. 7
( 11)


evidence for this is so far lacking. The only information we have about iron-
smelting technology in Punic North Africa dates from as late as the third
century bc (Niemeyer 2001), long after the period which is of prime interest
if we are to evaluate whether such expertise could have spread from here
to more southerly parts of Africa.
While the Phoenicians controlled the maritime trade of the western
Mediterranean, including the African coast west of Tripolitania, Greek
colonies were established on the coast of Cyrenaica from the late seventh
century bc onwards (Boardman 1999). The region takes its name from that
of the principal Greek colony, Cyrene. The southernmost point of the Gulf
of Sirte was eventually accepted as the boundary between the Greek and
Phoenician spheres of in¬‚uence. The prosperity of Greek Cyrenaica was
based on agriculture, the surplus of which was exported.
By the third century bc, Rome was challenging Carthaginian supremacy
in the western Mediterranean. Several of the Berber kingdoms were won
over as allies of Rome, the most notable being that ruled by Massinissa
(201--148 bc) in what is now eastern Algeria (Law 1978). Following the defeat
of Carthage by the Romans in 146 bc, these allies were rewarded with tracts
of formerly Carthaginian territory. Rome now became the main external
power and trading partner in the area, but political annexation followed
slowly, and it was some two and a half centuries after the sack of Carthage
before Rome established herself as mistress of all the Maghreb north of the
Atlas and of the coastal strip further to the east.
The heyday of Roman North Africa was in the second century ad, when
major cities were built, public works including roads, aqueducts and irriga-
tion schemes undertaken, and agriculture developed to such an extent that
the area became a major economic force in the Roman empire (Fig. 107).
The colonists often regarded Africa as their permanent home, and their
society fused with that of the Berber ©lite (Decret and Fantar 1981; Daniels
1987; Raven 1993). The economy of the colonies was ¬rmly based in their
agriculture (Mattingly 1989, 1995) and, to a lesser extent, on trans-Saharan
220 afric an archaeolog y

Fig. 107: Ruins of
the Roman city
of Timgad in
eastern Algeria.
Note the regular
street grid
centred on forum
and theatre.

Fig. 108: The
Roman emperor
Severus (ad
193--211) was of
North African

trade (Law 1967). In the third century more than one emperor of Rome was
descended from North African stock (Fig. 108). Christianity gained many
adherents both in the cities and in rural areas during the second and third
centuries. North Africa at this time was a Roman province as effectively
colonised as at any subsequent period of the region™s history: its economy
and industries were essentially those of the empire of which it formed an
integral part. However, by the late third century and throughout the fourth,
Berber uprisings led to a reduction in Roman in¬‚uence. North Africa did
not escape the collapse of Roman power in the west which came early in the
¬fth century. In ad 429 Vandals, raiders from the Baltic area, crossed from
Spain; Carthage fell to them six years later and they took control of what
was left of the urban settlements, while the more distant Roman estates
were once again appropriated by the Berbers (Pringle 1981).
Iron-using peoples before ad 1000 221

Egypt and the Arab invasion
To the east, the coming of iron meant that Egypt also lost her indepen-
dence, though in a signi¬cantly different manner. The pharaohs of the
Twenty¬fth Dynasty were the Nubian kings of Kush; they were replaced
by an invasion of Assyrians in 671 bc. Despite the fact that the country
frequently had tributary status to a succession of foreign powers, the tradi-
tional culture of ancient Egypt survived with few important modi¬cations
for over six centuries (Lloyd 1983; Nicholson and Shaw 2000). The develop-
ing Greek-dominated trade networks of the eastern Mediterranean brought
Egypt, through her entrepˆt at Naukratis in the Nile Delta, into closer con-
tact with Europe. Herodotus, who visited Egypt in the ¬fth century bc, has
left us (in his Book II) a detailed and informative account both of what he
saw and of what he learned about the Egypt of earlier times.
Late in the fourth century bc, the Macedonians from northern Greece
under their king, Alexander the Great, destroyed Persian power in the east-
ern Mediterranean. They conquered Egypt in 332 bc. The succession of
pharaohs, after almost three thousand years, was now brought to an end,
and one of Alexander™s generals was appointed ruler of Egypt as Ptolemy I.
He and his descendants controlled Egypt until 30 bc, severing the formal
link with Macedon and building the country to a pre-eminent commercial
and cultural position in the Greek-speaking world (Bowman 1986) while
emphasising, as did Roman emperors in later times, their status as suc-
cessors to the pharaohs. The Ptolemaic capital at Alexandria (Fraser 1972)
became not only a great centre of learning (Fig. 109) but also a port from
which traders sailed throughout the Mediterranean as well as through the
Red Sea to India and far down the East African coast (Burstein 1989; Salles
1996; see also Manzo 1996).
The Roman conquest of Egypt came in 30 bc. The country™s wealth was
partly drained in tribute and through taxation, but this did little to shake
Alexandria from its position of in¬‚uence throughout the eastern, Greek-
speaking, part of the Roman empire. Extensive archaeological research, com-
plemented by study of the very numerous written documents that have sur-
vived in Egypt™s arid conditions, gives a uniquely comprehensive picture of
everyday life at this time (N. Lewis 1983), illustrating the interplay between
traditional and foreign elements. Roman technology modi¬ed many earlier
practices (e.g. Peacock and Max¬eld 1997). After the Jewish revolt late in the
¬rst century ad, many refugees settled in Egypt. Among these people and
others Christianity rapidly took root (Watterson 1988; Finneran 2002). It was
by way of Egypt that Christianity was passed southwards to Nubia and, less
directly although at an earlier date, to Ethiopia.
Fig. 109: Map of Africa, according to calculations made by Claudius Ptolemy of Alexandria in the second century ad. This version was printed in Rome at the
end of the ¬fteenth century.
Iron-using peoples before ad 1000 223

Fig. 110: The
mosque of Sidi
Okba at
founded ad 670,
rebuilt and
enlarged in the
ninth century

Following the division of the Roman empire in the fourth century, Egypt
was included in the eastern (Byzantine) hegemony, where it remained until
the Arab invasion of the seventh century. Byzantine-ruled Egypt was so weak-
ened by internal dissension that it offered little resistance to the invading
Arabs in 639. Conversions to Islam proceeded steadily, and Arabic replaced
Greek as the language of government. The Arab invasion did not stop in
Egypt, but extended rapidly westwards through North Africa (Hitti 1963;
Brett 1978). By 647 the Arabs had reached Tunisia and defeated the Byzantine
forces, who themselves had ousted the Vandals early in the sixth century.
Later in the seventh century the Arab conquest of the area was consolidated,
and a local headquarters established at Qayrawan (Fig. 110). In the eighth
century the front was pushed further west to Morocco and northwards into
Spain. Islam spread more rapidly in North Africa than in Egypt, but some
224 afric an archaeolog y

Christian communities survived for several hundreds of years. Technolog-
ically, this period saw the eclipse of Roman skills, and the resurgence of
native North African ones. Many aspects of Arab culture were introduced
and ¬‚ourished, advances in architecture, science and literature being par-
ticularly noteworthy. As in the days of the Phoenicians, Tunis again became
an entrepˆt for trade with the south: goods from West Africa and from
Europe were to be found in the markets of the North African coast. This
trade was now in the hands of itinerant Arabs as well as Berbers, and was
institutionalised by the establishment of regular journeys by large camel
caravans. Through these, Islam was introduced to the areas south of the
desert, as will be described below; and these same areas became known to
Arab geographers (Levtzion 1978; Insoll 2003).

The Sudan
Following the expulsion from Egypt of the kings of Kush after the Assyrian
invasion of 671 bc, the Kushites retreated to their former homeland. Their
rule in the Sudan continued for another thousand years from a capital
which was initially located at Napata near the Fourth Cataract, but which
was moved about 600 bc upstream to Meroe, beyond the Nile--Atbara con-
¬‚uence (D. N. Edwards 1996; Shinnie 1996; Welsby 1996; see Fig. 113 below).
Meroe lies to the south of the most arid stretch of the Nile Valley, in an
area that appears to have been well wooded during the last millennium bc.
The transfer of the state™s capital to Meroe (T¨r¨k 1997), for whatever rea-
son it was undertaken, had three important results. It marked an effective
break from dependence upon Egypt, it brought the capital within reach
of the fuel that was needed to maintain an iron-smelting industry which
soon arose, and it provided the town with surroundings that enabled crops
and herds to be raised on a scale suf¬cient to feed its growing population.
Millet and sorghum were cultivated; herds consisted primarily of cattle,
humpless short-horns being depicted on painted pottery. Horses were in
use, but little is yet known about the introduction of the camel, which is
attested at Qasr Ibrim in Egyptian Nubia by about the ninth century bc
(Rowley-Conwy 1988). Meroe could also take advantage of new trade routes
that more than replaced the old Nile Valley route downstream into Egypt.
To the east was the way to the Red Sea, while the valley of the Atbara
led southeastwards to the highlands of Ethiopia where a distinct urban
civilisation was arising during the last millennium bc (see below). To the
south the Nile ¬‚owed through fertile plains where the mixed farming life-
style of earlier times continued, and supported substantial settlements such
as those at Jebel Moya and Jebel et Tomat (Addison 1949; J. D. Clark and
Stemler 1975). Westwards the dry plains stretched away to Darfur and,
Iron-using peoples before ad 1000 225

Fig. 111: The
Meroitic temple
at Naqa, erected
in the reign of
King Natakamani
and Queen
Amanitore (late
¬rst century bc /
early ¬rst
century ad)

beyond, to Lake Chad. Egyptian connexions were still emphasised in the
monumental architecture (Fig. 111), but Meroe was now able to strengthen
its links with peoples to the south of the Sahara and to import luxuries
from the north and east via the Red Sea ports. Gradually Meroitic replaced
Egyptian as the language of monumental inscriptions. By the last two cen-
turies bc, Egyptian hieroglyphs had been replaced by a local cursive script
(Fig. 112). The meaning of these Meroitic inscriptions cannot yet be fully
Archaeologists formerly attached great importance to the evidence for
Meroitic iron-working which, to judge from the size of the slag heaps that
are to be seen on the site, was at some period carried out on a very sub-
stantial scale. Subsequent research (Shinnie and Bradley 1980) has shown
that although some smelting may date back close to the seventh or sixth
centuries bc -- about the time of the establishment of the royal capital --
the industry did not reach a massive scale until the last centuries bc or
the beginning of the Christian era. The furnaces that have been excavated
date from this late period and were cylindrical structures rather over one
metre high, ¬red with the aid of bellows. It appears that much of the iron
produced at Meroe was dispersed through trade, since relatively few iron
objects -- hoes, axes, arrowheads and the like -- have been found on the site
itself. Despite the large scale of Meroitic iron-working and trade, the late
date now demonstrated does not support the view previously held that it
played a major role in the transmission of metallurgical knowledge to more
226 afric an archaeolog y

Fig. 112: Cursive
Meroitic script
on a sandstone
funerary stela
in the shape
of an offering
table, from
width: 34 cm

southerly parts of Africa (cf. Fig. 113). The imposition of Roman rule over
Egypt was re¬‚ected in the Sudan by luxury imports (Kirwan 1977) and by
technological changes, as in methods of quarrying (Harrell 1999).
In the second century ad the prosperity of Meroe began to decline rapidly.
Environmental deterioration, perhaps partly brought about by over-grazing
and deforestation, may have been a contributing cause; another was the rise
to trade-based prosperity of the kingdom of Aksum in northern Ethiopia, dis-
cussed below. It may have been King Ezana of Aksum who ¬nally destroyed
Meroe in about ad 350 (Kirwan 1960).
Downstream of the Third Cataract, the immediately post-Meroitic cen-
turies are marked by sites attributed to the poorly understood Ballana Cul-
ture (˜X-Group™) -- possibly Nubians whom the Romans used as a buffer to
protect their southern Egyptian frontier (Kirwan 1974; Welsby 1996, 2002).
The sites which have been most intensively investigated are royal graves at
Ballana and Qustul in Lower Nubia (Emery and Kirwan 1938) and tumuli at
Iron-using peoples before ad 1000 227

0 1000 km

0 500 miles
Present vegetation


il e

Qasr Ibrim
Debeira West


El Hobagi


Jebel Moya Aksum
Jebel et Tomat

Fig. 113:
Wun Rok
Africa, showing
some important
sites with
evidence for early
with the modern

El Hobagi in the vicinity of Meroe itself (Lenoble and Sharif 1992; Lenoble
et al. 1994); at the latter place there is evidence for continuity with Meroitic
funerary practices. Early in the sixth century Christianity was introduced,
and shortly afterwards Arab expansion cut the Nubians off from their co-
religionists in Egypt. Remarkably, the Nubians were able to maintain their
Christian culture for some seven hundred years; their artistic accomplish-
ments are best illustrated in the frescoes (Fig. 114) recovered from the cathe-
dral at Faras near Wadi Halfa (Michalowski 1967; Finneran 2002). Similar
buildings were erected as far to the south as the modern Khartoum, at Soba
which was the capital of the southern Nubian kingdom of Alwa (Welsby and
Daniels 1991; Welsby 1998). Village life in Christian Nubia, as illustrated by
228 afric an archaeolog y

Fig. 114: Fresco
of an archangel
from Faras
cathedral, tenth
century ad

excavations at Debeira West, followed a pattern of irrigation agriculture
which has continued into recent times (P. L. and M. Shinnie 1978).

Ethiopia and adjacent regions
The northern Ethiopian highlands during the last millennium bc saw a
gradual in¬‚ux of Semitic-speaking peoples across the Red Sea from south-
ern Arabia. The new arrivals appear to have encountered a settled, agri-
cultural population, perhaps Cushitic-speaking, such as those illustrated by
recent excavations at Kidane Mehret and Beta Giyorgis near Aksum in Tigray
(D. W. Phillipson 2000; Bard et al. 2000). Dated between the eighth and fourth
centuries bc, these settlements comprised clusters of stone buildings whose
inhabitants herded cattle, sheep and goats, and cultivated a range of crops
including wheat, barley and teff. They made distinctive pottery and mode-5
lithic artefacts; they were familiar with copper but not, it appears, with
iron. By the eighth or seventh century bc, the South Arabians established
in Tigray a literate urban culture which retained many features -- artis-
tic, architectural and technological -- derived from their homeland (Anfray
1990; Fattovich 1990). Iron-working was probably introduced into Ethiopia
at this time (cf. Mapunda 1997), as were the worship of the South Arabian
Iron-using peoples before ad 1000 229

Fig. 115: A
altar from Addi
showing the
symbol of the
moon-god and
part of a
inscription (after
Sergew 1972)

moon-god, symbolised by the crescent and disc, and also writing, in the
form of the Himyaritic script (Fig. 115). It seems likely that this culture of
South Arabian af¬nity was originally restricted to a relatively small num-
ber of settlements, most notably Yeha near Adwa (Anfray 1963, 1995; Robin
and de Maigret 1998), and that it only slowly in¬‚uenced the lives of the
indigenous Cushitic-speaking population.
By the ¬rst century ad, Aksum, some 50 kilometres southwest of Yeha,
developed as the capital of an extensive state in which there was a fusion
of indigenous Ethiopian and South Arabian cultural elements. The farm-
ing economy of earlier times continued largely unchanged (Michels 1994).
Through its port of Adulis on the Red Sea coast, Aksum was in trade con-
tact with the Roman empire, exporting ivory and skins in exchange for
manufactured luxury goods. The Aksumite state incorporated several urban
centres in addition to Aksum itself, most notably Matara on the plateau of
what is now Eritrea (Anfray 1967). Ge™ez -- basically Semitic but with a strong
Cushitic element -- seems to have been the general language of Aksum, but
Greek was also in use for commercial purposes. Coins were struck at Aksum
(Fig. 116); on the earlier issues the king™s name and titles were given in Greek.
230 afric an archaeolog y

Fig. 116: The
¬rst Aksumite
coinage: a gold
coin of King
Endybis, late
third century ad

In the fourth century ad Aksumite power seems to have reached its peak. It
was at this time that King Ezana is believed to have conquered Meroe. It was
also in Ezana™s reign that Christianity became the state religion of Aksum:
on his later coins the crescent and disc of the moon-god were replaced by
the cross (Munro-Hay and Juel-Jensen 1995: Hahn 1999).
Aksumite technology and material culture were complex and sophisti-
cated, but the civilisation™s indigenous roots are demonstrated by the contin-
uation from earlier times of long-established cultivation and herding prac-
tices and of mode-5 lithic technology (L. Phillipson 2000). The monumental
architecture and tombs of Aksum are relatively well known (Littmann et al.
1913; Munro-Hay 1989; D. W. Phillipson 2000; Fattovich et al. 2000). Great
stelae -- obelisks hewn from a single piece of rock up to 33 metres high and
carved as stylised representations of multi-storey buildings -- were erected
in a burial area on the edge of the town (D. W. Phillipson 1997; Fig. 117).
At Aksum, Matara and other sites, large rectangular buildings, sometimes
interpreted as palaces, reached heights of several storeys. The architecture of
these buildings, like that represented on the stelae, shows several features
which may be paralleled in southern Arabia and which have also continued
into more recent times in Ethiopian ecclesiastical buildings (Buxton 1970;
Garlake 2002). Ethiopian Christian manuscript painting has also been traced
back to late Aksumite times (Mercier 2000; D. W. Phillipson 2003b).
Aksum was from early times a major trading and imperial power. With
the eclipse of Meroe it effectively controlled the trade between the Red Sea
and the rich hunting grounds of the Sudanese Nile Valley, whence came
the ivory which was one of Aksum™s major exports. In the third century,
and again in the sixth, the kings of Aksum held sway also over parts of
southern Arabia. When, in the seventh century, Arabs gained control over
the Red Sea ports, Aksum was cut off from much of the trade on which its
prosperity had depended, and the kingdom rapidly declined into obscurity
(Sergew 1972; Munro-Hay 1991; D. W. Phillipson 1998).
Iron-using peoples before ad 1000 231

Fig. 117: The
Aksumite stela
(early fourth
century ad) still
stands, 20.6
metres high
232 afric an archaeolog y

The Red Sea trade of the ¬rst few centuries ad was not restricted to
Aksumite commerce (Horton 1996a). Most of our information comes from
a Greek work, The Periplus of the Erythraean Sea, which describes trade, origi-
nating in Alexandria, extending into the Indian Ocean (Casson 1989). There,
contact was made with sailors from Yemen who penetrated down the East
African coast as far as a port called Rhapta. The exact location of this place
is unknown (Horton 1990), but it is generally believed to have been some-
where on the coast of modern Tanzania (Chami 2001). The traders brought
metal and glass objects which they exchanged for gums, spices, ivory and
rhinoceros horn. This commerce continued through the early centuries ad
and is well illustrated at Berenike on the Egyptian Red Sea coast (Sidebotham
and Wendrich 1998). Excavations at Hafun on the coast of Somalia some
150 kilometres south of Cape Guardafui have revealed traces of a trading
settlement which probably dates back to this period (Chittick 1976; M. C.
Smith and Wright 1988). To the south, however, no coastal settlements in
East Africa that may have been visited by these early seafarers have yet been
discovered, although ¬nds of Roman-like pottery and beads on Zanzibar and
in the vicinity of Dar es Salaam provide some archaeological evidence for
the contacts reported in the written sources (Juma 1996; Chami 1999).
The later archaeology of southern Ethiopia remains largely unknown,
investigations having been largely restricted to burials marked by cylindri-
cal stelae (Azaïs and Chambard 1931; Anfray 1982; Joussaume 1995). In the
southern Sudan, likewise, research is at such a preliminary stage that few
¬rm conclusions can yet be drawn (Mack and Robertshaw 1982). It appears
that iron tools remained extremely scarce until relatively recent times and
that in some areas mode-5 lithic technology was practised well into the sec-
ond millennium ad. At Jebel et Tomat, between the Blue and White Niles
not far south of Khartoum, an extensive settlement was occupied through
the ¬rst ¬ve centuries ad. Cattle, sheep and goats were herded, sorghum was
cultivated, and food was also obtained through hunting, fowling and ¬sh-
ing. The pottery tradition at this site appears to have been a continuation
of that which had prevailed in far earlier times in the Sudanese Nile Valley,
and also shows connexions with contemporary Meroitic wares. Iron tools
were rare, and backed microliths were in use throughout the occupation
(J. D. Clark and Stemler 1975). In the absence of evidence to the contrary,
it is tempting to suggest that Jebel et Tomat represents a type of rural set-
tlement that was widespread at this time both in the central Sudan and
further to the west and southwest (cf. Fig. 118).
In the southern Sudan, the Wun Rok mounds 150 kilometres north of Wau
provide a view of the iron-using communities who inhabited the plains of
the Bahr el Ghazal during the ¬rst millennium ad. They herded humpless
cattle, hunted and ¬shed. Iron was used primarily for personal adornment
Iron-using peoples before ad 1000 233

Fig. 118: Stone
bead, lip-plugs,
sherd disc, shell
pendant and
fragment of iron
bangle from Jebel

and many tools were made of bone. Throughout the second half of the ¬rst
millennium ad these people produced roulette-decorated pottery similar
to that made in more recent times by Nilotic-speakers (David et al. 1981).
Roulette-decorated pottery (Soper 1985; Desmedt 1991) is largely but not
exclusively a second-millennium phenomenon; investigation of its possible
link with Nilotic-speaking peoples (cf. Ehret 2003) represents a major lacuna
in current archaeological research.
In northern Kenya, herding peoples, probably speaking languages of both
the Nilotic and Cushitic types, using stone tools and evidently still unfa-
miliar with iron-working technology, continued the life-style of their prede-
cessors throughout the ¬rst millennium ad. Extensive settlements existed
beside permanent sources of water, as at North Horr. In more arid areas,
nomadic pastoralism was now the only effective means of subsistence. To
the south, in the highlands west of the Rift Valley, it seems likely that
iron-users of southern Sudanese af¬nities, speaking Nilotic languages, were
established by early in the ¬rst millennium ad. The evidence for this is
primarily linguistic; and archaeological con¬rmation is not yet forthcom-
ing (Ehret 1974). In the Rift Valley itself, the herders described in chapter 6
appear to have continued in occupation until late in the ¬rst millennium ad,
perhaps obtaining some knowledge of iron -- which, however, remained very
rare -- towards the end of that period (Ambrose 1984b; Bower 1991). The com-
ing of iron to the Bantu-speaking parts of East Africa is discussed below;
234 afric an archaeolog y

elsewhere, informative evidence for use of metal prior to ad 1000 remains
extremely scanty.

West Africa
Presentation of a coherent overview of the evidence for early iron-using
peoples in West Africa is hampered by the very incomplete and uneven
coverage of the research that has yet been undertaken. This is, of course, a
dif¬culty in all areas and periods of African prehistory, but the problems for
West Africa are particularly acute. Such research as has taken place has been
concentrated on sites which have yielded art objects or which are connected
with the trading states mentioned in foreign written accounts. There has
been virtually no integrated investigation of technological and economic
development, or of state-centralisation processes, although pioneering work
around the Inland Niger Delta in Mali (S. K. McIntosh 1999b), in the sahel
(Gronenborn 1998) and in Senegal (S. K. McIntosh and Bocoum 2000) amply
demonstrates the potential for such research. As a result, whole areas remain
effectively unknown archaeologically (cf. S. K. McIntosh 1994).
Two areas of West Africa have received particular attention as provid-
ing indications that copper may have been worked there before iron. Near
Akjoujt in southwestern Mauritania, copper ore was mined and smelted by
the ¬fth century bc. Around Agadez in Niger, furnaces of elongated plan
were used for smelting copper at a similar period (Tylecote 1982). Claims
that simple pit furnaces were used at an even earlier time for melting native
copper are not supported by more recent research (Killick et al. 1988). In both
areas copper oxides and carbonates were reduced in simple furnaces and the
resultant metal hammered into small tools such as arrowheads and spear-
points (Lambert 1971, 1983). At Akjoujt contact with North Africa at this
time may be demonstrated archaeologically, and similarities with Agadez
suggest that both centres may have derived elements of their metallurgical
expertise from that source (S. K. and R. J. McIntosh 1988). It thus appears that
the advent of both iron and copper to these areas should be dated around
the middle of the ¬rst millennium bc, which is in accord with evidence
from southern Aïr, also in Niger. Yet further signs of early iron-working in
the immediately sub-Saharan latitudes come from Rim in Burkina Faso, in
a context of the ¬rst centuries ad showing continuity from earlier times.
The earliest occupation of the large settlement at Jenne-Jeno, in the Inland
Niger Delta of Mali, dates from the last two centuries bc when use of metal
tools may have facilitated the ¬rst exploitation by farming people of the
Inland Delta™s heavy clay soils and the cultivation of African rice (R. J. and
S. K. McIntosh 1981; R. J. McIntosh 1998). The clear implications of these
Iron-using peoples before ad 1000 235

Fig. 119: Nok
terracotta heads

sparse and broadly scattered data are that use of iron in the southernmost
Sahara and adjacent northern savanna of West Africa dates back to the
last few centuries bc, and was widespread (Posnansky and McIntosh 1976;
Calvocoressi and David 1979; Togola 1996). There is no evidence to suggest
that any major replacement of population accompanied the beginning of
metal-working, although expansion into previously unexploited areas may
have been facilitated.
In more southerly parts of West Africa, the earliest evidence for iron so
far known is that associated with the Nok ˜culture™ found in a restricted
area on the southern and western slopes of the Jos Plateau in Nigeria (Shaw
1978; Jemkur 1992), and from sites near Nsukka (Okafor and Phillips 1992).
Nok artefacts were originally discovered during mining operations in river
gravels which, as noted in chapter 4, have also yielded abundant remains
from earlier periods. The most striking discoveries were of pottery ¬gures,
mostly of humans, some of which are life-size (B. Fagg 1979; de Grunne
1999). These terracottas show great technical competence and an artistically
accomplished, though idiosyncratic, style of modelling. Examples are shown
in Figure 119. Particularly characteristic are the elaborate hairstyles and the
treatment of eyes, which are shown as sub-triangular areas delineated by
grooves, with a deep circular hole for the pupil. The faces are unmistakably
negroid. The presence of parts of limbs and torsos as well as heads suggests
236 afric an archaeolog y

Fig. 120:
Reconstruction of
an iron-smelting
furnace at
Taruga (after
Tylecote 1975)

that some specimens were fragments of complete ¬gures, although few of
these have survived intact. Particular attention was paid to reproducing
physical peculiarities and deformities.
The attributes of some of the Nok ¬gures provide information about their
makers™ material culture: one man carries a hafted axe, others sit on stools
or wear beads and pendants, a ¬‚uted pumpkin is represented. Because they
came from disturbed contexts, which also yielded archaeological material of
several distinct periods, there was initially considerable controversy about
the date of the Nok ¬gures, or even as to whether they were earlier or
later than the local beginning of iron-working. The situation has now been
clari¬ed, because Nok settlement sites at Taruga and Samun Dukiya have
been found and investigated. These have yielded radiocarbon dates between
the ¬fth and the third centuries bc, associated with fragments of typical Nok
¬gures, domestic pottery and the remains of furnaces which conclusively
demonstrate that iron was worked by the Nok people (A. Fagg 1972; Tylecote
1975). The shallow pit furnaces with cylindrical clay walls are of particular
interest for comparison with later examples from other parts of sub-Saharan
Africa (Fig. 120). Knives and points for arrows and spears were the principal
types of iron artefact produced, together with occasional bangles. The ˜Nok
culture™ continued to ¬‚ourish into about the second quarter of the ¬rst
millennium ad.
Unfortunately we know nothing about the local predecessors of the Nok
people, and we cannot evaluate the impact which the start of iron-working
made on their way of life. It is possible that the Nok tradition of naturalistic
clay modelling may have begun before iron was known, but the few early
radiocarbon dates come from disturbed contexts and their evidence cannot
be regarded as conclusive. In fact, the only excavated sites in West Africa
where the beginning of iron-working may be pinpointed in a continuous
Iron-using peoples before ad 1000 237

sequence are in plains bordering Lake Chad in the extreme northeast of
Nigeria and in neighbouring states (cf. pp. 200--1 above). Here, only 800
kilometres from the Nok area and a similar distance from Aïr, iron seems
not to have been known before the ¬rst millennium ad -- perhaps as much
as eight hundred years later than at Taruga. At Daima, for example, the
economy of the early iron-workers, like that of their predecessors, was based
on their herds of domestic cattle. Sorghum was cultivated, as it may have
been in earlier times. There was no signi¬cant change in the associated
pottery at the time when iron-working began, but more substantial houses
were now built of mud rather than of wood and grass. Some time after the
initial appearance of iron a new population seems to have arrived at Daima,
and the site became part of a more extended trade system (Connah 1976,
1981). The poorly documented Sao sites (Lebeuf 1962) of southern Chad are
probably similar to, and broadly contemporary with, Daima although there
are indications that the advent of iron may in this more easterly area have
been somewhat earlier (see also Breunig et al. 1996; Gronenborn 1998; Holl
2002). In the Koro Toro area of north-central Chad extensive iron-working is
attested from the ¬fth century ad.
In Ghana, iron-working near Begho, as at Atwetwebooso, extends back to
around the second century ad, but there is a dearth of other excavated sites
that have been dated to the ¬rst millennium. In the same general area, sites
which are traditionally associated with the origins of Akan groups are dated
from the ¬fth century ad onwards and show a pottery style which developed
into those produced by Akan in later times. Rockshelter sites in Ghana,
notably Akyekyemabuo, show that microlithic industries and ground-stone
artefacts continued in use through the ¬rst millennium ad (Anquandah
1982; Stahl 1994).
In more westerly regions only isolated discoveries have so far been
reported, and no comprehensive account can be presented. It appears, how-
ever, that iron-working was not widely adopted prior to the middle centuries
of the ¬rst millennium ad, this being the range of the earliest relevant
radiocarbon dates from Liberia, Sierra Leone and Senegal. The majority of
these dates come from rockshelters or coastal shell middens where there is
evidence for continuous occupation from earlier times.
The Nok sites were followed by those of other iron-using communities
in various parts of Nigeria later in the ¬rst millennium ad, as is shown at
several settlements that were investigated in the area now ¬‚ooded by the
waters of the Kainji Dam on the Niger. Near Yelwa, for instance, a small
village probably of some eighty inhabitants seems to have been occupied
through most of the ¬rst seven centuries ad by people who herded domestic
animals and made stone beads as well as pottery vessels in two successive
238 afric an archaeolog y

styles. They also made clay ¬gurines which, in comparison with those of
the Nok ˜culture™, are small and lack re¬nement. Stone tools had almost
completely fallen out of use, while those of iron -- notably axes, knives, ¬sh-
hooks and heads for spears and arrows -- were abundant (Breternitz 1975).
In view of the major developments in art, technology and the socio-political
¬elds which, to judge from the evidence of later times, must have taken
place during the ¬rst millennium ad, it is unfortunate that the archaeology
of this period in Nigeria remains so little known and has been the focus of
so little co-ordinated research.
In eastern Nigeria, south of the Benue, there is no evidence that iron
was known before the early centuries ad. However, the discoveries from an
apparently ninth-century context at Igbo Ukwu near Onitsha show that, by
the end of the ¬rst millennium, a great concentration of wealth was in
the hands (or at the disposal) of a minority who held considerable religious
power and perhaps, to judge from later parallels, political authority also. The
Igbo Ukwu site was a burial place where persons of great importance had
been interred in an elaborate manner and accompanied by rich belongings.
Meticulous excavation permitted the principal burial to be reconstructed in
considerable detail. The corpse, sitting on a stool, dressed in and surrounded
by regalia of of¬ce, was placed with three ivory tusks in a deep, earth-
dug, wood-lined burial chamber which was then roofed over. In the upper
chamber thus created were placed the remains of at least ¬ve attendants,
the whole grave being then ¬lled with earth. Nearby, two further caches of
artefacts were discovered; in one case the objects appeared to have been laid
out in a relic house; in the other they had been unceremoniously buried in
a pit. There could be little doubt that the whole site represented a single
burial complex; radiocarbon dates provide good evidence that it belongs to
the ninth century ad, although some archaeologists initially argued for a
later date. The excavator has suggested that the person buried at Igbo Ukwu
may have held a position analogous to that of the Eze Nri of the recent Ibo
people (Shaw 1970).
Without doubt the most remarkable features of the Igbo Ukwu discover-
ies are the superb bronze castings that were found in all three parts of the
site (Fig. 121). They were produced by the ˜lost wax™ method, in which a wax
model is encased in clay to produce a mould, the wax then being replaced
by molten metal. The bronzes include a series of elaborate vases with del-
icate surface decoration, bowls, and models of shells, as well as a breast-
ornament, ¬‚y-whisk handle and other regalia which accompanied the main
burial. Their style and the nature of their elaborate decoration have no close
parallels in other West African bronze-casting traditions, although some fea-
tures do recur on pieces attributed to the later and ill-de¬ned ˜Lower Niger™
Iron-using peoples before ad 1000 239

Fig. 121: Lost-
wax bronze
group of bronzes (Willett 1967). Despite their technological and stylistic
castings from
Igbo Ukwu sophistication, the Igbo Ukwu bronzes have no known antecedents. There
can, however, be virtually no doubt that they were manufactured locally.
There are strong stylistic similarities between the sites™ bronzes and pottery,
both of which display an iconography which shares features with more
recent Ibo symbolism, and it is now known -- contrary to previous belief --
that the metal from which they were made could have been obtained in the
region itself (Chikwendu et al. 1989; Craddock et al. 1997). The Igbo Ukwu
discoveries provide a glimpse of a local development of wealth and crafts-
manship within the southeast Nigerian forest region. This had probably
been under way for some time before the ninth-century date now accepted
for Igbo Ukwu. The extent to which long-distance trade -- re¬‚ected by the
presence of over 150,000 glass beads at Igbo Ukwu -- may have stimulated or
contributed to this development (Sutton 1991, 2001) remains an open ques-
tion, consideration of which will require discussion of contemporary events
in the savanna regions to the north.
240 afric an archaeolog y

Kumbi Saleh Koro Toro
El Oualedji
Bambuk Rim


Samun Dukiya
Nsukka Nana Mode
Igbo Ukwu
Bioko I. Maluba



Tchissanga Gombe
0 500 1000 km

0 500 miles

Fig. 122:
Principal West
The Igbo Ukwu discoveries indicate that craft specialisation and the con-
and central
centration of wealth which are seen in later times, in the Ife and Benin
African sites with
evidence for the
kingdoms for example (see chapter 8), were already taking place by the end
early use of
of the ¬rst millennium ad. It is logical to conclude that state-formation pro-
cesses were also taking place, at least in some parts of West Africa, during
the ¬rst millennium, and this is rendered more probable by evidence from
further to the north. By the eighth century ad, when Muslim traders ¬rst
reached the southwestern Sahara, they found that the gold trade with the
north was controlled by the powerful kingdom of Ghana, centred between
the upper reaches of the Niger and Senegal rivers in the borderlands of
the modern southern Mauritania and southwestern Mali (Fig. 122). Ancient
Ghana was thus far removed from the modern state which has taken its
name. The principal gold¬eld exploited at this time was located to the south,
around Bambuk in the Senegal headwaters, but the kingdom of Ghana
Iron-using peoples before ad 1000 241

controlled its access to the trading centre of Awdaghast, at the southern
end of the main camel-caravan route from the north. Its site has now been
convincingly identi¬ed with the ruins of Tegdaoust, where excavations have
revealed prolonged occupation from at least as early as the seventh cen-
tury, although the main stone buildings are signi¬cantly later (Polet 1985;
Robert-Chaleix 1989). There has been much controversy surrounding the
location of the capital of ancient Ghana (R. J. McIntosh 1998), one candidate
being the ruins at Kumbi Saleh in the extreme south of Mauritania, which
cover more than 2 square kilometres, the adjacent cemeteries being even
more extensive (Thomassey and Mauny 1951, 1956; Berthier 1997). The date
of the establishment of this state remains uncertain, but must have been
well before the eighth century, when Muslim visitors from the north were
impressed by the power and wealth of its ruler (Levtzion 1973). This view is
supported by radiocarbon dates from Kumbi Saleh (Sutton 1982) which show
that occupation there had begun by the sixth century ad. There can thus be
no doubt that the kingdom of Ghana was essentially a local development,
based upon local resources (MacDonald 1998), whose resultant wealth sub-
sequently attracted the attention of trans-Saharan traders. The archaeology
of Muslim peoples in West Africa is further discussed in chapter 8 (see also
Insoll 2003).
Attention has already been drawn (p. 234) to Jenne-Jeno, a mound-site in
the ¬‚ood plain of the Inland Niger Delta. It provides the clearest evidence
yet available for the early development of West African urbanism, at a date
long before the advent of Muslim traders. Founded late in the ¬rst millen-
nium bc as an agricultural village covering a few hectares, Jenne-Jeno rapidly
became a true urban centre at the focus of settlement in the Delta region.
As the excavators have noted, ˜the ability to create food surpluses for export
along the river to less favoured regions permitted the inhabitants to procure
through trade the raw materials, such as iron, copper and stone, that the
Inland Delta lacked™ (S. K. and R. J. McIntosh 1984: 87). By ad 300 Jenne-Jeno
had expanded to 25 hectares in extent, and excavations have yielded evi-
dence for both herding and ¬shing, as well as for the cultivation of millet,
sorghum and African rice. By the seventh or eighth century, gold was among
the items brought from beyond the Delta to the city, which now reached
its greatest extent -- 33 hectares -- and was surrounded by a wall almost 2
kilometres in circumference (Fig. 123). It was at about this time also that
many peripheral sites in the ¬‚ood plain were abandoned, emphasising the
emergence of a truly urban system (S. K. and R. J. McIntosh 1980, 1984; S. K.
McIntosh 1999b).
Further states arose to the east of ancient Ghana before the close of the
¬rst millennium ad. Among these, adjacent to Lake Chad, was Kanem, which
grew rich through trade with the north, primarily to Tunis and Tripoli.
Fig. 123: Location and plan of Jenne-Jeno (after S. K. and R. J. McIntosh 1980)
Iron-using peoples before ad 1000 243

Fig. 124:
Megalithic stone
circle at Sine
Saloum, Senegal

Traditions suggest that the Zaghawa rulers of Kanem originated in the south-
ern Sahara. Their penetration southwards to establish themselves as over-
lords of settled farming people in the savanna may have been stimulated by
continuing desiccation of the desert and also by desire to control the source
of the slaves and ivory on which their prosperity depended (H. F. C. Smith
1971). It is tempting to link this process with the change in artefact types
at Daima which, as noted above on pp. 200--1, accompanied evidence for an
expansion and intensi¬cation of trade late in the ¬rst millennium ad. The
later history of these sudanic states is described in chapter 8.
A remarkable series of megalithic monuments -- settings of large, erect
stones (Fig. 124) -- in the Gambia, adjacent parts of Senegal and extending
eastwards into Mali, have burials within them that date from the second
half of the ¬rst millennium ad (Thilmans et al. 1980). It is tempting to link
them with the large burial mounds which occur at numerous places in the
savanna from Mali to Senegal (S. K. and R. J. McIntosh 1993). These clearly
represent non-Islamic interments, some being of very wealthy individuals,
and are best dated between the sixth and fourteenth centuries ad (Connah
2001). At El-Oualedji in Mali a 12-metres-high earthen tumulus was found to
cover a wooden burial chamber containing the remains of two people and a
variety of grave-goods (Desplagnes 1951). As further discussed in chapter 8,
there are close parallels between these graves and the funerary customs
recorded by al-Bakri in the eleventh century.
244 afric an archaeolog y

The more easterly regions of the northern savanna, stretching from
Cameroon into the Central African Republic, remain very poorly known
archaeologically (Asombang 1999). It was noted in chapter 6 that, probably
before the coming of iron, some of the stone-tool-using farming peoples of
the eastern Nigeria/Cameroon area had begun to expand southwards into
the forest, and probably also eastwards along its northern margin. To the
north, at Nana Mode in the Central African Republic, an iron-using settle-
ment in the savanna is dated to about the seventh century ad; the pottery
from this site was decorated by means of carved wooden roulettes (David
and Vidal 1977).
In attempting a provisional overall picture of cultural trends in West Africa
around the beginning of the Christian era, it is essential to bear in mind the
very incomplete coverage of the research that has so far been undertaken.
An understanding, however provisional, of West African developments is,
however, particularly important in view of the indications, to be cited below,
that this region played a major role in the transmission of iron technology
and associated culture to more southerly parts of the continent.
The ¬rst point is that it seems no longer possible to argue that the devel-
opment of metal-working technology in West Africa was due exclusively
to transmission from the north. It now appears that little reliance can be
placed on claims that the working of copper in parts of Mauritania and
Niger began signi¬cantly before iron. There, as in more southerly parts of
West Africa, both innovations are now believed to date from around the
middle of the ¬rst millennium bc. At ¬rst sight, the most plausible explana-
tions for the beginning of West African iron-working seem to be either that
it was a local invention, or that some relevant knowledge was transmitted
south of the Sahara from the north, being then progressed and elaborated
in a speci¬cally African manner. These hypotheses are not, of course, mutu-
ally exclusive. It is arguably very unlikely that no-one in West Africa had
any knowledge of contemporary technological achievements north of the
Sahara. This would not mean that the greater part of West African metal-
lurgical expertise was not a local development. Further consideration will
have to await the recovery of more detailed evidence as to the smelting
methods that were practised in these two regions.
There can be little doubt that iron was adopted in the West African
savanna by peoples whose ancestors had lived there for many centuries
previously. Savanna--forest contacts at this time have been only cursorily
investigated, but continuity of settlement may well have prevailed in the
latter area also. This conclusion is supported by the observation that,
although metal-working was known from the mid-¬rst millennium bc, some
groups retained their stone-tool technology for at least a thousand years
Iron-using peoples before ad 1000 245

Archaeological research in West Africa has been concentrated on sites
which illustrate the history of the known states, especially on those which
have also yielded art objects. This has tended to obscure the fact that the
earliest iron-using communities were composed essentially of peasant farm-
ers, and that such people have probably formed the majority of the West
African population throughout the last 2000 years. Very little research has
been undertaken on the subsistence economies of these peoples. One must
assume that there was a major contrast between the herding and -- in
places -- cereal agriculture of the savanna regions, and the yam cultiva-
tion which, with arboriculture, provided the mainstay for life in the more
easterly forest and adjacent woodlands. The extent to which African rice was
grown at this time remains unknown, the primary evidence being restricted
to the Inland Niger Delta. Inter-regional exchange of commodities must be
seen as a major factor during this period and one that acted as a signi¬cant
stimulus to the processes of urbanisation and state-formation, long before
the development of formal trans-Saharan trade.

Central Africa
It is convenient to discuss separately the sparse archaeological evidence for
early iron-using peoples that is now available from this vast, largely forested
region, the greater part of which lies within the basin of the Congo River.
The northwestern part of the region, in the southern half of what is now
Cameroon, is contiguous with -- and in many respects a part of -- the forest
belt of southeastern Nigeria which has been discussed above. At Obobogo,
near Yaounde, a series of village settlements was occupied during the ¬rst
millennium bc (de Maret 1989). These sites, each covering an area of about
2 hectares in what was then a forest clearing, are marked by ¬‚at-based
pottery, ground-stone axe/hoes, grindstones and chipped-stone artefacts. A
series of deep pits, of unknown function, contained these characteristic arte-
facts as well as shells of the edible nut Canarium schweinfurthii and of the
oil palm Elaeis guineensis. Some, but not all, of the pits contained a few
pieces of iron slag. It appears that iron came into use on a small scale dur-
ing the occupation of Obobogo: the date at which this occurred cannot yet
be determined precisely but was probably around the fourth century bc.
It is noteworthy that ground-stone axe/hoes continued in use after the
appearance of iron.
This last point is relevant to an evaluation of discoveries at Batalimo (de
Bayle 1975), located in the extreme south of the Central African Republic
near the Lobaye-Ubangi con¬‚uence, some 500 kilometres east of Obobogo.
This extensive site appears to have been a factory for the production of
stone axe-hoes. The pottery is ¬‚at-based, like that at Obobogo, although the
246 afric an archaeolog y

Fig. 125: Pottery
vessel and stone
axes from
Batalimo (from
Eggert 1987a,
after de Bayle)

decoration is distinct (Fig. 125). A date around the fourth century ad is
indicated by both thermoluminescence and radiocarbon analyses (de Maret
1985a). Despite this relatively late date, there is no evidence for the use of
iron at Batalimo.
The off-shore island of Bioko, Equatorial Guinea, provides additional evi-
dence for the late continuation of stone-tool use, there being no source of
metal in its exclusively volcanic rocks. Stone axe/hoes were produced in large
numbers on Bioko and, it appears, traded to adjacent areas of the mainland.
Iron-using peoples before ad 1000 247

Fig. 126: Pits at
Oveng, Gabon

The known pottery sequence on the island shows certain similarities with
that of the mainland, but seems to date only from the second half of the
¬rst millennium ad: it is not clear whether the earliest phase has yet been
identi¬ed (de Maret 1982b; de Maret and Clist 1987).
The beginning of iron-working in the forest regions of Central Africa is
best illustrated by recent archaeological investigations in Gabon and Congo.
It seems that, at least in the former country, the new technology was
adopted in inland regions signi¬cantly before it appeared on the coast. There
are sites with iron-working dated to the last 500 years bc in all the inland
Gabonese provinces, but none near the coast before the ¬rst century ad (Clist
1989a). It has been suggested that knowledge of metallurgy may have been
introduced from the savanna areas of Congo, subsequently spreading across
Gabon by way of the Ogooue River system (Oslisly and Peyrot 1992). Few
details of these early sites have yet been published, but the iron-smelting
furnaces seem generally to have comprised a clay-lined pit surmounted by a
clay shaft. By early in the Christian era several contemporaneous groupings
may be recognised among the iron-using inhabitants of what is now Gabon.
One of the best known is the Oveng group, named from a site near Libreville
(Clist 1989b) where quantities of marine shells indicate a major source of
food, and deep pits (Fig. 126) are reminiscent of those at Obobogo. Remains
of pole-and-mud buildings suggest that the site was occupied on at least a
semi-permanent basis. Despite variations in the decoration, it is signi¬cant
that the pots in all these groups have ¬‚at bases.
248 afric an archaeolog y

Most of our knowledge about Congolese archaeology of this period comes
from ¬eldwork conducted by Denbow (1990) in the vicinity of Pointe Noire.
Both open grassland and dense forest environments are represented in this
coastal region. At Tchissanga, two phases of occupation have been recog-
nised, both being marked by artefacts of the general type noted above. In
the earlier phase, securely dated to the sixth century bc, there is no evidence
for the use of iron, but several fragments were recovered from deposits of
the later phase which belongs to the late fourth century bc. Although iron
may have been brought to Tchissanga from elsewhere, smelting is clearly
attested on the Congo coast during the second or third centuries ad at
the large village-site of Madingo-Kayes. The technology which they exhibit
is identical to that employed in later times further to the south (D. Miller
1998). Ground-stone tools also continued in use throughout this period.
The earliest iron-using peoples in the Lower Congo region of D. R. Congo
were those of the so-called Kay Ladio group. At Sakuzi there is evidence
for continuity of occupation from the preceding Ngovo group (see p. 202)
and, indeed, the pottery of the two groups shows signi¬cant similarity. Asso-
ciations of Kay Ladio pottery with fragments of iron are dated between the
mid-¬rst and the early third centuries ad (de Maret 1986). Broadly contempo-
rary settlement by iron-using peoples is known further up the Congo River,
at Gombe Point near the modern Kinshasa (Cahen 1981).
The vast interior equatorial forest of D. R. Congo has only recently begun
to yield archaeological remains of this general period (Eggert 1992; Wotzka
1995). Pottery very similar to that from Batalimo (see pp. 246--7 above) has
been found to extend far into the forest along the Ubangi River. The most
informative occurrence, dated to the last two centuries bc, is at Maluba in
deep pits where oil palm nuts were also recovered (Eggert 1987a). It seems
reasonable to conclude that the Batalimo pottery tradition lasted for some
500 or 600 years, the name-site being a late occurrence. None of the forest
sites has yielded traces of iron, but these are unlikely to have survived in
the environmental conditions which prevail there. Sites along the southerly
Congo tributaries, notably the Momboyo, are marked by another style of
¬‚at-based pottery, named Imbonga, which is likewise generally preserved in
deep pits containing oil palm and Canarium nuts. These sites probably date
to the late ¬rst millennium bc or early centuries ad. Again, no iron was
preserved, although it is likely to have been in use at this time. The far east
of the Congo Basin lies within the area, centred on Lake Victoria, which
was occupied rather more than 2000 years ago by the iron-working Urewe
people, discussed below (pp. 250--2).
There can now be little doubt that the forest regions of the Congo Basin
were occupied during the last millennium bc by people who established
village settlements marked by the presence of deep pits containing their
Iron-using peoples before ad 1000 249

characteristic pottery, the af¬nities of which are with the northern and
northwestern forest fringe. Some of the sites in the northwest are earlier
than those in other parts of the region, and it is in the former area that
their antecedents may be recognised. Clear evidence for the use of iron,
from about the ¬fth century bc onwards, is at present only available in the
northwest and west, from Cameroon to Bas Congo; it was probably used also
in the central forest region but has not survived in the local archaeologi-
cal record. It is tempting to interpret these sites as evidence for expansion
through the Central African forest by the ancestors of the recent Bantu-
speaking people (Vansina 1990). This proposed correlation will be discussed
and evaluated below (see pp. 261--5).

Eastern and southern Africa
The earlier iron-using communities over an enormous area of eastern and
southern Africa show a very remarkable degree of homogeneity, to the
extent that archaeologists generally attribute them to a single complex,
here named Chifumbaze.2 Radiocarbon dates indicate that the complex ¬rst
appeared on the west side of Lake Victoria around the middle of the last
millennium bc, extending by early in the Christian era to the northern and
eastern side of the basin. By about the third century ad it had expanded
southwards as far as KwaZulu-Natal (D. W. Phillipson 1975). The archaeo-
logical sites and artefacts of the Chifumbaze complex make a marked con-
trast with those that had gone before, and contain the ¬rst evidence in
these southerly latitudes for the cultivation of crops, for the herding of
domestic animals, for settled village life, for metallurgy and, south of
Tanzania, for the manufacture of pottery. In each case, these are cultural
features for which there is archaeological evidence in the more northerly
or northwesterly regions of Africa in earlier times; they were probably more
signi¬cant in the long term than was metallurgy (cf. Segobye 1998). The
fact that so many important aspects of culture were introduced more-or-less
together over such a wide area and so rapidly makes it highly probable that
these innovations in sub-equatorial Africa were brought about as a result
of the physical movement of substantial numbers of people. It cannot be
stressed too strongly that, as explained below, these innovations were not

2 Following my earlier proposal (D. W. Phillipson 1968a), the interim term ˜Early Iron Age™
was formerly applied to this complex. However, now that its parameters are relatively well
known, and to avoid confusion with contemporary early iron-using societies in other parts of
Africa, it is preferable to use a more distinctive designation. The name ˜Chifumbaze™, from the
rockshelter in Mozambique where pottery of this complex was ¬rst excavated (D. W. Phillipson
1976), has been widely adopted and is retained here. Other terms that have been proposed,
such as ˜Early Farming Communities™ (Morais 1988) or ˜Early Iron-Working™ (Chami 1995), are
felt to be insuf¬ciently distinctive.
250 afric an archaeolog y

Fig. 127: Urewe
ware from sites
in southwestern
Kenya (after M. D.
Leakey et al. 1948)

inseparably linked in a single ˜package™; although often broadly concurrent,
they represent separate processes of cultural change.
The ¬rst use of iron in the Lake Victoria area is attributed to the Urewe
group, named after a site in southwestern Kenya. Its characteristic pot-
tery (Fig. 127), in which several local sub-styles may be recognised, is
found in Rwanda and adjacent parts of D. R. Congo, in southern Uganda,
Iron-using peoples before ad 1000 251

northwestern Tanzania and around the Winam Gulf in southwestern Kenya
(Van Noten 1979; Van Grunderbeek et al. 1983). In Buhaya on the southwest-
ern shore of the lake, extensive settlements and iron-smelting sites, such as
Katuruka, date at least to the very beginning of the Christian era and may
be several centuries older (Schmidt 1978; Clist 1987). The iron-smelting tech-
nology of these sites has attracted considerable attention (Van Grunderbeek
et al. 2001). Although some claims for its sophistication may have been exag-
gerated, it was clearly both complex and accomplished (Schmidt and Childs
1985; Eggert 1987b). It produced on a substantial scale what was technically
a type of steel, although this may be of little signi¬cance since much of the
bene¬t appears to have been lost in the subsequent forging process.
No effective research has yet been undertaken to illustrate the subsistence
economy practised by the makers of Urewe ware. There are some indications
that they were probably herders of domestic cattle and cultivators of both
¬nger millet and sorghum (Van Noten 1983). Sediments in mountain swamps
and on the bed of Lake Victoria contain datable pollen which suggests
that there was a signi¬cant reduction in forest vegetation around the lake
about the middle of the last millennium bc (Kendall and Livingstone 1972;
D. Taylor and Marchant 1995). This could, of course, have been brought about
by climatic causes or by charcoal-burning for iron-smelting, rather than by
ground-clearance for agriculture, as discussed by Schmidt (1997b).
Much more research needs to be done before we shall be able to under-
stand the antecedents of the Urewe group. Their predecessors in parts of
the Lake Victoria area appear to have been the makers of Kansyore ware,
described in chapter 6, and the relationship, if any, between the two popu-
lations remains unknown (cf. MacLean 1996). There can be little doubt that
the Urewe sites represent a sharp discontinuity in the local archaeological
record, but we are not yet in a position to estimate the extent to which this
was due to rapid in situ development or to contact with other regions. No
close parallels are known for Urewe ware, although some similarities have
been suggested with pottery from Chad, far to the northwest (Soper 1971),
and also with certain recently discovered material from the equatorial forest
region. Linguistic studies may throw some light on this problem, and will
be discussed below.
Studies of pottery typology suggest that the earliest iron-using commu-
nities of more southerly latitudes may have been derived, at least in part,
from those of the Urewe group, and a large number of radiocarbon dates
con¬rm that this is probable on chronological grounds. Although points
of controversy remain, there is broad agreement among many archaeolo-
gists that the Chifumbaze complex to the south of the Urewe area may
best be considered as representing two or three separate facies or ˜streams™,
of which the more easterly probably derived directly from the Urewe group
252 afric an archaeolog y

(D. W. Phillipson 1977a). The ancestry of the western facies also may have had
Urewe connexions, but its origins remain less well understood: it probably
incorporated local elements from the western D. R. Congo/northern Angola
region. The distributions of these facies and of their constituent groups, as
currently known, are shown in ¬gure 128.
The easternmost facies extended by about the second century ad to the
coastal regions of southeastern Kenya and the adjacent parts of northeastern
Tanzania, where it is represented by sites yielding the characteristic type of
pottery named Kwale ware, after a site southwest of Mombasa. Settlement
seems here to have been restricted to the relatively well-watered hilly coun-
try where the villages of the region™s present Bantu-speaking farming com-
munities are concentrated (Soper 1967a, 1967b, 1982). It may be noted that
in the Rift Valley territory of the stone-tool-using herders no contemporary
traces of iron or of Kwale-related pottery have been discovered. Presum-
ably either this comparatively arid country was unsuited to settlement by
the Chifumbaze people, or the herders were able to prevent penetration by
the newcomers. Whatever the reason, it is probable that the Kwale indus-
try was introduced to the East African coast by a southerly route through
what is now central Tanzania, where Chifumbaze sites such as Lelesu have
been reported (Smolla 1956; Sutton 1968). Recently discovered sites of the
Chifumbaze complex in the vicinity of Dar es Salaam (Chami 1995, 2001)
may represent the common source both of Kwale and of its counterparts
further to the south.
Around the second century ad there took place an extremely rapid disper-
sal of iron-using farmers of the Chifumbaze complex through a wide area
extending southwards through Mozambique, Malawi, eastern Zambia and
Zimbabwe into Swaziland and adjacent parts of South Africa. The eastern-
most manifestation was largely restricted to the coastal lowlands and clearly
sprang from the Kwale group settlements of East Africa. More than 3000
kilometres appear to have been covered in less than two centuries. In the
south, the characteristic pottery by which this coastal dispersal may readily
be recognised is known as Matola ware, after a site near Maputo (Cruz e
Silva 1980; Sinclair 1986). South of the Zambezi, pottery of this general
type is now represented at numerous sites in the coastal regions of southern

Fig. 128: Sites of the Chifumbaze complex in eastern and southern Africa
1 Ben¬ca 7 Katuruka 13 Nakapapula Silozwane
2 Broederstroom 8 Kwale 14 Nkope Situmpa
3 Chifumbaze 9 Lelesu 15 Sakwe Thandwe
4 Kalundu 10 Lydenburg 16 Salumano Toutswe
5 Kapwirimbwe 11 Makwe 17 Sanga Tzaneen
6 Katoto 12 Matola 18 Schroda Urewe
254 afric an archaeolog y

Mozambique, in the Mpumalanga lowlands, and in KwaZulu-Natal at least
as far as latitude 31—¦ south (Maggs 1984). Villages, up to 2 hectares in extent,
were generally sited a few kilometres inland of the Indian Ocean shore, in
locations suited to the cultivation of cereals. The inhabitants of these short-
lived villages exploited iron and other local resources, obtaining most of
their protein from shell¬sh and other marine foods. Domestic animals, if
kept, were probably few. Af¬nities with Kwale ware are particularly strik-
ing in the pottery from Tzaneen in Mpumalanga Province (Klapwijk 1974;
Klapwijk and Huffman 1996). The archaeology of this time is poorly known
in the coastal regions between the Kwale and Matola areas, but compa-
rable material has been reported from the Nampula region of northern
Mozambique (D. W. Phillipson 1989a; Sinclair et al. 1993a). The involve-
ment of the East African coastal regions in Indian Ocean maritime trade
is demonstrable around the ¬rst century ad and again from the eighth cen-
tury onwards; indeed, continuity through the ¬rst millennium cannot be
ruled out but has not yet been thoroughly investigated (Chami 1995). For
convenience, the period of Islamic settlements on the East Coast, beginning
in the eighth century, is here discussed in chapter 8.
Further inland, in Malawi, eastern Zambia and much of Zimbabwe, the
Chifumbaze complex is represented both by substantial villages and by
pottery occurrences in rockshelters that apparently continued to be fre-
quented by stone-tool-using hunter-gatherers throughout the ¬rst millen-
nium ad, making only occasional contact with the farming communities
(D. W. Phillipson 1976, 1977a; Crader 1984). These inland occurrences are
attributed to the Nkope and Gokomere/Ziwa traditions (Fig. 129). Their rela-
tionship with the Matola tradition remains unclear and it is uncertain
whether they should be regarded as representing coastal and inland mani-
festations of an eastern facies of the Chifumbaze complex, or as two distinct
facies (cf. Huffman 1982, 1989).
The archaeology of the more westerly regions has so far been much less
intensively investigated, being well known only in the Katanga Province of
D. R. Congo and in central Zambia, where iron was worked as early as the
second century ad by villagers who cultivated sorghum and cowpeas. By the
¬fth century, domestic cattle were kept at Kapwirimbwe near Lusaka, and
other settlements had been established close to the rich mineral deposits
of the Zambia/Katanga Copperbelt (D. W. Phillipson 1968c, 1972a; Anciaux
de Faveaux and de Maret 1985; Robertson 2000). Most of our knowledge of
the late ¬rst millennium ad in Katanga comes from a series of cemeteries,
notably Sanga and Katoto in the valley of the upper Lualaba (Nenquin 1963;
Hiernaux et al. 1972; de Maret 1985b, 1992). By the closing centuries of the
¬rst millennium, copper -- which had been worked on a small scale for some
Iron-using peoples before ad 1000 255

Fig. 129: Pottery
of the eastern
facies of the
complex, from
Malawi (after
Robinson and
1968, Robinson
256 afric an archaeolog y

hundreds of years -- was used for a variety of display and prestige purposes.
This trend continued and expanded in more recent times, as is discussed in
chapter 8 (see also Herbert 1984).
There are indications that pottery which closely resembles that of the
Chifumbaze complex was made in parts of northern Angola by the early
centuries of the Christian era, as at Ben¬ca near Luanda (dos Santos and
Ervedosa 1970). The resultant suggestion that the Chifumbaze complex™s
beginnings in the western part of the sub-continent may have been at least
as early as in the east has been con¬rmed by investigations in the upper
Zambezi Valley (Katanekwa 1981; D. W. Phillipson 1989b). Several sites in this
area have yielded pottery which clearly belongs to the Chifumbaze complex,
in contexts now dated at Situmpa and Salumano to the third or second
centuries bc. These are thus the earliest manifestations of the Chifumbaze
complex yet known from the southern savannas. Iron artefacts have not
been recovered from the small-scale excavations so far conducted, but the
almost complete absence of ¬‚aked-stone artefacts suggests that metal was
indeed used by these sites™ inhabitants. Bones of domestic cattle and, less
certainly, ovicaprids at Salumano indicate the practice of herding.
Later manifestations of the Chifumbaze complex are known across north-
ern Botswana (Denbow 1990; Campbell et al. 1996), those in the east and
adjacent parts of the Limpopo Valley being particularly signi¬cant. In the
former area developments from around the seventh century onwards are
attributed to the Toutswe tradition, marked by a settlement system with
a small number of large easily defended hill-top sites surrounded by many
smaller ones generally located close to sources of water (Denbow 1984, 1986).
Cattle provided the mainstay of the system™s economy and there are good
reasons for tracing back to this period the important role played by cattle
in the culture of southern Bantu-speaking people. The settlement pattern
re¬‚ects this emphasis, with dispersed cattle camps located for water and
grazing availability, and larger sites serving as foci and centres for the com-
munity as a whole, as is still the case with modern Tswana peoples in this
area (Segobye 1998). It is clear that the large central sites were subject to
prolonged, permanent occupation: this, with the evidence for the herding
of very large numbers of cattle, suggests the rise of an ©lite such as may be
recognised also in rather later times at sites in the Limpopo Valley, notably
Schroda (Hanisch 1981; Fig. 130). Schroda was a large village of some 12
hectares, occupied between the eighth and the tenth centuries, and com-
prising a livestock enclosure, circular thatched houses and grain-bins. A par-
ticular area of the site was used for the working of ivory, which may have
been exchanged for imported glass beads, several hundred of which were
recovered. In another, more central, area were found several caches of clay
¬gurines, both human and animal, which strongly resemble those made in
Iron-using peoples before ad 1000 257

Fig. 130:
Schroda, viewed
from an
overlooking cave.
The Limpopo
River is visible in
the background.

Zimbabwe in later times. Subsequent developments in this Limpopo Valley
area are illustrated at the sites of Bambandyanalo and Mapungubwe (Meyer
1998; D. Miller 2001), further discussed in chapter 8; there is evidence for
further development of long-distance trade in ivory and, from the eleventh
century, in gold (Huffman 1982; Maggs 1984).
The sites of the ¬rst millennium ad in the highlands north of the Vaal
and in parts of KwaZulu-Natal are for the most part attributed to the
Msuluzi/Lydenburg tradition of the Chifumbaze complex. Opinion is divided
as to whether this tradition arose, around the ¬fth century, from the ear-
lier Matola tradition of the Mozambique and KwaZulu-Natal coastlands, or
whether its connexions lay further inland, perhaps with the western facies
of the Chifumbaze complex (Huffman 1982; Maggs 1984; D. W. Phillipson
1989a). By the seventh century, evidence for early farming settlement in
KwaZulu-Natal extended inland into the uKhahlamba/Drakensberg foothills
about as far as the 1000-metre contour, and for some 300 kilometres fur-
ther south than the Matola sites had done, penetrating what is now the
Eastern Cape Province in the vicinity of East London at latitude 33—¦ south
(Maggs 1980, 1995; Loubser 1993; Whitelaw 1994; Nogwaza 1994; Binneman
1996). Villages were often substantially larger than their earlier counter-
parts, located on colluvial soils near valley bottoms, and were supported by
well-developed mixed farming, with cattle and ovicaprids herded and a vari-
ety of crops cultivated including sorghum, ¬nger millet, bulrush millet and
258 afric an archaeolog y

pumpkin. The broadly contemporary and analogous occupation in the high-
land regions north of the Vaal is best known from sites at Broederstroom,
west of Pretoria, and near Lydenburg. A large village at the former site was
occupied around the sixth century, its inhabitants smelting the local iron
ore and relying for their food on the cultivation of cereals and the herding
of small stock, with perhaps a few cattle (Mason 1986). Around Lydenburg,
the mixed farming economy was more broadly based, but particular interest
attaches to the recovery, from a ¬fth-century context, of a series of life-sized
terracotta human heads (Fig. 131), which presumably served some ritual or
religious function (Maggs and Davison 1981; Evers 1982; Whitelaw 1995).
One of the heads was surmounted by a model of a domestic bovid, suggest-
ing both that herds occupied an important position in the lives of the site™s
inhabitants, and also that the heads may have been in some way linked
with practices intended to ensure the maintenance of the herds. Fragments
of similar objects have been found at other sites nearby and in KwaZulu-
Natal, but the Lydenburg heads are without parallel elsewhere south of the
Throughout the area inhabited by the Chifumbaze people, iron-working
was practised on a scale adequate to ensure that they only rarely, if ever,
used stone tools. In regions where the appropriate mineral deposits occur,
copper and gold were also worked. Mining appears to have been restricted
to small-scale operations during the ¬rst millennium ad; we have as yet
little reliable information concerning the types of smelting-furnaces that
were used, except in the interlacustrine region. Iron artefacts were mainly
utilitarian: axes, hoes, arrowheads and the like. Copper was used for ban-
gles and items of personal adornment; by ad 1300 in Katanga it was being
cast into small cross-shaped ingots which may have served as a form of cur-
rency (Bisson 1975; see also chapter 8). Gold-working, effectively restricted
to Zimbabwe, is attested only from the end of the ¬rst millennium ad
and seems largely to have been geared to an export trade through the
ports of the Indian Ocean coast, whence imported glass beads and, perhaps,
other more perishable items found their way inland (Summers 1969; Swan
Detailed evidence concerning the farming economies of the Chifum-
baze communities, and the plant and animal species on which they were
based, has only rarely been recovered. While the presence of iron hoes
and of numerous grindstones may indicate agriculture, conclusive proof
of this in the form of speci¬cally identi¬able plant remains comes from
disappointingly few sites, mostly in Zambia, Zimbabwe and South Africa.
The species for which we so far have evidence include bulrush and ¬nger
millet, sorghum, cowpeas and varieties of squash and beans. There can be
little doubt that most, if not all, of these crops were ultimately derived from
Iron-using peoples before ad 1000 259

Fig. 131:
Terracotta head
from Lydenburg

those which, as shown above in chapter 6, were originally brought under
cultivation in the areas to the north of the equatorial forest. More detailed
botanical studies will be needed before this statement can be ampli¬ed.
A small number of crops, notably the banana, appear to have been intro-
duced to eastern Africa from across the Indian Ocean. The date or dates at
which this took place remains effectively unknown, those most frequently
260 afric an archaeolog y


. 7
( 11)