Types and Varieties. Students move happily between levels of classification, and have no problems with their definition and differentiation.

Intuitive students can recognise incomplete motifs, once they knows what the complete ones look like. They recognise mistakes and deletions by comparison with norms from the site, as they recognise that some motif types are combinations of constituent motifs.

Archaeologists do the same by invoking models: If this one is one of those, it’s incomplete. They can specify and identify the lack of completion.

Star-crescent 1855 is missing a line

Star-crescent j 191910 has some points in the wrong places

Star-crescent 38 G19 3 has ornate decoration in the form of dots in the crescent begun but not completed

The star of 1919 WELERE is elaborated symmetrically, and an (unsatisfactory ?) inscription has been deleted from the crescent

Once artefacts are typed and classified, both the artefacts themselves, and their associations and distribution patterns are open for study and interpretation. Students explain many varieties as resulting from faults or quirks of the artist, (lapses of memory; failure to concentrate; getting muddled; making a mistake; starting something which is never finished; deleting offensive work; not taking trouble to rule a line straight or measure consistent angles before carving; obsessive correctness); others are technical: (adaptation of positioning, size, or details to accidents of the rock; skilled or less skilled use of tools and techniques); and some are free choice: (selection of subject; decoration for its own sake; choice of subject).

Students’ effective and efficient intuitive methods are excellent at generating hypotheses, whose demonstration unfortunately requires the boring effort of rational archaeology. Nowadays computers can help with the work, and may produce results (Clegg 2000a and 2000b; Barry, Clegg 1978, Higgs 2002) which are novel, even to intuition. As the illustrations demonstrate, it is possible to recognise decoration/variation for its own sake; which aspects of the artefacts express style and which express meaning, and the relationships of pictures’ form to their tools and support, and several other topics. Insights like these are available to intuitive methods, though they may not be worth the effort of rational archaeology.

Classification and typology

The processes of description, classification and ordering proved fruitful in the many sciences, and necessary to their development. Consider the development of chemistry and the periodic table of the elements that Mendeleev began; Linnaeus’ classification of living things.

Once a classification system is developed with its attendant nomenclature, questions of the sort “What is this?” could be answered within the profession through the use of a key, and reported by nominating its species and variety; “Xenopus levi”, “Mousterian side scraper”, “Hawkesbury sandstone”, “Rembrandt self-portrait”, “Federation villa”, “Chlorine thirty-eight”, “Windies opener”. Such answers are applicable within a profession, and may be opaque as well as uninformative elsewhere, where “frog, stone tool, rock, picture, house, isotope, cricketer” might be more useful. The professional understands from the naming what sort of thing is under discussion, whereas other people may have other interests and needs.

Even formal archaeologists have become less interested in typology and classification since they gave up misunderstanding it as a source of information about dating and stone tools’ functions when carbon dating and various forms of use analysis and diagnosis gradually became available in the second half of the 20th century. This is a valuable clearance of misleading interpretation and eliminates a difficult and inscrutable subject from examinations, but with it may have gone a fundamental investigative tool, classification and nomenclature. As we will see, there are enormous theoretical, semantic and practical difficulties in diagnosing what (if anything) a picture is of, while a formal typology or classification is possible and almost as informative.

This discussion will mention formal ways of classifying rock art, then move on to discussion about how to diagnose whether or not pictures are depictions, if so how to recognise subjects of depictions.

Classification of rock art: Maynard’s scheme.

There are many classification systems of rock art; most begin with a distinction between pigment art/pictographs and engravings/petroglyphs. Whatever words are used, the fundamental difference is between the marks that consist of material added to the natural surface/support, and those made by removing material.

Beyond that, the systems go in different ways. My favourite system was devised my Lesley Maynard (1977).

Her scheme considers first technique, then to form, then motif. It is only at the level of motif that things are named in a way that refers them to real-world objects, so that, given a motif type classified as kangaroo for the good mnemonic reason that its form resembles some aspect of the animal, a cursory reader may believe that the intention is to declare that the motif represents a kangaroo, or is a depiction of such an animal.

Table 1: Technique, Maynard p. 391

Table 2, Form, Maynard p.394

Table 3, Motif, p 395

Prehistoric pictures and depictions.

Presumably for the apparently simple reason that many motif types are easily recognised as depictions of things, few theoretical or practical discussions realise the enormous and complex issues involved in asserting that prehistoric pictures are necessarily depictions, and that much of their meaning resides in their subjects. I find these issues so difficult, that I have sought other means of dealing with motifs.

Typological methods can answer many of the archaeological questions usually posed of rock art. Its theory and practice are well worked out in archaeology (cf stone tools; Clarke on Analytical archaeology) and other disciplines (Sokal and Sneath on Numerical Taxonomy ). It can be performed on artifacts considered as physical objects, without the addition of questionable interpretive procedures.

Case example 3: figurative, non-figurative.

Below I quote at length what an entertaining and informative debate that shows that, whereas “figurative” and “non-figurative” may be useful categories for analysis, the categories need not mean or assume what the labels seem to assert; the division can be, should be, and usually is, made on the shapes of the artefacts.

On Wednesday, January 29, 2003, Andrea Arcà sent a message to the Rock-Art list. There had been a long and acrimonious debate, which is not relevant to this paper, so I have edited out a few sentences.


From the description given it is clear that we are speaking of complexes and vast associations of cup-marks and channels, one of the most diffused European patterns (I say European because I’m not sure for other areas, but I think that it is also possible to say world-wide).

In my mind this kind of petroglyphs takes part of a non-figurative category, which should be distinguished from a figurative one. It’s not a matter of value (both of them can be an invaluable archaeological and cultural source of data), but a question of organising a different way of an interpretative path. It seems clear for me that this pattern of cup-marks and channels bears a function, the obvious function of containing something; it could be the “cup” for an offer on a rocky altar, it could be the sign for a boundary limit, like the Latin “termen”, it could be the tool for foreseeing the future like the Etruscan Haruspex (prognosticator, this use is well explained in some Roman inscriptions of the III century AD). On the contrary figurative complexes don’t bear a function but an icon, which in many cases is the only faded image we have of our ancient dead and fossil cultures. The large part of the most known rock-art UNESCO protected heritage sites pertain to this last category.

In this sense non-figurative and figurative complexes have been differently evaluated in all rock art historical studies (I’m referring to European areas). Often non-figurative complexes (“schematic”, “cup-marks and other” and so on) have been shifted into the category of “minor signs”. This is clearly incorrect, because both can give important (but different) data.


I’d prefer not to define rock art in terms of it’s function or use, or of it’s figurative or un-figurative status until we have ways to recognise – or even define and debate such things.


I define figurative this:
or this:

on the contrary un-figurative this:
or this:

I thinks that images are the first (or one of the) way to recognise and after to debate.
Obviously, as I wrote, it’s no more than my opinion.



Andrea Arcà has proposed that early in rock art studies, the pictures should be diagnosed as figurative or not. I disagree.

I’m grateful to Andrea Arcà for providing in an earlier post such a fine way of defining “figurative” and “non-figurative”, by providing photos of both sorts of piccies. If we consider his first illustration, which he uses to illustrate and define “figurative”,

Italia, Valcamonica, Redondo rock 20

Warriors findings in the Rock 20 of Redondo (photo A. Fossati-Footsteps of Man 1993)

We may observe three rectangular-bodied soldiers having a go at each other with spears and daggers, a shield and cloak. One is being struck by a bolt of lightning addressing his crotch. Below them is a pushmepullyou, an animal with a head at each end of its body, which is a character in some of the childrens’ books written by Hugh Lofting, starring Doctor Doolittle, who can talk to the animals. Several movies have been made of them.

To the right are several circular depressions, which look quite similar to the heads of the battlers. Such depressions are quite common in rock art, and variously diagnosed, classified, and named. If they are almost hemispherical depressions, like an egg-cup, they are known as “cupules”, and seldom treated as figurative or representations. Some are shallow peckings, like the ones shown.

Inferring from their associations, appearance, and form, together with Arcà’s definition of them as figurative, we may enquire what they represent, and infer they are either the bodiless heads of defeated warriors, or the footballs. The whole story is surely that the fans attending a football match disliked some aspect of it and engaged in a stoush.

Very similar depressions in Australia are sometimes known as emu-eggs (there is a picture at Mutawintji where a clutch of them is placed correctly and to scale proximate to a small figure of an emu). Alternately they may “be” groups of kangaroo-droppings, which are about the same shape and size, and occur in similar groups, so they could mark roo-pads, where roos have slept through the day. This hypothesis has testable scenarios at a site that has lots of roo-track like engravings, in lines as though in a trail, as well as groups of small circular depressions.

The whole hypothesis goes something like
“IF the things that look like trails of kangaroo tracks represent kangaroo trails, AND the things that look like kangaroo droppings represent kangaroo droppings, THEN there may be significant statistical associations between these engraving types, AND PERHAPS provide a clue to our reading of the site.”

I once tried such a statistical test on Sturts Meadows data, and failed to find any such significant association.

But at that site, there are whole areas (each perhaps 10 hectares, thousands of engravings) that have groups of such small circles (which we call “scat groups” for convenience), and others that do not. The present model is that the engravings in these areas were made at different times, scat groups were made at some times, not at others.

Sorry this has been long-winded. I hope it explains why I prefer not to begin an investigation of rock art with assumptions about figurative, iconic, or representative status.


Dear John,
I think your message is to be read carefully: the subject is very interesting and your words are deeply exposed. I need obviously more time to write a complete reply (my mother tongue doesn’t help me in finding all the words I need). I’ll do it ASAP.
For the moment I post few other proposals:



Only a short add: I intend as “Figurative” images (carved, engraved, picted) which do really represent something (object, tool, animal, human …) , “Non-Figurative” images which do not. This second case could embrace many different typologies, most of them having a function (like cup-marks, i.e. tools), not an iconography. Obviously not all is white or black: this is the case of symbolic figures (what is a boundary Christian cross? A figure or a tool or a sign or a mark?) . This rough division could help in understanding different kind and periods of figures (in the Alps there is a huge difference between figurative engravings, which are concentrated only in few places – like Valcamonica and Mt. Bego, were rocky blackboards are more suitable to be pecked-, and non-figurative ones, like crosses and cup-marks, which are scattered all along the slopes).

But I hope we all can agree on a point: to know, to see and to show is crucial. Only in this way we will be able to build our net of interpretations. Internet can help a lot in this.


Dear John, the figure I posted is a picture of the Rock 20 of Redondo, Valcamonica (Italia). You can find the entire record (with many others) in the EuroPreArt online site ( , please click on “Data”, then on “Search”. Write “Redondo” in the field name and you will find it. Obviously I didn’t want to post a quiz or whichever similar. You can find the entire (reduced) tracing at:

this is the work we do while recording and studying engraved rocks. It must be noticed that the particular is very clear (but reduced) : it is a representation of a duel, like other tens of thousands we have in Valcamonica. It belongs to the II period of the Iron Age style. i.e. around the VI century B.C. This chronological frame is obtained because we are in front of a figurative panel, by comparison of depicted weapons ad tools with the real ones found in archaeological strata and by the help of the superimposition (I don’t intend to open here the subject of chronology and style…). For this reason I (and all the scholars who are working in this “huge” area) speak of “figurative” engravings.

I would like to point to some particulars: the animal with a double head is indeed a horse with a tail (the tail seems a head because it has been pecked over three little natural cracks, it is difficult to detect them as the photo is reduced), and the depressions don’t represent anything in the mean I intend as figurative (obviously it could in the past, but we can’t detect it for the moment). We have plenty of them, they are common in all the pecked surfaces, they probably constitute only the result of a sharpening action, as the tool was a quartz pebble. They are not cupules (cup-marks), because a cup-mark is a tri-dimensional object on flat surfaces most of times (deeply) carved and smoothed. They are only adjoined pecks, obtained very quickly by repeatedly pecking the surface in the same point to reshape the edge. Obviously no one would never say that they represent footballs (I apologise, but it’s hard to me understanding when you’re joking or not, and I’m missing the word “stoush”…) , only because in this period football is not testified and because duelling warriors don’t play football with several balls at the time.

So we have figurative elements (the warriors, the animals, and others not included in the pictures, like scratched Iron Age knives ) and other “figures”, executed with the same technique (and at the same time), wich don’t represent anything. As they are only a technical result of the pecking activity (in some other cases they can be inserted in some topographical compositions or maps, but not here) they are traced and recorded, but included in the class of the so called “no meaningful figures”. This a standard of the alpine scientific recording method, which is applied in the same way in the Mt. Bego area.

So “Figurative” and “Non-figurative” define only very general categories (sometimes the limit is hard to define, but most times it is very clear), including in each of them a plenty of types and sub-types. As we are treating the engraved signs like archaeological finds (and in this sense I have really no difficulty “with the fact that pictures are artefacts”, on the contrary I think they are, but sometimes they are mental artefacts), I think that a clear division in classes and categories is the first method to be applied to organise the whole subject, like in all archaeological finds. Obviously classes and categories may be (highly) questionable, but there is no reason of eliminating them, surely to discuss.

Regarding the other side, the “Non-figurative” one, I think that this definition is very useful because, when a suitable engraved sign is found, it can be argued that it was a tool (a boundary limit, a sign of property, a cup to contain something exactly like in many cup-marks cases) and like it studied. We have plenty of cadastral signs in the Alps (three parallel lines) or cross-like figures that for too much time were interpreted like anthropomorphic, not being able to understand their real “status”.

Rock art is clearly related with the culture which produced it but also with the surrounding landscape: a clear distinction of what was depicted for iconographical purposes and of what was executed for utilising it as a tool (like cup-marks are) seems to me fundamental for a first understanding of the (Alpine) rock art.

Obviously I don’t want to generalise and to overpass this area, but the tens of thousands of figures I’ve traced with the team I work with can re-enforce this point of view. If on the contrary we want to say that nothing is really white or really black, of that we will be never be able to define our human shape or to agree on figures in the clouds, in this case I completely agree with you, but I look at this discussion as distant from the archaeological material I’m treating, near on the contrary to my (and your) human nature.

I must apologise for replying in such a partial and limited way to your numerous points of discussion. This is for moment what I have to offer, I hope soon to focalise more.

Obviously once more I outline my statements as being only points of view and of debate, I understand how differently the matter could be treated.

Best of all.


Thanks very much, Andrea, for your careful and considered response.

It was unforgivable of me to use obscure dialect such as “stoush” when referring to what happens art some football matches. I apologise. Partridge gives : “Fighting, violence, anything from fisticuffs to a great battle. Aus: C.20”

I hoped to turn the discussion towards rock art by questioning any habit of commencing a study of rock art by deciding its figurative status. His reply has shown me that whatever was actually written in his original post, in Andrea’s domain many processes and decisions precede decisions about the figurative status of a picture.

A couple of examples may suffice to suggest that what seems so clear to one school of rock-art researchers may be unacceptable to others.

The question of Natural marks. In both US and Australia, as well as Euro-American fine arts, natural or accidental phenomena may be an important part of the picture: see my suggestion of a bolt of lightning.

Dating by interpretation. Arcà points out that their pictures can be dated to the Iron Age because of depictions of iron age daggers. In Australia a Cape York depiction of an extinct animal called a Diprotodon was held to date a picture, until it became likely that the picture was actually of a pig, introduced to Cape York when Captain Cook landed there in the late 18th century.

Another favourite of mine is the clear depiction of a Jurassic dinosaur, which is more correctly known as Goodbullboon, the butcher-bird.

I am profoundly interested in the assertion that the small circular peckings were made to sharpen the pecker, and are thus not figurative. If this is so, there should be evidence in the tool-marks, which should show a tool-point getting sharper. There are indeed some tools that get sharper through use; a scissors is one, and an impact-drill working on certain sorts of sandstone another. But if use sharpens a pecker, there would be no need ever to sharpen one… .. …

Sorry if you cannot tell when I am joking. As with many postings to this list, many comments are both comic and true. The motivation of the author – joking or not – is irrelevant.

Thanks for your time and attention, those who read this far!

John Clegg

In his response Arcà invited Clegg to write what has become the present paper.

The geometry of knobs and blobs

I have shown (Clegg, J. (1978). figs 7.10, 7.11, 7.12, 7.13, 8.1, 8.2) that it is possible to refer to the shapes of motifs without using anatomical references, (heads, tails, feet and so on).

My suggestion is the “knob and-blob” approach. It is dependent on old-fashioned stone tool analysis for its distinction between round and long knobs, as in flakes and blades. The geometrical construction involves first reducing the overall shape of an outlined area to a simple oval “blob” by conceptually making the shape a handy size of a brittle substance, and putting it in a clothes drier until the fragile knobs have all broken off. The ends of the blob are defined by its longest dimension, as is its length and breadth, ends and sides, and hence the locations from which the various knobs have broken. A profile animal thus has knobs on the ends and one side of the blob; a full-face standing human has knobs on both sides. This simple model can be extended to define places from which measurements can be made consistently. So it is possible to describe, count, and measure, the shapes of rock art.

This system is such a nuisance to use that it is better to go the easy way of using familiar referent words. I remain very concerned to distinguish between using such names as mnemonics for the shapes of forms and identifying the subjects of depictions, but the various devices I have tried (such as using “kangaroo” or !kangaroo or even kangarooomorph) are all so conceptually difficult that they have occasionally been ridiculed. (See the several refs to Clegg in the index of Morwood 2002)

Case example 4: small daras?

Here is an example. The problem concerned large (3 metre plus) profile (having protuberances on one side, not both) figures that are characterized by emu-bums and human feet (which can be described without reference to emus, bums, humans or feet). There were two questions about possible similar engravings: do smaller versions of the same type exist? Are there female versions? (with breast; below arm; both again definable as a round knob on the knobbed side between 2 long knobs). See Higgs 2002, Clegg 2002.

The Higgs thesis demonstrated that, so far as the available data allows, small versions and “female” versions exist; the small versions occur over the whole area, the “female” ones only in the Hunter Valley.

Case example 5: Diagnosis of representations: thylacines and zoologists

This discussion has shown that it is possible to archaeologise pictures without knowing what they represent, or even that they are representations. The relevant techniques are of classification and typology, which require the ability to compare pictures. It is of course possible to compare pictures that carry labels with other pictures that don’t.

In 1972 and 1973 several papers were published reporting rock art pictures that were held to look like thylacines. There was a built-in and unstated assumption that they are depictions. This assumption satisfactorily and transparently begs many questions. It works with minor modifications also as a definition of depiction.

A picture is a depiction insofar as it resembles an object or concept. The general assumption is that a picture is of that target object which it resembles most closely. (Clegg, 1978:104). A very slight modification reveals a way to diagnose the subjects of depictions: compare the picture in question with a choice of other pictures whose subjects are known; in this case zoologists’ drawings of possible mammals. The zoologists’ captions provide the identification.

The following case example compared zoologists’ pictures of Australian striped mammals to rock art, testing the hypothesis of the diagnosis of thylacines.

In 1972 and 1973 questions of the age of rock art were of great interest in Australia. Possible depictions of extinct animals fostered this interest. Of particular importance was thylacinus, a marsupial carnivore that had gone extinct on the mainland some thousands of years ago, although it survived in Tasmania till 1935. This animal was dog-like, but striped.

Brandl and Wright published pictures they thought were of thylacines, and therefore made before the extinction. I felt that the weak point in the discussion was an essentially subjective (if shared) approach to the question of resemblance, and sought a transparent and repeatable way to measure resemblance between pictures.

My solution was:

    1. reduce the field, considering only Australian striped mammals. (This was supplemented by the inclusion of feral cats.).
    2. collect several (5) labeled zoologists’ drawings of every candidate species
    3. define comparable points that could be recognized on the drawings
    4. devise many independent measurements that could be taken from those points. (head size and tail length may both depend on the overall size of the drawing, whereas a ratios of head length / body length, tail length /body length, although dependent on body length, are independent of overall size)
    5. determine from the zoologists’ drawings which measurements are diagnostic of which species
    6. measure the species-diagnostic elements on the candidate pictures.

This process ended up awarding the candidate motifs with marks, (one scored 13/17 cat, 15/22 thylacine).

This process has the enormous advantages of being transparent, so it can be repeated independently by other workers, who can choose their own zoologists’ drawings and measuring points.

Wood Jones’ numbat and thylacine


This example has significantly advanced the discussion, by providing a definition of depiction and a means of choosing between subjects. The definition of depiction may be unsatisfactory to some people, because it begs the questions of artists’ intent.

The above discussion can be applied to all sorts of recognition, and still need not assume any link between artists’ intent and viewers’ diagnosis. A search for such a link could revert to the consideration of criteria for defining artefacts: and invoke an hypothesis that if there is no such link, and the resemblances are chance, there should be many pictures with other degrees of resemblance.

The Polonius/Hamlet problem can happen between archaeologists. My solution is to compare target pictures (those whose diagnosis interests us) with pictures whose diagnosis is known, for example those in zoological texts. A few imponderables may be reduced by comparing target drawings with diagnosed drawings, through measuring as many comparable independent attributes as possible; i.e. allowing for differences in manner of drawing. (See Clegg 1978: 103-115) The general assumption is that a picture is of that target object which it resembles most closely. (Clegg, 1978:104). This assumption satisfactorily and transparently begs many questions. It works with minor modifications also as a definition of depiction.

A picture is a depiction insofar as it resembles an object or concept.

Epilogue Perhaps some people, even archaeologists, believe it may be possible to understand the past, and its rock art, and that archaeology can foster that unerstanding.
… for lines and colours arranged in various ways, even though apparently pleasing, are not enough. They must express meaning… … This meaning, however, cannot be grasped apart from the culture to which the art is an aspect, and the artist a medium. … … Actually, the only objects of native art which we can really appreciate, as distinct from admiring, are those we know or have known from the pulsating context of native life.
Few who study rock art can be so privileged. Even when old rock art has a place in contemporary culture, contemporary understanding may not coincide with the understanding of its makers or first users.

Given all the difficulties I’ve mentioned in this paper, many techniques may be needed to help appreciate rock art. Perhaps the most important are not often considered by archaeologists. Managers may work hard to make the surroundings of a site approximate those it had at a time in the past thought relevant to the rock art; and various literatures may inform about what is known of the cultures and practises of those who made and used the art.

But professional skills and staging may be needed to put potential appreciators into an appropriate state of consciousness for them to appreciate it. Simon McBurney is expert at facilitating almost shamanistic alterations to consciousness, as his installation “The Vertical line” and other works including “The three lives of Lucy Cabrol” and “Mnemonics”, showed. McBurney’s father, C.B.M.McBurney, was an archaeologist who took his children on some of his expeditions.
Over four nights in February 1999, the writer and art historian John Berger, Theatre de Complicite’s director Simon McBurney and the actress Sandra Voe conducted an intimate 30,000 year old journey, inscribing a downward line through time 30 metres below central London.

Part theatrical event, part archaeological dig, the Vertical Line was an oratorio of faces, voices, darkness, and light: a one-off excavation for small groups down 122 spiral steps into the bowels of the disused Strand tube station, where a sequence of audio-visual installations culminated in a live performance on seven occasions.
(Michael Morris 1999, “Introduction” in The Vertical Line, John Berger and Simon McBurney CD Artangel, Theatre de Complicity and Somethin’ Else, London)

The Strand station had two platforms. “One closed in 1907, the other in 1994: on the same date, we later discovered, as three French speleologists led by Jean-Marie Chauvet first opened the cave in the Ardeche gorge”. (Morris) “First, opened” that is, for 30,000 years or so.

† John Clegg


  • Anon 1951 Manual of seamanship. London, Her Majesty’s Stationery Office
  • Campbell, N. A. (1990). Biology . Redwood City, Benjamin/Cummings
  • Clarke, David L. 1968 Analytical Archaeology . London, Methuen.
  • Clegg, J. (1976) Aboriginal Art Calendar, 1977 . Balmain, Clegg Calendars
  • Clegg, J. (1977) Aboriginal Art Calendar, 1978 Balmain, Clegg Calendars
  • Clegg, J. (1978). Mathesis Pictures, Mathesis Words . Balmain, Clegg Calendars.
  • Clegg, J. (1998). Making sense of obscure pictures from our own history: exotic images from Callan park, Australia. The Archaeology or Rock-Art . C. Chippindale and PSC Taçon. Cambridge, Cambridge University Press: 336-345.
  • Clegg, J. (2000). Theory and common practise in rock art. International Prehistoric Art Conference Proceedings . J. A. Sher. Kemerovo, Siberian Association of Prehistoric art Researchers. 2: 5-16.
  • Clegg, J. (2001). Rock art studies: theory. Theoretical Perspectives in Rock Art Research . K. Helskog. Oslo, The Institute for Comparitive Research in Human Culture: 40-67.)
  • Clegg, J. (2002). “Variability and Information in Rock Art.” Tempus 7(7): 101-8
  • Cronin, Leonard 2001 Australian Reptiles and Amphibians Annandale, Envirobook : 101
  • Elkin, A. P., C. Berndt, et al. (1950). Art in Arnhem Land . Melbourne, F.W.Cheshire
  • Feagin, S. L. and P. Maynard (1997). Introduction. Aesthetics . S. L. Feagin and P. Maynard. New York, Oxford.
  • Gombrich, E. H. (1960). Art & Illusion: A study of the psychology of pictorial reresentation . Oxford, Phaidon.
  • Leroi-Gourhan, A. (1957). The Beginnings of Art. Larousse Encyclopedia of Prehistoric and Ancient Art . R. Huyghe. London, Paul Hamlyn: 26-29.
  • Maynard, L. (1977). Classification and terminology in Australian rock art. Form in indigenous art . P. J. Ucko. Canberra, Australian Institute of Aboriginal Studies: 387-402
  • McCarthy, F.D. 1983. Catalogue of the rock engravings in the Sydney Hawkesbury district of NSW National Parks and Wildlife Service, Sydney 3 Volumes.
  • Morwood, MJ 2002 Visions from the Past . Allen and Unwin, Crows Nest
  • Michael Morris 1999, “Introduction” in John Berger and Simon McBurney The Vertical Line CD Artangel, Theatre de Complicity and Somethin’ Else, London
  • Nisbett, R. E. (2003). The Geography of Thought: How Asians and Westerners Think Differently … … and Why . London, Nicholas Brealey Publishing.
  • Sokal, RR, and Sneath, PHA 1963 Principles of numerical Taxonomy. San Franscisco: WH. Freeman and Company
  • Wood Jones, F. 1924 The Mammals of South Australia. Part II, Adelaide, Government printer:

(3502 views in previous TRACCE PHP-Nuke version, 2002-2011)

Leave a Reply

Your email address will not be published. Required fields are marked *