Professional Documents
Culture Documents
The Harris Matrix was developed in the 1970s to correctly interpret the sequence of data
derived from archaeological excavations. When layers of pigment are applied over
surfaces to make rock art, they also form sequences through time. Understanding motif
superimpositions is a key to understanding sequential changes in rock art repertoires.
The use of the Harris Matrix in rock art research was first proposed by Chippindale and
Taçon in the 1990s and was used to derive a firm sequence for western Arnhem Land
rock art in northern Australia. Their work was amplified in subsequent larger projects in
South Africa that clearly demonstrated the potential of the Harris Matrix in rock art
studies. Despite these successes, the Harris Matrix has been little employed elsewhere;
this chapter is a timely re-evaluation of the method and its underlying principles.
Introduction
The Harris Matrix is a diagrammatic way of representing how layers are superimposed:
those superimposed layers may be of soil in the ground, of accretions on objects, or—the
one of central importance to rock art studies—of paintings on walls. In many cases, the
superimposition sequence will be simple and the sequence of layers readily obvious.
Where the patterning is more complex—too difficult to make rapid sense of by simple
inspection—a graphic representation can greatly assist in visualizing the pattern of
superimposition, making the chronology easier to analyse. Harris Matrices provide such a
method.
Page 1 of 19
PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.
On a painted panel, the various layers of paint are micro-strata, equivalent to soil layers
in an archaeological site. Superimposed motifs can be sequentially ordered into a relative
sequence; then, if an absolute age is known for any given motif or layer (such as by
radiocarbon dating a particular image made of charcoal), the relative sequence can be
‘locked in’ to an absolute time frame. Although Harris Matrices have been applied to soil
layers dating back to the 1970s and to rock art by a few researchers since the 1990s, the
method remains underutilized in rock art research. Here, we describe the origins of the
Harris Matrix in ‘dirt archaeology’, explore its application in rock art research, and
demonstrate its usefulness by analysing the sequence in a complex rock art site in
northern Australia.
The discipline of archaeology evolved from a primary interest in the collection of artefacts
from archaeological sites. Deposits, or ‘layers’ which contained the artefacts, were often
summarily removed in order to retrieve the objects of interest. Often, therefore, the
stratification of the site was not fully recorded (but see Browman & Givens 1996 for some
notable exceptions). In the 1950s, archaeological publications began to exhibit a better
understanding of the principles of archaeological stratigraphy, in particular with regard
to the nature of surfaces. The stratigraphic importance of surfaces, however, did not
come into full realization until the 1970s (Harris 2013). Thus, in that 20-year period,
controversy arose between the recording of pit sections, or profile drawings of
stratification, that showed a ‘naturalistic’ image in which surfaces were not defined by
hard lines, and those in which they were. What was not explicitly stated, and perhaps not
fully understood at the time, was that surfaces rather than deposits were the key to the
interpretation of stratification, for without the identification of the archaeological
surfaces, the ‘stratigraphic sequence’ for a site (rather than the sequence of
stratification, as shown in a profile through the depth of a site) could not be compiled.
The development of stratigraphic theory continued through the 1960s, resulting in the
identification of features such as ‘robber trenches’, describing their structural nature as
holes rather than emphasizing the stratigraphic surface unit that was created in their
making. In other words, there was a coming realization that surfaces were important for
stratigraphic analyses, but that understanding remained rooted in the mainly deposit-
oriented view of the stratification. Since the 1980s, stratigraphic principles have been
Page 2 of 19
PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.
The value of the surface aspect of stratigraphic units in an archaeological context, not
fully understood until the last quarter of the twentieth century, expanded our
understanding and interpretation of archaeological stratigraphy. The reconstruction of an
archaeological site, of deposits on a painted panel, is the reconstruction of its surfaces;
that is to say, of the topography (spatial configuration) of the site. Such a topographic
reconstruction can only occur through reference to the site’s ‘stratigraphic sequence’, a
sequence made clear and visible through the medium of a Harris Matrix (Figure 1).
Page 3 of 19
PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.
The difference between stratification (all units stuck together) and a stratigraphic
sequence (units separated into relative time relationships) can be illustrated by analogy
with a spring-loaded toy, usually of an animal standing on a base that contains the spring.
The upright shape and individual parts of the animal, say a giraffe, are held together by
strings running through its body, those strings being akin to the lines of a Harris Matrix
connecting the boxes of the stratigraphic units. When the base of the toy is pressed
upwards, the animal collapses into a heap of ‘deposits’ that in effect comprises a mass of
stratification wherein the parts of the animal have many more physical relationships with
one another, most of which have to be discarded when the Law of Stratigraphical
Page 4 of 19
PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.
Succession is applied. That happens when the spring in the base of the toy is released
and the giraffe stands upright, thus creating its stratigraphic sequence from the
compacted stratified version of the animal. The main task for the archaeologist in any
stratified context is to create such a stratigraphic sequence from the compact mass of
stratification, be it composed of usual surfaces and deposits or the overlaid ‘masses’ of
rock art images or ‘deposits’. As with usual units, or rock art units, some units will not be
in superposition (physically overlaid) with others in the site, and thus multilinear
stratigraphic sequences usually pertain in most archaeological contexts. The term
‘superpose’ in geology and archaeology means to cover one thing with another (such as a
soil layer). ‘Superimpose’ is used in graphics to indicate the occurrence of one motif over
another, but where the underlying motif is largely visible (Murray & Murray 1972).
The compilation of Harris Matrices is based on a question of relative time; that is to say,
between any given units in superposition, the archaeologist asks: ‘Which came first?’ The
means by which the answers are compiled into a Harris Matrix, that is, a stratigraphic
sequence, are detailed in the second edition of Principles of Archaeological Stratigraphy
(Harris 1989), which is freely available online at www.harrismatrix.com in a number of
languages.
Sections illustrate the thickness or depths of deposits or art layers on a particular plane;
that is one of their values. They are of limited value in the reconstruction of a site
because that task rests with surfaces, which of course must be recorded in plan form if
the topographic history of a site is to be presented in a contoured format. So, sections
provide the depth, or layer, dimension of a site; plans fulfil the second and third
dimensions of length and width, or area; and the stratigraphic sequence supplies the
Page 5 of 19
PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.
fourth dimension of relative time. One of the roles of the contents of the art layers is to
impart dates or intervals of time to the stratigraphic sequence so that it may be
positioned in relation to gradations in absolute time or calendar years.
The remains present within a deposit are not necessarily the most reliable evidence for
the integrity of a site’s or panel’s stratigraphy because artefacts can move more or less
imperceptibly within and between layers, and art motifs can be destroyed by natural
causes before a subsequent layer is produced. Following recording of the rock art, the
resultant stratigraphic sequence becomes the main ‘pattern’ to which the analysis of the
motifs must conform. That is the ultimate value of a stratigraphic sequence: the relative
time sequence signalled by a site’s stratigraphy, as expressed in a Harris Matrix diagram.
Stratigraphic sequences have import for rock art sites and others that cannot be
excavated in the normal manner of archaeological excavations. Yet the stratification of
such sites can be arranged into stratigraphic sequences based upon the superposition, or
not, of the individual deposits/surfaces of paint. Once compiled, the extension of the
sequence to other parts of the site will depend upon analysis, for example, as that of the
imagery and chronology, in order to understand the development of the site and its
stratigraphic sequence in absolute time. The types of analysis required to achieve this
end are discussed in this chapter, but if the archaeological tasks are to be completed with
competency, the task must have a stratigraphic sequence against which those analytic
results can be verified.
The eminent nineteenth-century geologist Charles Lyell aptly noted that stratification is
‘undesignedly commemorative of former events’ (Lyell 1875: I, 3). Thus, when expressed
as a stratigraphic sequence, the sequence is an unbiased record of the past against which
all later analyses must be tested in part or in the whole. Or, to rephrase the matter:
stratigraphic sequences, being composed from the evidence of undesigned stratification,
are the unbiased testing patterns for archaeological sites. The Harris Matrix is the only
way to properly graphically represent such relative time sequences from rock art and
other archaeological sites, no matter the origins of the deposits or the nature of the
cultural and ecological material remains contained within the deposits. Harris Matrices
are of universal application and thus should be considered the ‘industry standard’ for any
archaeological site or context that has stratigraphic entities, including rock art.
Page 6 of 19
PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.
In her pioneering study, Russell (2000) examined four panels of San rock paintings
involving 255 motifs at the single site of Main Caves North in KwaZulu-Natal, South
Africa. She examined whether ‘groups of variables tended to occur in the same position in
the painted sequence and could be defined as a particular painted period’ (Russell 2000:
62). The Harris Matrix was able to highlight discernible trends in the art. Russell’s
‘periods’ correlated well with the proposed sequences of earlier researchers working in
adjacent areas of the Drakensberg, thus confirming the likely reliability of their findings.
The Harris Matrices further revealed that some select artistic traits were better
indicators of relative chronology than others (polychrome art was better than
monochrome, and paintings of eland were better than human figures for determining
relative chronology). Russell did warn, however, that changes in the manner of depiction
(usually glossed as ‘style’) from one layer to the next were not necessarily indicative of
distinct stylistic or chronological breaks, citing Australian studies that show how art
styles can differ within the context of site use and that more than one art style (such as a
manner of representation) could be found within a single period of time (Rosenfeld &
Smith 1997).
Following Russell (2000), Swart (2004) recorded and compared two widely separated
sites in the uKhahlamba-Drakensberg Park (UDP), again in South Africa, utilizing Harris
Matrices. She then compared her results with sequences from previous studies elsewhere
Page 7 of 19
PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.
in the UDP, finding regional continuity and change and some general patterns in the
structure of the art. Swart then attempted to correlate her findings with absolute dates
derived from pigment and oxalate crusts (Mazel & Watchman 1997, 2003). The results
were promising, but she suggested that ‘more site-specific absolute dating, based on
relative sequences, is necessary’ (Swart 2004: 33).
The potential of Harris Matrices to assist in the interpretation of rock art sequences has
therefore been well demonstrated, but to date remain little used in the development or
clarification of regional rock art sequences. Importantly, a method for resolving temporal
relationships between units (motifs) within multilinear rock art sequences that do not
have superimpositional links has not been fully developed. Unlike artefacts excavated
from stratified deposits, rock art is rarely recovered in readily observed, uniform layers.
Harris Matrices allow for the recognition of relationships between individual motifs but
only as individual entities. To construct motif layers or strata from the artwork (i.e.,
aggregating motifs from the same time period) from which ‘broad generalised painting
periods or phases’ (Swart 2004: 15) might be constructed, remains the most subjective
part of the interpretative process. Common artistic attributes are the usual basis for
aggregation (Chippindale & Taçon 1993; Keyser 2001). For example, Chippindale and
Taçon (1993: 36) used technique, pigment, and manner (following Chaloupka’s promoted
styles) to establish ‘equivalence’ (where motif sequence was not demonstrated). However,
Chaloupka’s proposed styles are, if at all, poorly defined, and their identification can vary
from researcher to researcher. In contrast, Russell (2000: 61) and Magar and Davila
(2004: 131) used various artistic and technical traits of the artworks for the purpose of
aggregation:
Russell (2000) restricts ‘style’ to motifs produced with similar pigment colour and
application (e.g., monochrome, polychrome, shaded polychrome, etc.), thus permitting
one or more styles to be present within a single ‘painting period’. Swart (2004: 20–21)
selected subject (motif type) and colour to correlate her ‘phases’ from two sites to
produce an overall descriptive sequence that was then compared and contrasted with the
sequences of previous researchers in the region. Consequently, motifs on each panel that
hold a common position in the panel sequence can be aggregated or grouped into layers
(strata) according to common artistic traits or attributes, such as colour, form, and
degree of preservation (Harris 1989: 105–119; Russell 2000: 61). The layers, then, could
consist of groups of individual motifs, interrelated ‘compositions’ (cf. Haskovec & Sullivan
1989: 68–70), ‘scenes’ (cf. Domingo-Sanz 2011), or composite ‘motifs’ (Maddock (1970):
450). At one level, a ‘visual composition’ entails the graphic unity of a suite of
contemporaneous motifs such that ‘the whole is very much more than the sum of the
parts’ (Murray & Murray 1972: 95). There are examples, however, in rock art where
Page 8 of 19
PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.
compositions may involve motifs from different chronological periods, where a later artist
has added motifs to the work of an earlier artist to produce a new composition. Or a
particular trait, such as certain distinctive manner of depicting a human figure, may not
be confined to a single brief period: it may have persisted over a extended span of time or
gone out of use and been revived at a later period. The identification of contemporaneous
compositions is in some cases readily apparent, while in others the defining relationship
of relevant motifs remains much more difficult and may require additional explanation.
Maynard (1977: 398) recognized that, whereas the majority of motifs within Australian
rock art could be readily accommodated to ‘types’, within each type certain idiosyncratic
features could be identified from an individual motif or amongst one or more groups;
these could be taken as unifying traits to identify a ‘subgroup’ of the type. Maynard
restricted her features to the exceptional and visually obvious (i.e., to those that made a
motif stand out from others at a site or from a regional pattern). Such features could be a
distinctive form (e.g., variations on X-ray), colour (e.g., repeated use of distinctive colour
combinations), size (e.g., unusually large or small motifs with other associative aspects),
or additional motif elements (e.g., ‘figure with hat’). Maynard’s choice of ‘idiosyncratic
features’ appears to have been largely intuitive because no sources are referenced.
It is these idiosyncratic features, however, that are particularly important in the next step
to interpret the Harris Matrix of an art panel. If a cluster of motifs—say, a line of human
figures in the same colour and form, painted in a similar manner, and with similar states
of preservation—appear to have been part of a single composition, then it is reasonable to
assume they were painted at the same time by the same painter. If two of these images
(say, H1 and H2) are superimposed by two fish motifs (F1 and F2), and two others of the
human figures (H3 and H4) are superimposed over two macropods (R1 and R2), then we
can safely assume that, within the site’s sequence, F1 and F2 also postdate motifs R1 and
R2, even though they are not in direct superimposition. This is a method that Chippindale
refers to as ‘finding a cultural horizontal’ (personal communication, 2016), a procedure
that complements the vertical stratigraphy of the matrix. The procedure is not perfect
because it does not take into account the possibility that H3 and H4 could have been
painted by a later painter closely copying the earlier H1 and H2 pair of figures to produce
the present row of figures. Under ideal research conditions, one could make the Harris
Matrix, then identify any contradictions or impossible stratigraphies within the vertical or
horizontal relations, and, after removing those insecure relations, repeat the analysis
with ‘cleaner’ data.
Page 9 of 19
PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.
of the facts, and its validity is arrived at, in part, through the experience of the
investigator in that field (see Sebeok & Sebeok 1981).
In undertaking comparative analyses of rock art, Chippindale and Taçon (1993: 39)
further stress the importance of studying the various motif traits individually, in order to
notice the presence of any co-variance between motifs, an all-inclusive approach also
advocated by Swart (2004: 16).
Provided by authors.
Page 10 of 19
PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.
Page 11 of 19
PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights Reserved. Under the terms of the licence agreement, an
individual user may print out a PDF of a single chapter of a title in Oxford Handbooks Online for personal use.
As a simple example, Panel M3 is one of 41 art panels on the ceiling of the large rock art
site of Nawarla Gabarnmang, located on the Arnhem Land plateau in northern Australia
(see Figure 2). Nine motifs have been identified from this small panel (see Table 1).
Following the careful recording of the attributes of each motif (often requiring the use of
the digital enhancement programme DStretch; Brady & Gunn 2012; Gunn et al. 2010;
Gunn, Douglas, & Whear, 2014; Harman 2008), a Harris Matrix was constructed by
plotting the superimposition sequence of every individual motif on the panel according to
its position below or above other motifs. In all, seven instances of superimposition were
noted (see Table 2). Two of the motifs (Motifs 13 and 21) do not occur in superimposition
with other motifs (see Figure 3). Interpretation of the Harris Matrix was then undertaken
to aggregate the motifs into sequential layers (see Figure 4). As mentioned earlier, the
aggregation of the motifs remains the most subjective part of the interpretative process.
On Panel M3, the attributes of colour and preservation were selected as the most
pertinent elements for assigning motifs to a common layer.
Provided by authors.
Table 2. Motif superimpositions on Panel M3. Note that Motifs 13 and 21 are not
involved in superimposition.
17 14, 15, 16
18 16
19 17, 18
20 17
Following the development of layers, it may be apparent that adjacent layers contain
motifs with similar attributes yet, through superimposition and the different positions of
the motifs within the Harris Matrix, the motifs are assigned to different layers. These
different layers may then be combined into a single Art Phase, such as Layers 2 and 3 on
Panel M3 that both consist of elongated figures in white with red outline and fine linear
Page 12 of 19
PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.
infill (see Figure 4). In some cases, these two layers may bracket a third layer that is
dissimilar to the other two, but, given that the upper and lower layers contain similar
artistic conventions, it must also be included within this one Art Phase. Layers that do not
have a ‘companion’ layer are considered to represent a distinct Art Phase.
Page 13 of 19
PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.
Provided by authors.
Page 14 of 19
PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.
Page 15 of 19
PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.
during the period lasting a maximum circa 85 years, between 1845 (the underlying horse
painting) and 1935 (the uppermost recent paintings). As some motifs both below and
above the horse image were undertaken in a similar style to each other, the painting of
the horse, representing a foreign animal that came ‘from the outside’, does not mark a
distinctive break in the art style repertoire (Gunn et al., in press). We can then construct
a Harris Matrix for Panel D at Nawarla Gabarnmang that incorporates actual ages (or age
ranges) for specific motifs, as shown in Figure 7. This Harris Matrix then becomes useful
not only for understanding the history of artworks within Panel D, but also for cross-
correlating art styles and motifs from different panels across the site where other
individual motifs with associated radiocarbon dates can potentially add further to the
site’s chronological resolution.
Conclusion
Understanding motif superimpositions has long been held to be a key to understanding
sequential changes in the rock art repertoire of a site or region. Yet despite the potential
of Harris Matrices for rock art research first being flagged by Chippindale and Taçon in
1993, the method has rarely been applied to this field of research. The construction of
Harris Matrices allows stratigraphically associated motifs to be grouped into layers that
themselves form sequences. Where absolute ages can be attributed to individual motifs,
the various layers of a sequence can be attributed absolute ages or age ranges.
The use of Harris Matrices to help visually structure and interpret rock art sequences
should be considered a standard component of rock art research whenever complex
superimpositioning is involved. As Le Quellec, Duquesnoy, & Defrasne (2015) have
emphasized for the use of DStretch digital enhancement, use of a Harris Matrix allows
readers to follow the logic of the purported associations. Nevertheless, and keeping in
mind that it is the researcher who chooses which variables to analyse, when relating
layers chronologically, behaviourally, and/or thematically, ‘we should not forget that
“reading” rock figures will always involve subjectivity’ (Le Quellec et al. 2015: 11).
References
Brady, L. M., & Gunn, R. G. (2012). Digital enhancement of deteriorated and
superimposed pigment art: Methods and case studies. In J. McDonald & P. Veth (Eds.), A
companion to rock art (pp. 627–643). Oxford: Wiley-Blackwell.
Browman, D. L., & Givens, D. R. (1996). Stratigraphic excavation: The first ‘New
Archaeology’. American Anthropologist, 98(1), 80–95.
Chaloupka, G. (1984). From paleo art to casual paintings. Monograph Series 1. Darwin,
AUS: Northern Territory Museum of Arts and Sciences.
Page 16 of 19
PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.
Chaloupka, G. (1993). Journey in time: The World’s longest continuing art tradition.
Chatswood, AUS: Reed.
Chippindale, C., & Taҫon, P. S. C. (1993). Two old painted panels from Kakadu: Variation
and sequence in Arnhem Land rock-art. In J. Steinberg, A. Watchman, P. Faulstich & P. S.
C. Taçon (Eds.), Time and space: Dating and spatial considerations in rock-art research
(pp. 32–56). Occasional AURA Publication No. 8. Melbourne: Australian Rock Art
Research Association.
Domingo-Sanz, I. (2011). The rock art scenes at Injalak Hill: Alternative visual records of
indigenous social organisation and cultural practices. Australian Archaeology, 72, 15–22.
Edwards, R. (1979). Australian aboriginal art: The art of the Alligators Rivers region,
Northern Territory. Canberra: Australian Institute of Aboriginal Studies.
Gunn, R. G., David, B., Delannoy, J. -J., & Katherine, M. (in press). The past 500 years of
rock art at Nawarla Gabarnmang, central-western Arnhem Land. Accepted by Terra
Australis, Canberra.
Gunn, R. G., Ogleby, C. L., Lee, D., & Whear, R. L. (2010). A method to visually rationalise
superimposed pigment motifs. Rock Art Research, 27, 131–136.
Gunn, R. G., Douglas, L. C., & Whear, R. L. (2014). ‘Interpreting’ polychrome paintings
using DStretch. Rock Art Research, 31, 101–104.
Harman, J. (2008). Using decorrelation stretch to enhance rock art images. Abstract
retrieved from http://www.dstretch.com/AlgorithmDescription.html.
Harris, E. C. (2013). Urban sites and the stratigraphic revolution in archaeology. Papers
from the Institute of Archaeology, 23(1), 20. doi:http://doi.org/10.5334/pia.439
Haskovec, I. P., & Sullivan, H. (1989). Reflections and rejections of an Aboriginal artist. In
H. Morphy (Ed.), Animals into art (pp. 57–74). London: Unwin Hyman.
Hutton, J. (1788). Theory of the Earth. Transactions of the Royal Society of Edinburgh,
1(2), 209–304.
Keyser, J. D. (2001). Relative dating methods. In D. S. Whitley (Ed.), Handbook of rock art
research (pp. 116–138). Walnut Creek, CA: Altamira.
Page 17 of 19
PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.
Le Quellec, J. -L., Duquesnoy, F., & Defrasne, C. (2015). Digital image enhancement with
DStretch: Is complexity always necessary for efficiency? Digital Applications in
Archaeology and Cultural Heritage, 2(2015), 55–67. http://dx.doi.org/10.1016/j.daach.
2015.01.003i
Lewis, D. (1988). The rock paintings of Arnhem Land, Australia. BAR International Series
415. Oxford: BAR.
Maddock, K. (1970). Imagery and social structure at two Dalabon rock art sites.
Anthropological Forum, 2, 444–463.
Magar, V., & Davila, V. (2004). Considerations on the dating of rock art from the Sierra de
San Francisco, Baja California. Rock Art Research, 21(2), 129–136.
Mazel, A. D., & Watchman, A. L. (2003). Dating rock paintings in the uKhahlamba-
Drakensberg and the Biggarsberg, KwaZulu-Natal, South Africa. Southern African
Humanities, 15, 59–73.
Murray, P., & Murray, L. (1972). A dictionary of art and artists. Ringwood, AUS: Penguin.
Rosenfeld, A., & Smith, C. (1997). Recent developments in radiocarbon and stylistic
methods of dating rock-art. Antiquity, 71, 405–411.
Russell, T. (2000). The application of the Harris matrix to San rock art at Main Caves
North, Kwaazulu-Natal. South African Archaeological Bulletin, 55, 60–70.
Russell, T. (2012). No one said it would be easy. Ordering San paintings using the Harris
matrix: dangerously fallascious? A reply to David Pearce. South African Archaeological
Bulletin, 67, 267–272.
Sebeok, T., & Sebeok, J. U. (1981). ‘You know my method’: A juxtaposition of Charles S.
Peirce and Sherlock Holmes. In T. Sebeok (Ed.), The play of musement (pp. 17–52).
Bloomington: Indiana University Press.
Swart, J. (2004). Rock art sequence in uKhahlamba-Drakensberg Par, South Africa. South
African Humanities, 16, 13–35.
Page 18 of 19
PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.
Edward Harris
Robert G. Gunn
Monash Indigenous Studies Centre, Monash University; ARC Centre of Excellence for
Australian Biodiversity and Heritage
Page 19 of 19
PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.