Anda di halaman 1dari 73

THE EVALUATION OF THE SUPERPOSITIONING

SEQUENCE OF PAINTED IMAGES TO INFER


RELATIVE CHRONOLOGY
DIEPKLOOF KRAAL SHELTER AS A CASE STUDY

A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS


FOR A BACHELOR OF ARTS (HONOURS) DEGREE IN THE DEPARTMENT OF
ARCHAEOLOGY, UNIVERSITY OF CAPE TOWN

SIYAKHA MGUNI

NOVEMBER 1997

Copyrighted Material: University of Cape Town (1997)


Dedication

This is dedicated to my late grandfather, Msombuluko Jim Mguni whom I never met during
his lifetime, to my parents, Susan and Jacob and to my two brothers, Bhekinhlanhla and
Banele for all their love and support.

i
Acknowledgements

I am very grateful to a number of people who contributed to the success of this project. I
would especially like to thank Professor John Parkington for suggesting the project and for
his constant guidance throughout the project. Dr Christopher Chippindale, Dr Justin Hyland
and Anthony Manhire made insightful contributions on our visit to Diepkloof. Great thanks
are also extended to Dawn Fourie for the preparation of the document and to Royden Yates
who assisted me with the Harris programme.

I declare that this thesis is my own work unless otherwise stated


and acknowledged. It is submitted for the degree of Bachelor of
Arts in the Department of Archaeology, University of Cape Town. It
has not been submitted for any degree or examination before.

........................SIYAKHA MGUNI

..............................................................

Signed on...08 November 1997...at the University of Cape Town.

ii
TABLE OF CONTENTS

Dedication (i)
Acknowledgements (ii)
Table of contents (iii)

1. INTRODUCTION

Motivation and Objectives 1-5


A Survey of Previous research 5 - 12

2. DESCRIPTION OF THE STUDY AREA

The Choice of Diepkloof Kraal Shelter 13 - 14


Environmental context of DKS 14 - 16
General research context of DKS 16 - 19

3. METHODOLOGY

Background to the Harris Matrix System 20


Application to DKS rock art analysis 21 - 24
Fieldwork at DKS site 24 - 27

4. ANALYSIS OF THE SUPERPOSITIONING SEQUENCE 28 - 34

5. DISCUSSION 35 - 42

6. CONCLUSION 45 - 44

7. REFERENCES 45 - 48

iii
8. ILLUSTRATIONS Page

Figure 1: Diagram showing panel 2 superposition in 3 layers at Tandjesberg, Orange 8-9


Free State (from Loubser 1993).
Figure 2: The Harris matrix method. Three recognized relationships (A, B and C) 19-20
between stratification units. I, II and III show how the diagram is generated.
Figure 3: Shaded areas are red ochre and stippled images are brick red ochre. No case 30-31
of sequence could be resolved between images 164 and 106.
Figure 4: Multiple superposition from four different sets. Stippled images (63) and (46) 31-32
are in black and brick red ochre respectively. Solid areas are red ochre and
unshaded area is white.
Figure 5: The Harris matrix diagram showing the sequence of DKS painted images 32-33
(overleaf).
Figure 6: Finely stippled areas denote white, solid areas red and areas in large dots and 32-33
stipples indicate superposition. Unshaded areas are either faded or never had
paints applied to them.
Figure 7: Solid areas indicate red ochre. Unshaded geometric is brick red ochre. The 33-34
stippled image 77 is white; stippled image 78 is in red, but has been contrasted
with red handprint 209 underneath.
Map 1: The Elands Bay area in the west coast showing the location of DKS. 12-13
Matrix Diagram: The Harris matrices for Diepkloof Kraal Shelter sequence. 33-34

9. APPENDIX

Table 1: The list of images at Diepkloof Kraal Shelter and relations of and equivalence. 49-56
Table 2: The observed 65 cases of superpositions at Diepkloof Kraal Shelter site. 57
Table 3: The summary of information on sets and subsets of images at Diepkloof. 58
Table 4: Proportions in percentage terms of imagery involved in superpositions. 59

iv
CHAPTER ONE

INTRODUCTION

Motivation and objectives

Three main points form the basis upon which this project largely draws its impetus and
they are:

a.) the general lack of a systematic dating method for rock art, b.) the observations from
previous research in the southwestern Cape, and c.) the need for rock art preservation. As
will be explained below, these points have, for convenience, been condensed to the
following two main objectives of the project:

1. To carry out a systematic evaluation of the superpositioning of images at


Diepkloof Kraal Shelter site;

2. To map out the pattern of sequence of the paintings with a view to inferring
relative chronology; and,

3. To relate superpositioning to other general and compositional attributes.

The determination of the antiquity or age of rock art still remains largely a matter of
conjecture and circumstance. A chronometric method has yet to be developed which
would be regularly and widely utilised to accurately and reliably secure absolute or direct
dates for rock art, specifically rock paintings. The emphasis here is particularly on the rock
paintings because some progress in the dating of the other category of prehistoric rock art,
petroglyphs, has been made in the last few years (Bednarik 1979, 1992; Dorn 1990). The
difficulties with absolute chronology have been variously expressed in recent literature
(Thackeray 1983; Yates et al. 1993, 1994). Chaloupka notes that the dating of rock
paintings is problematic because the residual pigments as [Page 1 End]

1
[Page 2 Top]
found on the painted rock surfaces retain very tiny quantities of organic substances upon
which most conventional dating procedures are based (Chaloupka 1993).

Lorblanchet and his co-workers (1990) and McDonald and colleagues (1990) who have
worked in European and Australian contexts respectively, expressed reservations
regarding the accuracy in the application of AMS techniques on the direct dating of
organic materials in the pigments (Yates et al. 1993). In 1987, van der Merwe and his
colleagues reported the only radiocarbon date on an authentic rock painting, at that time,
in the Southern African context. A charcoal finger-painted human at Sonia’s upper (Site
Number BTJ 30) in the Boontjieskloof area was dated by radiocarbon AMS to an age of
500 + 140 years B.P. (OxA―515) (van der Merwe et al. 1987). This painting, being a
carbon residue, favourably lent itself to a dating procedure which requires tiny quantities
of sample material. Gillespie (1997) has argued that charcoal is a favoured material due to
desirable properties in it such as high carbon content and chemical stability. Regarding the
confidence in the above date the research authors caution that, being a single
determination it means very little by itself. The demonstration that rock paintings can be
dated in itself is quite significant. Support for this date, however, from more such
determinations is required to provide a more reliable body of evidence. In the same vein,
as science demands, the procedure should be replicable. As Bednarik (1992) notes there is
preference for techniques that could be standardized.

More recently dates have been obtained from paintings in the French Palaeolithic art from
suitable sample material. Jean Clottes and co-workers obtained dates for the Cosquer Cave
near Marseilles in France. Twelve determinations in 1992 (5 from charcoal in the cave
deposits and 7 from the paintings themselves) gave a range of 18, 500 BP to 27, 500 BP.
Ten more determinations are reported ranging from 14.050 + 180 (Gifa―96101) to 28,
370 + 440 (Gifa-96074) at the same site (Clottes et al. 1997). More determinations are
apparently in press.

This research project, therefore, seeks to investigate the Diepkloof Kraal Shelter site
(hereafter referred to as DKS) as a case study for the establishment of a temporal sequence
and relative age of the images at the site. This is necessary especially with the [Page 2
End]

2
[Page 3 Top]
above mentioned absence of direct dating methods for paintings. Analysis of the
succession of image-making at DKS with due consideration given to the compositional
associations of images (superpositions and juxtapositions) and a preliminary statement on
the different pigments is the major focus of this study. The use of Harris matrices, perhaps
better described as diagrams, in this study will not only permit the integration and
interrogation of different genres or assemblages of images at DKS, but will also allow for
the details of sequence of the panorama to be presented graphically. The stratigraphic
sequence of paintings at the site will show which images are earlier and which ones are
later in the succession, albeit with uncertainties of intervals between painting episodes, or
the nature of the temporal overlaps.

Various prehistoric research projects have been carried out in the southwestern Cape since
the 1950s (Johnson et al. 1959, 1963; Parkington 1976, 1981; Manhire and Parkington
1983; van Rijssen 1984; Yates et al. 1993, 1994). Particularly interesting for this study is
Parkington and colleagues frequent observation that in the southwestern Cape rock art,
sites where detailed fine-line paintings occur on the same panel with handprints (decorated
and/or undecorated), the former are always overlain by the latter, where preservation
allows superpositions to be detected and resolved. This observation is also attested by
separate research by Anthony Manhire (1981). Gavin Anderson (1991, 1996), on his
analysis of Andriesgrond paintings where there are also handprints, came to the same
conclusion. John Parkington and his colleagues, therefore, contrary to some opinions of
earlier investigators (Willcox 1959) concluded that handprints, always the uppermost in the
sequence, post-date the fine-line painting tradition (Yates et al. 1993). The point of
departure for this project, therefore, is this observation used here as the fundamental
working hypothesis. The project sets out to systematically test its applicability at DKS and
to build on it. Also important will be to map out the pattern of sequence of superpositions
of different painted images at the site with a possibility of using the outcome as a key
sequence for the southwestern Cape.

The field of rock art studies has a long history in southern Africa, presumably because of
the occurrence of this ancient San/Bushman art in great profusion and ubiquity in the sub-
region. South of the Limpopo River into South Africa, the first written/published reports
on the subject seem to be those by Ensign Beutler in the 1750s [Page 3 End]

3
[Page 4 Top]
(Deacon 1993) while in Zimbabwe the earliest reports were published in the 1890s
(Willcox 1984). Of particular note is that these early investigations tended to delve much
into the motivation and possible meaning of rock art to the extent that it tended to exclude
other interests. In keeping with this was (and still is) the investigation of the belief systems,
mythology and symbolism of the past San societies who are believed to have executed the
art, with of course the aid of ethnographic parallels from the extant San people within (in
recent years) the framework of the shamanistic hypothesis (Lewis-Williams 1981, 1984;
Lewis-Williams and Dowson 1989; Vinnicombe 1976; Parkington 1989). Recently some
rock art studies have focused attention on interpreting the paintings in terms of gender
and power relations deduced from sexual symbolism and related metaphors (Solomon
1989, 1994). A very impressive depth of knowledge is reflected in the extensive corpus of
information which has accumulated over the years on the subject. While rock art
researchers have for a long time realised the need for the conservation of the art their
analyses, however, have largely concentrated on developing methodological frameworks
for its interpretation rather than for its preservation.

Comparing rock paintings with engravings Rosenfeld (1987) concluded that the former are
generally more susceptible to weathering than the latter. Very early in this century
researchers in South Africa commented on the vulnerability of the rock paintings in the
wake of various deterioration factors (Stow 1905; Bleek 1932a). Since then, there has
gradually been a growing concern for rock art preservation (Rudner 1989).
Notwithstanding, only few isolated cases can be illustrated where that concern was really
transformed into practice. One such case was the removal of paintings from Linton and
sent to the South African Museum in 1910 (Deacon 1993) arguably for better preservation
there. Similar actions have been taken on a few sites near the Lesotho border in
northeastern Cape (Parkington: Pers. Comm.). Surface coatings and silicone preparations
have been used experimentally on paintings in South Africa, albeit unsuccessfully (Rudner
1989). Presumably, out of sheer ignorance and misguided efforts, some attempts elsewhere
aimed at conservation have culminated in total destruction of the paintings. The
Pomongwe site in Zimbabwe is a case in point where the total spalling and fading of the
pigments was initiated by the coating of paintings [Page 4 End]

4
[Page 5 Top]
with varnish in the 1960s (Garlake 1987). In recent years studies have accelerated in the
field of rock art preservation and management with pioneering work done in Australia
(Rosenfeld 1987). Janette Deacon has contributed a great deal in the conservation of rock
art in South Africa. In the early 1990s she worked with a team on a
recording/documentation programme of rock paintings in the Cedarberg area of the
southwestern Cape region and drew up guidelines for their long term conservation
(Deacon 1993). More work is still on-going under the National Monuments Council,
particularly in the area of graffiti removal from painted panels.

It has become an axiom in rock art conservation studies that the biogeochemical processes
acting on the painted rock surfaces need to be understood before preservation measures
can be undertaken. In the same vein, this study is partially a step in that direction with the
assessment of superpositioning and the implication on the preservation of pigments,
which will help understand the “paint-on-wall” situation.

A survey of previous research

The idea of analysing superpositions and manner of depiction of rock paintings and
formulating temporal sequences is not new. Superpositions have been variously
interpreted in past researches. David Lewis-Williams (1974) analysed superpositioning
from a different perspective from the present one of determining relative chronology, and
demonstrated that it was not a random phenomenon but a significant element of
shamanistic and ritualistic symbolism. In his quantitative study of the 2, 361 rock paintings
from thirty eight shelters in the Barkly East District, he used a card system, to record
superpositions with each case on a card, paying particular attention to the aspects of:
species, sex, techniques, dress, preservation, head type, weapons and equipment, elevation,
size and colour.

In his study, superpositioning meant a case where the upper painting is executed directly
and substantially on top of the lower painting while in cases where a small portion of a
painting marginally merges onto another it was construed as overlapping and therefore not
recorded. The reason for this was that the latter could easily result [Page 5 End]

5
[Page 6 Top]
from chance whereas superpositioning had to be an intentional or deliberate phenomenon,
in keeping with his interpretive methodological framework. The present study considers
overlapping as important cases of superpositioning, because regardless of whether they
were produced as a matter of chance or intention they bear chronological signatures
necessary for the construction of a temporal sequence. Lewis-Williams, however, has later
tended to dismiss such quantitative approaches which were initially thought to be the
answer in the elimination of subjectivity in analyses. He argues that the frequency of
occurrence of a motif “is no clue to its meaning’’ (Lewis-Williams 1990). As cited in
Anderson (1996), Lewis-Williams (1983, 1984a) also argues that rock art quantification can
result in downplaying metaphors in those representations.

Chains of superpositions were treated as thematic cases in contrast to Pager’s (1976)


scheme where such chains would be broken down into binary cases. Pager regarded chains
such as “eland on man on eland” as a binary case of “eland on man” and “man on eland”
and a composition of “one eland on three men” as three cases of “eland on man”. Lewis-
Williams’ scheme would treat the same as two thematic cases of “eland on man” and “man
on eland” and “eland on man” respectively (Lewis-Williams 1974: 94). Various categories
of subject matter were defined as well as recording the observed frequency of occurrence
in superpositions. Sixty-nine cases of two element superpositions were established with the
theme of “eland on eland” having the highest frequency of 25 cases, followed by “eland on
human figure” with 20 cases and 6 cases of “the buck on human figure”. In the final
analysis, it was observed that there were certain favoured combinations while others were
ignored. It was concluded, therefore, that superposition was an element deriving from a set
of rules, to be viewed not in instrumental terms ( as was previously common place) but as
a rendition of a value statement about the preferred subjects (e.g., eland and its “social and
economic significance” (Lewis-Williams 1974). This shows that his approach emphasizes
meaning more than chronology which can be deduced from superpositioning.

Vinnicombe (1976) analysed the Underberg paintings and attempted to reconstruct the
relative chronological sequence based on colour schemes, and techniques used in the
[Page 6 End]

6
[Page 7 Top]
execution of the images. Four phases were recognized from a large sample comprising 77
rock art sites. Here Phase One is dominated by animals as subject matter with no clear-cut
and definitive deduction of any colour scheme due to deterioration. Phase two comprises
animals and humans in mostly bichromes ― red ochre and white sometimes. Phase three is
represented by shaded polychromes of both animals and humans. The final phase was
observed as showing regression from earlier finely rendered depictions to colonial imagery
with a noted increase in the utilization of black and yellow ochre pigments (Vinnicombe
1976).

In the Southern African context, rock paintings have been studied and superpositions
analysed in attempts to determine periods in the development of the art and to build up
regional sequences, some based on colour schemes while some on technique and on
subject matter and manner (Stow 1905; Battiss 1939, Breuil 1930; Pager 1971, 1976;
Vinnicombe 1976, Cooke 1963). Replicability has been one of the major problems, with
some sequences only applicable to small localities. Schemes based on the subject-matter
and technique to establish clear-cut rock art developmental phases have been marred by
considerably overlaps between established categories.

In the 1970s and 1980s Chaloupka (1977, 1985) proposed a chronological sequence of the
Arnhem Land plateau rock art in Australia. From a survey of 1 400 sites various individual
art manners were identified and an attempt was made to establish the sequence of
superpositions. The identified manners were arranged into distinctive groupings which
were assigned to art periods and phases defined from known climatological,
geomorphological, archaeological and historical data and by zoological and botanical
evidence (Chaloupka 1985). Four art periods; the “Pre-Estuarine”, “Estuarine”,
“Freshwater” and “Contact” were established from the analysis. In defining his various
styles, Chaloupka drew from McCarthy (1958, 1962) and Maynard’s (1979) characteristics
of technique, form, colour, subject matter and size which they had used in their
classificatory schemes for recording and analysis of Australian rock art (Chaloupka
1985).[Page 7 End]

7
Figure 1: Diagram showing panel 2 superposition in 3 layers at Tandjesberg, Orange Free State
(Loubser 1993).
[Page 8 Top]
Chaloupka’s chronological sequence has been subjected to critical review. Particularly
queried was whether his manners of depiction are indeed distinctive entities occurring in
the putative order he proposed (Lewis 1988, quoted in Chippindale and Tacon 1993).
Lewis is also cited as having expressed reservations on the possibility of constructing
chronology from superpositions due to the multiplicity of obstacles involved in observing
sequence. In recent years, Chippindale and Tacon (1993) analysed the sequence of
superpositions of the Kungurrul and Brockman panels in the Arnhem Land, which
together comprise of over 300 individual images. While their study is an extension of the
existing large corpus of information on the Arnhem Land paintings which has
accumulated over several decades, it is particularly pioneering and significant in the
considerable creativity shown in their methodological approach.

In the synthesis of the superpositioning sequence, Harris Matrices were utilised for the
first time to order sequence in rock art (Chippindale and Tacon 1994). Two broad
divisions of art were defined ― the “old” and “new”, where the former encompasses
Chaloupka’s manners of depiction in the Pre-Estuarine period while the latter entails
manners in the Estuarine, Freshwater, and Contact periods. Various classes of images were
established from analysing the characteristics of technique, manner and pigment used for
individual paintings. A chronostratigraphic sequence was constructed based on observed
patterns of superpositions between images in different classes.

In South Africa in the Orange Free State, Johannes Loubser (1993) analysed the rock art
of a shelter in the Tandjesberg farm. A total of 530 motifs was analysed 407 of which were
superimposed. Similar observations to Lewis-Williams’s (1974) conclusions were drawn.
Human figures and eland are argued to have been “undoubtedly superimposed according
to some prescriptions and rules” (Loubser 1993: 350). Individual images were assigned
numbers and panel 2 of the eleven panels he defined had a diagrammatic representation of
stratigraphic relationship of motifs because it is the most detailed panel in the shelter.
Loubser constructed a “generalized” three tiered sequence of motifs (Fig. 1). [Page 8
End]

8
[Page 9 Top]
The earliest generalised motifs at the bottom of the sequence consisted of faint red
remains of a ‘bovid’ (antelope) covered by a maroon painting of a human figure with the
back of a bovid-like animal” (Loubser 1993: 354). Three shaded red ochre and white eland
followed immediately above in the sequence. The top or latest layer has “three elephants
and shaded ochre or red human figures with bows and arrows” (Loubser 1993). He argues
that the time lapse between the layers is uncertain but that they all could have been
produced by the same individual on the basis of similarities exhibited. Some associated
white figures are thought to be of later artists. The whole sample of images is in the fine
line manner. The observed layers of paintings are interpreted in the framework of the
shamanistic hypothesis. Some animal motifs are interpreted as rain animals being
controlled by associated human figures.

In the southwestern Cape superpositions have been dealt with and commented upon
(Yates et al. 1993, 1994, Manhire 1981; Anderson 1991, 1996) but have not received
treatment in great depth. Observations, however, contrary to Lewis-Williams’s (1974)
conclusions on superpositions in the Drakensberg region, show that the south western
Cape superpositions of handprints on detailed fine-line images “do not appear to be the
result of ritual determination rather than chronology of production per se” (Yates et al.
1993: 61). There is no definitive evidence to suggest ritual symbolism although the
possibility can not be discounted. This in part reflects a difference in approach. Three
broad categories of rock paintings have been tentatively established (Yates et al. 1993,
1994). These sets appear to fall within different temporal frames. This project will, it is
hoped, show if these are really distinct categories representing coherent blocks of time
rather than having certain images occurring in and out of given sets.

The detailed fine line paintings are clearly in the majority of the rock art in the
southwestern Cape region (Yates et al. 1993). This set is rendered in great detail with finer
definition of outlines, and usually infilled with ochreous wash. The skilful rendition of
images in this set suggests the use of a brush or some similar fine instrument as applicator.
A range of ochreous reds and yellows were used, while fugitive pigments like white and
black which appear to have been used quite often have largely succumbed to deterioration
agencies. It is believed that this set is linked to a [Page 9 End]

9
[Page 10 Top]
predominantly hunter-gatherer life-style and that this shows evidence of shamanistic ritual
symbolism because of the depicted subject matter (animals and humans shown in various
scenes and activities) (Yates et al. 1993). In most sites in the southwestern Cape, fine-line
paintings are overlain by all subsequent sets, suggesting that it is the earliest rock art
category in the region.

Handprints, technically not paintings in the true sense of the term, but which are positive
images of hands (left and right) are another set occurring widely in the study area as
decorated (usually nested curves or U-shapes) or undecorated (plain) forms (Figs 3 & 7).
On visiting the DKS site, Chippindale suggested these could be paintings of hands or
palms rather than prints, which is plausible enough. On closer examination of most
individual images of hands, however, there appears to be persuasive evidence to suggest
that they really are prints. A substantial number of these images have in sufficient detail
the inner architecture of the hand, with finer lines or prints of the skin clearly visible.

Suffice it to say that they are handprints, probably produced by dipping or placing the
hand into the wet pigment and then pressing it against the rock surface. The decorated
forms appear to have been produced by scraping off a desired pattern on an already paint
smeared palm/hand prior to imprinting (Yates et al. 1994). Alternatively, a design may have
been painted on the hand prior to offsetting or imprinting it on the rock face (Chippindale
1997). Handprints are almost always executed in ochreous reds, although in other sites like
the Elands Bay Cave yellow pigment was used. There is perhaps one recorded exception
of a black handprint although the site details and location are not provided (Yates et al.
1994). The spatial distribution of handprints was earlier shown to be concentrated in the
sandveld (sand country), but recent investigations show that similar proportions also occur
in the mountains (Manhire: Pers. Comm.).

The third category comprises images of colonial character. While they are commonly of
cruder rendition, possibly produced using the finger, at other sites like Stompiesfontein
(Site Number STK1) some appear in the form of refined and detailed representations.
Technically, crudeness distinguishes this set from the detailed fine line [Page 10 End]

10
[Page 11 Top]
representational paintings, as well as the colonially-related subject matter. Images of
colonial material culture include wide brimmed hats, high-heeled shoes, smoking pipes
(DKS Site), ships (Heidedal) wagons, mounted men, women in dresses, rifles/muskets
(Stompiesfontein) and quite often indeterminate animal species. There are, however, few
sites with paintings representing this set. With the exception of Diepkloof Kraal Shelter,
Porterville ‘galleon’, Klipfonteinrand, and Putslaagte 44 (Anderson 1996) sites with
paintings depicting colonial material culture are restricted to the Bokkeveld (Yates et al.
1993). Occurring in association with this category are a variety of geometric designs
ranging from gridded rectangular and squared forms to circular images usually with a cross
inside. They are largely finger painted in the Sandveld and the Bokkeveld, but at
Stompiesfontein and in some sites in the Cedarberg more finely rendered examples are
found (Anderson 1996).

Another interesting category can be added onto the above three sets. It consists of names
or signatures of individuals and/or dates and single continuous sinuous and looped lines
sometimes forming recursive circular shapes. These are also schematic representation and
imitative paintings and/or drawings of the earlier fine-line representational images. A
variety of media were used in the execution of images in this set. Unlike for prehistoric
rock art where it is not often possible to distinguish paintings (use of wet pigments) from
drawings (pigment applied dry) (Bednarik 1992), drawings in this set can be easily
identified from the media used like charcoal crayons, commercial paint, scratching
ochreous rocks on the cave/shelter face and other such materials. While these lines and
images were initially considered to be vandalism, their consistency does not warrant the
lumping them together under the term “graffiti” which is defined in chapter five. They are
part of a painting tradition (Anderson 1996).

The different categories of images discussed above can be tentatively assigned to a broad
chronological frame. Congruity between the observed categories and the temporal frame is
dependent, though not tightly, upon subject matter, technique, pigments/media and the
correlative relationship which some genres appear to have with entities that have been
established in the archaeological sequence and by environmental evidence. The category
with names and signatures of individuals is self-evident as it is [Page 11 End]

11
[Page 12 Top]
mostly accompanied by dates. Within a range of media used in this category commercial
paint and pencils can be easily identified and all this attests to them having been produced
in recent years. Colonial era imagery points to a terminus post quem for its development, by
which time the western material culture, such as depicted, began to infiltrate into the
southwestern Cape and indeed the whole of South Africa.

The detailed fine line representational set discussed above as having been mainly produced
by hunter-gatherer societies is argued to be the earliest to develop (Yates et al. 1993). It
must be realised that there could have been substantial overlaps in terms of chronology of
production between different sets, which complicates the possibility of inferring clear-cut
developmental stages and duration of intervals, if indeed they were any between painting
episodes. The Harris matrices as an analysis tool enables us to interactively check such
complexities usually manifested in the form of inconsistent or contradictory relations,
where superpositions can be detected.

As a relevant study Anderson (1996) analysed a sample from the Bokkeveld paintings and
attempted to demonstrate that finger paintings (colonial imagery, finger smears, dots)
differ in context and manner from fine line images. Three categories of images; fine line
images, finger paintings and colonial images were isolated and analysed. Using a statistical
procedure (Chi-squared test) specific depictions were compared in order to show which
images were often reused more than others and to demonstrate differences between the
fine line and finger painted images. Superpositioning, smearing, repainting, and rubbing
considered to be categories of reuse are suggested to also reveal the authorship of the art.
The relationship between the three image categories can be understood from their relative
time sequence (Anderson 1996). Chronology of image production is extended to
demonstrate differences between the paintings done by the San hunter-gatherers and those
by the Khoe herders. The chronology that Anderson observed will be briefly discussed in
Chapter Five. [Page 12 End]

12
Map 1: The Eland Bay area in the west coast of South Africa showing the location of DKS
(Diepkloof Kraal Shelter) site near the Verlorenvlei marshes.
CHAPTER TWO

ENVIRONMENTAL AND RESEARCH CONTEXT

Diepkloof Kraal Shelter (DKS)

Diepkloof is a high and freestanding sandstone kopje on the southern bank of the
Verlorenvlei, a coastal estuarine lake, river and reed-swamp system (Meadows et al. 1996).
This landform stands about 120 metres above the Verlorenvlei stream level (Parkington
1977b). Diepkloof comprises two large overhangs next to one another. The one facing the
northeast has deep occupation deposits covering an area approximately 200 m2
(Parkington and Poggenpoel 1987). In this shelter there are few paintings. The other
shelter facing the southeast, which is the study site for this project, is approximately 20
metres wide, + 15 metres high at the front and the sandy deposit is deep in parts (Site
record forms 1981). The front of this shelter has small to medium sized quartzite boulders.
The two shelters are also known as DK1 and DK2 respectively but the present study uses
the acronym DKS for the latter. This shelter is substantially decorated from side to side
with numerous paintings of various manners executed mostly in ochreous reds, although
this is not immediately obvious on arrival at the site. The Diepkloof site commands a
superb panoramic view across the Verlorenvlei and its precincts towards the visible
horizon of the sandveld landscape. The site is 18 Km southeast of Elands Bay or Cape
Deseada (Parkington 1977b) and around 180 Km from Cape Town.

The DKS site (Map 1) is chosen in this project as a case study for the following three
reasons:

a.) The site has paintings of various manners ranging from detailed fine line images,
through handprints to images of colonial character and names or signatures of people.
[Page 13 End]

13
[Page 14 Top]
b.) The site exhibits a fairly good number of images in different sets which are involved in
a sequence of superpositioning particularly between handprints and fine line imagery.

c.) The site provides an enormous potential for the establishment of relative chronology as
a key sequence for the area, and,

d.) The paintings are structured in a manner that makes the analysis easy (lines of
handprints and eland).

The southwestern Cape and indeed everywhere else in South Africa suffers the dearth of
sites bearing colonial imagery which, where they occur, are often dispersed. The
occurrence and distributions of colonial imagery and hand-printing, unlike the fine line set
and signatures, are spatially restricted to certain isolated geographical locations (Anderson
1996) and these two sets do not always coincide in a single panel. DKS, therefore, presents
an opportunity to construct a temporal sequence of the development of the art, involving
various sets which exhibit a fairly reasonable stylistic variation of imagery. It is hoped that
the DKS sample is sufficient to enable the establishment of meaningful superpositional
and juxtapositional patterns from which to draw a relative chronological sequence.
Superpositioning between various sets is recognisable at the site, which is an advantage.

Environmental context of DKS

Geologically, the Diepkloof site lies within an expanse of lowland coastal plain called the
Strandveld or sandveld. The sandveld consists of the Tertiary and Quaternary white to
slightly red sands capping a gently undulating landscape with large stretches of the area
having elevations of less than 70 metres. While most sandhills and ridges have been
stabilised by the vegetation cover, the tenacious dunal topography has encroached upon
large expanses of the terrain. The monotony of the sandveld landscape is interrupted
intermittently by angular quartzite geomorphological landforms, ‘outliers of the mountain
system to the east’ (Talbot 1947, in Parkington 1977b), such as the Table Mountain
sandstone outcrops along the Verlorenvlei. [Page 14 End]

14
[Page 15 Top]
The Cape Fold Belt forms the eastern boundary to the sandveld. This mountainous belt is
made up of a quartzite rocks, other fine-grained rocks of the Malmesbury Group and
shales of the Klipheuwel Formation (Verlorenvlei (BW 13) 1986). The geological process
of folding and the subsequent fluvial etching transformed this Cape system into a very
irregular and steep topography of parallel, north-south aligned rocks with highest peaks
attaining attitudes of over 2000 metres in parts. To the west, the colossal water body of the
Atlantic Ocean forms the boundary. The drainage of the area comprises an 87 Km long
catchment from the coast towards the inland “aligned along a north-west/south east
structural trend” (Miller 1987, in Meadows et al. 1996).

The Verlorenvlei catchment area originates in the Piketberg and the Olifants River
Mountains and covers an area of 1 890 Km2 (Meadows et al. 1996). The system has many
watercourses and streams traversing east-west across the coastal foreland. The
Verlorenvlei river which culminates in an open water lake that extends for 13 Km and
attains a maximum width of 1, 5 Km, linking the former with the sea. Water flows during
the wet season in winter roughly between April and September and the water courses dry
up in summer. The Verlorenvlei forms a sand-bar during reduced water flow in the dry
season and is only occasionally broken by winter flooding or high spring tides (Meadows et
al. 1996). This pattern has led to the formation of an estuary, largely freshwater in nature;
but with salinity increasing towards the mouth (Parkington 1977b). West of Diepkloof, the
Verlorenvlei is reed-laden, forming an estuarine ecology supporting a diverse marine
(aquatic) avian fauna. A variety of other aquatic floral species occur in this reed-swamp
system, which is part of the Cape Floristic Region (Meadows et al. 1996) which is part of
the strandveld.

The climate of the sandveld presents a semi-arid to sub-arid image with very little rainfall, a
situation exacerbated by the porous surface sediments which allow for rapid surface water
drainage. While the frequency and intensity of precipitation can be quite variable from year
to year, annual average rainfall figures range below 300 mm yr-1, of which between 70-80%
is received during winter in this Mediterranean type of climate. Temperatures in the
summer season are very high subsequently leading to high [Page 15 End]

15
[Page 16 Top]
evaporative losses from the watercourses and inland basins like the Verlorenvlei system
which induce saline conditions. The Cape Fold Belt in the east forms a zone of high
precipitation with rivers that flow for most of the year.

The semi-arid climate of the Sandveld supports a remarkably diverse flora of


predominantly drought deciduous vegetation of typically hardy shrubs ‘with leathery
leaves, often containing resin or oil’ (Parkington 1977b), geophytes and grasses and,
especially succulents such as Euphorbiaceae and Mesembryanthemaceae (Meadows et al.
1996). This is known as strandveld but also referred to as the sandveld fynbos, xeric
fynbos, in places. The Lowland fynbos, adapted to the deep and sour sandy soils deriving
from the rocks of the Cape system is dominated by Restionaceae, drought-resistant
Proteaceae and other sclerophyllous fynbos shrubs, such as Ericaceae, Iridaceae and
Gramineae. Also present is the dry mountain fynbos typified by small-leaved Cape macchia
favouring sandstone outcrops such as the Diepkloof. Forest vegetation is circumscribed in
deeper valleys and protacted revines towards the mountains (Wicht 1945, quoted in
Parkington 1977b), where elements comprising Olea europaea africana subspecies, Maytenus
oleoides and Podocorpus elongatus can be identified (Meadows et al. 1996)

General research context of DKS

In a research programme designed to investigate the large Diepkloof shelter in order to


infer seasonal patterns of resource utilization in the Western Cape, Parkington and
Poggenpoel excavated the site in 1973. The Middle Stone Age was demonstrated to be
deep and dating of charcoal from near the base of the test excavations gave an age of
about 45, 270 years (Pta 1, 054). Carbon dates were obtained for various levels within the
deposit with the youngest being around 390 + 30 (Pta 1, 055) for grass bedding towards
the top of the sequence from the rear of the shelter. The MSA lithic assemblages which are
characteristic of the Howieson’s Poort were found and mostly utilizing silcrete (Parkington
and Poggenpoel 1987). A note is made that Diepkloof exhibited very few LSA (Later
Stone Age) formal lithic tools. Finely polished and shaped bone points were recovered.
[Page 16 End]

16
[Page 17 Top]
The faunal assemblage is represented by micro-faunal material from moles and rodents
argued to be non-human in origin, the domestic sheep bones (teeth and post-cranial
material) indicating considerable access to herded stock. Two charred sheep bones were
associated with the 1, 590 + 85 BP carbon date (Parkington and Poggenpoel 1987). Small
ground game comprising tortoise, dassie (rock hyrax), grysbok, steenbok and some small
carnivores were recovered in significant proportions. A range of marine resources
(animals) recovered from the deposits suggested substantial utilization of marine foods.
Plant remains, well preserved within the grass bedding, were quite variable and substantial
although the sampling and interpretation of the modes through which these were
accumulated has presented problems (Parkington and Poggenpoel 1987). While the
Diepkloof Shelter sample has been argued to demonstrate reliance on both animal and
plant foods, the problem has been in the assessment of the relative contribution of each of
these sources. Evidence of settlement suggests there was repeated occupation at Diepkloof
with a hiatus of about 30, 000 years until the last 2, 000 years of the Holocene when
occupation resumed. The site is argued to have been visited by the San hunter-gatherers
who had a close relationship with herders or pastoralists.

In the past few years there has been a significant development in the study of the late
Quaternary palaeo-environments of the southwestern Cape region. Multidisciplinary
approaches are being applied in the reconstruction and interpretation of palaeo-ecological
information of the late Pleistocene and Holocene periods (Meadows et al. 1996).
Consequently, there has been intensification of the search for sites bearing undisturbed
and suitable fossil pollen sediments in the lowlands of the sandveld area. For many years
this focus has been at the backwaters in such palaeo-environmental studies. The focus of
attention has been the mountains where relatively humid climates have ensured the
accumulation of substantial sequences of organic sediments. Also to be noted, perhaps, are
the analyses of wood charcoal evidence (Cartwright and Parkington 1997) and micro-
mammalian remains (Avery 1990) in reconstructing and interpreting palaeo-environments
and results of more such applications are apparently in press. [Page 17 End]

17
[Page 18 Top]
Grootdrift farm site, on the lower south-eastern tip of the Verlorenvlei estuarine-lake has
been sampled for fossil pollen preserved in the sediments (Meadows et al. 1996).
Investigations included palynology, sedimentology, and geochemistry of the accumulated
organic sediments retrieved from several cores of varying depths. Considerable
information has been gleaned from the above analyses despite the evidence of either the
cessation of sedimentation or subsequent erosion of sediments which had accumulated,
after 4, 300 BP in the sample area. The environment/ecology of the Verlorenvlei and its
immediate precincts is shown to have been largely influenced at different stages in the
Holocene by a “complex interplay of changes in moisture availability, sea-level
(fluctuations) and degree of anthropogenic activity” (Meadows et al. 1996: 93)

Fossil pollen analysis of three cores (GDV1, GDV2, and GDV4), anticipated to produce
reliable models of past vegetation conditions demonstrated a significant influence of sea-
level fluctuation on sedimentation and fossil pollen accumulation in the latter part of the
Holocene (Meadows et. al, 1996). The Verlorenvlei hydrological shifts were between the
fresh-water/terrestrial and hypersaline-estuarine/marine conditions. The mid-Holocene
coincided with the rise in sea-levels of about 2.8 metres (c. 4, 000 BP) and 3 metres (c. 5,
000 BP) (Meadows et al. 1996; Miller et al 1993; Yates et al. 1986). The pollen spectra in
these areas indicates the abundance and dominance of Poaceae (grasses) more than the
present distributions suggest, as well as significant proportions of Chenopodiaceae and
Amaranthaceae pollen in the sequence. The species decrease significantly upwards the
sequence more visible especially in core GDV4 where a 50% decline is indicated. The
decrease is suggested to be a consequence of human impact on the landscape by way of
overgrazing in the much latter periods of the Holocene (Meadows et al. 1996).

Anthropogenic influences on the Verlorenvlei environs are also indicated by the presence
of alien floral species in the sediments. Included are the Pinus spp. and annual or weedy
elements like the Scrophulariaceae, Oxalidaceae and Stoebe - type” (Meadows et al. 1996),
while the GVD4 core representing the period circa 1, 600 AD to the present is marked by
an increase of Asteraceae, Acacia as well as Pinus spp. toward the top. [Page 18 End]

18
[Page 19 Top]
Human disturbance on the landscape increased significantly in this period by way of agro-
pastoralism; overgrazing impact, altered fire regime and large scale agricultural clearing of
the indigenous vegetation.

Several cave sites including Diepkloof in the vicinity of Verlorenvlei and Elands Bay Cave
further afield to the north are argued to have been occupied sporadically by hunter-
gatherer communities who had access to domesticated stock, in the Holocene (Parkington
1977b). This episodic occupation of sites in the sandveld biome was consistent with their
seasonal food procurement strategies, themselves partially dictated by the periodic pulsing
of the Verlorenvlei environments between high sea-level saline estuarine conditions and
the open fresh water/lacustrine hydrology. The palynology studies of the sedimentary
sequence at Grootdrift provide a background against which successive modes of human
occupation of the area have occurred (Meadows et al. 1996). The high sea-level during the
mid-Holocene possibly induced unpalatable habitation conditions in the Verlorenvlei
environs which were not particularly attractive to human occupation. This period is
marked at Elands Bay Cave by a hiatus in the sequence and a remarkable absence of
evidence of human occupation in the area. [Page 19 End]

19
Figure 2: The Harris matrix method. Three recognised relationships (A, B and C) between
stratification units. I, II, and III show how the diagram is generated (Harris 1989).
CHAPTER THREE

METHODOLOGY

Background to the Harris matrix system

The Harris matrix system was developed to resolve complex stratigraphic relationships of
depositional units in archaeological sites into relative sequential order. It has only been
recently applied to the analysis and recording of superpositioning sequences in the rock art
(Chippindale and Tacon 1993, Loubser 1993; Anderson 1996). The Harris matrices can be
better described as a diagrammatic format for expressing stratigraphic sequences and their
relationships. The term “Harris matrices” does not carry any mathematical connotations
whatsoever. It is important to note that the Harris matrix methodology is applicable to any
units of stratification whatever their thickness, from larger units of several meters deep in
archaeological sites to densely overpainted rock faces with pigments therein measuring to
only a few microns or millimetres thick.

The Harris matrix formalism recognises three possible relationships between two given
strata (Harris 1989). Two strata may not have any direct stratigraphic relationship or may
be in superposition (above/below relationship) or finally, may only be correlated (equated
by the “=” sign) as separate parts of a once whole deposit or image (Fig. 2). The Harris
diagram, therefore, was devised to show such superpositional and juxtapositional
relationships schematically. Regarding relations of equivalence, instead of correlating
separate parts of a once whole unit in that relation, for rock art analysis single and discrete
images are equated with one another on the basis of defined criteria discussed below.
Separated portions of a once whole image, where it can be established, are identified as a
single motif. [Page 20 End]

20
[Page 21 Top]
Application to DKS rock art analysis

The underlying tenet in the analysis of superpositional sequences in rock art is that images,
as seen today, accumulated on the rock substrate in phases. It is hoped that the Harris
matrix methodology will provide the means with which to isolate the phases that are there.
The hypothesis is that every observed set in the sequence represents a new phase in the
development of the painting tradition through time.

The DKS study shows some deviations from earlier studies in the manner in which
relations between different imagery are analysed. DKS exhibits sets and subsets of imagery
(Table 3) which are categories of images exhibiting certain common compositional and
individual attributes. As shown in Chapter one, there are signatures and dates, colonial era
imagery, hand prints and fine line imagery. There are some groups of depictions which can
be regarded as a single composition. A line of 12 eland at DKS can be argued to be a
single composition on the basis of very similar, if not the same, pigments same scale, in a
single file or procession, same manner of depiction and technique. On the analysis of
sequence using the Harris matrix method, Chippindale and Tacon argue that:

“When distinct figures appear to belong together, as in a pair of images


which make up a set of macropod footprints, or as in a line of adjacent
human figures closely similar in subject, manner of depiction and
pigment, we have allocated them different numbers. These affinities
need to be taken note of. To indicate they are equivalent to each other
in the Harris matrix formalism, we have added relations of
equivalence.” (Chippindale & Tacon: 1993: 35)

There seems to be emphasis on deducing equivalence in those circumstances where images


form a single composition. At DKS this can easily be applied to the line of eland as shown
above some portions in the different lines of handprints. This approach seems to exclude
those images which are not adjacent or juxtaposed with one another, or not in the same
line or procession but showing common attributes. At DKS, the colonial [Page 21 End]

21
[Page 22 Top]
era imagery is characterised by mostly individual or separate images scattered across the
painted surface. Despite their placement on different parts of the painted area, they form a
good coherent set on the basis of pigment, manner of depiction and technique and
humans in this set are not significantly different in scale. There is no evidence to suggest
that these images form a single composition. This study construes these to be sufficiently
involved in relations of equivalence. However, motifs are regarded as equivalent if they
depict the same subject matter; that is, humans, geometrics and animals. This is not to say
there is anything wrong with focusing on compositional groups, but it is to expand the
approach in order to accommodate certain circumstances. Regarding signatures and dates,
I assigned them single numbers, whatever their length, if they appear in the same material
and colour and same manner of writing. Similarly, crayon lines which are often disjointed
or discontinuous, but on the basis of colour and size of line, have been assigned a single
number.

The various sets discussed in Chapter One into which various distinctive images are
classified appear to be coherent categories. The criteria according to which the different
sets were defined include; pigment/colour, subject/motif, technique, manner of depiction,
size/scale, and for some sets of images I included their placement in relation to one
another as when in a procession or in the same line. An individual image does not
necessarily have to satisfy all the listed criteria, but it must satisfy at least three in order to
qualify into a given set. These are not water tight divisions or groupings. Images may have
been executed in different pigments or may have varying shades of the same pigment
(either an intended effect by the ancient artist or due to differential preservation), but on
the basis other criteria like technique, manner, scale, motif and placement can be
reasonably related, under a single category.

In this study, relations involving a painting executed on top of another or marginally


impinging upon another (overlapping) define superposition. The correspondence between
such images is expressed in the form of “relations of equivalence” and are therefore
equated by the “=” sign in the Harris diagram. This entails establishing contemporaneity
between various images and demonstrates if indeed they are a coherent category to be
placed in a single developmental episode. The images that are [Page 22 End]

22
[Page 23 Top]
not involved in stratigraphical associations with others tell very little in terms of the
sequential order of the paintings and establishment of relative chronology. Their position
in the sequence, however, can be established from assumptions derived from those they
are in equivalence with, themselves involved in superposition with images of other sets.
That is, their temporal frame is inferred from those in the set which have clear
superpositional associations with images in other sets. This is the situation with colonial
imagery at DKS which has only one case of superposition with a handprint and occurs
spatially sparse across the rock substrate.

Intriguing in the context of this study is how superposition can be approached within a
single set like the fine line imagery or handprints. I am inclined to believe that at DKS
these superpositions within sets have limited potential in constructing relative chronology
partly due to the very small nature of the sample. An eland in a procession or handprint in
a line may partially impinge upon another in the succession; this merely being incidental or
the result of chance as, may be, the artist, possibly, was not worried about precision in the
placement of the images in relation to one another. Earlier reconstructions of art
sequences appealed to stylistic variations in the motifs in addition to colour schemes. But
if the motifs happen to be the same while occurring in observable superpositions this
limits the possibility of reconstructing a revealing chronological sequence. Rosenfeld and
Smith (1997: 408) have argued elsewhere that a stylistic analysis directed at chronology
requires the isolation of traits that can be given temporal value. Conceivably, within a
single set that exhibits a high degree of stylistic coherence one may have problems in
trying to isolate traits that have chronological signatures. Where good preservation allows,
there could be temporal patterns within individual sets, like the fine line tradition, and the
Harris matrix method may be very useful in revealing such relations.

In keeping with the methodology adopted in this study, such compositional associations
will have each image assigned a number as an identifier and subsequently receiving
separate consideration in the Harris matrix diagram. It seems very unlikely that any
meaningful temporal sequential phases could be discerned from within-set
superpositionings. Temporal sequences and phases of the development of the art have
[Page 23 End]

23
[Page 24 Top]
been approached elsewhere (Vinnicombe 1976) within the fine line tradition by definition
and isolation of distinctive colour schemes and style. As hinted above, however, this would
require a substantial sample; large enough to demonstrate repetitive frequencies of certain
associations and patterns on which to construct meaningful chronological patterns.

Fieldwork at DKS site

Three separate field trips were conducted to the DKS site, the first of which was to
introduce myself to the site and the paintings therein. While many colour slide
photographs were taken of the panel during the trip, only a few were scaled which meant
that they were not very useful for reproduction purposes. My trial exercise on tracing and
construction of the painting sequence using Harris matrices was based on those slides. The
line of eland superimposed by handprints on the far left section of the panel (Fig. 6) was
traced and all the images were numbered. It was apparent that the more one scrutinised
the slides during tracing, the more one recognised the details not seen at the site. It must
be stressed though that it is only after the second or better still, the third close examination
of the actual paintings on site that further meaningful details can be discovered. The copies
of tracings were taken back to the site for the verification of their correctness as a trial
attempt. It was clear that this is a tiresome task which exerts considerable physical stress
on the body and eyes.

On the second field trip, a series of scaled colour slide photographs was systematically
taken from side-to-side of the DKS panel. As a result of poor lighting conditions at the
time of the day they were taken, most slides did not come out clearly. They were usable;
however, as the reproduction exercise was based on this collection of slides. In the
workroom at the university, the tracings were reproduced from slides at 40% of the
original size onto 1.3 metre wide soft newsprint paper roll cut into 2-metre lengths for
manageability. On transferring the photographic mosaic capturing the DKS images onto
paper, sufficient overlaps were created in order to allow for accurate overlay of images
from one slide to the next in the successive order they were taken. This enabled [Page 24
End]

24
[Page 25 Top]
the construction of a continuous mosaic of tracings showing varying compositional
associations of the DKS images from side to side of the whole site.

The rationale behind using tracing as a method was that a near accurate reproduction of
the paintings would allow for an almost accurate integration and investigation of
superpositional and juxtapositional relationships between various distinctive images in the
panel in an exhaustive way. In the analysis of the old Arnhem Land art, Chippindale and
Tacon (1993) used soft pencil and soft drawing paper to draw the paintings from colour
slide photographs; they made use of field sketches and notes to ensure a reasonable and
acceptable level of accuracy. Varying strengths of pigment and discernible differences
within areas of overlap were indicated using a range of pencil tone. I believe that my
tracing method is a “soft technique” sufficient to convey varying tones and strengths of
pigments if carefully and skilfully executed. Tracings also allow the tracer to closely and
actively interact with the paintings thereby enhancing greater understanding of their
production in the past and to discover what was painted.

Areas of superpositioning and other meaningful detail are not always immediately obvious
the first time at the site. A closer examination of the slides during tracing enables their
detection, and can then be verified on the subsequent re-examination of the panel. As has
been discussed elsewhere, the importance of tracing as a method cannot be
overemphasised:

Far from being a mechanical documentation of ‘facts’, tracing is a form


of analysis that is more than merely descriptive. (Lewis-Williams 1990:
127)

Admittedly, tracing is an arduous task which places considerable strain on the eyes; it
demands patience and painstakingness. Tracing has shortcomings in the form of human
and instrumental imperfections always at play in reducing the level of accuracy. Even a
skilled, meticulous and experienced tracer cannot completely eliminate the inaccuracies,
but can only keep them to a minimum. To begin with, photography on which tracing or
drawing can be based is not a ‘complete and objective’ documentation method from [Page
25 End]

25
[Page 26 Top]
which to make reproductions. I will not labour to discuss the limitations of photography as
this has been discussed exhaustively over the years regarding photographic recording of
rock art (Vinnicombe 1960; Schoonraad 1965, in Lewis-Williams 1990). Colour
photography is ineffective in cases where the deterioration agencies have left us very faint
silhouettes of this formerly radiant ancient art. It has been argued elsewhere that, “colour
photography, however good, does not convey the faint markings that are part of any
palimpsest on a much-painted panel” (Chippindale and Tacon 1993: 34). In the same vein,
Vinnicombe (1960) notes the impossibility to tell whether a colour in the photograph is
faded paint or a natural rock discoloration. The principal advantages of photography
cannot be underestimated, however, and despite its inevitable compacting effect on any
relationships within painted images, it has been quite useful in the identification of all
those areas of the site where overlapping images occurred. Even tracing from the rock face
is no way ‘complete and objective’ reproduction as it involves reducing a surface
characterised, in almost all instances, by irregularities, unevenness and convolutions onto a
flat sheet of paper.

One mechanical problem encountered during the tracing exercise was that the slide
projector was constantly altering the scale of images each time the slides were run through.
This introduced discrepancies of varying magnitudes in matching up overlaps from one
side to the next. The images meant for overlap to ensure continuity in the mosaic from
side to side of the panel could not directly overlie one another. As a solution to this
problem, the projector was constantly adjusted and shifted forwards, backwards, and
sideways so that images meant for overlap finally had their edges within, at most, between
20-25 mm of each other. This method of distributing the error proved effective in
ensuring continuity, without much distortion. Close-up slides of some important
compositional associations of the panel were traced to scale in the addition of finer details
to those sections of the mosaic they represented.

All the reproduced images were numbered for identification purposes in the analysis. For
practical purposes, as Loubser (1993) notes this ensures that motifs are not counted more
than once or left out and also simplifies the generation of Harris diagrams. No particular
order in the numbering was followed, although images in a procession or a [Page 26 End]

26
[Page 27 Top]
line could easily be assigned numbers in some continuous fashion in their successive order.
This was the case with the only line of eland in the site and the handprints. The names and
signatures as well as their related dates are assigned single numbers rather than having each
and every letter or numerical character given a number. The crayon lines, whether
continuous or discontinuous, are encompassed into a single image with a single number on
the basis of colour, material used, style, and manner. In cases where images have been
rendered discontinuous due to weathering processes they are assigned a single number, if
indeed it can be established that the two or more disjointed parts originally formed one
whole image.

This study also involved tabulation of the DKS information. Three tables were produced;
one showing all the recorded images at DKS, their relations of sequence and of
equivalence and other attributes (Table 1). Table 2 was generated as a summary of the
relations of sequence at DKS which has been useful in the reconstruction of the sequence.
The final Table 3 also gives summary information on the sets and subsets of imagery at
DKS. All the fine line images are designated as Set “A”, handprints as Set “B”, colonial era
imagery as Set “C” and finally, signatures and plates as Set “D”. Some of these sets are
further broken down into subsets. [Page 27 End]

27
CHAPTER FOUR

RESULTS AND ANALYSIS OF THE SUPERPOSITIONING SEQUENCE

Altogether 219 images including smudges and smears (10 in number) and 1 splash were
recorded and analysed at DKS (Table 1). Many more faded ochre images and later
crayon markings and drawings were excluded from the analysis because they are not
involved in superpositioning in any recognisable or useful way. Adopting a scheme used
by Chippindale and Tacon (1993) for the Kungurrul and Brockman panels in Arnhem
Land, I have characterised each individual image in terms of traits such as: manner in
which it was depicted, technique and pigments used. These are particularly important
diagnostic features which facilitate the grouping of discrete motifs or images into
distinctive sets which, at DKS, reveal interesting temporal patterns. As shown above in
Chapter One, the painted images at DKS lend themselves to straightforward
categoriazation.

Manner of depiction

Three manners of depiction have been identified at DKS and are recognizable in
different sets of images which have been defined in chapter one from observations
made by Yates, Manhire and Parkington (1993, 1994). These are images which are
rendered in the fine-line manner (some of these are detailed), some images are in offset
print manner and others in crude finger-painted manner. The other image types have
been classified as having no defined manner, and they include crayon lines, writings,
drawings, smears and smudges, and a splash. In terms of numbers, there are 34 images
in the fine-line manner, which include animals and humans, 133 plain and decorated
handprints in offset print manner, 19 images of colonial character depicting humans
with European material cultural objects and geometrics (grids) as well as animals and all
are rendered in crude finger-painted manner. The remaining 34 motifs are of no
definable style of execution. [Page 28 End]

28
[Page 29 Top]
Techniques used in making images

DKS has two recognisable techniques which were utilized in the production of images.
The predominant technique is the conventional wet technique which involved mixing
powdered pigment with a liquid binder before application on to the rock substrate. A
total of 203 images/motifs are in this wet technique. The remaining 16 motifs were
produced using the dry technique in which crayons of charcoal, ochreous material and
commercial chalk (i.e., yellow line 58) were scratched or rubbed on the rock substrate to
make the images.

Pigment used for imagery

The red ochre pigment which occurs in various shades or hues is the dominant colour at
DKS rock paintings. All the colonial era images appear in monochrome brick red ochre
pigment. It is tempting to associate this pigment with the ochreous material which
derives from the seams at the bottom of the DKS rock shelter. Research is necessary to
determine if indeed the pigment for colonial imagery was obtained ad hoc on site. All the
handprints in the analysed sample are in red ochre of varying shades, mainly near brown
to near purple or maroon. Overall, they are barely distinguishable and differential
preservation could be the main reason for the observable slight variations in hue of
pigmentation within this set. The 10 smudges and smears appear in the same colour as
handprints. All the 13 eland are rendered in tri-chrome pigments of red (torsos), white
(head, neck, belly and the limbs) where preservation permits and black (fine outline)
which is barely visible in some images due to fading. Some of the fine line images were
executed in black (4 humans and 1 animal), yellow ochre (2 animals) and white (2
animals and 1 identified peculiar splash). Signatures and dates, crayon lines and
drawings are in black (charcoal), brick red ochreous material and one instance of
commercial yellow colour. One noteworthy material which was used in writing
signatures and dates is a pitch black substance crudely and thickly applied on the rock.
It has not been established yet what it really is, nor its possible source. [Page 29 End]

29
[Page 30 Top]
Defining colours or pigmentation of various images is a difficult process particularly
where preservation is poor, as is invariably the case with many sites. At DKS fading is a
limiting factor and is compounded by the presence of a white to off-white or creamish
layer/coat which masks most of the images especially in the middle sections of the site.
This coating apparently emanates from the ledges above the shelter where the overhang
projects outwards and is possibly associated with urine from dassies (Procavia capensis)/or
birds.

Evidence of co-variation in manner, technique and pigment

There is evidence in the sequence of DKS of the close relatedness of the attributes of
manner, technique and pigment. All the colonial era images are in the crude finger-
painting manner, wet technique and brick red ochre pigment. All the handprints are in
offset print manner, wet technique and red ochre pigment. While the earliest set of
imagery at DKS is in fine line manner and wet technique, the pigments used vary from
shades of red ochre, yellow ochre, black and white colours between the individual
monochrome, bichrome and trichrome images. There is also variation within the latest
set comprising signatures, dates, crayon lines and drawings where various materials of
different colours were utilized as crayons, and all except three cases share the same
attributes of no defined manner and dry technique. The other three cases are in the
crude finger painting manner, wet technique and black pigment which are the signatures
and dates executed in a thick black substance.

The sequence at DKS

Discerning superpositioning at DKS has been in part affected by preservation problems


of paintings. There is ample evidence that images/motifs here, especially handprints do,
in fact, overlap onto one another in many instances, but the difficulty is in attempting to
deduce with certainty which image overlies or underlies which. In this study, such
doubtful relationships have been excluded from analysis and attention was placed on
these images where the sequence is clearly resolvable. This study [Page 30 End]

30
Figure 3: Shaded areas are red ochre and stippled images are brick red ochre different from the
red pigment used for other images. No sequence could be resolved between images 164
and 106.
[Page 31 Top]
focused on as much images as could be recorded within the main painted area primarily
to get the important relationships of superposition and equivalence and secondarily their
general spatial associations. Images occurring above or below the main painted area
have not been taken into account because they are mainly widely separate and not
involved in chronologically useful relations. For the purposes of reconstructing relative
chronology, only the relations of sequence and of equivalence are essential.

Of the 219 recorded images, 24 are floating or not involved in any relation of sequence
or equivalence with other images, they were initially thought to be, then discarded. From
this number, two images that would have increased the reliability and confidence of the
sample, involving a fine line animal (no. 149) with a handprint (no. 99) and a crude
finger-painted bird (no. 143) with a charcoal drawing (no. 142) (chicken?) and a
potentially useful relation of colonial era human (involved only in equivalence relation
with the other humans in the set) with a handprint have been excluded from the analysis
due to their unclear sequence (Fig. 3). The other 195 images in the sample are either in
the relations of equivalence, or sequence or both. The dominant image type is the
handprint totaling 133 in number with also the largest chain of equivalence relations
involving 101 handprints. This is the main line of handprints which occupies the site
from side-to-side.

Overall, 65 relations of sequence have been identified at DKS (Table 2). They are
mostly of two-element superpositioning (i.e. one to one relation of sequence) which is
one image above or below another image. Multiple-element superpositions have also
been identified; those images overlapping with two or more other figures. The largest
one of these involves a single crayon line motif which overlies 9 other individual images.
In this study such a relation is construed as nine instances of crayon line motif (No. 58)
over those respective images which it overlies (i.e. 6 handprints, 2 eland and 1
geometric). Another long chain is that of a signature (No. 200) which overlies 7 other
images (i.e. 4 handprints, 2 fine line animals, and 1 fine line human). Three instances of
three-element superpositioning and five cases of four-element (Fig. 4) superpositioning
have been recorded. The occurrence of multiple [Page 31 End]

31
Figure 4: Multiple superposition from four different sets. Stippled images (63) and (46) are in
black and brick red ochre respectively. Solid areas are the red ochre pigment and the
unshaded area represents white.
[Page 32 Top]
superpositions becomes inevitable in situations where over-painted and a composite
number of images exist in a restricted or small area.

The Harris matrix diagram (Fig. 5) has revealed that there are four or five layers/tiers of
superimposed motifs on which some confidence can be placed in the building up of a
chronological sequence. It has also related different image types into distinct sets in the
relative chronology at DKS by virtue of their superpositionings. The diagram has been
generated from the images that are involved in relations of sequence within the stratified
sets at DKS. This is because only these relations can inform us on the relative
chronology of the painting traditions. The observed sequence from top (latest set) to the
bottom (earliest set) is as follows:

1. In the upper layer, the observed motifs include; signatures (names), dates crayon
lines and charcoal drawings which seem to mimic images from earlier painting
traditions. There is no instance where images in this set are overlain by other images
from other distinctive sets. Conversely, some motifs from all other sets are
superimposed by images from this layer. This set is involved in 38 relations and
sequence; 15 instances of it overlying the set of fine line imagery, 18 instances above
handprints, 2 cases above smudges/smears and 3 cases of it superimposed on colonial
era imagery (Table 2).

2. The second layer comprises the colonial era imagery which includes; humans,
animals (also birds) and geometric designs. All images in this layer are in monochrome
brick-red ochre pigment. There are three instances of the upper layer set of signatures
and crayon lines overlying colonial images. They are 63/46 (Fig. 4), 142/144 which are
described as crude finger-painting manner, wet technique, brick red pigment (colonial
images) overlain by no defined manner, dry technique, black (charcoal drawings). The
third is crayon line motif 58 over geometric (no.59). Images which are overlain by this
set include 3 handprints, 1 fine line animal and 1 smudge. [Page 32 End]

32
FIG 5: The Harris matrix diagram showing the sequence of DKS painted images
(overleaf).

Key to the Harris matrix diagram:

SD = signatures and dates

DR = charcoal drawings

CR = crayon line motifs

CA = colonial era animals

CH = colonial era humans

CG = colonial era geometrics

SM = smudges/smears

HP = Handprints

SP = Splash

FE = fine line eland

FA = fine line animals

FH = fine line humans


Figure 6: Finely stippled areas denote white, solid areas are red and areas in large dots/stipples indicate superposition. Unshaded areas are either
faded or never had paints applied to them.
[Page 33 Top]
3. The third layer is made up of smudges and smears all of which are in red ochre
pigment similar to that used for handprints. This set occurs immediately above the
handprints, overlying them in 5 instances. The fine line images are overlain
superimposed by this set in 3 instances.

4. Handprints (decorated and undecorated) are found in this layer and appear in
varying shades of red ochre pigment. This set forms a useful datum as it is a continuous
line across the painted area of the site. Other image sets in the DKS sequence are either
above or below this datum. Handprints are overlain on 15 instances by various images
from the distinctive sets described above. The handprints overlie fine line eland 12 times
(Fig. 6), and on 4 instances occur above other fine line images. One case of a handprint
has been recorded overlying a peculiar white splash (Fig. 7).

5. The bottom-most layer comprises the earliest fine line image set which includes;
eland, other animals (black, shaded red ochre, shaded yellow ochre and white pigment)
and humans (shaded red ochre and black pigment). There is no observable case where
an image from this set overlies any image from other defined sets. The observed pattern
is that of other images from the top layers in the DKS stratified sequence overlying the
fine line imagery. Fine line images are 16 times overlain by handprints, 3 by
smudges/smears, once by a colonial era image and 14 times by motifs from the topmost
layer.

What I have presented above are the different layers in the sequence into which the
defined image sets seem to fit well. The Harris matrix diagram has shown that the
different image sets are internally coherent and can be used to deduce a chronological
sequence from DKS. These image sets, however, should not be taken to mean a kind of
single coherent compositions. There is a probability of the existence of sub-sequences
within these distinctive image sets defined above. The small nature of the sample,
however, does not allow a finer resolution of sequence beyond the categories which this
study has defined. A bigger sample from a number of sites where sequence [Page 33
End]

33
Figure 7: Solid areas indicate red achreous pigment. Unshaded geometric is brick red ochre.
Stippled image 77 is white; stippled image 78 is in red, but has been contrasted with red
handprint 209 underneath.
[Page 34 Top]
is observable and resolvable is necessary if one requires high resolution information
about sequence within the level of a single distinctive set of imagery.

The Harris matrix proves to be a powerful tool in sorting out and analysing
relationships of imagery in overpainted panels. It pays particular attention to discrete
images as units of stratification in the sequence, if indeed they are involved in above or
below relations. Images belonging together in one stratum are represented at different
levels within the final diagrammatic layout of the sequence. Relations of equivalence,
signified by=sign; do not appear in the diagram. These relations were numerous and
tended to complicate the diagram and hence I decided to exclude them. The
manipulated images are only those involved in relations of sequence (Table 2), which are
essential in relative chronology reconstruction. [Page 34 End]

34
CHAPTER FIVE

DISCUSSION OF THE RESULTS

The working hypothesis presented in this study is that the fine line imagery is earliest in
the sequence and that handprinting emerged later than this tradition. Whether
handprinting emerges when it had ceased completely or not is uncertain. The defined
stratified sets of images discussed above represent phases in the production of paintings
throughout the sequence at DKS. This study, therefore, tested the idea that the sets:
signatures and crayon lines, colonial era imagery, handprints and fine line imagery are
indeed distinctive entities with strong temporal coherence. The Harris matrix
methodology was applied to resolve the images in the sample which are involved in
relations of sequence into a single successive order. It has also helped to check if there
are any inconsistencies in the form of contradictory relations.

Of importance in this study is the nature of relations between the defined stratified sets
of imagery. On the use of the Harris matrices, Chippindale and Tacon have pointed out
that:

“Figures of class A are always stratified over figures of class B, then classes A and B may
indeed be distinct entities in a chronological sequence. If figures in class A are
sometimes over and sometimes under figures of class B, then something is amiss: the
superpositions may have been mis-read or the classes A and B ― if entities at all ― are
not chronologically distinct” (1993:3:39).

Regarding DKS paintings, the evidence provided by the Harris matrices is that the
defined image sets are cohesive entities, occurring within different chronological phases.
The temporal lapse between the different sets is uncertain. There could also have been
temporal overlaps which are not discernible because of the small nature of the sample.
What can be established are broad generalized time frames, some of which can be linked
to historical evidence with absolute time depths. [Page 35 End]

35
[Page 36 Top]
The fine line imagery occurring at the bottommost layer is the earliest painting tradition
phase. This argument that the fine line imagery set predates all the other defined sets is
supported by the absence of any motif from other image sets shown to be overlain by
fine line images. This set, also referred to as the detailed representational art (Yates et al.
1993) utilizes a range of ochreous reds and yellow, black and white pigments. It is the
detailed fine line manner of execution which sets apart this category from other image
sets. The time depth of this painting tradition still remains unknown due, largely, to the
present lack of appropriate absolute dating procedures or methods for rock paintings as
discussed in chapter one.

The depictions of fat-tailed sheep in fine line representational manner in the south-
western Cape is argued to be evidence of the persistence of this tradition until the
period around 1, 900―1, 600 BP when pastoralism was introduced in the region (Yates
et al. 1993). One fine line image at DKS is arguably identified as a sheep painted in black
pigment. If indeed this image is a sheep, one could make an argument that the period
around 1, 900―1, 600 BP provides a terminus ante quem for the disappearance of fine line
tradition at DKS. It should be noted, however, that Manhire and his colleagues (1986)
reported that no definitive examples of fat-tailed sheep were located from searches
within the coastal foreland.

Another coherent set of imagery which occurs above the fine line images in the DKS
sequence is that of handprints. The absence of evidence of handprints overlain by fine
line images and the occurrence of instances where the latter are overprinted is clear
proof that the handprinting tradition was practised when the fine line tradition had
ceased, and therefore later in the sequence. Otherwise, fine line images would be
expected to be sometimes overprinted by and at other time superimposed on the hand
prints. An argument against this inference could be presented that the two sets of
imagery fall within the same broad coherent time frame. The observed pattern of
sequence then argued as being, for one reason or the other, a question of avoidance to
superimpose fine line images on handprints while at the same time the reverse being
encouraged. This is not a preferred argument in this study and can be refuted by the
[Page 36 End]

36
[Page 37 Top]
observed absence of superpositionings in some sites where these two sets of imagery
exist side-by-side (Yates et al. 1993).

Handprinting, while largely controversial, has variously been associated with pastoralists,
people of Khoi or Khoisan stock (Van Rijssen 1994; Anderson 1996), particularly Khoe
females during their initiation rites to womanhood in the context of the latter study.
Yates and co-workers (1993) suggested circumstantial association of handprints with
pastoralism, mainly due to the existence of high frequencies of occurrence of this image
type at the coastal areas where, then, evidence of pastoralism was argued to be earliest,
most intensive and its impact greater than in the mountains within the region. Recent
observations show that the distribution of sites with handprints is in almost equal
proportions in the sandveld and mountains (Manhire: Pers. Comm.).

Smudges and smears appear in the Harris diagram to be the next set of imagery above
the handprints. These images are not artistic representations of anything. They occur in
the form of amorphous shapes or red ochre stains of various sizes. They have been
identified overlying handprints (5 times) and sometimes fine line imagery (3 times). They
are themselves overlain by signatures and dates (3 times) and once by a colonial era
geometric. While at some sites finger smears have been found to overlie colonial era
imagery this is not so with the analysed sample at DKS. As noted above, this set of
imagery uses a similar pigment to the one used for handprints―red ochre. It is not clear
whether it is a significant and chronologically distinct entity. It is difficult to draw
meaningful inferences of this nature from such a small sample.

The other stratified set in the sequence is the colonial era imagery, which at DKS is only
rendered in the crude finger painting manner. Examples of colonial era subject matter
depicted in fine line manner have been identified at Stompiesfontein 1 (STF1) and other
localities in the southwestern Cape (Anderson 1996). The colonial era in the
southwestern Cape is taken to mean the period from 1652 to the 1850s AD when
sustained contact occurred between the Europeans and the local indigenes (Yates et al.
1993). The subject matter depicted in this set of imagery is self-evident of this period
[Page 37 End]

37
[Page 38 Top]
as it invariably shows Europeans (farmers?) with western material culture such as
brimmed hats, smoking pipes, rifles or muskets and high-heeled shoes. At DKS, also
associated with colonial era humans are exaggerated genitalia and inanimate depictions
like geometrics (rectangular, square and circular in shape with grids and crosses inside)
and one fork-like design (Fig. 4) and a likely unfinished motif.

The colonial era imagery set is distinctive and temporally coherent in terms of both the
attributes of manner, technique and pigment and the subject matter. A distinctive brick
red ochre pigment was used with the crude finger painting manner of depiction and wet
technique which have all necessitated connections or linkages to be established between
these otherwise spatially separate or isolated images. They have been merged into a
single contemporary stratum and chronologically self-contained in the position they
occupy within the DKS sequence. This should not be equated with a single
compositional group, but an image-making tradition within a given time frame as
defined above.

The uppermost stratum which succeeds, if indeed it can be so defined, the colonial era
imagery in the sequence of DKS as revealed by the Harris diagram is the signatures,
dates, crayon lines and charcoal drawings. There is no evidence of any other image set
overlying this set, an observation also made by Anderson (1996) from a number of sites
he studied. This is the latest phase of image-making at DKS and appears to be active
even in modern times. There are dates of 1832, 1910, 1930 or 1932 and another date is
not clear whether it is 1917 or 1.9.17 and whatever such numerical notation would
mean.

This last set of imagery is interesting and Anderson notes that it is often thought to be
“graffiti” or “modern art” or often considered to be scratching by recent young herders
or shepherds and thus not art per se (1996:71). He argues that there is same internal
consistency within this imagery set, which considering it idle scratching would be
improper. This view that this set should be regarded as part of an image-making
tradition worthy of study is adopted in this present analysis. This is especially true in
attempting the establishment of relative chronology of the painting traditions. The
[Page 38 End]

38
[Page 39 Top]
Dictionary of Art (1996) gives a definition of graffiti as “an arrangement of
institutionally illicit marks in which there has been an attempt to establish some sort of
coherent composition.” While this phenomenon in all its various forms is viewed as an
illegitimate art form or vandalism there is growing interest in its study from different
perspectives. In this study I have avoided the use of the term “graffiti” because of the
connotations embedded in it. A preferred reference to it is particularly descriptive of the
nature of the subject matter and form of the imagery in this set.

The results presented in this study are in agreement with Anderson’s (1996)
observations and conclusions from the Bokkeveld sample. He argued that the fine line
images from his sample are superimposed by all other image sets. Where fine line
images were observed superimposed on other images, those images were themselves
depicted in the fine line manner. Above this set in the sequence that he deduced were
handprints which he divided into small and medium size (earlier) and large handprints
(later) in their production. This is predictable from his interpretive framework which
addressed the question of authorship and pastoralists, particularly the female Khoe
populations are implicated. The present study is cautious about this subject which has
been a matter of constant debate, albeit with less information based on definitive
evidence forthcoming. Results from recent investigations, contrary to earlier
suggestions, show that the distribution of sites with handprints is almost in equal
proportions in the sandveld and the mountains (Manhire: Pers. Comm.). Earlier
suggestions were that these sites occur in superabundance in the sandveld, particularly
along the coast where pastoralist activities had a great impact.

Anderson (1996) suggests that smearing occurred after the fine line tradition had ceased,
which has also been demonstrated by the Harris matrix diagram generated for the DKS
sample. From the present study, smears occur above the handprints, although this
forms a tiny sample to be very reliable. Anderson argues that smears are in fact in the
same phase with the small and medium handprints, finger painting and certain colonial
imagery, but not with the later large handprints. His most recent phase comprises large
handprints and some colonial imagery and black charcoal drawings. [Page 39 End]

39
[Page 40 Top]
The Harris matrix has shown that image sets at DKS are distinct and self-contained
categories. While Anderson’s (1996) sequence shows some image sets like the
handprints and colonial imagery to have some of their individual members (images)
occurring in different chronological phases, the present study demonstrates that all the
identified image sets are chronological coherent entities. These sets occur in those
positions in the sequence such as revealed by the Harris matrix diagram. This means
that, for DKS, one does not find some images from a given set in a given position
within the sequence also occurring in other different stratigraphic levels associated with
other sets.

One notable difference in approach which could, in part, be responsible for the
difference in our results is the definition of categories/sets. The present study attempts
to simplify the image type categories for practical reasons. Anderson’s (1996) study
tends to give a complex breakdown of image sets; the handprint set has been sub-
divided on the basis of size measurements. Fine line images are sub-divided into those
that are finger-smeared and those that are not ― and the same distinction is carried over
to the colonial imagery.

Handprints are here regarded as a single coherent set. A fairly reasonable interpretation
on the authorship is that the descendants of the former San hunter-gatherer groups who
produced the fine line images are probably the ones who made the handprints. Size
measurements have shown a wide range of variation including handprints of children at
other sites (Manhire: Pers. Comm.). It appears, therefore, that this tradition was a
communal activity, perhaps associated with some sort of ritual or otherwise, where
different age groups seem to have partaken in the production of this set. Regarding
whether females or only males were involved, Manhire (Pers. Comm.) is very cautious
due to the present lack of any definitive evidence on the differences, based on sex, of
the handprints in rock shelters. From a small sample experimental approach involving
female and male students at UCT, differences in the prints they produced were noted.
For obvious reasons, this cannot [Page 40 End]

40
[Page 41 Top]
be translated onto the handprint images in rock shelters. This area of rock art studies
definitely requires more attention than it has previously received.

A general statement can be made about the pigments at DKS. There seems to be a shift
in the range and nature of pigments utilized through time and reflected throughout the
DKS sequence. In the earliest fine line image making tradition, a good range of
pigments comprising various shades of red ochre, yellow ochre, black and white
pigments was employed in the execution of images. This is in keeping with the finesse
with which these images were depicted. In the second level, that of handprints, the only
pigment utilized is red ochre with slight variations in hue which, as shown above, could
be due to differential preservation, while at other sites, yellow ochre (Elands Bay Cave),
has been used ― with one black print and white handprints identified elsewhere
(Manhire: Pers. Comm.). DKS has only handprints executed in red ochre. The reason
for this is uncertain and needs investigation. Smears and smudges above the handprints
utilize the same kind of red ochre pigment. Interestingly, the colonial imagery set utilizes
a distinctive brick red ochre pigment which can not, in all probability, be associated with
the sources from which other red ochre pigments discussed were obtained. The seams
of the bottom of the DKS shelter sandstone produce soft ochreous material which is
the same colour as that used for colonial imagery. One is tempted to assume that this
was the source of the material used for colonial era imagery. It is uncertain why the
earlier image sets did not utilize this source of pigment, available right at the rock
substrate suitable for painting? Perhaps this may be explained in terms of logistical
considerations and the cognitive realms within which these earlier societies operated.
As I have said above, this is another area that needs more detailed attention. The final
set at the top of the sequence is that of signatures, dates and crayon lines as well as
charcoal drawings. A variety of media are utilized in the execution of the motifs in this
set. These include charcoal crayons, ochreous crayons with the same colour as the
material deriving from the seams at the bottom of DKS shelter, commercial chalk and
an unknown distinctive thickly applied black substance. [Page 41 End]

41
[Page 42 Top]
Conservation of paintings at DKS

The paintings at DKS appear to have endured through time but signs of deterioration
are showing up. While the shelter is generally protected from the direct impact of the
elements, the paintings are mostly faded. The microclimate within the shelter appears to
be largely influenced by the air flow and draughts generated in the process. The fauna
(dassies) and avian fauna (raptors) habiting the ledges and cleavages above the shelter
also have a negative impact on the preservation of the DKS paintings as evidenced by a
white or creamish coating that is over the middle section of the site. These are general
observations not based on any specific investigations.

The superposition phenomenon discussed above undoubtedly has an effect on the


preservation of paintings, although the scale cannot be immediately demonstrated or
established. Some handprints at DKS appear to have flaked off from the rock surface as
a result of the detachment of pigments constituting the underlying images of eland. This
is a two-way process. The execution of the handprint superimposed on earlier fine line
eland could have weakened their bonding on the rock and initiated a gradual process of
decay. On finally flaking off the rock substrate ― the earlier pigments come off with
portions of pigment of images superimposed above them. In some cases the pigment of
later images applied on earlier images were not successful in penetrating the pigment of
these images to form a strong bond with the rock substrate. In this way the later
paintings or their portions are, therefore, in a weakened position right from the time of
execution, and at the same inducing the deterioration of earlier superimposed images
which have stronger bonds with the rock below them.

Overall, the red ochre pigments appear to be strongly bonded to these quartzite
sandstones which form a good host rock matrix for the absorption and stabilization of
ochreous pigments. I believe that DKS also offers some revealing sets of circumstances
which, when properly investigated, will inform rock art conservation studies. [Page 42
End]

42
[Page 43 Top]
CONCLUSION

The sequence of the painting tradition at DKS confirms the early observations that
handprinting post-dates the fine line imagery (Yates et al. 1993, 1994). It has also
demonstrated that there is an observable shift through time from one image set to
another. The succession of image production phases is from the fine line images (some
of which are detailed) through the handprints and smears/smudges to colonial era
imagery and finally the signatures, dates and crayon line motifs and charcoal imitative
drawings of earlier motifs at the top. The Harris matrix diagram demonstrates that these
are coherent sets or categories of images. The cohesive nature of these sets is strongly
supported by the lack of inconsistencies in terms of reversed sequence or any
contradictions in the relationship structure in the sequence. More importantly, these
distinctive image sets are also chronologically coherent; they have been produced in
different phases although these should not necessarily be tight time frames. While these
broad temporal phases can be envisaged, through such innovative methodological
approaches like the Harris matrix analysis, the nature of the intervals from one episode
of image-set production to another, cannot be demonstrated. The possibility that cannot
be discounted is that there were temporal overlaps from one image set to another. This
sort of overlap is attested by cases where some colonial era images are rendered in fine
line manner ― which is evidence that the fine line tradition endured until the Colonial
period. Similarly cases where fat-tailed sheep are rendered in fine line manner suggest
that this painting tradition continued until the time when pastoralism was introduced.

The chronology presented in this study did not take into account overlapping within
single distinctive sets identified above. That is, no deduction of sequence was attempted
within the fine line image set or the handprint set or any other set at DKS. This is
because there were no obvious or observable instances of stratification within individual
sets such as the fine line image sets for example; a good number of overlaps could be
observed in the handprint set but due to fading no clear sequence could be resolved.
Overall, regarding this, the sample is too small to provide any meaningful chronological
information from the level of a single set (intra-set stratification or sub- [Page 43 End]

43
[Page 44 Top]
sequence). The analysis of this nature can be productive at inter-site level where larger
and reliable samples can be obtained.

44
[Page 45 Top]

REFERENCES

Anderson, G.C. 1996. Andriesgrond revisited: material culture, ideology and social change.
Unpublished B.A. (Hons) Thesis: University of Cape Town.

Anderson, G.C. 1996. The social and gender identity of gatherer-hunters and herders in
the southwestern Cape. Unpublished MPhil. Thesis: University of Cape Town.

Battis, W.W. 1939. The amazing Bushman. Pretoria, Red Fawn Press.

Bednarik, R.G. 1979. The potential of rock patination analysis in Australian Archaeology-
art 1. The Artifact 4: 14-38.

Bednarik, R.G. 1992. A new method to date petroglyphs. Archaeometry 34: 279-291.

Bleek, D.F. 1932a. A survey of our present knowledge of rock paintings. South African
Journal of Science 29: 72-83.

Cartwright, C. and Parkington, J.E. 1997. The wood charcoal assemblages from Elands
Bay Cave, southwestern Cape: principles, procedures and preliminary
interpretation. South African Archaeological Bulletin 52: 59-72

Chaloupka, G. 1983. Kakadu rock art: its cultural, historic and prehistoric significance.
In Gillespie, D. (ed.), The rock art sites of Kakadu National Park: pp. 3-33.
Canberra: Australian National Parks and Wildlife Service.

Chaloupka, G. 1985. Chronological sequence of Arnhem Land Plateau rock art. In


Jones, R. (ed.), Archaeological research in Kakadu National Park: pp. 269-280.
Canberra: Australian National Parks and Wildlife Service.

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

Chippindale, C. and Tacon, P.S.C. 1993. Two old painted panels from Kakadu: variation
and sequence in Arnhem Land rock art. In Steinbring, J., Watchman, A.,
Faulstich, P. & Tacon, P.S.C. (eds.), Time and Space: dating and spatial considerations
in rock art research: pp. 32-56. Melbourne: Australian Rock Art Research
Association.

Clottes, J., Courtin, J., Collina-Girard, J., Arnold, M. and Vallaclas, H. 1997. News from
Cosquer Cave: climatic studies, recording, sampling, dates. Antiquity 71: 321-326.

Cooke, C.K. 1963. The painting sequence in rock art of Southern Rhodesia. South African
Archaeological Bulletin 18: 172-175.

[Page 45 End]

45
[Page 46 Top]
Deacon, J. 1993. Management guidelines for rock art sites in two wilderness areas in the
Western Cape. Cape Town: Project Report, National Monuments Council.

Dorn, R.I. 1990. Rock varnish dating of rock art: state of the art perspective. La Pintura 7:
49-73.

Garlake, P. 1987. The Painted Caves: an introduction to prehistoric rock art of Zimbabwe. Harare:
Modus Publications.

Gillespie, R. 1997. On human blood, rock art and calcium oxalate: further studies on
organic carbon content and radiocarbon age of materials relating to
Australian rock art. Antiquity 71: 430-437.

Harris, E.C. 1989. Principles of archaeological stratigraphy (2nd edition). London: Academic
Press.

Heydorn, A.E.F., and Morant, P.D. 1986. Estuaries of the Cape: Verlorenvlei. Stellenbosch:
University of Stellenbosch.

Lewis-Williams, J.D. 1974. Superpositioning in a sample of rock paintings from the


Barkley East District. South African Archaeological Bulletin 29: 93-103.

Lewis-Williams, J.D. 1981. Believing and seeing: symbolic meanings in southern San rock paintings.
London: Academic Press.

Lewis-Williams, J.D. 1984. The empiricist impasse in southern African rock art studies.
South African Archaeological Bulletin 39: 58-66.

Lewis-Williams, J.D. and Dawson, T.A. 1989. Images of power: understanding Bushman rock art.
Johannesburg: Southern Book Publishers.

Lewis-Williams, J.D. 1990. Documentation, analysis and interpretation: dilemmas in rock


art research. South African Archaeological Bulletin 45: 126-136.

Loubser, J.H.N. A guide to the rock paintings of Tandjesberg. National Museum,


Bloemfontein 9(2): 346-384.

Manhire, A.H. 1981. Rock art of the Sandveld. Unpublished BSc. (Hons) Thesis:
University of Cape Town.

Manhire, A.H. and Parkington, J.E. 1983. A distributional approach to the interpretation
of rock art in the southwestern Cape. South African Archaeological Society Goodwin
Series 4: 29-33.

[Page 46 End]

46
[Page 47 Top]
Manhire, A.H., Parkington, J.E., Mazel, A.D. & Maggs, T.O’C. 1986. Cattle, sheep and
horses: a review of domestic animals in the rock art of southern Africa. South
African Archaeological Society Goodwin Series 5: 22-30.

Meadows M.E., Baxter, A.J. and Parkington, J.E. 1996. Late Holocene environments at.
Verlorenvlei, Western Cape Province, South Africa. Quaternary International 33: 81-
95.

Miller, D.E., Yates, R.J., Parkington, J.E. and Vogel, J.C. 1993. Radiocarbon dated
evidence relating to a mid-Holocene relatie high sea level on the
southwestern Cape coast, South Africa. South African Journal of Science 89: 35-44.

Pager, H. 1971. Ndedema. Graz. Akademische Druck.

Pager, H. 1976. The rating of superimposed rock paintings. Almagoren 5: 205-218.

Parkington, J.E. and Poggenpoel, C. 1987. Diepkloof rock shelter. In Parkington, J.E. and
Hall, M. (eds.). Papers in the prehistory of the Western Cape, South Africa: 269-293.
Cambridge: Cambridge University Press.

Parkington, J.E. 1989. Interpreting paintings without a commentary. Antiquity 63: 13- 26.

Parkington, J.E. 1977b. Follow the San. Unpublished PhD Thesis: University of
Cambridge.

Philips, S.A.(In press). The Dictionary of Art. London: Macmillan Publishers.

Rosenfeld, A. 1987. Rock art conservation in Australia. Australian Government Publishing


Service.

Rosenfeld, A. and Smith, C. 1997. Recent developments in radiocarbon and stylistic


methods of dating rock art. Antiquity 71: 405-411.

Rudner, I. 1989. The conservation of rock art in South Africa. Cape Town: National
Monuments Council.

Solomon, A. 1989. Division of the earth: Gender, symbolism and the archaeology of
Southern San. Unpublished MA Thesis: University of Cape Town.

Solomon, A. 1994. ‘Mythic women: a study in variability in San rock art and narrative. In
Dowson, T.A. and Lewis-Williams, J.D. (eds.), Contested Images: pp. 331-371.
Johannesburg: University of the Witwatersrand Press.

Stow, G.W. 1905. The native races of South Africa. London: Swan Sonnenschein.

[Page 47 End]

47
[Page 48 Top]
Thackeray, A.I. 1983. Dating the rock art of southern Africa. South African Archaeological
Society Goodwin Series 4: 21-26.

Van der Merwe, N.J., Sealy, J. and Yates, R.J. 1987. First accelerator carbon-14 date for
pigment from a rock painting. South African Journal of Science 83: 56-57.

Van Rijssen, W.J. 1984. South-western Cape rock art-who painted what? South
African Archaeological Bulletin 39: 125-129.

Van Rijssen, W.J. 1994. Rock art: the question of authorship. In Dowson, T.A. and Lewis-
Williams, J.D. (eds.), Contested Images: pp. 159-175. Johannesburg: University
of the Witwatersrand University Press.

Vinnicombe, P. 1960. The recording of rock paintings in the Upper reaches of Umkomaas,
Umzimkhulu and Umzimvubu rivers. South African Journal of Science 56: 11-14.

Vinnicombe, P. 1976. People of the eland. Pietermaritzburg: Natal University Press.

Willcox, A.R. 1959. Hand imprints in rock paintings. South African Journal of Science 55: 292-
298.

Willcox, A.R. 1984. The Rock Art of Africa. Johannesburg: Macmillan.

Yates, R.J., Manhire, A.H. and Parkington, J.E. 1993. Colonial era paintings in the rock art
of the southwestern Cape: some preliminary observations. South African
Archaeological Society Goodwin Series 7: 59-70.

Yates, R.J., Manhire, A.H. and Parkington, J.E. 1994. Rock painting and history in the
southwestern Cape. In Dawson, J.A. and Lewis-Williams, J.D. (eds.), Contested
Images: pp.29-60. Johannesburg: University of the Witwatersrand Press.

[Page 48 End]

48
49
50
51
52
53
54
55
56
57
Table 3: The data is a list of sets and subsets of imagery at Diepkloof Kraal Shelter (DKS) and the
attributes that define each layer.

Summary data on imagery sets at Diepkloof Kraal Shelter site


Set Numbers Criteria
SET A All images in fine line manner (some are detailed) and are
representational

Subset A1: 1-12 & 148 All are eland; same pigments (red, white, and black), All
are eland; same pigments (red, white, black), procession -
unidirectional.
Subset A2: 48, 49, 83, 202, 203, 210, 215, 218 All are humans (partial and complete); same technique
(red ochre and black pigments).
Subset A3: 45, 47, 147, 149, 150, 173, 193, 201, Animals (most are indeterminate species) similar in all
207, 213, 219 respects. Pigments include yellow and white.

SET B 13-44, 65, 66, 68-75, 79-82, 89, 90, 94- All are handprints (decorated and plain) in a line, with the
108, 110-139, 153-168, 177-192, 195- same techniqe and manner, and are all in red ochre.
199, 205, 209, 214

SET C Colonial era images in crude finger painting manner, same


techniqe and pigment
Subset C1: 140,141, 144, 145, 171, 172, 174, 206 European farmers have exaggerated genitals, same scale,
brimmed hats, smoking pipes, rifles or muskets and high
heeled shoes.
Subset C2: 46, 143, 146, 152 Animals (birds included) - horned bovids
Subset C3: 59, 76, 169, 170, 175, 193 Geometrics (square, rectangular and circular designs) with
grids and crosses inside.

SET D Signatures, dates, crayon line motifs and charcoal


drawings
Subset D1: 50, 55-57, 64, 87, 88, 91, 151 Signatures and dates
51-54, 200, 204 Executed in thick pitch black substance crudely applied
on rock
Subset D2: 58, 67, 85, 86, 92 Charcoal crayons of brick red ochre and yellow chalk

Subset D3: 63 & 142 Imitative drawings of earlier motifs in charcoal

58
Table 4: The diagrams below show proportions in percentage terms of individual images
involved in superpositions at DKS.

NB: The vertical axis shows the imagery type in a relative order they occur in the sequence and in
relation to each other, on the horizontal axis are the frequencies as percentages of the total number of
imagery analysed.

59