Rock Art Analysis and Issues



To start with a personal note, I should to point out that rock art analysis is a vast and ever-changing subject, and that this page is simply incapable of doing full justice to it.  What I have tried to do here is give an overview of some of the approaches and problems faced by rock art experts.  I am not suggesting that this is anything more than a superficial insight into some of the many approaches, issues and challenges that are being confronted by those who are truly expert in this field.

Philip Smith, writing in 1968, believed that northern African rock art offered one of the best opportunities of any world region for interpreting rock art and using it to supplement the archaeological and historical data already available (p.2).  He believed that due to lack of natural defacement, most northern African rock art would have survived and it would therefore “be accepted as reasonably representative of all periods in the past when rock art was being done”.  He was attracted by the variety of themes depicted, and viewed them as documents which could lend themselves to decipherment and interpretation.  He described the number of areas about which they could supply information, including climate, social relationships, physical types, sexual division of labour, religious behaviour, subsistence practices and social stratification (p.3-4). 

However, since Smith was writing in the late 1960s the difficulties in using these painted and engraved images to supplement archaeological and historical have become more prominent, the key issues being interpretation, dating and distribution.  A short but incisive statement of some of the difficulties faced by researchers comes from John Pfeiffer:  “It should be clear that investigators are long on guesswork and short on evidence.  All they have to go on is a mass of unanalyzed observation and imagination” (Pfeiffer 1970, p.237).  Matters have  improved considerably since this time, with new approaches and methodologies, but his words can certainly serve as a useful warning that speculation needs to be back up with solid evidence.   Balancing this are  pragmatic approaches taken by modern petroglyph analysts.

As Chippindale and Nash say in their introduction (1999) “Each class of archeological material has its own character, and with each character come the special strengths and weaknesses of that personality”.  Rock art certainly has its own personality. On this page I try to look at some aspects of this personality.  The examples of analysis and difficulties with rock art research described in this section are generic problems with rock art world-wide, and it would be artificial to discuss them with reference only to Eastern Desert examples, although issues relevant particularly to the Eastern Desert are highlighted.

Fashions and trends in archaeological research

Different approaches to archaeological material have a very different impact on opinions about how certain datasets can or should be used and what they can reveal.  Briefly, and simplifying a complex subject, two conspicuously different philosophies have dominated modern archaeological thinking about the possibilities offered by archaeological analysis of symbols and motifs. 

One, typical of the so-called New Archaeology approach or Processualist, considers that thoughts and beliefs, ideas and values, cannot be retrieved from archaeological data, because the data links back to what people do, not what they think.  The archaeological record, in this view, is the outcome of processes of economic and industrial activity, not ideological conceptualization. 

Another and more recent approach has been to attempt to get to the meaning behind these actions.  Perhaps the best known of these Post-Processualist writers who have encouraged development in this field are Ian Hodder, Colin Renfrew and Stephen Mithen, but there are a great many more. 

For more information about the different approaches to archaeological data see Trigger (1989) or Johnson (1999).  Both are excellent summaries, weighing both the benefits and disadvantages of past and present approaches.

Asking the right questions

A great many rock art surveys have taken place, globally, sometimes without specific research questions or detailed consideration of what rock art research is really able to contribute.

First, we need to look at how rock art is being approached at the moment:

  • What information can rock art provide by itself?
  • What can rock art contribute to research in other fields, particularly archaeology?

There are different answers to these  basic questions, although they are not approached in all of the literature.  The different answers inevitably lead to conflicting ideas on how rock art can or should be assessed and incorporated into research projects.  Some of the different approaches are described next. They will be investigated further in the Why Study Rock Art section.

Methods of Analysis

Methodological approaches have been much-discussed in archaeology over the last few decades, and need to be developed further for rock art research.  In spite of difficulties, considered studies by more recent writers have offered more interesting and useful approaches. 

The following paragraphs under this heading are largely taken from the University of New England’s useful World Rock Art web pages at:

The UNE’s rock art pages provide an overview of rock art research as follows:

    “Specific approaches to the analysis of prehistoric art are both numerous and varied. Similarly, there are many different ways of classifying rock art analyses. However, at the most general level rock art analyses can be divided into two broad categories – ‘Descriptive’ and ‘Comparative’ . . . . The analyses in each general category also share underlying assumptions and problems, and these can be illustrated by a relatively limited number of case studies, rather than an exhaustive (and exhausting) review of the world literature.”

Descriptive analysis looks at the subject matter of figurative rock art and attempts to use the information to reconstruct aspects of the society and environmental contexts that produced it.  Whilst this may reveal elements of life not available from other archaeological data, it does depend on tying in the rock art to a particular cultural period.  It can also be difficult to understand the relationship between artistic traditions and conventions, and life as it was actually lived. Rock art is also selective – it does not represent life in its entirety, but emphasizes particular aspects of it.  In Eastern Desert rock art, for example, certain animals are represented (e.g. bovids) whereas others are ignored (e.g. sheep and goat) and plants are not represented at all.  However, as Smith points out (1968, p.21) “Regardless of the interpretations of certain scenes, there is a hard core of precise information on such matters as clothing, ornaments, certain implements and tasks, and even on group interactions, which are unlikely to be preserved in any other way”.

Comparative analysis looks at a broader range of data, tying in both figurative and non-figurative rock art with artefactual, zoological and environmental data, amongst others. Aspects of it are also come under the borad heading of “cognitive archaeology” (Renfrew and Bahn 200, p.385).  The method is used to explain content or structure of archaeological assemblages, with a view to identifying broad behavioural patterns.  A number of examples of comparative studies from different parts of the world are provided on the UNE Analysis of Rock Art web page.  Other examples are described later on this page.

Rock art can be highly contextualized.  In Upper Palaeolithic France highly skilled and beautiful paintings are often associated with extensive cave systems where the interiors are in complete darkness (Leroi-Gourhan 1968).  Lewis-Williams (2002) sees these cave systems as a highly structured way of using space for shamanistic activities.  In northern Britain, Bradley (1994) has demonstrated that patterns in motifs change in and between upland and lowland zones.  In the Eastern Desert the context is less obviously specific and observations of distribution are not conclusive or particularly helpful, but the images do form part of a landscape which may be made up both of landmarks and perceived values and connections.

Rock art is often studied in regional isolation.  It would be more useful to compare regions with a view to understanding relationships between them (e.g. the Eastern Desert with sites in areas like the Negev and the Western Desert).  As Andrew B. Smith (2005) has demonstrated, both ethnographic and archaeological data indicate that mobile groups can cover huge territories, and that rock art should be considered over a very broad geographical range.  Unfortunately, it is very rare that this happens, mainly because there is not sufficient survey data available for comparative analysis.  At the moment Egypt’s rock art is divided into regions which are not only rarely compared with each other, but are not considered in the overall view offered by northeastern Saharan archaeology and rock art.  With reference to the Tassili paintings of Algeria, Smith says that the paintings “need to be analysed as a group, and not just as individual panels.  As pastoralists use paths to connect places, and these enter stories and narratives, it is possible that the paintings constitute a larger whole, and can be linked together in the wider ritual landscape” (p.49).

Ethnographic analogies have also been used, with varying degrees of success, to develop ideas about  the meaning of rock art.  Early 20th Century attempts to interpret rock art sometimes lifted concepts of totemism from published ethnographic analyses (UNE).  This resulted in the blanket interpretation of much rock art as deriving from totemic magic. Smith (1968, p.21) warns about work that has used ethnographic data:  “Ethnographic data is notorious for revealing at times as much about its observers as its makers”.

Ethnographic data can also be used to assist in considering the broader role of the rock art within the social landscape:

    “This approach to the ethnography of rock art is based on the notion that rock art data should be treated as part of broader cultural systems of meaning and, as such, should be analysed in a similar manner to other archaeological data — that is, in terms of spatial and temporal patterning and relationship to other archaeological material. The argument here is that the basic archaeological problem of establishing links between material evidence and the original cultural context remains the same regardless of whether the evidence being examined consists of stone artefacts, food remains, sediments, or rock art. Middle Range Theory is often used as a device for establishing links between rock art and cultural context and for producing hypotheses which can be tested in the field” (UNE).

One of the most promising and influential approaches is described by Alison Wylie.  Her approach brings together a number of different strands of data and binds them together in order to build a stronger model of a given situation, keeping research relevant and focused by preventing hypotheses from going off at tangents (Wylie 1989). If you don’t have access to academic journals, Lewis-Williams provides a good overview of numerous different approaches to rock art data (2002, p.102-103).

Barfield and Chippindale (1997) propose three simultaneous approaches for the analysis of rock art: consideration of the context, focus on common motifs rather than unusual ones, and application of only secure cross-cultural analogy.

Renfrew and Bahn offer a list of how rock art is actually used by the people who create it, and this would provide a useful measure.  These give some idea of the range of expression and intention that symbols can incorporate and some of the difficulties in developing an archaeology of symbols and motifs from purely archaeological data (p.391):

  • Establishment of place
    • Using markers to outline the range of a community establishing a link between community and their own landscape. 
  • Measurement
    • To define units of time, length, weight etc.
  • Instruments for conceptualizing plans and modeling the world
    • Modeling building plans, maps etc.
  • Organizing and regulating relations between people and groups
    • Money
    • Hierarchical rankings
  • Organization of human relations with the supernatural world

They go on to explain that the New Archaeology approaches, which followed some of the early attempts to find meaning in rock art, largely rejected rock art because of the difficulties in interpreting it.


Interpretative Comparative Approaches


Second only to problems with assigning dates to rock art, interpretation of images and scenes is the real can of worms of rock art research. 

The term interpretation refers to the meaning of rock art, its purpose and what it tells us about the people who created it.  It comes into what the UNE rock pages refer to as a comparative approach, and places it in the realm of what is sometimes referred to as cognitive archaeology (e.g. Renfrew and Bahn 2000, p.385).  It does not refer to other data that might be acquired from the depictions, like details of clothing or weaponry.  Describing themes is one thing, interpreting them is quite another.  And as anyone who has read the literature on rock art in other parts of the world will know, there is a vast leap between offering intelligent speculation and achieving academic agreement on any interpretation, whether ideas are of a pedestrian or innovative nature.  It is very difficult to actually test hypotheses about the meaning of petroglyphs.

It has to be accepted that ideologies, traditions and belief systems are probably incorporated into the reasons and motifs for producing the rock art.  However, it is all too easy for modern minds to have culturally determined ideas on ancient art that was produced by very different cultures and lifestyles.  Ideologies and values of the modern onlooker may be very alien to those of the artists, and we have no way of knowing what types of image are purely representational or symbolic – and if symbolic, we have no access to the thought behind the symbols.  Smith explains this well:  “Perhaps the greatest difficulty in getting at the meaning of prehistoric art is that we do not know the symbolic conceptions which were involved even in naturalistic representations.  Are these to be taken literally, that is as signs?  Or are they loaded symbols, part of a code to be broken?” (1968, p.30).

The fact that no-one knows for sure why the images were executed has not prevented a lot of both scholarly and amateur speculation on the subject.  A lot of the early work on rock art made assumptions about its role. Early interpretations included sympathetic magic, totemism (based on the close relationship between certain animals and human groups), religion (the formalized organization of belief), and ideology (thoughts and concepts that underly human social organization and behaviour), all of which are perfectly valid approaches to take when questioning the function of rock art (discussed in more detail by Huyge 2002, p.192) and art in its truest sense.  However, some of the least well thought through of the ideas were assumptions and speculations rather than informed opinions based on observation of the rock art itself or measurement of ideas against good ethnographic data.  Even the excellent Cyril Aldred, for example (1998) referred to Egyptian rock art as “magic-working rock drawings” (p.66).  As Renfrew and Bahn point out, these theories were often speculative and some were undisciplined, and they were in danger of producing, ‘imagining’ what ancient people must have thought or believed” (2000 p.385).

Searight (2004, p.17) says that “A basic assumption in relation to rock art is that it was done with a purpose.  If it was merely the doodling of idle shepherds the usefulness of the study would be reduced to an appreciation of the artistic talents – or their absence – among prehistoric populations” (2004, p.17).  In many cases this “basic assumption” seems fair.  In cases where rock art consists of simple sketches with no visible underlying structure, this assumption becomes less obviously acceptable.  It must not be ignored that “art” could well be an elaborate term of a doodle or graffiti, done to pass the time or to stamp an individual’s mark on a place. Without detailed analysis, and on the basis of observation alone, much of the “art” looks less like hunting magic or symbolic content, and more like random etchings, much like graffiti.  There are plenty of other examples of artistic graffiti in the Eastern Desert (e.g. etchings of four wheel drives), and inscriptions in ancient Egyptian, Latin, Greek, Aramaic, and modern scripts.  Site 24, discussed on this site by Tony Judd, is a perfect example.  This is not easy to believe of sites like the Upper Palaeolithic sites at Altamira or Lascaux, which are vibrant and detailed, and were composed in almost complete darkness, or of complex Aboriginal compositions in Australia, but it is entirely possible for rock art depictions in the Eastern Desert, made using simple techniques and stylistically indistinguishable not only from each other but also from images sketched in other regions of the world. Searight, quoted above, questions the value of “doodling” to archaeological research (2004, p,17), but sketches and doodles might still have a perfectly valid role – to help to chart areas used and useful to those who occupied the desert at different times.  Rock art, even doodles, shows that people were occupying certain areas, and the more rock art that is present the more important these areas might be.  The presence of any type of rock art is a good starting point for developing hypotheses about land use.  Further, the subject matter of any doodles tells us something about the interests of the artists.

Many studies have attempted to examine rock art as a symbolic components of a wider set of religious expressions.  However, Searight points to problems of analyzing apparently representational motifs  – what she refers to as the “difficult recognition that a concrete image (a dagger) might well stand for an abstract idea (power) or perhaps both at the same time” (p.139).  She goes on to offer the following useful warning:  “As even the most simple images lend themselves to a symbolic interpretation, one runs the risk of seeing symbols everywhere” (2004, p.140).  She admits that many of her own proposals are speculative. 

Craig Alexander provides a nice summary of how interpretations can come into conflict when meaning is at the heart of a discussion, in a summary of his own approaches to Italian rock art:  “Anati (1994) sees the petroglyphs as quite straightforwardly representational of belief systems, a view strongly contested by, for example, Bradley who refers to ‘grandiose interpretations of the imagery based on the literature of comparative religion (1997:8) and Chippindale and Nash, who write that Anati’s approach has ‘too much faith in unproven social universals’ and ‘too little respect for the particular course of social change in any place or region’ (2004:16)” (Alexander 2006 WR).

Mithen suggests that a visual symbol has 5 critical properties that define it (1996, p.157):

  • May have no relationship to its referent
  • Created with intention of communication
  • There may be a significant temporal gap between an event/subject and the time at which it was depicted
  • A symbol may have different symbols for different people
  • A symbol with a given meaning may be executed differently by different people

Mithen (1996) observes that symbols may mean different things to people depending on their access to the knowledge that enables them to read or decipher a given symbol or set of symbols, and that some symbols may only be identifiable precisely by context amongst other symbols. Symbols and motifs, in Mithen’s framework, form a grammar in their own right.  Mithen emphasizes that archaeologists have no way of reconstructing the exact message represented by some rock art without a suitable record of continuity into more recent historic times.  Lewis-Williams gives the example of the use in North American art of bighorn sheep, which were considered to be spirit helpers for rain shamans (2002, p.169).  this type of incorporation of natural motifs into the supernatural experiences makes the presence of a religious spiritual meaning much more difficult to identify.  Searight raised the problem in her extensive analysis of Moroccan rock art when she said that “Unfortunately it is impossible to say whether Moroccan rock art image were coded for a few initiates, readily available to all members of a group, or easily understood by all passers by”.  

Ethnographic data has been used quite extensively for interpreting rock art, and has highlighted some serious difficulties.  Morphy (1991) describes the complexity that rock art can incorporate, including how it encodes information about social organization, political relations and rights to resources, but it cannot be used unambiguously to interpret ambiguous prehistoric or historic data. Many of these ethnographic studies highlight the difficulties of offering any coherent form of interpretation of rock art symbols and motifs.  Fuchs (1991) highlighted this problem with specific reference to the rock art in the Wadi Barramiya, but his is just one voice in a cacophony of research, speculation and frustration. 

Another difficulty with interpretation of rock art in a broad archaeological context lies in the bias shown in the  topics represented – for example, in the Eastern Desert animals, both wild and domesticated, are represented  but plants are not.  What we have is a very partial image of life as represented by the rock art. This means that we may have aspects of life that were considered to be particularly important, has a specific social or symbolic status, or that we may have aspects of life which are simply relevant to the immediate occasion – it is really not possible to know. 

Attempting to determine economic information can be hazardous too, because there is no guarantee that the rock art is a faithful representation of economic activities, or that it is representational of anything material.  As stated above there are very obvious omission, and even where certain domesticated animals are represented,  others are missing, and no plants are shown.

Some writers have attempted to interpret rock art in terms of the outcome of a specific set of activities. A notable example, whether or not one agrees with his approach and conclusions, is Lewis-Willams (2002) who has compared San and North American rock art with French Upper Palaeolithic cave art and has concluded that some rock art may be closely associated with shamanic activities in certain cases.  His conclusions have been met with criticism, but have also evoked considerable interest and discussion.

The idea of rock art as art for its own sake needs to be considered.  Lewis-Williams (2002, p.42) describes art for the sake of art as “an innate desire to produce beautiful things that can be indulged only in leisure time”. Mithen (1996) points out that as with words like mind, language and intelligence, the term art defies an easy definition, because it is “culturally specific” (p.154).  He goes on:  “Membership of the elite group of artifacts that we call ‘art’ must go to those which are either representational or provide evidence for being part of a symbolic code, such as by the repetition of the same motifs” (p.155).  If this definition is accepted, Eastern Desert rock art could fall into both categories of representation and being part of a symbolic code, but it is remarkably difficult to tell. Recurrent themes could simply record recurrent interests, without being either artistic or symbolic.

Caroline Malone warns against intellectual arrogance with respect to the interpretation of rock art:  “Rock art studies in many continents have become very fashionable of late, and it is tempting to over-interpret it and assume that in the twenty-first century we can ‘read’ the meaning of patterns and symbols” (Malone 2001, p.253).


The difficulties described above should not be underestimated, but recent ideas and approaches have emerged in recent years which accept these difficulties, but attempt to find ways around them.

There are a number of approaches possible when considering an interpretation of rock art.  Here are a few examples:

  • Looking at rock art images as a collection of symbols and trying to get to the meaning behind that symbol
  • Looking at patterns of structure within the symbols without trying to understand the thoughts behind them, but getting some sense of dichotomy or linearity in the ideas expressed
  • Looking at how rock art functions in terms of perception of landscape
  • Looking at the symbols embedded into rock art to attempt to derive meaning (rarely possible)

An early attempt to make sense of Franco-Cantabrian rock art was made by Laming-Emperaire (1962)  Another early attempt look at the underlying structure of rock art was made by Leroi-Gourhan (1968b). His approach was designed to address two issues – a) the suggestion by Abbe Breuil that the images were made on an individual basis over a long period of time and were not a single coherent composition and b) that rock art could reveal some of the meaning behind its creation.  He believed that he could find structure, and beyond that, a scheme, in the organization of rock art:  “If, rather than working haphazardly, the men of the Palaeolithic consciously – or even unconsciously – introduced order into the way their pictures are positioned, then an analysis of where various animal paintings are located in a sizable number of caves (say 50 or more out of the 100-odd sites) should reveal what general scheme, if any, the artists had in mind” (1968b).  He analyzed over 2000 images in over 60 caves, and was able to demonstrate that there was a pattern, relating to the overall organization of the cave system themselves, which could be identified and described in detail.  He went on to propose a binary system of image organization, based on gender division. He differs from Laming-Emperaire in the nature of his gender-based scheme, which sees two pairs of binaries into a type of visual communication based on male and female elements which he believed were associated with specific parts of the caves, which itself formed part of a grammar describing ideas about the natural and/or supernatural world. The advantage of approaches like this is that they analyze data rather than making assumptions.  they attempt to address the data in a systematic way in order to tease information from it. However, the difficulty with many of these schemes is that they are impossible to evaluate empirically. (1968b, p.57).  Other writers in more recent years have also considered rock art in terms of gender differentiation (e.g. Robb 1994; Bevan 2000), and the art of the Franco-Cantabrian painters continues to be discussed extensively (e.g. Lewis-Williams 2002).

A number of other writers have tried to investigate beyond the immediate images to determine an underlying language of symbolism (e.g. Leroi-Gourhan 1965; Gimbutas 1989; Tilley 1991; Lewis-Williams 2002; Lewis-Williams and Pearce 2005).  Some of these are discussed below.

Matters are complicated by the way in which images are organized on a rock surface:  as isolated images, combinations of naturalistic images in a sequence that may or may not be narrative or employ some form of underlying grammar, and those which are not intended to be naturalistic, but are quite clearly representative of some idea or concept.

Mithen (1996) says that there are two main symbolic relationships between humans and animals – anthropomorphism (assignment of human attributes to animals) and totemsim (embedding humans into the animal world).  Mithen interprets the rise of art and symbolism with the rise of religion, religions being sets of recurring ideas that may incorporate a belief in non-physical beings, a survival of a non-physical aspect of an individual after death, the belief that certain individuals can be in touch with the dead spirit, or other supernatural agencies, and that acts conforming to specific formulae can influence the natural world (p.175-175).  Lewis-Williams’s The Mind in the Cave, which deals with methodological issues and tackles concepts alien to Western thoughts like altered states of consciousness, shamanism and the role of the supernatural in economic life.  Whether or not you agree with his findings, discussions of modern South African San rock art do suggest that many some of the less straight forward images, even sketches, that can appear on rock art in many different parts of the plant may represent either supernatural beings that mediate between the natural and the supernatural, or patterns generated by the brain in the form of hallucinations (which could be experienced naturally or psychotropically).  Some of these hallucinations may involve audio as well as visual components and it is possible that sometimes sound may be represented graphically.  It is clear, however, from modern ethnographic parallels that however abstract or hallucinatory the origin of an image might be, it is still incorporated into a firm socioeconomic framework.  Religions usually depict an overall picture of the universe, and are not independent of it.

Jean McMann takes the idea of symbols as language and discusses it in relations to European prehistoric rock art.  She says that linguists believe that many Mediterranean abstract art signs have no phonetic associations, which she supports:  “if only for the obvious reason that their arrangement is so limited.  it is difficult to conceive of a language with as few different signs occurring in a given geographical area” (1980, p.146).  She believes that a stage beyond this is the refinement of fixed signs that become a conventional way of communicating precise and unequivocal concepts, and that these are a step on the way to the evolution of writing “ refined, specific, evolved ideograms which could symbolize both concrete objects (such as the axe or the sun) and bas tract ideas such as the spiral may have represented” (1980, p.146).  She believes that symbols refer to intuitive knowledge which does not limit conceputal thinking or expression, whereas worlds refer to concepts which are limited by the specific task of static formulation.

Taking McMann’s idea one step forward, it is interesting that even fully evolved ancient Egyptian hieroglyphic writing has determinative symbols which are used to represent the concept of an abstract idea, differentiating abstract ideas from concrete items.  This sense of concrete versus abstract does appear to be a recurrent theme.

McMann’s idea that symbols may be a pre-literate form of conceptual grammar is encapsulated in work elsewhere in Europe. Herity, for example, presents a scheme for Irish rock art associated with Boyne tombs in Ireland which attempts to distinguish between simple elements, motifs and symbols, and combinations of elements.  Lewis-Williams (2002) also looks at the potential of a grammar underlying symbols:  “Cultural inventories select items from a full repertoire of potential mental images and this selection process takes place within a social community:  the spectrum of consciousness is simply on of the raw materials that society uses to construct itself through the intervention of individuals” (Lewis-Williams 2002, p.172).

Most of the approaches that examine rock art agree that symbols need contexts.  They cannot be seen in isolation. Rock art should only ever be seen in the context of other data wherever this is possible in order to contextualize it properly. This is phrased eloquently by Renfrew and Bahn:  “It is usually impossible to infer the meaning of a symbol within a given culture from the symbolic form of the image or object alone.  At the very least we have to see how that form us used, and see it in the context of other symbols.  Cognitive archaeology has therefore to be very careful about specific contexts of discovery:  it is the assemblage, the ensemble that matters, not the individual object in isolation” (p.386).  In rock art terms there are two ways in which these requirements can be fulfilled. 

  • Rock art should motifs and symbols should be seen in the wider context of other examples
  • Rock art data should be seen in the context of other archaeological data sets, both with and without overtly symbolic content

It is important to grasp that areas used by humans, even on a seasonal and temporary basis, as “landscapes”.  Landscapes, made up of topography, vegetation and seasonally changing features, can be a human construct, and are certainly part of human perception. People interpret them in various different ways, observing landmarks, embuing particular sites with both functional and spiritual meaning, and employing natural features as useful devices (e.g. caves for shelter, narrow passes for hunting traps).  Barry Cunliffe has articulated this as follows:  “Environment . . . like space exists without dependence on the observer.  Place, on the other hand, is specific to time and people.  It is enmeshed in a network of beliefs and values:  it is a personal construct.  One can fairly argue, as others have done, that a landscape is a network of personal places” (2000, p.111).  Later in the same paper he says that “for people, their personal landscapes have many faces:  they constrain and legitimize through their boundaries; they inform through their symbols; they instill awe through their waywardness; and they inspire through their distant vision of futures” (2000, p.113-4). 

The more mobile the population, the more important and wide-ranging the territory and, therefore, the more important the concept of landscape.  Holl and Dueppen (1999, p.24), talking about northeast African rock art, express this as follows:  “Pastoral-nomadic territories are not self-contained and banded large spatial units, but mainly networks of places (water places, streams, grazing lands, shrubs, groups of trees, meeting places, religious places etc.) selected according to their pastoral and effective resources, social, ritual and symbolic importance in the actual cultural landscape”.  Into this framework they suggest that rock art in certain areas may have been used as territorial markers.  Mithen (1996) offers a similar interpretation: “Hunter-gatherers do not just live in a landscape of animals and plants, rocks, hills and caves. Their landscapes are socially constructed and full of meaning” (p.166).

Craig Alexander has begun his research into Alpine rock art in Italy by asking a series of questions, which directly explore the relationship of rock art with the landscape:  “Whilst we cannot identify the ‘hand’ of an individual carver and so tell directly whether multiple people carved at a given site we can look for concentrations of petroglyphs that can be dated to a given period. Are they at similar altitudes? Are they near known settlements? Are they on particular soil-types? Does their distribution coincide with likely activity areas? Do they coincide with expected paths between settlements (or, more broadly, the valley floor) and higher land? Are the sites intervisible? Are they interaudible?” (2006, p.20 WR).  He goes on to look at diachronic issues:  “which sites were re-used over time? Are they simply the easiest of access? What are their characteristics? Are they close to water? Close to paths? At places of greater than average viewshed? At places of enhanced visual exposure? Do certain petroglyphs from one period typically co-occur with certain types from earlier periods? Do the locational characteristics of sites with a predominance of later petroglyph forms differ from those with a predominance of earlier forms? What does this tell us about the changing interrelationships of people, activities and land?” (Alexander 2006, p.21 WR).

Mixing datasets has sometimes helped.  Bradley (1994) used his knowledge of the archaeology of northern Britain to form hypotheses about expected differences in pattering of rock art assemblages, finding that there were differences between upland and lowland areas which were used at different times of the years in a mobile lifestyle.  Searight’s (2004) attempts to analyze the rock art of Morocco in a thorough review of all the available data has produced some interesting results (see also Distribution).  She used archaeological data to draw conclusions about the nature of the rock themes, and how they related to what the groups that occupied the area actually used.  Like most rock art studies, Searight’s is plagued by a lack of direct dates, but she was able to compare animal remains found in archaeological deposits with animal depictions, and found that “of the wild animals, only antelopes figured at the top of both lists.  The other animals engraved were rarely found during archeological excavation while, on the contrary, excavations brought to light remains of animals very rarely engraved” (2004 p.85). Although no firm conclusions can be drawn from this finding, it again shows how archaeological data can be used to establish relationships between different datasets.  Jean McMann suggests that many other disciplines can be brought to bear on rock art research:  “Contemporary developments in linguistics and semiotics, neuro-physiology and even physics, as well as studies in comparative religion and anthropology can be explored for answers to questions about the symbols” (1980, p.146).

Wilkinson (2003) explores the possibility  of continuity from the Predynastic to Pharaonic periods for Eastern Desert rock art,  but his ideas have not met with universal approval, due to the nature of the evidence that he uses.  Huyge (2002) also considers rock art in terms of later developments, suggesting that a religious explanation for some rock art is plausible given that the religious expression of the Early Dynastic and Old Kingdom periods had to have its origins in Egypt’s Predynastic past.  In both the cases of Wilkinson and Huyge, however, the evidence can be interpreted in more than one way, and the parallels drawn between rock art images and the more formal iconography of different periods are not always obvious.

Huyge (2002) raises the question of a regionally differentiated view of rock art.  Huyge has created a detailed diachronic and interpretative scheme for the rock art of Elkab in Egypt north of Edfu, from Naqada I to the post-Pharaonic periods, and raises the important point that schemes based on motif and symbol can only be applied to other areas with extreme caution.  His point is that geographical and functional differences in population groups must be taken into account:  “In this sense the rock art of quarry workers and herdsmen will in all probability be thematically distinct, while at the same time possessing similar meanings and motivations, and subordinate to the same syntactical rules” (2002, p.204-205).

As already highlighted, Lewis-Williams has suggested parallels between the art work of the San group of South Africa and the Upper Palaeolithic work in western Europe, and he offers a shamanistic interpretation, involving altered states of consciousness, for the production of Upper Palaeolithic art (Lewis-Williams 2002) but does not suggest that all rock art can be explained in this way (Lewis-Williams 2005, p.165-6).  Nor does he suggest that shamanism is a simple explanation:  “The making of San rock paintings was essentially (or ‘principally’) associated with a range of shamanistic beliefs, rituals and experiences and was situated within a tiered shamanistic cosmology and complex social relations” (2002, p.166).

In the future, statistical studies should help to analyze and compare locational choices, orientation of particular motifs and association of motifs with each other.  This is probably going to be a very important approach in the future, helping to highlight statistically significant aspects of rock art associations, which will be a considerable assistance in rock art analysis.

It is clear that rock art cannot be examined out of the context of other data – environmental, ecological, archaeological and historical.  To see it in isolation is unrealistic, because it was created within the context of socioeconomic activities and beliefs.  It can help to raise questions for which answers may be located in other datasets, but it cannot be seen as an information source in isolation.


Introduction to rock art dating

Dating is essential for any meaningful interpretation of rock art.  Without knowing which art depictions were contemporary with others it is impossible to determine which themes were portrayed at a given time.  At the same time, it is impossible to tie in the rock art with archaeological and historical data, meaning that it is unusable as supplemental data.  It is known from various parts of the world that rock art may be considerably old.  Examples in France date back to the Upper Palaeolithic, at a time when some of the earliest burials were also being made.  However, there is also a considerable amount of rock art of considerably more recent date, some of it historical, and some of it modern.  In Egypt the modern Bedouin still engrave the rocks with clan symbols, and I have photographs of carvings quite clearly left by modern expeditions, as mentioned above.

Smith (1968, p.5) highlights some of the difficulties presented by rock art, when he says that where establishing a chronology is concerned “there is still a good deal of disagreement on details, on the position of individual figures or sites, and especially  on the absolute (chronometric) dating of periods, styles and figures”.  He goes on to compare the situation in Africa with that in Spain, where rock art studies have been menaced by controversies regarding dating schemes:  “If such uncertainty can exist in an area such as eastern Spain, which has been methodically explored and studied for over half a century and where the purely archaeological sequence is fairly well known, we need hardly wonder that in North Africa we are still very far from certainty, particularly in the Sahara, where intensive study of the art is a recent feature and where archaeological investigations orientated to rock art problems have only recently advanced beyond the surface collecting stage” (Smith 1968, p.7).  Susan Searight offers a further comment:  “Without a preliminary chronological setting, any attempt at understanding the way of life of the engravers or interpreting their work runs the risk of mixing incompatible elements” (2004, p.125).  All of these are serious impediments to incorporating rock art data into the overall framework of the archaeological context.  Different schemes and approaches to the whole dating problem are described below.

Most of the systems described below are relative dating systems, which attempt to slot sites into a place within an existing chronological framework.

Relative Dating

In some very rare cases an image or scene may be sealed beneath more recent archaeological material, providing a terminus ante quem. If a decorated fragment falls on top of an archaeological context, it provides a terminus post quem for the decoration. A famous example of rock art sealed within earlier contexts is the site of La Mouthe in France, where deposits of Palaeolithic and Neolithic date completely sealed the entrance to a cave where paintings and engravings were found.  The findings were presented in 1896 and convinced many archaeologists and geologists who had previously doubted the antiquity proposed by some scholars for the rock art on the basis of artefacts found within the caves (Daniel 1981, p.99).  Another example was the cave of Le Tuc, also in France, where two stalagmites had to be removed before the art could be reached (Daniel 1981, p.99).

Patination has been used by some scholars to give a sense of the relative age of different petroglyphs.  Sweinfurth (1912) was one of the first to use patination to address the issue of grouping engravings chronologically.  Winkler formed a chronology of archaeological occupation of the Eastern Desert based on the strength of patination over rock art scenes, using a relative rating scheme to help him – a 0 rating was high patination and a rating of 10 was shallow patination (1938).  Winkler’s chronological scheme was innovative but has been attacked on many occasions since.   Patination is still being used in rock art research, and there are examples where patination clearly marks chronological differences – for example at the EDS site ED-1 a horse and rider are depicted near much darker boats and animals. However, Wilkinson (2003) offers warnings about this, based on his survey work in the Eastern Desert.  He gives the example of a single composition in the Wadi Umm Salam where a scene showing a line of ibex, nose to tail, were half dark-patinated and half light-patinated:  “At first we could not understand this difference, until we noticed that the overhanging rock ledge caused half the rock surface to remain in shade throughout the day.  The other half was exposed to the full glare of the sun . . . This was a cautionary lesson:  it meant that patination could not be taken as a reliable indicator of age, even within a single ‘canvas’ or rock art  (2003 p.55).  Other problems include the impact of the underlying mineral content of the rock itself, which might give rise to different effects of patination, and the way in which different scenes are exposed to different levels of weathering.  Anati highlights the fact that a number of processes can be at play to alter the appearance of the patinated surface:  “Depending on the composition of the rock and its exposure to the sun and wind, the surface patina can be multi-hued” (1999, p.25).  Even where patination actually is accepted as a tool for distinguishing which engraving is older than another, it is of no help for indicating a possible time period for the gap between one engraving an another – a gap between one image or another could be measured in months, years, decades or centuries.

Superimposition, when one drawing is etched on top of another, establishes that one image is more recent than another, but doesn’t give any information about the time gap between the two images. At EDS site ED-1 an image of a camel is superimposed over much darker animals, and the combination of superimposition and patination seems to argue that the two are some distance apart in terms of age. In general terms, however, there is always the possibility that an artist superimposed one image over another deliberately, and that they are of identical or similar age. 

Weathering is the process by which chemical and physical processes impact an image after it has been created.  Lorblanchet (1992) suggested that differential weathering could provide an indication of the relative ages of different rock art images.  Unfortunately, weathering does not necessarily occur at a consistent rate across all rock art surfaces in a given area, and is not always a reliable way of determining relative ages. A more scientific approach to weathering is micro-erosion analysis, pioneered in Australia, which examines the surface of a rock using a microscope.  The rate of weathering is is measured in a given area by sampling a number of rock surfaces. A rock carved with petroglyphs is then analyzed in the light of this data.  The weathering of rock surfaces are measured giving a duration of time since the rock surface was manually eroded during the process of etching the petroglyphs (Wilkinson 2003).

Style has sometimes been offered as a way of dating sites, with sequences developed using a number of different techniques (e.g. superimposition, weathering and cross-dating with motifs with known dates). The best known of these systems is that developed by Leroi-Gourhan (1968b) for Upper Palaeolithic French cave paintings.  Fuchs (1989), offers a warning about using style for dating rock art offering the example of a scene at site EG-A/WB5 in the Wadi Barramiya, which shows archers hunting ungulates with the assistance of dogs who surround the game.  Fuchs says that the “stylistic variations observable in the representations of the hunters – from realistic to geometric – demonstrate the need to be cautious whenever style is considered to be characteristic of a specific period” (p.145).  MacDonald (1998) also expresses concerns:  “Since the 1950s Saharan rock art studies have evolved from the diffusionist ramblings of early explorers to gradually more sober evaluations of the problem.  Still, until some form of rock art dating is attempted in the Sahara (e.g. Van der Merwe et al 1987), one must worry that new chronologies and stylistic groupings are mere sophistic triumphs for the field’s practitioners”.  My own concerns about many of the simple linear rock carvings of the Eastern Desert variety are that any concept of style is exceedingly difficult to define.  Only a handful of motifs have distinctive features which appear to relate clearly to each other and their creators.

Comparison with similar dated motifs to cross date rock art have been attempted.  The principle is to locate clearly recognizable and distinctive motifs that are common to rock art scenes, but which also occur on artefacts, usually from the Upper Egyptian Nile, which have been tied into a secure chronological sequence. In the Eastern Desert this has been considered to be a potentially useful line of inquiry, but remains full of pitfalls.  See the discussion on the Eastern Desert Rock Art page for details.

Donald and Susan Redford studied rock art technique, and suggested that pecked petroglyphs are earlier than carved ones (1989) but there is still too little data to confirm or deny this hypothesis.

Using the images of animals and equating them to past environmental regimes is another system for attempting to tie in rock art images with a chronological place holder.  A number of studies in the Sahara have been based on time ranges associated with specific animal species. Animal types can be viewed as assemblages, with particular combinations indicating particular environmental conditions.  These can then by fitted into known environmental sequences.  Muzzolini (1986) has attempted this for the Tassili region (eastern Algeria). However, even though some animals ceased to be present at certain times, due to changes in environmental conditions, there is nothing to say that certain species were not represented by people who had seen them elsewhere.  Smith (1967) says that “We only need to remember the cases in southern Africa of paintings made by Bushmen observers of non-Bushmen peoples to recognize that we cannot take for granted an absolute identity between artists and subjects” (p.25).  He also points out that there are good palaeontological and stylistic reasons to believe that “the only rhinoceros so far found in the rock art of the Upper Nile Valley of Upper Egypt is such a caricature, done from memory” (p.28). Fuchs (1989) says that this point can be demonstrated by the species depicted in Wadi Barramiya.  Another problem is that some species endured in areas for long periods of time, and others have migrated in and out of areas due to short term seasonal variations or longer term environmental changes.   A good example of this type of problem comes from Hierakonpolis site HK64, where an ostrich image, typical of those found both on D-Ware and in the Eastern Desert was found superimposed on a cartouche of Amenhotep I (Friedman et al 1999, p.23 and Friedman 1999b).   Oddities like these undermine attempts to date rock art by animal specie.  A more bizarre problem is that some animals may have been briefly introduced to the area for specific reasons.  The elephant images may belong to early periods when conditions could have supported them, but as Morrow and Morrow (2002, p.16) point out, later carvings “may have been inspired by noteworthy events, such as occurred during the Ptolemaic era when a train of elephants captured near the Ethiopian coast were walked from the Red Sea through the Wadis Barramiya/Abbad to the Nile”.

Extinct species are sometimes considered to be relevant. For example, the mammoth is a frequent topic of Cantabrian Upper Palaeolithic rock art.  As the mammoth died at by the end of the last ice age, this provides a terminus ante quem for depictions of mammoth.  There is nothing as clear cut as this for Egypt, although in the Sahara the Bubaline (Epipalaeolithic/Early Neolithic) period of rock art is named for the extinct giant African buffalo Bubalus antiqus (MacDonald 1998).

At the other extreme, items and animals that are known to have appeared only recently can eliminate early dates for certain scenes.  Obvious examples are those showing four-wheeled drives, camels, horses with riders, saddlery and chariots, and modern weapons.

Nearby archaeological remains may or may not be associated with a site, and even if the same types of site and artefact are consistently located near to the same sort of rock art, this might simply indicate that the same types of location were preferred by two separate and chronologically disparate groups, rather than representing one group or many groups of one period.  However, the presence of archaeological remains does encourage research to explore possibilities of a connection between the two datasets.  As Searight says (2004, p.126) “It is clear that the mere presence of a dated object near a rock art site cannot be taken as indicative of the age of the engravings.  At the best it imparts the rather banal but not un-useful information that people of such a period were in the area”.  Art usually exists within a cultural context (Anati 1999, p.22).  Whilst petroglyphs cannot be dated solely by the context within which they are found, it would be irresponsible to ignore a context completely.

Spatial analysis was recognized as important by Philip E. L. Smith in the 1960s (Smith 1968, p.27).  Smith highlighted the value of comparing different regions, and of understanding differential survival of styles, techniques and motifs in different regions. Spatial analysis is also considered to be a possible avenue for future consideration, as emphasized by the UNE: “At some sites differential weathering and superimpositions indicate that ‘bursts’ of artistic activity occurred over considerable time periods. In these cases intra-site patterning can reflect chronological patterning. By implication, there may be sites which were used for a short period only in which techniques, motifs and colours in use at one time have been ‘stranded’ by previously held cultural values determining site significance. If so, trends in the inter- site distribution of artistic variables can also provide evidence for sequence, assuming other determinants of such patterning can be monitored and taken into account” (UNE).  Bradley (1994) and Searight (2004) have used this type of approach.

Clearly the best solution is to combine as many of these devices as possible.  Chippindale and Tacon (1998 WR) have taken a range of devices to determine a chronological system that they refer to as “cabling” after Wylie (1989), mentioned in brief above:  “As a strong cable is made by combining together many threads, each one individually weak, so can a variety of forms of evidence, and individually weak deductions drawn from them, be tied and pulled together, collectively to build a reasonable secure knowledge” (Chippindale and Tacon 1998 WR).  They use archaeological and environmental data to “bridge” between  rock art and other types of data.  Searight has also attempted to combine several of the above approaches in her survey of Moroccan rock art, resulting in a chronological framework which she herself describes as “very tentative” (p.138).  Her main points of reference are environmental data (identifying periods when the desert would have supported pastoral communities), wild animal species, domesticated animal species, the presence of metal weapons, shields, chariots and inscriptions. Muzzolini, who acknowledges that there is considerable bias against using rock art as a dataset due to the difficulties in dating (1992, p.147) has also taken the approach of pulling together a number of different strands in an attempt to provide a relative sequence for rock art.  He classifies rock art by technique, patination, anthropological type represented and style, and uses patination, superimposition and seriation of fauna and themes to tie rock art into well dated environmental sequences.  Huyge (2002) has attempted to create a chronology of 350 rock drawings for the El Kab area of Egypt (on the east bank of the Nile, north of Edfu).  He uses superimposition, rock patina, iconography, and archaeozoology to create a sequence which he then analyzes in terms of theme, technique and style, and then goes on to use the chronological framework to interpret the art.  The understanding that interpretation can only take place within a diachronic framework is explicit.

In a article in the journal Antiquity (Huyge et al, 2007) a collection of rock art images located near the Egyptian village of Qurta on the Northern edge of the Kom Ombo Plain has been dated to 15,000 years old, which equates archaeologically to the late Palaeolithic Ballanan-Silsilian.  The themes of the rock art are listed as “Bovids are largely predominant (at least 111 examples), followed by birds (at least 7 examples), hippopotami (at least 3 examples), gazelle (at least 3 examples) and fish (2 examples)” (Antiquity 2007 WR).  All are engraved, none painted.  On the grounds of style (the art is considered to be completely different from other rock art considered to be Predynastic), the species represented, and nearby archaeological data, the researchers conclude that the rock art is Late Palaeolithic, corresponding to the Ballanan-Silsilian occupation of the area.

Finally, in all many of the above approaches, the creation of typologies has been an essential part of the system.  It seemed worth mentioning that even when typologies are disputed on chronological grounds, they may be invaluable for creating a grammar of rock art.  Just as ancient Egyptian hieroglyphs can be discussed by reference to a common system, like Gardiner’s sign list, typologies which assign codes to each type of motif allow both conversation without reference to individual rock art photographs or diagrams, and also allow stylistic/thematic standardization to be recognized and recorded.  Huyge makes frequent reference to Cervicek’s 1974 typology for referring to boat types – he is not using Cervicek’s chronological scheme, but he uses his numbering system as an aid to discussion (2002, p.197-205).

Scientific dating techniques

Chronometric techniques have proved to be useful for dating rock art only in some cases, and scientific dating has only been attempted in one area of Egypt – at El Hosh in Upper Egypt, where Accelerator Mass Spectrometry Carbon 14 methods have been employed.  Chronometric dates can be employed where organic materials have been used in the production of rock art – for example, paints, paint binders, human blood, charcoal, plant fibres and dyes, amino acids (Watchman 1993, p.61).  They can also be employed where rock art surfaces have been covered by organic or mineral deposits where a minimum age can be obtained – oxalates, mineral crusts, algae and silica skins (Watchman 1993, p.62) . 

One of the biggest problems  of direct dating of materials is the risk of contamination.  There is also the problem associated with the damage to the image that is potentially caused by removal of the samples required for direct dating: “The dilemma for archaeologists is the necessity to minimize sample sizes in order to protect the art versus the futility of collecting a sample which is too small to contain sufficient organic material for dating. This highlights the necessity for sampling to be undertaken only with expert technical assistance” (UNE).  Less invasive methods are being investigated (Dorn et al 1992).

Accelerator Mass Spectrometry Carbon 14 has been used apparently with a number of sites globally. The technique was, pioneered for dating rock art in Australia. Dorn et al (1992) say that “The great advantage of AMS is its ability to measure the radiocarbon content in samples of only a few milligrams” (p.136).  Dorn et al (1992) have used the technique in dryland environments to date organic inclusions trapped in pockets under rock varnish, a coating that appears on exposed rocks in dryland environments, providing minimum ages for rock art surfaces.  In Egypt, AMS has been applied at the site of El Hosh (Huyge et al 2001), as follows, although the results have not been as definitive as initially hoped.  A summary of the study follows:

    El Hosh lies c.30km south of Edfu on the west bank of the Nile.  It was decided to use AMS to make age determinations of the El-Hosh petroglyphs.  The petroglyphs include abstract forms, boats, human figures and animals. Huyge and his team hypothesized that whilst a lot of images were probably, to judge by their subject matter and execution, prehistoric and Predynastic, other more abstract forms belonged to a different period. Therefore, whilst giraffe forms were tentatively assigned on stylistic grounds to the Naqada I period, the curvilinear designs capped with protuberances (possibly fish traps) were thought on the basis of patination to belong to the early Predynastic dating to the 6th or 5th millennium B.C. (Huyge et al 2001, p.69).

    In order to test these hypotheses, carbon-bearing substances in patina and rock varnish that formed within the petroglyphs themselves were sampled, and AMS carbon 14 was then applied to the samples.  Analysis of the surfaces before the samples were taken indicated whether they were formed by accretion of aeolian components or whether they resulted from leaching.  Samples were only taken from those surfaces that could with confidence be determined to have formed by build up rather than leaching.  The dates obtained provided termini ante quem for the petroglyphs.  Huyge et al summarize the dates that resulted as follows:  “They document different stages of varnish formation.  Whereas it is evident that the rock art must be older than 2450+/-320/2280+/-320 C14 years BP (Third Intermediate Period to Ptolemaic period at 68% probability) and even 3740+/-300 C14 years BP (Old Kingdom to Middle Kingdom at 68% probability), the date of 6690+/-270 C14 ears BP (5900-5300 cal BC at 68% probability) is outstanding.  This age estimates indicates that the El-Hosh ‘fish-trap’  motifs are well beyond the age of any other graphic activity known in the Nile Valley”. 

    Although the authors speculate on just how much older than the above date the fish trap motifs may be, they acknowledge that it is not possible to establish a clear vintage for them.

It has also been suggested that cation ratio techniques can be used to date desert varnish which sometimes overlies rock art images. 

Working without dates or chronological frameworks

Some rock art studies have proceeded without the benefit of chronological frameworks, and have simply provided descriptions of motifs or comparison between motifs of similar types without attempting to draw conclusions about any chronological organization.  These often have a limited application, and some are more successful than others, but they should be mentioned.

One example is Mary Aitken Littauer’s overview of rock art of Transcausia, central Asia and Outer Mongolia (1977).  Her view is that “lacking archaeological contexts, most of them have been dated only roughly by their publishers on the basis of typology” (p.243).  Her view is that the carvings were too important to wait for further publications, due to their potential contribution to knowledge about chariotry.

Dating Summary

Relative dating is still controversial and is quite clearly not without difficulties, although some approaches offer hope in general terms. Watchman offers a cautiously optimistic opinion: “Challenges for rock art chronologists are to define protocols for sampling, to refine and develop dating techniques and to select appropriate paintings to establish regional chronologies” (1993, p.63).  Muzzolini (1992) also believes that there is genuine potential for establishing rock art chronologies in the Sahara, and sets out to demonstrate how it can be achieved.  He observes that very few archaeologists use rock art because of the perceived difficulties in dating it, and that rock art experts are often thought to use guesswork rather than a disciplined approach (p.147).  He accepts that dating rock art by association with nearby archaeological remains is “mere conjecture” (p.147).  Instead his approach, like that described by Chippindale and Tacon (1998 WR) weaves together many threads of data towards establishing a sequence.  Smith concludes (1967 p.11) that providing a date for the earliest Saharan rock art “still hangs in mid-air”, but advises against ruling out an Epipalaeolithic or even Palaeolithic date for some Saharan rock art on the grounds that the archaeology is certainly there, even if it cannot yet be linked to the rock art.

Direct dating, although offering considerable potential, still seems to have a long way to go. 

Racial Determinations

Attempts to determine racial characteristics from rock art representations, which might help to identify movements of people through space and time, have been plagued by difficulties.  Of the different suggestions by different writers summarized by Smith for north Africa (1968, p.23), the main notable feature is the lack of consensus. 


When looking at rock art in a geographical or archaeological context, it is necessary to consider the idea that looking at the location of rock art can reveal something about its role or its meaning.  In this sense there is considerable overlap with research in the interpretation of rock art. 

Looking at the distribution of rock art may or may not help to understand why it was created in certain locations.  Understanding rock art frequently involves chicken-and-egg situations, and this is one of them.  Lewis-Williams (2005) states that geographic location cannot reveal the meaning of an art – he says that to attempt to do so “would be to fall into the well-known trap  of empiricism:  one cannot logically induce rock art explanations from supposedly theory-free data”.  In his own research into San rock art of South Africa, Lewis-Williams was unable to find any significant pattern of distribution “beyond the self-evident observation that the painted sites were important points in the landscape” (2005, p.169).  However, looking at distribution may not explain the art but it may help to explain how the landscape itself is used, which is of value archaeologically.

Other issues are practical ones.

Fuchs eloquently expresses one of the key problems with considering the significance of rock-art distribution (1989, p.149):  “The distribution map of rock art sites in the Eastern Desert and in the Nile Valley (Resch 1967) reflects research activities, itineraries of expeditions, the existence of suitable rocks and probably the geological structure;  it cannot be expected that it shows the real distribution of sites, nor their numbers”. 

Mike and Maggie Morrow point out that it must be remembered that terrain can change over long periods of time, creating blockages across previously passable routes (2000, p.180).  In the Eastern Desert, rock falls, sand dunes, flash floods, and heavy storms may all have altered the way in which the landscape looks today.  Unless the original landscape can be reconstructed, for the time periods concerned, our understanding of why sites were located in certain areas may be coloured by the changes that have occurred since they were created.

However, analysis of distribution has been shown to bear fruit in some research areas.

The University of New England website looks at how rock art has been married with archaeological data in the past:

    “In most situations detailed ethnographic information on rock art is not available and it is not possible for a researcher to obtain specific meanings. Nevertheless, it is still possible to study the role of rock art in past societies by analyzing the content as well as the natural and cultural context of the art. For instance, in the Central Queensland Highlands there is a consistent association between rock art and burials. It is clear that rock art production in this region was closely associated with disposal of the dead (Morwood 1984). This approach to the ethnography of rock art is based on the notion that rock art data should be treated as part of broader cultural systems of meaning and, as such, should be analyzed in a similar manner to other archaeological data — that is, in terms of spatial and temporal patterning and relationship to other archaeological material. The argument here is that the basic archaeological problem of establishing links between material evidence and the original cultural context remains the same regardless of whether the evidence being examined consists of stone artefacts, food remains, sediments, or rock art. Middle Range Theory is often used as a device for establishing links between rock art and cultural context and for producing hypotheses which can be tested in the field.”

Richard Bradley has looked at the landscape of the British Neolithic from many different angles.  One of the features of the landscape is a type of abstract rock art forms known as “cup and ring marks”.  Cup and ring marks are carved into exposed rock faces.  Bradley looks at how people perceive their own landscapes and how they may mark them according to their usage of it at different times of the year (Bradley 1994). He brings together land-use studies and rock art in an attempt to answer the question “how did prehistoric people perceive the landscape in which they lived?” (1994, p.96).  He describes two obvious problems – a) difficulty of analyzing the prehistoric perception of the landscape, b) difficulty of interpreting systems of petroglyphs, especially when those petroglyphs are abstract.  His aim, however, is to “devise explicit analytical procedures that can be replicated by other workers” (1994, p.96), and that is what is of interest here, particularly in terms of non-sedentary communities.  The engravings date to the Late Neolithic/Early Bronze Age when there is little sign of cereal cultivation or large permanent settlement, but a lot of evidence for domesticated animals.  Without going into all of Bradley’s methodology, observations and analyses (see his paper, 1994) Bradley finds clear patterns in the distribution of the rock art.  Bradley’s hunt for geographical/ topographical variation in rock art distribution begins with hypotheses about how the use of rock art as a communication device will impact the distribution of different types of rock art motif and assemblage. He comments as follows:  “the distribution of the petroglyphs may stretch from regions that were capable of sustaining year-round land use to others which were better suited to seasonal occupation, as summer pasture and/or hunting rounds.  In either case we would expect the audience for the carvings to have changed:  from people who might have come into contact on a day-to-day basis in the lowlands, to those who came into the uplands only occasionally, and from a range of different home areas.  If so, the petroglyphs on the higher ground would probably have been addressed to a larger and more varied audience” (1994, p.101).  Bradley was able to divide rock art into two groups – those within 7.5km of freely draining lowland sites, and those which are further away.  They conform to a type of behaviour that he predicted.  This approach may have a lot of potential for other areas where abstract or highly stylized rock engravings occur. Bradley suggests that rock art is associated with post-domesticated mobility in some areas, including Britain, Scandinavia and Iberia.

Searight’s work (2004), which took into account many different projects initiated by Bradley,  incorporated not just a medium sized area like northern Britain (Bradley 1994), but the vast area defined by Morocco’s political boundaries.  She divided Morocco into zones to enable her to detect regional differences in much the same way that Bradley approached  a much smaller area, and took a comparative approach to the data.  She found that “certain types of engravings were concentrated in certain areas and that little mixing took place” (2004, p.121).  In other words, like Bradley in northern Britain, she found an underlying pattern in the themes chosen in her study of Morocco. Finally, she concludes that in most cases “the rock art sites were associated with ‘useful places’, high altitude pasturages and passes in the mountains, passages between mountain chains, ridges providing vantage-places from which to observe the surrounding plains, or water sources . . . . It was proposed that the engravings could define territories, proclaim ownership, indicate the presence of water (not always immediately visible), commemorate heroes or battles” (p.153).  She believes (p.153) that the location of rock art can throw light on the activities of mobile populations who used the regions analyzed, but she found nothing to tie in the rock art directly with the archaeological material.  She also points to the fact that she is unable to explain why some ridges and passes were not marked.

Finally, Caroline Malone makes a point about rock art distribution (in this case in the UK, but applicable elsewhere) that might seem obvious but again points to how rock art can offer, even at a minimal level, some important information:  “Apart from the sheer interest of these sites, the presence of rock art in the remotest places of Britain shows that all the landscape was known and visited.  No part of it was excluded, and Neolithic people exploited, identified, marked and knew their land very well” (2001, p.256).


Although many survey teams have been successful in finding and recording 1000s of petroglyphs, Kleinitz (2006 lecture) makes the point that it is entirely usual to systematically survey the same area many times and still make new discoveries.  This is not due to a failure of the techniques used, or any lack of precision, but to the fact that different lighting can have an enormous impact on the visibility of engravings.  Morrow and Morrow (2002, p.14) also highlight this problem:  “The direction and quality of the natural sunlight near the tropics can result in extremes of contrast and often there is only a short period, perhaps a matter of minutes each day, when a petroglyph is clearly lit, or in some cases visible at all.  Certain carvings appear markedly different under changed lighting conditions.  Orientation, date and time are therefore crucial when deciding the ideal time to visit a particular site” (2002, p.14).

There is a particular problem in cases when an engraving is heavily weathered (Fuchs 1991, p.62).

Similarly, Morrow and Morrow (2002, p.13) highlight the difficulties that time constraints can impose on surveys.  Their 2000 and 2001 surveys involved using slow-moving vehicles to observe cliff faces look for suitable surfaces.  When these were identified, the rock faces were then examined in person.  However, the writers are very aware that this approach is a “limiting factor” which may have left important sites in concealed situations undiscovered.

Blending Approaches

A recurring theme through all these pages is the use of a number of different strands of data woven together to try to assemble a complete picture of a given area or site.  Herity (1974) provides a very useful example in his analysis of highly ornamental and abstract southern Irish Boyne art.  In Chapter 5 of his book Irish Passage Graves (p.151-180), he asks the question “was there a Boyne culture?”.  It is a very valid question because it addresses the issue of how the builders of these tombs related to each other, and to what extent they shared a language of symbol and belief.  He addresses the following data:  chronology, similarities/differences with other British groups, distribution of tombs, settlement data, artefactual data and potential food supply.  More about this subject is phone in a following section.


Although I have been impressed with the number of ways of looking at rock art, and of trying to extract information from it that can provide data to archaeologists, I have finished  writing this section with a sense of frustration.  There are many approaches, and many attempts have been made to analyze different types of art from all over the world, but the solutions remain unassailably elusive.  The two key problems are interpretation and dating.

Reading a number of articles which attempt to interpret rock art, I have been struck by how few of them reach any useful conclusion whatsoever.  Renfrew and Bahn, in their well written introduction to cognitive archaeology (2000) provide no examples of convincingly useful approaches to understanding rock art analysis in anything other than structural ways.  In spite of emphasizing that cognitive elements should be seen in the light of other data, they do not demonstrate how this can be achieved for rock art studies.

Christopher Tilley’s work with Swedish rock art (1991), which I acquired after it was mentioned in Renfrew and Bahn (2000), is one of many frustrating examples of the type of positive and innovative attitude and well thought out approach which ends with completely inconclusive results.  Tilley looks at rock art as text and grammar, and attempts to find a language in the rock art that he analyses (some of which is stylistically and technically similar to Eastern Desert types of depiction).  However, he is unable to reach any definitive conclusions from all his research.

As far as interpretations concerned, Renfrew and Bahn conclude, somewhat depressingly, “The analysis by archaeologists of information contained in visual form in artifacts and on representations is still in its infancy.  Prehistorians are much less practiced at this kind of study than are art historians and Classical archaeologists.  There is still much to learn” (2000, p.419).

Most promising are the approaches characterized by, amongst many others adopting similar approaches, Richard Bradley (1994) and, in his footsteps, Susan Searight (2004). These approaches accept the difficulties of interpreting some rock art, and look instead at analyzing the distribution of rock art within a given landscape and finding an underlying structure which can be linked into a region’s economic activities.  Again, dating provides a challenge.  Whereas Bradley can engage with his material confident in its Neolithic date, Searight has the same problems that have been described above – it is difficult to discuss distribution of engravings when the engravings are undated. 

Where dating is secure, Bradley’s approach offers a way of incorporating rock art into archaeology, but only in terms of identifying a structure in rock art use and distribution – as Bradley acknowledges in his own particular case, it cannot get to the meaning behind it.  What research projects by Bradley and others do tell us is that the specific character of rock art assemblages may depend on locations, so this clearly mattered to their creators and was representative of some idea that they were expressing or communicating.

Dating is another key issue.  There are some promising direct dating approaches being tried out in Australia and elsewhere, but the one attempt to use this method in Egypt, at El Hosh, produced somewhat inconclusive results.  It is probably fair to say that there is no immediate prospect of any of these methods being attempted in the foreseeable future on any Eastern Desert rock art.  That leaves us to fall back on relative dating techniques, and these have their own legions of problems associated with them.  The page on Eastern Desert Rock Art will consider some of the techniques applied to the Eastern Desert, and it is quite clear that the most promising approaches are those that bring together a number of strands of information about dating to establish a relative date.  Direct dating has potential, but it is still in its early days.

Herity’s analysis of Irish tombs and associated rock art (1974) appears to present a useful approach, making use of different datasets to form a view of rock art in a wider context of significance.

I am absolutely longing to find rock art analysis that convinces me that rock art can contribute in a major way to archaeological research.  At the moment, I am more concerned with its limitations.  Having said that, I’m scarcely the best read person in this field, and as I read more specialized volumes and papers I may change my tune.  Here’s hoping!



Copyright Andie Byrnes 2007