. 4
( 11)


edge-retouched pieces made on ¬‚akes removed from radial cores, bear-
ing some resemblance to the Bambata material from Zimbabwe, discussed
below. The sequence in the north is well seen at two sites near Mokopane: the
Cave of Hearths (Sampson 1974) and Mwulu™s Cave (Tobias 1949). Broadly sim-
ilar industries are recorded in the Western Cape, but they are from poorly
documented contexts and their position in the general sequence of mode-
3 industries is unclear; possibly they belong to the period between about
40,000 and 20,000 years ago, when sites such as Nelson Bay Cave and Klasies
River Mouth were unoccupied.
It is in the context of these late blade-based mode-3 industries that the
local origins of the mode-5 backed-microlith technological tradition must
be sought. As in other parts of Africa, recent research has shown that the
process involved was generally one of gradual transition without a clearly
de¬ned interface, except in the case of the Howieson™s Poort industry, dis-
cussed above. Blade-based occurrences are reported from rockshelters in the
mountains of Lesotho, at Ha Soloja, Sehonghong and Moshebi™s, dating from
more than 43,000 until about 30,000 years ago (Carter and Vogel 1974; Carter
et al. 1988; Mitchell 1988, 1994, 1996). The blades show a progressive diminu-
tion in size, and backed retouch is clearly attested. Clearly, the role played
by this cold high-altitude region in the development of southern African
Regional diversification and specialisation 105

mode-5 industries requires further investigation. To the south, Strathalan
Cave near Maclear has preserved what is described as a transitional
(mode 3--mode 5) industry dating to about 22,000 years ago, associated with
well-preserved plant materials (Opperman and Heydenrych 1990).
There are indications from Botswana that mode-5 technology may have
begun there considerably more than 25,000 years ago. Many areas may have
been uninhabited during the hyper-arid interlude around 18,000 years ago,
except around the Tsodilo Hills and the Okavango Delta. Surface water
remained at Tsodilo long afterwards (Brook et al. 1992), where it sustained
a local ¬shing adaptation based on the use of barbed bone points (Robbins
et al. 1994, 2000; see also p. 145 below).
Broadly contemporary is the Robberg industry of southernmost South
Africa, dated between 19,000 and 12,000 years ago at sites such as Nelson
Bay Cave (Klein 1974; J. Deacon 1978) and Boomplaas (H. J. Deacon 1979),
and comprising many tiny bladelets with few standardised retouched tools
(Wadley 1993; Binneman and Mitchell 1997). Its makers, like their predeces-
sors, hunted the grassland fauna of the coastal plains. Similar microlithic
industries are now known from more northerly areas, as in Lesotho
(Mitchell 1988) and at Bushman Rock in Mpumalanga (Beaumont and Vogel
Between about 12,000 and 8000 years ago several sites widely dispersed
across southern Africa show occupation by makers of poorly understood
industries with many large edge-retouched ¬‚akes but virtually no microliths
or backed pieces (Fig. 45). Although regional variants may be recognised,
these industries are conveniently grouped together as the Oakhurst Com-
plex (Sampson 1974; Mitchell 2002). At Nelson Bay Cave the abrupt change
from the Robberg to the Oakhurst Complex (here represented by the Albany
industry) appears to have coincided with the rise of sea level to its maxi-
mum post-glacial height, with a corresponding change in faunal availability
(J. Deacon 1984).
The early appearance and development of southern African mode-5 indus-
tries is particularly problematic. Their oldest attestation is in the Howieson™s
Poort horizons at Klasies River Mouth and, probably, Border Cave, now dated
about 70,000 years ago. In the southeastern highland areas there is evidence
for continuous occupation by makers of industries comprising small blades
with backed retouch for most of the last 45,000 years and perhaps longer.
By 30,000 years ago these industries show features recognised as ancestral
to the later mode-5 traditions. This technology is not ¬rmly attested in post-
Howieson™s Poort contexts in the Western Cape until some ten millennia
later, around 20,000 years ago, but this period remains poorly known as
several major sites were not then occupied. Inter-regional comparisons are
106 afric an archaeolog y

Fig. 49: Kalemba

hampered by inconsistent use of terminology, by the still imprecise dating
of some occurrences, and by uncertainties surrounding the effects of raw
material on lithic technology and typology.
From about 8000 years ago, microlithic industries became virtually ubiq-
uitous in southern Africa (Mitchell 1997, 2002), conceivably linked with
widespread adoption of the bow and arrow (cf. Binneman 1994). Many of
them have been designated Wilton after a site near Alicedale in Eastern
Cape Province (J. Deacon 1972; H. J. Deacon 1976). However, when a num-
ber of these assemblages are compared, a great range in the form and fre-
quencies of tool types is observed (J. Deacon 1984; Inskeep 1987; Opper-
man 1996). At coastal sites this development correlates with the ¬nal rise
of the sea to approximately its present level, and there is evidence that
marine food resources were once again intensively exploited. Exceptional
areas included the greater part of the Kalahari, which seems to have been
largely uninhabited from about 9500 until about 4500 years ago (J. Deacon
1974), and the upper Zambezi Valley, where the river margins may have
seen a late persistence of mode-3 stone-working techniques (L. Phillipson
A substantial number of graves throw light on the burial customs of this
time. Generally, the dead appear to have been buried within the settlement
area, whether this was a cave or rockshelter (Fig. 49) or in the open air. Grave
goods in the form of tools, items of adornment or other personal belong-
ings are often present and may indicate a belief that the dead would have
Regional diversification and specialisation 107

some use for such objects. Antelope horns or warthog tusks were also some-
times buried with the deceased. Items of personal adornment are frequently
encountered in graves as well as on settlement sites. Beads and pendants
of bone or shell are widespread. Rock paintings suggest that such beads
were sometimes sewn onto clothing or worn in the hair, as well as being
threaded in strings. Ochre and other colouring matter were probably used
for cosmetic and other purposes as well as for mural decoration. Much vari-
ation is apparent in the goods placed with different interments (S. Hall and
Binneman 1987; S. Hall 2000). As will be shown below, painted gravestones
were in vogue in the south coastal area.
Many of the dated archaeological occurrences of the period between
10,000 and 2000 years ago in southern Africa are from sites where little
has been preserved apart from the stone artefacts and occasional associated
faunal remains. Only rarely has vegetable matter survived, leading to an
unbalanced representation of the tool-kits and diet of the people. The dry
cave deposits of southernmost Africa (H. J. Deacon 1976) are therefore par-
ticularly signi¬cant. From these sites we can see that, at least in some areas,
wood was used for bows, arrows, digging sticks, pegs and wedges. Bark was
used for trays. Bags and clothing were made of sewn leather. Leaves were
used as a wrapping material for valuables; while grass and soft undergrowth
were collected for bedding. Vegetable foods were varied and often assumed
considerable importance in the total diet, as they do among most mod-
ern tropical hunter-gatherer groups. Remains of plant foods provide vital
information about the seasonality of settlements, con¬rming the often
less detailed evidence of the faunal remains. In the southwestern Cape,
for example, it has been argued that some hunter-gatherer groups may
have moved regularly between their coastal winter settlements and summer
haunts further inland (Parkington 1986, 2001; Parkington and Hall 1987; but
see H. J. and J. Deacon 1999). Seasonal movement is also indicated in the
uKhahlamba/Drakensberg region (Carter 1970).
As archaeological survey becomes more comprehensive, and the empha-
sis of research extends beyond technology and subsistence economy, it has
become possible in several areas of southern Africa to recognise the social
implications of group mobility. Certain large sites, often prominent rockshel-
ters, have been interpreted as places of regular social aggregation, generally
at a particular season of the year (e.g. Wadley 1987). A similar phenomenon
has been recorded among certain recent San populations, the aggregations
serving an important function in commodity exchange (hxaro), the mak-
ing of matrimonial and other alliances, and suchlike activities. At other
seasons the populations appear to have dispersed to smaller, more tempo-
rary encampments such as are also now recognised in the archaeological
108 afric an archaeolog y

South-central Africa
North of the Limpopo, the post-Acheulean archaeology of the region cen-
tred on Zimbabwe and southwestern Zambia differs in important respects
from that in South Africa. As noted in chapter 3, this region fell within
the distribution of Sangoan industries into which, it has been suggested,
should be subsumed those sometimes designated Charaman. These even-
tually gave way to lithic industries which display a well-developed mode-3
technology. The immediately post-Sangoan stone industries are now seen to
have depended to a far larger extent on the production of ¬‚ake-blades than
was previously recognised. In central Zambia, Barham (2000, 2001, 2002b;
see also J. D. Clark and Brown 2001) has shown how re-investigation of key
sites and the application of recently developed dating methods can provide
a wholly new understanding of this period. At Mumbwa and at Twin Rivers,
artefact assemblages of Lupemban af¬nity (see pp. 117--20 below) have been
recognised, probably dating to the general period 250,000--170,000 years ago;
they include a few small and unmistakably backed pieces such as would pre-
viously have been regarded as restricted to very much later periods. Extensive
use of pigment is also indicated at this time (Barham 2002a); the signi¬cance
of this is evaluated below.
In Zimbabwe and some adjacent areas, on the other hand, the mode-3
industries are generally known as Bambatan after a cave in the Matopo
Hills near Bulawayo. Informative occurrences have been investigated also
at Pomongwe (C. K. Cooke 1963), at Redcliff in central Zimbabwe (C. K.
Cooke 1978) and at Kalemba in eastern Zambia (D. W. Phillipson 1976).
Unifacially trimmed sub-triangular points were produced, with occasional
bifacial forms, together with a variety of edge-retouched ¬‚akes. These tools
may well have served functions similar to those of their counterparts in the
more southerly, and probably earlier, Pietersburg industries, but they clearly
belong to a different technological and stylistic tradition. Radiocarbon dates
for the Bambatan occurrences at these sites between 45,000 and 30,000 years
ago probably indicate the minimum age of the industry. The faunal remains
suggest that the makers of these artefacts exploited a wide range of animal
species for their food. At several sites there is evidence that the climate was
becoming progressively wetter during this occupation (Brain 1969). Partic-
ularly interesting is the discovery, within the Bambata industry, at three
widely separated sites in Zimbabwe, of horizons marked by a proliferation
of small blades (C. K. Cooke 1971, 1973). These occurrences, which are prob-
ably signi¬cantly older than 40,000 years, are possible counterparts of the
more southerly Howieson™s Poort. Little research has been undertaken on
this material since the 1970s; Larsson (1996) has drawn attention to the
need for new ¬eldwork and synthesis.
Regional diversification and specialisation 109

Fig. 50: Early
mode-5 artefacts
from Kalemba
(after D. W.
Phillipson 1976)

Some isolated but signi¬cantly earlier occurrences notwithstanding, it is
in the context of the Bambata-related industries that we see widespread
local evidence for the adoption of mode-5 techniques of backed-microlith
manufacture. In most areas these industries show, with the passage of time,
a signi¬cant reduction in mean artefact size. Flakes and blades with par-
tial backing occur in most Bambata-related assemblages, becoming progres-
sively more numerous, and the backing more extensive; crescent-shaped
segments of blade with continuous backing along the curved edge fre-
quently occur. In Zimbabwe these later ill-de¬ned industries have sometimes
been named Tshangula after a cave site in the Matopo Hills. Broadly sim-
ilar occurrences are known in several areas of south-central Africa; they
have been dated within the time-span 25,000 to 14,000 years ago (e.g. C. K.
Cooke 1963; D. W. Phillipson 1976), although some may be signi¬cantly
North of the Zambezi the earliest true backed-microlith industry yet dis-
covered had developed by about 19,000 years ago, as seen, for example,
at Leopard™s Hill near Lusaka and at several sites in northern and eastern
Zambia. Tiny pointed backed bladelets and varied scrapers are the character-
istic tool types of these so-called Nachikufan I industries (Fig. 50), together
with bored stones, the larger examples of which resemble objects which
are known to have been used in later periods as weights for digging sticks,
although the functions of the smaller ones remain unknown (S. F. Miller
1971, 1972; D. W. Phillipson 1976; Musonda 1984). It seems that the develop-
ment of microlithic technology proceeded at differing speeds in the various
Later microlithic industries were characterised by geometrical backed
forms, chie¬‚y crescents, which replaced the single-pointed types of
Nachikufan I and its counterparts. As in South Africa, these industries show
considerable regional variation, the signi¬cance of which is not yet appar-
ent. In some areas, such as southern Zambia, small convex scrapers far
outnumber the backed microliths. Despite this variability, there has been
a tendency among archaeologists to label most of these industries with
the generic name Wilton (cf. p. 106 above). This has served to obscure the
110 afric an archaeolog y

Fig. 51: Artefacts
from Gwisho
(after Fagan and
Van Noten 1971):
1, 5, wooden
arrow-points or
link-shafts; 2, ¬re
drill; 3, digging
stick; 4, wooden
6, 7, backed
8, grinding stone

very real differences between most of the assemblages so designated, while
exaggerating the idiosyncrasy of those, such as the later Nachikufan phases
of northern Zambia (S. F. Miller 1972) and the Kaposwa from Kalambo Falls
( J. D. Clark 1974), which have been given different names. The waterlogged
sites at Gwisho hotsprings in southern Zambia (Fagan and Van Noten 1971)
are of particular importance, organic materials having been exceptionally
well preserved (Fig. 51), including wooden pieces interpreted as digging-
sticks, heads or link-shafts from arrows, and possibly part of a bow. Some
sequences in Zimbabwe, notably Pomongwe in the Matopo Hills, include
occurrences attributed to the Oakhurst Complex (N. J. Walker 1995), better
known from sites further to the south (p. 105 above). It now appears that
economic, cultural, social and geological factors all in¬‚uenced the types
of tools that were used. At sites with a strati¬ed sequence of microlithic
industries, there has been noted a general decrease in artefact size with the
passage of time (D. W. Phillipson 1977a).
Regional diversification and specialisation 111

Rock art in southern and south-central Africa
No account of the later prehistory of this part of Africa would be com-
plete without reference to the remarkable rock art -- both paintings and
engravings -- that is so abundantly preserved there and which has gen-
erated a profuse and varied literature (see Summers 1959; Willcox 1963;
Vinnicombe 1976; Lewis-Williams 1983; Lewis-Williams and Dowson 1989;
Dowson 1992; Dowson and Lewis-Williams 1994; Garlake 1995; Solomon
1997; Russell 2000). During the past quarter-century great progress has been
made in at least partially understanding the original signi¬cance of this art,
but the full implications of these advances can only be appreciated with
knowledge of its varied age and authorship. Without this background we
cannot hope to make full use of this potentially most illuminating evidence,
which relates to aspects of prehistoric life that would otherwise remain
Virtually all surviving southern African rock art occurs on the walls
of shallow caves or rockshelters or on rock outcrops in the open air. It
has inevitably suffered from exposure to the elements (unlike almost all
European Palaeolithic paintings, which are in deep caves); it thus seems
probable that most of the still-extant paintings were executed within the
past few thousand years. Such arguments do not necessarily apply to the
engravings which, although often fully exposed, are less susceptible to oblit-
eration by the elements. As with many categories of archaeological evidence,
much more art, executed in different circumstances or in earlier times, may
not have survived, as is suggested, for example, by the frequent occurrence
of ochre on settlement and burial sites.
There is an increasing body of evidence that, although much of the extant
art is probably of relatively recent date, it belongs to a very ancient tra-
dition. Painted and engraved stone slabs have been excavated from dated
deposits at several southern African sites. At ˜Apollo 11 Cave™ in southern
Namibia, detached slabs bearing naturalistic paintings of animals (Fig. 52)
occur in levels dated as long ago as 28,000 years, associated with a late
mode-3 industry, indicating that some southern African rock art has an
antiquity comparable with that of its European and Australian counterparts
(Wendt 1976). Rather later in date, but showing much greater variety, are
the painted stones found in graves of the makers of microlithic industries
in caves on the south coast, as at Klasies River Mouth and Coldstream Cave
(Singer and Wymer 1969; Rudner 1971). Engraved stones have been excavated
at Wonderwerk Cave near Kuruman in contexts of around 10,000 years ago
(Thackeray et al. 1981), while Butzer et al. (1979) have argued that some
of the open-air engraving sites may be more than 4000 years old. In the
112 afric an archaeolog y

Fig. 52:
years old on
stone slabs from
˜Apollo 11 Cave™,

uKhahlamba/Drakensberg, radiometric analyses suggest that some extant
paintings were executed some 3000--2000 years ago (Mazel and Watchman
It is sometimes possible to distinguish older and younger works in the
same site or area, either on the basis of differential weathering or through
the study of superimpositions, where one image has been made over an
earlier one. Con¬rmation of such sequences and indications of age can
Regional diversification and specialisation 113

Fig. 53:
Naturalistic rock
paintings at
Makwe near

sometimes be obtained when the artist has depicted phenomena such as
domestic animals or European colonists, whose dates in southern Africa are
known from other sources (cf. B. W. Smith and Van Schalkwyk 2002). By these
means it has been demonstrated that, in several parts of southern Africa,
rock art continued to be executed into the last few centuries, often main-
taining stylistic traditions that had been established in far earlier times. It
is to these late expressions that most of our data regarding the meaning
and signi¬cance of the art relate, but it is pertinent to cite such evidence
here since it may perhaps provide an indication of the original function of
the more ancient art.
Most southern African rock paintings are naturalistic representations
of people and animals (Fig. 53); other natural or arti¬cial objects, other
than personal accoutrements, are only rarely depicted. The wide variety
of animal species shown, both identi¬able and ˜mythical™, and the range
of human activities, equipment and clothing, has attracted considerable
attention. Less interest has until recently been accorded to the varied geo-
metrical or schematic motifs, presumably because these lack the obvi-
ous aesthetic appeal of the naturalistic art. The latter paintings show
a tendency to develop from simple outlines or ¬‚at silhouettes towards
use of shading in monochrome or polychrome and an increased mas-
tery of perspective: these latest features are best represented by paintings
(Fig. 54) in the uKhahlamba/Drakensberg which probably represent the ¬nest
114 afric an archaeolog y

Fig. 54: Rock
painting at

achievements of African rock artists. Among the engravings, animals and
schematic motifs predominate, human ¬gures being rarely shown. Here,
¬nely engraved outlines are generally earlier than are pecked silhouettes
(Figs. 55, 56).
Several writers, viewing southern African rock art from a primarily foreign
viewpoint, have stressed its undoubted aesthetic qualities and suggested that
the painters™ main object was to create a thing of beauty. This ˜art for art™s
sake™ explanation can no longer be accepted in view of detailed comparisons
that have been made between, on the one hand, the arrangement and sub-
ject matter of the paintings and, on the other, ethnographic records relat-
ing to southern San-speaking peoples in the nineteenth and twentieth cen-
turies (Vinnicombe 1976; Lewis-Williams 1981; Lewis-Williams and Dowson
1989; cf. also Solomon 1997). The eland is the animal most frequently rep-
resented in the paintings but not, by contrast, in the faunal remains pre-
served at occupational sites; this species is known to have occupied an impor-
tant place in the belief systems and symbolism of recent San communities.
Close parallels may be drawn between certain painted scenes and rites prac-
tised by nineteenth/twentieth-century San on occasions such as puberty and
marriage. Many paintings have otherwise inexplicable features that may be
understood in terms of shamanistic trance, such as played an important rˆleo
Regional diversification and specialisation 115

Fig. 55:
from Doornkloof,
South Africa
(after Willcox

Fig. 56: Pecked
Northern Cape

in the lives of the southern San. Only three examples need be cited. Paintings
of human ¬gures in the Western Cape and KwaZulu-Natal provinces of South
Africa often show lines descending from the nostrils: trance among the
San is frequently accompanied by nose-bleeding. A strange, forward-leaning,
half-crouching stance, not infrequently represented in the paintings, is
116 afric an archaeolog y

Fig. 57: Rock
showing people
in trance (after

identical to that adopted by dancing San when entering trance states
(Fig. 57). San emerging from trance have described their experience as
feeling like riding on the back of a serpent; and paintings in the Matopo
Hills of Zimbabwe show an enormous snake, sometimes double-headed,
with numbers of people standing on its back. These interpretations should
not, of course, detract from appreciation of the paintings™ aesthetic quality
although such appreciation is inevitably tied to the cultural background of
the individual beholder.
The lines and dots which comprise most of the schematic or non-
representational designs are interpreted by Lewis-Williams and Dowson
(1988) as ˜entoptics™ -- visual patterns or hallucinations experienced under
certain circumstances by people of many cultural backgrounds. More
recently, it has been persuasively demonstrated that much schematic art
(Fig. 58) may be attributed to herding peoples rather than to hunter-
gatherers (S. Hall and Smith 2000; B. W. Smith and Ouzman 2004). Inde-
pendent studies attribute an analogous signi¬cance to much of the rock art
further to the north, as will be described in chapter 7. It seems appropriate
to conclude that rock art can only be interpreted adequately with reference
to the belief systems of the artists. It is fortunate that some of the southern
African art was executed by societies suf¬ciently recent for these systems
to be at least partially understood, but untested assumptions that super¬-
cially similar rock art in other parts of the world necessarily has similar
shamanistic connotations should be regarded with grave suspicion (Clottes

Central Africa
Environmental change in central Africa resulted in major ¬‚uctuations in
the extent of the equatorial forest (cf. Brooks and Robertshaw 1990). Such
processes are as yet well documented only for the last 20,000 years, for which
Regional diversification and specialisation 117

Fig. 58:
Schematic rock
paintings in
South Africa™s
Northern Cape
Province of the
type now
attributed to
herding peoples

a series of maps has been published by Bonne¬lle (1999). There is no reason
to believe that similar changes did not take place in earlier times also. Since
stone is extremely rare in many parts of this region, it is likely that some
groups depended largely on artefacts made of perishable materials which
have not yet been recovered by archaeologists. It appears that even the areas
of densest forest were at least occasionally subject to human habitation
during this general period (Fiedler and Preuss 1985; Lanfranchi 1996; cf.
also Fig. 59).
By about 250,000 years ago, if not earlier, Sangoan industries (pp. 81--4)
gradually gave way to those known as Lupemban. Classic Lupemban arte-
facts are characterised by re¬ned bifacial stone-working techniques, seen
especially on core-axes and on long double-ended points which may have
been mounted for use as spearheads. They are found abundantly in the
area of the Plain of Kinshasa beside the Malebo (Stanley) Pool (Van Moorsel
1968), and also in the river gravels of the Dundo area in northern Angola
which have been mined extensively for diamonds ( J. D. Clark 1963). It is
now recognised that these classic forms may owe much to geological fac-
tors exempli¬ed by the availability of ¬ne-grained isotropic raw material in
large pieces. As noted above, industries of Lupemban af¬nity have recently
been recognised in central Zambia, provisionally dated between 250,000 and
170,000 years ago; occasional small backed tools have been recognised in a
Lupemban context at Twin Rivers, Lusaka (Barham 2000, 2001, 2002b; J. D.
Clark and Brown 2001). A similar age is probable at Kalambo Falls, where
Lupemban-type artefacts overlie the Sangoan horizons. Unlike much of the
Tamar Hat,
Kebibat Sale
el Arbi
Dar es Soltan Taforalt Columnata Bir el Ater
El Guettar
Jebel Ighoud Haua Fteah
et Tera

Wadi Kubbaniya
Bir Tarfawi
Bir Sahara Wadi Halfa
Arkin Khor Musa
Adrar Bous
Cap Vert
Kourounkorokale Anqqer Baahti
Middle Awash sites Porc Epic
Yengema Lada Oda
Melka Kunture
Iwo Eleru Gadamotta
Shum Laka
Afikpo Njuinye Omo-Kibish

Guli Waabayo
Gogoshiis Qabe
Katanda Yala A™lego Muguruk
Gamble™s Cave
Sango Bay Nderit Drift
Enkapune ya Muto
Kagera Nasera
L. Eyasi Olduvai
Gombe Point, Singida Kondoa
Malebo Pool

Kalambo Falls


Luangwa Valley
Broken Hill Rivers
Mumbwa Leopard™s
Gwisho Hill


Cave of Hearths Mwulu's Cave
Wonderwerk Rock
˜Apollo II™ Border Cave
Florisbad Cottage
Moshebi™s Sibudu
S e Ha Soloja
Strathalan h o n g h ong
Montagu Wilton
m Klasies River Mouth
C o ld s

0 2000 km

0 1000 miles

Fig. 59: Occurrences of post-Acheulean/Sangoan stone industries discussed in chapter 4
120 afric an archaeolog y

earlier material at Kalambo, the Lupemban does not occur on undisturbed
¬‚oors, but is incorporated in a thick rubble layer representing scree and
erosion in the valley after a period of downcutting ( J. D. Clark 1964, 2001a).
Here and at several north-Angolan sites pollen has been preserved which sug-
gests that the vegetation and climate at this time did not differ signi¬cantly
from those of the present. Also provisionally attributed to the Lupemban,
although the characteristic bifacial points are not represented in this spe-
cialist context, is a butchery site at Mwanganda in northern Malawi ( J. D.
Clark and Haynes 1970), for which an age of at least 300,000 years is now
considered probable. The implements, scattered among the bones of a sin-
gle elephant, consist almost exclusively of edge-retouched ¬‚akes, with a few
It is now recognised that there is no adequate basis for the multi-phase
sequence of post-Acheulean industries formerly proposed for the Kinshasa
area (Cahen and Moeyersons 1977; Cahen 1978). The artefacts occur in poorly
strati¬ed sands, as at Gombe Point, and it is now clear that considerable mix-
ture has taken place, both between artefacts originally deposited at different
levels, and of the materials sampled for radiocarbon dating. Consequently,
neither the typological composition of the industries nor their chronology
should be regarded as securely established.
At Katanda on the Semliki River north of Lake Edward/Rutanzige in east-
ernmost D. R. Congo, a series of mode-3 industries has been recovered from
contexts for which an age between 170,000 and 80,000 years is indicated
(Brooks et al. 1995; Yellen 1996). Apparently associated with these stone
industries are barbed bone harpoon-heads akin to those previously recorded
from nearby Ishango and for which a much later date has been widely
accepted (Brooks and Smith 1987; see also pp. 157--8 below). Controversy con-
tinues to surround the high antiquity claimed for the Katanda harpoons,
which has been proposed as supporting the case for the early development
of modern human behaviour in sub-Saharan Africa, discussed and evaluated
below (pp. 143--5). In view of the widespread distribution of such harpoons in
later times (Yellen 1996), it would be unwise to accept the great age proposed
for the Katanda examples until further evidence is available.
The development of microlithic mode-5 technology in central Africa has
conventionally been considered in two regional subdivisions (cf. Cornelissen
2002). In the savanna regions such as those in much of Zambia and southern
Angola, on the one hand, true mode-5 industries appeared at an early date,
as noted on p. 109, having their roots in the earlier industries designated
Bambata. An early microlithic occurrence at Matupi in the extreme north-
east of D. R. Congo (Van Noten 1977; Mercader and Brooks 2001) is often con-
sidered in this connexion. Probably extending back for some 40,000 years,
Regional diversification and specialisation 121

it belongs to a time when, in contrast with its present situation, Matupi
apparently lay a short distance outside the equatorial forest. On the southern
fringes of the forest, on the other hand, a local largely microlithic industry
known as the Tshitolian gradually developed from the preceding Lupemban
tradition. A gradual reduction in artefact size is apparent through the
Lupemban--Tshitolian sequence, together with a shift in emphasis from
¬‚akes struck off radial cores to parallel-sided blades. Backed microliths are
frequent in the Tshitolian, especially the ¬‚ared trapezoidal form known
as petits tranchets that may have been hafted as transverse arrow-points
(Figs. 43, 51); these became smaller and more numerous with the passage of
time (Cahen and Mortelmans 1973; S. F. Miller 1988). Pointed core-tools and
leaf-shaped points continued, generally smaller than previously. In south-
ern D. R. Congo it is noteworthy that sites in the more densely forested
river valleys have yielded a higher proportion of backed microliths, while
assemblages from the more open plateaux contain larger bifacially worked
tools. The distribution of comparable industries extends to the northwest
into Congo, Gabon and Cameroon (Clist 1989a; Lanfranchi and Schwartz
It may now be suggested that the dichotomy between the Tshitolian and
its more dominantly microlithic counterparts may not have been so clearly
de¬ned as was once thought. The classic Lupemban industries are now seen
as a localised manifestation of a very widespread technological tradition, its
apparent idiosyncrasy perhaps partly due to the raw material that was avail-
able. In the broadest sense, environmental factors will have had some effect
on human activities and the types of tool that were required to perform
them, but the overall picture that is now slowly emerging is a mosaic of
variation on a widespread common theme, rather than two clearly de¬ned
Studies of the archaeology of central Africa are greatly hampered by the
sparseness and uneven coverage of research (cf. Fig. 59), by the almost com-
plete absence of faunal and ¬‚oral remains and the total lack of hominid
fossils. Such rock art as has been located in this area (where rocky outcrops
are often rare) appears to be of late date; it is discussed in chapters 7 and 8.
In addition, as has been shown above, doubt has been cast upon the
stratigraphic integrity of many of the open-air Kalahari Sand sites which
formed the basis for much of the sequence previously proposed. Despite
these problems, the region is seen as having occupied a key position in
continent-wide trends during the later stages of the Middle Pleistocene and
the early part of the Late Pleistocene (see Fig. 7 on p. 25). Recent research has
emphasised both the wide distribution and the signi¬cance of Lupemban
industries. No longer recognised simply by the presence of large foliate
122 afric an archaeolog y

bifacial points, these are now seen as occupying a crucial position in the
development of inter-regional variation and of the transition from mode
3 to a mode-5 (microlithic) technology (Barham 2002b). Lupemban af¬ni-
ties are recognised as far south as Namibia and, as will be argued below,
the North African Aterian may have greater af¬nity with the sub-Saharan
Lupemban than with broadly contemporary European or Levantine techno-
logical traditions. It may be anticipated that, as research progresses, geo-
graphical and chronological subdivisions of the Lupemban tradition will
be recognised, and their contribution to later developments more widely

Eastern Africa
Some areas, notably in Kenya, have been intensively examined, but else-
where enormous areas remain virtually unexplored by archaeologists, whose
attention has tended until recently to be concentrated on the very rich sites
belonging to the earliest Stone Age periods. Although in some instances
they are not well documented, eastern Africa has yielded a number of fos-
sils which may be regarded as early representatives of anatomically modern
humans. Potentially the most signi¬cant is the recently announced discov-
ery at Herto in the Middle Awash Basin of three skulls, apparently dated
between 160,000 and 150,000 years ago (White et al. 2003). They display
physical characteristics which clearly belong to Homo sapiens, with some
archaic features, being signi¬cantly more modern, for example, than the
Bodo, Broken Hill or Hope¬eld specimens. Unfortunately, the archaeologi-
cal associations of the Herto skulls remain unclear, both Acheulean-type and
mode-3 artefacts being reported from the same horizons ( J. D. Clark et al.
2003). Particular interest also attaches to three specimens from deposits of
the Kibish Formation in the lower Omo Valley of Ethiopia (Day and Stringer
1991; Day et al. 1991). Neither their stratigraphic positions nor their cul-
tural associations are known with any clarity, but direct dating suggests
Late Pleistocene ages analogous to those from Klasies River Mouth (p. 101;
Klein 1999). An earlier and even more poorly documented discovery from
Singa near the Blue Nile in east-central Sudan (Stringer 1979; Stringer et al.
1985; McDermott et al. 1996) shows comparable modern features. Previously
considered recent, it is now recognised as belonging to the general period
around 150,000 years ago. Probably broadly contemporary are a skull from
Ngaloba at Laetoli (Magori and Day 1983; Klein 1999) and isolated teeth from
Mumba rockshelter (Br¤uer and Mehlman 1988), both in Tanzania.
Eastern Africa is a region of great environmental variation, from
glaciated mountains, to semi-arid deserts, to mangrove-fringed ocean shores
Regional diversification and specialisation 123

Fig. 60:
artefacts: 1, 2,
from Musolexi,
Angola (after J. D.
Clark 1963); 3--5,
from Kalambo
Falls (after J. D.
Clark 2001a)

(W. M. Adams et al. 1996). These features have themselves varied through
time but are, throughout, re¬‚ected in the archaeology. In the area around
Lake Victoria, which was for much of this time an internal drainage area
with no outlet to the north, the immediately post-Acheulean industries were
of Sangoan type, discussed in chapter 3. Presumably later are Lupemban-type
industries (Fig. 60), best known from Muguruk (McBrearty 1988) and nearby
sites such as Yala Alego (O™Brien 1939; S. Cole 1964). This sequence is paral-
leled by abundant discoveries in Rwanda and Burundi (Nenquin 1967; Van
Noten 1982). Throughout this area the published data are unfortunately vir-
tually restricted to typological descriptions; there are few if any undisturbed
assemblages from sealed strati¬ed contexts and no reliable age determi-
nations. Recent research on the Kapthurin Formation near Lake Baringo
in Kenya has, however, done much to clarify these matters and to pro-
vide con¬rmation both of the high antiquity of these developments and
124 afric an archaeolog y

of their relationship with the Sangoan (Cornelissen 1992; McBrearty et al.
1996; McBrearty 1999, 2001; Deino and McBrearty 2002).
In addition to the Middle Awash Basin, where research is ongoing (Kalb
et al. 1982; White et al. 2003), scattered localities in Ethiopia and Eritrea have
yielded lithic industries which may be regarded as early manifestations of
mode-3 technology; unfortunately they are at present incompletely docu-
mented. At Melka Kunture near Addis Ababa, horizons strati¬ed above the
Acheulean material noted in chapter 3 contain a series of assemblages in
which bifacial tools became progressively rarer and smaller, being gradu-
ally replaced by sub-triangular points and ¬‚akes with shallow retouch along
their longer sides (Hours 1973; Chavaillon 1979). Full details of these discov-
eries have not yet been published, nor are age determinations available. It
is, however, relevant to note potassium/argon dates about 180,000--150,000
years ago for a site used for the manufacture of mode-3 artefacts on obsidian
outcrops at Gadamotta near Lake Zwai, some 70 kilometres south of Melka
Kunture (Wendorf and Schild 1974).
Mode-3 assemblages, characterised by the presence of sub-triangular
points and edge-retouched pieces, often made on ¬‚akes struck from radial
cores similar in general terms to those from south-central Africa attributed
to the Bambata industry, are widely distributed in eastern Africa. As argued
above, the distinction between the industries designated Bambata and those
whose Lupemban af¬nities are now recognised is less clearly de¬ned
than the conventional nomenclature would imply. In eastern Africa, the
chronology of these industries, like the activities and life-styles of their
practitioners, awaits detailed investigation (but see Barut 1994; Merrick et al.
1994; Ambrose et al. 2002), although they appear to have been broadly anal-
ogous to those of their counterparts further to the south. Industries of this
type have been noted in Ethiopia and Somalia ( J. D. Clark 1954; Brandt
1986), where they may extend back into the relatively moist conditions of
the Last Interglacial period. At Porc Epic cave near Dire Dawa such an indus-
try is associated with a human mandible of near-modern form ( J. D. Clark
and Williams 1978; Br¤uer 1984). At Olduvai Gorge such material occurs
overlying Bed IV in the Ndutu Beds (M. D. Leakey et al. 1972), where its
age may exceed 100,000 years. At nearby Lake Eyasi a similar industry with
some Sangoan-like artefacts is associated with fragmentary human crania
which show morphological features akin to those of the specimen from
Broken Hill mine, Kabwe, attributed to Homo heidelbergensis (Mehlman 1987).
Many further occurrences await detailed publication (cf. Anthony 1972; Kalb
et al. 1982). Related mode-3 industries also occur in the early stages of several
of the local sequences which have been established on the basis of cave or
rockshelter excavations, notably those at Nasera (Mehlman 1977) on the east-
ern edge of the Serengeti Plain, at Lukenya Hill (Gramly 1976; Barut 1994)
Regional diversification and specialisation 125

near Nairobi, and at Enkapune ya Muto west of Lake Naivasha (Ambrose
In some parts of eastern Africa, as in certain more southerly regions, the
mode-5 techniques of bladelet production and of backing retouch can now
be traced back to far earlier periods than were previously considered rele-
vant. New data are available relating to these developments in the central
Rift Valley and adjacent highlands of Kenya (Kyule et al. 1997; Ambrose 1998b;
Ambrose et al. 2002). Technological developments here, although idiosyn-
cratic, are easily discerned since the majority of artefacts are made from
obsidian. Deep deposits at Marmonet Drift in the Naivasha-Nakuru basin pre-
serve a mode-3 industry interstrati¬ed with layers of volcanic ash for which
absolute dates have not yet been announced. The archaeological sequence
at Enkapune ya Muto rockshelter, high on the Mau escarpment, extends
from at least 70,000 years ago into recent times. A mode-5 industry desig-
nated Sakutiek, with numerous small edge-retouched tools, backed ¬‚akes
and bladelets, began there as much as 50,000 years ago, replacing a mode-3
By at least 20,000 years ago, as in northern and eastern Zambia, there
is evidence that microlithic industries containing quantities of backed
bladelets and some geometrical microliths were widespread in the high-
lands of Tanzania and southern Kenya. The best-described occurrences are
still those from Kisese rockshelter in central Tanzania (Inskeep 1962) and
at Lukenya Hill (Gramly 1976) where such artefacts are associated with a
fragmentary human skull showing -- it has been claimed -- features resem-
bling those of recent negroid populations (Gramly and Rightmire 1973). A
similar but rather later industry comes from Buvuma Island in Ugandan
waters of Lake Victoria (Van Noten 1971). Somewhat earlier material from
the Naisiusiu Beds near the top of the Olduvai Gorge sequence (M. D. Leakey
et al. 1972) may partly owe its distinctive appearance to the fact that many
of its artefacts are made of obsidian.
In the more northerly parts of eastern Africa, there are very few dated
sequences that can be compared with those noted above. At Laga Oda in the
escarpment of the southeast Ethiopian plateau near Dire Dawa, a backed-
microlith industry was established at least 16,000 years ago ( J. D. Clark and
Williams 1978). In more northerly parts of Ethiopia an industry based on the
production of large blades may have intervened between the mode-3 and the
backed-microlith industries, as at Gobedra and Anqqer Baahti rockshelters
near Aksum in Tigray (D. W. Phillipson 1977b; Finneran 2000). J. D. Clark™s
(1954) proposal of a similar ˜Hargeisan™ stage in northern Somalia is not
borne out by more recent research (Brandt 1986).
A comparable blade industry is clearly demonstrated further to the south:
the Eburran industry, formerly known as the Kenya Capsian, is restricted
126 afric an archaeolog y

Fig. 61: 1--5,
artefacts of
modes 4 and 5
from Gobedra
(after D. W.
1977b); 6--9,
Eburran artefacts
from Gamble™s
Cave (after L. S. B.
Leakey 1931)

to a small area near Lake Nakuru in the eastern Rift Valley of south-
central Kenya. It is best represented at Gamble™s Cave and Nderit Drift,
dated between 13,000 and 9000 years ago (L. S. B. Leakey 1931; Isaac et al.
1972; Ambrose et al. 1980), although related industries continued into more
recent times, as at Enkapune ya Muto (Ambrose 1998b). The ¬ne large arte-
facts of the early Eburran (Fig. 61) -- it is typi¬ed by large backed blades,
crescent-shaped pieces, edge-retouched ¬‚akes and ˜burins™ with a sharp trans-
verse edge produced by removal of a narrow spall -- are almost all of ¬ne
obsidian; the similarity with more northerly blade industries may be largely
By 10,000 years ago backed-microlith industries were generally ubiquitous
in eastern Africa. These industries, despite their general similarities, show
a confusing complexity of variation; no really convincing and meaningful
classi¬cation of them has yet been proposed. In the north our knowledge
comes from widely scattered localities: Gobedra and Laga Oda in Ethiopia,
for example, as well as from Gorgora near Lake Tana (L. S. B. Leakey 1943).
In Somalia, the Doian industry has unifacial and bifacial points alongside
Regional diversification and specialisation 127

backed microliths, notably at Gogoshiis Qabe and Guli Waabayo ( J. D.
Clark 1954; Brandt 1986). Further to the south were the specialised ¬sh-
ing settlements of this period beside Lake Turkana, discussed in chapter 5.
Elsewhere, except in the parts of the Rift Valley where obsidian was plen-
tiful and whence it was sometimes evidently traded to neighbouring areas
(Merrick and Brown 1984; Merrick et al. 1994), the microlithic industries are
generally of quartz and of rather informal character. The longest and best-
documented sequences are those at the Lukenya Hill (S. B. Kusimba 2001),
Kisese and Nasera sites noted above. Similar material comes from northern
and western Uganda (e.g. Nelson and Posnansky 1970).
The makers of the microlithic industries of between 10,000 and 2000 years
ago were probably responsible for the earliest extant East African rock paint-
ings. Such paintings are only surely known from north-central Tanzania and
show some stylistic parallels with their counterparts in southern Africa.
Isolated naturalistic animal ¬gures and stylised humans are the most fre-
quent motifs, shown either in outline or in ¬‚at monochrome. The best-
known sites are in the Kondoa and Singida areas (Fosbrooke et al. 1950;
Masao 1979; M. D. Leakey 1983; Anati 1986). The naturalistic art tradition
did not continue here into such recent times as it did in southern Africa,
and its interpretation is more problematic. In general terms, however, it is
possible that it ful¬lled much the same functions; certain features, such as
the frequency of eland representations, are common to both areas, although
it would be premature to suggest that shamanistic trance experience con-
tributed as much to rock art in East Africa as it did further to the south. It
should not be assumed that the belief-systems and practices of the southern
African San were ubiquitous in what is now Tanzania (Schepartz 1988; see
also Morris 2003).
Although the data are from widely scattered sites in quite different envi-
ronments, and although long detailed sequences are so far lacking, the suc-
cession of post-Sangoan stone industries in eastern Africa is now seen to have
been broadly analogous to, and lasted for about as long as, that revealed by
more intensive research further to the south. With the possible exception of
Mumba in northern Tanzania (Mehlman 1991) there is, however, no evidence
for any very early appearance of full-¬‚edged mode-5 technology such as that
represented by the Howieson™s Poort industries south of the Limpopo. The
earliest post-Acheulean industries are not based to the same extent on ¬‚ake-
blade production but, in contrast to the situation in the south, true mode-4
industries are attested in the more northerly parts of eastern Africa. The
reasons for these distinctions will not be properly understood until compre-
hensive data relating to the economy and life-style of this period have been
recovered from eastern African sites.
128 afric an archaeolog y

West Africa
Like those of the Acheulean and the Sangoan, the subsequent stone indus-
tries of West Africa have not been thoroughly investigated, although abun-
dant remains are known to exist (Fig. 62). It is clear that the region saw
extensive environmental changes involving signi¬cant shifts in vegetational
zones. Considerable confusion surrounds the sequence of West African stone
industries in post-Sangoan times because so much of our information is
based upon undated collections from disturbed contexts or surface expo-
sures. However, there can be little doubt that in several areas, such as
Cap Vert in Senegal, southwestern Mali and central Guinea, there existed
assemblages typi¬ed by large, lance-shaped points akin to, but generally
cruder than, those of the central African Lupemban. It is dif¬cult to be
certain about the true distribution of this material, because of the super¬-
cial similarity between some of its artefacts and those of much later times
(cf. MacDonald and Allsworth-Jones 1994). More clearly recognised are the
generalised mode-3 industries based on radial cores and ¬‚akes which were
made into points and edge-retouched ¬‚akes rather like those from more
easterly and southerly regions, and also showing super¬cial similarity to
artefacts from further to the north that have been designated Levalloiso-
Mousterian (see p. 131). In West Africa, such material occurs widely in river-
gravel deposits in Sierra Leone and on the Jos Plateau of Nigeria, where
they belong to a later cycle of erosion and deposition than does the local
Acheulean. A particularly rich occurrence is in the outwash gravels below
Zenebi Falls, for which a very late radiocarbon date is most unlikely to repre-
sent the true age, which remains unknown since no undisturbed assemblage
of this type has been excavated in situ anywhere in West Africa (Soper 1965;
Shaw 1978; Allsworth-Jones 1986, 2001).
There has been more research into later periods, at least in some areas,
and a rather fuller picture may thus be drawn. The only sequence which
illustrates the early development of mode-5 industries in West Africa is at
the rockshelter of Shum Laka in the Grass¬elds of northwestern Cameroon
(de Maret et al. 1987; Lavachery 2001; Cornelissen 2003). Here, a microlithic
industry in quartz was being produced signi¬cantly before 30,000 years ago
and continued with little discernible change through a period that was envi-
ronmentally relatively stable until less than 10,000 years ago. The date of its
¬rst appearance is not yet known, but earlier industries of Sangoan type are
recorded in the same general region at Njuinye (Mercader and Marti 2003).
At the rockshelter of Iwo Eleru in the now forested zone of southwestern
Nigeria, a microlithic industry is known from about 12,000 years ago (Shaw
and Daniels 1984). This early horizon at Iwo Eleru also yielded a human
Regional diversification and specialisation 129

Fig. 62: Mode-3
artefacts from
Nigeria (after
Shaw 1978)

burial which is stated to show negroid physical features (Brothwell and Shaw
1971). The stone assemblage, which included a very low proportion of inten-
tionally retouched tools, was characterised by crescent-shaped, triangular
and trapezoidal microliths. It continued with only minor change for some
9000 years. Around 5500 years ago, pottery and ground-stone artefacts made
their appearance. These late pottery-associated industries are described and
discussed in chapter 6. Comparable microlithic industries without associ-
ated pottery are known elsewhere in Nigeria, notably at Mejiro near Oyo
Ile (Willett 1962), at Rop on the Jos Plateau (B. Fagg et al. 1972), probably in
the lower levels at A¬kpo rockshelter in eastern Nigeria (Andah and Anozie
1980), and also at Bingerville in Ivory Coast where its age is similar to that
at Iwo Eleru (Chenorkian 1983).
Elsewhere in West Africa, the earlier microlithic industries, before the
start of pottery manufacture, have only been found in a few areas, but
may have been originally widespread (Shaw 1981). Such an occurrence at
Rim in Burkina Faso has an age of over 5000 years. In Ghana there are
numerous surface ¬nds which may belong to this period, as at Adwuku,
but none has been dated precisely. By contrast, sites in Guinea and Sierra
Leone, such as Yengema (Coon 1968), have yielded crudely ¬‚aked core-tools,
edge-retouched ¬‚akes and a few backed blades. Such material, like analo-
gous ¬nds from the Congo Basin, apparently represents an adaptation to
forest life; if so, the presence of a true backed-microlith industry at Iwo
Eleru remains to be explained, unless that site™s surroundings have become
only recently afforested. Material from Kourounkorokale near Bamako is
described in chapter 5. In the far west, in Senegal, the lagoonside Tiemassas
130 afric an archaeolog y

Fig. 63: West
African mode-5
artefacts: 1--3,
bifacial points
from Tiemassas
(after Dagan
1956); 4--8,
microliths from
Rop (after Shaw

site 80 kilometres south of Dakar has yielded a pre-pottery microlithic
occurrence. Backed blades and crescents here are large and crude: they are
associated with bifacially ¬‚aked leaf-shaped points (Fig. 63). More than one
phase of occupation is probably represented in these extensive exposures
for which, unfortunately, no absolute dating is currently available (Dagan
1956, 1972).
Despite the very incomplete nature of our evidence, the following tenta-
tive synthesis can be offered. It is clear that mode-5 technology in West
Africa began more than 30,000 years ago, although it is not yet ¬rmly
attested west of Cameroon until about 12,000 years ago. For many mil-
lennia prior to this date the Sahara was largely uninhabited, and so the
microlithic industries to the south, as in other areas of sub-Saharan Africa,
were probably indigenous developments. In some densely forested regions
of West Africa, such as southeastern Nigeria, Sierra Leone and Guinea, non-
microlithic industries continued until the last few millennia (Shaw 1981;
MacDonald and Allsworth-Jones 1994). The characteristic artefacts of these
industries are crude core-tools, hoe-like or axe-like in form, perhaps used for
forest clearance and for digging. Cultivation of tubers may have begun at
an early date in those West African areas where yams are today the staple
food. The plausibility of such a contrast, perhaps between about 6000 and
4000 years ago, between an essentially hunting life-style in the West African
savanna and a vegecultural one in the forests, will be further explored in
chapter 6.
Regional diversification and specialisation 131

North Africa and the Sahara
The archaeology of the northern parts of Africa has for the most part
been studied in an essentially Mediterranean, rather than an African, con-
text. Both the conceptual framework and the terminology conventionally
employed thus differ in some important respects from those used in other
parts of the continent (cf. Kleindienst 2001). In the present work it is the
African aspects and connexions that are stressed, and North Africa will be
shown to have occupied a more central and innovative place in prehistory
than some earlier accounts would suggest.
As the following account will make clear, North Africa and the Sahara have
seen environmental changes at least as great as those in any other part of
the continent, and the region™s prehistory must be viewed in that context.
The time of the Acheulean in the Sahara appears to have been followed
by an arid phase when many previously inhabited areas were abandoned,
which may explain the episodic nature of the archaeological record in many
areas (e.g. Tillet 1985). Sites in Morocco which have yielded very small hand-
axes and cleavers as well as ¬‚ake-tools appear to belong to this arid period:
similar artefacts also occur at a few places in the Sahara, notably in south-
eastern Libya (Biberson 1967; Arkell 1964; J. D. Clark 1980). With the return
of moister conditions more than 130,000 years ago, settlement again became
widespread. Best known from the Algerian and Moroccan Maghreb, the arte-
facts then in use were of the mode-3 type generally known as Mousterian
or Levalloiso-Mousterian, after their closely similar and broadly contempo-
rary European and Levantine counterparts. Although small, heart-shaped
handaxes occur in some of these North African Mousterian assemblages,
light-duty tools are the most characteristic element, being made generally
on ¬‚akes removed from prepared or radial cores (van Peer 1991). Long edge-
retouched ¬‚akes and sub-triangular points, perhaps for projectiles, were the
most common implement types.
The industries which subsequently became widespread have, despite their
diversity, been conventionally grouped together under the name Aterian,
after Bir el Ater, near Tebessa, Algeria (Camps 1974; J. D. Clark 1980;
Kleindienst 2001). The Aterian is generally de¬ned by the presence of a vari-
ety of ¬‚ake tools possessing well-worked tangs, which may have facilitated
hafting (Fig. 64). These include not only specimens which resemble projec-
tile points, but also varied shapes of edge-retouched ¬‚akes and pieces which,
bearing little intentional retouch other than that which forms the tang, may
be broken or reworked points. There are, of course, major problems with
de¬ning a distinct cultural or even industrial entity on the basis of a single
morphological feature, especially one whose function is so poorly under-
stood. Artefacts designated Aterian are encountered through most of the
132 afric an archaeolog y

Fig. 64: Aterian
artefacts (after
Camps 1974: 1--5,
from Bir el Ater;
6, 7, from Adrar

Sahara, from the Atlantic coast almost as far east as the Nile. In several parts
of this vast region, as at Adrar Bous in Tenere, bifacial points are another
regular component of the Aterian assemblages ( J. D. Clark et al. 1973). Assem-
blages from later sites in both areas include numerous parallel-sided blades
in addition to ¬‚akes struck from radial cores. This marks a technologi-
cal development parallel to, and perhaps connected with, the advent of
mode-4 industries in Cyrenaica (see p. 134). These general observations apart,
although several local versions have been recognised (e.g. Deb©nath et al.
1986), regional and temporal variation within the Aterian complex remains
very poorly understood, as does its dating. Whatever its status, the Aterian
seems to have been an exclusively African phenomenon, no trace of it
Regional diversification and specialisation 133

having been found on the Iberian peninsula. South of the Sahara, anal-
ogous industries have been reported from northern Ghana (Davies 1967).
In the east, confusion has arisen over the terminology applied to mode-3
industries from the oases of the Western Desert in Egypt (Wendorf et al.
1993; Kleindienst 2001), but it seems clear that those designated Aterian
according to the generally accepted criteria did not extend so far as the Nile
Chronologically, the Aterian seems to belong to a period when reduced
temperatures prevailed in northern Africa, resulting in glaciers on the High
Atlas mountains and a general spread of vegetation zones to lower altitudes.
In the Sahara, evergreen vegetation, primarily of Mediterranean species,
grew in the highlands, and lower evaporation rates ensured that the rivers
¬‚owing from these highlands watered the adjacent parts of the otherwise
relatively dry intervening plains (Van Campo 1975). In such a situation was
the site of Bir Tarfawi in the Western Desert of southwestern Egypt where,
probably between 130,000 and 70,000 years ago, people living beside a shal-
low lake were able to hunt a variety of animals, including gazelle, warthog
and ostrich (Wendorf et al. 1993). Species included ones which are now
restricted to the Mediterranean zone as well as some of more southerly
af¬nities; rhinoceros remains have been found on Aterian sites as far to the
north as El Guettar in southern Tunisia (Camps 1974). In southwestern Libya,
the Aterian appears to be older than 60,000 years (Cremaschi et al. 1998;
Cremaschi 2002), while in Morocco it may have continued until the last
glacial maximum as steadily increasing aridity made much of the Sahara
progressively unsuited to human settlement. Neither the Nile nor the Niger
had by then attained its present course or extent: the upper reaches of the
Niger ¬‚owed northeastwards into the present inland delta where its waters
were lost by evaporation. Lake Chad was probably almost completely dry
(Williams and Faure 1980). Such conditions, which coincided with a severe
period of glaciation in northern Europe, continued in the Sahara until about
13,000 years ago.
The identity of the people responsible for the North African and Saha-
ran mode-3 industries has recently been clari¬ed. At Jebel Irhoud, Kebibat
and Sale in Morocco and at Haua Fteah in Cyrenaica human fossils
recovered in association with Levalloiso-Mousterian artefacts were formerly
believed to resemble contemporaneous European neanderthalers; they are
now regarded as closer to anatomically modern people (Ennouchi 1962;
McBurney 1967; Hublin 1993, 2001; Klein 1999). The implication of this
re-assignment is that there can have been little population transfer across
the Straits of Gibraltar at this time, despite the apparent similarity of
the pre-Aterian stone industries on the two sides during the ¬nal Middle
Pleistocene (but see Barton et al. 2001). By later, Aterian, times, a fully modern
134 afric an archaeolog y

population is attested in North Africa -- as at Dar es Soltan, Morocco -- of the
heavily built type sometimes designated ˜Mechta-Afalou™, which continued
to inhabit the Sahara until much later millennia (Hublin 1993). Continu-
ity between these two human types seems likely but need not be assumed
for, at the few sites (such as Adrar Bous in Niger and Bir Sahara in the
Egyptian Western Desert) where Aterian artefacts are strati¬ed over those of
Levalloiso-Mousterian type, there was once again an arid intervening period
during which much of the area may have been uninhabited.
Mention has already been made of the archaeological sequence in
Cyrenaica. This part of northern Libya possesses, at the great cave of Haua
Fteah, the most complete succession of Upper Pleistocene industries known
from any part of North Africa (McBurney 1967; Close 1986). The oldest levels
at this site have not yet been excavated, but it is known that Levalloiso-
Mousterian occupation of the cave was established by at least 60,000 years
ago. This industry continued through the period when its southerly Aterian
counterpart ¬‚ourished in the Sahara. It was preceded, in the lowest levels yet
investigated at Haua Fteah, by an apparently quite distinct industry, which
has been named the Libyan Pre-Aurignacian, based upon the production of
parallel-sided blades struck from prismatic cores (Fig. 65). The makers of
this mode-4 industry were accomplished hunters of wild cattle, gazelle and
zebra; unlike their successors they also collected sea-food. This industry is
not known from any other African site, although broadly comparable tech-
nology is attested in South Africa at an analogous time-depth (cf. pp. 98--102
above). Fairly close parallels have also been suggested in the Levant. Its local
precursors in Cyrenaica remain unknown.
After the Libyan Pre-Aurignacian phase, Levalloiso-Mousterian occupation
of Haua Fteah continued for more than 20,000 years until about 40,000
years ago when it was abruptly replaced by a more developed blade indus-
try, called Dabban after another Cyrenaican site. The Dabban clearly belongs
with the great complex of broadly contemporary mode-4 industries in
Europe and western Asia conventionally known as the Upper Palaeolithic.
Its most characteristic features were backed blades, edge-retouched ¬‚akes
and narrow, chisel-like ˜burins™. Here again, the closest connexions of this
phase of the Haua Fteah sequence appear to be with the Levant, but it
should be noted that it ¬rst appeared in Cyrenaica at broadly the same
time as the beginning of the Saharan desiccation which led to the eclipse
of the Aterian. Cooler conditions at that time are likewise indicated at Haua
Fteah. With relatively little change, the Dabban occupation continued until
about 14,000 years ago, when it was abruptly replaced, possibly with some
overlap, by a mode-5 industry known as the Eastern Oranian, characterised
by large numbers of small backed bladelets.
Regional diversification and specialisation 135

Fig. 65: Artefacts
from Haua Fteah
(after McBurney
1967): 1--3, Libyan
4--6, Levalloiso-
7--9, Dabban

The Eastern Oranian, known also at other Libyan sites such as Hagfet et
Tera, takes its name from its apparent close similarity to an industry of the
Maghreb farther to the west. This Oranian industry was widespread in the
North African coastland and hinterland from south of Rabat as far east as
Tunis (Camps 1974, 1975). In recent years the Oranian has been more gen-
erally referred to as ˜Iberomaurusian™. This term does not follow the stan-
dard practice of naming an archaeological industry after a site at which
it has been recognised and described, and it carries unwarranted implica-
tions concerning connexions between Africa and the Iberian peninsula; the
original name ˜Oranian™ is here retained. There is no evidence for continuity
between the Aterian and the Oranian; probably there was a long intervening
136 afric an archaeolog y

period when much of the region was uninhabited. So far, the earliest known
Oranian occurrences are those located in Morocco and Algeria; at Tamar
Hat on the coast of eastern Algeria an occupation dated some 20,000--
16,000 years ago is attributed to this industry (Saxon et al. 1974; see also
J. Roche 1971). The industries here designated Oranian and Eastern Oranian
are restricted to zones of Mediterranean vegetation along the North African
littoral in the Maghreb and Cyrenaica, being interrupted only at the Gulf
of Sirte where desert conditions extend almost to the shore. At the time
when these industries were made, sea levels were signi¬cantly lower than at
present, and their distribution may have been continuous across the coastal
plain that was then exposed.
Remains of the Oranian people have been discovered in several extensive
cemeteries, notably at Taforalt, Columnata some 200 kilometres southwest
of Algiers, and Afalou bou Rhummel, adjacent to Tamar Hat. They belong
without exception to the anatomically modern Mechta-Afalou type, noted
above, which had a long ancestry in North Africa (Chamla 1978). This is a
further, albeit inconclusive, argument in support of a local origin for the
Oranian industry. They were buried in extended or, at later sites, contracted
positions. At Columnata, several skeletons were covered with settings of
large stones, in one case capped by horns of wild cattle, Bos primigenius. The
presence in the graves of red ochre and perforated shell indicates, appar-
ently for the ¬rst time in the Maghreb, the practice of personal adornment
(Fig. 66).
During the Oranian occupation, species of pine and oak which are now
restricted to high altitudes in the Atlas Mountains were more widely dis-
tributed, indicating cooler climatic conditions during the ¬nal stages of the
last glaciation. In this environment abundant animal species were available
for hunting; this activity may be re¬‚ected in the survival of numerous small
open-air sites, evidently brie¬‚y occupied, which contrast with the large,
repeatedly re-occupied cave sites. Barbary sheep were intensively hunted and
there is disputed evidence that, at Tamar Hat, the herds may have been
managed by selective culling at a time when grazing on the coastal plain
was curtailed by rising sea levels (Saxon et al. 1974; Klein and Scott 1986;
di Lernia 1999). Land and water molluscs were also collected, particularly in
later Oranian times, following the rise in sea level.
By about 10,000 years ago the Oranian unity had broken down and sev-
eral local short-lived blade industries, such as that named Columnatan, are
found in the Maghreb. Of particular interest is the so-called Typical Capsian,
attested from about 8500 years ago in a restricted area of the Algeria/Tunisia
border country south of Tebessa, as at Relilai. It is characterised by the large
size of its artefacts, among which burin-shaped pieces and backed blades
varyingly predominate. Many of its sites are middens of land-snail shells.
Regional diversification and specialisation 137

Fig. 66: 1--4,
Oranian artefacts
from Taforalt;
5--12, Capsian
artefacts from
Relilai and
Mechta el Arbi
(all after Camps
1974). 1 and 12
are bone points,
5 is a fragment
of decorated

Far more widespread, being found as far a¬eld as western Algeria, is the
more microlithic Upper Capsian, which is somewhat earlier in date than
the Typical Capsian (P. E. L. Smith 1982) and is also frequently found on
shell-midden sites. At Columnata it replaced the Columnatan about 7500
years ago; at Relilai the corresponding event seems to have taken place a few
centuries earlier. Crescentic, triangular and trapezoidal backed microliths,
backed bladelets and notched or denticulated ¬‚akes are the most frequently
encountered stone tools, together with a variety of bone artefacts. Much
138 afric an archaeolog y

Fig. 67: Capsian
carvings on stone
and ostrich-
eggshell (after
Camps 1974)

artistic ingenuity was applied to the engraved decoration of ostrich-
eggshells, and to the carving of small stone ornaments (Fig. 67). In addi-
tion to the collection of land-snails, hunting is well attested, in an environ-
ment which shows reduction of forest cover. Industries related to the Upper
Capsian (which, like its predecessors, may have been the work of a Mechta-
Afalou population, although it has been suggested that some skeletons show
af¬nity to more recent Mediterranean peoples) are also found in the adja-
cent Saharan regions (Sheppard and Lubell 1990) and at Haua Fteah, where
it has been called Libyco-Capsian. The greater part of the Sahara remained
uninhabited until around 12,000 years ago; discussion of its resettlement
after that time is best postponed until chapter 5.
The archaeological sequence of the Nile Valley, even at this early period,
differs suf¬ciently from those of neighbouring areas that it is best consid-
ered separately. At least during the Pleistocene and Holocene phases when
better-watered conditions prevailed, communication between the central
Saharan highlands and the Nile would have been facilitated by the now-dry
Wadi Howar, ¬‚owing eastwards from Ennedi to join the Nile near Dongola
(Keding 1998). As a result of investigations which originated as rescue
operations prior to the ¬‚ooding of Lake Nasser, the prehistory of Nubia
is more completely understood than that of neighbouring regions. Since
the 1970s, however, most research in this region has concentrated on later
Regional diversification and specialisation 139

Post-Acheulean industries named, following Levantine terminology,
Levalloiso-Mousterian are best known from the area upstream of Luxor as
far as Sudanese Nubia. Traces of their counterpart further downstream have
presumably been destroyed by erosion or deeply buried by accumulations
of silt. Several successive stages have been recognised, of which the earli-
est, designated Mousterian, shows considerable diversity, with at least three
variants that are believed to re¬‚ect different activities (Wendorf and Schild
1976). In one of these variant Mousterian industries small handaxes occur,
although never in large numbers. Wild cattle (Bos primigenius) was the pre-
ferred prey at these sites, which are too old to be dated by the radiocarbon
Later Mousterian assemblages show some Aterian af¬nities and may be
broadly contemporary with that complex in the desert to the west. These
artefacts show accomplished use of a prepared-core technique, with rare
tanged pieces and bifacial leaf-shaped points similar to those of the southern
Saharan Aterian. At Arkin near Wadi Halfa stone-knapping production sites
are located at the raw-material outcrops, where the leaf-shaped points were
roughed out (Chmielewski 1968). Nearby, at Khor Musa, is a contemporary
occupation site where, for the ¬rst time in this region, most of the food
debris comprises the remains of ¬sh (Wendorf 1968). Further to the south,
near Khartoum, a comparable industry has large, elongated foliate points
akin to those of the Lupemban, possibly suggesting an extension of the
equatorial forest environment along the valley of the White Nile at this
time ( J. D. Clark 1980).
Another site at Khor Musa has given its name to the ¬nal phase of this
group of industries, where the earlier artefact types are accompanied by
blade tools and burin-shaped pieces of the same general type as those in
the Libyan Dabban industry. The Khormusan is now considered to be broadly
contemporary with the Dabban, at least 35,000 years old (Fig. 68). It occurs
on extensive sites where both ¬sh and land mammals were eaten (Wendorf
1968; Wendorf and Schild 1976). Further downstream, near Qena, chert for
tool-making was obtained from underground mines (Vermeersch et al. 1990;
Vermeersch 2002); a child™s burial, described as ˜anatomically modern™, was
found nearby, associated with mode-3 artefacts dated between 80,000 and
50,000 years ago (Vermeersch et al. 1998).
Following the Khormusan, at least 25,000 years ago, began a period where
the prehistory of the Nile Valley was characterised by even more diverse local
industries of restricted temporal and spatial distribution. One of the earli-
est, the Halfan, is found on hunting/¬shing camps in the Wadi Halfa area.
Its tools, made on small blades, are in clear contrast with those of the
Khormusan, which it eventually replaced (Marks 1968). Downstream, in
Upper Egypt, the time-span of the Halfan was taken by non-microlithic
140 afric an archaeolog y

Fig. 68: Artefacts
from Nubia (after
Wendorf 1968):
1--4, Khormusan;
5--7, Halfan

industries grouped under the name Edfuan, in which blade technology
accompanied a continuation of the Levallois technique. At Wadi Kubbaniya
near Aswan, intensive use of vegetable foods is attested about 18,000 years
ago, although the claim for the early use of barley at this site has been dis-
proved by subsequent research (Wendorf et al. 1989; Barakat 2002). From
about 17,000 years ago the local industries became progressively more
microlithic, as with the Fakhurian which was contemporary with the later
Edfuan and Halfan (Lubell 1974).
These markedly disparate industries co-existing at the same time so close
together suggest that we are dealing not merely with activity variants, but
with the presence of distinct population groups whose boundaries were
Regional diversification and specialisation 141

not necessarily coterminous with those of the stone-tool industries cur-
rently recognised. Perhaps, in the closely circumscribed habitat provided by
the Nile Valley, pressure of numbers was already stimulating technological
innovation as part of the competition for control of, or access to, resources.
Alternatively, diverse populations may have been brought together by shared
environmental pressures into a circumscribed area. Such processes may have
stimulated social contact and reinforced group differentiations, identity and
rivalry. This situation continued into the period from about 15,000 to 11,000
years ago when water levels in the Nile, like those in the Saharan and East
African lakes, rose. The important economic innovations which accompa-
nied these developments are the subject of chapter 5.
The foregoing survey of the post-Acheulean stone-tool makers in North
Africa, the Sahara and the Nile Valley has brought out two main points.
First, throughout the whole of this area except the Mediterranean littoral,
the distribution and nature of human settlement has been largely depen-
dent upon changing environmental conditions. Major manifestations of
this dependence are that settlement apparently ceased in much of the
Sahara during the arid period following the ¬‚orescence of the Aterian, and
the dense concentration of occupation in parts of the Nile Valley. Secondly,
and in part causally related, at a remarkably early date a widespread series
of innovative practices evolved to maximise the food-yielding capacity of the
natural environment. These developments were intimately connected with
the presence of closely circumscribed communities which, as is re¬‚ected
in their material culture, showed considerable ingenuity in adapting their
life-style to varied situations. Subsequently, as will be shown in chapter 5,
similar developments are found over a far wider area.

Changing life-styles and technology
At ¬rst sight, this chapter has presented a bewildering mass of data of very
varying quality, ranging from the results of modern, scienti¬c excavations
to details pieced together from observations of undated surface occurrences.
The data relate to three strands of enquiry which have often been followed
independently: hominid fossils, human genetics and material archaeology.
It is now necessary to stand back from the details in an attempt to gain an
overall picture of human development in Africa during the period between
the ¬nal Acheulean/Sangoan industries around 350,000--250,000 years ago
and the beginnings of permanent settlement.
Throughout the period discussed in this chapter archaeological sites in
most parts of Africa are both larger and more numerous than those of
earlier times. This cannot be explained solely by the fact that they have been
preserved with less disturbance or obliteration than older sites. Virtually all
142 afric an archaeolog y

parts of the continent were now at least sporadically occupied, some of them
for the ¬rst time. There is evidence that a wider range of environments was
subject to human exploitation, and that there was a progressive increase in
the size of hominid populations both at a local and at a continental level
(Lee 1963). This may indicate that people had become more adaptable and
consciously responsive to environmental and other pressures. One result of
this adaptability was the steady proliferation of local industries and the
accompanying faster rate of cultural change that may be recognised in the
archaeological record. Demographically, it was accompanied by increases
both in the size of individual groups and in overall population levels. A
corresponding acceleration in the development of non-material culture is
indicated, for example, by the burial customs and artistic traditions, as well
as by the personal adornment that has been preserved. By the end of the
period many of the foundations for the diverse richness of later African
cultures had already been laid.
The African hominid fossils of this period have been evaluated by, among
others, Klein (1999) and Rightmire (2001). The picture that emerges is one of
generally uniform development from Homo heidelbergensis to an early form of
H. sapiens. The designation H. helmei is sometimes applied to an intermediate
species. The model proposed above in chapter 2 demonstrates that it would
be inappropriate to seek a de¬nite point in time or place where this transi-
tion took place; it is suf¬cient to note that the presence of archaic H. sapiens
is indicated in most regions of Africa by the end of the Middle Pleistocene.
Subsequent further developments led to the emergence of ˜anatomically
modern people™ in eastern and southern Africa, and probably also in North
Africa and other parts of the continent, between 200,000 and 100,000 years
Genetic studies have not yet been undertaken on African human fos-
sils and are based almost exclusively on recent samples. Such studies have
focussed largely on attempts to illustrate the origins of modern populations,
with a particular focus on Europe. As noted above (pp. 95--6), this has led
to a tendency to attribute human genetic modernity to a rapid and recent
transformation (e.g. Klein 2000a) which is supported neither by other lines
of enquiry nor by theoretical considerations. Since genetic mutations occur
primarily in individuals rather than across populations, it has proved dif¬-
cult to produce a diachronic interpretation of modern data which does not
involve improbably small populations or bottlenecks. Despite uncertainty
about the details and the chronology, there is broad acceptance of the view
that anatomically modern people ¬rst evolved in Africa and spread from
there to other parts of the world.
Turning to the archaeological record, it is clear that, despite local


. 4
( 11)