. 6
( 11)


be recognised, as may the practice of arti¬cial deformation of the horns
(Schwabe 1984; cf. also J. Brown 1990), which has continued in the Sahara,
as in parts of East Africa, into recent times. Milking and use of cattle for
riding are also depicted.
To the southwest the beginning of herding seems to have taken place
rather later, as shown in the Tilemsi Valley which enters the Niger from the
north near Gao. In the upper part of the valley, cattle herders lived at Asselar
in about 3300 bc, but they probably did not penetrate the previously unin-
habited ¬‚ood plains to the south until early in the third millennium, when
the Karkarichinkat sites were occupied. At these sites cattle are represented
both by abundant bones and by clay ¬gurines. Fishing, hunting and fowl-
ing are also attested, but there was no evidence for the cultivation of any
vegetable foods. This movement down the Tilemsi Valley may be regarded
as the ¬rst stage of the southward spread of cattle-herding into West Africa
(A. B. Smith 1974, 1980b; Van Neer 2002; see also p. 203).
The results of research at Dhar Tichitt, located near the southern edge
of the desert in south-central Mauritania, have often been cited as sug-
gesting that cereal cultivation in the western Sahara was a relatively late
180 afric an archaeolog y

Fig. 83: Dhar
Tichitt stone
enclosures: (top)
from the air, and
(below) as
preserved on the

development. This view must now be modi¬ed, and the adoption of culti-
vation seen in its socio-political context (Holl 1989). The site lies within the
natural distribution area of wild varieties of both sorghum and bulrush
millet. Remains of extensive villages, comprising numerous stone-walled
compounds (Fig. 83), occur along a limestone cliff over a distance of some
40 kilometres. Numerous radiocarbon dates indicate that the occupation of
Early farmers 181

the complex extended from the mid-third to the early ¬rst millennia bc,
a period which saw progressive desiccation of the local environment and
shrinking of the lakes at the foot of the cliff. Bones of cattle and goats
are preserved, while the pottery retains impressions both of wild grains
and of cultivated bulrush millet (Pennisetum). Initial research by Munson
(1976) appeared to indicate that the farming of millet at Dhar Tichitt began
shortly before the middle of the second millennium bc and did not achieve
economic importance until several centuries later. Reinterpretation by Holl
(1985a, 1985b) suggests, however, that the differences in food sources
do not re¬‚ect a chronological sequence so much as varying, seasonally
determined economic practices. If this was indeed the case, the Dhar
Tichitt sequence would ¬t more conformably within the general picture
now emerging in several parts of the Sahara (Amblard 1996; Wetterstrom
Detailed conclusions should not be drawn from the very incomplete cover-
age of the research that has so far been undertaken in the Sahara. A general
picture is, however, beginning to emerge. Rich concentrations of resources
in often hostile environments stimulated, as was described in chapter 5,
the development of a sedentary life-style at several places in the Egyptian
Western Desert, leading in due course to the domestication of local cattle
and the use of sorghum and millet. By the beginning of the sixth mil-
lennium bc these innovations had spread to several widely dispersed parts
of the Sahara. It was probably shortly afterwards, around 5800 bc, that
domestic ovicaprids were introduced from the Levant into Africa, where
their remains have been discovered in Cyrenaica, in the Red Sea Hills and,
some centuries later, in the Nile Delta. Further discussion is best postponed
until data relating to the Nile Valley have been presented.

The Nile Valley
Despite the evidence for early experiments, described in chapters 4 and 5,
the permanent adoption of farming techniques in the Nile Valley is known
only from a relatively late date. In the Khartoum area, for example, there is
no indication that any form of farming was practised before the late ¬fth
millennium bc (Hassan 1986a; Haaland 1992). Recent research has shown
that the settlement pattern at this time was far more complex than had
originally been anticipated. The ¬rst relevant site to be excavated was that
of Esh Shaheinab, 50 kilometres north of Khartoum (Arkell 1953), where
the material culture is clearly a development of that represented at Early
Khartoum (pp. 153--4 above). The site had a river-bank location and ¬shing
was evidently of major importance. Barbed bone harpoon heads were now
pierced at the base for attachment of the line, and shell ¬sh-hooks were
182 afric an archaeolog y

Fig. 84: Artefacts
from Esh
Shaheinab (after
Arkell 1953):
1, barbed bone
harpoon head
with perforated
base; 2, bone
adze; 3, backed
4, shell ¬sh-hook

also in use. Axes and adzes, ¬nished by grinding, were fashioned from both
bone and stone (Fig. 84). Both the microlithic stone industry and the pot-
tery resemble those from Early Khartoum and some Saharan sites, but the
pottery was now burnished and also included new types resembling contem-
poraneous Pre-Dynastic Egyptian wares. Beads were made from amazonite
which appears to have been brought from Tibesti, over 1700 kilometres
distant. The animal bones recovered were originally believed to represent
wild species, with a few small goats, but it was subsequently recognised
(Peters 1986) that domestic animals, including cattle, were present in larger
It now appears that farming in the central Sudanese Nile Valley dur-
ing the closing centuries of the ¬fth millennium bc was both intensive
and complex (Haaland 1987). Much of the evidence comes from Kadero,
a site which covered an area of 4 hectares on the edge of the ¬‚ood plain
northeast of Khartoum (Krzyzaniak 1978, 1984). Here, at the same time as the
occupation of Esh Shaheinab, cattle were herded in very large numbers
along with sheep and goats; about 90 per cent of the animal bones recov-
ered from Kadero were of domestic species. There were enormous num-
bers of heavily used grindstones. Grain-impressions on pottery showed
that sorghum, ¬nger millet and panicum were the principal cereals used.
Early farmers 183

Although not morphologically distinguishable from wild grains, they do
appear to have been collected and processed on a substantial scale and may
have been cultivated. Hunting and ¬shing were both of marginal impor-
tance. The dead were buried within a circumscribed area. The Kadero site,
by its size and evidence for prolonged use, may be regarded as a base set-
tlement. Other sites in the locality were occupied by smaller groups on a
seasonal basis and used for different economic activities, including ¬shing.
The overall picture which is emerging from current research is of a commu-
nity which occupied several different sites on a seasonal cycle, using vari-
ous food resources according to their availability, possessing large numbers
of domestic animals and perhaps also cultivating cereals at some of their
locations. Further evidence in support of this view comes from El Geili,
some 30 kilometres north of Kadero, across the river from Esh Shaheinab
(Caneva 1988). This complex situation provides an excellent demonstration
of the dangers in drawing detailed conclusions from the investigation of
single sites which may yield very incomplete pictures of ancient life-styles
(Mohammed-Ali 1982).
Away from the Nile, there are indications that a similar situation prevailed
along the now-dry Wadi Howar (Richter 1989), which extends eastward from
the highlands of eastern Chad to join the Nile near Dongola. As research
continues in this area, it may be expected more clearly to demonstrate con-
nexions between the Sahara and the Nile Valley. During the better-watered
conditions that prevailed for much of the mid-Holocene, these connexions
would have been more readily made than has been possible more recently, as
indicated also by settlement in more northerly areas that are now virtually
inaccessible desert (Schuck 1989).
Downstream, pottery which seems to be related to that from Esh Sha-
heinab occurs at a number of sites in the Dongola region and extending
into Nubia, where it is sometimes associated with the stone industry known
as Abkan. In Lower Nubia the earliest fully farming community known to
archaeologists is the so-called A-Group, which probably arose around the
middle of the fourth millennium bc (Fig. 85). The Abkan af¬nities of the
stone artefacts found on A-Group sites, together with some features of their
pottery, suggest that the A-Group was of indigenous Nubian origin; it was
subsequently much in¬‚uenced by contact with Pre-Dynastic peoples of Egypt
(pp. 189--91; O™Connor 1993a). Trade between the two areas was extensive,
with Egyptian ¬‚int and a wide range of manufactured goods including stone
vessels, copper tools, palettes, amulets and the like ¬nding their way south-
wards. It was presumably raw materials that went to Egypt in exchange:
commodities such as ivory, and ebony from further to the south. The
A-Group people are known mainly from their graves, which occur in large
184 afric an archaeolog y

Fig. 85: A-Group
artefacts from
the Wadi Halfa
area (after
Nordstr¨m 1972):
1--3, types of
pottery vessels
(not to scale);
4, 5, human

cemeteries and show burial customs similar to those which prevailed in
Egypt (H. S. Smith 1991). Sheep and goats were herded, with smaller num-
bers of cattle. Wheat and barley were cultivated; linen cloth was in use,
but may have been imported from Egypt rather than made from locally
grown ¬‚ax. Fishing and hunting were both practised. Although many peo-
ple lived in insubstantial shelters, others -- perhaps an ©lite -- were housed
on a grander scale, as at Afyeh near the First Cataract, where an A-Group set-
tlement consisted of rectangular houses with up to six rooms, dating from
Early farmers 185

about 3000 bc (W. Y. Adams 1977). The reasons for the end of the character-
istic A-Group settlement at about the time of the end of the First Dynasty
in Egypt (approximately 3100--2900 bc) are poorly understood, but may have
been connected with growing imbalance of authority as centralised Egypt
grew more belligerent and substituted raiding for trade. A Fourth-Dynasty
incursion (2600--2500 bc; see pp. 189--90 below) is recorded as having resulted
in the capture of 7000 people and 200,000 domestic animals (Breasted
For several centuries Egyptian contact with Nubia was at a reduced level.
From Sixth Dynasty times (2300 bc) onwards Nubian archaeology is again
better known, the indigenous population being known as the C-Group
(Trigger 1976; J. H. Taylor 1991). Particularly in the south, there is evidence
for cultural continuity between the latter and people living east of the Nile,
in the Red Sea Hills. Although domestic cattle, represented in grave-goods
and depicted on pottery, clearly occupied a more important place in the
lives of the C-Group than in those of their predecessors, faunal remains
suggest that their herds also included numerous small stock. Settlements
were initially small and consisted mainly of large circular houses with the
bases of their walls built of stone. More complex structures are attested
in later times. Graves, as at the Aniba cemetery, were sometimes marked
by stelae. Luxury goods of Egyptian origin were obtained on a moderate
scale, but local crafts, notably potting, were highly developed. There is evi-
dence that Egyptian contacts now extended further upstream than pre-
viously, probably at least as far as the Dongola Reach, beyond the Third
In Middle Kingdom times (2000--1600 bc) the Egyptians established a mil-
itary occupation of northern Nubia and erected a series of massive forts,
as at Semna south of the Second Cataract, to secure Egyptian control
of trade and access to the area™s gold deposits. This period saw the rise
of a rich culture at Kerma, located in the most fertile part of Sudanese
Nubia at the northern end of the Nile™s Dongola Reach. The formative
stages at Kerma are not fully understood; probably an essentially indigenous
society gradually became subject to increasing Egyptian in¬‚uence, although
its speci¬cally Nubian features remained dominant (cf. O™Connor 1993b). For
reasons that are not known to us, but which may be connected with polit-
ical troubles in their homeland, the Egyptians retreated from Nubia under
the Thirteenth Dynasty, late in the eighteenth century bc, and this provided
Kerma with a further boost. Herding of cattle and ovicaprids contributed
greatly to Kerma™s prosperity, as indicated both in the settlement sites and
in the grave-goods placed in the extensive surrounding cemeteries. Kerma
was both a political capital and a religious centre; among its most impres-
sive monuments are huge brick structures known locally as defuffas, and
186 afric an archaeolog y

Fig. 86: The brick
substructure of
one of the royal
burial mounds at

the graves where its powerful rulers were buried, accompanied by the bod-
ies of numerous retainers, under large mounds up to 80 metres in diameter
(Fig. 86). Great wealth was evidently accumulated through Kerma™s control of
wide-ranging trade, and a remarkable level of craftsmanship was attained,
particularly in pottery. Egyptian stylistic in¬‚uences remained strong but
many local features are apparent (Reisner 1923; Dunham 1982; Bonnet et al.
1990; Bonnet 1991, 1992).
Early in the Eighteenth Dynasty (around 1500 bc) Nubia was re-occupied
by the Egyptians, the old forts were repaired, and Kush was conquered.
This time the Egyptianisation of Nubia was cultural as well as political, as
is witnessed by the temple of Ramesses II at Abu Simbel, erected in the
thirteenth century bc (Fig. 87). Both the C-Group culture and its Kerma
manifestation withered away. From about 1000 bc, Kush ceased to be ruled
by Egyptian-appointed viceroys and became an independent kingdom. In
the eighth century bc the king of Kush conquered Egypt and established
the Twenty¬fth Dynasty.
It is now necessary to turn northwards, to the Egyptian Nile Valley, and to
return to the period which saw the beginning of farming there. As will be
apparent, the picture contrasts markedly with that obtained in the Sudan. It
was probably owing to the abundance and reliability of wild food resources
there, and despite much early experimentation, that the beginning of
Early farmers 187

Fig. 87: The
temples at Abu
Simbel, before
clearing and
restoration (after
Gau 1822)

farming on a signi¬cant scale in the Egyptian Nile Valley seems to have
been abrupt and late in comparison with its inception in the Western Desert
(pp. 173--5; cf. Hassan 1986b, 1988b; Wetterstrom 1993). In the Delta and
in the Fayum Depression small villages of farming people broadly resem-
bling those from adjacent parts of southwestern Asia were established from
about 5200 bc (Fig. 88). That at Merimde on the western side of the Nile
Delta, dated around 4800 bc, may be cited as an example (Baumgartel 1955;
Hawass et al. 1988). It covered an area of some 18 hectares and consisted of
small oval dwellings measuring only about 2 by 3 metres, built of lumps
of mud mixed with straw, sunk slightly into the ground and presumably
originally roofed with reed thatch. These structures were set on either side
of narrow lanes and interspersed with mud-lined storage pits, basket gra-
naries and open shelters which appear to have been used as workshops. The
dead were buried within the settlement in mat-lined graves. Bifacially ¬‚aked
stone tools and undecorated pottery serve to link this settlement with its
counterparts in the Fayum Depression (Caton-Thompson and Gardner 1934;
Brewer 1989). The economic basis for these Lower Egyptian settlements was
the cultivation of barley, emmer-wheat and ¬‚ax. Cattle, sheep, goats and pigs
were kept, as were dogs. The earliest Nile Valley attestation of the domestic
donkey, an indigenous African species, dates from about this time. Hunt-
ing and ¬shing continued to be practised. Although the Fayum settlements
188 afric an archaeolog y

Fig. 88:
artefacts (after
Arkell 1975): 1--3,
ground stone
axe, arrowhead
and pot from
Fayum; 4, Nakada
I pot; 5, Badarian

may have been of short duration, that at Merimde and allied sites such as
those near Helwan were clearly occupied on a permanent basis, with farm-
ing on a substantial scale supporting communities which probably num-
bered well over a thousand persons (Hoffman 1980). In such circumstances
specialist craftsmen were able to establish themselves, and non-utilitarian
products such as stone vessels and objects for personal adornment
Subsequently there arose the Egyptian cultures conventionally known
to archaeologists as ˜Pre-Dynastic™, because they ¬‚ourished prior to the
sequence of numbered dynasties that form the chronological framework for
Early farmers 189

the literate civilisation of ancient Egypt. For many years the Pre-Dynastic cul-
tures were known mainly from their graves, and their ordering and chronol-
ogy were based upon detailed typological studies of their pottery. A succes-
sion of closely related industries was thus proposed, consisting of Badarian,
Nakada I (Amratian), Nakada II (Gerzean) and Nakada III. The earlier phases
occur in the 200-kilometre stretch of the Nile Valley south of Asyut; their
descendants spread rapidly both to the north and to the south. It is only in
recent years that signi¬cant settlement sites of the Pre-Dynastic period have
been investigated. Of prime importance is the town at Hierakonpolis, 100
kilometres north of Aswan: estimates of its population vary greatly but it is
likely to have held well over 5000 people. Other settlements include those
of Badari and Hamamiya. Both round and rectangular houses are attested.
The farming economy followed the pattern of the earlier sites, with the cul-
tivation of barley and emmer-wheat and the herding of domestic cattle and
ovicaprids, supplemented with hunting, ¬shing and the gathering of wild
plant foods (Hoffman 1980; Hassan 1988a).
Technological advances at this time included the development of meth-
ods of superb pressure-¬‚aking to impart a regular rippled ¬nish to stone
tools. Fine stone vessels were now carved from basalt, alabaster and even
porphyry. Occasional small copper objects appear early in the Pre-Dynastic
period, mainly pins and beads; they seem to have been produced by ham-
mering native (unsmelted) metal. In later Pre-Dynastic times, from Nakada II
(around 3600 bc) onwards, techniques of copper-smelting were introduced,
presumably from western Asia where they had been known for many cen-
turies, and ¬‚at axes, daggers and knives were cast. Fine basketry was made,
linen was woven and the simple black-topped pottery was now supplemented
by more elaborate wares with painted decoration (Nicholson and Shaw
The general trend through much of the fourth millennium bc was one
of steady development of centres such as Hierakonpolis, Nakada and This,
with increasing evidence for craft specialisation and social strati¬cation. It
appears that these centres, and many others, each became the nucleus of a
small state, with its own king and patron deity (Spencer 1993). By about
3100 bc, or slightly before, a uni¬ed kingdom was established over the
whole of the Egyptian Nile Valley north of Aswan, ruled by the pharaohs
of the ¬rst of the thirty dynasties (Fig. 89) which provide the conventional
framework for the history of ancient Egypt (James 1979). The uni¬cation of
Egypt was not achieved without con¬‚ict, and the First Dynasty pharaohs are
often depicted as conquerors or plunderers. Concurrently with these polit-
ical developments there was a marked ¬‚orescence in crafts and industries,
which must have been connected with the start of extensive trade in raw
190 afric an archaeolog y

Period Dynasty Date bc

Early Dynastic I c. 3100--2890
II c. 2890--2686
Old Kingdom III c. 2686--2613
IV c. 2613--2494
V c. 2494--2345
VI c. 2345--2181
First Intermediate VII--XI c. 2181--1991
Middle Kingdom XII 1991--1786
XIII 1786--1674
Second Intermediate XIV--XVII 1674--1567
New Kingdom XVIII 1567--1320
XIX 1320--1200
XX 1200--1085
Late Dynastic XXI 1085--945
XXII--XXIII 945--730
XXIV--XXV 730--656
Fig. 89: The
chronology of 664--332
ancient Egypt

materials. We now ¬nd for the ¬rst time evidence for contact with the older
literate civilisations of Mesopotamia, and certain Egyptian innovations in art
and technology -- including methods of building with bricks -- may owe much
to the latter area. It was even formerly suggested that the Egyptian state
system itself may have been of Mesopotamian inspiration, but it is important
to emphasise its unique local character. The same is true of the Egyptian
hieroglyphic script which developed under the First Dynasty (Fig. 90).
A huge amount of research has been devoted to the study of ancient Egypt,
often with little if any reference to its African connexions. This has become
such a specialised ¬eld that it is often designated a separate discipline: Egyp-
tology. A detailed description of ancient Egyptian civilisation lies beyond the
scope of this book, but some of its essential features may be summarised
(James 1979; J. D. Clark 1982b; Kemp 1989; Nicholson and Shaw 2000), not
only to redress the separatist trends of some studies but also to facilitate
an evaluation of the in¬‚uences which ancient Egypt and its African neigh-
bours exerted on one another (cf. O™Connor and Reid 2003). The ¬rst point
which requires emphasis is the civilisation™s remarkable continuity through
three thousand years. The second is its great material wealth, based both
upon the annual Nile ¬‚ood laying down fertile silts which supported the
agriculture needed to feed the population concentrated in the narrow val-
ley (Bowman and Rogan 1999), and upon the acquisition of raw materials
through large-scale external trade.
Early farmers 191

Fig. 90: Ancient Egyptian hieroglyphic writing. Egyptian hieroglyphic writing was developed
about 3100 bc and continued in use with remarkably little change until the end of the fourth
century ad. Its meaning seems to have been forgotten soon afterwards and it was not
re-deciphered until the early nineteenth century.
It incorporates some seven hundred signs. Most of these are ideograms -- simpli¬ed pictorial
representations of the concepts to which they relate. Some of these ideograms also had a
phonetic value representing one or more consonants. (Vowels were not indicated in ancient
Egyptian writing, so it is often not possible to ascertain the original pronunciation.) Often
ideograms and phonetic symbols were combined. For example depet -- meaning ˜boat™ -- could
(hand = d) (stool = p) (loaf = t) (boat). Reading is further complicated
be written
by the fact that words were not divided and that inscriptions could be written from left to
right, right to left, or vertically.
Royal names may be recognised by their inclusion in an oval shape or cartouche, as in the
examples above which give, from left to right, the names of the Eighteenth Dynasty pharaohs
Tuthmosis III, Akhenaten and Tutankhamun.

The head and epitome of the Egyptian state was the divine ruler, the
pharaoh. The whole complex bureaucracy of the state was ultimately respon-
sible to him and, particularly in the earlier periods, senior of¬cials were
often members of the royal family. The pharaoh was also the ¬gurehead of
the of¬cial religion, the personi¬cation of the sun-god Ra, counterpart of
Osiris the god of the land of the dead. Material preparation for life after
death was of immense importance to the ancient Egyptians, as is shown
by the complex efforts made to protect deceased bodies by mummi¬cation
and to immure them with many belongings in elaborate tombs. As a result,
archaeological research for many years tended to concentrate on the tombs
of the dead rather than on the settlements of the living. The royal tombs
in particular re¬‚ect the great wealth and concentration of resources, both
human and material, at the pharaohs™ disposal, whether they were buried
in the mighty pyramids of the Old Kingdom or in the hidden underground
chambers of the New Kingdom (Fig. 91).
In evaluating the structural achievements and technological skill of the
ancient Egyptians it is necessary to remember the limitations under which
they worked. The wheel was unknown before the New Kingdom, yet the
pyramids, for example, were built of stone blocks weighing over 2.5 tonnes,
192 afric an archaeolog y

Fig. 91: Egyptian
royal tombs (after
James 1979): top,
pyramid of
Khufu, Fourth
Dynasty; below,
rock-cut tomb of
Seti I, Eighteenth

presumably moved and erected with the aid of rollers and levers. Copper,
bronze and gold were effectively the only metals used, for iron did not come
into regular use before the Twentysixth Dynasty in the seventh century bc.
Much of our information about ancient Egyptian history comes from the
records that were carefully maintained by the Egyptians themselves, notably
by the priests who were regarded as the guardians of the state™s accumulated
wisdom. The value of this source of information was recognised in ancient
times, as by the Greek historian Herodotus in the ¬fth century bc. Idealised
scenes of everyday life, at least for the upper classes of society, were often
depicted on the walls of tombs. Here we see representations of the ships
which carried Egypt™s trade along the Nile and further a¬eld. We see the
huge bands of labourers -- enslaved foreign captives and peasants providing
work as a tax-payment -- on whom the state™s public works depended. It is
from this source also that we obtain our sole information about another
little-understood aspect of Old Kingdom culture: the capture and taming of
animals such as gazelle, oryx and even perhaps giraffe (J. D. Clark 1971; see
also Fig. 92). Ancient Egypt was responsible for major advances in knowledge
in such ¬elds as literature, mathematics, medicine and law (J. R. Harris
1971), discussion of which falls outside the scope of this book.
The political history of ancient Egypt may be summarised brie¬‚y. Since
much of our information comes from contemporary written sources,
Early farmers 193

Fig. 92: Relief
carving at
Kalabsha, Nubia
(after Gau 1822),
showing the
taming of wild
including gazelle
and other
antelope, ostrich,
monkeys and a
big cat, perhaps a

knowledge of this topic is of a different order of detail from that which
is available for other parts of Africa at this time; the chronology is also
known with far greater precision.
After the Early Dynastic period, when the Nile Valley was already largely
isolated by Saharan desiccation, and during which the uni¬cation of the
Egyptian state was consolidated, the accession of the Third Dynasty in about
2700 bc marks the start of the ¬rst great period of prosperity, the Old King-
dom. Through patronage and control of trade, power and wealth were effec-
tively concentrated in the hands of the ruling dynasty. This is re¬‚ected ¬rst
and foremost in the scale at which resources and manpower were devoted
to state works, notably to the construction of pyramids for the burial of
deceased pharaohs.
By later Old Kingdom times the pharaoh™s control over the state bureau-
cracy seems to have weakened, and the proportion of Egypt™s resources
that was devoted to royal works was consequently diminished; for example,
the Fifth-Dynasty pyramids were smaller than those of the Fourth Dynasty.
Shortly after 2000 bc, following a period of contraction from the peak of Old-
Kingdom prosperity and wide-ranging trade, Egyptian political unity broke
down during the First Intermediate period of some two centuries. Famine
may have added to the general impoverishment of this time. Reuni¬cation
under the Eleventh Dynasty, based at a new capital near Thebes, heralded
the Middle Kingdom. Egypt™s authority in Nubia was further strengthened
at this period, as shown by the erection of the forts which have been noted
The new-found stability was short-lived, however, and during the Thir-
teenth and Fourteenth Dynasties there was a rapid succession of pharaohs
as different factions competed for supremacy. Early in the resultant Sec-
ond Intermediate period a group of invaders from Palestine -- the so-
called Hyksos rulers -- took advantage of Egypt™s weakness and established
themselves in Lower Egypt as the Fifteenth Dynasty in about 1670 bc. The rise
194 afric an archaeolog y

Fig. 93: Crowned
head from a
green schist
statue of the
Dynasty pharaoh,
Tuthmosis III;
height: 45 cm

of independent Kerma, described above, may have been facilitated by Egyp-
tian weakness at this time. Increased frequency of trade-goods of Palestinian
origin, particularly in the Delta, indicates greater contact with western Asia
during the period of Hyksos rule.
Eventually, a dynasty (the Seventeenth) from Thebes in Upper Egypt
expelled the Hyksos rulers and re-established Egyptian unity and indepen-
dence. The New Kingdom which followed marked the greatest ¬‚orescence of
ancient Egyptian power and prosperity. Egyptian control was re-established
over Nubia as well as over substantial areas of the Levant, all governed
by a complex imperial bureaucracy set up by the pharaoh Tuthmosis III
(Fig. 93). Egyptian trade ranged far and wide, even to the Land of Punt far to
Early farmers 195

the southeast (p. 205). During the Eighteenth Dynasty occurred the remark-
able reign of the pharaoh Akhenaten, who attempted to impose monotheism
in place of the traditional religion. Akhenaten™s successor was the young
Tutankhamun, the only pharaoh whose tomb has been discovered virtually
undisturbed and unrobbed, to reveal the full richness and splendour which
surrounded the New-Kingdom rulers.
From the Twenty¬rst Dynasty onwards, Egypt™s cohesion once again broke
down, and from the eleventh to the seventh centuries bc Libyan, Asian
and Nubian contenders vied with Egyptians for control of the state. The
Twenty¬fth Dynasty originated in Kush and ¬nally, as will be described in
chapter 7, lost control of Egypt to an invasion from Assyria, after which
ancient Egypt was controlled by a succession of foreign rulers, with only
brief interludes of independence.
At this stage in the narrative, it is appropriate to offer a brief summary
of the evidence for early cultivation and herding in the Nile Valley. The ¬rst
point that requires emphasis is the late date at which these innovations
took place in comparison with the situation in the Sahara. The earliest
farming communities so far recognised in the Egyptian Nile Valley and its
immediate environs are those in the Fayum Depression and in the Delta.
Their cultivated wheat and barley and their domestic ovicaprids were all of
species farmed in earlier times in western Asia and it is presumably there
that their ultimate origin should be sought. Even then, their ¬rst attestation
in the Nile Valley, around 5200--4800 bc, is several centuries later than the
presence of ovicaprids both in Cyrenaica and in the Red Sea Hills. Somewhat
later, during the second half of the ¬fth millennium, cattle-herding was
also adopted, this time from the west, in most parts of the Nile Valley as
far south as the central Sudan. In the latter area it was accompanied by
the use (if not the formal cultivation) of locally available cereals such as
¬nger millet and sorghum. Subsequently, with progressive desiccation, many
parts of the Sahara became unsuited for farming. In the ensuing dispersal
of population, areas were settled where local cereals and other plants grew
more successfully than wheat and barley; and thus millet and sorghum were
eventually brought under more formal cultivation. Overall, there appears to
have been a gradual expansion of farming from the north and east to the
south and west. In the process domestic animals were introduced into areas
where their species were previously unknown, and indigenous African plants
were cultivated.

West and Central Africa
It has been shown botanically that many of the food-crops traditionally culti-
vated in the western sudanic region and the adjacent forest fringes belong to
species that are indigenous to the sub-Saharan latitudes. Among the cereals
196 afric an archaeolog y

Fig. 94: Sorghum

the most important are bulrush millet (Pennisetum), fonio and various types
of sorghum (Fig. 94). The homelands of these species extend in a broad belt
from the Nile Valley to Senegal, as indicated above in ¬gure 78 (p. 168).
Distinct techniques are used for the cultivation of African rice in the val-
leys of the Niger and Benue (R. J. McIntosh 1998). The propagation of yams
(Fig. 95), which is presumed to have originated near the northern fringes
of the West African forests, involves distinct methods yet again. It would
be unwise to assume a common source for these varied types of indige-
nous African agriculture. Unfortunately, conclusive primary archaeological
evidence for early cereal cultivation is scarce, for rice exceedingly rare and
for yams non-existent. In the last case, physical traces of the crop itself are
Early farmers 197

Fig. 95:
yam-storage in
eastern Nigeria

highly unlikely to survive in the archaeological record, and the artefacts
traditionally used are almost all perishable. Pending the re¬nement of tech-
niques for studying phytoliths, it is therefore necessary to rely upon less
certain indicators and upon inference, as will be attempted below. It will
be clear to the reader that, a few major projects notwithstanding, recent
archaeological research has been sparse in this region; parts of the account
that follows thus depend on widely dispersed and incomplete information.
In West Africa as a whole, predominantly mode-5 stone industries, as
described in chapter 4, continued in use until the beginning of iron-working.
However, an important change is attested from about the late ¬fth mil-
lennium bc onwards, when two previously unknown cultural items made
their appearance. These were pottery and ground-stone axe-like or hoe-
like implements (Shaw 1977, 1981). It seems likely that these innovations
were somewhat earlier in what is now Nigeria and Cameroon than they
were further to the west. There is no evidence for the practice of any type of
cultivation or herding before these items appeared and, in later times, they
were certainly used in connexion with such activities. It would be wholly
misleading, however, to assert that the presence of pottery and/or ground-
stone artefacts in an archaeological assemblage necessarily indicates that
its makers were farmers. This very tentative archaeological reconstruction
receives some measure of support from palaeoenvironmental investigations
(Sowunmi 2002) which indicate widespread reduction in forests in more
westerly regions after 3000 bc. This could be due to lower rainfall and/or
198 afric an archaeolog y

to clearance for agriculture; increased occurrence of pollen from oil palm,
which frequently colonises newly cleared ground, strengthens the latter
possibility. Archaeological evidence from Ghana provides some degree of
support for these hypotheses.
At Bosumpra Cave near Abeti¬ in Ghana, a microlithic industry basically
similar to that from earlier, pre-pottery, sites is associated with simple pot-
tery and ground-stone hoe-like or axe-like implements (Shaw 1944; A. B.
Smith 1975). This occupation had begun by the end of the ¬fth millen-
nium bc and lasted intermittently for over 3500 years, with pottery and
ground-stone artefacts becoming progressively commoner with the passage
of time. If they were not farmers (and there is no ¬rm archaeological evi-
dence that they were), the inhabitants of Bosumpra were presumably able in
some way to follow a reasonably settled life-style permitting the use of such
a fragile, heavy type of equipment as pottery. Comparable conditions also
prevailed by the mid-¬fth millennium on the Ghana coast, at settlements
where the economy was based on the exploitation of marine food resources
(Calvocoressi and David 1979).
At Iwo Eleru rockshelter in southwestern Nigeria (the lower levels of which
were described in chapter 4), pottery and ground-stone implements likewise
¬rst appear around the middle of the ¬fth millennium. At about the same
level there is reported the earliest occurrence of implements bearing so-
called sickle-sheen (see p. 149), but this does not necessarily mean that the
inhabitants harvested -- or cultivated -- cereals (Shaw and Daniels 1984). A
rockshelter at A¬kpo in southeastern Nigeria probably provides a counter-
part sequence (Andah and Anozie 1980).
In more westerly regions of West Africa, signi¬cant differences are appar-
ent. In the Ivory Coast, pottery and ground-stone artefacts are found in
association with microlithic industries both on inland sites and in coastal
shell mounds, but cannot yet be shown to be earlier than the late third
millennium bc (Mauny 1973). In Sierra Leone, however, similar associations
at Kamabai and Yagala rockshelters extend back to around the end of the
fourth millennium (Atherton 1972). Throughout this region discoveries of
ground-stone artefacts need to be evaluated with care because these objects
were extensively traded in ancient times from factory sites such as those
near Cape Three Points in Ghana, on Bioko Island and in the Oueyanko Val-
ley near Bamako in Mali. They have also been sought after and preserved
until recent times in the belief that they possess magical properties, being
preserved, for example, on altars at Benin (Connah [1964]). Reliable infor-
mation concerning prehistoric industries in Guinea and Guinea-Bissau is
almost totally lacking. In Senegal, on the other hand, extensive shell mid-
dens (Fig. 96) attest coastal settlement, with pottery, beside the Casamance
estuary and near Saint Louis; in the latter area the occupation dates
Early farmers 199

Fig. 96: Shell
middens on the
coast of Senegal

back to the fourth or late ¬fth millennium (Linares de Sapir 1971; Ravis©
Firm indications of cultivation or herding in West Africa are surprisingly
late in date. In Ghana, for example, the earliest primary evidence comes
from Kintampo-industry contexts, in about the eighteenth century bc; these
sites appear to be restricted to the forest margin and the southern part
of the woodland savanna to the west of the Volta. The Kintampo industry
(Stahl 1985, 1994; Casey 2000; d™Andrea and Casey 2002) presents a sharp
discontinuity with its predecessor and may indicate in¬‚uences from the
north and west, apparent both in the pottery decoration and in the typology
of the stone arrowheads. At one site bones of ovicaprids and, less certainly,
small domestic cattle have been reported: if con¬rmed, the cattle could be of
a type ancestral to the modern dwarf shorthorn breeds of West Africa. Wild
animals were also represented. Oil-palm nuts and cowpeas were preserved in
the Kintampo deposits. A highly characteristic but enigmatic artefact which
is frequently encountered on Kintampo sites is a soft stone slab with deeply
scored surfaces (Fig. 97). The purpose of such objects remains completely
unknown. Suggestions include use in pottery manufacture, or for grating
yams, or for removing hard skin from the feet. The Kintampo people lived
in villages with rectangular wattle and daub structures (Dombrowski 1980);
at Ntereso, overlooking the White Volta 50 kilometres west of Tamale, the
settlement covered an area of at least 750 square metres. The northern
200 afric an archaeolog y

Fig. 97: Artefacts
from Kintampo
sites (after Davies
1967): 1, stone
rasp; 2, decorated

af¬nities of the Kintampo industry seem to be particularly strong with the
Tilemsi Valley area north of the Niger bend in Mali (p. 179).
Archaeological evidence from these more northerly regions does not yet
allow any comprehensive assessment of early farming practices. An excep-
tion is provided by current research in Burkina Faso and adjacent regions
(Breunig et al. 1996; Gronenborn 1998; Frank et al. 2001; Breunig and Neu-
mann 2002), which has yielded evidence for encampments dating between
2200 and 1000 bc. There is no evidence for herding, but cultivation may
have begun around the mid-second millennium; the earliest remains of an
actual crop are of Pennisetum, about 1000 bc. In the Chad Basin, the early
and mid-Holocene was marked by a very extensive lake, ˜Mega-Chad™, with
which is associated a remarkable dug-out canoe 8.4 metres long found at
Dufuna in northern Nigeria and dated to about 7000 bc (Breunig 1996),
together with somewhat later shoreline settlements with pottery but no
evidence for cultivation or herding. This lake began to shrink after 5000
bc; eventually its dry ¬‚oor was settled by cattle herders of the Gajiganna
Complex, probably of more northerly origin. Present from about 1800 bc,
these herders made use of numerous wild grasses and began to cultivate
small quantities of Pennisetum from about 1200 bc. Later settlement of the
basin is best known from the Bornu plains in the extreme northeast of
Nigeria (Connah 1976, 1981), where a composite sequence at Daima (Fig. 98)
and neighbouring sites extends back to the late-second millennium bc; the
settlements comprised durable wooden-walled, clay-¬‚oored houses (see also
Holl 2002). In the absence of local stone, many tools were made of bone. The
pottery shows little change during the sequence. Domestic cattle and goats
were present throughout, although hunting and ¬shing were also impor-
tant activities. Similar mound-sites are known from north and east of the
lake in both Chad and Cameroon.
Early farmers 201

Fig. 98:
through the deep
accumulation of
deposits at

Further to the east, at Shum Laka near Bamenda in Cameroon, stamp-
decorated pottery seems to have been in use as early as about 5000 bc,
associated with chipped hoe-like stone artefacts (de Maret et al. 1987;
Lavachery 2001). Ground-stone implements are known from many areas of
Cameroon, from Gabon and as far south as the lower reaches of the Congo.
Their age in the north remains uncertain, but it has been suggested that
202 afric an archaeolog y

they may be as old as 3000 bc in the Estuaire and Ogooue provinces of Gabon,
where village sites have been discovered with pottery akin to contemporary
wares from Cameroon, grindstones and nuts of oil palm (Elaeis guineensis).
Particular interest attaches to the recovery of phytoliths at Nkang near
Yaounde in southern Cameroon which indicate the growing of bananas dur-
ing the last millennium bc (Mbida et al. 2000). Since the banana is thought
to have been introduced to Africa from a South Asian source (De Langhe et al.
1994--5), this date is remarkably early and raises the likelihood that bananas
may have been grown in more easterly parts of Africa in even earlier times
(cf. p. 260 below). Taken together, these observations suggest that, during
the last three millennia bc, the inhabitants of these coastal regions may
have adopted elements of settled life based on the intensive exploitation of
forest vegetable foods (Clist 1986, 1989a).
In the coastal areas north and south of the Congo River mouth it is now
known that pottery, whose ¬‚at bases serve to link it with more northerly
wares, together with ground-stone axe/hoes, was in use during the last
centuries bc, preceding by several hundred years the beginning of iron-
working in this region (Denbow 1990; de Maret 1986). Nuts of oil palm and
Canarium have been found on several of these Ngovo-group sites, suggest-
ing the practice of arboriculture and, possibly, other forms of non-cereal
horticulture. Linguistic studies suggest that some of the inhabitants of the
Cameroon/Congo region during the last millennia bc may have spoken Bantu
Current research on the megalithic funerary monuments near Bouar, in
the westernmost part of the Central African Republic, attributes to them a
date in the last millennium bc (David 1982; Zangato 1999). Well over one
hundred of these sites are now known, consisting of a rubble mound incor-
porating walls and cists of large, undressed stone slabs weighing up to 2
tonnes. Such monuments are widely dispersed across the sahel from Senegal
to the upper Nile; in the latter region they were still in use during the twen-
tieth century ad. No recent overview of their form, use or antiquity has been
Despite the incomplete and widely scattered nature of the evidence, the
general picture that emerges from recent research is that the forest-edge
peoples in the eastern half of West Africa began to make pottery and ground-
stone artefacts around the mid-¬fth millennium bc, possibly somewhat ear-
lier in Cameroon. Although both these technologies were known in the
Sahara in earlier times, there is -- in contrast with the later situation in
more northerly and westerly regions -- no reason to suppose that they were
other than independent innovations. It seems probable that yam cultiva-
tion near the forest/woodland-savanna ecotone may have begun at approxi-
mately this time. Once such a plant is collected, it proliferates and provides a
Early farmers 203

rich, readily available source of food (Hillman 1989; Chikwendu and Okezie
By contrast, in the more northerly savannas of West Africa, cereal cultiva-
tion and the herding of domestic animals were not begun until at least two
thousand years after these traits were attested in the Sahara. Their appar-
ent southward dispersal into West Africa seems to have occurred at about
the time of the major period of Saharan desiccation, when climatic and
vegetational zones would have shifted to the south. The effect which these
changes may have had on the distribution of tsetse ¬‚ies probably allowed
the eventual entry of domestic cattle into West Africa.
In view of the foregoing, it would clearly be wrong to suggest that all farm-
ing in West Africa began as a direct result of contact with more northerly
areas. It appears certain, however, that domestic animals were so derived,
and probably cereal agriculture also. Yam cultivation, however, although
nowhere conclusively attested in the archaeological record, may well have
been an indigenous development and perhaps one which pre-dated any
other form of farming in this region (cf. Coursey 1976). The antiquity of
West African rice cultivation also remains totally unknown. There are good
botanical and environmental reasons for regarding the Inland Niger Delta
as one of the centres for early domestication of this crop and, indeed, it is
there, at Jenne-Jeno, that its earliest archaeological attestation occurs: in a
context dated to the ¬rst century ad (R. J. and S. K. McIntosh 1981).

Ethiopia and the Horn
For a long time it has been recognised that the highland areas of Ethiopia
must have played an important part in the development of African farming,
particularly agriculture (Harlan 1969). Not only has there been considerable
diversity in recent Ethiopian agricultural practices, with wheat and barley
being grown, for example, alongside indigenous African crops; but there
are also several food-crops which are traditionally cultivated in Ethiopia
and nowhere else. The latter must presumably have been originally domes-
ticated there: they include the tiny but highly nutritious grain teff, the
banana-like plant enset, and the oil-yielding noog (Simoons 1965). Botanical
studies suggest that cultivated ¬nger millet may also have originated in the
lower regions of southwestern Ethiopia. Unfortunately, very little archaeo-
logical evidence has yet been recovered to illustrate the early development of
Ethiopian farming, so that many aspects, including its chronology, remain
poorly understood (Brandt 1984; D. W. Phillipson 1993a; Agazi 1997a; Barnett
Only two widely separated areas of Ethiopia have yielded archaeologi-
cal sequences which span the period when farming began. At Gobedra
204 afric an archaeolog y

rockshelter near Aksum, in the Tigray highlands of northern Ethiopia, the
earlier occupation of which was noted in chapter 4 (p. 125), pottery ¬rst
appeared in association with backed microliths at a level which may date
between the mid-¬fth and the third millennia bc. It is possible that the
camel was present at this early period, although seeds of cultivated ¬nger
millet that were excavated from the same layer are now known to be sub-
sequent intrusions. The presence of domestic cattle is indicated at a later
stage of the Gobedra sequence (D. W. Phillipson 1977b, 1993a).
Research on later sites in the neighbourhood of Aksum provides evidence,
described in chapter 7, that a varied mixed-farming economy was well estab-
lished by at least the middle of the last millennium bc (Anfray 1990; Bard
et al. 2000; D. W. Phillipson 2000). Although the local antecedents of these
communities remain unknown, there seems a strong probability that the
origins of their cultivation and herding practices may extend back to the
earlier times with which this chapter is concerned (Fattovich 1994, 1996).
The second Ethiopian sequence which is relevant to the present discussion
comes from the area around Lake Besaka near the escarpment west of Harar.
Here, the local backed-microlith industry was augmented early in the second
millennium bc by the production of large numbers of steeply retouched
tools that appear to have been used as scrapers. Domestic cattle made their
appearance at the same time. Excavations also yielded a fragment of a stone
bowl akin to those found on sites of the early East African herders (J. D.
Clark and Williams 1978).
Away from these two areas the evidence for early Ethiopian food-
production is indirect or circumstantial. The only other signi¬cant archae-
ological discoveries are of late date, such as those from Lalibela Cave, east
of Lake Tana, which included remains of barley, chickpeas, cattle and small
stock, dated to the middle of the last millennium bc (Dombrowski 1970).
It would be reasonable to propose that there were two main agricultural
traditions in Ethiopia during the last 3000 or 4000 years bc, that in the north
being based on cereal cultivation and that in the southwest on enset. In the
latter region large numbers of ground-stone hoe-like tools were found many
years ago (Bailloud 1959) in association with pottery and, only occasion-
ally, with metal objects, but their age and signi¬cance remain completely
unknown pending further research. The northern, cereal, zone is more likely
to have had contact with the Nile Valley, but so far the only early sites which
provide plausible evidence for such connexions are at Agordat in Eritrea and
around Kassala in the Sudan. Publication of the Agordat sites (Arkell 1954)
includes no absolute dates or excavation results, but the surface ¬nds of
ground-stone tools and ornaments, including an ox ¬gurine, may possibly
show some features in common with those of the Nubian C-Group in the late
Early farmers 205

Fig. 99: Rock
paintings of
domestic cattle
at Genda Biftu,
Ethiopia (after
J. D. Clark 1954)

third millennium bc (p. 185). Fattovich™s (1994; see also Sadr 1991) research in
the Gash Delta near Kassala close to the Sudan/Eritrea frontier has revealed
an archaeological sequence from the sixth millennium bc onwards. Cattle
were herded before 3000 bc, the associated pottery resembling that from the
Nile Valley; a few centuries later both cultivation and herding were ¬rmly
established, upright stone stelae were used to mark graves, and links both
with Agordat and with later sites in Eritrea and northern Ethiopia may be
Rock paintings (Agazi 1997b) provide another possible source of infor-
mation concerning domestic animals, notably cattle which are frequently
depicted (Fig. 99). Unfortunately the paintings cannot be dated directly.
Humpless longhorned cattle are shown in paintings in Eritrea and around
Harar, as well as in Somalia (Brandt and Carder 1987). Ancient Egyptian
records extending back to the Old Kingdom include references to the Land
of Punt, now generally believed to have been in what is now northern Eritrea
and adjacent regions of Sudan (Kitchen 1993; Phillips 1997). Eighteenth-
Dynasty (mid-second-millennium bc) carvings at Deir el Bahari depict the
presence in Punt by that time of domestic small stock, two breeds of cattle,
and cultivated cereals (Naville 1898; see Fig. 100).
Linguistic research offers some supplement to the scant ¬ndings of archae-
ology. Vocabulary relating to the cereal--plough agriculture complex in
northern Ethiopia appears to be of Cushitic origin and to pre-date the
arrival of Semitic-speakers during the earlier part of the last millennium
bc. Taking the story further back, intensive cereal use and the subsequent
early stages of cultivation are, both in northeastern Africa and in adja-
cent parts of the Levant, ¬rst encountered in areas occupied by speakers
of Afroasiatic languages (see chapter 1, pp. 6--8). This language family prob-
ably has a time-depth in the order of 10,000--15,000 years, which effectively
covers the formative phases of incipient farming as revealed by archaeology
206 afric an archaeolog y

Fig. 100: Punt
and its
inhabitants, as
represented in
ancient Egypt
(after Phillips
1997): 1, houses;
2, a ship and its
cargo; 3, Puntites
delivering their

(Ehret 1980, 2003). The most likely location of the homeland of the Afroasi-
atic languages, on the southwestern side of the Red Sea in Ethiopia and
easternmost Sudan, suggests that we still have much to learn about the
beginnings of Ethiopian farming.

East Africa
The earliest evidence for herding in East Africa comes from the plains of
northern Kenya. Domestic cattle and sheep/goat are represented at three
sites in the Ileret area on the northeast shore of Lake Turkana, dated to the
middle of the third millennium bc (Owen et al. 1982; Marshall et al. 1984;
Early farmers 207

Fig. 101: Stone
bowl from North
Horr (after D. W.
Phillipson 1977a)

Barthelme 1985). Associated artefacts are stone bowls and pottery, includ-
ing vessels with jabbed decoration and internal scoring which resemble
the Nderit ware (see Fig. 103 on p. 210) of southern Kenya. Similar pot-
tery has also been reported from sites west of Lake Turkana, which appear
long to pre-date the local beginning of herding, although the age of this
material cannot be regarded as certain. Fishing remained an important
source of food for the early Ileret herders. These people exploited a more
extensive territory than their ¬shing and hunting predecessors, and they
brought obsidian from distant sources to use in making their microlithic
Inland, east of Lake Turkana, an extensive and long-occupied settlement
was established during the third millennium on the shores of the then
shallow but extensive Chalbi Lake at North Horr (D. W. Phillipson 1977a,
1979), where the pottery, stone bowls and microliths may be compared with
those from Ileret (Fig. 101). Information about the economic developments
which took place at this time in more arid areas has been obtained from a
rockshelter at Ele Bor, near the modern border between Kenya and Ethiopia
(D. W. Phillipson 1984; Gifford-Gonzalez 2003). Here, until very recent times,
meat was obtained mainly by hunting, but ovicaprids and, interestingly,
camels were present in small numbers from about 3000 bc, which is also
the time of the ¬rst appearance of pottery. This evidence for camels helps to
substantiate their apparently early date at Gobedra in Ethiopia, noted above.
Particularly important is the evidence at Ele Bor from seeds and numerous
grindstones for the intensive exploitation of cereals -- presumably wild -- at
this same period, which faunal evidence shows to have had a climate some-
what wetter than that of today. Subsequently, as the climate deteriorated,
cereal use was abandoned but ovicaprids continued to be herded in small
The available archaeological evidence from northern Kenya shows that
in some areas a settled life-style based at least in part on the herding of
208 afric an archaeolog y

domestic stock continued until early in the present millennium. Sites of
this period such as, for example, a second settlement at North Horr, contain
abundant pottery. Even at this late date iron was evidently unknown or
exceedingly rare and microlithic artefacts continued in use. Further climatic
deterioration, leading to today™s arid conditions, caused the people to adopt
their present nomadic pastoral way of life within the last few centuries,
involving abandonment of the use of pottery in favour of lighter and more
easily transportable receptacles (D. W. Phillipson 1984).
Further to the south, the highlands of southern Kenya and northern
Tanzania were the scene of one of the best-known farming complexes
which pre-date the beginning of iron-working (Gifford-Gonzalez 1998). It
¬rst appears in the archaeological record late in the second millennium
bc. Claims (e.g. Bower and Nelson 1978) that domestic animals were herded
in this region in far earlier times are now generally discounted (Owen et al.
1982; Robertshaw and Collett 1983; Bower 1991; Marshall 2000). These indus-
tries have been named collectively ˜Pastoral Neolithic™ in recognition of their
status as herders, but it remains an open question whether they were true
pastoralists in the sense generally accepted by anthropologists or whether
they practised any form of cultivation (Bower et al. 1977; Bower 1991). The
term is therefore not used in the discussion which follows.
The archaeological material relating to these early East African herders
seems to divide most of them into two groups. One industry, known as
Elmenteitan, has a very restricted distribution in the high-rainfall area on
the west side of the Kenyan Rift Valley; its ¬‚aked-stone artefacts are char-
acterised by large double-edged obsidian blades and its pottery by plain,
mostly bowl-shaped vessels. Shallow stone bowls, akin to those noted above
from earlier times in northern Kenya, are also present on Elmenteitan sites.
An unusual feature was the practice of cremating the dead, as at Njoro
River Cave where each burial was accompanied by a stone bowl, pestle and
mortar. Charred remains of a gourd and of an elaborately carved wooden
vessel were also recovered from this site, as were large numbers of stone
beads (M. D. and L. S. B. Leakey 1950). Dated to about the twelfth century bc,
Njoro River Cave may be one of the earliest Elmenteitan sites. The industry
appears to have continued well into the ¬rst millennium ad (Robertshaw
The second major grouping is less well de¬ned, with both stone indus-
tries and pottery types showing considerable variation (Fig. 102). This may
be at least partly due to inadequate consideration of the signi¬cance of such
variability, particularly in the case of the pottery. Several so-called ˜wares™
have been recognised and named after such sites as Nderit, Narosura, Akira
and Maringishu (Bower et al. 1977; Wandibba 1980). The signi¬cance of this
stylistic diversity is far from clear, for none of the ˜wares™ has well-de¬ned
Early farmers 209

Fig. 102:
Artefacts from
Hyrax Hill, Kenya
(after M. D.
Leakey 1945): 1,
pottery of
ware™; 2, 5, stone
bowls; 3, stone
pestle; 4, pottery
of Nderit ware;
6, backed

geographical or chronological parameters and frequently more than one
˜ware™ seems to have been in use in the same place at the same time.
The scheme was developed with exclusive reference to pottery from the
Central Rift Valley, and it has not proved easy to extend it to neighbouring
areas (cf. Chami 2003). Retention of this scheme has hindered improved
understanding of the East African archaeological record of this period.
Nderit ware (formerly known as Gumban A), characterised by decoration pro-
duced by jabbing large areas of a pot™s surface with a wedge-shaped object
and often by scoring of the vessel™s interior surface (Fig. 103), is the only type
which has clear af¬nities with material from other areas. As noted above,
closely similar pottery occurs in association with evidence for herding in
contexts of the third millennium bc in northern Kenya, while west of Lake
Turkana it may extend back into yet earlier times, when domestic animals
were apparently still unknown.
210 afric an archaeolog y

Fig. 103: Nderit
ware pottery
bowl from
Stable™s Drift,
Kenya; diameter
21 cm

Since no clear subdivisions (other than Elmenteitan) are apparent, the
material relating to the stone-tool-using herders of southern Kenya and
northern Tanzania during the last millennium bc must -- if only provi-
sionally -- be discussed together. There are, in fact, many common fea-
tures. The dead were buried, without cremation, under stone cairns or
in crevices between rocks. Stone bowls occur in settlements as well as on
burial sites. There is considerable variation in size of site and in the propor-
tion of the faunal remains represented by domestic animals. It is tempting
to suggest that there was some seasonally changing settlement-pattern in
which, at certain times of the year, the whole community was dependent
upon the produce of the herds, while at others smaller groups obtained
their livelihood by more varied means. This is the pattern with several
recent pastoralist societies, and there are indications, as noted above for the
Khartoum area, that it is one of considerable antiquity (Robertshaw 1989).
The most comprehensive picture yet available for East Africa comes from
research by Robertshaw (1990b) in the Lemek area of southwestern Kenya,
where there is evidence for intensive occupation and where faunal remains
support a comprehensive reconstruction of exploitation and management
Early farmers 211

It was formerly believed that, within East Africa, only the Rift Valley high-
lands were settled by stone-tool-using herders. This is now known not to
have been the case, traces of comparable and broadly contemporary occu-
pation having been recognised in several surrounding regions including
the Lake Victoria Basin and the coastal hinterland from southern Kenya
to central Tanzania. Around the western, southern and eastern sides of
Lake Victoria, a distinctive type of pottery called Kansyore ware has been
reported (Chapman 1967; Soper and Golden 1969). There are indications
that it may have originated in the third millennium bc or even earlier
(Bower 1991). Its predominantly stamped decoration comprises a variety
of motifs, some of which may be derived from earlier pottery styles fur-
ther to the north. Particularly in view of the fragmentary nature of most
of the ¬nds, such conclusions must be regarded as highly tentative, as
must claims for sherds of Kansyore ware from far to the east and south-
east of Lake Victoria. So far, Kansyore ware has not been found with evi-
dence for the working of iron, and the only indication that its makers
may have practised any form of farming is its apparent association with
remains of domestic cattle at Gogo Falls in southwestern Kenya (Robertshaw
1991; Karega-Munene 2002).
It will be instructive now brie¬‚y to compare the archaeological evidence
for the beginnings of farming in East Africa with that which has been
deduced from linguistic investigations (cf. Ambrose 1984a). Study of mod-
ern linguistic distributions and loanwords indicates that much of highland
southern Kenya and northern Tanzania now settled by Nilotic- and Bantu-
speakers was formerly occupied by people who spoke languages that may
be classi¬ed as Southern Cushitic (Ehret 1974). The vocabulary that has
been reconstructed indicates that these Southern Cushitic-speakers were
herders of domestic stock who milked their cattle and who seem also to
have possessed some knowledge of agriculture. Two points require emphasis
here. First, although purely linguistic considerations can provide only a very
approximate estimate of the time-depth at which these Southern Cushitic
languages were spoken, the indications that we possess are that their antiq-
uity is broadly the same as that of the stone-tool-using herders. Secondly, the
area where the former presence of the Southern Cushitic-speakers is attested
is approximately the same as that covered by the main central distribution
of sites which have yielded evidence for domestic animals at this time. It
therefore seems reasonable to accept as a working hypothesis the view that
most inhabitants of these highland sites may have spoken Southern Cushitic
languages. The linguistic af¬nity of the outlying communities, including the
makers of Kansyore ware, remains a matter for speculation pending further
research; it is by no means unlikely that some were Nilotic-speakers.
212 afric an archaeolog y

These conclusions have in the past been used to support the view that
early East African domestic livestock was derived from Ethiopia. However, it
is now known that northern Kenya was itself an important dispersal area for
Cushitic speech, being, for example, the region whence the Somali languages
were derived (Heine 1978). There is also evidently a very long history of
Nilotic/Cushitic contact in the area around Lake Turkana. All this would be
fully in keeping with the archaeological indications for the relatively high
antiquity of farming in northern Kenya. It suggests that a predominantly
Ethiopian origin for East African livestock should not be accepted as proved,
and that Sudanese connexions may eventually be shown to have been at
least equally important. As noted above, such derivation need not imply
large-scale human migration; on the contrary, strong elements of continuity
within East Africa have long been recognised (D. W. Phillipson 1979; Karega-
Munene 1996).
The evidence of physical anthropology, although far from conclusive, is
in keeping with these conclusions. Human skeletons from sites of the early
East African herders have been known for some decades; early descriptions,
which emphasised their caucasoid features, were misinterpreted as suggest-
ing European af¬nities. More recent investigations have demonstrated that
negroid features are in many cases dominant, and that such non-negroid
characteristics as are present would be in keeping with Northeast African
caucasoid af¬nities (Rightmire 1975).
Although the archaeology of central and southern Tanzania remains
poorly investigated, there are no indications from sites of the ¬rst mil-
lennium bc that domestic animals were ever acquired by the stone-tool-
using peoples of inland regions to the south of the Serengeti Plain. In more
southerly latitudes, indeed, it seems that farming was not practised before
the beginning of iron-working. For reasons that are imperfectly understood,
the relatively rapid diffusion of farming techniques through the stone-tool-
using populations of the northern half of Africa seems here to have come
to a temporary halt. Although there is little archaeological evidence from
the forested regions of the Congo Basin, it appears that the only parts of
sub-equatorial Africa where farming was practised prior to the ¬rst millen-
nium ad were in southern Kenya and northern Tanzania and, less certainly
(p. 202), in a restricted area around the lower reaches of the Congo River.
Possible factors which may have hindered the adoption of farming further
to the south at this time include rich and reliable food supplies from wild
sources, coupled with very low human population densities. Also relevant
are the traditionalism and well-adapted hunter-gatherer life-styles of the
KhoiSan-speaking peoples who were probably the exclusive inhabitants of
the more southerly latitudes. Later history, as will be described below, shows
Early farmers 213

how reluctant these well-adapted and conservative peoples have often been
to adopt alien life-styles and economic practices. Whatever the reasons, the
beginning of farming in the southern half of Africa had to await the large-
scale population movements which there accompanied the beginning of
Iron-using peoples before AD 1000
The greater part of Africa differs from most other regions of the Old World
in that there was (except in Egypt and some other areas of northern and
northeastern Africa) no distinct ˜Bronze Age™ or ˜Copper Age™ during which
softer metals, often including gold, were utilised but when techniques of
smelting iron had not yet been mastered. In the sub-Saharan latitudes, iron
was the ¬rst metal to be brought into use; the working of copper and gold
began at the same time or somewhat later (van der Merwe 1980; D. Miller
2003). In most of Africa south of the equator, the beginnings of farming
and of iron-working took place at approximately the same time. In the lat-
ter area there was thus a pronounced contrast between the metal-using
farming people and their stone-tool-using hunter-gatherer neighbours and
immediate predecessors.
The beginnings of metallurgy cannot be said to have made such a great
impact on prehistoric life-styles as did the advent of cultivation and herd-
ing. The civilisation of ancient Egypt (pp. 189--95 above) provides a vivid
example of the technological achievements that could be attained with vir-
tually no use of iron. Basically, the advantages of iron are ones of increased
ef¬ciency. The clearance of forest, the working of wood, the quarrying and
carving of stone, the cultivation of ground and the slaughter of enemies
may all be accomplished more effectively and with less effort by people
who are equipped with iron tools and weapons. These advantages may serve
to explain why the knowledge of iron-working techniques spread so rapidly
through Africa -- as this chapter will demonstrate -- in comparison with the
slow and often hesitant processes by which cultivation and herding were
adopted. They also help to account for the great prestige which, in many
African societies, is traditionally associated with the knowledge and owner-
ship of iron (Herbert 1993). The word which the early Bantu-speaking people
used to signify iron seems originally to have meant ˜a thing of value™.
The technology required to smelt iron -- to produce workable metal
from the naturally occurring ore -- is complex and highly labour-intensive
(Wertime 1980; D. Miller 2002). The ore, having been extracted from the
ground and broken up, must be heated to a temperature of at least 1100
degrees centigrade under carefully controlled conditions. To achieve such
temperatures, aided only by the natural draught of a clay-built furnace and,
Iron-using peoples before ad 1000 215

Fig. 104:
a re-enactment in

usually, by hand-operated bellows, is a major task in itself (Fig. 104). Once
smelted, the usable metal has to be separated from the waste-products -- the
slag -- and brought to its desired shape by repeated heating and hammer-
ing. This last process is known as forging. African iron-working technology
(Haaland and Shinnie 1985; D. Miller and van der Merwe 1994; J. Brown 1995;
Schmidt 1997a) did not include the melting and casting of iron although
copper and its alloys, with their much lower melting temperatures, were
successfully cast.
As later sections of this chapter will make clear, the signi¬cance of metal-
working for African communities has extended far beyond the technological
sphere. Its cultural and social implications (Herbert 1984, 1993; Schmidt
1996; Vogel 2000) have been at least equally signi¬cant and must be given
full weight in any consideration of early African metallurgy.
Knowledge of how to smelt iron on any signi¬cant scale seems ¬rst to have
been discovered in western Asia early in the second millennium bc (Jean
2001). Iron-working was probably brought to North Africa, west of Egypt, by
Phoenician colonists in about the eighth century bc (van der Merwe 1980).
At about the same time iron objects ¬rst came into common use in Egypt
(Nicholson and Shaw 2000). The theory that iron technology spread, with
great rapidity, from these two sources through the rest of the continent has
met with much opposition, not all of it based on archaeological evidence.
Because the associated technology is so complex, and in earlier African
societies no other process involved heating materials to such high
216 afric an archaeolog y

temperatures, we have to consider the possibility of a northerly source for
sub-Saharan iron-working knowledge rather than duplicate independent dis-
covery. The radiocarbon dates that are now available for early iron-working
in Africa lend only modest support to the view that there could have been an
independent development of metal-working technology south of the Sahara
(cf. Woodhouse 1998). There is also the problem, rarely considered, of de¬n-
ing exactly what is meant by ˜independent™. The question usually posed
ignores the need to consider whether, to be truly independent, such a devel-
opment must be shown to have been made by people who were totally igno-
rant of related events elsewhere (cf. Gallay 2001). It is perfectly possible that
future research will yield evidence which demonstrates that knowledge of
iron technology could have been introduced to sub-Saharan Africa from the
north. It is particularly dif¬cult to establish a precise chronology for the
period which saw the beginning of iron-working because, for most parts of
Africa, it is necessary to rely almost exclusively on radiocarbon dates: calibra-
tion of such dates (see pp. 5--6) is exceptionally uncertain at just the period --
the last millennium bc -- with which we are here concerned. It follows that
the radiocarbon ages cited in this chapter -- although an attempt has been
made to calibrate them -- should be regarded as approximations. With this
proviso, we may now examine the archaeological and other evidence for
these processes, and follow the history of the early iron-using societies in
Africa up to about the year ad 1000.1

North Africa
During the last millennium bc, the indigenous inhabitants of North Africa
west of Egypt were essentially farmers, probably for the most part transhu-
mant herders. Cultivation was concentrated in the river valleys of Libya
(Barker et al. 1996), Tunisia and the Maghreb. There can be little doubt that,
at least in the latter area, these farmers may already be recognised as the
ancestors of the modern Berbers (Camps 1982; Brett and Fentress 1996).
Such archaeological work as has been done on Berber sites of this period
has been concentrated on the stone-built funerary monuments which are by
far their most visible component (Camps 1961, 1986), and we know far more
about the colonies that were established on the North African coast by a
succession of trading peoples from the north and east. A certain amount of
information about the early Berbers may, however, be gleaned from the writ-
ings of Greek authors, notably Herodotus, whose work dates from the ¬fth

1 This date has been chosen somewhat arbitrarily and is applied loosely in certain regions where
a more signi¬cant break appears to have occurred rather earlier or later. For example, the
Islamic settlements of the East African coast, although traceable back to the eighth century,
are all considered in chapter 8.
Iron-using peoples before ad 1000 217

Fig. 105: Rock
painting of a
chariot, Acacus
(after Mori 1978)

century bc. Another Saharan people were the Garamantes in Libya (Mattingly
et al. 1998). The Berbers and Garamantes penetrated much of the Sahara;
it has generally been assumed that they used mules as beasts of burden,
although Liverani (2000) argues that camels were already in use. There has
been much speculation about the use of the horse-drawn chariots which
are frequently depicted in Saharan rock paintings (Fig. 105) and engravings
and which are also noted by Herodotus (IV.183), who states that they were
used by the Garamantes for hunting ˜Troglodyte Ethiopians™. It is invariably
¬‚imsy two-wheeled vehicles that are shown: they must have carried people
rather than heavy goods and it seems highly probable that metal was used
in the construction of their wheels. No actual remains of such vehicles have
ever been found, so our information about them comes exclusively from the
rock art. Attempts to trace trans-Saharan ˜chariot routes™, based on the dis-
tribution of these representations (Mauny 1978), are now widely discounted.
It seems likely that the paintings and engravings, which are generally dated
to around the ¬rst half of the last millennium bc, were done in many areas
that were not actually penetrated by such horse-drawn vehicles. Regular
traverse of many desert areas, other than on foot, probably did not take
place before the arrival of the camel, perhaps during the ¬rst half of the
last millennium bc. There is good archaeological evidence for the presence
of camels in Egyptian Nubia by about the ninth century bc (Rowley-Conwy
1988) and, as shown in chapter 6, possible indications in Ethiopia and East
Africa at a signi¬cantly earlier date.
Through most of the last millennium bc, metal was a very scarce commod-
ity among the indigenous peoples of North Africa. Knowledge of the working
of copper is likely to have reached a limited area of northwestern Africa from
218 afric an archaeolog y

Hippo Utica
Lixus Hadrumetum
Ptolemais Apollonia
Lepcis Cyrene

Fig. 106: Greek
and Phoenician 0 1000km Greek colonies
colonies in North Phoenician colonies
0 500miles

Spain during the second millennium bc, as is suggested by rock engravings
in the Atlas Mountains which show metal tools similar to Iberian examples
(Malhomme 1959--61; Camps 1974). Elsewhere, although copper and bronze
ornaments were occasionally obtained through trade, metal objects gener-
ally were very rare; North African warriors of this period were famous in the
ancient Mediterranean world for their use of ¬re-hardened wooden spears
without metal points (Herodotus VII.71).
Such were the indigenous peoples amongst whom successive colonists
settled. The ¬rst to arrive were the Phoenicians who, from their homeland
in the coastlands of the eastern Mediterranean, penetrated by sea as far
to the west as Mogador in southern Morocco (Markoe 2000). Phoenician
expansion into the western Mediterranean may have begun as early as the
end of the second millennium bc, but was placed on a ¬rm footing by
the foundation of Carthage near the modern Tunis, traditionally dated to
the ninth century bc. The map (Fig. 106) shows the location of the principal
Phoenician settlements on the African coast. The type of square-rigged ship
used by the Phoenicians, which could not sail close into the wind, makes it
highly unlikely that any of their vessels would ever have been able to return
from a voyage which penetrated further down the coast of northwestern
Africa than southernmost Morocco, and this is in fact the limit of well-
documented archaeological evidence for a Phoenician presence. The claim
mentioned by Herodotus (IV.42), who himself seems to have disbelieved it,
for a Phoenician circumnavigation of Africa from the Red Sea to the Atlantic
should not be dismissed out of hand (cf. Cary and Warmington 1929); it is
incapable of proof or disproof, although objections about the dif¬culty of a
northward voyage along the northwestern coast still apply.
The Phoenicians were great merchants, and their colonies provided a stim-
ulus and end-point for the Berbers™ trade with the central Saharan highlands
and, perhaps, West Africa. Salt, ivory, animal skins and slaves were the major
Iron-using peoples before ad 1000 219

items that found their way northwards; manufactured goods, including pot-
tery, glass and metalwork, probably went south in exchange. The Phoenician
colonies prospered on the pro¬ts of this trade, aided by cereal cultivation
on the Tunisian plains, most of which was controlled by Carthage from the
¬fth century bc onwards (Fantar 1993). As a result, the settled Berbers of the
coastal regions were drawn into the Phoenician cultural and technological
sphere; Berber kingdoms came into existence, nominally independent but
often Phoenician clients (Brett and Fentress 1996).
There can be little doubt that both copper and iron were introduced to
much of North Africa by the Phoenicians, but conclusive archaeological


. 6
( 11)