ArticlePDF Available

Early Metal Working in Sub-Saharan Africa: A Review of Recent Research



This paper is a review of the course of research during the past decade into the history of indigenous metal working in sub-Saharan Africa. It comprises three sections: a summary of the chronology of early metallurgy and the spread of metal working; a description of African metal working in terms of mining, smelting and smithing, with particular emphasis on recent interpretations of the iron-smelting technology; and a conclusion summarizing the main developments and some lines of future enquiry. A glossary of technical terms used in this paper is appended.
Journal oJ AJrican
Copy'rigl-rt O ,99-+ Historl', SS ( r qq+), pp. r-36
Cambridge lJniversity Press
Uniaersity of Cape Town
TuIs paper is a revierv of the course of research during the past decade into
the history of indigenous metal working in sub-Saharan Africa. It comprises
three sections: a summary of the chronology of early metallurgy and the
spread of metal working; a description of African metal working in terms of
mining, smelting and smithing, with particular emphasis on recent inter-
pretations of the iron-smelting technology; and a conclusion summanzing
the main developments and some lines of future enquiry. A glossary of
technical terms used in this paper is appended.
In the first half of this century writing on African indigenous metallurgy
r,r,as dominated by discussion of the possible role of Egypt in the advent of
iron metallurgy in the old world and ethnographic accounts of smelting
practice.l With the burgeoning of African Iron Age archaeology in the r 96os
more intense interest developed in the economic role of metals production in
fooci producing and trading societies.2 Despite pioneering metallurgical
analysis of southern African and Egyptian metal artefacts3 technical studies
of the products of African metallurgy remained rare until the r98os, when
the pace of publication on all aspects of African metal working increased
dramatically. A major development during the past decade has been the
* Financial support from the Foundation for Research Development, CSIR (South
Africa) and from Anglo-American De Beers Chairman's Fund Educational Trust is
acknorvledged gratefully. The lJniversity of Cape Town provided facilities and ad-
ministrative support for this research. We wish to thank Judith Sealy for her very
thorough reading of a draft and suggestions for improvement. The constructive criticism
of anonl'mous referees is acknowledged lvith thanks.
1 For summaries and discussions of this literature see: W. B. Cline, Mining and
Nletallwgy in I'legro AJrica (Menasha, t%7); P. L. Shinnie (ed.), The African Iron Age
(Oxford, tgTr); F. J. Kense, Traditional AJrican Iron Workizg (University of Calgary
Department of Archaeology, African Occasional Papers, No. r, r983); R. Haaland and
P. Shinnie (eds.), African Iron lVorking - Ancient and Traditional (Oslo, r9B5).
: T. A. Wertime and J. D. Muhly (eds.), The Coming oJ the Age of Iron (New Haven,
r98o); Haaland and Shinnie, African Iron Working;8. W. Herbert, Red Gold of AJrica
(London, rSB+).
3 L. Frobenius, Erythriie, Lii'nder und Zeiten des heiligen Kdnigsmordes (Beriin, rg3r),
zj7*g3; G. H, Stanle1,, 'Some products of native iron smelting', South African Journal
oJ' Science, xxvlII (rg3r), 66-8; E. H. Schulz, 'Zusammensetzung und Aufbau einiger
lVletallfur-rde der Afrika-Expedition von Leo Frobenius rgz8/3o', Paideuma, Y (rg5o),
59-65; D.E.N'Iiller, 'Pioneering metallographic analyses of indigenous metal artefacts
from southern Africa-material collected by the Frobenius expedition in rgzglrg3o',
South AJ-rican Archtteological Bulletin, xLvrr (rggz), io8-r 5 ; C. H. Desch, 'Report on the
nretallurgical examination of specimens for the Sumerian Committee' , Reports of the
British AssociationJor the Adaancement of Science (London, r936), 3o8-ro.
J e n n{i-jeno
-*+j;ir|,, -' . :, ;,r1 g6g5:g&i.i*,*,*4*e'€r4r*t:F4.4i#Ee'{r{*l#
Tchissanga & Madingd
r Kasungu
r E nkwazin i
7 Mzonjane
Fig. r. A map of Africa showing the approximate locations of sites and regions
mentioned in the text. For the sake of clarity the sites of Silver Leaves and
Eiland, both located betr,veen Phalaborwa and Messina in the eastern Transvaal,
have been omitted.
establishment of the discipline of comparative archaeometallurgy,a an in-
tegration of social anthropology and metallurgical science. This field of
research promises rich insights into the nature of technological change and
the social and economic parameters of metals production in pre-colonial
Africa, as well as providing functional models for the interpretation of
archaeological remains of processes long since extinct elsewhere.5 This ap-
4 P. D. Glumac (ed.), 'Recent trends in archaeometallurgical research' , MASCA
Research Papers in Science and Technology, vlu (r) (rggr); R. NI. Ehrenreich (ed.),
'Nletals in society: theory beyond analysis', MASCA Research Papers in Science and
Technology, vrlr (z) (rSSr).
t D.J.Killick, 'Technology in its social setting: bloomery iron smelting at Kasungu,
N{alawi, r86o-194o'(Ph.D. thesis, Yale University, r99o);D.J.Killick,'The relevance
t" -r\
---rr AJ
& Nqom\ r Greit Zimbabwe
\. Mabveni \
M a pu n g u bwei-; M essi n a'
Rooiberg. . Phalaborua
Hufrat en-Nahas
ta Debbat El Eheima
r Dimi
' \, *l Katansa
Benguela rr Xaniaristri
proach has led to the recognition that African metal working is much more
complex and diverse than previously recognized and necessitates a revision
of some of the earlier claims for African uniqueness.
Deciphering the appearance and development of metal-working tech-
nology in Africa has been a background to two interrelated strands of
inquiry.6 One is concerned with the social, cultural, economic and en-
vironmental aspects of the introduction and practice of metallurgy.T The
other describes the chronology, technology and mechanisms of indigenous
metals production itself.s Here we deal primarily with the second topic and
of recent African iron-smelting practice to reconstructions of prehistoric smelting
technology',IVIASCA Research Papers in Science and Technology,vrtt (l) (i99r), 47*54.
o N.J.r,an der,'1lhe advent of iron in Africa', in Wertime and Muhly (eds.),
The Coming of the Age of lron,463-5o6; F. J. Kense, 'The initial diffusion of iron to
Africa', in Haaland and Shinnie (eds.), African Iron Working, rt-27.
t E.g. J. P. Warnier and I. Fowler, 'A nineteenth-century Ruhr in Central Africa',
AJricct, xr- (r97o), 3zg*47; C. Goucher, 'Iron is iron'til it is rust: trade and ecology in the
decline of West African iron-smelting', J.Afr. Hist., xxrr (r98r), r7g-89; C. Goucher,
'The iron industry of Bassar, Togo: an interdisciplinary investigation of African
technological history' (Ph.D. thesis, University of California-Los Angeles, rgB+);
L. NI. Pole, 'Decline or survival? Iron production in West Africa from the seventeenth
to twentieth centuries', j. Afr. Hist., xxtu (r982), 5o3-r3; H. M. Friede, 'Socio-
economic aspects of metal production in the western Transvaal (South Africa)' , Nyame
Ahttma, xxr (r982),4r;H. Sassoon,'Kings, cattle and blacksmiths: royal insignia and
religious symbolism in the interlacustrine states', Azania, xvttt (rg8:), g: ro6; Herbert,
Red Gold; N'i. Wright, 'Iron and regional history: report on a research project in
southwestern Tanzania', African Econoruic History, xIv (rg8S), r+7-6S; R. Haaland,
'Iron production, its socio-cultural context and ecological implications', in Haaland and
Shinnie (eds.), African Iron TVorking, So-72; P. de Maret, 'The smith's myth and the
origin of leadership in central Africa', in Haaland and Shinnie (eds.), AJrican lron
Working, 73-87; P. de Barros, 'Bassar: a quantified, chronologically controlled, regional
approach to a traditional iron production centre in West Africa', Africa, lvI (l986),
r18-73; R. Larick, 'Iron smelting and interethnic conflict among precolonial Maa-
speaking pastoralists of north-central Kenya', The African Archaeological Reaieu, w
(1986), 165-76; N.J.r,an der Merwe and D. Avery, 'Science and magic in African
technology: traditional iron smelting in Malarvi' , Africa, LVII (rg8Z), r43-72;
L. Digombe, P. R. Schmidt, V. Mouleingeu-Boukosso, J.-B. Vlombo and M. Locko,
'The development of an Early Iron Age prehistory in Gabon' , Cttrrent Anthropology, xxlx
(r988), r7g-84; B. M. Perinbam, 'The political organization of traditional gold mining:
the Western Loby, c. r85o to c. rgro', J. Alr. Hist., xxrx (r9BB), $7-62; Killick,
'Technology'; S.T. Childs, 'Style, technology, and iron smelting furnaces in Bantu-
spealiir-rg Africa', Journal of Anthropological Archaeology, x (r99r), 33z-Sg; S. T. Childs,
'Transformations: iron and copper production in central Africa', MASCA Research
Papers in Science and Technology,vtrt (r) (i99r), n-46; S.T.Childs,'Iron as utility or
expression: reforging function in Africa', MASCA Research Papers in Science and
Technology, vrtl (z) (rggr), 57-68.
' E.g. P. R. Schmidt and D. H. Avery, 'Complex iron smelting and prehistoric culture
in Tanzanta', Science, cct (rgZ8), lo85-9; P.R.Schmidt and D. H. Avery, 'More
evidence for an advanced prehistoric iron industry in Africa' , Jottrnal of Field Archaeology,
x (1983), 4zr-3+; D.H.Avery and P. R. Schmidt, 'A metallurgical study of the iron
bloomery particularly as practiced in Buhaya', Journal of Metals, xxxr (r979), r4-zo;
N. J. van der Merr,r,e and D. H. Avery, 'Pathways to steel', American Scientist, t-xx
(r9Bz), r,+6-55; R.B.Gordon and N. J. van der lr'Ierwe, 'Metallographic study of iron
artefacts from the eastern Transvaal, South Africa', Archaeometry,xxvt (rg8+), to8-27;
have restricted this discussion to the archaeology of the physical aspects of
metals production.e This includes the chronology of smelting in Africa, the
process of smelting itself and the practice of forging or smithing.
The intensification of research in the past r 2 years is due partly to the fact
that African metal working reflects aspects of past technological and social
organization central to current issues in African history.lo With tine realiza-
tion that all living memory of indigenous smelting is vanishing rapidly with
the demise of the last of its former practitioners, there has been a vigorous
attempt to record as much of this ethnographic information as possible, both
in print and on fiIm.11
Until the middle of the past decade a number of researchers emphasized
the apparently unique or advanced nature of many indigenous African metal-
working practices.12 With recent more rigorous re-evaluation there has been
a reduction in emphasis of African technological primacy or uniqueness.
This is evident in critical reviews of earlier claims of anomalously early or
superior African metal working technology.l3
With these more recent developments we are rescued from a proliferation
P. R. Schmidt and S. T. Childs, 'Innovation and industry during the Early Iron Age in
East Africa: the KIVIz and I{M3 sites of northwest Tanzania' , The African Archaeological
Reviezu, rrr (rg85), fi-g4; H. o. Kiriama, 'Archaeo-metallurgy of iron smelting slags
from a Vlr.vitu Tradition site in Kenya', South AJrican Archaeological Bulletin, xLtr
(rg8Z), rz;-3o; N. David, R. Heimann, D. Killich and M. Wayman, 'Between bloomery
and blast furnace: IVIafa iron-smelting technology in North Cameroon' , The African
Archaeological Reaiezu, vII (rg8g), r83-zo8; D.H.Avery, N. J. van der Merwe and
S. Saitor,vitz,'Tlne metallurgy of the iron bloomery in Africa', in R. Maddin (ed.), The
Beginning of the Use of Metals and Alloys (Cambridge, MA, r98B), z6t-82; D. J. Killick
and R. B. Gordon, 'The mechanism of iron production in the bloomery furnace', in
R. N{. Farquhar, R. G. V. Hancock and L. A. Pavlish (eds.), Proceedings of the z6th
International Archaeology SymposhLm (Toronto, r988), rzo-3; D.J.Killick, 'A com-
parative perspective on African ironworking technologies' (paper presented at The
African Studies Association, Chicago, r988); D. J. Killick, 'A little-known extractive
process; iron smeltir-rg in natural-draft furnaces', Journal of Metals, xlrrr (r99r),62-4;
D. J. Killick, 'Relevance'; D. J. Killick, 'A tin lerale frorn the Soutpansberg, northern
Transvaal, South Africa', South African Archaeological Bulletin, xLyr (r99r), r37-4r.
s We have not attempted to evaluate the arguments based on linguistic reconstructions
fbr the spread of metal-using peoples in Africa. E.g. D. W. Phillipson, 'Notes on the later
prehistoric radiocarbon chronology of eastern and southern Africa', J.Afr.Hist., xt
OgTo), r-r5; D. W. Phillipson, 'The chronology of the Iron Age in Bantu Africa',J.Afr.
Hist., xvl (1975), 3zt-12; P. de Maret and F. Nsuka, 'History of Bantu metallurgy',
History in AJrica, w (1977), q-65; J. Vansina, ''Western Bantu expansion', J. Afr. Hist.,
xxv (r984), rzg-45; J.R Denbow,'Congo to Kalahari: data and hypotheses about the
political economy of the westem stream of the Early Iron Age', African Archaeological
Reaiezu, vtlt (rg9o), r39-?6. 1o Killick,'Technology'.
t' D. J. Killick, 'Recent films and publications of African metallurgy' , Nyame Akuma,
xxx (1988), :g-+r; E. W. Herbert and L. M. Pole, 'African iron working on film and
video', Nyame Akuma, xxxl (r989),47-g.
12 Schmidt and Avery,'Complex iron smelting'; Schmidt and Avery,'lVlore evi-
dence' ; Avery and Schmidt, 'l\r'Ietallurgical study' ; van der Merwe and Avery,
'Pathr,vays'; Schmidt and Childs,'Innovation'.
" J.E.Rehder, 'l]se of preheated air in primitive furnaces: comment on the vier,vs of
Avery and Schmidt' , Journal of Field Archaeology, xrrr ( r 986) , 3Sz_3; M. Eggert, ' On the
alleged complexity of early and recent iron-smelting in Africa: further comments on the
preheating hypothesis', Journal oJ Field Archaeology, xlv (r987), 377-82; Killick,
of discoveries of 'special' circumstances and can begin to appreciate the
social contexts of indigenous metal productionla and the real diversity of
African metal-working techniques, especially the numerous variants that
make up the bloomery process for producing iron and steel.15 These variants
are far from fully understood, and the fundamentai documentation of them
is only just underway.
Processes conventionally described as forging have been shown to be an
integral part of indigenous metals production.16 Nevertheless, the study of
forging or smithing practice generally has been neglected in favour of
studying the smelting process. IJndoubtedly, the near-indestructibility of
smelting slags has contributed to this situation but, as was pointed out by
Stanley in rgzg, so has the understandable reluctance of curators to submit
metal artefacts for destructive metallographic analysis.l? The sporadic
publication of isolated analyses or the analysis of very small collections of
artefacts has been the norm for the past fifty years. This situation has
changed only recently, with the analysis of large assemblages from Zairerg
and Botswana.le
The radiocarbon-dated chronology of sub-Saharan prehistory is summarrzed
conveniently in the lists of radiocarbon dates and their associated comments
published intermittently in the Journal of African History.zo Other sub-
'Comparative perspective'; D J.Killick, N. J. van der Vlerwe, R. B. Gordon and
D. G16b6nart, 'Reassessment of the evidence for early metallurgy in Niger, West Africa',
Journal of Archaeological Science, xv (r988), S6l-g+.
11 S. K. Mclntosh and R. J. lVlclntosh, 'From stone to metal: new perspectives on the
later prehistory of West Africa', Journal of World Prehistory, Ir (r988), B9-r::;
G. C. Thomasson, 'Primitive Kpelle steelmaking: a high technology indigenous knowl-
edge system for Liberia's future' , Liberian Studies Jottrnal, xII ( r g87), 149-64;
G. C. Thomasson, ' Liberia's seeds of knor,vledge' , Cttltttral Suraival Qttarterly (Summer
rggr), z3-8. 15 Killick, Comparative perspective'.
16 Avery et al.,'Vletallurgy'; David et al.,'Bloomery and blast furnace'.
17 G. H. Stanley, 'Primitive metallurgy in South Africa: some products and their
significance', South African Journal oJ Science, xxvr (rgz9), 732-+8.
1s S. T. Childs, W. Deu,ey, VI. wa B. Kamwanga and P. de Maret, 'Iron and Stone
Age research in Shaba Province, Zaire: an interdisciplinary and international effort',
Nyarne Ahwna, xxxrr ( r 989), S4-g; Childs, ' Transformations '.
'n D. E. Nliller, ' Iron Age metal working at the Tsodilo Hills, northwestern Botswana'
(Ph.D. thesis, University of Cape Torvn, tggz); D. E. Miller and N. J. van der Merwe,
'Early Iron Age metal rvorking at the Tsodilo Hills, north,"vestern Botsr,vana', Journal of
Archaeological Scie'nce (in press); D.E.Miller and N. J. van der Merrve,'Late Iron Age
netal rvorking at Nqoma, Tsodilo Hills, northwestern Botswana', Historical Metallurgy
(in press).
?0 B. NI. Fagan, 'Radiocarbon dates for sub-Saharan Africa', J. Afr. Hist., Ir (r96r),
r37-g; B. NI. Fagan, 'Radiocarbon dates for sub-Saharan Africa: II', j. Afr. Hist., tv
(rg6:), tz7-8; B.M.Fagan, 'Radiocarbon dates for sub-Saharan Africa: III', J. Afr.
Hist., vr (r965), ro7-16; B.M.Fagan,'Radiocarbon dates for sub-Saharan Africa: IV',
J. Afr. Hist., vri (r966), ,t95-5o6; B.M.Fagan, 'Radiocarbon dates for sub-Saharan
Africa:V',). Afr. Hist., vrrr (r967), 513-27; B.M.Fagan,'Radiocarbon dates for sub-
stantial reviews of the chronology of sub-Saharan metallurgy have been
published elsewhere.2l Because of the inherent and significant imprecision of
radiocarbon dates relating to the African Iron Age22 we quote approximate
dates only throughout this section. Because of the lack of uniformity in the
citation of dates in the literature the reader is urged to refer to the original
papers for details of sampling, measurement statistics and calibration.23
Traditionally, Africa has been divided into two parts when discussing its
historical metallurgy, with lVlediterranean North Africa, the Nile Valley, and
the Red Sea coast considered separately from sub-Saharan Africa because
Bronze Age metal working was confined to the Nile valley.2a This review is
concerned primarily with sub-Saharan Africa, but the advent of metallurgy
in Eg1,pt is important to consider in assessing the role of iMero6 in the
development of metallurgy elsewhere in Africa.
Salraran Africa: VI', J Af, Hist., x (r969), 149-69; Phillipson, 'Notes'; Phillipson,
'Chronology' ; F. A. Willett, 'A survey of recent results in the radiocarbon chronology of
western and northern Africa', J.Afr.Hist., xtr (tg7r), Tg-jo; J.E.G. Sutton, 'New
radiocarbon dates for eastern and southern Africa', J.Afr.Hist., xttt (tg7z), r-24;
C. Flight, 'A survey of recent results in the radiocarbon chronology of northern and
western Africa', J. Afr. Hist., xrv (rg73), 53r*S4; R. C. Soper,'New radiocarbon dates
for eastern and southern Africa', J. Alr. Hist., xv (tg74), r7s-gz; M. Posnansky and
R. N'IcIntosh, 'New radiocarbon dates for northern and western Africa', J.Alr.Hist.,
xvtt (r976), t6rg5; P. de Maret, F. van Noten and D. Cahen, 'Radiocarbon dates for
West Central Africa: a synthesis',J. AJr. Hist., xvrrr (r977),48r-5o5;T. Maggs,'Some
recent radiocarbon dates from eastern and southern Africa', j. Alr. Hist., xvttt (tg77),
16r-9r; D. Calvocoressi and N. David, 'A new survey of radiocarbon and thermo-
lr-rminescence dates for West Africa', j. Afr.Hist., xx (r979), t-zg; A. E. Close, 'Current
research and recent radiocarbon dates from northern Africa', j. Afr" Hist., xxr (r98o),
115-67; NI. Hall and J. C. Vogel, 'Some recent radiocarbon dates from southern Africa',
J. Alr. Hist., xxr (rg8o), 43r-55 ; G. G. Y. Mgomezulu,'Recent archaeological research
and radiocarbon dates from eastern Africa', l. Afr. Hist., xxrr (1981), $5-56; P, de
IVIaret, 'Nerv survey of archaeological research and dates for West-Central and North-
Central Africa', j.Afr.Hist., xxrrr (1982), ,-r5;J.E.G. Sutton,'Archaeology in West
Africa: areviewof recentworkandafurtherlistof radiocarbondates', J.Afr.Ilisl., xxrrr
(Ig8z), zgr-3r3; A. E. Close, oCurrent research and recent radiocarbon dates from
northern Africa, II', J. Afr. Hist., xxv (r984), r-24; P. Robertshaw, 'Archaeology in
eastern Africa: recent developments and more dates',J.Afr.Hist.,xxv (rg8+), S6g-gS;
P. de Nlaret, 'Recent archaeological research and dates from Central Africa', J. Afr.
Hist., xxvl (r985), tzg-48; S. K. Mclntosh and R. J. Mclntosh, 'Recent archaeological
research and dates from West Africa', j.AJr.Hist., xxvlr (r986),4t3-42; J. Parkington
and IVI. Hall, 'Patterning in recent radiocarbon dates from southern Africa as a reflection
of prehistoric settlement and interaction',J.Afr.Hist.,xxvtu (rg87), r-25;A. E. Close,
'Current research and recent radiocarbon dates from northern Africa, III',y. Afr. Hist.,
xxlx (1988), t45-76; T. Maggs and G. Whitelaw, 'A review of recent archaeological
research on food-producing communities in southern Africa',1.Alr.Hist., xxxrr (rggr),
3-24; P J.J.Sinclair, 'Archaeology in eastern Africa: an overview of current chrono-
logical issues', j. Afr. Hist,, xxxrr (r99r), 17g-ztg.
21 van der Vlerrve, 'Advent'; Herbert, Red Gold; Kense, 'Initial diffusion'.
22 Vlclntosh and Mclntosh, 'Recent archaeological research'.
23 The interpretation of African Iron Age radiocarbon dates is a topic beyond the scope
of this paper. For discussion of this problem, see Nlclntosh and Vlclntosh, 'Recent
archaeological research'; D.J. Killick, 'On the dating of metallurgical sites', Nyarne
Akttma, xxvrrr (rgB7), zg-3o; I(illich et al., 'Reassessment'.
'4 \,an der lVlerwe, 'Advent'.
Egypt appears to have participated early in the use of copper, with early
finds dated from 5ooo B.c. to 4ooo B.c.25 By 3ooo B.c. arsenical copper was
being produced, and by about z6oo s.c. bronze was in use, possibly using
metal from Sinai, where it is thought to have been mined since about
4ooo B.c.26 The earliest known worked iron in Africa comes from the late
prehistoric period and early Bronze Age of Egypt." The context of these
objects indicates that iron had ritual significance, probably reflecting its
rarity.28 The Middle Bronze Age (from about zoooB.c. to 16oon.c.) is
generally poor in finds of iron, with only one socketed spear head, of doubtful
authenticity, reported from Egypt." After about r57o 8.c., with the rise of
the New Kingdom in Egypt, iron appears more frequently in texts and in
archaeological sites.30 These finds include a dagger blade and miniature iron
and copper artefacts buried with Tutankhamun about r34on.c.31 The
analysis of a number of Egyptian iron artefacts, dating from about r2oo B.c.
to the Roman period, showed signs that some of them had been quenched
and tempered.32 If this identification is correct, it is significant because these
heat treatments were not practised systematically elsewhere in Africa until
European colonization.
The social turbulence of Egypt during the first half of the first millennium
s.c. led to the rvidespread adoption of iron by the middle of the millennium,
possibly stimulated by the invasion of the Assyrians, whose army was well
equipped with iron, and by the Mediterranean mercenary settlements
present in northern Egypt." By this time metallurgy was being practised in
West Africa. Egypt does not appear to have had a direct formative influence
on the advent of metallurgy elsewhere in Africa other than at Mero6.3a
Copper and iron metallurgy seem to have appeared together in sub-
Saharan Atrica without a prior period of copper and bronze manufacture.3s
The lack of any persuasive evidence for copper metallurgy in sub-Saharan
Africa before the advent of iron working in the region impiies the sim-
ultaneous introduction of copper and iron smelting technology.36 During the
past decade the discovery of structures interpreted as furnaces, found in the
25 R. F. Tylecote, A History of Metallurgy (London, r976), r-r82.
26 B. Rothenberg, R. F. Tylecote and P. J. Boydell, 'Chalcolithic copper smelting',
IAIVIS Archaeo-Metallurgy Monographs, I (r978), r-5 r ; B. Rothenberg (ed.), The Ancient
tletallurgy of Copper (London, rggo), r-r9r.
" J.C.\4/aldbaum, 'The first archaeological appearance of iron and the transition to
the Iron Age', in Wertime and lVluhly (eds.), The Coming of the Age of lron,6g-tz6; van
der Nlerrve, ' Advent'.
28 Desch, 'Report' ; Waldlaum, ' First archaeological appearance' ; E. S. El Gayar and
NI. P. Jones, 'Vletallurgical investigations of an iron plate found in r837 in the Great
P-vramid at Gizeh, Egypt', Journal of the Historical Metallurgy Society, xxllr (r989),
75-83; N. H. Gale, H. G. Bachmann, B. Rothenberg, Z. A. Stos-Gale and
R. F. Tylecote, 'The adventitious production of iron in the smelting of copper', in
Rothenberg (ed.), The A'ncient Metallurgy of Copper, t8z-gr. 2e lbid.
30 van der N,Ierwe, 'Advent'.
31 H. H, Coghlan, 'Copper artefacts from Tutankhamun's tomb', Journal of the
Historical Nletallurgy Society,Iv (r975), 65-y; van der Merwe, 'Advent'.
32 \,an der lVlerwe,'Advent'. 33 Tylecote, History; van der Menve,'Advent'.
3r P. L. Shinnie and F'. J. Kense, 'Vleroitic iron working' , Meroitica, tv (tg9z), q-z$;
van der Vlerwe,'Advent',4'7r. 35 Tylecote, History; van der Merwe,'Advent'.
36 r,an der Nlerr,ve and Avery, 'Pathways'.
Agadez region of Niger and dating to the late second and early third
millennium B.c.,37 fueled speculation about an 'Early Copper Age' in North
Africa.38 The interpretation of these structures and theii associated radio-
carbon dates has been reassessed in detail recently, with the telling conclusion
that there is no substantial evidence of smelting activity before rooo B.c. at
Agadez.3e Some of the fused materials formerly identified as slags transpired
to be vitrified soil. At least one of the assumed furnace structures was
probably a burnt-out tree stump used two millennia later for iron smelting,
and the possible use of 'old wood'no -uy be responsible for the very early
radiocarbon dates for smelting at Agadez.aL This latter source of potential
error should be stressed, because it is relevant to the interpretation of all the
charcoal radiocarbon dates used to construct the chronology of the in-
troduction of metal working to Africa. In the future, the routine sampling of
ceramic furnace materials for thermoluminescence dating may help to
corroborate the more precise but not necessarily more accurate radiocarbon
dates in studies of smelting technol ogy.42
The independent discovery of iron smelting in sub-Saharan Africa has
been postulated in a number of contexts.a3 Nevertheless, the autochthonous
development of iron smelting in central West Africa from a hypothetical
earlier phase of copper smelting does not appear to be supported by the
available archaeological evidence.a4
In the light of the reassessment of evidence for very early smelting in
Agadezab it is reasonable to believe that the knowledge of metai workingias
introduced to sub-Saharan Africa from outside, despite the paucity of
archaeological evidence in those areas that might have acted as conduits for
the spread of this technological knowledge.aG Several possible routes have
been suggested for the diffusion of metal-working technology to sub-S ainaran
Africa, based on geographical feasibility and classical references. The first is
south and west from the Kushitic capital of Mero6 on the Nile; the second
is across the Sahara desert from the Gulf of Sirte in modern Libya to Gao
37 Calvocoressi and David, 'New survey'; Sutton, 'Archaeology'; R.F. Tylecote,
'Early copper slags and copper-base metal from the Agadez region of Niger' , Journal of
the Historical Nletallurgy Society, xvr (r9Bz), S8-6+.
38 Herbert , Red Gold; D. N. H. Gr6bdnart, 'Les mdtallurgies de cuivre et du fer autour
d'Agadez (Niger), des origines au d6bout de la pdriode m6di6vale' , Mdmoires de la Socidtd
des Africanesles, rx (lg8:), rog-zs; D.N.H. Grdbdnart, 'Characteristics of the final
Neolithic and Vletal Ages in the region of Agadez (Niger)', in A. Close (ed.), Prehistory
oJ Arid North Africa: Essays in Honour of Fred Wendorf (Dallas, \997), z87-316.
3e Killick et al., ' Reassessment '.
40 NI. B. Schiffer,'Radiocarbon dating and the "old wood" problem: the case of the
Hoholram chronology ' , Jotunal of Arcltaeological Science, xrrr ( r 986), r 3-3o; Killick, ' on
the dating of metallurgical sites'.
al Calvocoressi and David, 'New survey'; Killick et al.,'Reassessment',
a2 Killich et al. , ' Reassessment ' .
{3 See: Kense, 'Initial diffusion' ; B. W. Andah, ' Iron Age beginnings in West Africa:
reflections and suggestions', West AJrican Journal of Archaeology, tx (tg7g), r35-5o;
Schmidt and Avery, 'More evidence'; I.R.Amadi,'The northern factor in the history
of sub-Saharan Africa: the Hamitic hypothesis revisited', Transafrican Journal of
History, xvlll (r989), 8o-9.
r{ \,an der IVIerwe, 'Advent'; Killick et al.,' Reassessment'; Mclntosh and Mclntosh,
'Stone to metal'. 4-o Killick et al., 'Reassessment'. a6 Kense,'Initial diffusion'.
on the Niger bend; the third is from Morocco to Timbuktu; and the fourth
is via maritime East Africa.a?
Iron smelting was carried out at Merod with increasing intensity from
about 2oo B.c" until it was sacked sometime after a.o. 3oo,a8 although the
earliest slag has been recovered from a level dating to the mid-first
millennium B.c.4e The well-preserved furnaces that have been excavated
dated from the later period and were domed slag-tapping furnaces con-
structed rvith mud brick and driven by clay bowl bellows"s0 The iron objects
recovered from graves at Merod were mostly small trinkets and arrow heads
which mimicked those produced locally in copper and bronze.5l There is
little similarity between the furnace design at Mero6 and that used con-
temporaneously elsewhere in Africa.52 The earliest dates associated with iron
smelting at lVIeroE are too late for the site to represent a primary source for
the introduction of metal working to West and East Africa. About 6oo km to
the south, at the site of Debbat El Eheima on the White Nile, iron points
have been dated by association with charcoal to the ninth or eighth century
8.c., several centuries earlier than the earliest signs of iron working at
Mero6.53 The iron smelting technology at Merod relates to Egyptian and
Roman technology and appears to have no bearing on the development of
metallurgy in West or East
At Akjoujt in western Mauritania there is evidence for large-scale copper
smelting beginning in about the sixth century B.c.55 Iron smelting appears to
have been practised contemporaneously at Do Dimmi and at Agadez in
Niger,56 and the oldest iron-smelting furnaces of Taruga and Nsukka in
Nigerias? have been dated to approximately the sixth century 8.c.58
The 'old wood' problem5e makes it impossible to know with certainty that
the u,ood used to date any particular smelting furnace is contemporary with
that furnace. This, in conjunction with the inherent imprecision of radio-
1' \,an der Mer'"ve,'Advent';van der Merwe and Avery,'Pathr,vays'.
48 B. C. Trigger, 'The myth of Merod and the African Iron Age', AJrican Historical
Studies,II (r969), z3-so; R. F. Tylecote, 'Iron working at Mero6, Sudan', Bulletin oJ the
Historical Nletalhtrgy Group, rv (rg7o), 67-72.
as P. Shinnie,'Iron working at Merod', in Haaland and Shinnie (eds.), African lron
Working, z8-:S.
50 R. F. Tylecote, ' Iron smelting at 'laruga, Nigeria' , Journal of the Historical
lVletallurgy Society,lx (r975), 49-56; R.F.Tylecote, 'The origin of iron-smelting in
Africa', lVest African Journal of Archaeology, v (rg7S), r-9; Shinnie and Kense,
' N'Ieroitic' ; Shinnie,' VIerod',
51 G. A. Wainwright, 'Iron in the Napatan and Meroitic Ages', Sudan Notes and
Records, xxvr (r945), q-36; Trigger, 'Myth of lVleroe'; Shinnie and Kense, 'Meroitic'.
52 lbid. 53 Robertshaw, 'Archaeology'.
5r Tylecote, 'Iron smelting at Taruga'; Tylecote, History; Shinnie and Kense,
' lVleroitic ' ; Shinnie, ' N4eroE '.
55 Posnansky and i\y'Iclntosh, 'New radiocarbon dates'; Herbert, Red Gold.
56 Calvocoressi and David,'New survey';Mclntosh and Mclntosh,'Stone to metal'.
5? R. F. Tylecote, 'Iron smelting on the Nigerian Early Iron Age site at Taruga, Abuja
Emirate', Bulletin of the Historical Metallurgy Group, u (r968), 8r-z;Tylecote, 'Origin
of iron-smelting' ; E. E. Okafor and P. Phillips. 'New 1aC ages from Nsukka, Nigeria, and
the origins of African metallurgy', Antiquity, LxvI (rggz),686-8.
58 Sutton, 'Archaeology'; Okafor and Phillips, 'Nerv laC ages',
5e Schif}'er,'Radiocarbon dating'.
carbon dating, militates against the clear resolution of the possible routes of
introduction and the chronological sequence of the development of smelting
technology. This is true even in West Africa, which has a fair number of sites
with early dates for metal working.60
It is evident that on the basis of the available chronological evidence the
introduction of copper smelting and iron smelting to West Africa cannot be
distinguished. Copper and iron working were probably introduced sim-
ultaneously to West Africa in the middle of the first millennium n.c., and
along the same, but possibly multiple, routes from North Africa.61
The introduction of iron working has often been linked to the advent of
agriculture. In West Africa the transition from Stone Age subsistence
strategies to economies depending on iron has been recorded in the
archaeological sequence between the mid first millennium B.c. and early first
millennium A.D. at a number of sites: Daboya in northern Ghana and at
Jenn6-jeno in Mali,62 at Atwetwebooso in Ghana and at Daima in Nigeria.63
Nevertheless, at others, like the shell mound at N'Gaty in Ivory Coast, the
introduction of iron does not appear to have led to major changes in
subsistence economy.6a Evidently the economic and social impacts of the
introduction of metals were not uniform in West Africa or, by implication,
anywhere else in Africa. This is a potentially rich topic for further
archaeological research.
The cultural relationship between West and East Africa during the second
half of the first millennium B.c. is obscure, but the spread of iron metallurgy,
by whatever routes, to encompass the whole of western, central and eastern
Africa must have taken place rapidly during the final centuries of the first
millennium 8.c.65
Dates associated with the earliest phase of iron smelting in Gabon lie
between the seventh and first centuries 8.c.66 with the most probable date for
the inception of iron smelting being the second to third century B.c.67 This
precedes the first reliable occurrences of iron smelting in the East African
Interlacustrine Region. Despite the acknowledged archaeological obscurity
of events in Central Africa during the late first millennium B.c., it is difficult
to reconcile these relatively early dates in Gabon with the suggestion that
60 Mclntosh and Mclntosh,'Stone to metal'; Kense,'Initial diffusion'.
61 Tylecote , History; van der Merwe, 'Advent' ; J. E. G. "Sutton, 'West African metals
and the ancient lVlediterranean', Oxford Journal of Archaeology, tt (rq8l), r8r-8; Kense,
'Initial diffusion'.
62 Kense,'Initial diffusion'; S. I(. Mclntosh and R. J. Mclntosh,'Current directions
in West African prehistory', Annual Reaiew of Anthropology, xll (rg83), zr5-58;
NIcIntosh and McIntosh,' Recent archaeological research'.
63 Calvocoressi and David, 'New survey'.
6a Vlclntosh and Mclntosh, 'Recent archaeological research'.
65 \,an der lVlerwe, 'Advent'.
66 P. R. Schmidt, L. Digombe, NI. Locko and V. Mouleingui, 'Newly dated Iron Age
sites in Gabon' , Nyarne Akttma, xxvr ( r g85), r 6-1 8 ; B. Clist, 'Early Bantu settlements in
west-central Africa: a review of recent research', Current Anthropology, xxvrlr (rg8Z),
3 8o-2.
67 L. Digombe, P. R. Schmidt, V. Mouleingui-Boukosso, J.-B. Mombo and
NI. Locko, ' Gabon; the earliest Iron Age of West Central Africa' , Nyame Akuma, xxvrrr
( r S8Z), 9-r r ; Digombe et al.,'Development of an Early Iron Age prehistory' ; R. Oslisly
and B. Peyrot, 'L'arriv6e des premiers m6tallurgistes sur I'Ogoou6, Gabon', African
Archaeological Reaiew, x (rggz), ng-38.
smelting techniques employed in Gabon may represent the 'spread from the
Interlacustrine Region back toward already settled Bantu speakers in West
Central Africa'.68
Dates from the early first millennium a.o. have been obtained in Rwanda
from charcoal associated with furnaces built from brick-like structural
elements.6e (The description of these structural elements as 'bricks' has led
to some confusion about their nature and hence about their significance. The
unqualified use of 'bricks' implies their intentional manufacture for furnace
construction,T0 rather than their being by-products which may have be re-
used incidentally.Tl) Anomalously early dates ranging from iate second
millennium to mid-first millennium B.c. have also been reported for iron
smelting sites in Rwanda and Burundi,T2 but these may represent a forest fire
stratum or the use of old wood.73 At present, clairns for the seventh-century
B.c. appearance of iron technology in the Interlacustrine RegionTa appear to
be unsupported.
The evidence for early iron smelting in East Africa also dates from the
close of the first millennium B.c.75 Iron artefacts \,vere found associated with
the floor of a house dated to about zoo 8.c.,?6 and a cluster of radiocarbon
dates ranging from the first to the sixth centuries A.D. was associated with
smelting furnaces at the sites of KMz and KVI3 in Tanzanta." A furnace
from Nkese in north-eastern Tanzama dated to between the first and the
third centuries A.D.,t8 and evidence of the earliest known iron production on
the East African coast has been dated to between the first and the fourth
centuries A.D.7e These dates concuruvith those of Matola tradition settlements
located further south in Mogambiques0 and attest to the rapid southward
spread of Iron Age peoples on the east coast of Africa.
68 Digomb e et al., ' Gabon', r r.
6e F. r,an Noten, 'The Early Iron Age in the Interlacustrine region: the diffusion of
iron technology', Azania, xrv (r979), 6r-8o; F. van Noten, 'Ancient and modern iron
smelting in Central Africa'. Zaire, Rwanda and Burundi', in Haaland and Shinnie (eds.),
African Iron lVorking, roz-zo; J. Raymaekers and F. van Noten, 'Early iron furnaces
rvith " bricks " in R"vanda: complementary evidence from Mutwarumbona' , Azania, xxr
(r986), 6s-8+.
'o E.g. Schmidt and Childs,'Innovation and industry', 59;T.Maggs,'The Iron Age
south of the Zarnbezi', in R. G. Klein (ed.), Soalhern African Prehistory and Palaeo-
enoironrnenrs (Rotterdam, r984), 34r.
71 'Since reports of these furnace "bricks"/"briques" have occasioned some mis-
understanding, it should be explained that they were not in general pre-fired. As
illustrated belolv, the shafts of the furnaces (or "kilns") were built up rvith short lengths
and rolls of rvet clay arranged in coils. These were baked to brick consistency during t},e
smelt. Some thus fired were re-used in building the lower courses of new furnaces or in
renovating old ones,'ibid.65. ?2 Schmidt and Childs,'Innovation and industry'.
13 lbid.
'* E.g. F'. r,an Noten and J. Raymaekers, 'Early iron smelting in Central Africa',
Scientffic Arnericant, ccLVIII (r988), r04*r r.
75 Schnidt and Avery,'Complex iron smelting';Avery and Schmidt,'Metallurgical
study' ; van der Nlerwe and Avery,'Pathrvays'; Sinclair,'Archaeology'.
?6 Schmidt ar-rd Avery, 'Vlore evidence'.
?? Schmidt and Childs, 'Innovation and industry'.
?8 P. R. Schmidt, 'Eastern expressions of the "Mwitu" tradition: Early Iron Age
industrl, of the lJsambara mountains, Tanzania', Nyame Akuma, xxx (r988), 36-7.
?e F. A. Chami, 'Excavations of a coastal Early Iron Age site in Kisarawe District,
Tanzanitr', I'{yame Aktma, xxx (r988), 34-5. 80 Sinclair,'Archaeology'
The Early Iron Age archaeology of the Zaire basin and Central Africa
proper is poorly understood, but the first Early Iron Age sites in the region
appear to date from the fourth to first centuries 8.c.81 The most recent
summary of Early Iron Age archaeology of the West Central African region
is that of Denbow.82 The introduction of iron at Tchissanga East on the coast
of Congo has been dated to the early fourth century n.c" This early
appearance of iron is succeeded at the nearby sites of Tandou-Youmbi,
Nleningue and Madingo-Kayes by mid-second-century B.c. to early third-
century a.o. dates associated with ceramics distinct from Tchissanga East
and with a more extensive metals assemblage. The ceramics from these and
related sites appear to have similarities to those from the reportedly mid-
second-century e.o. site of Benfica south of Luanda, possibly indicating a
western origin for the southern African Early Iron Age. This proposition is
strengthened by the fact that the ceramics and metal artefacts from the Early
Iron Age site of Divuyu in the Tsodilo Hills west of the Okavango swamps
in Botswana also appear to relate to those from the sites further north.83
Dates for the first appearance of Early Iron Age peoples in eastern
Southern Africa range from the second to the third centuries A.D. at Matola,
Enkr,vazini,Mzonjani, Silver Leaves and The site of Zttundo on the
southern border of Mogambique has large quantities of slag and tuybre
fragments dating to the late second century A.D.85 Although iron smelting in
the form of slag is clearly associated with the Matola tradition from the coast
of Mogambique,s6 the earliest known preserved metals in southern Africa are
from the fifth-century a.o. site of Broederstroom in the Transvaals? and from
Divuyu in the Tsodilo Hills, with a range of dates from A.D. 550 to e.o. 73o'88
Copper and iron beads have been found at Mabveni in Zirnbabwe and may
date to the second century A.D.,tn but two of the three dates from Mabveni
are late sixth-century a.o.e0 and the dating of this metal work is thus insecure.
Trying to trace the detailed course of the spread of metal working through
Africa from radiocarbon dates is frustrated by a number of systematic and
interpretational problems. The possible use of old wood or ancient charcoal
for smelting, or the contamination of archaeological samples with charcoal
produced in forest fires, compound the interpretational problem of validating
the archaeological association in the case of smelting furnaces and metal
artefacts dated by contiguous charcoal. The systematic imprecision of
radiocarbon dating often produces wide probability ranges for individual
dates, and resolving the first occurrence of a phenomenon is not aided by
caiculating weighted averages where a number of supposedly related dates
are available. The correlation of dates from different sites and regions is
compromised by the uneven application of calibration, which can also alter
any particular radiocarbon date and its standard deviation considerably.
81 Kense, 'Initial diffusion'; J.R.Denbow, pers. comm.
82 Denbow, 'Congo to Kalahari'. 83 lbid.
8{ J. N{orais, 'lVlozambican archaeology: past and present', African Archaeological
Reoiezo, rr (r984), rr3-28;Maggs,'Iron Age south of the Zarnbezi',329-6o; Maggs and
Whitelarv, 'A review'. 85 Morais,'Mozambican archaeology'.
86 Nlaggs and Whitelaw,'A review'; Sinclair, 'Archaeology'.
8? H. VI. Friede and R. H. Steel,'Notes on Iron Age copper-smelting technology in
the Transvaal', Journal oJ the South African Institute of Mining and Metalllrygy, Lxxv
(rg75), zzr-3r. 88 Denbow, 'Congo to Kalahari'. 8e Friede and Steel, 'Notes'.
so Nlaggs, 'The Iron Ag.'.
These factors all conspire to reduce the resolution one can expect from the
relatively sparse radiocarbon dates available for the earliest appearance of
metal r,vorking in different parts of Africa. The problem of the unavoidably
poor resolution of radiocarbon dates has been discussed recently with a
graphic illustration of the 'stunning imprecision of calibrated dates for early
metallurgy in West Africa' and the conclusion that 'it is high time that the
focus of early iron research... shifted from when and where iron first
appeared to how and why it spread and subsequently developed'.e1
A broad trend is obvious though, with dates in the middle of the first
millennium B.c. for the appearance of copper and iron working in Sahelian
West Africa; the spread of metal working during the last centuries B.c. to
encompass the whole of West Central Africa, the Interlacustrine Region and
the East African coast; and the subsequent passage of metal working
southward by eastern and western coastal routes to appear in Botswana, the
northern Transvaal and Natal by the middle of the first millennium a.o. The
paths travelled are no doubt ramified and complex,e2 and considerably more
u,ork is required to establish a reliable detailed chronology. The necessary
resolution r,vill not be achieved with radiocarbon dating alone, and an
increasing emphasis will have to be placed on thermoluminescence dating
and ceramic typology to resolve cultural change.e3
It is not possible to present a detailed chronology of indigenous mining
because not enough studies of African mining have been carried out to allow
one to construct such a history. Nevertheless, there are numerous records of
indigenous mining in Africa; indeed in southern Africa early European
prospectors found that nearly every viable outcrop of gold- or copper-bearing
rock had been exploited in the past.ea The earliest indigenous mining was
probably for minerals used in cosmetics and body paint, including galena,
malachite,es pyrolusite, ochre, and specularite.e6 Some of the southern
African pigment mines have very early dates of about 4o,ooo 8.c.," but the
s1 lVlclntosh and lVlclntosh,'Stone to metal', ro7.
e2 \,an der lVIerrve,'Advent'; Kense,'Initial diffusion'.
e3 lVlclntosh and Mclntosh, 'Stone to metal'.
ea R. Summers, Ancient Mining in Rhodesia and Adjacent Areas (Salisbury, 1969);
Herbert, Red Gold.
e5 A. A. Hassan and F. A. Hassan, ' Source of galena in predynastic Egypt at Nagada',
Archaeoruetry,xxtu (r98r), 77-8"; V.E. Chikwendu, R. M. Craddock, R. W. Farquhar,
T. Sharv and A. C. Umeji, 'Nigerian sources of copper, lead, and tin for the Igbo Ukwe
bronzes', Archaeometry, xxxtr (rqSq), 2716.
e6 T. G. Trevor, 'Some observations on ancient mine workings in the Transvaal',
Journal oJ the Cheruical, Nletallttrgical ancl Nlining Society of South Africa, xtt (r9rz),
z6j-7 S; P. B. Beaumont, ' The ancient pigment mines of southern Africa', South African
Journal of Science, lxrx (r973), r4o-6; G. Cohen, 'The ancient workings at Gakgale',
Botswana Notes and Records, rx (r g77), t7-tg; A. L Thackeray, J. F. Thackeray and
P. B. Beaumont, 'Excavations at the Blinkklipkop specularite mine near Postmasburg',
Sottth African Archaeological Bulletin, xxxvIII (rg8:), r7*zs.
e7 Beaumont, 'Ancient pigment mines'.
dating is based on charcoal found lying on worked floors and has not been
Not surprisingly the earliest available dates for mining with metallurgical
intent are from West Africa, where there are indications that copper mining
at Akjoujt began in the first millenniurn The onset of coppei extraction
at Agadez was roughly contemporaneous.lO0 The two other locations of
extensive indigenous copper mining in North Africa, at Nioro-Siracoro in
northeru Mali and at Hufrat en-Nahas in the Sudan do not appear to have
been dated.1o1
Historical records attest to the pre-colonial exploitation of the copper
deposits of Niari in the Congo Republic, at Bembe and Benguela in Arrgola.o,
and at Tsumeb in Namibia, where supposedly 'Bushman' miners traded
copper ore to the local Ovambo"lo' At the time of the first European
settlement at the cape of Good Hope, the local indigenes traded copper,
brass, and iron from the Dutch and the English.l0a Buiin r66r the quantity
of copper evident in the first contact with the Little Namaqua north of the
colony was a surprise. Pieter van Meerhoff reported to the Cape governor
Van Riebeeck:
Their dress consists of all kind of beautifully prepared skins ... gorgeously
ornamented rvith copper beads... Their locks they thread with copper beads,
covering their heads all over. Around their necks they have chain., rl,rrrg round.
them fifteen or sixteen times. Many have round copper plates suspended from
these chains. On their arms they have chainr of .opi.r and iron beads which go
round their bodies thirty or forty times ... their legs are encased in plaited skins,
ornamented r,vith beads... Their only industry is working in copper and iron, from
rvhich they make very neat beads and chains.r05
It is unlikely that this represented material bartered from the Cape, because
rumours of a 'copper mountain' in Namaqualand were borne out by Simon
van der Stel's expedition of r685. No unequivocal evidence of indigenous
mining or smelting is recorded from the area, although indication. of pr.-
European mining in the Richtersveld have been reported anecdotally by the
prospector Cornell.106
The situation in Namaqualand is archaeologically anomalous because
almost all other major copper-producing areas in central Southern Africa
have clear evidence of early exploitation.l0? Mining and smelting rubble at
Kansanshi, only one of a large number of prehistorically worked copper
'o R.J.Vlason, 'Prehistoric mining in South Africa and Iron Age copper mines in the
Drvarsberg, Transvaal', Journal of the South African Institute oJ Mining and Metalhrgy,
LxxxII (rg8z), t34-+4. ee Calvocoressi and David, ,New survey,.
100 S. Bernus and N. Echard, 'Metalworking in the Agadez Region (Niger): an ethno-
archaeological approach', in P" T. Craddock and M. J. Hughes (eds.), Furnaces and
smelting Technology in Antiqttity (London, rg85), 7r-8o. 101 Flerber:t, Red Gold.
\02 lbid.
103 c. J. Ar-rdersson, Lake Ngami (London, r856), tlz_zo3; G. Btirg, Die ruttzbaren
lVlinerallagerstiitten aon Deutsch-Siidwestafrika (Berlin, rg4z), 66-9, toz-3, rzo-r;
B. Heintze, 'Buschmiinner und Ambo' , Journal, Siidwest-Afrikanische Wissenichaftliche
G e s e llschaft ('f,Vindho ek), xxvr (r 97 z), 45- 56.
10r A. J' H. Goodwin, 'lVletal working among the early Hottentots', South African
Archaeological Bttlletin, xr (tg56), 46_5r. Loi lbid. 4g.
106 F. C. Cornell, The Glamour of Prospecting (Cape Town, r986), 146_7.
107 Herbert, Red Gold.
sources in Zarnbia and southern Zaire, has yielded radiocarbon dates ranging
from about A.D.4oo to A.D. t655.tot Some of the trenches at Kansanshi were
so narrow that they probably could have been worked only by children.
lVlalachite ore was transported about a kilometre to smelters operated by
three successive cultural groups distinguished on the basis of distinctive
pottery.loe Further records of copper mining in Central Africa have been
summarized comprehensively by Herbert.l10
The record of early copper mining rn Zirnbabwe, eastern Botswana and the
l['ransvaal is particularly rich. The ancient copper mines and mining
techniques of Zimbabwe have been described extensively by Summers,1l1
with an estimate of 5oo ancient mines devoted to copper exploitation in
Zimbabwe and eastern Botswana. More recent reports have confirmed the
efficiency of prehistoric copper prospecting in Botswana.l12
The ancient copper mines of the Transvaal attracted considerable scientific
attention from mining engineers early this century."' Early workings are
scattered throughout the Transvaal,lla but the main centres of copper
production were Messinalls and Phalaborwa.ll6 The ancient workings at
i08 NI. S. Bisson,'Prehistoric copper mining in northwestern Zambia', Archaeology,
xxvlr (r 974), z4z-8.
10e Bisson, 'Prehistoric copper mining';lVI. S. Bisson,'Continuity and discontinuity
in Copperbelt and North-western Province ceramic sequences', Nyame Akttma, xxxt
(r989), 43-6. 110 Herbert, Red Gold. 111 Summers, Ancient Mining.
i12 Tr. G. Niolyneux and T. Reinecke, 'Ancient ruins and mines of the Tati region of
north-east Botslvana', South African Archaeological Bttlletin, xxxvIII (rq8:), gg;
R. S. Johnson, 'llhe Phoenix and Selkirk nickel-copper sulphide ore deposits, Tati
greenstone belt, eastern Botswana', in C. R. Anhaeusser and S. Maske (eds.), Mineral
Deposits oJ Southern Africa (Johannesburg, r986), 243-8; C. van Waarden, 'Prehistoric
copper minir-rg in Botsr,vana' (Paper presented to the Archaeometallurgy Workshop,
Southern African Association of Archaeologists, Pretoria, rg87).
113 G. H. Stanle1,, 'Notes on ancient copper workings and smelting in the northern
Transvaal', Proceedings of tl're Uniaersity of Durham Philosophical Society, ttt (r9ro),
3a7-tz; F. E. B. Fripp, 'An engineer's notes on old mine-lvorkings of Transvaal and
Rhodesia', Proceedings of the Rhodesia Scientific Association, xI (rgrz), r68-8r;Trevor,
'Some observations'; '1. G. Trevor, 'Some observations on ancient mine workings in the
Transvaal: repll,to discussion' , Journal of the Chemical, Metallurgical and Mining Society
of South Africa, xltt (r9r:), 148-9; T.G.Trevor, 'Some observations on the relics of
pre-European culture in Rhodesia and South Africa' , Journal of the Royal Anthropological
Institute, r-x ( r 93o), :8S-qg.
11r T. lVI. Evers and R. P. van den Berg, 'Ancient mining in southern Africa, with
reference to a copper mine in the Harmony Block, north-eastern Transvaal', Journal of
the South African Institute of Nlining and Nletallurgy, Lxxrv (tg7+), zt7-26; R. Mason,
'Background to the Trar-rsvaal Iron Age - new discoveries at Olifantspoort and Broeder-
stroonl', Journal of the South African Institute of Mining and Metallurgy, Lxxrv (tg7+),
zrr-16 ; lVason, 'Prehistoric mining' ; H. iVI. Friede, ' Iron Age mining in the Transt aal',
Journal of the South AJrican Institute of Nlining and Metallurgy, Lxxx (rg8o), r56-65;
R. H. Steel,'Iron Age mining and metallurgy in South Africa', in R. K. Hitchcock and
N{. R. Smith (eds.), SettlenLe'nt i'n Botszuana (Johannesburg, tg9o), g4-7.
115 B. J. de Vaal, 'Die ou kopermyne van Nlessina', Lantern, r (I952), 4o5-B;
E. O. l\'I. Hanish, 'Copper rvorking ir-r the Messina district' , Jottrnal oJ the Sortth African
Institute of lVlining and Nletallurgy,Lxxrv (tgl+), zso-3; N4. Wilson,'Historical aspects
of mining in and around Venda', Newsletter of the South African Deaelopment Trust
Corporation Ltd., w (r989), 6-7.
116 C. IVI. Schrvellnus,'Short notes on the Phalaborwa smelting ovens', South AJrican
Journal of Science, xxxlII (rgS7), go4-rz; W.J.Verwoerd, 'Sekere produhte van
lVlessina were destroyed by modern mining before they could be studied by
archaeologists, but the extensive copper workings at Phalaborwa were
studied by van der Merwe and Scully.117 They identified the ores mined as
malachite and azurite, removed from complexes of chambers, adits and
shafts (some of them so narrow that they also were probably worked by
children) and recovered charcoatr from the floor of a freshly opened gallery
which dated to about a.o. 77o.r18 This is the earliest reliable date for mining
in southern Africa.lle
lVlining for copper in southern Africa appears to have been restricted to the
oxidized surface zone of the ore bodies, with the ultimate depth of mining
determined by the water-table. The mining was done by hand using stone
hammers and iron gads. There is no archaeological evidence for mechanical
aids for hoisting ore, and this must have been done manually, presumably in
baskets. Wooden ladders and pit props have been found in situ in some
workings, and charcoal attests to the use of fire-setting to break rock from the
face or to produce updraughts for ventilation.l20 It has been argued that these
mining techniques relate to Indian mining practicelzl but in reality their
similarities simply reflect the geological constraints imposed on the ap-
plication of a rudimentary mining technology.
The southern African records of indigenous copper mining almost always
include accounts of gold mining, and in some cases copper and gold were
possibly recovered together.122 A small number of ancient gold mines have
been reported from the Transvaal.rzs By far the majority of the ancient mines
in Ztrnbabwe, an estimated 4ooo, were for gold, which was recovered mainly
from auriferous quartz veins.12a This gold played an integral role in the
establishment of major trading towns like Mapungubwe and Great Zirn-
babu'e, which dominated the Indian Ocean trade in southern Africa during
the first half of the second millennium A.D.125
primitielr,e koper-, yster-, en bronssmeltery in Oos Transvaal met besonder verrnysing na
Phalaborwa', Teg'nikon, tx (tg56), 9l-ro4; N. J. van der Merwe and R. T. K. Scully,
'The Phalaborwa story: archaeological and ethnographic investigation of a South African
Iron Age group', World Archaeology, llt (r97r), r7g-zoz; C.E.More, 'Some observa-
tions on "ancient" mining at Phalaborwa', Journal of the South African Institute of
Nlining and Metallurgy, Lxxtv (rg74), zz7-32.
r17 \,an der Merwe and Scully, 'llhe Phalaborwa story'. rLB lbid.
lls lVlason, 'Prehistoric mining'. 120 Summers, Ancient Mining.
121 lbid.; C.A.Hromnik, Indo-Africa (Cape Town, r98r),45.
122 Trevor,'Some observations'; Summers, Ancient Mining.
r23 Trevor, 'Some observations'; Fripp, 'Traditional smithing'; Vaal, 'An
ancient mine in the Transvaal', African Studies, t (tg4z), rSt-z; R. Mason, Prehistory of
the Transaaal (Johannesburg, r96z),4r9; G. M. Thain, 'Ancient gold-mining sites in the
Transvaal ' , Journal oJ the South African Institute of Mining and Metallurgy, Lxxry (tg7 +),
243-4; Friede,'Iron Age mining'. 12a Summers, Ancient Mining.
r25 lbid.; T.N.Huffman, 'Ancient mining and Zirnbabwe', Journal of the South
African Institute of Mining and Metallurgy, Lxxrv (rg7+), 238-42; T.N.Huffman,
'Archaeology and ethnohistory of the African Iron Age' , Annual Reaiew of Anthropology,
xI ( r 98z), r 3 3-5o ; L R. Phimister, ' " Ancient " mining near Great Zimbabwe' , Journal of
the South African Institute of Nlining and Metalhugy,Lxxty (tg7+),43-7; I. R. Phimister,
'Pre-colonial gold mining in southern Zambezia: a reassessment', African Social
Research, xxt (r976), v3o; Maggs, 'The Iron Age';A. Oddy, 'On the trail of Iron Age
The knolvn ancient tin sources in Africa are restricted to the mines of
Rooiberg in the central Transvaal,t'u u single possible tin-producing mine in
Zwnbabwe,lz7 and the alluvial deposits of central Nigeria.1z8 In some
instances in Nigeria tin was produced as a by-product of indigenous iron
smelting,l2e r,vhich suggests that initially the presence of tin may have been
discovered by smelting cassiterite-bearing gossans for their iron content.
A11 the metals required for the creation of the tenth-century A.D. bronzes
from Igbo-Ukwu (and possibly also the later Nigerian bronzes) have been
shorvn to be available locally in the Benue Rift zone of Nigeria.13o Recent
research has been aimed at locating the ancient mines and sourcing the
metals by trace element analysis.131
The extensive early tin mines at Rooiberg in the Transvaal received some
attention early this century,t" but most of the evidence of early mining has
been destroyed.133 Dates of about A.D. r 445 and a.o. r485 have been obtained
retrospectively from a log found in the mine and from charcoal embedded in
a tin ingot.l3n Much of the estimated r8,ooo tonnes of tin ore mined
prehistorically at Rooiberglss must have been exported, because very little
tin or bronze has been found in local archaeological sites. Nevertheless some
tin may have been used locally to make bronze and bronze-like alloys.136 A
gold', Transaaal Nlttsettm Bulletin, xrx (r983), z4-6; A. Oddy, 'Gold in the southern
African Iron Age', Gold Bttlletin, xvu (1984), 7o-8; M. Hall, The Cha'nging Past:
Farrners, Kings and Traders in Southern Africa, AD zoo-t86o (Cape Town, rg87), r-r6r.
126 Friede, ' Iron Age mining' ; Steel, ' Iron Age mining' ; M. R. Grant, 'A radiocarbon
date on a tin artefact from Rooiberg' , South African Journal oJ Science, LXxxvI ( r 99o), 63.
12? Summers, Ancient Mining.
128 D. E. S. King, 'The provenance of tin metal in working for alluvial cassiterite in
Nigeria', Journal oJ the Historical Metallurgy Sociely, xxIII (rg8g), 35-4o; Chikwendu el
rz/., ' Nigerian sources '. 12e King, 'The provenance of tin'.
130 E. E. Okafor,'Ancient mines in the north of Igboland', Nyame Akuma, xxtv/xxv
(rq8+), r.f-rS; Chihwendu et al.,'Nigerian sources'.
r31 C. L. Goucher, J. H. Teilhet, K. R. Wilson, and T. J. Chow, 'Lead isotope studies
of metal sources for ancient Nigerian " bronzes " ' , Nature, ccLXII (t976), r 3o-r ;
P. T. Craddock, 'lVledieval copper alloy production and West African bronze analyses-
part I', Archaeometry, xxvu (rg85), r7-+L; P.T.Craddock and J. Picton, 'Medieval
copper alloy production and West African bronze analyses-part II', Archaeometry,
xxvIII (r986), :-:z; Chikwendu et al.,'Nigerian sources'.
132 N,I. Baumann, 'Ancient tin mines of the Transvaal', Journal of the Chemical,
Nletallurgical and Nlining Society oJ South Africa, xtx (r9r9), rzo-32,; Trevor,'Some
observations' ; Frobenius, Erythrde, Lrinder and Zeiten des heiligen Krinigsruordes,
133 G. St. J. O. Oxland and H. White, 'Ancient mining practices in the Rooiberg area' ,
Journal oJ the South African Institute of Mining and Metallurgy, Lxxrv (rg7+), z+5;
NI. R. Grant, pers. comm 3a Grant, 'A radiocarbon date'.
135 Baumann, 'Ancient tin mines'; H.M.Friede and R. B. Steel, 'Tin mining and
smelting ir-r the Transvaal during the Iron Age', Journal of the South African Institute of
fuIining and Metallurgy, LXXvr (rg76), 46v7o; Grant, 'A radiocarbon date'.
136 P. A. Wagner, 'Bronze from an ancient smelter in the Waterberg District,
Transvaal', South AJrican Journal oJ Science, xxttl (r926), 899-9oo; P.A.Wagner and
H. S. Gordon, 'Further notes on ancient bronze smelters in the Waterberg District,
Transvaal' , South AJrican Journal oJ Science, xxvr (r929) , 563-74; H. M. Friede, 'Notes
on the composition of pre-European copper and copper-alloy artefacts from the
number of tin ingots and a handful of bronze artefacts have been found at
northern Transvaal sitesl3? and at Great Zimbabwe.l38 Ongoing trace
element analysis of these artefacts and tin ingots from Rooiberg is aimed at
sourcing this material.l3e Although a few brass artefacts do occur in the
archaeological record in southern Africa, there is no evidence that zinc was
being produced loca11y,1a0 which implies the importation of zrnc, perhaps
from the long-established sources in India.1al
Iron ores are more ubiquitous,la2 but indigenous mining for iron ores is
usually less conspicttous and has attracted less archaeological attention than
mining for non-ferrous ores. The variety of material mined, or collected, was
considerable and ranged from high grade magnetitela3 to low grade laterite.laa
This range in composition necessitated a diversity of ore preparation and
smelting techniques. In some cases lateritic ores apparently too low in grade
to produce free iron by direct reduction were exploited.lab These were
partially reduced at relativeiy low temperatures to upgrade the ore in a prior
pre-smelting process,146 or smelted without allowing the reaction between
the iron oxide and the coarse quartz gangue to proceed to completion, hence
excluding much of the gangue from the process.l47
Iron ore mining techniques in Africa varied widety depending on the
nature of the ore.'n8 For instance, at Sukur in north-eastern Nigeria, women
Transvaal' , Journal of the South African Institute of Miruing and Metallurgy, LXxv (rg7S),
137 L, C. Thompson, 'Ingots of native manufacture', Natiae Affairs Department
Anrutal, xxvl (1949), 7-zz; L.C.Thompson, 'A native-made ingot', Natiae Affairs
Departrnent A'nnual, xxxl (r 954), 4o-r.
138 Stanley, 'Primitive metallurgy'; G. H. Stanley, 'The composition of some pre-
historic South African bronzes with notes on the methods of analysis' , South African
Journal of Science, xxvr (r929),4+-g; G. Caton-Thompson, The Zimbabwe Culhre : Ruins
and Reactiozs (Oxford, r93r), zoo-28.
l3e Grant, 'A radiocarbon date'; VL R. Grant, pers. comm.
140 lJ. S. I(r.isel, 'Extractive metallurgy in Iron Age South Africa' , Journal of the South
African Institute of Mining and Metallurgy, Lxxrv (rq7+), 246-9.
141 P. T. Craddock, I. C. Freestone, L. K. Gurjar, K. T. M. Hegde and V. H.
Sonar.vane, 'Early zinc production in India', Mining Magazine, clrr (1985), 4S*5";
I. C. Freestone, P. T. Craddock, K. T. M, Hegde, M. J. Hughes and H. V. Paliwal,
'Zinc production at Zawar, Rajasthan', in Craddock and Hughes (eds.) Furnaces and
Srnelting Technology, zzg-44; H.V.Paliwal, L. K. Gurjar, and P. T. Craddock,'Zinc
and brass in ancient India', Btilletin oJ the Canadian Institute of Mining and Metallurgy,
Lxxrx (r986), 75-9. 142 Friede, 'Iron Age mining'.
'*' E.g. N. J. van der lVlerwe and D. J. Killick, 'square: an iron smelting site near
Phalaborr,va', South African Archaeological Society Goodutirt Series, rrl (r979), B6-Sf .
'nn E.g. D. B. Wenner and N. J. van der Merwe, 'Mining for the lowest grade ore:
traditional iron production in Malawi', Geoarchaeology, rr (r986), tgg-2t6.
1{5 L Keesman, J. Preuss and J. Endres, 'Eisengewil-rnung aus laterischen Erzen,
Ruki-Region, Pror.inz F,quateurfZaire', Offa, xl (rg83), r83-9o.
'46 Wenner and van der lVlerwe, 'lVlining'.
1{7 D. Killick, 'Recent iron-smelting in central Malawi' , Nyame Akuma, xxvrrr (rgB7),
z7-g; Killick and Gordon, 'The mechanism of iron production'; Avery et al.,'Metal-
148 W. Rostoker and B. Bronson, Pre-industrial Iron: its Technology and Ethnology
(Philadelphia, rggo), t-232.
collected magnetite sand from stream beds and washed it to remove the
unwanted quartz.lae At Bandjeli in Togo women mined haematite ore from
shallor,v surface pits and trenches.150 On the Nyika plateau in Malawi mining
for lateritic ore took place in large open pits and from underground galleries
and tunnels;tt' while at Phalaborwa locally abundant magnetite pebbles were
simply collected from the surface.l52 The diversity of exploited ore sources,
and the related diversity of furnaces used, must indicate a complex (and
virtually unknown) history of prospecting, experimentation, technical inno-
vation and local development.
The literature describing indigenous smelting in Africa is vast and cannot be
reviewed here comprehensively. For more extensive lists of references
dealing r,vith the topic, see the bibiiographies of Cline153 and Miller and
The operations of sub-Saharan smelters were associated almost ubiquit-
ously with extensive ritua1.155 The ethnographic record reveals that ritual
was aimed primariiy at warding off malevolent influences so as to protect the
smelters involved in an enterprise perceived as dangerous and to ensure the
successful outcome of the smelt. The smelter's role was identified as special,
rvith the smelter accorded particular status, either elevated or in some cases
as a member of an outsider caste. This reflected respect, and often fear, for
the smelter's ability to manipulate the dangerous forces of generation. The
reproductive symbolism of the smelting process was often made explicit by
the anthropomorphic design of furnaces found ranging from Nigeria to
southern Africa.156
Several detailed technical descriptions of various aspects of the processes
involved in indigenous African copper and iron smelting are avatlable,l57
while a number of descriptions dealing specifically with the mechanism of
the bloomery iron process are relevant here.r58 This discussion will con-
lre H. Sassoor-r, 'Iron-smelting in the hill village of Sukur, north-eastern Nigeria',
NIan, txl (tg6+), 174-8. 150 de Barros, 'Bassar'.
151 Wenner and van der Vlerwe, 'Mining'.
,.2 \,an der lVlerwe and Killick,'square'. 153 Cline, Mining and Metallurgy.
15{ D. E. Miller and T. Vlaggs, 'Pre-colonial metalworking in Africa, especially
southern Africa: a bibliography' (Department of Archaeology, IJniversity of Cape Town,
rggr), r-6r. 155 van der Merwe and Avery, 'Science and magic''
156 van der Vlerrve, 'Advent' ; L. M. Pole, ' Furnace design ar-rd the smelting operation :
a slrr\/ey of rvritten reports or iron smelting in West Africa', in Haaland and Shinnie
(eds,), African lron Working, 14z-63.
15? N. J. r,an der Vlerrn'e, The Carbon'r4 Dating of Iron (Chicago, ry6g), r-r37; van der
Nlerrve,'Adt,ent';N.J.van der Merwe,'Production of high carbon steel in the African
Iror-r Age: the direct steel process', in R. E. Leakey and B. A. Ogot (eds.), Proceedings of
the Pan AJrican Congress of Prehistory and Quaternary Studies (Nairobi, rg77),33I-4; van
der N'Ierwe and Avery,'Pathways' ; Avery et al.,'Metallurgy'; Killick"A comparative
perspective' ; Killick, 'Technology' ; Killick, 'A little-known extractive process'-
158 R. Pleiner, 'Problem of direct steel production in early ferrous metallurgy' , Steel
Times,xx (rq6B), 3rz_l8; R. E. Clough, 'The bloomery process - observations on the use
of rich ores and the production of natural steel', in B. G. Scott and H. Cleere (eds.), The
centrate mainly on aspects of research into copper and iron smelting in Africa
that are currently topical and will deal with the two metals ."pu*t"ly.
The furnaces used for copper smelting varied widely in their design and
size and presumably in their operation. 'Ihese included simple conical non-
slag-tapping shaft furnaces of clay like those used in the first millennium B.c.
in the Agadez region of Niger;15e shallow basins surmounted by piles of
anthills and more substantial clay brick ovens in Katanga;160 the simple bowl
and shaft construction identified at phalaborwa;tu, and the more complex
domed furnaces described from Uitkomst Cave and Buispoort in the
Transvaal.162 The ores used appear to have been exclusively the hydrous
carbonates malachite and azurite.163 In some instances the ores were roasted
before smelting, driving offthe water and incidentaliy oxidising any sulphide
fragments, but th9 systematic processing of massive sulphid" or., by
roasting seems to have been unknown.l.a The number and form of the
tuybres used in copp-el smelting aiso varied, from flared clay pipesr.b to
hollowed out anthillsl66 and tubes fashioned out of stone.162
Iron oxide was added on occasion as a flux,168 but the possible addition of
other fluxes such as lime in the form of shells or calcrete is debatable.l6e The
mechanisms of copper smelting in Africa have received remarkably little
attention despite the recogmzed indigenous importance of the metal.l?O
Successful experimental smelting of copper was ca.ried out by Friede and
Steel in a variet), of furnace designs - a ceramic low shaft fu.nate, a crucible
furnace with the crucible supported by rocks and a reconstruction of the
domed, so-called Uitkomst type furnace.l71 Nevertheless, there are no
available data on yields or quantified comparisons between different tech-
niques, and there is clearly the opportunity for far more experimental work.
The chronological issues associated with the confusing claims for very
early copper working in the Agadez region appear to have been resolved, bui
a technical puzzle remains.172 If copper smelting was introduced to Africa in
the first millennium B.c. it postdated the knowledge of sulphide ore roasting,
already being practised in Europe.l73 Perhaps the relatively plentiful supply
of surficial carbonate ore in Africa was sufficient to satisfy lrocal needs,^and
the more complicated and hazardous mining techniques needed to mine the
more deeply seated sulphide ores would have involved an unnecessarily
taxing escalation in mining activity.l?4
Crafts of the Blacksmith (Belfast, r98+), rg-27; N. Bjorkenstam, ,prehistoric and
medieval iron production: reaction processes in the reduction of iron ores in low, shaft
f\rnaces', Journal of the Historical Metallurgy Society, xrx (r985), fi6*gz; Killick and
Gordon, 'The mechanism of iron production'; Rostoker and Bronso n', Pre-industrial
Iron. rse Gr6bdnart,'characteristics'; Killick et al.,'Reassessment'.
160 Herbert, Red Gold.
161 Schwellnus,'Short notes';van der Merwe and Scully,'The Phalaborwa story'; der N'Ierrve, 'Advent'. 162 Friede and Steel, ,Notes,.
163 van der Merwe, 'Advent'; Herbert, Red Gold. 164 lbid.
165 van der lVlerwe and Scully, 'l['he Phalaborwa story'. 166 Herbert, Red Gold.
16? B. H. Sandelowsky, 'Prehistoric metal-working in South West Africa', Journal of
the South A/rican Institute of Mining and Metallurgy, Lxxtv Og7d, $S-6.
168 van der Merwe, 'Advent'. 16e Herbert, Red Gold. L7o lbid.
171 Friede and Steel, 'Notes'.
r?s Tylecote, History. 172 Killick et al.,'Reassessrnent'-
174 Fferbert, Red Gold.
The details of furnace construction and iron smelting techniques vary
widely throughout Africa. These differences reflect the diversity in ore types
exploited. They also gave rise to products that vary significantly in form and
carbon content. This technological diversity attests to local experimentation,
regional specialization and a rich but largely unexplored history of tech-
nological change.
Iron smelting can be achieved by a continuum of processes ranging from
the bloomery furnace producing low-carbon iron at one end of the scale to
the blast furnace producing high-carbon cast iron at the other. Steel (i.e. an
alloy of iron r,vith various amounts of carbon up to about z pff cent, which
can be added either in the furnace or in the subsequent smithing process) of
intermediate carbon content can be made in a variety of ways. These include
hybrid smelting techniques, which produce inhomogeneous bloorns of
variable carbon composition; cementation in a reducing atmosphere to
increase the carbon content of bloomery iron; decarburtzatton of cast iron in
an oxidizing atmosphere; and the addition of bloomery iron to cast iron to
dilute the carbon content of the latter.l75 The bloomery process for
producing low-carbon iron has been termed the 'direct' process, and the
routes using cast iron as the starting material have been calied the 'indirect
process',176 while the production of high-carbon blooms in Africa has been
termed the 'African direct steel process'.1?7 The 'African direct steel
process' has since been shown to be a normal variant of steel production in
the bloomery furnace, and the term is redundant.l78 The terms'direct'and
'indirect' are used more precisely to classify the partial reduction reactions
u,hich take place in iron smelting.l?e Direct reduction involves the reduction
of the oxides of iron by solid carbon, in the form of added charcoal or
precipitated lamp black180 according to the following suite of reactions:
+ 4FerOn + COz
+ 6F'eO + COz
> zFe + COz
Indirect reduction takes place in the presence of an excess of carbon
monoxide according to the following reactions:181
3FerO, + CO -- zFeeO+ + CO2
The practical validity of this distinction under furnace operating condi-
tions is doubtful,182 and both sets of reactions may take place in volumes with
175 Rostoker and Bronson, Pre-industrial lron.
1'6 \,an der Nlerwe, Carbon-t4 Dating; Rostoker and Bronson, Pre-indnstrial lron.
177 van der lVlerwe, 'Production of high carbon steel'; van der Merwe and Avery,
' Pathrvays '.
178 Killicli, 'A comparative perspective' ; Killick and Gordon, 'The mechanism of iron
17e G, R. lVlorton and J. Wingrove,'The efficiency of the bloomery process', Bulletin
oJ the Historical Nletallurgy Grottp, III (1969), 66-Z; Bjorkenstam, 'Prehistoric and
medieval iron production'. 180 Morton and Wingrove, 'EfHciency of the bloomery'.
181 lbid.; Rostoker and Bronson, Pre-indttstrial lron.
182 Pleiner, 'Problem of direct steel production'.
6FerO, + C
zFerOn + C
zFeO + C
differing atmospheric conditions in the same furnace. It is realistic to view
the different steel producing processes as part of a continuum and to
recognize that the distinction between the 'direct' and 'indirect' processes
stems from the historical distinction in Europe between the production of
bloomery iron (produced directly) or wrought iron (produced indirectly,
through the decarburization of cast iron).
The metallurgy of the iron bloomery is not simple. A complex set of
interactive mechanisms must operate to various degrees in different parts of
individual furnaces and in furnaces of different designs to accommodate the
observed variations in the ore composition, the furnace operating parameters
and the composition of the final bloomery product. There is still considerable
divergence of opinion about these mechanisms, but the combination of
archaeological description, ethnographic reconstruction and controlled
experimentation is clarifying the picture.183
Indigenous African iron smelting produced material with a range of
carbon compositions ranging from conventional, nearly carbon-free
bloomery iron through steel to rare instances of cast iron.18a It has been
recognized elsewhere from the archaeological record that ancient bloomeries
must have been capable of producing steel,185 and this has been confirmed
through experiment.ls6 F'or a detailed bibliography of experimental bloomery
smelting, see Tylecote and lMerkel.l87
It has in fact been known for a long time that the carbon content of a bloom
r83 lbid.; Morton and Wingrove, 'Efficiency of the bloomery'; R. F. Tylecote,
J. NI. Austin and A. E.Wraith, 'The mechanism of the bloomery process in shaft
furnaces',Journal oJ the Iron and Steel Institute, cctx(r97t),3+2-63; H.M.Friede and
R. H. Steel, 'An experimental study of iron-smelting techniques used in the South
African Iron Age ' , Journal of the South African Institute of Mining and Metallurgy, Lxxvrr
(tgll),233-42; van der ly'Ierwe,'Production of high carbon steel';van der Merwe and
Avery, 'Pathlvays'; D. H. Avery, 'The iron bloomery', in T. A. Wertime and
S. F. Wertime (eds.), Early Pyrotechnology : the EaolLttion of the First Fire-using Industries
(Washington, r gBz), zoS-r4; Clough, 'The bloomery process' ; H. M. Friede,
A. A. Hejja, A. Koursaris and R. H. Steel,'1'hermal aspects of the smelting of iron ore
in reconstructed South African Iron Age furnaces' , Journal of the South African Institute
of Nlining and LVletallurgy, LxxxIV (rg8+), z8S-g7; Bjorkenstam, 'Prehistoric and
medieval iron production'; Avery et al.,'Metallurgy'; Killick and Gordon,'The
mechanism of iron production'; Rostocker and Bronson, Pre-indttstrial Iron.
18{ C. V. Bellamy, 'A West African smelting house', Journal of the Iron and Steel
Institute, lvl (rgo4), gg-rz7; G.H.Stanley, 'Some products of native iron smelting',
SoLtth AJricanJournal of Science, xxvIII (r93r), r3r-4;E. H. Schulz,'Zusarnrrrensetzung
und Aufbau einiger Metallfunde der Afrika-Expedition von Leo Frobenius r9z9/3o',
Poideuma, v (rg5o), 59-65 ; Avery and Schmidt,'Metallurgical Study' ; van der IVIerwe,
'Advent'; P.R.Schmidt,'Further evidence for an advanced prehistoric technology in
Africa' , Journal of Field Archaeology, x ( r g83), 4zt-34; Schmidt and Childs, ' Innovation
and industry'; J.A. Todd, 'Studies of the African Iron Age', Journal of Metals, xxxt
(rct7q),39-45; Killick, 'A comparative perspective'; Killick and Gordon, 'The mech-
anism of iron production'; David et al.,'Bloomery and blast furnace'.
185 Pleiner,'Problem of direct steel production'; R. Pleiner,'Early iron metallurgy in
Europe', in Wertime and Muhly (eds.), The Coming of the Age of lron,375-4r5.
186 Pleiner, 'Problem of direct steel production'; Tylecote et al.,'Mechanism of the
bloomery'; R. F. Tylecote and J. F, Merkel, 'Experimental smelting techniques:
achievements and future', in Craddock and Hughes (eds.), Furnaces and Smelting
Technology, 3-zo. 187 Ibid.
can be varied at will by manipulating the reducing conditions through
control of the fuel to ore ratio and the air flow.188 Outside of Africa this
process is documented for pre*Roman Celtic smelting,lse for South Asia and
the Tatara furnace process in Japanle0 and for the bloomeries in the
Adirondacks in North America.lel All of these applications of the bloomery
process are no longer extant.
African iron smelting took place in bloomery furnaces, producing het-
erogeneous blooms u,ith variable carbon content.le2 The furnaces themselves
display a wide range of forms, some of them adapted to processing specific
ores and some constructed to take advantage of the flexibility of bloomery
smelting to make specific products. Brief summaries of several smelting
reconstructions, described fairly recently in the literature, will illustrate the
Iron smelting by the Dimi of Ethiopia took place in a forced draught, non-
slag-tapping turnace about r metre high with a o'5 metre pit below."' This
rvas filled rvith grass, which was burnt before the furnace was charged with
alternate layers of ore and charcoal. Thirty-six tuybres were used to direct air
from clay pot bellows into the furnace. After four hours the furnace was
allor,ved to cool and the bloom removed for forging in a separate oxidizing
forge. A knife produced from this bloom had a carbon content that varied
from o'8 per cent carbon to r.irtually carbon-free iron.
The furnace used by the Haya tnTanzania was a forced draught bowl shaft
furnace constructed out of slabs of refractory slag bound together with clay.lea
Reeds were burnt in the furnace before inserting the tuybres and charging the
furnace with charcoal and roasted ore. Eight tuybres were placed with much
of their original length inside the furnace, which has led the investigators to
believe that significant preheating of the incoming air had taken place.1e5
Their arguments have been criticized vigorously, particularly on
thermodynamic grounds.le6 The only experimental evaiuation of this pre-
188 Clough, 'The bloomery process' 18e Pleiner, 'Early iron metallurgy'.
1e0 Kiilick,'A comparative perspective';Rostocker and Bronson, Pre-industrial lron.
1s1 Killick and Gordon, 'The mechanisnt of iron production' ; R. F. Allen,
J. C. Dar.vson, NI. F. Glenn, R. B. Gordon, D. J. Killick and R. W. Ward, 'An
archaeological survey of bloomery forges in the Adirondacks', TheJournal of the Society
Jor Indttstrial Archaeology, xvr (r99o), 3-2o; R.B.Gordon and D. J. Killick, 'The
metallurgv of the American bloomery process', Archeomaterials, vt (tggz), t4t 67.
1e2 van der Vlerwe,'Advent';van der Merwe and Avery,'Pathways';Rostoker and
Bronson, Pre-industrial lron.
,,, J.A. llodd and J. A. Charles, 'Ethiopian bloomery iron and the significance of
inclusion analysis in iron studies ' , Journal oJ the Historical Metallurgy Society, xu (, g7 8),
6Z-81; Todd, 'Str-rdies of the African Iron Age'; J.A. Todd, 'Iron production by the
Dimi of Ethiopia', in Haaland and Shinnie (eds.), AJrican lron Workizg, 88-ror.
re+ Schmidt and Avery, 'Complex iron smelting'; Avery et al.,'Vletallurgy'.
1e5 Schmidt and Avery,'Complex iron smelting'; Schmidt and Avery,'More evi-
dence'; Avery and Schmidt, 'Metallurgical study'; D.H.Avery and P. R. Schmidt,
'The use of preheated air in ancient and recent African iron smeltiug furnaces: a reply to
Rehder ' ,Journal of Field Archaeology,xttt ( I g86), SS+-Z; Schmidt and Childs, ' Innovatiou
and industry'.
1eo Eggert, 'Aileged complexity' ; Rehder, 'IJse of preheated air'; J. E. Rehder, 'Reply
to Avery and Schmidt', Journal of Field Archaeology, xrv (1987), rzz; Killick,'A
comparative perspective'.
heating hypothesislsT suffered from the same problems in measuring the air
temperature in tuyires as the field reconstruction and is not considered a
valid test.1e8
The relatively high carbon content of the final bloom product from the
Haya furnace was ascribed to carburization by carbon precipitated as lamp
black in cracks in the ore during the prior roasting step and to entrapped
charcoal lrom the charred grass.lee Despite claims by Avery and Schmidt to
the contrary, the iron crystallizatiorr mechanism and products of this furnace
were not significantly different from those of bloomery furnaces operated
elsewhere.200 The bloom had a carbon content of o'4 per cent carbon to o.6
per cent carbon, and the metal was almost completely decarburized in the
Natural draught furnaces are not confined exclusively to Africa, as there is
one example known from Burma,202 but they were used extensively in the
West African forest, the savanna woodlands of the Sahel and in southern
Central Africa,z03 perhaps as far south as Ztrnbabwe.204 They vary widely in
shape and size but tend to be larger than forced-draught furnaces. Most were
slag-tapping, but some not, and they produced inhomogeneous blooms with
up to o'8 per cent carbon.205 Particularly well described examples are from
oyo in Nigeria,206 Koni in the Ivory Coast2,? and Kasungu in Malawi.2,8
Reconstructions involving two different types of natural draught furnace
have been studied in Malawi. The Cew'a furnace was a 2.5-metre tall, two-
chambered, non-slag-tapping clay furnace with nine tuybres,load,ed with a
lor,v grade lateritic ore and charcoal.2oe After five days, and the consumption
of about r,455 kilograms of charcoal, the furnace produced a bloom of low-
carbon iron which was then resmelted in a small, forced draught furnace.210
The Phoka furnace was a r'5-metre high conical clay furnace with numerous
tuyires and was charged with specifically selected, very low grade laterite
ore.211 The furnace was fired for five days, consuming r,ooo kilograms of
charcoal, and produced a bloom which was broken up and most of it
resmelted in a small forced draught furnace to produce a dense, metallic
secondary bloom of unspecified carbon content. These two-stage processes
are variants that accommodated the low-grade lateritic ore used and enabled
the beneficiation of the ore in the natural draught furnace by separating iron-
rich slag from unreacted gangue.2"
le? S. T. Childs and P. R. Schmidt, 'Experimental iron smelting: a genesis of a
hypothesis r.vith implications for African prehistory and history', in Haaland and Shinnie
(eds.), African lron working, tzr-4r. 1e8 Killick, 'A comparative perspective'.
les Avery and Schmidt, 'Metallurgical study'.
200 Pleiner, 'Problem of direct steel production'; Tylecote et al., 'Mechanism of the
bloomery' ; Eggert, 'Alleged complexity' ; Killick and Gordon, 'The mechanism of iron
production'. 2ol Avery et al.,'lVletallurgy'. 202 Tylecote, History.
203 I{illick, 'A little-known extractive process'.
201 NI. D . Prendergast, ' A new furnace type from the Darwendale Dam basin ', Journal
of the Prehistory Society of Rhodesia, vrr (r975) , t6*zo.
205 Killick, 'A little-known extractive process' ; I(illick, 'The relevance of recent
African iron-smelting practice'. 206 Bellamy, 'West African smelting house'.
207 S. Zacharias and H.-G. Bachmann, 'Iron smelting in lVestern Africa: Ivory Coast',
Journal of the Historical Metallurgy Society, xvu (r983), r-3.
208 Avery et al.,'Metallurgy'; Killick, 'A little-known extractive process'.
eoe Avery et al.,'Metallurgy'.
2rr Wenner and van der lVlerwe, 'Mining' ?to lbid.
2r2 Avery et al.,'Metallurgy'
The Mafa of North Cameroon employed a forced down-draught, single
tuybre furnace described as a hybrid between bloomery and blast furnace.213
The pendanttttybre, suspended from the top of the furnace to direct the air
blast vertically dor,vnwards, lost about a third of its length during the
smelting procedure to contribute to the slagging of the predominantly
magnetite ore. The tuybre had to be trimmed of viscous slag which blocked
the air hole intermittently, and slag was also removed from a series of vent
holes cut into the furnace wall at increasing heights, tracing the build-up of
the bloom beneath the tuybre. After ten-and-a-half hours, and after con-
suming 8z'3 kilograms of charcoal, a rS'7 kilogram bloom mass was
recovered from r8'o kilograms of ore. The bloom varied from ferrite with
less than o'o5 per cent carbon to cast iron with more than 4 per cent carbon.
About half the mass was cast iron, decarburized in the oxidizing forge with
unsorted fragments of the other metallic products in an open crucible. The
final product, manufactured into a hoe, consisted of steel with a carbon
content between o'z per cent carbon and o'8 per cent carbon.214
Attempts at describing the mechanisms operating in African bloomery
turnaces which produce steel products began with that of Harbord, who
recognized that the Oyo steel bloom consisted of partly decarburized cast
iron.215 A hypothesized two-stage mechanism whereby a cast iron product
was decarburized in the Oyo furnace was summarized by Rostocker and
Bronson2l6 as follows.
The high carbon flux was added every two hours. During the period between flux
additions, there was a continuing reaction between the iron oxide-rich slag and the
carbon dissolved in the metal pool below and in contact with the slag pool. The
slag/metal reaction slowly removed the carbon present near the surface of the
gradually accreting lump of iron. As the lump was decarburized, the iron oxide
content of the slag rvas simultaneously reduced to iron, which helped to increase
the metal yield. Since this decarburization reaction generated no heat, fuel had to
be added with the flux to keep both slag and metal molten.
If this interpretation is correct, then the technical distinction between
bloomery and blast furnace is invalid in at least some African contexts.
Four possible mechanisms for the direct production of steel in a bloom
were described by Rostoker and Bronson.2l7
i. Charcoal fragments trapped in the bloom carburize the metal directly.
ii. Reduced iron particles are carburized by gas in the stack, followed by
rapid oxidation by the flame zone so that composite particles falling into
the slag pool separate, with the oxides entering the slag and the carbon-
rich iron cores accreting on the bloom.
iii. High-carbon metal particles are coated with partially reduced slag, which
protects the iron from the oxidizing flame.
ir,. Droplets of cast iron produced on the reducing periphery of the junction
betr,veen the oxidizing high temperature flame and surrounding zone of
carbon monoxide settle through the slag to form localized high-carbon
regions in the predominantly lower-carbon bloom.
The range of carbon dioxide concentration required to avoid the production
213 David et al.,'Bloomery and blast furnace'.
!15 Bellamy, 'West African smelting house'.
lro Rostocker and Bronson, Pre-industrial lron, t38.
2r4 lbid.
217 lbid.
of either bloomery iron or cast iron (i.e. to produce steel) at rzoo oC in a shaft
furnace is very narrow (betr.veen about r per cent and 5 per cent), making the
direct reduction of ore to steel very difficult to control. Thus, it is argued that
it is likely that steel production in the bloomery furnace most probably
results from the in situ secondary decarburization of cast iron.218
A generalized mechanism for bloomery iron production based on the East
African and lVlalawian smelting reconstructions was presented by Avery et
al.zts This involved the initial formation of iron spherules by a process of
accretion into tiny spheres. At higher temperatures the partially reduced
spherules were consumed by rapidty growing iron plates with a pure ferrite
growth front. At yet higher temperatures this structure 'began to organize
itself into islands... These islands gradually assumed a concentric ar-
rangement of platelets parallel to the surface, and the inner plates dis-
solved...leaving dense iron nodules of a few millimetres in diameter with
hollow or internally fissured interiors'.220 A general sequence was proposed
of fine precipitation, platelet formation and nodule development. At higher
temperatures still the smalier iron particles dissolved in the liquid slag and
\,vere reprecipitated onto larger particles. In the forced draught furnaces,
attaining the highest temperature, direct reduction by finely divided carbon
was thought to have taken place, enabling iron to precipitate out of the
supersaturated slag to form large equiaxed grains. Fluctuating temperatures
were seen as essential to this slag solution-transport-reprecipitation mech-
This account is probably erroneous because it failed to recognize the
pisolitic structure of the lateritic ore. The spherical structures described by
Avery et al. could have been inherited from the structure of the ore. The
concentrically banded spherulitic structures illustrated by Avery et al.22r
appear to be pseudomorphous after the concentrically banded iron oxide and
silicate pisoliths and not to be the result of transport and accretion of iron in
the furnace.222 The postulated mechanism of precipitation of iron from the
slag is feasible, but it is not necessary to explain the observed microstructure
of the bloom. Large equiaxed crystals are typical of iron cooled slowly from
high temperatures, as these blooms were.
Killick and Gordon223 have presented an alternative description of the
mechanisms operating in the bloomery process, drawing on examples from
Kasungu in lVlalawi, from the historic Adirondack bloomeries in New York
State and from an experimental bloomery at Colonial Williamsburg. Their
concise description of the bloomery process warrants being quoted in full.22a
The accepted theoretical basis of the bloomery process is that the ore is reduced by
carbon monoxide in the sequence, hydrated iron oxides - haematite - magnetite -
wustite [sic] - iron. The equilibrium conditions for complete reduction are attained
r,vhen the temperatures are greater than 8oo oC and the CO/CO, ratio greater than
2t8 lbid. 21e Avery et al.,'Metallurgy'. 220 lbid. 276. 22r lbid. z7B.
222 The reduced pisolith illustrated in Fig. r of Killick and Gordon ('The mechanism
of iron production', rzz), captioned 'Kasungu, Malawi. Concentric shells of ferrite,
pser.rdomorphous after a laterite pisolith, in slag of fayalite, glass and undissolved qvartz' ,
isalmostidenticaltothoseinFigs. z4.5cand24.5dof Avery etal.('Metallurgy',278),
captioned 'Concentric formation of platelets into protonodules, Chulu ng'anjo' and
'Hollow nodule' respectively.
sP3 Killick and Gordon, 'The mechanism of iron production' 224 lbid. rzo.
75 per cent. To effect separation of the iron from the gangue in the ore, some of
the iron oxide in the ore is sacrificed to form a fluid slag with a composition near
that of fayalite so as to have a solidus temperature of about rzoo oC. If ali of the
gangue is to be removed in this way, and if slag-forming reactions approach
equilibrium, the ore must contain at least 49 per cent iron for there to be any yield
of metal. Because reduction of iron oxide to iron can be carried out in a bloomery
at temperatures too low for elements such as manganese, silicon, and titianium that
may be present in the ore to be reduced, relatively pure iron can be produced.
Since the fuel used is alr,vays charcoal, bloomery iron is usually free of sulfur but
phosphorus in either the ore or the fuel ash may be reduced and enter the iron.
Either steel or iron can be made by adjusting the ore/fuel ratio and the
According to Killick and Gordonzzs the Malawian bloomeries were used
for smelting low grade laterite ores with 20 per cent to 3o per cent elemental
iron, too low to have produced any metallic iron if equilibrium conditions
obtained in the furnaces. The concentrically banded pisoliths were reduced
by carbon monoxide successively to magnetite-wristite-iron and formed
concentric bands of metallic iron in a nodular structure directly pseudo-
morphous after the original pisolith. The fluid slags formed subsequently
from the silicate groundmass consisted of fayalite, glass and unreacted quartz
and drained easily from the reduced iron, which coalesced due to surface
tension. The accretion and consolidation of the iron to form a bloom of
variable carbon content took place under the r,veight of the material above it
in the stack. This was summarrzed as an example of simple direct gas
reduction of a porous lateritic ore.226
The diversity of mechanisms and the inherent