ArticlePDF Available

Review Essay. What do we know about African iron working?

Journal of African Archaeology Vol. 2 (1), 2004, pp. 97-112
Les Routes du Fer en Afrique (
The Origins of Iron Metallurgy: Proceedings of the First International Colloquium on The Archaeology of Africa
and the Mediterranean Basin/Aux Origines de la Métallurgie du Fer: Actes de la Première Table Ronde
Internationale dArchéologie (LAfrique et le Bassin Méditerranéen). Jean-Paul Descoeudres, Eric Huysecom,
Vincent Serneels & Jean-Louis Zimmerman (eds.). Mediterranean Archaeology volume 14, 2001. (ISSN 1030-8482;
309 pages, figures, 8 plates. Price Australian $ 55,-).
Aux Origines de la Métallurgie du Fer en Afrique: Une Ancienneté Méconnue. Afrique de lOuest et Afrique
Centrale. Hamady Bocoum (ed.). Paris: Éditions UNESCO, 2002, ISBN 92-3-203807-2. (240 pages, 13 figures, 87
photographs. Price EUR 22,-.)
David Killick
David Killick
Department of Anthropology
University of Arizona
Tucson, AZ 85721-0030
The UNESCO project Les Routes du Fer
en Afrique
In 1988 the General Assembly of the United Na-
tions designated the next ten years as the Decade of
Cultural Development. As its contribution to this cam-
paign, UNESCOs Division of Intercultural Dialogue
invited proposals for educational initiatives based
upon cultural themes whose scale was larger than
that of individual nation states. The most widely known
of these initiatives was the acclaimed international
touring exhibition on the histories and cultures of the
regions along the central Asian Silk Route that for-
merly linked China to the Mediterranean. Among other
proposals submitted was one entitled Les Routes du
Fer en Afrique, proposed in 1991 by the Mozambique
National Committee for UNESCO, with the active sup-
port of France, Belgium, Portugal, Venezuela and the
Organization of African Unity.
The planning document for Les Routes du Fer en
Afrique emerged from a conference held in Maputo in
December 1991. Ten of the eighteen participants were
from southern Africa (Mozambique, Swaziland, Zimba-
bwe); the remainder were from the UNESCO Secretariat,
Senegal, Burundi, Algeria, Zaïre and Belgium. The rec-
ommendations of the conference are summarized on
the project web site and also in the volume edited by
BOCOUM (pp. 219-224). The major proposed activities
were the following:
uto investigate the role of iron in the history of Af-
uto create a bibliography of prior studies, both pub-
lished and unpublished (e.g. student theses);
uto solicit financial support from industry for new
research ;
uto undertake new research upon all aspects of iron
in Africa, specifically including the geology of iron
ore deposits, smelting and forging technology,
economic aspects (agriculture, trade, currencies),
social roles (e.g. bridewealth, symbols of power),
comparative linguistic studies of terms for iron and
iron working, and the role of iron in the colonial
and post-colonial industrialization of Africa; and
uto consider whether indigenous African tech-
niques of iron production could be revived in re-
mote rural areas as a cost-effective alternative to
importing iron from outside the continent.
Journal of African Archaeology Vol. 2 (1), 2004
D. Killick
The goals proposed by the Maputo Conference
were accepted by the Division of Intercultural Dia-
logue at UNESCO, which appointed an international
Scientific Committee of nineteen individuals to direct
the project. Among these were six archaeologists
(Hamady Bocoum, Pierre de Maret, Jean Devisse, Lech
Krzyzaniak, Mohamed Ould Khattar and Kléna
Sanogo); the other appointees were five historians, two
linguists, two sculptors, a conservator, a geologist, an
engineer, and a former Minister for Culture in Nigeria.
Thirteen of the nineteen were resident in Africa.
UNESCO then commissioned a feasibility study,
which was submitted by Susan Legêne of the Royal
Tropical Institute in October 1993. This recommended
the preparation of a bibliography, a travelling exhibi-
tion, a documentary film, and publications and radio
programs aimed at schools and the general public. All
of these would be intended primarily for African audi-
ences. A very limited role was proposed for academic
inquiry, with only a single conference to establish the
state of the art, to discuss theoretical and methodo-
logical issues, and to prepare a publication.
The feasibility study was circulated for comment
to several authorities on the subject of indigenous Af-
rican iron working, one of whom was myself, and bibli-
ographies of the relevant literature were commissioned.
A limited selection of references, and a link (no longer
functional) to the on-line bibliography prepared by
Duncan Miller and Tim Maggs, can be found on the
project web site, as can a directory of researchers in
the field of African metallurgy with their contact infor-
mation and relevant publications up to 1994. Plans for
a travelling exhibition were also begun, under the co-
presidency of Hamady Bocoum and Pierre de Maret,
assisted by a Commissariat Scientifique of no fewer
than thirty-one scholars, all but three of whom were
from Francophone nations.
In February 1995 the project sponsored a work-
shop in Abuja on the theme 25 centuries of iron work-
ing in Nigeria. Three papers presented at the workshop
(those by Joseph Jemkur, Edwin Okafor and Isaac
Akinjogbin) are included in French translation in the
volume edited by Bocoum. In June 1999 UNESCO was
one among several sponsors of the conference LAfrique
et le Bassin Méditerranéen: aux Origines de la Métallur-
gie du Fer at the University of Geneva, and since pub-
lished as a special issue of Mediterranean Archaeology
(volume 14, 2001, though it did not appear until December
2002). Twenty of the twenty-nine contributions to this
issue discuss iron metallurgy on the African continent.
In November 1999 an exhibition entitled Les
Routes du Fer en Afrique was presented at UNESCO
headquarters in Paris, with financial support from
UNESCO, the French steel company USINOR and sev-
eral other sponsors. Although this exhibition is still
advertised on the project web site as a travelling exhi-
bition, my own inquiries have turned up no evidence
that it has travelled beyond Paris. Much of the text and
a selection of the images from the exhibition can how-
ever be downloaded from the project website as a 50-
page Adobe pdf document. This has five sections. The
first is a display of African ironwork as art, juxtaposing
historic forged iron objects with the work of living art-
ists from Burkino Faso and Madagascar. The second,
entitled Les débuts du fer en Afrique, presents as
established fact the claim that iron working was in-
vented in Niger in the third millennium BC. (This claim
derives from the research of Gérard Quéchon and Alain
Person, which will be critically examined below). The
third section, Savoir le Fer, provides basic informa-
tion on iron ores in Africa, on ancient and modern tech-
niques of mining them, and a few illustrations of indig-
enous African smelting techniques. The longest sec-
tion is devoted to the role of iron in society. It touches
on the past importance of iron in agriculture and war-
fare, on the economic and ritual roles of ironworkers,
and on the continuing importance of blacksmiths, work-
ing with recycled steel, in Africa today. The fifth sec-
tion, Lire le Fer, is a very brief discussion of the aims
and methods of archaeometallurgy. Tacked onto the
back, without explanation, is a schematic diagram of a
state-of-the-art USINOR steel plant. A symposium was
held in conjunction with the exhibition at which papers
by Hamady Bocoum, Pierre de Maret, Gérard Quéchon,
David Aremu and Bruno Martinelli were presented. All
of these are included in the volume edited by Bocoum.
In April 2000 some members of the Scientific Com-
mittee joined UNESCO officials and selected African
politicians in Addis Ababa to issue a list of recommen-
dations, addressed to African governments, for fur-
ther research into African iron working, for conserva-
tion of important iron working sites, and for study of
the revival of indigenous African iron working for rural
development. Another conference was scheduled for
late 2001, but it is not clear whether this ever took place,
as the Routes du Fer en Afrique web site on the
UNESCO server has not been updated since July 2001.
It appears that UNESCO is no longer funding the
As this summary makes plain, few of the ambitious
goals proposed by the 1991 Maputo conference or in
Journal of African Archaeology Vol. 2 (1), 2004 99
Book review essay: What do we know about African iron working?
the 1993 feasibility study have been realized. Apart
from the exhibition, which was aimed at a general audi-
ence, the major products of Les Routes du Fer en
Afrique thus far have been papers generated by schol-
arly conferences. In the remainder of this review I ex-
amine the two published works that derive from con-
ferences funded in whole or in part by UNESCO.
The origins of iron working around the
Mediterranean and the Nile
This special issue of Mediterranean Archaeology
is particularly welcome because it combines fourteen
papers on iron working in sub-Saharan Africa with up-
to-date summaries of the evidence for the earliest iron
working in the Aegean, Anatolia, Egypt, France, Spain
and North Africa. For the last two decades Africanists
have relied upon the relevant chapters in The Coming
of the Age of Iron (WERTIME & MUHLY 1980) for summa-
ries of research outside Africa. This work has served
us well, but may now be retired.
Two chapters in Mediterranean Archaeology ex-
amine the supposed role of the Hittites in the innova-
tion of iron metallurgy. Eric Jean provides an anno-
tated list of the earliest iron objects recovered in exca-
vated contexts in Anatolia, while Jana Souckova-
Siegelová summarizes accounts of iron in Hittite and
Old Assyrian documents. Iron was certainly known
much earlier than Hittite times. A dozen iron objects,
including two remarkable sword blades, have been re-
covered from deposits of Early Bronze Age II and III
(2800-2100 BC), and some three dozen, including sev-
eral samples of steel, from the Late Bronze Age (ca.
1700-1200 BC), which coincides with the span of the
Hittite Empire1. Jean emphasizes that iron of any kind
was very rare in comparison to copper and bronze.
Souckova-Siegelová notes that the many references
to iron in Hittite documents are almost all from royal
records, and show that iron was in Hittite times a rare
and very valuable material reserved for royal use as
ritual objects, or given as gifts to other kings. Not until
the New Hittite era (1400-1200 BC) were iron weapons
worn by palace elites, and iron objects are not as abun-
dant as those of bronze in non-elite Anatolian sites
until the Middle Iron Age (after 850 BC). Both authors
agree that earlier writings, which attributed Hittite mili-
tary success to the possession of iron weapons, have
no basis in fact.
Indeed, Anatolia lagged behind both Cyprus and
Greece in the widespread adoption of iron, as Jean-
Louis Zimmermann shows in his chapter on the Aegean.
Although the earliest objects of iron in Greece and
Cyprus are dated only to the thirteenth century BC,
iron appears to have come into widespread use by the
tenth century. Metallography has revealed that a few
of these weapons and tools were of quenched and tem-
pered steel. There is unfortunately no chapter on the
earliest iron in the Levant and Palestine  an omission
that detracts considerably from the utility of this vol-
Michel Valloggias survey of Egyptian inscriptions
emphasizes (1) the lack of specificity in Egyptian hi-
eroglyphic and hieratic terms for metals, which makes
it difficult to determine when iron was first smelted in
Egypt, and (2) that there are no certain references to
the working of smelted iron (as opposed to iron mete-
orites) before the sixth century BC. He makes a strong
argument that the magnificent iron-bladed dagger, with
gold hilt and scabbard, from the tomb of Tutankhamen
(14th century BC) was not made in Egypt, but was a gift
from Anatolia. Michel Wuttmann, who contributes a
short chapter on the archaeological evidence, concurs
with this assessment. He also lists all reported finds of
iron before the advent of the New Kingdom. Most of
these were found in dubious context or were mere piles
of rust. Only two pieces have trustworthy chemical
analyses; both contain more than 7% nickel and were
probably forged from iron meteorites. Tutankhamens
tomb contained nearly two-thirds of all iron objects
from contexts older than 1300 BC, but unfortunately
none of these have been subjected to scientific analy-
sis. His rule marks the beginning of a period in which
iron objects occur more frequently, though it is not yet
known what proportion of these are local products and
what are imports. The first iron tools are in contexts
dated after 1000 BC, but it was not until the 26th and 27th
Dynasties (663-405 BC) that tools were commonly
1Throughout this review historical dates (or dates de-
rived from historical dates by crossdating) are cited as
calendar years BC or AD, uncalibrated radiocarbon ages
are cited as radiocarbon years BP, and calibrated radio-
carbon dates as calendar years cal BC or cal AD. All
calibrations cited were made with the computer pro-
gram OXCAL 3.9 (BRONK-RAMSEY 2003) at two sigma
(two standard deviations, 95.4% confidence interval).
2Reviews of the evidence for the origins of iron in the
Near East, the Indian subcontinent and China may how-
ever be found in the volume edited by PIGOTT (1999).
Journal of African Archaeology Vol. 2 (1), 2004
D. Killick
made of iron. Wuttmann notes that we know little more
about Egyptian iron technology now that we did in the
Patrice Lenoble provides a superb, though depress-
ing, commentary on the history of research into iron
working at Meroe. Ninety years after Sayces ill-con-
ceived labelling of Meroe as the Birmingham of an-
cient Africa, Lenobles highly critical review shows
that we still understand almost nothing about iron
working in ancient Nubia. Neither the chronology of
iron working at Meroe nor the scale of production have
been firmly established, still less the questions of state
control of the industry, or of its possible environmen-
tal impact. Nor do we even know how iron was used,
except for its deposition in royal funerary rites. Lenoble
argues convincingly that the abundance of iron arrows
in some royal tombs is a symbolic statement of the
power of the ruler, and cannot be taken as an index of
the abundance of iron in Meroitic society at large.
Although little progress has been made in the study
of Egyptian or Nubian iron working since the publica-
tion of The Coming of the Age of Iron, there have been
many recent finds at the western end of the Mediterra-
nean. Proponents of an indigenous invention of iron
working in West Africa have long emphasized the lack
of evidence for Phoenician iron working in North Africa,
but that objection is no longer valid. The evidence from
Carthage, summarized in a chapter by Hans Georg
Niemeyer, now includes an entire iron working quarter
used between the fifth and second centuries BC (exca-
vated by Serge Lancel), and two separate dumps of
tuyères and slag, both dated to the eighth and seventh
centuries BC (excavated by Niemeyer and by F. Rakob).
Chemical and mineralogical analyses of the earlier slags
from are presented in a chapter by Ingo Keesmann. He
concludes that both the copper and the iron slags from
these deposits derive from secondary processing (cru-
cible melting of copper, forging of iron). No furnaces
were found in association with the earlier dumps, but
Niemeyer includes a reconstruction of a slightly younger
(late seventh/early sixth century BC) Phoenician iron
smelting furnace that he excavated at Toscanos in
Spain. This is reconstructed as a domed slag-tapping
Carme Rovira Hortalàs chapter is a comprehen-
sive discussion of the earliest iron in the Iberian Penin-
sula. A large number of finds have been published
since 1980, and allow the first two centuries of the use
of iron to be traced in some detail. There are rare finds
of iron objects in Late Bronze Age contexts, dated to
the tenth and ninth centuries BC. Most scholars think
that these items were traded from the eastern Mediter-
ranean. On the south and south-eastern coast of Spain
iron objects first appear in quantity in the tombs of
Phoenician colonists during the eighth century BC. Iron
has been found in both Phoenician and non-Phoenican
contexts in present Portugal and in the interior of Spain
that are dated to the seventh and sixth centuries. In
Catalonia and southern France, around the vicinity of
modern Marseilles, the first iron objects date to the
eighth century BC, but precede the first appearance of
Phoenician trade goods. Scholars are divided over
whether these are Greek trade goods or whether they
reflect an in-situ development of iron technology.
A characteristic item of material culture has been
excavated from the earliest Phoenician metallurgical
sites in both Carthage and Iberia, and is well illustrated
in the chapters by Niemeyer and Rovira. This is a unique
type of fired clay tuyère, with rectangular cross-sec-
tion enclosing two parallel channels of circular sec-
tion, apparently designed for the simultaneous opera-
tion of two pairs of bellows. I will return to this item
below after reviewing the chapters on iron in sub-Sa-
haran Africa
The origins of iron working in sub-Saharan Africa
The special issue of Mediterranean Archaeology
(MA hereafter) contains eight chapters (not counting
summaries) on the origins of iron working in Africa,
and the UNESCO volume edited by Hamady Bocoum
(UN hereafter) has seven. There is some duplication;
the chapter by Alain Person and Quéchon in each vol-
ume is essentially identical, and that by Gérard
Quéchon differs only in the preface.
The chapters in MA attempt a survey of the evi-
dence for the spread of iron working throughout sub-
Saharan Africa. The coverage is not comprehensive,
as there is no chapter on the earliest iron in Eritrea,
Ethiopia or Somalia, though the data from crucial re-
gions are as yet too scarce to allow any firm conclu-
sions to be drawn. Marie Claude van Grunderbeek, Emile
Roche and Hugues Doutrelepont review the evidence
excavated in Rwanda and Burundi between 1960 and
1986.3 Their discussion of the earliest radiocarbon dates
3Those who do not read French will soon have the option
of van Grunderbeeks excellent Ph.D. thesis, to be
submitted in English to the University of Cape Town.
Journal of African Archaeology Vol. 2 (1), 2004 101
Book review essay: What do we know about African iron working?
from this region is more cautious than in their earlier
articles. They now reject the date of 3615 ± 205 BP for
the Muganza I furnace, and caution that the lack of
visible stratigraphy makes it difficult to be certain that
three other dates, ranging from ca. 3200 BP to ca. 2850
BP, really were associated with the iron slag and Urewe
ceramics excavated with them (p. 276). Two other dates
of 2815±165 BP and 2635±95 BP are in certain associa-
tion with smelting residues, but the huge standard de-
viations result in calibrated age ranges of 1450-500 cal
BC and 1000-400 cal BC respectively. Clearly neither of
these dates is of much use in determining when iron
working actually began in this region. The technology
employed in the earliest furnaces in Rwanda/Burundi 
truncated conical shafts of clay bricks erected over
a slag pit. This is the same as in the adjacent region of
Buhaya, Tanzania, where SCHMIDT (1997) suggests that
iron working may have begun as early as 600 cal BC,
though he acknowledges that this dating is far from
Even if the origins of iron working in this region
are still obscure, it should be stressed that in no other
part of sub-Saharan Africa has iron working in the first
millennium cal BC been as carefully studied, or so fully
published, as in Buhaya and Burundi. Unfortunately
there is almost nothing to which we can compare these
studies. Research on early iron working technology in
the upper Nile, Ethiopia, Eritrea or in the Arabian Pe-
ninsula has barely begun, so there is simply no point
as yet in debating whether iron working was independ-
ently developed in, or diffused into, the regions around
the Great Lakes.
Although Pierre de Maret (UN, p. 125) states that
that there is a consensus assez général that iron
working began throughout Central Africa (Cameroon,
Gabon, the Central African Republic and Rwanda) by
the late ninth century BC, some of those whom he cites
appear not to agree. I have already noted the reserva-
tions of van Grunderbeek and colleagues regarding the
earliest dates from Rwanda. Bernard Clists abstract
(MA, p. 269) places the start of iron working in
Cameroon at ca. 450 BC, and that in Congo (Brazzaville)
and Gabon at around 200 BC, while Richard Oslisly
(MA, pp. 263-268) favors a start in Cameroon at ca. 2600
BP, reaching Gabon around 2500/2400 BP and Congo
around 2200 BP. Clist, Oslisly and de Maret are draw-
ing on the same corpus of published radiocarbon dates,
so why do their estimates for the first iron working
differ so much? The differences reflect their individual
decisions about the presentation of radiocarbon dates.
Clist and Oslisly are both citing uncalibrated radiocar-
bon dates, and ignoring the associated standard er-
rors, while de Maret appears to be arbitrarily picking
the older end of the two-sigma probability range of
calibrated dates. A great deal of confusion would be
avoided if archaeologists in general, and Francophone
archaeologists in particular, would consistently use the
established conventions  BP (and only BP) for
uncalibrated radiocarbon ages, and cal BC and cal AD
(or cal BCE and cal CE, if preferred) for calibrated dates.4
De Maret would like the first use of iron in Central
Africa to be even older than this, and proposes (UN,
p. 125) that the sudden disappearance (disparition
brutale) of stone tools on open sites in Congo
(Brazzaville) and Cameroon after 3500 BP is attribut-
able to the adoption of iron. Forced to confront the
complete absence of evidence for iron or iron working
in these regions before the first millennium BC, he ar-
gues that the metal was imported (from where?) and
that recycling of iron and corrosion have left none for
the archaeologist to find! His hypothesis has the great
advantage of being completely untestable, but the form
of argument seems better suited to theology than to
Nothing useful is known about the beginning of
iron working in Congo (Kinshasa) or Angola, but
Duncan Miller (MA, pp. 229-234) picks up the trail again
in southern Africa. There is secure evidence of iron
working in eastern half of southern Africa dated be-
tween the second and sixth centuries cal AD, but the
earliest known evidence from the western half dates to
the sixth century cal AD. These may represent two
separate introductions of iron technology  the former
from Tanzania via Zambia and Zimbabwe, the latter from
Congo via Angola  but the evidence is as yet inad-
equate. Millers chapter draws upon a substantial quan-
tity of new data, much of it generated by himself in
more than fifty publications over the last twelve years.
He notes that there appears to be little change in the
techniques for smelting or forging iron over the 1800
years after their earliest occurrence, and offers some
preliminary thoughts on why these technologies were
so stable over time. He concludes by noting the diffi-
cult middle course that responsible scholars of Afri-
can technology must steer between those who see the
record as one of technological stagnation and those
4Archaeologists should also state which calibration chart,
or software package, was used to generate the calibra-
tion, and whether the calibrations are at one or two
Journal of African Archaeology Vol. 2 (1), 2004
D. Killick
who make wildly exaggerated claims for ancient Afri-
can achievements in science, mathematics and tech-
There is very little new data on early iron working
in West Africa in either volume, but the UNESCO vol-
ume does perform a valuable service by bringing to-
gether several important studies in an attractive pack-
age at a very reasonable price. This includes a French-
language summary of Edwin Okafors excellent 1992
Ph.D. thesis for Sheffield University on the develop-
ment of iron working around Nsukka, Nigeria. Unfortu-
nately Okafor has not, since his return to Nigeria, had
access to either the funding or to the scientific facili-
ties that would have enabled him to apply his training
to new research.
The most controversial chapters are those by
Gérard Quéchon (one written with Alain Person, one
without) on their discoveries in the Termit massif of
eastern Niger. As long ago as 1974 Quéchon and Jean-
Pierre Roset published a radiocarbon date (on char-
coal) of 2630 ± 120 BP (Dak-145) for one of a group of
22 very small slag-pit iron working furnaces at Do Dimmi
15, and a date of 2925 ± 120 BP (Dak-147) for charcoal
found with iron objects on a deflated surface site
(QUÉCHON & ROSET 1974). Although these dates were
noted with interest, claims for an independent inven-
tion of iron working in Niger (QUÉCHON & ROSET 1974;
DIOP 1976) were not widely accepted for lack of sup-
porting evidence. In the mid-1980s Danilo Grébénart
claimed to have discovered even earlier evidence for
metallurgy in Niger. His research west of Agadez in-
cluded the excavation of more than two dozen fired
clay structures, from which he obtained some thirty
radiocarbon dates. Using these dates, the morphology
of the structures, and partial chemical analyses of the
residues excavated from them, he defined four phases
of metallurgy in the Agadez region (GRÉBÉNART 1985:
Figures 275 and 276). These were Cuivre I (ca 4200 BP-
ca 3000 BP), Cuivre II (ca 2800 BP- ca 2000 BP), Fer I
(ca 2500 BP  ca 2000 BP) and Fer II (after ca 2000 BP).
Grébénarts claim that Cuivre I represented an in-
dependent invention of copper metallurgy in Niger was
almost immediately contradicted by petrographic stud-
ies of samples of supposed slags from Cuivre I struc-
tures, which showed that these were just soil partly
vitrified by intense heat (KILLICK et al. 1988). Our re-
sults appear to confirm the view of Suzanne BERNUS
(1983: 169) that the irregular furnaces from which
these samples came, and the charcoal for the radiocar-
bon samples, were actually the remains of burned-out
tree stumps and roots from savannah forest killed by
the dessication of the northern Sahel between 4000
and 3000 BP. I warned that unexpectedly early radio-
carbon dates for metallurgy in the Sahel may reflect the
use of old wood, or old charcoal, and suggested two
ways to detect, or avoid, old wood errors (KILLICK
1987; KILLICK et al. 1988). The first is to compare dates
on wood charcoal with thermoluminescence dates on
samples of fired clay from the same furnace; the sec-
ond is to avoid dating bulk wood charcoal wherever
possible by dating samples of annual plants instead.
The first recommendation has been ignored, but
Person and Quéchon, aided by Jean-Pierre Saliège of
the University of Paris radiocarbon laboratory, have
found a way to implement the second by extracting
charred annual plants (grass, seeds or chaff) from pot-
sherds collected from surface sites around Termit. In
their essentially identical chapters in the UA and MA
volumes they list 23 radiocarbon dates on annuals ex-
tracted from ceramics, along with eleven conventional
radiocarbon dates on charcoal. The ceramic temper
dates have very poor precision (90-300 radiocarbon
years at one sigma) but there is no reason to question
their accuracy.5 They range from ca 7000 BP to ca 2000
BP and, with a few exceptions, correspond to the four
very broad ceramic groups defined by Quéchon and
his colleagues. Objects of iron and copper have been
recovered on sites with pottery of type post-néo
phase 1, which yielded eight ceramic temper dates
ranging from 3300 ± 120 BP to 2430 ± 110 BP, or with
type post-néo phase 2, with two ceramic temper dates
of 2270 ± 90 BP and 2095 ± 200 BP.
All of the Termit pottery temper dates are from sur-
face scatters. In five instances it was possible to com-
pare dates on pottery temper with dates on wood char-
coal from the same surface scatter. The comparison is
presented as a table in both volumes (UN, p. 110; MA,
p. 251). Although Quéchon finds the results of the com-
parison très convaincants (MA, p.250), this is not at
all obvious from the table. In two cases (Gara Tchia Bo
176 and Termit Ouest 8B) the dates from bulk charcoal
and ceramic temper are within one sigma of each other,
though the standard deviations on the dates are large;
5These dates were obtained by decay-counting, not by
accelerator mass spectrometry (AMS), and the poor
and variable precision reflect the fact that only small
(and variable) amounts of carbon could be extracted
from the sherds. Much better precision could now be
obtained by dating organic temper by AMS, which re-
quires less than a milligram of carbon per sample.
Journal of African Archaeology Vol. 2 (1), 2004 103
Book review essay: What do we know about African iron working?
in a third (Gara Tchia Bo 20) the pottery temper date is
within two sigma of one charcoal date, but separated
by more than four sigma from the other; in the fourth
(Do Dimmi 16) the two dates are barely within three
sigma; and in the fifth case (Gara Tchia Bo 75) pottery
and charcoal dates are more than five sigma apart,
and each is at least two thousand radiocarbon years
younger than expected.
These dates are nevertheless acclaimed by sev-
eral of the contributors to the UNESCO volume as proof
that iron working was invented in sub-Saharan Africa.
The chapters by Person and Quéchon provide graphi-
cal comparisons of uncalibrated radiocarbon dates
older than 1000 BP that have been reported as associ-
ated with iron objects, or with evidence of iron work-
ing, in Niger, Nigeria, Cameroon, Sudan (Meroe) and
Rwanda/Burundi. This exercise shows that the oldest
radiocarbon dates are from Termit, and Person and
Quéchon seize the opportunity to insist that an inde-
pendent invention of iron working in Niger is proved.
Their claims are accepted without comment by de
Maret (UN, p.125) and are extended in the brief chapter
by Marie-Louise Maes-Diop that concludes the
UNESCO volume. Her table of the earliest appearance
of iron in various regions of the world (UN, pp. 189-
190) leads her to claim that iron working at Termit is at
least as old as any in the world. This is a pointless
exercise, as her table makes no attempt to distinguish
between iron forged from meteorites and iron smelted
from the ore. Both the preface and the back cover of
the UNESCO volume proclaim that lindustrie du fer
est un marqueur lourd de la civilisation, a sentiment
that Maes-Diop clearly endorses. But even if one ac-
cepts this statement (was the Incan empire, which
lacked iron, not a civilization?) it can only apply to the
mastery of iron smelting, for the forging of iron-nickel
meteorites requires no advanced technological skills.
Historians of metallurgy are always careful to distin-
guish between meteoritic iron and smelted iron, but
Maes-Diop does not appear to be aware of the impor-
tance of this distinction. Nor, it appears, has Quéchon
even considered the possibility that iron objects from
Termit might not be smelted iron, for there is no men-
tion in these chapters of any metallographic or chemi-
cal study of these artefacts.
The two chapters by Hamady Bocoum (MA, pp.
235-245; UN, pp. 93-103) also accept, and try to rein-
force, the claims of Person and Quéchon. Both briefly
review the early history (to 1980) of the debate be-
tween those who have concluded that iron metallurgy
diffused to sub-Saharan West Africa from Phoenician
North Africa (Mauny, van der Merwe) and those who
have argued for independent invention (Lhote, Diop)6,
but this review is merely background to Bocoums main
concerns. These are to refute the two major objections
that have been raised about claims for independent
invention of iron smelting in West Africa. His first tar-
get is my suggestion that that some radiocarbon dates
from Niger may be misleading because early metallur-
gists may have used old wood or old charcoal for fuel.
Bocoum argues that since the available radiocarbon
dates from ceramic temper and from charcoal on sur-
face sites around Termit and Egaro agree, there is no
old wood problem in Niger, and thus the chronology
of metallurgy proposed for the Agadez region by
Grébénart should be reinstated (UN, p. 96; MA, pp.
239).7 If there is no old wood problem, he argues,
then the best way to determine when iron working be-
gan in Africa is by statistical examination of the whole
corpus of radiocarbon dates associated with iron ob-
jects or evidence of iron working (UN, p. 99).
Bocoums second aim is to explain how the smelt-
ing of iron could have developed in Niger within a
neolithic context. Person and Quéchon (whose train-
ing, like that of Grébénart, is in paleolithic archaeol-
ogy) offer no explanation at all, but Bocoum has suffi-
cient knowledge of archaeometallurgy to understand
that some plausible explanation must be offered. His
solution is to reinstate Grébénarts Cuivre 1, and with
it Grébénarts proposal that copper smelting provides
a technological bridge in Niger between the open fir-
ing of ceramics and the reduction of iron ores using
furnaces and forced-air blast (MA, pp. 241-242; UN,
pp. 99-100).
Evaluating claims for an independent invention of
metallurgy in sub-Saharan Africa
Since van Grunderbeek and collaborators are now
reluctant to accept the earliest dates for iron working
in Rwanda and Burundi, the argument for an independ-
ent invention of iron working in sub-Saharan Africa
6The chapter by Joseph Fazing Jemkur in the UNESCO
volume provides a fuller history of this debate, but does
not register any of the publications on iron working at
Carthage that appeared during the 1990s.
7Bocoum does however accept my argument that some
of the supposed furnaces in Grébénarts Cuivre 1 are
in fact burned-out tree stumps (UN, p. 95, footnote 2).
Journal of African Archaeology Vol. 2 (1), 2004
D. Killick
hinges upon the evidence from Niger. How conclusive
is this evidence? It should be noted that the radiocar-
bon dates for Termit were originally published a dec-
ade ago (PARIS et al. 1992), and that no new data has
been presented since then. The set of chapters argu-
ing for independent invention in Niger should there-
fore be seen as a renewed effort to convince those who
were not persuaded the first time around.
The major objection to the claims for Termit is still
the lack of proof that the radiocarbon samples are truly
associated with iron and copper samples. Unfortunately
all of the Termit and Egaro sites except the smelting
site Do Dimmi 15 have been deflated by the action of
wind on bare soil, so that the sherds, charcoal and
metal samples now lie together on the ground surface.
Quéchon argues that each of these sites was originally
a single component (UN, p. 113), but his more detailed
arguments for this position are still not persuasive. I
have already above noted that the agreement between
radiocarbon dates on both charcoal and ceramic tem-
per from the same scatters (UN, p. 110; MA, p. 251) is
not as close as Quéchon states; in only two of six cases
are all the dates from a site within two standard devia-
tions of each other. No site has more than one date on
pottery temper or more than three dates in total.
Quéchons second argument for treating these
surface scatters as single component sites is that the
metal objects are always found with the same range of
pottery types. The scarcity of illustrations of pottery
from these sites makes it impossible to check this claim,
but even if true it does not prove that these are single
component sites. There is another plausible explana-
tion. By about 4000 BP the Termit region was very arid,
and there would have been few sources of water. These
would have been the locations at which broken pot-
tery would have accumulated, and would have been
visited repeatedly over time, but the pottery broken on
these visits has since been scrambled together by de-
flation. This would explain why the paired pottery and
charcoal dates are in agreement at some sites but not
at others.
It is unfortunately impossible to distinguish be-
tween these possibilities from the available data, but
there is no necessary association between the metal
objects, the dated sherds and the charcoal samples in
surface scatters.8 Metal objects can only be conclu-
sively dated by radiocarbon when: (1) they are found
with short-lived carbon samples (annual plants, or or-
ganic temper in pottery) in undisturbed stratified con-
texts, none of which have yet been discovered at Termit
or Egaro; or (2) in the case of iron objects, when they
are directly dated by AMS radiocarbon dating. The
second option would only be feasible if these samples
still contain small patches of steel, but has been suc-
cessfully performed in much more humid environments
than that at Termit (e.g. KUSIMBA et al. 1994). It would
be worth attempting on the samples from Termit.9
There are other problems with claims for independ-
ent invention of metallurgy in Niger by 3300 BP. Let us
suppose that the earliest iron objects recovered at
Termit and Egaro are indeed of this age. If the objects
were smelted from ore, as Quéchon, Person and their
supporters assume, where is the evidence for their pro-
duction? Iron slag and fired furnace ceramics are virtu-
ally indestructible, and would be difficult to miss in
this barren and deflated landscape. Yet the only evi-
dence of smelting or forging recovered at Termit or
Egaro is a crescent-shaped array of 22 very closely
spaced furnace bottoms at Do Dimmi 15 (UN, Photo 73;
MA, Plate VIII:2) from which dates of 2630 ± 120 BP,
2500 ± 70 BP and 2065 ± 60 BP were obtained.10 And if
iron was really smelted at Termit by 3300 BP, why is
there no radiocarbon evidence for iron or iron working
before 2500 BP in the many excavated sites to the south
and west of Agadez (GRÉBÉNART 1985: 263-348)? These
sites are only 300-450 km west of Termit.
Bocoums argument for even earlier copper metal-
lurgy in the Agadez region is equally problematic. He
asserts that the ceramic temper dates from Termit show
that the radiocarbon date (on charcoal) of 3650 ± 50 BP
from Afunfun 175 Furnace 8 is not an old-wood date,
and thus that iron technology in Niger developed out
of a prior copper working technology. But the dates
from Termit cannot prove that any date, or set of dates,
from the Agadez region is acceptable or unacceptable;
8A recent review of the archaeology of Niger reaches a
similar conclusion (HAOUR 2003: 217), as does a exten-
sive analysis by Susan MCINTOSH (in press) of the claims
for early metallurgy at Termit.
9COOK et al. (2003) have shown that even steel that
appears to be totally corroded may yield reliable radio-
carbon dates, as the cementite (iron carbide) phase of
steel is much more resistant to corrosion than the low-
carbon (ferritic) phase.
10 The spacing and similarity of the furnace bases suggests
to me that they the product of a single group of metal-
workers and were produced in a short span of time,
perhaps a single year. It is possible that the older dates
reflect some old wood effect, so it is unfortunate that
thermoluminescence dating was not attempted.
Journal of African Archaeology Vol. 2 (1), 2004 105
Book review essay: What do we know about African iron working?
each piece of evidence must be evaluated on its own
merits. Although fourteen of the eighteen fired clay
structures excavated at Afunfun 175 appear to be
burned-out tree stumps or fallen trunks (KILLICK et al.
1988)  an interpretation that Bocoum accepts - Fur-
nace 8 is certainly an artificial furnace or firepit, as are
Furnace 3, which gave a date on charcoal of 3680 ±50
BP, and the undated Furnaces 6 and 18 (GRÉBÉNART
1985: 113-160). The reported chemical compositions
of the slags excavated from these structures
(GRÉBÉNART 1985: 158-160; KILLICK et al. 1988) show
that almost all are baked soil, but two samples (one
each from Furnaces 6 and 8) contain more than 1%
copper and are certainly evidence of some form of cop-
per metallurgy.
So far so good, but when one broadens the focus
from Furnace 3 to the site as a whole, the picture be-
gins to blur. The radiocarbon dates on Furnaces 3 and
8 are both within one standard deviation of those on
adjacent furnaces (2, 4, 5, 7 and 12) that are clearly
fired earth casts around former tree stumps or trunks.
There is also evidence of later metallurgical activity
on this site. Furnace 6 yielded a nodule of tin bronze
(15% Sn), while Furnace 1 (almost certainly a burned-
out tree base, with three radiocarbon dates between
4140 ± 90 BP and 3680 ± 60 BP) contained several pieces
of iron slag, one of which was directly dated to 1500 BP
from an entrapped piece of charcoal (KILLICK et al.
With this additional information in hand, the most
likely site history for Afunfun 175 is as follows:
(1) A small forest once grew on this site, but burned
at some time after 3100 ± 70 BP (the youngest
radiocarbon date obtained by Grébénart on a fired
clay structure). Combustion of some bases and
fallen trunks was incomplete, leaving variable
amounts of charcoal within the fired clay exterior
casts of the former trees.
(2) Later in time, metal-using groups visited the site
and constructed firepits and furnaces to work
copper and iron. At least some of these activities
made use of charcoal from the burned-out trees.
One of these visits took place around 1500 BP,
during which one of the burned-out stumps
served as a dump for some iron slag. The timing
of the other visits cannot be established by
radiocarbon dating, as the radiocarbon age of the
samples is that of the trees that furnished the
charcoal - not the age of the furnace or firepit.11
In spite of the confident assertions of Quéchon,
Person, and Bocoum, there is therefore no proof that
iron was independently invented in Niger, or anywhere
else in sub-Saharan West Africa. It is equally true there
is no proof that iron smelting was introduced from any-
where else. We still lack conclusive evidence for either
position. There are however a couple of lessons that
we can learn from this review. One is that attempts to
infer the place and time of the earliest metalworking by
graphical or statistical manipulation of all available ra-
diocarbon dates for metalworking are absolutely use-
less. Nothing is gained by analysis of data of dubious
quality. Some archaeologists in Europe are developing
numerical systems for ranking radiocarbon dates, us-
ing such criteria as reliability of association, sample
material and precision (PETTIT et al. 2003). African ar-
chaeologists would be wise to follow this lead, and to
use only those dates of the highest rank in arguments
about origins. None of the earliest dates that we cur-
rently possess for metallurgy from Rwanda/Burundi or
Niger would qualify for high ranking.
But even when we do find samples of carbonized
annual plants in indisputable stratigraphic association
with the metals or metalworking debris, we will not be
able to avoid the flattening of the radiocarbon calibra-
tion curve in the mid-first millennium BC. Table 1 and
Figure 1 both show the calibrated age ranges at two
sigma (95.4% probability of inclusion) for dates of fairly
good precision from 2200 BP to 2800 BP. Note that from
2300 BP through 2600 BP the calibrated age range is
never less than 300 calendar years, and may be as much
as 500 years. Almost all of the earliest radiocarbon dates
for the earliest iron metallurgy in Rwanda/Burundi,
Buhaya, Congo, Central African Republic, Gabon and
Cameroon fall in this range, and many of the reported
dates older than 2600 BP extend throughout this inter-
val after calibration because they have large standard
deviations (>100 radiocarbon years).12
The inescapable conclusion is that we will never
be able to trace the spread of iron working through
West, Central and north-eastern Africa by radiocarbon
dating. Unfortunately the only alternative methods -
11 The ages of these furnaces and firepits could be established
directly by thermoluminescence dating of fired clay, but
this technique was not employed by either Grébénart or
Quéchon in their respective research projects.
12 For example, the much cited date from Do Dimmi 15 in
Termit of 2630 ± 120 BP calibrates at two sigma to a
calendar range from 1050 cal BC to 400 cal BC.
Journal of African Archaeology Vol. 2 (1), 2004
D. Killick
2200±50 BP 390-110 cal BC
2250±50 BP 400-190 cal BC
2300±50 BP 510-200 cal BC
2350±50 BP 800-350 cal BC (88.0%), 300-200 cal BC (7.4%)
2400±50 BP 770-610 cal BC (28.6%), 600-390 cal BC (66.8%)
2450±50 BP 770-400 cal BC
2500±50 BP 800-480 cal BC (87.7%), 470-410 cal BC (7.7%)
2550±50 BP 820-510 cal BC (94.4%), 440-410 cal BC (1.0%)
2600±50 BP 900-870 cal BC (1.8%), 840-750 cal BC (68.4%), 720-540 cal BC (32.2%)
2650±50 BP 920-760 cal BC
2700±50 BP 980-950 cal BC (2.4%), 940-790 BC (93.0%)
2750±50 BP 1000-810 cal BC
2800±50 BP 1130-820 cal BC
Tab. 1. Hypothetical radiocarbon dates from 2200±50 BP to 2800±50 BP, calibrated at two standard deviations (95.4%) with
OXCAL 3.9 (BRONK-RAMSAY 2003). Where the calibrated ranges are discontinuous, the probability for each portion is listed,
with the probability of all portions summing to 95.4%.
Atmospheric data from S tuiver et al. (1998); OxCal v3.8 Bronk Ramsey (2002); cub r:4 sd:12 prob usp[chron]
Ca librat e d date
Fig. 1. text
Journal of African Archaeology Vol. 2 (1), 2004 107
Book review essay: What do we know about African iron working?
thermoluminescence (TL) and optically-stimulated lu-
minescence (OSL)  are at present no more precise than
radiocarbon in this time range (GODFREY-SMITH & CASEY
Does this mean that we can never know whether
the knowledge of iron working was independently de-
veloped in sub-Saharan Africa, or whether it spread
from elsewhere? This is not necessarily the case. We
could still find evidence of iron in excellent stratigraphic
association with short-lived radiocarbon samples, pre-
cisely dated to before 2600 BP, though at present there
is not a single case that satisfies all these conditions.
But even if iron working did spread through the north-
ern half of sub-Saharan Africa during the radiocarbon
black hole between 2300 BP and 2600 BP, we may still
be able to track its spread if we can make use of infor-
mation other than radiocarbon dates. Even if we can
never know the precise absolute ages of sites in this
interval, we may still be able to infer relative ages from
the associated pottery. This of course will require that
detailed sequences of ceramic styles be established in
each area by excavation and patient analysis of pot-
tery from deeply stratified sites, as Graham Connah did
at Daima, as Susan McIntosh has done in the middle
Niger and middle Senegal River valleys, and as the
Frankfurt team are doing in Burkina Faso and northern
Nigeria. Francophone archaeologists in sub-Saharan
Africa have shown little interest in constructing ce-
ramic sequences, but they will have to do so from now
on if they wish to understand the archaeology of the
first millennium BC.
We may also be able draw some inferences from
the comparative study of iron smelting technology. As
noted above, there is now firm proof of Phoenician
iron working at Carthage and in the Iberian Peninsula
by at least the eighth century BC. We do not yet know
whether the seventh/sixth century Phoenician smelt-
ing furnace from Toscanos, Spain (illustrated by
Niemeyer in MA, p.87, Figure 3) is typical, but it is
clearly very different from the oldest known iron smelt-
ing technology in sub-Saharan Africa. Almost all pub-
lished iron smelting furnaces of the first millennium cal
BC from Rwanda/Burundi, Buhaya, Nigeria, Niger,
Cameroon, Congo, Central African Republic and Ga-
bon are slag-pit furnaces, which are so far unknown
from this or earlier periods in the Middle East or North
Africa. Early Phoenician tuyères, which have square
profiles enclosing two parallel (early) or converging
(later) narrow bores are also quite unlike those de-
scribed for early sites in sub-Saharan Africa, which are
cylindrical with a single and larger bore. While this
comparison is hardly conclusive, it does suggest that
the earliest Phoenican technology was probably not
the model for the earliest iron working in sub-Saharan
West Africa. We cannot yet make comparisons with
other potential source areas for lack of data. These
include the Arabian Peninsula, pre-Axumite Eritrea,
Ethiopia, and Greek colonies in North Africa. Nor is
anything known about early iron working technology
in the southern Sudan, which may yet prove to be ear-
lier than that in Nubia.
Later prehistory, history and ethnography of
African iron working
The UNESCO volume also includes several regional
reviews of research on later iron working in West Af-
rica, all in French. David Aremus chapter provides an
inventory of known iron working sites in Nigeria, fol-
lowed by a synthesis of historic iron working proce-
dures derived from his own oral histories, and a sum-
mary of the rituals dedicated to the deity Ogun (god of
fire and of iron) in Yoruba-speaking regions. This is
complemented by a short chapter by Isaac Adeagbo
Akinjogbin on the recent history of Yoruba iron work-
ing. Joseph-Marie Essombas chapter, reprinted from
an earlier work, summarizes archaeological and oral his-
torical studies of iron working in central Cameroon.
Those who have dismissed the possibility of an old
wood offset for radiocarbon dating in the more humid
areas of Africa (e.g. DE MARET & THIRY 1996) should
examine the twelve dates obtained by Essomba on wood
charcoal from a single furnace at Oliga (UN, pp. 137-
138). These range from 1860 ± 70 BP to 2820 ± 70 BP!
Some of the richest and most insightful studies of
recent African iron working have been written by the
anthropologist Bruno Martinelli, who has for more than
thirty years been blending ethnography, oral history
and archaeology to recreate the social and economic
roles of iron in west African societies. His chapter in
the UNESCO volume (pp. 165-188) is a superb study of
specialization in iron working within the Yatenga king-
dom, which extended from the Bandiagara escarpment
of Mali, east of Mopti, into the plains of modern Burkina
Faso. Martinelli notes the emergence, probably in the
13 Recent developments in OSL dating promise much better
precision in the near future, but will require in-situ
measurement of radioactive dose rates for each sample
by gamma-ray spectrometry (Jean-Luc Schwenninger,
personal communication, February 2004).
2200±50 BP 390-110 cal BC
2250±50 BP 400-190 cal BC
2300±50 BP 510-200 cal BC
2350±50 BP 800-350 cal BC (88.0%), 300-200 cal BC (7.4%)
2400±50 BP 770-610 cal BC (28.6%), 600-390 cal BC (66.8%)
2450±50 BP 770-400 cal BC
2500±50 BP 800-480 cal BC (87.7%), 470-410 cal BC (7.7%)
2550±50 BP 820-510 cal BC (94.4%), 440-410 cal BC (1.0%)
2600±50 BP 900-870 cal BC (1.8%), 840-750 cal BC (68.4%), 720-540 cal BC (32.2%)
2650±50 BP 920-760 cal BC
2700±50 BP 980-950 cal BC (2.4%), 940-790 BC (93.0%)
2750±50 BP 1000-810 cal BC
2800±50 BP 1130-820 cal BC
Journal of African Archaeology Vol. 2 (1), 2004
D. Killick
18th century, of an endogamous occupational caste in
the center of Yatenga dedicated to full-time specializa-
tion in the production of steel, which their kin traded to
the neighbouring Dogon, and as far away as modern
These steel bars were, Martinelli asserts, welded
by blacksmiths in these various regions to their own
locally-forged iron, forming composite tools with steel
cutting edges (tranchants). To make this steel, the
Yatenga ironworkers developed the tallest furnaces
known in Africa, varying from five to six and a half
meters in height. These were charged with about 2000
kg of charcoal and ore, and smelted very slowly by
natural draft to produced 150-200 kg of bloom per
charge over a period of five to seven days. Former iron
workers built a furnace of this type for Martinelli and
smelted two charges, one for three days and the other
for four, each of which is fully documented. It has been
already been established that the long slow combus-
tion in natural draft furnaces favors the production of
steel blooms (KILLICK 1991), but this is the first docu-
mentation of a specialist industry in sub-Saharan Af-
rica dedicated to the production of steel. Unfortunately
Martinellis fine field work does not appear to have
been followed by full metallurgical study of the smelt-
ing process and products. If blacksmiths throughout
the region were indeed routinely welding steel edges
to iron tools, this would be the first documented case
of this technique in sub-Saharan Africa. This needs to
be confirmed by metallographic analysis of tools from
archaeological and historic contexts.
Martinellis chapter is nicely complemented by a
reconstruction of smelting by natural draft among the
neighboring Dogon. Eric Huysecom provides a sum-
mary in MA (pp. 73-82), with colour plates, and has
also made an excellent film (HUYSECOM & AGOSTINI 1996)
that shows the whole process from the digging of a
17.5 m shaft to extract the ore, the rebuilding of a ruined
furnace, and the smelting process itself. The natural
draft furnace was about 2.3m high and consumed a
charge of about 460 kg of charcoal14 and ore in 40 hours
to produce a bloom of about 70 kg. Temperatures were
measured, but it appears that the equipment was not
working correctly, as the peak temperature reported
when slag was flowing (1021°C) is below the melting
point of any possible mixture of iron oxide, silica and
alumina!15 Metallographic analysis of the products
showed, not surprisingly, that the bloom contained a
generous proportion of steel. Evidently, the Yatenga
specialists did not have a monopoly on steel produc-
tion in this region, as Martinelli seems to imply, but
they do seem to have manufactured it on an unusually
large scale.
Most of the papers reviewed here are the product
of Francophone scholarship. The special issue of Medi-
terranean Archaeology does however include three
reflections by senior Anglophone scholars, each of
whom has made major contributions to our current un-
derstanding of African metallurgy in all its aspects 
technological, economic, social and conceptual. In her
chapter, the historian Eugenia Herbert argues that ...the
preoccupation with dates and origins obscures ques-
tions of greater historical interest (p. 42). As exam-
ples of more interesting work, she focuses upon two
recent studies  Métallurgie et politique en Afrique
centrale (DUPRÉ & PINÇON 1997) and Iron Technology
in East Africa (SCHMIDT 1997). Both are given a sympa-
thetic but critical reading, concentrating on whether
the evidence can bear the weight of the ambitious so-
cial and political models employed in each case. While
she salutes these scholars for asking the right ques-
tions, she laments that (p)erhaps it will never be pos-
sible to write a history of African metallurgy that satis-
fies the historians inordinate greed for both generali-
zation and specificity (p. 48). Perhaps not, but the
same can be said for almost any work of precolonial
African history. Is Jan Vansinas Paths in the Rainfor-
est really richer in evidence than the work of anthro-
pologists like Martinelli, Schmidt or Nicholas David on
African metallurgy? I think that the reverse may be
true. Each of these anthropologists employs all of the
evidence and the techniques of the historian of pre-
colonial Africa (written documents, oral histories, struc-
tural analysis, linguistic evidence), but they are also
able to draw upon archaeological and scientific evi-
dence, which very few African historians are compe-
tent to evaluate.
14 The video shows that this charcoal was produced by
carbonizing an entire standing dead Prosopis africana,
with a basal diameter of more than one meter. If sam-
ples of the charcoal have been retained, it would be most
instructive to run a series of radiocarbon dates on these,
and to publish the distribution of dates obtained.
15 OSBORN & MUAN (1960, plates 6 and 9). The actual
temperatures were probably similar to those obtained by
Martinelli (UN, p. 182). His graph shows temperatures
above 1400°C at tuyère level while slag was flowing
from the furnace.
Journal of African Archaeology Vol. 2 (1), 2004 109
Book review essay: What do we know about African iron working?
Nicholas Davids long chapter revisits many ear-
lier studies of African iron working in search of an ap-
propriate balance between what he labels naturalist
and antinaturalist approaches. Broadly speaking, the
former have been more interested in documenting the
technology and economics of African iron working,
while the latter have concerned themselves more with
symbolism and ideology  with what the words and
gestures of iron-workers may reveal about their con-
cepts of social order and disorder. None of the works
that he surveys, including my own, have in his opinion
managed to get the balance between these approaches
quite right, though he suggests that those of Schmidt,
Avery and Childs in Buhaya come closest. (He could
also have mentioned his own, but modesty prevailed).
How should scholars of African iron working reconcile
their naturalist and antinaturalist tendencies? David
suggests (as Alain Gallay does, quite independently,
in his chapter) that the conceptual framework of the
French Anthropologie de Techniques school is per-
haps the best available at present  a proposal with
which I am in full agreement.
Peter Schmidts contribution to MA is aimed as
much at historians of Europe and of the Middle East as
at Africanists. He argues that the European passion
for classification has distorted our understanding of
the history of technology in Africa. All African iron work-
ing processes have been lumped together as variants
of the bloomery process, thus assigning all African
smelting processes to a rung near the base of a global
evolutionary ladder of metallurgical progress. In fact,
as Schmidt notes, students of African metallurgy have
documented an amazing variety of processes, many
with no known counterparts on other continents  a
spectrum of variation of such diversity as to suggest
that the term bloomery no longer does justice to
the range of evidence (p. 220). Earlier writers, blinded
by their urge to classify, managed to miss most of the
evidence for innovation in African metallurgical tech-
nology. This was an error that Schmidt himself has done
much to correct. His is a profound critique, and one
with which a growing number of archaeometallurgists
worldwide would tend to agree. Schmidt goes on to
argue that continuous innovation was the normal state
in African iron smelting, with each iron worker impro-
vising off a preexisting repertoire of techniques  much,
I suppose, like a jazz musician improvising off a stand-
ard melody. Im not sure that I agree with this latter
point; iron working can succeed only within a very
narrow window of temperature and gas composition,
which tends to impose rather strict limits upon indi-
vidual departures from a successful process.
We owe much of our current knowledge of the
variety of African iron smelting techniques to interdis-
ciplinary collaboration, in both the field and the labora-
tory, between anthropologists or archaeologists on the
one hand, and metallurgists or geologists on the other.
Chemical or metallographic analyses of African iron
and iron smelting residues had been undertaken inter-
mittently since 1904, but the analysts had, in most
cases, no appreciation of the cultural context of the
samples. They produced technical reports that the ar-
chaeologists or anthropologists could not understand,
and thus were usually reproduced verbatim as appen-
dices. Collaboration in the field, whether in ethno-
graphic or archaeological context, was a crucial step
forward. When these first began, from the late 1960s
(Nikolaas van der Merwe with Revil Mason at
Phalaborwa, Ron Tylecote with Bernard Fagg at Nok,
Don Avery with Peter Schmidt in Buhaya) there was
much mutual misunderstanding of aims and methods.
But each learned from the other, and the result was the
birth of archaeometallurgy as an area of specialization.16
There are too many areas of specialization in ar-
chaeology for any one person to master them all, so
specialists owe generalists an explanation, in non-tech-
nical language, of what they know and how they know
it. Each of these volumes contains an exceptionally
clear and informative essay on the archaeometallurgy
of iron working. Any archaeologist who has ever won-
dered how to distinguish smelting slags from forging
slags, or wrought iron from steel, should obtain a copy
of Philippe Fluzins excellent chapter in the UNESCO
volume. This is illustrated with about thirty color pho-
tographs and is by itself worth the cost of the book. I
hope that the English translation of this volume, prom-
ised for 2002, will eventually appear so that Anglophone
archaeologists can also have the benefit of Fluzins
experience. The Mediterranean Archaeology volume
contains a chapter, in French, co-authored by Fluzin
with the Swiss archaeometallurgist Vincent Serneels.
This concentrates more upon the interpretation of slags
than does the UN chapter, reflecting Serneels special
expertise in this area, but unfortunately is much less
useful to archaeologists because it provides only a
single photographic plate.
16 I do not mean to imply that archaeometallurgy devel-
oped solely in Africa; similar developments were taking
place at the same time in Europe, the Soviet bloc, the
Near East and South America.
Journal of African Archaeology Vol. 2 (1), 2004
D. Killick
The present state and future prospects of research
on African iron working
The two volumes reviewed here complement each
other to a substantial extent, and together provide a
very useful snapshot of the present state of research
of African iron working. The volume edited by Bocoum
will undoubtedly be more widely read in Africa because
it is distributed by UNESCO, is attractively produced
with abundant illustrations (many in color) and is inex-
pensive. The less accessible special issue of Mediter-
ranean Archaeology is however the more useful for
African archaeologists because it covers almost all of
the continent, includes much comparative material from
beyond Africa, and contains the valuable overviews
by Gallay, Herbert, David and Schmidt. The UNESCO
volume suffers in comparison because many of its chap-
ters are recycled from previous publications or were
written in the mid-1990s, so some of the information
given  particularly that relating to the Phoenicians - is
out of date. The bibliography in the UNESCO volume
is also unreliable, with many incomplete or otherwise
inaccurate citations, particularly for works originally
published in English.
The central theme of all of the products directly
created by the UNESCO project Les Routes du Fer en
Afrique  the book edited by Bocoum, the Paris exhi-
bition and the web site  is that it there is definite proof
that iron working was independently invented in West
Africa by 3300 BP. I have argued here that the support-
ing evidence falls far short of proof. These claims rest
entirely upon interpretation of radiocarbon dates, and
in all cases there are potential problems with the radio-
carbon evidence. In some there may be an old wood
problem and in others there is no proof that the metal
objects are of the same age as the dated pottery sherds
that lay with them on deflated surfaces. Furthermore,
some of the Francophone archaeologists of Africa who
have contributed chapters to the UN and MA volumes
do not appear to fully understand how to interpret cali-
brated radiocarbon dates. Calibrated dates are ranges
of calendar ages, and there is absolutely no scientific
justification for selecting the older end of the range as
the real date.
In summary, I have suggested that the quality of
the evidence is so poor that we cannot yet establish
when iron working began in sub-Saharan Africa. Sta-
tistical manipulation of the current corpus of radiocar-
bon dates for the earliest metallurgy is absolutely use-
less  as statisticians are fond of saying, garbage in
means garbage out. We must accept the limitations of
the evidence, and keep searching until we find short-
lived carbon samples in undeniable stratigraphic asso-
ciation with certified iron working debris. Even then
we must respect the inherent limitations of radiocar-
bon calibration. Some portions of the calibration curve
yield narrower intervals of calendar age, some yield
wider ranges, and if our radiocarbon dates fall between
2350 and 2600 BP, we must simply accept that we can
do no better than a 300-year range of calendar age.
Meanwhile there are more interesting issues to be
investigated. The publication of these volumes effec-
tively marks the end of the era of ethnographic and
oral historical studies of African iron working, since
almost all of the potential informants are now dead.17
Studies of the symbolic and conceptual frameworks of
African iron working, which have been such a promi-
nent feature of the literature of the last thirty years, will
not completely disappear, but will be largely restricted
to reworking the case study material that has already
been collected. From this point on almost all studies of
African iron working will be archaeological.
The main thrust of future archaeological work
should, I suggest, be directed towards reconstructing
economic aspects of past African iron working. For
any given region, like Meroe or the Senegal River val-
ley, we need to reconstruct how much iron was pro-
duced, by what technical processes, over what period
of time, how production was organized, where the prod-
uct went, and how it was consumed. These are the
answers that will be of most value in the long run to
African archaeology and African history. Work on the
origins and spread of African iron working will and
should continue, but is likely to be limited by our in-
ability to date sites with useful precision in some por-
tions of the radiocarbon calibration curve.
All of these tasks require that archaeologists be
able to work closely with specialists in dating tech-
niques and in archaeometallurgy. A major failure of the
UNESCO project is that it appears to have invested
little effort in identifying priorities for future research,
or in assessing the resources and technical facilities
that would be required. The facts of the matter are that
very few archaeologists who live in Africa have access
17 The only region of the African continent where iron is
still being smelted by indigenous techniques is in the
very remote Dime territories west of the Omo River in
Ethiopia (R. Haaland, personal communication).
Journal of African Archaeology Vol. 2 (1), 2004 111
Book review essay: What do we know about African iron working?
to the technical assistance that they would need to
undertake research on either the origins or the eco-
nomics of iron working, and that the situation is dete-
riorating. The reopening of the Dakar laboratory has
doubled the number of radiocarbon facilities in sub-
Saharan Africa, but otherwise the picture is bleak. Sev-
eral African students have done postgraduate theses
in archaeometallurgy at European universities, but have
been unable to make use of their training after return-
ing home for lack of access to funding and technical
facilities. The only specialist archaeometallurgy labo-
ratory in sub-Saharan Africa, at the University of Cape
Town, has been crippled in a thoughtless act of aca-
demic sabotage.18 In retrospect, UNESCOs money
would have been better spent in Africa, upon assess-
ment of research infrastructure, than in funding an ex-
pensive meeting and exhibition in Paris.
Nor is the picture much brighter overseas. Many
of the leading scholars in this field have retired, or are
close to retirement, and there are few students in train-
ing who could replace them. Since most research on
African iron working will, for the foreseeable future, be
conducted with foreign funding, the scarcity of train-
ees in Europe and America should be a cause for con-
cern. The years between 1965 and 2000 will be seen in
retrospect as the Golden Age for studies of African
iron working, and many of the accomplishments of this
era are cited in the volumes reviewed here. What will
the next thirty-five years bring? From the present van-
tage point, I fear that I identify with the pessimism of
the ancient Greek writer Hesiod (8th or 7th century BC),
who, in looking back to the glorious past ages of gold,
silver and bronze, lamented that he himself should have
to live in the age of iron.
This review has been greatly improved by the criti-
cal comments of Stanley Alpern, Susan McIntosh and
Duncan Miller on an earlier draft.
18 Throughout the 1990s the director, Duncan Miller, raised
all salary and research funding for the laboratory by
submitting proposals directly to private foundations.
The University of Cape Town ruled in 2000 that all
proposals should be submitted by the University, not by
individuals, and has since 2002 declined to submit any
proposals on Dr. Millers behalf.
Bernus, S. 1983. Découvertes, hypothèses reconstitution et
preuves: le cuivre médiéval dAzelik-Takkeda. In: Echard,
N. (ed.), Métallurgies Africaines: Nouvelles Contributions.
Mémoires de la Société des Africanistes No. 9, Paris, pp.
Bronk-Ramsay, C. 2003. Oxcal 3.9. (
Cook, A.C., Wadsworth, J., Southon, J.R. & van der Merwe, J.R.
2003. AMS radiocarbon dating of rusty iron. Journal of
Archaeological Science 30, 95-101.
De Maret, P. & Thiry, G. 1996. How old is the Iron Age in
Central Africa? In: Schmidt, P. (ed.), The Culture and
Technology of African Iron Production. University of
Florida Press, Gainesville, pp. 29-39.
Diop, Cheikh Anta 1976. Lusage du fer en Afrique. Nyame
Akuma 53, 93-95.
Dupré, M.-C. & Pinçon, B. 1997. Métallurgie et Politique en
Afrique: Deux Mille Ans sur les Plateaux Bateké.
Harmattan, Paris.
Godfrey-Smith, D.I. & Casey, J.L. 2003. Thermoluminescence
chronology for Early Iron Age smelting technology on
the Gambaga Escarpment, Ghana. Journal of
Archaeological Science 30, 1037-1050.
Grébénart, D. 1985. La Region dIn Gall-Tegidda n Tesemt
(Niger), Programme Archéologique dUrgence 1977-
1981. II: Le Néolithique Final et les Débuts de la
Métallurgie. Études Nigeriennes, No. 49. Institut de
Récherches en Sciences Humaines, Niamey.
Haour, A. 2003. One hundred years of archaeology in Niger.
Journal of World Prehistory 17, 181-234.
Huysecom, E. & Agustoni, B. 1996. Inagina, lUltime Maison
du Fer. (Video). Also issued (1997) as Inagina, the Last
House of Iron. University of Geneva, Department of
Anthropology (
Killick, D. 1987. On the dating of African metallurgical sites.
Nyame Akuma 28, 29-30.
Killick, D. 1991. A little known extractive process: iron smelting
in natural-draft furnaces. JOM (Journal of the Minerals,
Metals and Materials Society) 43 (4), 62-64.
Killick, D., van der Merwe, N.J., Gordon, R.B. & Grébénart, D.
1988. Reassessment of the evidence for early metallurgy
in Niger, West Africa. Journal of Archaeological Science
15, 367-394.
Kusimba, C.M., Killick, D. & Cresswell, R.G. 1994. Indigenous
and imported metals in Swahili sites on the Kenyan coast.
In: Childs, S.T. (ed.), Society, Culture and Technology in
Africa. MASCA Research Papers in Science and
Archaeology, Supplement to vol. 11. The University of
Pennsylvania Museum, Philadelphia, pp. 63-77.
Journal of African Archaeology Vol. 2 (1), 2004
D. Killick
McIntosh, S. K. (in press). Archaeology and the reconstruction
of the African past. In: Philips, J. (ed.), Writing African
History. University of Rochester Press, Rochester.
Osborn, E.F. & Muan, A. 1960. Phase Equilibrium Diagrams of
Oxide Systems. The American Ceramic Society, Columbus.
Paris, F., Person, A., Quéchon, G. & Saliège, J.-F. 1992. Les
débuts de la métallurgie au Niger septentrional (Aïr,
Azawagh, Ighazer, Termit). Journal de la Société des
Africanistes 62, 55-68.
Pettit, P.B., Davies, W., Gamble, C.S. & Richards, M.B. 2003.
Paleolithic radiocarbon chronology: quantifying our
confidence beyond two half-lives. Journal of
Archaeological Science 30, 1685-1693.
Pigott, V.C. (ed.) 1999. The Archaeometallurgy of the Asian
Old World. University Museum Monograph 16/MASCA
Research Papers in Science and Archaeology 16.
University of Pennsylvania, Philadelphia.
Schmidt, P.R. 1997. Iron Technology in East Africa. Indiana
University Press, Bloomington and James Currey, Oxford.
Quéchon, G. & Roset, J.-P. 1974. Prospection archéologique du
massif du Termit (Niger). Cahiérs ORSTOM, Série
Sciences Humaines 11, 85-104.
Wertime, T.A. & Muhly, J.D. 1980. The Coming of the Age of
Iron. Yale University Press, New Have.
... That first spur was followed by a period of critical evaluation (Killick 2004;Killick et al. 1988;McIntosh and McIntosh 1988). A reassessment of Niger furnaces' radiocarbon dates showed a handful of dates to be erroneous. ...
Full-text available
The “Three Age System” designed in the middle of the 19th century framed the general pattern of universal technological evolution. It all started with the use of stone tools in the very long “Stone Age”. The much shorter “Bronze Age” followed, to be capped by the even shorter “Iron Age”. This evolutionary taxonomy was crafted in Scandinavia, based on evidence from Denmark, and Europe by extension. Patterns of global long-term technological evolution recorded in Africa are at variance with this Stone-Bronze-Iron Age sequence; there is no Bronze Age yet. The advent of copper and iron metallurgy is one of the most fascinating debate taking place in African archaeology at the beginning of the 21st century. The debate on the origins of African metallurgies has a long history with multiple implications. It is anchored on 19th century evolutionism and touches on the patterns and pace of technological evolution worldwide. It has also impacted the history of discourses on human progress. As such, it has strong socio-political implications. It was used to support the assumption of “Africa backwardness”; an assumption according to which all important material and institutional inventions and innovations took place elsewhere – in the Near-East precisely – and spread from there to Africa through demic and/or stimulus diffusion. Does such a scheme capture the global human technological history or is it a specific case of local areal development? That is the core of the current debate on the origins of African metallurgy.
... The Nok Culture is also known, however, for its iron metallurgy, which is among the earliest securely dated evidence for ironworking in Africa (see e.g. Killick 2004;Alpern 2005;Pringle 2009; Clist 202, Eggert 204). In the 960s, the Nok Culture was thought to have lasted from 500 BCE to 200 CE (Fagg 962: 445). ...
The Central Nigerian Nok Culture has been well known for its elaborate terracotta sculptures and evidence of iron metallurgy since its discovery by British archaeologist Bernard Fagg in the 1940s. With a date in the first millennium BCE, both, sculptures and ironworking, belong to the earliest of their kind in sub-Saharan Africa. After a period of destruction of Nok sites by looting, scientific research resumed in 2006, when a team of archaeologists from Goethe University in Germany started to explore different Nok Culture aspects, one of which focused on chronology. Establishing a chronology for the Nok Culture employed two approaches: a comprehensive pottery analysis based on decoration and form elements and a wealth of radiocarbon dates from a large number of excavated sites. This volume presents the radiocarbon dates and the methods, data and results of the chronological pottery analysis, conducted within the scope of a dissertation project completed in 2015. Combining the two strands of information, a chronology emerges, dividing the Nok Culture into three phases from the middle of the second millennium BCE to the last centuries BCE and defining seven pottery groups that can be arranged to some extent in a chronological order.
... Rappelons qu'il existe une variété quasi infinie de techniques métallurgiques et de pratiques socio-techniques -en fonction des ethnies, des lieux et des époques -, ce dont témoigne le mieux l'énorme quantité de travaux consacrés à la production de fer en Afrique (p. ex . Killick 2004b ;Mapunda 2000 ;Rehren et al. 2007 ;Schmidt 1997). Cette diversité est une chance car, sans elle, nous ne serions pas capables de définir des styles technologiques et de retracer des réseaux de transmission technologique (p. ex. Charlton et al. 2010 ;Killick 2004a ;Kuhn 2004). Il demeure néanmoins quelques constantes physico-chimiques univ ...
Full-text available
This article is a French translation of the English language original, published in Asian Perspectives in 2009. The original manuscript was written in 2008 but the core ideas stand, governed, as they are, by thermodynamics.
... De ce fait, l'apparition et le développement du fer en Afrique sont largement étudiés de nos jours et deux modèles s'opposent à ce sujet (e.g. Bocoum, 2002 ;Killick, 2004 ;. Selon le modèle diffusionniste, l'exploitation du fer serait apparue au moyen orient et aurait été propagé en Afrique puis en Afrique sub-saharienne et en Afrique de l'ouest pour enfin se diffuser jusque dans le sud du continent. ...
Full-text available
L'objectif de ce travail de thèse a été de développer l'utilisation des isotopes du fer pour le traçage des métaux anciens, principalement ferreux. Notre approche méthodologique s'est articulée autour de trois axes majeurs. Le premier axe a consisté à étudier l'influence potentielle des processus métallurgiques sur la composition isotopique du fer des métaux produits. Pour cela, nous avons mesuré la composition isotopique du fer d'échantillons issus d'expérimentations de réduction de minerai de fer en bas fourneau, réalisées sur un site sidérurgique majeur de la période romaine (la Montagne Noire, Sud-Ouest de la France). Le second axe a eu pour objectif de valider ce nouveau traceur en mesurant la composition des isotopes du fer de minerais, scories et objets en fer issus d'un contexte archéologique connu et dont la provenance avait déjà été identifiée par des analyses élémentaires. Nous avons alors mesuré la composition des isotopes du fer de barres de fer principalement retrouvées dans des épaves romaines au large des Saintes-Maries-de-la-Mer (Bouches-du-Rhône, France). Enfin, le troisième axe a eu pour vocation d'estimer la pertinence, mais aussi les limites de ce nouvel outil en l'appliquant à deux terrains archéologiques très différents, où aucune étude de traçage classique n'avait été utilisée précédemment. Ces deux terrains concernaient la sidérurgie ancienne au Togo et la production de plomb argentifère médiévale au Maroc. Les résultats obtenus montrent que la composition isotopique du fer de la scorie et du métal produit est similaire à celle du minerai correspondant. Il n'y a donc pas de fractionnement des isotopes du fer tout au long de la chaîne opératoire de production de fer. De plus, et contrairement à certains traceurs élémentaires, les isotopes du fer ne sont pas contaminés par la paroi du four très pauvre en cet élément durant la réduction. Ceci permet ainsi d'établir des liens de provenance directs entre un objet en fer et un minerai. L'application de cette méthode de traçage à un contexte archéologique déjà largement étudié a permis de valider les hypothèses de provenance d'objets archéologiques. En outre, les isotopes du fer peuvent être plus discriminants que les éléments en trace car ils permettent notamment de différencier des productions de fer temporellement et géographiquement très proches. Le traçage est ainsi affiné. Enfin, nos résultats préliminaires suggèrent que les analyses des isotopes du fer pourraient également être appliquées à l'étude de la production de métaux non ferreux. Cette étude offre ainsi de nombreuses perspectives, telles que l'étude de la provenance de pièces de musée étant donné la faible quantité de matière nécessaire, l'établissement d'une base de données de composition isotopique du fer de minerais archéologiques et l'élaboration d'une méthode de traçage commune aux métaux ferreux et non ferreux.
... In technological and economic terms, this period is marked by the appearance of iron technologies (MacEachern 1996) Of these technologies, horses would have originally come from North Africa and/ or the Nile valley, but we have as yet no evidence for their route of introduction into the southern Lake Chad Basin. The question of origins for African iron technology is a perpetually vexing issue, but it increasingly seems likely that at least the slag-pit smelting process in use across the continent is distinctively African (Craddock 2009;Killick 2004). The late Holocene spread of rouletting on pottery across Africa appears to originate in the Tichitt-Méma-Niger Bend area and spread into the Lake Chad Basin from the west, as part of a much larger process through which this form of ceramic decoration would eventually be found between the Atlantic and East Africa (Livingstone Smith 2007). ...
Full-text available
The distribution of Chadic languages in Africa is extremely diverse, including the widely dispersed Hausa language, the more restricted Central Chadic languages in the southern Lake Chad Basin, and the poorly understood Eastern Chadic languages in Chad. These distributions are disjunct in complex ways, and the relationships between Chadic and neighboring language families is extremely complicated. The genesis of these distributions lies in the mid-Holocene, with the occupation of the Lake Chad Basin by populations faced by the desiccation of the Sahara and the opening of arable lands further south. Further differentiation of Chadic languages appears to be associated with sociopolitical developments in the region, especially over the last 1,000 years. This chapter will consider the methodological challenges associated with studying the history of these populations using archaeological, linguistic, and genetic data, as well as providing an initial framework for understanding the social dynamics within which these linguistic distributions emerged.
... While the basic requirements for bloomery smelting are essential, there exists an enormous range of technical choice depending on the character and availability of resources, tradition, and knowledge (Killick 2004;Rehren et al. 2007). Experimental smelting therefore provides a means of exploring the possibilities created by the array of potential choices, to shed light on the practices wrought by metallurgists in the past (Killick 1991). ...
Full-text available
The Royal City of Meroe, ca. 200 km north of Khartoum in the modern-day Republic of the Sudan, was an ancient capital of the Kingdom of Kush. From the 3rd century B.C. to the 4th century A.D., Kushite rulers controlled significant territory from the banks of the Nile at Meroe, in part through their ability to ensure the production of significant quantities of iron. The extensive archaeological remains of Meroitic iron production have been investigated over decades, and recently a series of experimental iron smelts in a replica Meroitic furnace has shed new light on the archaeometallurgical evidence. The data generated during the smelting campaigns has provided an understanding of the type of iron ore used, the construction and operating parameters of the furnace, and the workshop space created by the ancient iron smelters during the later and post-Meroitic times.
Mbinga is one of many areas in Tanzania with deep-rooted evidence of archaeology. Archaeometallurgically, the area has remained terra incognita. This paper focuses on Mbinga’s technological change and continuity of iron production techniques in the first and second millennium AD. To achieve this, archaeological data were collected through ethno-history, archaeological survey, excavation, and physical attribute methods, while archaeometric data were generated through optical microscopy, electron microscopy, and metallographic techniques. The results indicate that over the last 1500 years, there has been technological change in iron smelting techniques in terms of furnace charging platform, slag-pit provision, slag tapping, use of multiple tuyères per tuyère port, reduction efficiency, utilized iron ore, and final smelting product. While change took place, there has been technological continuity in the furnace construction materials, forced air supply, and spatial organization of iron smelting activities. The change and continuity in the technology of iron production in Mbinga were probably driven by demand for iron tools for socio-economic purposes, environment, technological efficiency, and the demand for production of carbon-rich steel tools.
Cambridge Core - Archaeology of Asia, Sub-Saharan Africa and The Pacific - The Worlds of the Indian Ocean - by Philippe Beaujard
Full-text available
Three distinct primary pathways to steel production have been identified, namely the Mediterranean, Chinese and African pathways. This paper focuses on African steel production, which has been argued to yield carbon-rich steel as a smelting product. Earlier work has, however, dealt with, and assumed the existence of, a single pathway to steel production in Africa, i.e. steel production as a smelting product whether in forced or natural draft furnaces. Because of this conviction the possibility of a secondary pathway to African steel production, with steel as a refining product, has been discounted. This paper presents ethnographic and archaeological evidence for a secondary (refining) pathway to steel production using miniature (vintengwe) refining furnaces in southwestern Tanzania. It also presents chemical and mineralogical data generated from refining slag materials in the vintengwe. The ethnoarchaeological and scientific data indicate that vintengwe refining furnaces were used to produce secondary carbon-rich steels through carburisation of a soft (low-carbon) iron product from smelting furnaces (malungu). These data constitute the first evidence for a secondary pathway to steel production in Africa as opposed to the well known primary pathway to steel through the use of ore smelting furnaces.
Full-text available
L’ancien royaume Kongo a non seulement donné son nom au fleuve et aux deux pays modernes dont ce cours d’eau constitue une frontière naturelle, mais aussi à la langue bantoue qui fut et est toujours parlée sur son territoire. Ce qui est généralement appelé le kikongo renvoie en fait à un ensemble de plusieurs langues étroitement apparentées, plutôt qu’à un seul et unique parler.
Full-text available
Over the last twenty years, there has been a discernible increase in the number of scholars who have focused their research on metal production, working and use in antiquity, a field of study which has come to be known as archaeometallurgy. Materials scientists and conservators have worked primarily in the laboratory while archaeologists have conducted fieldwork geared to the study of metal technology in a cultural context with laboratory analysis as one portion of the interpretive program.
Full-text available
A large number of structures that appear to be forges or smelting furnaces have been excavated by D. Grébénart in the Agadez region of Niger. Many of the calibrated radiocarbon dates from these structures fall in the second and third millennia BC, more than a millennium older than the earliest previous dates for metallurgy in sub-Saharan Africa. Chemical and microstructural studies of the fused materials from these structures show that most of the samples dated prior to 1000 be are partially vitrified soil and cannot be positively associated with a metallurgical process. The only positive evidence for metallurgy in this region in the second/third millennium BC is a single radiocarbon date of 1710 ± 110 be (GIF-5176) for a copper-working furnace. This date may reflect the use of old charcoal and should be viewed with caution until thermoluminescence dates can be obtained for this furnace.After 1000 BC, native copper and copper oxide minerals were processed in non-tapping shaft furnaces. Calcite, dolomite and aluminosilicate gangue minerals have combined to produce unusual red melilite slags. The scale of production appears to have been very small. Iron smelting came into general use in this region around 500 BC, but the origins of this technology are still unclear.
Full-text available
Les recherches menees entre 1976 et 1981 dans les regions meridionales du Sahara, au Niger ont reactualise la question de la metallurgie posee par les decouvertes faites quelques annees plus tot a Termit. Les âges anciens alors obtenus ont conforte l'idee d'un possible foyer d'invention dans ces regions, avec une succession de differents episodes technologiques, tant pour le fer que pour le cuivre. Les resultats recemment acquis, sur le terrain comme en laboratoire, nous conduisent a discuter ici certaines de ces notions et a reposer le probleme de l'anciennete et de l'originalite technique de cette metallurgie.
Full-text available
It is now three decades since Waterbolk introduced evaluation criteria to 14C chronology. Despite this, and other subsequent attempts to introduce quality control in the use of 14C data, no systematic procedure has been adopted by the archaeological community. As a result, our databases may be significantly weakened by questionable dates and/or questionable associations between dated samples and the archaeological phenomena they are intended to represent. As the use of chronometric data in general becomes more ambitious, we must pause and assess how reliable these data are. Here, we forward a set of evaluation criteria which take into account archaeological (e.g. associational, stratigraphic) and chronometric (e.g. pre-treatment and measurement) criteria. We intend to use such criteria to evaluate a large 14C dataset we have assembled to investigate Late Glacial settlement in Europe, the Near East and North Africa, supported by the Leverhulme Trust. We suggest that the procedure presented here may at least form the basis of the development of more rigorous, scientific use of 14C dates.
The role of rust in the radiocarbon dating of iron artefacts has been examined experimentally. This was accomplished by re-measuring, using modern accelerator mass spectrometry techniques, the radiocarbon dates of ancient iron-based artefacts that were originally radiocarbon dated 30 years ago by beta counting. The samples were dated with the rust that had accumulated over the last 30 years, which in some cases had consumed most of the original metal. The study also afforded the opportunity to compare beta counting with accelerator mass spectrometry, large and small sample sizes, and old versus new carbon extraction methods. The results demonstrate that in at least some circumstances the carbon in rust can be reliably used for radiocarbon dating. Some experimental observations on iron microstructure and oxidation processes which support these results, are reviewed. The issues of sample size and the role of rust are important because they open up new possibilities for dating iron-based artefacts that had previously been assumed to be unavailable.
We present luminescence dates which demonstrate a full-blown smelting technology in Northern Ghana, West Africa, during the Early Iron Age. Our chronology is based on thermoluminescence (TL) dating of quartz grains extracted from the walls of three iron smelters located at the Birimi site in the Northern Region of Ghana. Two of the smelters yielded statistically indistinguishable ages of 1080±70 and 1090±60 years, while the third yielded a higher age of 1600±100 years. All are significantly older than the sole direct radiocarbon date of 550–330 calBP (460±90 BP) obtained on a furnace in the Northern Region. The TL ages indicate that iron smelting was well established in the Northern Region before the middle of the first millennium AD, and corroborate the validity of a number of similarly early radiocarbon dates associated with Early Iron Age ceramics but not directly associated with smelting activity, from other sites in northern Ghana.The iron working remains at Birimi are located on a river terrace, only a few tens of meters west of a dense Kintampo occupation. Radiocarbon and TL dates for the Kintampo complex demonstrate that the Iron Age and Kintampo components at Birimi are clearly non-contemporaneous. The occurrence of two distinct layers of goethite within 150cm of the terrace surface, both of which are clearly exposed in the scarp of the terrace upon which the smelters are situated, indicates that the industrial site was deliberately sited to take advantage of the readily available source material. This premise is confirmed by geochemical analyses of sediment, goethite, and slag from the smelting site. We therefore propose that the frequent proximity of Kintampo and iron production sites is incidental, in the sense that each group exploited different resources offered by a riverbank location.
This paper considers, under rough chronological headings, work undertaken in Niger in the past century. Sites relevant to the Pleistocene occupation of the Sahara, to the adoption of elements of a Neolithic package, to the (perhaps misleadingly late) occupation of the Sahel, to alleged metalworking 3000 years ago, and to social complexity, are described and discussed. These data carry a relevance far beyond their immediate area, and the most fruitful application of the archaeology of Niger is to be found in theoretical rethinking.