Quaternary International xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

Quaternary International
journal homepage: www.elsevier.com/locate/quaint

A matter of diversity? Identifying past hunter-gatherer aggregation camps

through data driven analyses of rock art sites
Oliver Vogels∗, Eymard Fäder, Tilman Lenssen-Erz
University of Cologne, Institute of Prehistoric Archaeology, African Archaeology, Jennerstr. 8, 50823, Cologne, Germany

A R T I C LE I N FO A B S T R A C T

Keywords: During the last decades, ethnographic observations taught hunter-gatherer archaeologists to incorporate social
Rock art archaeology phenomena into their interpretative models of land uses, including rock art. Although ethnographic analogies
Quantitative analysis always allow only a rough approximation to the past, strategies such as the (seasonal) aggregation and dis-
GIS persion of San groups, and related exchange systems, have become an integral part of the interpretation of
Computational archaeology
variability in the spatial organisation of sites and the excavated material culture. Regardless of a long-standing
Namibia
Brandberg/Dâureb
dispute about its potential meaning, one widely accepted explanation for making rock art is that it was part of
the social contexts of past hunter-gatherer aggregation events. While motifs such as the jointly performed trance
dance seem to support this view, methods to identify aggregation and dispersion events within rock art on a
larger scale are poorly developed. The objective of this paper is to better understand the organisation of the rock
art sites at the Dâureb in Namibia using GIS and computational archaeology in the light of hunter-gatherer
aggregation and dispersion strategies.

1. Introduction surrounding Namib desert. However, as moisture in the form of rain or

fog is protected from rapid seepage by the rock, minimal quantities of
1.1. The study area 1 mm of rain are sufficient to collect hundreds of litres of water in pits
and scours (Pager, 1980: 352). Thus, the mountain is not only a fa-
The study area considered here is located in northwestern central vourable habitat in terms of climate but also hydrologically. This is
Namibia, at the Dâureb massif. Geologically, the Dâureb belongs to a expressed in the presence of various plant taxa usually native to Bots-
series of granitic intrusions (Diehl, 1990, Schmitt et al., 2000). Over wana or Angola, i.e. areas with much higher precipitation, or which are
millions of years, water formed some deep, cutting gorges and several endemic to the Dâureb (Craven and Craven, 2000; Burke, 2005). Si-
smaller ones. Since the formation of the Namib desert in the Late multaneously, the flora is largely protected from herbivores, as the
Miocene, erosion was limited to variations in temperature, causing mountain range prevents large mammals from entering. The largest
spheroidal weathering (Gutiérrez and Gutiérrez, 2016: 120), that in mammals in the habitat are the small klipspringer (Oreotragus oreo-
turn created the present aspect of the terrain, resulting in blocks and tragus) and the leopard (Panthera pardus), which, in addition to various
boulders as well as granite debris being scattered across the entire small mammals such as the rock hyrax (Procavia capensis), dassie rat
massif. Due to erosion, the foot of the Dâureb is rather steep, whereas (Petromus typicus) and mountain hare (Pronolagus randensis), char-
the area above about 1800 m, often referred to as the “Upper Dâureb”, acterise the wildlife of the Dâureb (Breunig, 1988: 60). Taken together,
is topographically relatively even. While occasional rains quickly run geology, vegetation and water availability are obvious subsistence re-
off the slopes, they remain within numerous plateaus in higher areas. lated reasons for humans to visit the Dâureb mountain oasis since the
These naturally collect water and store it over weeks or months in sinks, Middle Stone Age (Breunig, 1989: 26). Determining social and cultural
cracks or scours (Lenssen-Erz, 2001: 26). The exact inter-annual climate reasons as well as nature's impact on livelihood is the subject of various
diversity in the Dâureb is still little understood because field surveys in research programmes and an ongoing debate (Lenssen-Erz, 2000. 2001.
the massif are usually confined to few days or weeks (Craven and 2008; Kinahan, 2005. 2016. 2017).
Craven, 2000). With annual rainfall between 110 mm and 0 mm, the
interior of the Dâureb should correspond to the arid landscape of the

∗
Corresponding author.
E-mail address: oliver.vogels@uni-koeln.de (O. Vogels).

https://doi.org/10.1016/j.quaint.2020.05.057
Received 28 June 2019; Received in revised form 19 May 2020; Accepted 30 May 2020
1040-6182/ © 2020 Elsevier Ltd and INQUA. All rights reserved.

Please cite this article as: Oliver Vogels, Eymard Fäder and Tilman Lenssen-Erz, Quaternary International,
https://doi.org/10.1016/j.quaint.2020.05.057
O. Vogels, et al. Quaternary International xxx (xxxx) xxx–xxx

1.2. Theoretical and methodological considerations sociogram of the past hunter-gatherers on a more general basis. On a
structural level, on the other hand, the art allows to investigate the
Perhaps because processual methodology was favoured in archae- communicative methods of the past discourse.
ological research in the 1960s and 1970s, and perhaps due to the sheer One of the advantages of rock art over most other archaeological
amount of information embedded in rock paintings, the advanced early sources is that the pictures are today still in the same place as their
scholarly approaches in these days aimed at comprehensive doc- creators chose according to their own relevant considerations (Deacon,
umentation and quantitative analysis of southern African hunter-gath- 1988; Chippindale and Nash, 2004; Lenssen-Erz, 2008; Lewis-Williams,
erer art (Vinnicombe, 1967, Lewis-Williams, 1972. 1974, Pager, 1971. 2010). Therefore, rock art is particularly apt for analyses using geos-
1974). Statistical methods revealed differences in the regional dis- patial methods (Hartley and Wolley-Vawser, 1998; Russell, 2012;
tribution of rock art and semantic relationships between certain motifs. O'Sullivan, 2019), multivariate statistics (Sauvet and Wlodarczyk,
But soon it became apparent that a better understanding of the paint- 1995, 2006; McCall, 2010; Tratebas, 2012) and supra-regional con-
ings was possible only by incorporating traditional mythology and the siderations, as e.g. the integration of rock art in the “Kalahari debate”
world view of extant hunter-gatherer societies – the San (Vinnicombe, shows (Manhire et al., 1986; Smith and Ouzman, 2004; Lander and
1972. 1975. 1976. Lewis-Williams, 1981. 1982). The principal value of Russell, 2015). It must be emphasized here that quantitative analyses
the overall debate was that neither structuralist, quantitative nor are not an explanatory tool, as statistical methods do not explain the
semiotic studies alone were able to comprehensively and convincingly patterns they produce, and thus may lead to false conclusions (Lewis-
understand rock art in southern Africa. Instead, it is the scientific dis- Williams, 1984). Nevertheless, we follow Russell's (2012: 37) argument
course among these diverging approaches that has been fundamental that space in rock art research has not yet been exploited analytically to
for the development of rock art research over the past decades (Conkey, its full potential. Regarding the Dâureb this is even more so. Its western
2001: 279). part has been surveyed intensely over decades so that today 840 sites
Since the 1980s, rock art research became deeply rooted in hunter- are comprehensively recorded and published (Pager, 1974–2006). Only
gatherer archaeology in southern Africa (Mitchell, 2005a) as it became very few small rock art sites may have been missed by archaeologists in
obvious from the ethnographic record that rock art and excavated this region, therefore the rock art documentation provides a solid da-
material culture relate to a continuum of interrelated social and eco- taset for spatial analyses.
nomic strategies of past hunter-gatherer societies (Lewis-Williams,
1982; Ouzman and Wadley, 1997, see also Mitchell, 2005b). But 1.3. State of research
whether strategies of both kinds of media are related or complementary
often remains uncertain, mainly because there is hardly any evidence of Over the last fifty years, Later Stone Age Archaeology in Namibia
a direct chronological relationship between rock art and material cul- has aimed to follow the research strategy outlined above, towards a
ture on site. more complete archaeological model by associating hunter-gatherer
Based on ethnographic observations, excavated stone or bone arte- archaeology (Wendt, 1972; Richter, 1991; Breunig, 2003) with hunter-
facts, but also beads and mobile art, allow to reconstruct to some degree gatherer art (Scherz and Breunig, 1975. 1986; Pager, 1974–2006;
economic and social activities that were conducted on a site (Yellen, Lenssen-Erz, 1997) and with structural aspects of the physical land-
1977; Wadley, 1987; Mazel, 1988). In visual representations however, scape (Lenssen-Erz, 2001).
each individual figure can theoretically have its own meaning and Today, more than 1000 rock art sites are known within the Inselberg
could have been created in its own situational and presentational of the Dâureb mountain, of which 840 have been documented in full by
context. Interpreting rock art thus is particularly problematic in regions Harald Pager (1989, 1993; 1995; 1998; 2000; 2006). Due to Pager's
where no living painting traditions exist, i.e. where no further (ethno- unexpected death in 1985 the northeastern part of the Dâureb remains
historic) cultural knowledge about their meaning or situational contexts unrecorded, an area that contains ca. 15–20% of the mountain's art. By
can be acquired. In such cases, any approach to rock art must also rely annotating all individual figures of the documented Dâureb rock art
on ethnographic observations, bringing in all the blurriness of analogy (n = 39,075; about 6000 further individual figures in the Numas Gorge
(Taçon and Chippindale, 1998; Bednarik, 2011). This is not only the still await publication) in a relational data set, a distinctive oeuvre was
case with European Upper Palaeolithic rock art (Lewis-Williams, 1991), designed and established since the late 1980s (Lenssen-Erz, 1989, 2001)
but also with Later Stone Age (LSA) rock art in southern Africa. One of that invites in-depth analyses with multivariate statistical methods
the main questions in rock art archaeology today hence is concerned (Sauvet and Wlodarczyk, 1995., 2006). In the course of the outlined
with, or should be concerned with, how “formal” and “informed” research strategy, understanding the culture and chronological con-
methods (Taçon and Chippindale, 1998) can be brought together in nections between material culture and rock art are of particular im-
order to obtain new hypotheses and approaches to enrich the scientific portance and have been assessed with various approaches.
discourse. Direct dating attempts included direct radiocarbon dating and paper
Parkington and Mills (1991: 357) describe the spatial design of a chromatography. Acquiring an absolute age via traditional radiocarbon
southern African hunter-gatherer campsite as a “sociogram of hunter- dating of organic binder (lipids or other proteinaceous substances),
gatherer society”. Similarly, they argue rock art sites to be repositories sampled at rock paintings from six different sites at the Dâureb failed
with “socially informed images”. We argue that rock art sites yield due to absence of binder in the paint (Conard et al., 1988: 466). Paper
sociograms of hunter-gatherer society (including their entire life- chromatography conducted by Denninger et al. (1986) was more suc-
world), as it involves a discourse across generations. We further argue cessful in determining the presence of binders in the paint. However,
that not only the images are socially informed. Also, the choice of lo- the causes for the degradation of amino acids are unknown, Denninger's
cation, and semantic components such as superpositioning, scenic re- absolute time estimates thus cannot be used as an absolute chron-
lationships (cf. Lewis-Williams, 1972. 1992; Lenssen-Erz, 1992), but ological argument (Richter, 1991: 208–209; Thackeray, 1983). Re-
also single figures, and interactions between figures (Lenssen-Erz et al. cently, Bonneau et al. (2011. 2017) presented an improved protocol for
14
in press) are consciously used and thus culturally charged and recur- C dating for rock paintings containing carbon. Nevertheless, this
sively iterated or modified. This way, rock art functions as a commu- dating method is destructive to paintings and was therefore not pursued
nicative system – a discourse – that “updates” the present through re- in the study area.
cursive iteration or modification of identity, behaviour and also the Since an absolute date could not be determined, estimates about
communal cognitive map of the Dâureb in the light of the past (cf. age, chronology and cultural background of the rock art was based on
Lenssen-Erz, 2008). Instead of focusing on single paintings, analysing indirect dating techniques, residing on excavations at more than twenty
the Dâureb rock art as a complete population provides insights into the rock art sites in the Dâureb, Twyfelfontein region and the Erongo

2
O. Vogels, et al. Quaternary International xxx (xxxx) xxx–xxx

(Martin and Mason, 1954; MacCalman, 1965; Wendt, 1972; Jacobson that the majority of paintings in the Dâureb relates to the late Holocene
and Vogel, 1975; Jacobson, 1978; Wadley, 1976; Wadley, 1977; Vogel Later Stone Age occupations. Fig. 1 shows a summed probability dis-
and Visser, 1981; Burgess and Jacobson, 1984; Kinahan, 1984. 1990. tribution (Williams, 2012) of radiocarbon dates (excluding Early/
1991; Richter, 1991; Smith et al., 1995; Breunig, 2003; Breunig et al., Middle Stone Age dates and the early Holocene date) sampled in the
2019). According to these archaeological studies, almost the entire course of the archaeological investigations discussed above. The ma-
material culture emerged during Holocene Later Stone Age (LSA) oc- jority of radiocarbon dates lies between 4000 and 1800 calBP, marking
cupations between ca. 8000 calBP and historical times. Radiocarbon the widely assumed main painting phase in northwestern central Na-
dates were calibrated after Bronk Ramsey (2008) without normalisation mibia (Richter, 1991; Kinahan, 1991; Lenssen-Erz, 2001; Breunig,
(following Weninger et al., 2015) using R package rcarbon (Bevan and 2003). Due to a lack of a more fine-grained chronology, Kinahan re-
Crema, 2018) and ShCal13 calibration data (Hogg et al., 2013). Ex- cently attempted a microstratigraphic approach based on the observa-
ceptions in the material culture chronology are a few Middle Stone Age tion made at two sites that polychrome human figures superimpose
assemblages (Richter, 1991: 77–78, Breunig, 2003: 119–120; Breunig monochrome paintings (Kinahan, 2017: 556). Other studies (e.g.
et al., 2019: 183–184), a late Pleistocene date at 17786 calBP Lenssen-Erz, 2001: 252; Lewis-Williams, 1972. 1974) could show that
(17987–17578 at 95.4% probability; Breunig et al., 2019: 183) and two superimpositioning of figures often seems to have a semiotic back-
early Holocene LSA assemblage dated to 10669 calBP (10741–10584 at ground and may therefore be inconclusive for chronological analyses
95.4% probability; Richter, 1991: 129–130) and 11303 calBP (see also Pearce, 2010).
(11596–11557 at 95.4% probability; Breunig et al., 2019: 183) re- An initial approach (Lenssen-Erz, 2001) to the social and economic
spectively. Pottery findings occurred in the LSA assemblages around background of the Dâureb rock art took into consideration 300 (at that
2000 calBP in northwestern central Namibia (Richter, 1991: 192–193), time fully documented) sites (Pager, 1989, 1993, 1995), setting up an
possibly also domestic stock (Pleurdeau et al., 2012; Kinahan, 2016). “idealized land use model” that reflects Later Stone Age hunter-gatherer
During the last 1000 years the Dâureb has been inhabited by the so- mobility on the basis of structural spatial information, empirical ob-
called “Brandberg Culture” (Rudner, 1957: 33), a pastoral archae- servations and the ethnographic record (Lenssen-Erz, 2001: 324; 2004).
ological culture which has been determined by a technological shift According to archaeological research, few archaeological horizons
towards macrolithic tools (Breunig, 2003), construction of stone circles contain large numbers of blanks but only a small number of tools
(Kinahan, 1991: 53–55) and the presence of cattle (Breunig, 2003: (Kinahan, 1984; Breunig, 2003). In the absence of information derived
239). from artefact assemblages relating to social and logistic activities, the
The rock art itself is also chronologically indicative. With few ex- argument about the placement of rock art was based on a hypothetico-
ceptions, the Dâureb rock art on which we will focus, is made in the deductive model of economic and social activities derived from recent
hunter-gatherer fine-line style (Lenssen-Erz and Vogelsang, 2005; Smith hunter-gatherers in southern Africa (Lenssen-Erz, 2001: 267–270).
and Ouzman, 2004; Orton, 2013) and mainly depicts humans and wild Owing to the fact that painting traditions no longer exist in the research
animals (Scherz and Breunig, 1986; Pager, 1974–2006, Lenssen-Erz, area and therefore cannot be consulted to better understand the relation
2001). Exceptions are a couple of finger-paintings on the upper-most between rock art and the locations where it was placed, the theoretical
layer of superimpositions (Lenssen-Erz and Vogelsang, 2005: 59) and model was built on “dichotomic pairs” (Lenssen-Erz, 2004: 143–148).
four depictions of fat-tailed sheep in the prevailing fine-line style The first couple of pairs treat preference in economic and social activ-
(Lenssen-Erz, 2000: 112; Scherz and Breunig, 1986, Fig. 69), which are, ities: secular vs. religious; as well as spatial remoteness and visibility:
however, contrasted by 6409 depictions of wild animals. private vs. public (Lenssen-Erz, 2004: 145). Another pair treats the
Painting style, depicted topics, and the radiocarbon record suggest number of paintings as a rough gauge for occupation intensity. Based on

Fig. 1. Summed probability of calibrated radiocarbon dates from the Dâureb (coloured by archaeological culture) and wider regional archaeological sites (grey).
Radiocarbon dates were calibrated after Bronk Ramsey (2008) without normalisation (following Weninger et al., 2015) using R package rcarbon (Bevan and Crema,
2018) and ShCal13 calibration data (Hogg et al., 2013; artwork: Oliver Vogels).

3
O. Vogels, et al. Quaternary International xxx (xxxx) xxx–xxx

the assumption that each painting requires a certain amount of time, a 1.5. Materials and methods
higher number of paintings at a given site either requires many people
at once or longer occupations: regular visits or long-termed stay; few Methodologically this paper expands on a previous study (cf. section
figures could be made during a singular stay or short-termed visits. With 1.3) by including approaches of computational archaeology and sta-
the Dâureb being characterised both as a “sacred landscape” and a tistical rock art analysis. It is based on rock art recording (Pager,
“resource spot”, the next pair treats the basic motivation of visiting the 1974–2006) and spatial site characteristics gathered during field survey
massif: materialistic vs. idealistic (Lenssen-Erz, 2004: 145). The materi- in the Dâureb (Lenssen-Erz, 2004) with respect to theoretical aspects of
alistic motivation addresses rock art sites that were visited – and ethnographic observations and archaeological results. Recording
painted – because people entered the Dâureb for exploitation of re- methods and the data scheme have been presented elsewhere (Lenssen-
sources (water, plant food, small game). In contrast, the idealistic mo- Erz, 2001. 2004. 2006) and will not be discussed here in detail. The
tivation implies that sites were painted because people visited the complete data set including detailed description has been published
mountain mainly for certain spiritual reasons. Taken together, these digitally as an open-data online repository hosted at IANUS For-
four dichotomic pairs generate a complex matrix defining seven dif- schungsdatenzentrum and is freely accessible at https://dx.doi.org/10.
ferent classes of site use (Lenssen-Erz, 2001: Tables 44–46 2004: Tables 13149/r5f.3fyhip-x (Lenssen-Erz and Vogels, 2017). The main trans-
7.7–7.8): formation performed here is from quantifiable data to a qualifying ar-
gument. The formal approaches to studying rock art elements being
A landmark site used are founded on GIS (geographic information system) and geosta-
B short-term living site tistics. These approaches are quantitative and formal in nature, aiming
C long-term living site to test whether existing assumptions about rock art and its spatial
D aggregation camp context are verifiable, and to set up new hypothesis on the human use of
E casual ritual site space in rock art landscapes. In this respect, rock art – as a field of study
F deliberate ritual site within archaeology – is very useful, even gratifying, because it has some
G “sanctuary, hermitage” advantages over all other archaeological sources of information, espe-
cially when analysing it at the juncture with landscape. Our methodo-
This site classification scheme was mainly related to spatial site logical aim thus is to further develop methods that combine “dirt ar-
characteristics, the placement and quantity of rock art within the sites. chaeology” and rock art analysis, an approach that has been repeatedly
However, the classification is idealized, reflecting ethnographically demanded (e.g. Ouzman and Wadley, 1997: 387; Mitchell, 2005b: 165).
observed economic and social needs rather than individual site char- It may sound commonplace but should be emphasized none the less that
acteristics (Lenssen-Erz, 2001: 264–267). Furthermore, the depicted rock art is part of a landscape and also that the landscape makes up part
content has been included only on a summative level (Lenssen-Erz, of the meaning of rock-art (e.g. Bradley et al., 1994; Schaafsma, 1985;
2004), considering the absolute number of figures per site as well as an Ouzman and Wadley, 1997).
“action index” that represents the amount of body activity per figure,
averaged per site (Lenssen-Erz, 2001: 65–72; 280), where, e.g. a low 2. Identification of aggregation camps by spatial site
action index is also indicative of a high ratio of animals in the paintings characteristics
of a site.
Rainfall occurs highly locally in the central Namib, at the eastern
margin of which the Dâureb is located. Furthermore, the granitic massif
1.4. Aims and objectives has an extreme topography that raises up to 2000 m above the sur-
rounding area. Taken together, these characteristics render the Dâureb
This paper aims at investigating whether the spatial distribution of with clearly different environmental areas. Due to this it must be as-
rock art at the Dâureb can be explained in terms of aggregation and sumed that past inhabitants visited each location for a specific reason.
dispersion strategies by integrating spatial site characteristics, ex- Among these reasons are resource exploitation as well as events related
cavated material culture and rock art motifs. Today it is widely assumed to the production of rock art. Therefore, the question arises as to how
in hunter-gatherer archaeology that differences in artefact assemblages far artefact assemblage, spatial (structural) site characteristics and rock
reflect variability in economic activities. For instance, Richter (1991) art motifs are interrelated.
observed LSA assemblage variability in frequency per tool type, in Lee (Lee, 1972a. 1979), Marshall (1976), Yellen (1977), Bartram
northwestern central Namibia reflecting differences in economic ac- et al. (1991), and others observed that camp size and the spatial or-
tivities. These, in turn, indicate specific patterns of land use in the wider ganisation of groups cannot be regarded as uniform throughout the year
Dâureb area. Based on the observation that both, artefact assemblages especially in regions where seasonal variability is pronounced, as ob-
(Richter, 1991; Kinahan, 1984. 1991; Breunig, 2003) and rock art served in southern Africa. Ethnography shows that the degree of dis-
(Pager, 1974–2006) reflect lifeways of mobile hunter-gatherers, we persion and aggregation is highly dependent on the availability of re-
argue that both kinds of media were made by the same archaeological sources, which means in contrast that in areas where resources are
culture, albeit our knowledge of chronology and time-depth of painting more abundant aggregation and dispersion patterns may be less pro-
activities is based on statistical approximation of radiocarbon dates (see nounced. Besides this, it has been emphasized repeatedly that motiva-
section 1.3; Fig. 1). In this we follow the epistemological principle of tion for group aggregation is not only of economic but also of social
the “inference to the best explanation” (Lipton, 2000) which ac- nature (Lee, 1972b; Marshall, 1976; Silberbauer, 1981). Among !Kung
knowledges the possibility of other interpretations but opts for the most San, Richard Lee observed group aggregation to be a period of large-
straightforward one that does not require to construe a possibly con- scale trance dancing and curing, initiations, trading, storytelling, and
tradictious scenario. Therefore, one objective of this paper is to de- marriage brokering (Lee, 1972b: 345). But also, the workload per
termine differences in the rock art across sites and whether these can be person increases with group size: hunting and gathering activities re-
similarly explained by means of functional variability. Functional quire higher levels of cooperation and coordination (Lee, 1972b: 347).
variability here is not understood as a function of the placement of rock San groups of over 100 people, e.g. can be sustained for months but
art in general (Lewis-Williams, 1982; Vinnicombe, 1976: 356), or only at the cost of an increasingly high input of subsistence effort.
within its syntax or meaning (e.g. Lewis-Williams, 1972), but as dif- According to ethnographic observations the disadvantage of large
ferences in site use. groups is an increased probability of conflict. Especially large San ag-
gregations are “inherently unstable” and require stabilisation via trance

4
O. Vogels, et al. Quaternary International xxx (xxxx) xxx–xxx

Fig. 2. Summary statistics of number of depicted figures per site. Most sites have less than 260 depictions (dashed red horizontal line). Sites with more than 260
figures are considered as “main painting sites” (Artwork: Oliver Vogels). (For interpretation of the references to colour in this figure legend, the reader is referred to
the Web version of this article.)

dancing or other ritual activities (Lee, 1972b: 347). In case production larger hunter-gatherer groups has been observed to not only being di-
of rock art was part of such social or ritual activities, higher numbers of rected at negotiation of social tension only. The largest accumulations
paintings indicate longer stays or more frequent visits of a site. How- of !Kung San groups, e.g. came together every four or five years during
ever, a closer look at the distribution of quantities of paintings at the the winter (dry season) for men's initiation called choma. The number of
Dâureb sites shows that the majority of sites has rather few depictions groups participating in the choma was especially high because at least
(Fig. 2 bottom), as 625 sites yield 55 paintings or less (3rd quartile in seven boys of the correct age were needed to make it worthwhile to run
Fig. 2). These 625 sites contain only a total of 22.4 percent of the the elaborate six-week-long initiation programme (Lee, 1972b: 345).
overall amount of recorded paintings (n = 45,662, including sites in The families of the initiates camped together and provided food for the
the Numas Gorge). In contrast, the majority of paintings accumulated at camp during the six-week initiation period. According to Lee, the
a relatively small number of locations (Fig. 2, top right). Visual in- number of people in the adjacent camps must have been well over 200
spection shows a break in the trend at 260 paintings (dashed red hor- people, in cases when twenty or more boys were initiated together
izontal line in Fig. 2). Above this limit, very few sites yield extra- (ibid.).
ordinarily high numbers of paintings. At these 31 sites (4.2%), 32.8% of Such trance-dancing ceremonies also brought together medicine
the overall amount of paintings were placed. Given that larger ag- men from across the entire region because the curing medicine was
gregations entail an increase in social negotiation, the latter 31 sites thought to be especially effective when many performers entered a
were particularly dedicated to paintings activities (“main painting trance at the same dance (ibid). Such big trance-dances required a good
sites”), and thus also candidates for past group aggregation. organisation, as they could last between 12 and 36 h. Food and water
It must be noted that terrain in the Dâureb is very rugged and open had to be organised for the singers, dancers, and trance performers. Lee
space suitable to accommodate larger groups of people is scarce argued that at least fifty or more adults had to be in the camp when
(Fig. 3). Therefore, including structural site characteristics in the ana- fifteen or twenty adults were participating in trance-dancing at any one
lysis is reasonable, foremost shelter size (“spatiality”; Lenssen-Erz, 2001: time (ibid).
285–286; Lenssen-Erz, 2004: 146). But intriguingly, taken together, Following Lee's ethnographic observations, also the intensely
spaciousness and painting intensity appear to be mutually exclusive: painted but small sites discussed above (Fig. 4, top left) are eligible
only three out of 28 sites with more than 250 painted figures are spa- aggregation sites, in case they are located within site clusters. One of
cious enough to provide shelter for larger groups (Amis 10, Karoab 2 the most immediate problems in determining clustering is the choice of
and Umuab 21, highlighted in Fig. 4). scale (Barceló et al., 2015: 359). Whether clustering exists in the spatial
Another series of sites would be well suited for group aggregation as distribution of sites has been determined at different scales using Ri-
they are especially spacious but contain far less amounts of paintings: pley's K-function (Ripley, 1976; Conolly and Lake, 2006: 166–168) with
Hungorob 28 and 43, Naib 100 (I 100) and 170 (I 170), Numas 164 and 999 permutations using ESRI's tool “Multi-Distance Spatial Cluster
Orabes 7 (highlighted in Fig. 4). In contrast, other sites were painted Analysis (Ripleys K function)” in ArcMap. The resulting graphical
much more intensely, i.e. they yield a higher quantity of depictions, but model (Fig. 5) shows that site clustering is generally high at the Dâureb,
are physically too small to accommodate larger groups of people inside with a slight peak around 200 m, for the observed average distance (red
sheltered room (Fig. 4, top left). In this respect it is worth mentioning line) is way above the expectation (blue line). The result is statistically
that (Yellen (1977): 134) noted concerning space that individuals significant, as the observed average distances (red line) are higher than
cannot be treated as simple aggregations. Two people do not take twice the confidence envelope (grey dashed lines). The high degree of spatial
as much space as one, and it is not very helpful to divide total site area clustering indicates that the Dâureb is one large settlement area, which
into number of square meters per person (Yellen, 1977: 134). There- makes sense from a regional perspective. Delineation of site clusters on
fore, space inside a rock shelter alone is not a sufficient argument for an smaller scale has been computed in a second step using ArcMap and the
aggregation camp site in the Dâureb. largest empty circle approach (LEC) after (Zimmermann et al. 2009; see
It is the ethnographic record again that allows for some insights also Schmidt and Zimmermann, 2019; Maier and Zimmermann, 2017;
about aspects of group aggregation among hunter-gatherers today. Kretschmer et al., 2016). Based on Thiessen polygons created from rock
According to (Lee (1972a, b): 345), increased ritual activity among art site locations, the approach considers the distance of each polygon

5
O. Vogels, et al. Quaternary International xxx (xxxx) xxx–xxx

Fig. 3. Terrain in the Dâureb is characterised by countless rocks and boulders. The fact that only few of them have been painted (marked with arrows in this frame),
indicates that painted shelters and boulders were chosen by certain conditions (view to northwest; photograph: Marie-Theres Erz).

node to the next nearest site – the largest empty circle. Interpolation of structural characteristics, these places can be addressed as “deliberate
the radii of the largest empty circles via Kriging (Oliver and Webster, ritual sites“, i.e. hot spots of social or cultural activities. Site clusters
1990; Conolly and Lake, 2006: 97–100) allows to model a density-grid. including such a deliberate ritual site can thus be qualified as “extended
Isolines created on this grid then render certain intervals of site den- aggregation camps”, as they combine place for several groups and space
sities. The increase in sites per isoline occurred to be crucial to de- for group activity.
termine the “optimum isoline” (Schmidt et al. in press; Zimmermann Another noteworthy aspect is the existence “deliberate ritual sites”,
et al., 2009), which computed for the Dâureb sites results in a 200 m richly in paintings (260 and 800), but little spacious and isolated rock
isoline (Fig. 6). shelters. Such sites are Gaa'seb 12, Karoab 13, Karoab 24, and Porters 1
Colouring the sites in Fig. 4 by their clustered/isolated state shows (red dots top left in Fig. 7). Either these sites were visited regularly by
that most of the intensely painted but little spacious sites are in fact particularly small groups, or they provided space in another way. No-
located within groups of other sites (highlighted in Fig. 7). From their teworthy in this respect is that some of these isolated (as well as some of

Fig. 4. Rock art site classification by number of paintings per site and spatiality (point size = quantity of figures; artwork: Oliver Vogels).

6
O. Vogels, et al. Quaternary International xxx (xxxx) xxx–xxx

Fig. 5. Graphical model of Ripley's K-function. Observed averaged distances (red line) above the expected distances (blue line) indicate spatial clustering, observed
distances below the expectation indicate dispersion, the confidence envelope (999 permutations) is represented by grey dashed lines. Spatial clustering is statistically
significant in case the observed values (red line) are above or below the confidence envelope (artwork: Oliver Vogels). (For interpretation of the references to colour
in this figure legend, the reader is referred to the Web version of this article.)

Fig. 6. Delineation of clustered and isolated sites (A) using largest empty circle approach (LEC; cf. Zimmermann et al., 2009), and their spatial organisation in the
Dâureb massif (B; artwork: Oliver Vogels).

the clustered) deliberate ritual sites provide space in terms of an ad- of hunter-gatherers to rest at a communal place. The presence of an
jacent open field. Defined as an “open level plain of at least 10 × 10 m” open field at some of the little spacious and isolated main painting sites,
(Lenssen-Erz, 1989: 345), the presence of an open field permits the especially at Gaa'seb 12, Karoab 13, Karoab 24, and Porters 1, supports
performance of social activities. But with respect to group aggregation, their potential for aggregation, which is expressed by their high number
it is worth mentioning that a level plain does not only allow to conduct of figures (Fig. 8).
social activities such as dancing or playing, but allows groups or bands Main painting sites and less intensely painted sites taken together

7
O. Vogels, et al. Quaternary International xxx (xxxx) xxx–xxx

Fig. 7. Rock art site classification by number of paintings, spatiality and spatial configuration (clustered/isolated sites, see. Fig. 4; point size = quantity of figures;
artwork: Oliver Vogels).

Fig. 8. Rock art site classification by number of paintings, spatiality and presence of an open field (point size = quantity of figures; artwork: Oliver Vogels).

8
O. Vogels, et al. Quaternary International xxx (xxxx) xxx–xxx

allow to better understand how the landscape was used and organised archaeological relevance. Unfortunately, the number of excavated sites
mentally by the LSA hunter-gatherers. Assuming that the main painting in the Dâureb is small (Breunig, 2003; Kinahan, 1991; Kinahan, 1984;
sites would have had the greatest attraction – being “hot spots” of social Burgess and Jacobson, 1984). Establishing a land use model for the
or cultural activities according to their sheer mass of depictions and entire inselberg based on material excavated at single sites would surely
thus their symbolic power – it must be assumed that these places have be misleading. We therefore aim at integrating rock art analysis in our
been visited most regularly. Due to the topography of the granitic land use model. But due to the poor chronological connection between
massif, formation of extraordinary spacious sites, such as cave-like rock art and material culture (see section 1.3), archaeological horizons
cavities or large boulders, is confined to the intermediate areas – the cannot be directly linked with specific rock art elements. We assume
slopes. While caves are confined to the middle ranges, many boulders that both, Later Stone Age horizons and rock art, were produced
tumbled down the slopes and accumulated in the main drainages. roughly between 3800 and 1800 years calBP by the same LSA culture
Extraordinary spacious places are not numerous and naturally limited (cf. section 1.3). The following overview of the archaeological material
to the lower and middle ranges of the Dâureb. In contrast, smaller thus deliberately neglects individual occupation phases and attempts to
blocks, boulders and rock shelters are scattered across the entire massif elucidate whether conclusions can be drawn about the motivation to
(cf. Fig. 3). But intriguingly, paintings are not similarly randomly visit individual sites from their entire assemblages. This approach is
scattered across the Dâureb but remarkably often occur in vicinity to necessarily an approximation since there may have been a continuum of
the main drainages (cf. Fig. 6). Formed by water over millions of years motivations.
before the current hyper-arid conditions came into being, drainages, i.e. For instance, (Yellen (1977): 77–84), among others, observed that
the bottoms of the gorges, are among the most prominent topographic the longer a Kalahari !Kung San site is inhabited, the greater the like-
features in the Dâureb. Whether this preference reflects human move- lihood that any particular activity will take place there. With the
ment, the presence of natural resources such as water, or whether it is number, duration, and diversity of people at a certain site, the diversity
related to economically or socially/ritually motivated visits (cf. of the artefact spectrum will increase as well. Therefore, an increase in
Lenssen-Erz, 2001: 285–286. 2004: 146) is a matter of ongoing re- the variety in archaeological assemblage spectra has been interpreted as
search. an indication of past LSA hunter-gatherer aggregation camps (Brooks,
Both, intensity of ritual activity and spatial requirements in Lee's 1984: 46).
descriptions(Lee, 1972a, 1972b) descriptions (see above) about public Archaeological excavations in the Dâureb (cf. section 1.3) were
trance-dancing and large-scale initiation rites fit the structural condi- conducted preferably at spacious sites with painting quantities way
tions of the most intensely painted sites in the Dâureb quite well. above average and, above all, promising sedimentation. Unfortunately,
Therefore, it is argued here that these sites were visited in similar estimating inter-site differences in artefact diversity is widely im-
contexts of group aggregation and deliberate ritual activities. But based possible because the tool production and trimming stages (châine op-
upon structural site characteristics alone one can hardly determine the ératoire) have been analysed only at one of 10 excavated LSA sites,
precise social or religious purposes for which groups or families came Hungorob 28 (Falls Rock), where Kinahan determined a high degree of
together. Furthermore, the ethnographic record is not as decisive as it core preparation and reduction activities, but does not mention tool
appears. (Keeney and Keeney (2013): 117) note that especially initia- reshaping processes (Kinahan, 1984: 20). Due to the near-absence of
tion rites for girls and boys and their spatial requirements remain ob- information about the châines opératoires, it also remains unknown
scure because few if any investigators have seen a complete ceremony, whether tools were locally manufactured. Inter-site differences must
and even those who were allowed to participate are still uncertain therefore rely on the tool spectrum alone. Intriguingly, the quantity of
about some aspects of its meaning. In addition, it would be misleading tools at excavated rockshelters is generally low compared to the
to assume that cultural characteristics and traditions have remained quantity of blanks (< 1%) and does not correspond well with the
unchanged during the last 2000 years. number of paintings (Table 1). It must be noted that the proportion of
All in all, the given classification of rock art sites by structural the excavated areas vary considerably between the individual sites, as
characteristics alone remains vague, because they do not allow for a does the thickness of the existing sediment. Artefact counts therefore
determination of one specific site function. In fact, most structural provide only a rough measure of the full in-situ assemblages (Breunig,
characteristics, such as space, an adjacent water source or open field, 2003; Kinahan, 1984. 1991). This picture is somewhat contrasted by a
allow for certain uses but do not preclude others. Hence, it cannot be unique single phase LSA open air site (Umuab B-2, Breunig, 2003:
excluded that different activities took place each time a site was visited. 186–187). Here, 1715 stone artefacts were recovered from an area of
In consequence, only structural characteristics with a limiting quality or 13 m2 and a depth of 10 cm. In contrast to the multiphase rock shelters,
certain combinations of spatial characteristics allow with some con- a higher ratio of 4.6% of the assemblage are tools.
fidence to determine sites with a single function. For instance, site class In general, the spectrum of LSA tools in the Dâureb (Breunig, 2003;
“landmark site” is the most discrete site class, as corresponding places Kinahan, 1984. 1991) is similar to the surrounding areas (Richter,
are composed of paintings on a vertical wall or on a flat surface,
without habitation possibilities in the near vicinity (Lenssen-Erz, 2001: Table 1
285: 2004: 145–146). But in general, people may have started into the Number of paintings, blanks and tools (excluding Middle Stone Age) from ex-
massif once for social or religious reasons, another time for resource cavated rock shelters in the Amis Gorge (Breunig, 2003) and Hungorob Gorge
exploitation (Lenssen-Erz, 2001: 282–284), each of these motivations (Kinahan, 1991: 28; Falls Rock = Pager's Hungorob 28; Snake Rock = Hun-
could thus result in certain areas being visited and specific motifs or gorob 43).
themes being placed – and still motivation could have changed during a Site paintings blanks tools
visit. An important methodological step therefore is to include the
painted motifs in the analysis. Amis 6 635 362 5
Amis 10 1073 32652 133
Amis 11 20 6431 16
3. Diversity of rock art and material culture as a measure of group Amis 12 186 139 4
aggregation Amis 17 40 396 5
Umuab 21 1066 439 9
3.1. Artefact diversity Umaub 28 207 218 4
Umuab B-2 0 1715 79
Falls Rock 353 15076 157
Site classes derived from spatial rock art site characteristics in the Snake Rock 416 2149 54
previous section require some degree of evaluation to estimate their

9
O. Vogels, et al. Quaternary International xxx (xxxx) xxx–xxx

Table 2 stays. It remains questionable whether these stays are related to ag-
Number of paintings, ostrich eggshell fragments, bead pre-products and beads gregation of bands in the Dâureb. In Botswana, where Yellen studied
from excavated rockshelters in the Dâureb (data taken from Breunig, 2003; the Dobe !Kung San, longer stays and higher diversity of people usually
Kinahan, 1991: 37); processing states of beads not listed separately by Kinahan occurred together, as several groups came together at one of the few
for Falls Rock (Hungorob 28) and Snake Rock (Hungorob 43).
waterholes during the dry season. Due to the spatial distribution of
Site fragments pre-products beads resources settlement patterns in the Dâureb seem to be slightly dif-
ferent. Spatial point analysis could show that the entire Dâureb can be
Amis 6 393 220 274
seen as one huge settlement area (cf. Figs. 5 and 6), within which a
Amis 10 1944 835 327
Amis 11 1609 282 19 continuum of site densities indicates a complex settlement pattern.
Amis 12 37 2 3 Whether the inselberg has been inhabited preferably during the dry or
Amis 17 22 0 4 wet season remains unknown. But generally, spatial characteristics of
Umuab 21 97 10 12
rock art sites (cf. section 2) suggest that people did not necessarily stay
Umaub 28 12 4 10
Umuab B-2 86 10 2
at the supposed aggregation camps but perhaps met there for group
Falls Rock 1329 NA NA activity while staying in smaller bands or families at other spacious but
Snake Rock 86 NA NA less intensely painted sites (cf. Fig. 4). This pattern seems to be partly
reflected in the material culture, for an increase in space corresponds
with an increase in material culture but not necessarily an increase in
1991). This is worth mentioning since in contrast to its hinterland, only rock paintings.
small game inhabits the Dâureb massif (cf. section 1.1). Ethnographic Crucial could be again the presence of an open field at the vicinity of
observations suggest that tools do not necessarily indicate logistical painted rock shelters. There is reason to argue that some intensely
activities. Yellen (1977) for instance, observed that gift manufacture is painted rock shelters, such as Umuab 21, were dedicated to painting
one of the defining attributes of !Kung aggregation camps. In hunter- activities, while many other social activities, such as dancing, manu-
gatherer societies today gift items of the reciprocal hxaro exchange facture of gifts or food processing took place outside the shelter.
network can be any non-food items, such as beads, arrows, ostrich Unfortunately, this question remains unanswered because so far ar-
eggshells, clothing, blankets, bowls or pots (Wiessner, 1982: 70). In the chaeological excavations were conducted only within painted cavities.
archaeological record, an increase in manufacture of certain stone im- In summary, the current state of research does not allow to determine
plements, such as arrow tips, may thus indirectly point to the produc- past aggregation camps based on material culture alone.
tion and exchange of non-food hxaro gifts (Wiessner, 1983: 260), and
hence to a spot with a high degree in social networking. However, 3.2. Rock art diversity
arrow tips manufactured for gift exchange are hardly recognisable in
the archaeological record. This is different with beads and beadwork Very influential in attempts to determine group aggregation sites
made of ostrich eggshell, which are easily recognisable. Today these are through analysis of decorative art were (Conkey's, 1980: 616) criteria
particularly popular gifts, preferably made and worn by women established from Upper Palaeolithic mobile art (cf. Bahn, 1982;
(Marshall, 1976: 304–306), and somewhat related to social and re- Bourdier, 2013; Davidson, 2012). Influenced by Yellen's considerations
ligious events such as trance-dancing (Marshall, 1969: 358). According on hunter-gatherer artefact diversity, Conkey (1980: 612) argued that if
to Wadley (1987): 8. 1992, gifts are particularly manufactured and different hunter-gatherer groups came together from different loca-
exchanged during events of group aggregation, occasions during which tions, such aggregation camps should be traceable by a higher variety of
also trance dancing most likely occurs (Lee, 1972b). In contrast, gift decorated bone and antler pieces, assuming that each band brought in
manufacture is reduced to a minimum during the dispersal phase its own styles and design elements. From this she expected that
(Ouzman and Wadley, 1997: 388) and also other ritual activities, such (Conkey, 1980: 616):
as the trance dance, are largely absent (Marshall, 1969).
Ostrich eggs are commonly used as water containers (Marshall, a) The diversity of pieces at Altamira will be greater than at any other
1976: 77), a high number of fragments (Table 2) may thus indicate the single (hypothesized dispersion) site.
need to store water rather than on-site bead production. While on-site b) Most design elements of the core Magdalenian engraving repertoire
bead manufacture can be deduced from the presence of bead pre-pro- […] will be present everywhere or at least be widespread.
ducts, the presence of beads may originate from bead manufacture c) There will, however, also tend to be design elements and structural
processes as well as gift exchange or beadwork being worn during principles that are unique to Altamira.
special social or religious events (Marshall, 1969: 358). The number of d) Elements and principles lacking at Altamira should tend not to occur
beads and pre-products is high at the main painting sites Amis 6, Amis elsewhere.
10 (cf. section 2) but also at Amis 11, and according to Kinahan also at
Hungorob 28/Falls Rock (Kinahan, 1991: 37), while beads and pre- The second expectation (b) is considered satisfied here, since the
products are rare at other excavated sites such as Umuab 21, a site rock art of the Dâureb, as far as it is known today, comes from a
similarly intensely painted as Amis 10. homogenous cultural context and a comparably short period of time (cf.
There is a strong correlation between the quantity of ostrich egg- section 1.3). Expectations c) and d) – uniqueness of rock art elements or
shell fragments and pre-products. Fragments of ostrich eggshell are entire sites – are difficult to determine at the Dâureb. The Brandberg-
therefore likely to reflect intensity of bead production rather than a Daureb Database (BDDB; Lenssen-Erz and Vogels, 2017) lists 299 dif-
need for storing water. The correlation between remains of ostrich ferent types of figural subjects, such as human or springbuck (Antidorcas
eggshell and tool production (cf. Tables 1 and 2) is less strong, mostly marsupialis), as the most basic types of motifs. Each subject is further
due to a lack of bead pre-products (and eggshell fragments) in Umuab described by different types of features, body postures, and actions.
21, Umuab 28 and Snake Rock (Hungorob 43). Taken together, subjects and features amount to over 70,000 possible
Since bead manufacture is a highly gendered task (Marshall, 1976: variants (299 different subjects x 250 different descriptive terms),
414), large amounts of ostrich eggshell pre-products indicate pro- which – theoretically – could occur together. If furthermore aggrega-
nounced female manufacture activities. Assuming that these processes tions of single figures in compositions or scenes (Lenssen-Erz, 1992;
are likely to occur during longer stays, similar to tool reshaping pro- Lenssen-Erz et al. in press), are considered, the Dâureb rock art – the-
cesses (Richter, 1991: 234), rather than during rests at daily foraging oretically – becomes almost as complex as speech: recurring terms
trips, the presence of female manufacture activities may indicate longer (subjects and features) can appear in different combinations, thus

10
O. Vogels, et al. Quaternary International xxx (xxxx) xxx–xxx

Table 3 individuals (figures) per site, a tendency that is represented in Fig. 9 by

Frequency of motifs condensed to their most basic a black solid regression line. All seven motif categories together occur
entity named “subject categories” (total = 39,075). only at two sites. Following Conkey's argument, it is not surprising that
Subject Category Frequency one of them, Amis 10, is one of the supposed aggregation sites (cf.
section 2), as it has the highest number of subject categories and the
Human 23176 highest number of paintings in the entire collection (n = 1073). Sur-
Animal 6409
prisingly, however, the other site, Naib 20 (I 20), has a similar amount
Imaginary Being 1154
Abstract Form 1329 of depicted subject categories but only a small amount of paintings
Item 683 (n = 116).
Plant 58 Measuring inter-site motif diversity requires to take the number of
Weather 16
individuals per category into account. For this task the sites were dis-
Remains 6250
tributed along a diversity index. Among the two most common indexes
in archaeological studies are the Shannon index and the Simpson index.
transporting an almost unlimited number of different messages. On site The Shannon index ranks rare species higher while the Simpson index
level the complexity is so high that each site is virtually unique. ranks frequent species higher. In the present study differences between
The remaining point in Conkey's scheme concerns site diversity (a). both indexes turned out to be negligible and only the Shannon index
It basically refers to presence/absence of rock art elements and the will be discussed here. Inter-site or relative diversity (x-axis in Fig. 10)
number of the respective individuals. Due to the overall complexity of reveals a dichotomous distribution to be present among the rock art
the rock art, diversity measurements here concern the subject, the fig- sites, between more specialised sites (left) and more diverse sites
ure's most basic description in the BDDB (Lenssen-Erz and Vogels, (right). Most striking is that specialised sites are invariably comprised
2017). At this point it is important to note that figures painted in the of a high number of human figures, while diversification of sites seems
Numas Gorge are not yet recorded, the following analysis therefore mainly driven by the number of animals. The fact that human figures
relies on a reduced data set of 39,075 single figures. Among these, the dominate the more “specialised” sites (H’ > 0.5) suggests that the
most common motif is human, i.e. human figures without any sexual topics were not placed accidentally in the landscape, but with respect to
characteristics (n = 15,937). Since many of these figures are well the place (cf. Lenssen-Erz, 2008). Following on from this, the question
preserved, sexual characteristics must have been deliberately omitted arises as to what the choice of placement of the motifs was based on. We
(“zero-marked”), perhaps indicating the presence of a third gender assume this choice was influenced either by the existing paintings, by
(Lenssen-Erz, 2002) or a specific social role. The difference in numbers activities that took place at certain spots, or both. The fact that sites
to figures with unambiguous sexual characteristics (man = 1916; with the highest amount of paintings also yield the highest number of
woman = 1674) is above ten thousand. Compared to these numbers the human figures (marked in a red circle in Fig. 10), argues for a combi-
most often occurring animal species, such as springbuck (n = 711), nation of both scenarios: certain places evoked certain activities, which
antelope (n = 558), or giraffe (n = 484) are rare. Due to this, motif in turn evoked the placement of paintings with a similar (human) topic.
representations were condensed here to their most basic species or Both together may have evoked re-occupation for similar activities and
category in order to highlight the most basic trends in the spatial dis- again the placement of additional paintings with similar topics. We
tribution of the rock art (Table 3). further argue that these “special places” did not come into being acci-
At first sight, measuring presence/absence of subject categories dentally but, at least partly and besides ‘spiritual’ qualities, due to
(“species richness”) gives the expected results (Fig. 9): the number of specific spatial characteristics, such as the ability to house larger groups
species (subject categories) increases with the total number of of people, good water availability, an adjacent open field for group

Fig. 9. “Species richness” per site: rock art site classification by number of paintings (excluding “remains” = non-identifiable motifs) and number of motifs per site
(artwork: Oliver Vogels).

11
O. Vogels, et al. Quaternary International xxx (xxxx) xxx–xxx

29

Fig. 10. “Rock art diversity”: rock art site classification by number of paintings (excluding “remains” = non-identifiable motifs) and the Shannon index of diversity
(number of figures per motif per site; artwork: Oliver Vogels).

activity, etc. characteristics become relevant, of which spatiality seems to be among

According to the ethnographic record, assemblage diversity some- the most important ones for the selection of locations for group ag-
what reflects the variety of activities on site and thus allows to compare gregation. On the one hand, sheltered space is a prerequisite to ac-
assemblages regarding the number of persons and the diversity of ac- commodate larger groups. On the other hand, group aggregation evokes
tivities (e.g. Yellen, 1977; Conkey, 1980; Brooks, 1984). If the same social activities which require a level plain in the Dâureb's rocky
would hold true for rock art elements in the Dâureb, diversity of motifs landscapes. While a series of sites are particularly spacious (cf. Fig. 4,
should increase with group size, length of occupation and/or number of right), only a few of them were extensively painted. From this it can be
occupations. If this was the case, supposed aggregation sites should be concluded that events evoking extensive painting did not take place at
located in the top right corner of Fig. 10. But this is not the case. The every spacious site. The presence of an adjacent open field might in-
most likely past aggregation camp is Amis 10, as it combines all (ethno- stead explain why certain sites with little sheltered space are rich in
) archaeological requirements: it is physically spacious and yields large paintings. In general, spacious shelters mostly occur at the Dâureb's
numbers of paintings and material culture. However, its relative di- steep slopes, while level plains are most abundant on its plateau. In
versity of depicted topics (subject categories) lies in the middle range, consequence, both characteristics rarely coincide, then becoming at-
because human figures are strongly overrepresented. The same is the tractive for group activities (cf. Fig. 3). This circumstance might explain
case with most of the other main painting sites (Fig. 10). Although they why only a small group of sites is as richly painted (> 600 figures) as
are characterised by a high variety of motifs (cf. Fig. 9), the main focus Amis 10. These are either similarly spacious (Karoab 2), allow to ac-
is on humans. Exceptions to this rule are Amis 6, where the number of commodate bands or families at shelters in their vicinity (Umuab 21,
humans is exceeded by the number of animals, as well as Orabes 7, Naib 55 = I 55, Naib 19 = I 19, cf. Fig. 7), or on an adjacent open field
Hungorob 28, Hungorob 43 and Ga'aseb 1. Diversity of motif categories (Karoab 13, Naib 55 = I 55, Naib 19 = I 19, cf. Fig. 8). The fact that
is thus contrary to tool type diversity: the more figures were placed at a Umuab 21 is located within a cluster of other sites and has an adjacent
given location, the more they centre around topics related to the people open field at which logistic activities may have taken place, could also
themselves. explain the near-absence of material culture. However, more archae-
Finally, the question remains which of the sites can be regarded as ological excavations at other sites are needed to test this hypothesis.
past hunter-gatherer aggregation camps. Generally, we assume that an Among this group of sites (encircled in Fig. 10) a large number of
increase in paintings corresponds with an increase in presence of people paintings is combined with a strong bias towards depictions of human
(either synchronically or diachronically). But neither the material cul- figures. This kind of specialisation contradicts the common archae-
ture nor the quantity of rock paintings can prove whether a synchro- ological understanding of band aggregation, as these should result in
nous presence of larger aggregations of people took place, or recurring maximum diversity. Here, an important difference between material
visits of smaller bands. Intensely painted, spacious sites may imply the culture and rock art becomes apparent. While material culture reflects
aggregation of larger groups. It cannot be concluded from this, how- an increase in exchange of goods and logistic activities, rock art com-
ever, that richly painted sites with little accommodation space imply municates the people's life-world, which in the case of band aggregation
frequent diachronic visits by small groups. At this point spatial site may have been the power and concerns of social life (Lee, 1972b).

12
O. Vogels, et al. Quaternary International xxx (xxxx) xxx–xxx

In this respect it is certainly not a coincidence that some very spa- Bartram, L.E., Kroll, E.M., Bunn, H.T., 1991. Variability in camp structure and bone food
cious sites, such as Hungorob 28 (Falls Rock) or Hungorob 43 (Snake refuse patterning at Kua San hunter-gatherer camps. In: Kroll, E.M., Price, T.D. (Eds.),
The Interpretation of Archaeological Spatial Patterning. Plenum Press, New York, pp.
Rock), yielded a large quantity of material culture (cf. Tables 1–2) and 77–148.
were less intensely painted, but with a much higher diversity of topics Bednarik, R.G., 2011. Ethnographic analogy in rock art interpretation. Man India 91,
(cf. Fig. 10). Either band aggregation events occurred less regular at 223–234.
Bevan, A., Crema, E.R., 2018. rcarbon: methods for calibrating and analysing radiocarbon
these places, or they were dedicated to activities related to daily life dates. https://github.com/ahb108/rcarbon 12/2019.
rather than specialised social activities. From this it could be concluded Bonneau, A., Brock, F., Higham, T., Pearce, D.G., Pollard, A.M., 2011. An improved
that, at some times, aggregation and dispersion occurred simulta- pretreatment protocol for radiocarbon dating black pigments in san rock art.
Radiocarbon 53, 419–428. https://doi.org/10.1017/S003382220003455X.
neously, which also explains the presence of large-scale and small-scale Bonneau, A., Pearce, D., Mitchell, P.J., Staff, R., Arthur, C., Mallen, L., Brock, F., Higham,
rock art sites within one large settlement area, the Dâureb massif. T., 2017. The earliest directly dated rock paintings from southern Africa: new AMS
radiocarbon dates. Antiquity 91, 322–333. https://doi.org/10.15184/aqy.2016.271.
Bourdier, C., 2013. Rock art and social geography in the Upper Paleolithic. Contribution
4. Conclusion
to the socio-cultural function of the Roc-aux-Sorciers rock-shelter (Angles-sur-
l’Anglin, France) from the viewpoint of its sculpted frieze. J. Anthropol. Archaeol. 32,
In the ethnographic record, aggregations of hunter-gatherer groups 368–382. https://doi.org/10.1016/j.jaa.2013.05.005.
are a common phenomenon that is associated with two different cir- Bradley, R., Boado, F.C., Valcarce, R.F., 1994. Rock art research as landscape archae-
ology: a pilot study in galicia, north-west Spain. World Archaeol. 25, 374–390.
cumstances: water shortage during the dry period (Yellen, 1977; Lee, https://doi.org/10.2307/124888.
1972a; Bartram et al., 1991), and social activities that need to be Breunig, P., 1988. Botanisch-archäologische Beobachtungen in einem afrikanischen
conducted in larger groups (Lee, 1972b; Marshall, 1969), which often Hochgebirge. Aspekte zur prähistorischen Besiedlung eines ariden Gunstraumes.
Archäologische Informationen, vol. 11. pp. 53–73. https://doi.org/10.11588/ai.
occurred together. A small number of rock art sites at the Dâureb with 1988.1.26841.
painting intensity way above-average suggests that these sites were Breunig, P., 1989. Der Brandberg: forschungen in einem Hochgebirge am Rande der
visited by more people than others. But since water is often accessible Namib. Archaol. Deutschl. 1989, 42–44. https://doi.org/10.2307/26308722.
Breunig, P., 2003. Der Brandberg. Untersuchungen zur Besiedlungsgeschichte eines
for longer periods in the Dâureb, or large parts of it, than in the sur- Hochgebirges in Namibia, Africa Praehistorica. Heinrich-Barth-Institut, Köln.
rounding Namib, the necessity for groups to come together at certain Breunig, P., Behringer, J., Fels, M., Maidhof, J., 2019. West of the best: rock art and
places can be hardly explained by water shortage alone. More reason- archaeological discoveries in the Doro !Nawas region of northwest Namibia. Acta
Archaeol. 89, 174–192. https://doi.org/10.1111/j.1600-0390.2018.12199.x.
able is to assume that people came together at these places for certain Brooks, A.S., 1984. San land-use patterns, past and present: implications for southern
communal social activities such as powerful trance dances or initiation African prehistory. In: Hall, M. (Ed.), Frontiers : Southern African Archaeology
rites (Lee, 1972b; Marshall, 1969). In the ethnographic and archae- Today. Cambridge Monographs in African Archaeology 10, B.A.R. International
Series. B.A.R., Oxford, England, pp. 40–52.
ological record, one of the major indicators of aggregation sites are the
Bronk Ramsey, C., 2008. Radiocarbon dating: revolutions in understanding.
quantity and diversity of the excavated material (Yellen, 1977; Conkey, Archaeometry 50, 249–275. https://doi.org/10.1111/j.1475-4754.2008.00394.x.
1980; Brooks, 1984). However, this is not necessarily the case in Burgess, R.L., Jacobson, L., 1984. Cultural sediment formation in open-air sites and rock
Dâureb, as smaller groups were apparently dispersed to different places shelters on the Brandberg, Namibia. J. Field Archaeol. 11, 233–239. https://doi.org/
10.2307/529357.
and only came together for communal activities. This hypothesis may Burke, A., 2005. Biodiversity patterns in arid, variable environments. Mt. Res. Dev. 25,
explain why some sites with few paintings had large quantities of ar- 228–234. https://doi.org/10.1659/0276-4741(2005)0250228:BPIAVE2.0.CO;2.
tefacts, while some sites with a particularly large number of paintings Chippindale, C., Nash, G., 2004. Pictures in place: approaches to the figured landscapes of
rock-art. In: Chippindale, C., Nash, G. (Eds.), Pictures in Place: the Figured
yielded rather small artefact assemblages. Most importantly, in contrast Landscapes of Rock-Art. Cambridge University Press, Cambridge, pp. 1–36.
to the material culture, diversity of the depicted topics is rather low Conard, N.J., Breunig, P., Gonska, H., Marinetti, G., 1988. The feasibility of dating rock
with a strong bias towards picturing humans, thus underlining the so- paintings from Brandberg, Namibia, with 14C. J. Archaeol. Sci. 15, 463–466. https://
doi.org/10.1016/0305-4403(88)90043-X.
cial aspect of aggregation sites in this area. Conkey, M.W., 1980. The Identification of prehistoric hunter-gatherer aggregation sites:
the case of Altamira. Curr. Anthropol. 21, 609–620. https://doi.org/10.2307/
Data availability 2741828.
Conkey, M.W., 2001. Structural and semiotic approaches. In: Whitley, D. (Ed.), Handbook
of Rock Art Research. AltaMira Press, Walnut Creek, pp. 273–310.
The Brandberg-Daureb database on rock art (BDDB) containing the Conolly, J., Lake, M., 2006. Geographical Information Systems in Archaeology,
datasets related to this article can be found at https://dx.doi.org/10. Cambridge Manuals in Archaeology. Cambridge University Press, Cambridge.
https://doi.org/10.1017/CBO9780511807459.
13149/r5f.3fyhip-x, an open-source online data repository hosted at
Craven, P., Craven, D., 2000. The flora of the Brandberg, Namibia. Cimbebasia Mem. 9,
IANUS Forschungsdatenzentrum (Lenssen-Erz and Vogels, 2017). 49–67.
Davidson, I., 2012. Variation in Early Paintings and Engravings. In: McDonald, J., Veth, P.
Declaration of competing interest (Eds.), A Companion to Rock Art. Wiley-Blackwell, Malden, MA, pp. 51–68. https://
doi.org/10.1002/9781118253892.
Deacon, J., 1988. The power of a place in understanding southern san rock engravings.
The authors declare that they have no known competing financial World Archaeol. 20, 129–140.
Denninger, E., 1986. Naturwissenschaftliche Untersuchungen zur Altersbestimmung von
interests or personal relationships that could have appeared to influ- Felsmalereien in Südwest-Afrika. In: Scherz, E.R., Breunig, P. (Eds.), Felsbilder in
ence the work reported in this paper. Südwest-Afrika. Teil 3: Die Malereien, Fundamenta. Böhlau, Köln, pp. 65–82.
Diehl, M., 1990. Geology, mineralogy, geochemistry and hydrothermal alteration of the
Brandberg alkaline complex, Namibia. Geological Survey of Namibia Memoir 10,
Acknowledgements 1–55.
Gutiérrez, F., Gutiérrez, M., 2016. Landforms of the Earth: an Illustrated Guide. Springer,
We are grateful to the editors for the invitation to this special issue. Cham.
Hartley, R., Wolley Vawser, A.M., 1998. Spatial behaviour and learning in the prehistoric
We also thank the three reviewers for their very helpful comments and environment of the Colorado River drainage (south-eastern Utah), western North
suggestions. America. In: Chippindale, C., Taçon, P.S.C. (Eds.), The Archaeology of Rock-Art.
Cambridge University Press, Cambridge, pp. 185–211.
Hogg, A.G., Hua, Q., Blackwell, P.G., Niu, M., Buck, C.E., Guilderson, T.P., Heaton, T.J.,
References
Palmer, J.G., Reimer, P.J., Reimer, R.W., Turney, C.S.M., Zimmerman, S.R.H., 2013.
SHCal13 southern hemisphere calibration, 0–50,000 Years cal BP. Radiocarbon 55,
Bahn, P.G., 1982. Inter-site and inter-regional links during the upper palaeolithic: the 1889–1903. https://doi.org/10.2458/azu_js_rc.55.16783.
pyrenean evidence. Oxf. J. Archaeol. 1, 247–268. https://doi.org/10.1111/j.1468- Jacobson, L., 1978. A Study of Functional Variability in the Later Stone Age of Western
0092.1982.tb00313.x. Damaraland, Namibia. Master thesis. Universtiy of Cape, Town.
Barceló, J., Achino, K., Bogdanovic, I., Capuzzo, G., Del Castillo, F., Almeida, V. de, Pérez, Jacobson, L., Vogel, J.C., 1975. Recent radiocarbon dates from the Brandberg. South Afr.
J.N., 2015. Measuring, counting and explaining: an introduction to mathematics in J. Sci. 71, 349.
archaeology. In: Barceló, Juan A., Bogdanovic, I. (Eds.), Mathematics and Keeney, B., Keeney, H., 2013. Reentry into first creation: a contextual frame for the Ju/
Archaeology. CRC Press, Boca Raton/Florida, pp. 3–64. ’hoan bushman performance of puberty rites, storytelling, and healing dance. J.

13
O. Vogels, et al. Quaternary International xxx (xxxx) xxx–xxx

Anthropol. Res. 69, 65–86. https://doi.org/10.2307/24393640. Companion to the Philosophy of Science. Blackwell, Oxford, pp. 184–193.
Kinahan, J., 1984. On the relative homogeneity of a short Holocene sequence of stone tool MacCalman, H.R., 1965. Carbon 14 dates from south west Africa. S. Afr. Archaeol. Bull.
assemblages from the Central Namib Desert. South Afr. J. Sci. 80, 273–275. 20https://doi.org/10.2307/3887829. 215–215.
Kinahan, J., 1990. Four thousand years at the Spitzkoppe: changes in settlement and Manhire, A.H., Parkington, J.E., Mazel, A.D., Maggs, T.M.O., 1986. Cattle, sheep and
landuse on the edge of the Namib Desert. CIMBEBASIA 12, 1–14. horses: a review of domestic animals in the rock art of southern Africa. Goodwin
Kinahan, J., 1991. Pastoral Nomads of the Central Namib Desert. Namibia Archaelogical Series 5, 22–30. https://doi.org/10.2307/3858142.
Trust, Windhoek. Mazel, A.D., 1988. People Making History: the Last Ten Thousand Years of Hunter-
Kinahan, J., 2005. The late Holocene human ecology of the Namib Desert. In: Smith, M., Gatherer Communities in the Thukela Basin (PhD Dissertation). University of Cape
Hesse, P. (Eds.), 23° S: Archaeology and Environmental History of the Southern Town, Cape Town.
Deserts. National Museum of Australia Press, Canberra, pp. 120–131. Maier, A., Zimmermann, A., 2017. Populations headed south? The Gravettian from a
Kinahan, J., 2016. Archaeological evidence of domestic sheep in the Namib Desert during palaeodemographic point of view. Antiquity 91, 573–588. https://doi.org/10.15184/
the first millennium AD. J. Afr. Archaeol. 14, 7–17. https://doi.org/10.3213/2191- aqy.2017.37.
5784-10280. Marshall, L., 1969. The medicine dance of the !Kung Bushmen. Africa 39, 347–381.
Kinahan, J., 2017. The solitary shaman: itinerant healers and ritual seclusion in the https://doi.org/10.2307/1157382.
Namib Desert during the second millennium AD. Camb. Archaeol. J. 27, 553–569. Marshall, L., 1976. The !Kung of Nyae Nyae. Harvard University Press, Cambridge.
https://doi.org/10.1017/S0959774317000348. https://doi.org/10.4159/harvard.9780674180574.
Kretschmer, I., Maier, A., Schmidt, I., 2016. Probleme und mögliche Lösungen bei der Martin, H., Mason, R., 1954. The test trench in the phillips cave, ameib, Erongo moun-
Schätzung von Bevölkerungsdichten im Paläolithikum. In: Kerig, T., Nowak, K., Roth, tains, south west Africa. S. Afr. Archaeol. Bull. 9, 148–151. https://doi.org/10.2307/
G. (Eds.), Alles Was Zählt… Festschrift Für Andreas Zimmermann, 3886836.
Universitätsforschungen Zur Prähistorischen Archäologie. Habelt, Bonn, pp. 47–58. McCall, G.S., 2010. Changing views of Drakensberg San rock art: examining landscape
Lander, F., Russell, T., 2015. Fat-tailed sheep and thin-walled pots: contextualising rock use, ritual activity, and contact through multivariate content-based spatial analysis.
art and pre-agriculturist pottery within the last 3000 years in KwaZulu-Natal, South Am. Antiq. 75, 773–791.
Africa. South. Afr. Humanit. 27, 113–163. Mitchell, P., 2005a. Modelling later stone age societies in southern Africa. In: Stahl, A.B.
Lee, R.B., 1972a. !Kung spatial organization: an ecological and historical perspective. (Ed.), African Archaeology: A Critical Introduction. Blackwell Publishing, Oxford, pp.
Hum. Ecol. 1, 125–147. https://doi.org/10.1007/BF01531351. 150–173.
Lee, R.B., 1972b. The intensification of social life among the! Kung Bushmen. In: Spooner, Mitchell, P., 2005b. Why hunter-gatherer archaeology matters: a personal perspective on
B. (Ed.), Population Growth: Anthropological Implications. MIT Press, Cambridge, renaissance and renewal in southern african later stone age research. S. Afr. Archaeol.
pp. 343–350. Bull. 60, 64–71. https://doi.org/10.2307/3889119.
Lee, R.B., 1979. The !Kung San: Men, Women, and Work in a Foraging Society. Cambridge Oliver, M.A., Webster, R., 1990. Kriging: a method of interpolation for geographical in-
University Press, Cambridge. formation systems. Int. J. Geogr. Inf. Syst. 4, 313–332. https://doi.org/10.1080/
Lenssen-Erz, T., 1989. The catalogue. In: Pager, H. (Ed.), The Rock Paintings of the Upper 02693799008941549.
Brandberg I: Amis Gorge, Africa Praehistorica. Heinrich Barth Institut, Köln, pp. Orton, J., 2013. Geometric rock art in western South Africa and its implications for the
343–502. spread of early herding. S. Afr. Archaeol. Bull. 68, 27–40. https://doi.org/10.2307/
Lenssen-Erz, T., 1992. Coherence – a constituent of “scenes” in rock art. The transfor- 23631481.
mation of linguistic analytical models for the study of rock paintings in Namibia. O'Sullivan, R., 2019. Movement across a mountain barrier: mapping accessibility with
Rock Art Res. 9, 87–105. rock-art and GIS in the altai mountains, eastern eurasia. J. Archaeol. Sci.: Report 27,
Lenssen-Erz, T., 1997. Metaphors of intactness of environment in Namibian rock paint- 101979. https://doi.org/10.1016/j.jasrep.2019.101979.
ings. In: Faulstich, P. (Ed.), Rock Art as Visual Ecology. American Rock Art Research Ouzman, S., Wadley, L., 1997. A history in paint and stone from Rose Cottage Cave, South
Association, Tucson, Arizona, pp. 43–54. Africa. Antiquity 71, 386–404. https://doi.org/10.1017/S0003598X00084994.
Lenssen-Erz, T., 2000. Picturing ecology through animal metaphorism – prehistoric Pager, H., 1971. Ndedema. Akademische Druck- und Verlagsanstalt, Graz.
hunter-gatherers and the first pastro-foragers in Namibia. In: Bollig, M. (Ed.), People, Pager, H., 1974. Quantitative analysis elucidate the motives of the South-African rock
Cattle and Land: Transformations of a Pastoral Society in Southwestern Africa. painters. Almogaren 5, 205–218.
Rüdiger Köppe, Köln, pp. 95–118. Pager, H., 1980. Felsbildforschungen am Brandberg in Namibia. Beiträge zur Allgemeinen
Lenssen-Erz, T., 2001. Gemeinschaft – Gleichheit – Mobilität. Felsbilder im Brandberg, und Vergleichenden Archäologie 2, 351–357.
Namibia, und ihre Bedeutung. Grundlagen einer textuellen Felsbildarchäologie., Pager, H., 1989. The Rock Paintings of the Upper Brandberg I: Amis Gorge, Africa
Africa Praehistorica. Heinrich Barth Institut, Köln. Praehistorica 1. Heinrich Barth Institut, Köln.
Lenssen-Erz, T., 2002. The Third Gender: Human.Gender-related patterns of activity in Pager, H., 1993. The Rock Paintings of the Upper Brandberg II: Hungorob Gorge, Africa
the rock paintings of the Brandberg, Namibia. In: Banks, A. (Ed.), The Proceedings of Praehistorica 4. Heinrich Barth Institut, Köln.
the Khoisan Identities and Cultural Heritage Conference. Institute for Historical Pager, H., 1995. The Rock Paintings of the Upper Brandberg III: Southern Gorges, Africa
Research, University of the Western Cape, Bellville, South Africa, pp. 146–152. Praehistorica 7. Heinrich Barth Institut, Köln.
Lenssen-Erz, T., 2004. The landscape setting of rock-painting sites in the Brandberg Pager, H., 1998. The Rock Paintings of the Upper Brandberg IV: Umuab and Karoab
(Namibia): infrastructure, Gestaltung, use and meaning. In: Chippindale, C., Nash, G. Gorges, Africa Praehistorica 10. Heinrich Barth Institut, Köln.
(Eds.), The Figured Landscapes of Rock-Art: Looking at Pictures in Place. Cambridge Pager, H., 2000. The Rock Paintings of the Upper Brandberg V: Naib Gorge (A) and the
University Press, Cambridge, pp. 131–150. Northwest, Africa Praehistorica 12. Heinrich Barth Institut, Köln.
Lenssen-Erz, T., 2008. Space and discourse as constituents of past identities – the case of Pager, H., 2006. The Rock Paintings of the Upper Brandberg VI: Naib (B), Circus and Dom
Namibian rock art. In: Domingo Sanz, I., Fiore, D., May, S.K. (Eds.), Archaeologies of Gorges, Africa Praehistorica 20. Heinrich Barth Institut, Köln.
Art, Time, Place, and Identity. Left Coast Press, Walnut Creek, CA, pp. 29–50. Parkington, J., Mills, G., 1991. From space to place: the architecture and social organi-
Lenssen-Erz, T., Vogels, O., 2017. Brandberg-daureb datenbank. IANUS zation of southern African mobile communities. In: Gamble, C.S., Boismier, W.A.
Forschungsdatenzentrum. https://www.ianus-fdz.de 10.13149/r5f.3fyhip-x. (Eds.), Ethnoarchaeological Approaches to Mobile Campsites: Hunter-Gatherer and
Lenssen-Erz, T., Vogelsang, R., 2005. Populating no-man’s-land: rock art in northern Pastoralist Case Studies, International Monographs in Prehistory:
Namibia. Goodwin Series 9, 54–62. https://doi.org/10.2307/3858034. Ethnoarchaeological Series. Berghahn books, Ann Arbor (Michigan), pp. 355–370.
Lenssen-Erz, T., Fäder, E., Vogels, O., Mathiak, B., (in press). Scenic narratives of humans Pearce, D.G., 2010. The Harris Matrix technique in the construction of relative chron-
and animals in Namibian rock art – a methodological restart with data mining. In: ologies of rock paintings in South Africa. S. Afr. Archaeol. Bull. 65, 148–153. https://
Davidson, I., Nowell, A. (Eds.), Making Scenes: Global Perspectives on Scenes in Rock doi.org/10.2307/41408507.
Art. Berghahn Books, New York. Pleurdeau, D., Imalwa, E., Détroit, F., Lesur, J., Veldman, A., Bahain, J.-J., Marais, E.,
Lewis-Williams, J.D., 1972. The syntax and function of the Giant's Castle rock-paintings. 2012. “Of sheep and men”: earliest direct evidence of caprine domestication in
S. Afr. Archaeol. Bull. 27, 49–65. https://doi.org/10.2307/3888814. southern Africa at Leopard Cave (Erongo, Namibia). PloS One 7, 1–10. https://doi.
Lewis-Williams, J.D., 1974. Superpositioning in a sample of rock-paintings from the org/10.1371/journal.pone.0040340.
barkly east district. S. Afr. Archaeol. Bull. 29, 93–103. https://doi.org/10.2307/ Richter, J., 1991. Studien zur Urgeschichte Namibias, Africa Praehistorica. Heinrich-
3888335. Barth-Institut, Köln.
Lewis-Williams, J.D., 1981. Believing and Seeing: Symbolic Meanings in Southern San Ripley, B.D., 1976. The second-order analysis of stationary point processes. J. Appl.
Rock Painting. Academic Press, London. Probab. 13, 255–266. https://doi.org/10.2307/3212829.
Lewis-Williams, J.D., 1982. The economic and social context of southern San rock art. Rudner, J., 1957. The Brandberg and its archaeological remains. J. SWA (South West Afr.)
Curr. Anthropol. 23, 429–449. Sci. Soc. 12, 7–44.
Lewis-Williams, J.D., 1984. The empiricist impasse in southern african rock art studies. S. Russell, T., 2012. The position of rock art: a consideration of how GIS can contribute to
Afr. Archaeol. Bull. 39, 58–66. https://doi.org/10.2307/3888595. the understanding of the age and authorship of rock art. In: Smith, B.W., Helskog, K.,
Lewis-Williams, J.D., 1991. Wrestling with analogy: a methodological dilemma in upper Morris, D. (Eds.), Working with Rock Art. Recording, Presenting, and Understanding
palaeolithic art research. Proc. Prehist. Soc. 57, 149–162. https://doi.org/10.1017/ Rock Art Using Indigenous Knowledge. Witwatersrand University Press,
S0079497X00004941. Johannesburg, pp. 37–45.
Lewis-Williams, J.D., 1992. Vision, power and dance: the genesis of a Southern African Sauvet, G., Wlodarczyk, A., 1995. Eléments d’une grammaire formelle de l’art pariétal
rock art panel, Kroon-Voordracht gehouden voor de Stichting Nederlands Museum paléolithique. L’Anthropologie 99, 193–211.
voor Anthropologie en Praehistorie. Stichting Nederlands Museum voor Sauvet, G., Layton, R., Lenssen-Erz, T., Tacon, P., Wlodarczyk, A., 2006. La structure
Anthropologie en Praehistorie, Amsterdam. iconographique d’un art rupestre est-elle une clef pour son interprétation?
Lewis-Williams, J.D., 2010. The imagistic web of San myth, art and landscape. South. Afr. Zephyrus 59.
Humanit. 22, 1–18. Schaafsma, P., 1985. Form, content, and function: theory and method in north american
Lipton, P., 2000. Inference to the best explanation. In: Newton-Smith, W.H. (Ed.), A rock art studies. Adv. Archaeol. Method Theor. 8, 237–277.

14
O. Vogels, et al. Quaternary International xxx (xxxx) xxx–xxx

Scherz, E.R., Breunig, P., 1975. Felsbilder in Südwest-Afrika. Teil 2: Die Gravierungen im Vinnicombe, P., 1975. The ritual significance of eland (Taurotragus oryx) in the rock art
Nordwesten Südwest-Afrikas, Fundamenta. Böhlau. (Köln/Wien). of southern Africa. In: Anati, E. (Ed.), Actes Du Valcamonica Symposium ’72: Les
Scherz, E.R., Breunig, P., 1986. Felsbilder in Südwest-Afrika. Teil 3: Die Malereien, Religions de La Préhistoire. Edizioni del Centro, Capo di Ponte History of Religions,
Fundamenta. Böhlau, Köln. I.A. for the.
Schmidt, I., Hilpert, J., Kretschmer, I., Peters, R., Broich, M., Schiesberg, S., Vogels, O., Vinnicombe, P., 1976. People of the Eland. Rock Paintings of the Drakensberg Bushmen
Wendt, K.-P., Zimmermann, A., Maier, A. (in press) Approaching prehistoric demo- as a Reflection of Their Life and Thought. University of Natal Press, Pietermaritzburg.
graphy: proxies, scales and scope of the cologne protocol in European contexts. Phil. Vogel, J.C., Visser, E., 1981. Pretoria radiocarbon dates II. Radiocarbon 23, 43–80.
Trans.B. https://doi.org/10.1017/S0033822200037462.
Schmidt, I., Zimmermann, A., 2019. Population dynamics and socio-spatial organization Wadley, L., 1976. Radiocarbon dates from big elephant shelter, Erongo mountains, south
of the Aurignacian: scalable quantitative demographic data for western and central west Africa. S. Afr. Archaeol. Bull. 31https://doi.org/10.2307/3887735. 146–146.
Europe. PloS One 14, 1–20. https://doi.org/10.1371/journal.pone.0211562. Wadley, L., 1977. Some Aspects of Holocene Prehistory in Central South West Africa.
Schmitt, A.K., Emmermann, R., Trumbull, R.B., Bühn, B., Henjes-Kunst, F., 2000. With Especial Reference to Big Elephant Shelter, Erongo Mountains (Master Thesis).
Petrogenesis and 40Ar/39Ar Geochronology of the Brandberg Complex, Namibia: University of Cape Town, Cape Town.
evidence for a major mantle contribution in metaluminous and peralkaline granites. Wadley, L., 1987. Later Stone Age Hunters and Gatherers of the Southern Transvaal:
J. Petrol. 41, 1207–1239. https://doi.org/10.1093/petrology/41.8.1207. Social and Ecological Interpretation. BAR International Series. B.A.R., Oxford.
Silberbauer, G.B., 1981. Hunter and Habitat in the Central Kalahari Desert. Cambridge Wendt, W.E., 1972. Preliminary report on an archaeological research programme in
University Press, Cambridge. South West Africa. CIMBEBASIA B 2, 1–61.
Smith, A.B., Yates, R., Miller, D., Jacobson, L., Evans, G., 1995. Excavations at geduld and Weninger, B., Clare, L., Jöris, O., Jung, R., Edinborough, K., 2015. Quantum theory of
the appearance of early domestic stock in Namibia. S. Afr. Archaeol. Bull. 50, 3–20. radiocarbon calibration. World Archaeol. 47, 543–566. https://doi.org/10.1080/
Smith, B.W., Ouzman, S., 2004. Taking stock: identifying khoekhoen herder rock art in 00438243.2015.1064022.
southern Africa. Curr. Anthropol. 45, 499–526. https://doi.org/10.1086/422081. Wiessner, P., 1982. Risk, reciprocity and social influences on !Kung San economics. In:
Taçon, P.S.C., Chippindale, C., 1998. An archaeology of rock-art through informed Leacock, E., Lee, R.B. (Eds.), Politics and History in Band Societies. Cambridge
methods and formal methods. In: Chippindale, C., Taçon, P.S.C. (Eds.), The University Press, Cambridge, pp. 61–84.
Archaeology of Rock-Art. Cambridge University Press, Cambridge, pp. 1–10. Wiessner, P., 1983. Style and social information in Kalahari San projectile points. Am.
Thackeray, A.I., 1983. Dating the rock art of southern Africa. South Afr. Archaeol. Soc. Antiq. 48, 253–276. https://doi.org/10.2307/280450.
Goodwin Ser. 4, 21–26. https://doi.org/10.2307/3858098. Williams, A.N., 2012. The use of summed radiocarbon probability distributions in ar-
Tratebas, A., 2012. North American – siberian connections: regional rock art patterning chaeology: a review of methods. J. Archaeol. Sci. 39, 578–589. https://doi.org/10.
using multivariate statistics. In: McDonald, J., Veth, P. (Eds.), A Companion to Rock 1016/j.jas.2011.07.014.
Art. Wiley-Blackwell, Malden, MA, pp. 143–159. https://doi.org/10.1002/ Yellen, J.E., 1977. Archaeological Approaches to the Present: Models for Reconstructing
9781118253892. the Past. Academic Press, New York.
Vinnicombe, P., 1967. Rock-painting analysis. S. Afr. Archaeol. Bull. 22, 129–141. Zimmermann, A., Wendt, K.P., Frank, T., Hilpert, J., 2009. Landscape archaeology in
https://doi.org/10.2307/3888475. central europe. Proc. Prehist. Soc. 75, 1–53. https://doi.org/10.1017/
Vinnicombe, P., 1972. Myth, motive, and selection in southern African rock art. Africa 42, S0079497X00000281.
192–204.

15