Journal of African Archaeology Monograph Séries Vol. 2

Crossroads / Carrefour Sahel Cultural and technological developments in first millennium BC / AD West Africa

Développements culturels et technologiques pendant le premier millénaire BC / AD dans l'Afrique de l'Ouest

Edited by Sonja Magnavita, Lassina Koté, Peter Breunig& OumarouA. Idé

r

Iron Metallurgy in West Africa: An Early Iron Smelting Site in the Mouhoun Bend, Burkina Faso

Augustin F.C. HoU

Abstract The debate on the development of metallurgy in West Africa is a particularly interesting one. The three field seasons conducted by the Mouhoun bend archaeological project (MOBAP) team from 1997 to 2000 adds some new éléments to this dossier. A number of archaeological sites, ail consisting of more or less extensive mound-clusters, were surveyed, mapped, and a handflil excavated. AU the surveyed sites appear to be structurally similar and consistent, made of three distinct components: ( 1 ) A quarry for the procurement of both iron ore and building material; (2) spécial purpose mounds (eg: iron smelting, forge, cemetery, cloth dyeing; and, (3) generalized habitation mounds. The dynamics which generated and sustained the formation of thèse mound-clusters is still poorly understood. There are nonetheless a number of suggestions, ail mostly articulated on more or less extensive ethnographie analogs. This paper outlines some of the key issues in the archaeology of West African metallurgies and focuses on the new data from Tora Sira Tomo 1 (TST-1), an elaborate iron smelting site dated to the middle of the first millennium BC. It évaluâtes the importance of this new find in relation to the broader debate on the émergence of metallurgy in West Africa.

Résumé Le débat sur les origines de la métallurgie en Afrique de l'ouest est particulièrement intéressant. Trois campagnes de fouille menées dans la cadre du "Projet Archéologique de la Boucle du Mouhoun de 1997 a 2000 apportent de nouveaux éléments a ce dossier. De nombreux sites archéologiques organisés en ensembles de buttes anthropiques ont été prospectés, mesurés et relevés, et quelques-uns ont été fouillés. Tous les sites prospectés présentent des éléments récurrents: 1) une carrière pour l'obtention du minerai de fer et de la latérite pour la construction des planchers; 2) des buttes a fonction spéciale (réduction du fer, forge, travail des textiles, cimetière); et 3) des buttes d'habitation ordinaire. Les raisons sociales de la formation des ensembles de buttes demeurent obscures. Il existe cependant un certain nombre de suggestions, toutes plus ou moins inspirées des données de l'ethnographie récente. Cet article présente quelques points clefs sur l'archéologie des metallurgies Ouest-Africaines, expose les données nouvelles du site de Tora-Sira-Tomo I (TST-1), un atelier de réduction du fer datant du milieu du premier millénaire BC, et discute de l'importance de cette découverte dans le cadre du débat sur l'émergence de la métallurgie du fer en Afrique de l'Ouest.

INTRODUCTION

Research on West African early metallurgies is particu­larly fascinating and went through cycles of growth and slowdowns. The first growth spur took place in the 1970-1980s with the remarkable discoveries of Nicole LAMBERT in Mauretania (1975, 1983), the surveys and test excavations conducted by Quéchon and colleagues (PARIS et al. 1992; PERSON & QUÉCHON 2002; QUÉCHON 2002) in the Termit massif, and Danilo Grébénart in the context of the Programme Archéologique d'urgence de la région d'In Gall - Teggida-n-Tesemt (GRÉBÉNART 1985,1988). EssoMBA (1993) and CLIST (1989) provide good summaries of the state of the art on the archae­ology of metallurgy in Central Africa in the 1980s. They reviewed the material collected from sites in the Saharan and Sahelian part of the Chad Republic, and the rainforest in Southern Cameroon and Gabon. The discoveries from Taruga and Samum Dikuya in the Nok Culture area of Nigeria and the more accurate re-assessment of the antiquity of Meroe iron industry

opened the gâtes for a totally new perspective on the genesis of African metallurgical traditions (ECHARD 1983; HAALAND & SHINNIE 1985) .

This first peak of archaeological effort was fol-lowed by a period of critical évaluation (KILLICK et al. 1988; MCINTOSH & MCINTOSH 1988; KILLICK 2 0 0 4 ) .

A re-assessment of Niger fumaces radiocarbon dates showed a number of dates to be erroneous. This fact that prompted KILLICK et al. ( 1 9 8 8 ) to write that " the only positive évidence for metallurgy for this région... is a single radiocarbon date of 1 7 1 0 ± 110 B.C. (Gif 5 1 7 6 ) for a copper-working fiamace....We suggest that this date be viewed with great caution until it can be corroborated by another method such as thermolumi­nescence dating of the fired lining of the fumace.. .Until thèse are available, the évidence for metallurgy in Niger prior to 1000 B.C. remains in doubt".

The core of the issue is the reliability of radiocar­bon dates in the archaeological research on early West

59

A.F.C. Holl

African metallurgies. For KILLICK et al. (1988), the "old wood" problem is "so pervasive that attempts to infer the origins of metallurgy in Africa from the prés­ent radiocarbon base should probably be suspended" (in MCINTOSH & MCINTOSH 1988: 106). This loud and

clear call for a research moratorium is however not bal-anced by any practical suggestion on how to proceed from there. Enthusiastic researchers are simply asked to stop doing research while waiting for an expert to come, clear the way, and give the "green light". The handful of existing archaeometallurgical research fa-cilities are not only extremely difficult to access for "foot-soldiers" of West African archaeology, but also utterly expensive for the tiny research budgets gener-ally available. Those in control of thèse laboratories, in India, Germany, France, United Kingdom, and the Unités States, wield a tremendous influence on the nature and pace of progress that can be achieved in the field of African archaeometallurgy.

Another facet of this critical stance is developed by MCINTOSH & MCINTOSH (1988: 107) and stemmed from

the availability of a high-precision calibration curve for the years 250P BC to AD 1950. "The dramatic fiatten-ing to the calibration curve in the mid-fîrst millennium B.C. means that the majority of the dates for early met­allurgy in West Africa caimot be resolved to better than 800 - 400 B.C." {ibid). Radiocarbon dating techniques are being constantly improved. It is now possible to nm direct radiocarbon dates on iron artifacts themselves, a method tested on three iron artifacts from the Mouhoun Bend (COOK et al. 2003). Furthermore, some sources of pollution can be detected and neutralized but others are still difficult to handle. The assumption running through MCINTOSH & MCINTOSH (1988) papers is that ail early cases of West African metallurgies date from the middle of the first millennium BC, a time segment characterized by a flat calibration curve which makes any précise chronological assessment impossible. For sake of cohérence, i f the "flatness" of the calibration curve is only confined to the 800 - 400 BC time seg­ment, it means that dates of metallurgical installations older than 800 BC and younger than 400 BC are ac­curate. Considering the rather loose archaeological coverage of West Africa, the scale and resolution of surveys, and the sheer number of excavations which need to be done, one would expect scientists to be open minded because so few is known. The 500 BC base-line for the émergence of West African metallurgical traditions has no sound empirical foundation. Its only justification is to make the assumption of Carthaginian origins plausible.

The second growth peak in the archaeology of West African metallurgies, following the rejection of

early radiocarbon dates for fumace installations in the Termit massif, the In-Gall Teggida-n-Tesemt area, and Southern Cameroon rainforest, occurred in the 1990 - 2000s. New research was conducted in the Termit massif (PERSON & QUÉCHON 2002; QUÉCHON

2002) wi th the explicit goal of coUecting high quality samples for radiocarbon dating. Organic temper was extracted from pottery, charcoal samples were collected from the fumaces, and dates accuracy was assessed comparatively, i.e., when pottery dating was cohérent with fumace dating. The stunning discoveries of an original iron-smelting tradition in the Nsukka area of Nigeria (OKAFOR 1993, 2002), the finds from northem Mandara (MACEACHERN 1996), as well as ZANGATO

(1999) research in northwestem Central African Re­public were aspects of this new research round. This new wave of fieldwork confirmed the trends already emerging from the first growth spur. Iron metallurgy appears to date from 2000 to 600 BC in ail four study areas mentioned above.

The third round of interesting research is going on right now. Archaeological research on early met­allurgies has witnessed a number of interesting and surprising discoveries in the last few years. New work in the Middle Sénégal Valley, at Walalde in Morphil Island, has revealed an early iron smelting tradition dating as far back as ca 800 to 550 BC (DEME 2003; DEME & MCINTOSH 2006). The excavation probe did not

contain any direct évidence of iron smelting installation but with 47 tuyères fragments and 19 kilograms of slag collected, the iron-smelting workshop was very likely présent nearby but somewhat out of the scope of the excavation probe. The analysis of smelting débris done by Dr. D. Killick points to fumace température peaks ranging from 1200 to 1300° Celsius (DEME &MCINTOSH 2006: 336).

Sustained fieldwork resulted not only in the con­firmation of prior findings but also in the discovery of still older iron processing facilities at Oboui, located at 6° 03' N / 15° 20' E, at 1048 m above sea level in the eastem extension of the Adamawa plateau in Central Africa, and dated to ca 2200 - 1965 BC (SALIÉGE 2007: 135; ZANGATO 2007). Oboui is a small residential site of iron using communities "discovered" in 1992 after torrential rains eroded most the capping sédiments (ZANGATO 2007: 11). It took eight field seasons of two to nine months from 1992 to 1996 to probe some 800 m^ of the site's deposit. The data obtained from the excava­tion so far offer interesting insights into ail aspects of the chaîne opératoire of iron production but the initial mining and iron ore smelting. The site was used, very likely on an intermittent basis, from 2340 - 2058 BC to AD 346 - 544. The features of archaeological inter-

60

An Early Iron Smelting Site in the Mouhoun Bend, Burkina Faso

est include forge fumaces, hearths, trash pits, charcoal storage pits, and anvils, ail arranged in a more or less cohérent forge workshop.

Research conducted at Dekpassanware in north central Togo by P. DE BARROS (2003 ,2006 ) also produced surprising results. The researched site measures some 30 hectares in surface extent with 1.80 to 2 . 1 0 m thick archaeological deposit. The site consists of slag concen­trations, bare latente zones, one recorded burial pit, and bumt wattle and daub remains. The bottom deposit which appears to have accumulated between ca 800 and 400 BC attests to a Late Stone Age occupation. The top deposit is dated to a Late Iron Age occupation that took place between AD 1300 and 1600. The intermediate deposit contains concentrations of slag, iron ore, tuyères frag­ments, as well as poorly preserved iron artifacts belong-ing to an Early Iron Age occupation. The latter seems to have occurred between ca 4 0 0 BC and AD 100, " 1 0 0 0 years older than expected" (DE BARROS 2003: 75).

Bena, at 12° 0 4 ' 0 5 " N and 4° 1 1 ' 0 2 " W is lo­cated in the Bwamu in Westem Burkina Faso. The site includes a séries of natural draft semi-subterranean smelting fumaces dated to 3 6 0 - 2 2 0 BC (COULIBALY 2006; KJÉTHÉGA 2006) . Looked at from a broader West

African perspective, thèse new discoveries add sig-nificantly to the difficult problem of the émergence of West African metallurgies (HERBERT 1993; ZANGATO 1999, 2007 ; BISSON et al. 2000 ; BOCOUM 2002 , 2 0 0 6 ;

PRINGLE 2009) . The Niger, Nigeria, and Central African Republic early iron producing sites ail point to diverse craft traditions which started to develop at the begin-ning of the second millennium BC.

TECHNOLOGICAL INNOVATION

The dominant "Research and Development" — RAND — approach that characterizes contemporary tech­nology tends to distort our approach to early pristine technological developments. The advent of both copper and iron metallurgy have generally been conceived as the resuit of a drive toward greater efficiency, with such technological irmovations claimed to have had quasi-revolutionary conséquences.

"In the Southem half of Africa the story is consid-erably différent. Here the coming ofiron was a catalyst which woke up half of the continent from the slumber of the Stone Age. The food producing révolution of the Near-East, which set the stage for the ensuing Bronze Age, had penetrated as far as North Africa, the Nile Valley, and the Red Sea Coast." (VAN DER MERWE 1980:

464) . [emphasis mine].

Technological innovation is not always goal ori-ented. It took a number of years, i f not décades for some spectacular contemporary technological breakthroughs like X-ray, Laser, and computers to reach their fijll po-tentials (LEMONNIER 1992, 1993; LATOUR 1993) . This

means that technological innovations, borrowing, or émulation have to make social sensé. "Technologies are not only things and means used by societies to act upon their physical environment. For the ethnologist, and for the archaeologist and historian as well, technologies are — like myths, marriage prohibitions, or exchange Systems — social production in themselves" (LEMONNIER 1 9 9 2 : 1 - 2 ) . Technological Systems have a certain cohér­ence and différent levels of compatibility. They mobi-lize a number of variables, principally: matter, energy, objects, gestures, and spécifie knowledge (LEMONNIER 1992, 1993). The simple présence of a technique does not guarantee its fiill scale adoption. There are at least three sides to an innovation/invention. 1 ) Its potential can be grasped as a positive move and adopted quickly. 2 ) It can be felt as a threat for the social order and in that case, suppressed. 3 ) But most of the time, pristine innovations/inventions, could have been felt as neutral. Through cycles of trials and errors, an un-noticed tech­nological innovation can be refined and transferred to many other domains of application. The production of métal artifacts appears to have started with the manu­facture of status-related items, weapons and éléments of Personal adomment. It is only later in the séquence, when the production protocols were "routinized" that heavy-duty tools started to be manufactured. It is at this later stage that iron metallurgy had an impact on land clearings and agriculture in gênerai. Présent day ar­chaeological and archaeometallurgical research tend to focus on a very narrow portion of technological Systems. The issue is clearly a difficult one and one has to proceed from conjecture to refiitation, over and over again.

THE MOUHOUN BEND ARCHAEOLOGICAL PROJECT

The Mouhoun Bend Archaeological Project (NW Burki­na Faso) was designed to investigate the development of food-producing économies in relation to Holocene climatic change and the émergence and amplification of craft specialization. This part of West Africa was liter-ally an archaeological Terra Incognita when this project started in 1997. The Mouhoun River (formerly known as Black Volta) flows from the SW to NE, then winds its course in a U-shape bend to follow an almost N-S direction {Fig. 1). The Study area is delimited in the North and northeast by this meandering river course. The selected area measures 4 0 km East-West (3° 1 1 ' / 3° 3 2 ' longitude East) and 38 km North-South (12° 30 ' / 12° 4 5 ' latitude North). The land is flat in gênerai, with

61

A.F.C. HoU

iJ^v Approx. southem limil of désert fWK* Presenl-day northcni boundan of Ihick forcsls

Fig. 1. Location map of the Mouhoun Bend and study area.

altitude varying from 294 to 249 m above sea level. The végétation is characteristically a highly anthropic wooded savanna, with the protected shea butter tree {Butyrospermum parkii) largely prédominant, followed by différent kinds of Euphorbiacea.

to the Mouhoun river may have been a thriving trade "highway". There is im-pressive évidence of iron metallurgy ail over the study area. It consists of iron ore mines, some with complex System of connected galleries and tunnels, open-air mines and quarries, smelting sites with large amount of slag, fumaces installa­tions, and habitation sites, with varying quantity of iron artifacts.

A number of sites distributed ail over the Mouhoun Bend landscape have been mapped and tested (HOLL & KOTÉ 2000).

Late Stone Age scatters of lithic artifacts have been recorded on the Mouhoun river shore, and four mound clusters have been

tested. They range in chronology from ca 760 - 210 BC to A D 1410 - 1650 at Tora Sira Tomo, ca 350 BC - AD 120 to AD 1290 - 1450 at Kerebe Sira Tomo, and ca AD 440 - 750 to AD 980 ~ 1260 at Diekono {Tab. 1). This paper focuses on the iron-metallurgy side of the project, and more precisely on an early iron-smelting workshop found at TST-1.

The project examines change in subsistence Sys­tems characterized by a shift from hunting and gather-ing to food-producing societies, the onset, extension and intensification of iron production, and long-term cultural change. As shown by survey and test excava­tions carried out in the summer of 1997, the préserva­tion of the archaeological record is generally good. The results of the first séries of radiocarbon readings indicate that the tested mounds range in time from 7/600 BC to AD 1400 (HOLL & KOTÉ 2000). Up to a

certain point, it is clear that iron metallurgy is attested in ail the tested mounds, from the bottom to the top of the exposed sfratigraphic séquences.

The relatively rich local oral fradition emphasizes large scale migrations and movements of population from the core of the AD 1200-1400 Mali Kingdom along the Inland Niger Delta (the Marka) toward the periphery where the study area is located. The Marka (a Mande speaking ethnie group) are claimed to have been specialized merchants lineages [Dyula], and may have been involved in long-distance trade between the periphery and the core of the Mali Kingdom. It is well known that there is no iron ore in the Inland Niger Delta (MCINTOSH & MCINTOSH 1980; MCINTOSH

1995; TOGOLA 1996). Iron blooms and probably tools were imported from surrounding lands. The Mouhoun Bend may have been one important supplier of metals. The Sourou river that connects the Inland Niger Delta

Tora Sira Tomo mound complex

Tora Sira Tomo (TST) mound complex is located at approximately five kilometers southwest of Kerebe Sira Tomo. It is the largest of the settlement complex found in the study area with 17 distinct mounds spread over 900 m west-east and 500 m north-south. TST-3, the largest single mound of the whole complex is stretched along the northem edge and measures 260 m in length west-east, and 120 m in maximum width north-south. A i l 17 mounds of the settlement complex were tested after three field seasons which took place in 1997, 1999, and 2000. Four, TST-1, TST-2, TST-4, and TST-9 [Ring Site], are spécial purpose sites. The first is an iron-smelting workshop, the second a latérite quarry for both iron ore and constraction material extraction, the third a cloth processing and dyeing site, and finally, the fourth, a spécial burial site. The remaining thirteen sites appear to have been standard habitation mounds with varying intensity and length of occupation.

TST-1: The Iron-smelting workshop

Tora Sira Tomo (TST-1) is located at approximately two hundred meters southwest of the main mound (TST-3). It is sub-circular in shape, measures 18 m in length, 15 m in width and 1.30 m in maximum height. The total of the excavated sample amounts to 65 m^

62

An Early Iron Smelting Site in the Mouhoun Bend, Burkina Faso

gi(j Number of Mounds Excavation units Chronology

LSA Scatters Diekono

Tora-Sira-Tomo (TST)

Kerebe-Sira-Tomo (KST)

Total

6 17 8

31

4 sampled 3

18 11 32

unknown cû AD 550- 1250

ca 760 B C - A D 1650 ca350BC-AD 1450 ca 760 B C - A D 1650

Tab. 1. General outline of the mound-clusters excavated in the Mouhoun Bend (1997-2000).

and covers the whole range of archaeological remains associated with iron-smelting. A large number of clay vessels set in upside down position were found dis-posed off along the northem part of the iron-smelting complex {Fig. 2). The excavation of this portion of the site was divided into several sub-units, A in the north-west, E in the center-east, B and F in the east.

Probe A located in the northwest edge of the site measures 2 by 2 m. The archaeological deposit is 0.25 to 0.30 m thick and is comprised of three sub-horizontal layers: Layer 3: (0.00-0.15 m) Dark brown loose silty clayey

topsoil. Layer 2: (0.15-0.20 m) Compact and hard brown-

yellow clay. Layer 1 : (0.20-€.30 m) Brown-reddish latente gravels.

The uncovered archaeological remains were foimd in layer 3, deposited on layer 2 surface. They consist of five clay vessels set in upside-down position at 0.50 to 0.30 m from one to the next. The lower part of the recorded vessels was not preserved. They ail belong with minor variations to the same shape/size category, that of con-stricted neckpots with everted rim, decorated with twisted roulette impression. The mouth diameter ranges from 24 to 22 cm and wall thickness from 7.5 to 10 mm.

Probe B in the east measures 6 m^. Its stratigraphie séquence is similar to that of probe A and contains a concentration of 13 clay vessels ail found in upside down position. The curvilinear arrangement of the uncovered pottery suggests the présence of a round hut/shelter that may have measured 2.20 to 2.50 m in diameter. Three additional pots are found on the east fiank of the main concentration. The pottery sample from Probe B is more diverse in shape and size. It in­cludes a small pot with straight rim, globular pots with constricted neck and everted rim as well as a flat base spécimen. Décoration pattems which combine twisted roulette impression, horizontal, oblique, and vertical grooved Unes, and herringbone motifs are richer.

Probe E, oriented NW-SE, measures 5 m in length and 3 m in width and contains a total of 22 vessels more or less evenly distributed ail over the tested

15 m^ {Fig. 2). The stratigraphy is similar to that of probes A and B, with however a thinner topsoil that was probably more eroded on this gently sloping part of the site. The recorded pottery sample is comprised of two main size/shape catégories; one made of globu­lar to slightly elongated vessels with more or less constricted neck and flared rim, generally decorated with roulette impression and grooved Unes. The other includes globular pots with inverted rim with décora­tion made of herringbone, incised wavy or straight Unes, confined in most cases to upper part of the ves-sel. There are very few significant variations in the characteristics of the sherds collected in the archaeo­logical deposit from Probe E. The gênerai distribution of vessels parts, decorarion techniques, as well as sherd thickness présent a similar pattem of variation in both layers suggesting Utile change in the material culture of the user of this iron-smelting site.

Probe C — the flimace probe — is located at the center of the site and cuts through the thickest portion of the archaeological deposit. It measures approxi­mately 40 m^ and consists of a 7 x 2 m trench oriented NW-SE abutting a larger SW-NE 7.5 x 4 m excavation unit. Probe C stratigraphie séquence measures 1.50 m in maximum thickness corresponding to the highest élévation of the whole site, and made of 4 layers:

Layer 4 (0.00-0.60 m): Brown-grey silty clay topsoil partly termite mound.

Layer 3 (0.60-0.80/0.90 m): Light brown grey sédiment from the coUapsed fiimace, including slag, blow pipes, and flimace bricks fragments.

Layer 2 (0.80/0.90-1.30 m): Brown-yellow hard and compact clay.

Layer 1 (1.30-1.50 m): Light brown grey sédiment with latérite gravels.

The base of a relatively large iron-smehing fumace was found at 0.45 m below the surface {Fig. 3). A line of red bricks found along the trench section suggests that the uncovered fumace may have measured at least 3 m in height. A flre hardened surface was exposed on the eastem side of the fumace which mouth was ori­ented northeast. The remaining part of the flimace mea­sures 1.80 m in diameter. Its wall, 0.20 m thick, buih

63

A.F.C. Holl

Fig. 2. View of the pottery left on the north fiank of TST-1 iron-smelting workshop.

Fig. 3. View of the eight tuyères arrangement.

with superimposed irregularly shaped clay lumps in two layers, an inner and an outer one, was preserved up to a height of 0.25 m. A group of broken and com­plète tuyères was exposed on the north fiank of the fumace, and a set of three clay vessels was exposed on the fire-hardened surface on the east fiank.

Slag, bricks fragments, and broken tuyères filled the inner part of the fiimace. This deposit was accumulated on top ofa tuyères level at 0.60 m below the surface above a relatively thin bottom slag. Thèse eight relatively well preserved tuyères converge to the center of the flimace with in the middle a mass of bumt clay, bricks, and slag, capped with a whitish 5 to 10 cm thick circular chalky deposit {Fig. 4). The preserved segments of the eight tuyères are more or less organized into pair. They measure 30 to 40 cm in length and 10 cm in diameter at the proximal end and 5-6 cm at the distal one. The more or less bal-anced arrangement of the tuyères coupled with the northeastem orientation of the fumace mouth suggest that this installation was a natural draft one.

The circular white chalky deposit at the center of the fumace is presumably made of the remains of the fluxing material used in the smelting process. It is not yet known i f this was an intentional addition to the furnace, or the resuit of the types of wood used as fuel (HAALAND & SHINNIE 1985; SCHMIDT 1996). In any case, fluxing material optimizes the

64

An Early Iron Smelting Site in the Mouhoun Bend, Burkina Faso

Fig. 4. The base of the fumace view from the Northeast,

use of fuel by lowering the température at which the iron ore starts to melt. Surprisingly, the bottom slag is relatively small in size, and the structure of the fumace finally appears much more complex and interesting than thought previously. The remaining portion of the fumace was still to be found 1.20 m below the surface {Fig. 5), some 0.60 m below the level of horizontally laid tuyères. An underground chamber, tronconically shaped, was dug in the natural brown-gray silty clay deposit. It measures 0.60 in diameter at bottom with the base and wall lined with a mixture of cmshed latente gravel and clay. Fourteen vertical but slightly tilted tuyères arranged into two distinct sets were found in this part of the fumace installation. The westem set has six tuyères, and the eastern one eight {Fig. 5), with the central space filled with termite nest material. The tuyères found at the bottom of the fumace had no direct connection with the combustion chamber. The six spécimens from the westem set measure 26 to 14 cm in length and 10 to 13 cm in maximum diameter. The eight from the eastem set were longer on the average, 26 to 40 cm with a narrow diameter range of 11-12 cm. Further work on thèse tuyères has shown that they were not used in any smelting process yet and were filled with sédiment that was easily removed. This suggests that the iron-smelters had developed a System allowing for the production of ready to use blow-pipes, clearly an important economy of scale. A new supply of dry tuy­ères was set to be fired below the combustion chamber taking advantage of the high températures generated by

the fiimace during the iron ore smelting process. Once the process is completed, the flimace is left to cool, the bloom is collected and the new load of fired tuyères is recovered to ran the next shift.

The fumace stmcture uncovered at TST-1 is a rela­tively sophisticated pièce of craft engineering. There is no known case in African literature on metallurgy, in archaeology as well as ethnography (BISSON et al. 2000; COULIBALY 2006; KIÉTHÉGA 2006). The pyrotechnological performance of the whole installation has yet to be studied in détail. The fiimace was a natural draft one, operating without below blowers on the natural strength and per-sistence of the dry season northeastem wind (the Harmat­tan). That is why the fumace mouth is oriented NE.

However, the greatest surprise of the excava­tion of TST-1 smelting site is the radiocarbon date obtained from a large charcoal sample collected on the fire-hardened surface at less than one meter from the fumace mouth. The reading shows the smelting site to date to (ISGS - 4349, 2360 ± 70 bp), 501 - 386 B C (1 sigma) or 761 - 212 B C (2 sigma). This is clearly an early case of iron-smelting fumace in this part of West Africa, west of Niger-Nigeria where earlier iron-smehing sites have already been recorded (GRÉBÉNART 1985, 1988; OKAFOR 1993, 2002; HOLL 1997; HOLL & KOTÉ 2000). The technical expertise involved in the conception, constmction, and opération of the TST-1 fumace is very impressive indeed.

65

A.F.C. HoU

Fig. 5. The tuyères firing chamber un-demeath.

Probe D is located in the southwest of the smelting site and measures 2 by 1 m. The test has shown ail the southem half of the site to be a slag heap where ail the by-products of the smelting process were discarded. The exposed stratigraphie séquence is 1.00 m thick made of three levels:

- Level 3 (0-0.50 m): light brown grey sédiment made of slag, blow-pipes, and fumace wall's bricks.

- Level 2 (0.50-0.80 m): compact brown-yellow clay. - Level 1 (0.80-1.00 m): Brown-grey clay with gravels.

A number of miscellaneous finds were recovered from level 3 in Probe D, one hammer-stone in dolerite and a 8 cm long fragment of an undetermined iron object.

The site seems to be partitioned into two parts, with the fumace in the middle dividing east-west line. The set of complète or near complète vessels were found exclusively in the north half and iron ore smelt­ing by-products were discarded in the south half. This site layout can be explained by the intermittent, very likely seasonal, use of the smelting site. At the peak of the production season, the natural draft fumace can be in opération for days. Working crews take shifts in order to keep the momentum and as such may have camped nearby in lightly built shelters. The recorded pottery is predominantly made of relatively large liquid containers that provided the working crews with neces-sary beverages. The variation in décoration pattems can have many probable explanations: On the one the hand, working crew members may have been recmited from différent surrounding settlements, with some serviced

by différent potiers. They bring with them the material they use in their respective villages. In this case, the recorded diversity reflects the elasticity of the actual pottery offer. On the other hand, the smelting site may have been in use for a relatively long period that may have witnessed some variation in the pattems of pottery décoration. A varying combination of both suggestions outlined above is also a genuine possibility.

CONCLUSION

The data recorded in this part of the Mouhoun bend add a significant novelty to the debate of the development of West African metallurgies. The iron-smelting fiimace of TST-1 points very directly to a local case of technological innovation that took place in the middle of the first millen­nium BC, ca 700 - 200 BC. The local expert smelters have designed a dual-purpose fumace that produced important economy of scale. On windy days of the West African dry seasons, the Harmattan — northeasterly — can blow for days without intermption. Natural draft fumaces were de­signed to take advantage of this "wind-power" and be able to operate shift after shift without significant interruption but the removal of the bloom and the retrieval of a new batch of ready to use blow-pipes. After several épisodes of exposures to high températures, the tuyères, essential components of the venting System can crack and break, and/or be clogged by liquid slag. They could be replaced without the need to hait the smelting shift in opération. This stunning innovation took place during the middle of the first millennium BC and supports the previous finding from Bena, a site with natural draft iron smelting fumaces also located in the Mouhoun River valley.

66

An Early Iron Smelting Site in the Mouhoun Bend, Burkina Faso

ACKNOWLEDGMENTS

Fieldwork upon which part of the paper is based was funded by the Centre National de la Recherche Scien­tifique (CNRS), France; National Géographie Society Grant #6378-98, and the University of Califomia, San Diego start up funds. I am gratefiil to Dr. Lassina Kote for his help in the field and the opportunity he offered to conduct an exciting archaeological research project in the Mouhoun Bend.

REFERENCES

Ballouche, A., Kûppers, K., Neumann, K. & Wotzka, H.R 1993. Aspects de l'occupation humaine et de l'histoire de la végé­tation au cours de l'Holocène dans la région de la chaîne de Gobnangou, SE Burkina Faso. Berichte des Sonderforschun-gsbereichs 268 1, 13-31.

Barros, R de 2003. Récent Early Iron Age research in Bassar, Togo. Nyame Akuma 59, 76-78.

Barros, R de 2006. The surprising Early Iron Age site of Dekpassanware (Togo). Paper presented at the Society of Africanist Archaeologist Meetings, Calgary, June 22-26, 2006 (unpublished).

Bisson, M.S., Childs, S.T, Barros, R de & Holl, A.FC. 2000. Ancient African Metallurgy: The Sociocultural Context. AltaMira Press, Walnut Creek.

Bocoum, H. 2006. Histoire technique et sociale de la métallurgie du fer dans la Vallée Moyenne du Sénégal. Thèse Doctorat d'Etat, University Cheikh Anta Diop, Dakar.

Bocoum, H. (ed.) 2002. Aux origines de la métallurgie du fer en Afrique: une ancienneté méconnue. Editions de l'UNESCO, Paris.

Chst, B. 1989. Archaeology in Gabon 1986-1988. The African Archaeological Review 7, 59-96.

Cook, A.C., Southon, J.R. & Wadsworth, J. 2003. Using radiocar­bon dating to establish the âge of iron-based artifacts. http:// H-ww.tms.ore/pubs/iournals/JOM/0305/Cook-0305.html

Coulibaly, E. 2006. Savoir et savoir-faire des anciens métal­lurgistes d'Afrique. Karthala, Paris.

Deme, A. 2003. Archaeological investigations of settlement and emerging complexity in the Middle Sénégal Valley. Unpub­lished Ph.D. Thesis, Rice University, Houston.

Deme, A. & McIntosh, S.K. 2006. Excavations at Walalde: new light on the settlement of the Middle Sénégal valley by iron using peoples. Journal of African Archaeology 4 (2), 317-347.

Echard, N. (ed.) 1983. Metallurgies africaines: nouvelles contri­butions. Mémoire de la Société des Africanistes, Paris.

Essomba, J.M. 1993. Le frr dans le passé des sociétés du sud Cameroun. L'Harmattan, Paris.

Grébénart, D. 1985. La région d'In-Gall - Tegidda-n-Tesemt (Niger) II: le néolithique final et les débuts de la métallurgie. Etudes Nigériermes 49. Niamey.

Grébénart, D. 1988. Les premiers métallurgistes en Afrique oc­cidentale. Editions Errance, Paris.

Haaland, R. & Shinnie, RL. (eds.) 1985. African Iron Working: Ancient and Traditional. Norwegian University Press, Oslo.

Herbert, E.W. 1993. Iron, Gender, and Power: Rituals of Trans­formation in African Societies. Indiana University Press, Bloomington/Indianapolis.

HoU, A.F.C. 1997. Metallurgy, iron technology and Afncan late ho­locene societies. hi: Klein-Arendt, R. (éd.), Traditionelles Eisen-handwerk inAfrika. Heinrich Barth Institut, Kôln, pp. 13-54.

Holl, A.F.C. 2002. The LandofHoulouf: The Genesis ofa Chadic Chiefdom. Mémoire of the Muséum of Anthropology. Uni­versity of Michigan, Ann Arbor.

Holl, A.F.C. & Koté, L. 2000. Settlement pattems, food produc­tion, and craft specialization in the Mouhoun Bend (NW Burkina Faso: preliminary results of the MOBAP 1997-1999 Field Seasons. West African Journal of Archaeology 30(1), 69-108.

Kiéthéga, J.B. 1993. Le cycle du fer au Burkina Faso. ln:Decou-vertes du Burkina Faso II. Sepia, Paris, pp. 73-96.

Kjéthéga, J.B. 2006. La métallurgie lourde du fer au Burkina Faso. Une technologie a l'époque precoloniale. Karthala, Paris.

Killick, D. 2004. Review Essay: What do we know about Af­rican Iron Working? Journal of African Archaeology 2 (1), 97-112.

Killick, D., van der Merwe, N.J., Gordon, R.B. & Grébénart, D. 1988. Reassessment of the évidence for early metallurgy in Niger, West Africa. Journal of Archaeological Science 15, 367-394.

Lambert, N. 1975 Mines et métallurgie antiques dans la région d'Akjoujt. Annales de l'Institut Mauritanien de la Recherche Scientifique 1,6-24.

Lambert, N . 1983. Nouvelle Contribution a l'Etude du Chal-colithique Mauritanien. In: Echard, N . (éd.), Metallurgies africaines: nouvelles contributions. Mémoires de la Société des Africanistes, Paris, pp. 63-87.

Latour, B. 1993. Ethnography of a "High-Tech" case: about Ara-mis. In: Lemonnier, P. (éd.), Technical Choices. Routledge, London, pp. 372-398.

Lemonnier, P. 1992. Eléments for an Anthropology of Technol­ogy. University of Michigan Muséum of Anthropology, Ann Arbor.

Lemonnier, P. 1993. Introduction. In: Lemonnier, P. (éd.), Techni­cal Choices. Routledge, London, pp. 1-35.

MacEachern, S. 1996. Iron Age Begiimings north of the Mandara Mountains, Cameroon and Nigeria. In: Pwiti, G. & Soper, R. (eds.). Aspects of African Archaeology. University of Zimba­bwe Publications, Harare, pp. 489-^96.

67

A.F.C. Holl

McIntosh, S. (ed.) 1995. Excavations at Jénné-Jeno, Hambar-ketolo, and Kaniana (Inland Niger Delta), the 1981 Season. University of Califomia Press, Berkeley.

McIntosh, R.J. & McIntosh, S.K. 1980. Prehistoric Investiga­tions at Jenne-jeno (Mali). British Archaeological Reports, Oxford.

McIntosh, S.K. & McIntosh, R.J. 1988. From stone to métal: new perspectives on the later prehistory of West Africa. Journal of World Prehistory 2 (1), 89-133.

Okafor, E. 1993. New évidence on early iron-smelting in Southeastem Nigeria. In: Shaw, T., Sinclair, P., Andah, B. & Okpoko,A. (eds.), The Archaeology of Africa: Food, Metals, and Towns. Routledge, London, pp. 432^48.

Okafor, E.E. 2002. La réduction du fer dans les bas-foumeaux - Une industrie vieille de 2500 ans au Nigeria. In: Bocoum, H. (éd.). Aux origines de la métallurgie du frr en Afrique. Editions UNESCO, Paris, pp. 35^8.

Paris, F. 1984. La région d'In Gall-Tegidda-n-Tesemt (Niger) III: Les sépultures du Néolithique Final à l'Islam. Etudes Nigériermes, Niamey.

Paris, R, Person, A., Quéchon, G. & Saliège, J.F. 1992. Les dé­buts de la métallurgie au Niger septentrional: Aïr, Azawagh, Ighazer, Termit. Journal des Africanistes 62 (2), 55-68.

Person, A. & Quéchon, G. 2002. Données chronometriques et chronologiques de la métallurgie à Termit - Matériaux graphiques pour l'étude des âges anciens du fer. In: Bocoum, H. (éd.). Aux origines de la métallurgie du fer en Afrique. Editions UNESCO, Paris, pp. 115-122.

Pringle, H. 2009. Seeking Africa's first Iron Men. Science 323, 200-202.

Quéchon, G. 2002. Les datations de la métallurgie du fer à Termit (Niger): Leur fiabilité, leur signification. In: Bocoum, H. (éd.). Aux origines de la métallurgie du fer en Afrique. Editions UNESCO, Paris, pp. 105-114.

Saliège, J.F. 2007. Apport de la chronologie "C, la méthode et ses limites. In: Zangato, E., Les ateliers d'Oboui: pre­mières communautés métallurgistes dans le Nord-ouest Centrafrique. Recherches sur les Civilisations, Paris, pp. 131-135.

Schmidt, RR. (ed.) 1996. The Culture and Technology of African Iron Production. University Presses of Florida, Gainesvtlle.

Shinnie, P.L. 1985. Iron working at Meroe. In: Haaland, R. & Shiimie, P.L. (eds.), African Iron Working: Ancient and Tra­ditional. Norwegian University Press, Oslo, pp. 28-35.

Togola, T. 1996. Iron Age occupation in the Méma région, Mali. The African Archaeological Review 13, 91-110.

Van der Merwe, N.J. 1980. The advent of iron in Africa. In: Wertime, TA. & Muhly, J.D. (eds.), The Coming of the Age of Iron. Yale University Press, New Haven, pp. 463-506.

Zangato, E. 1999. Sociétés préhistoriques et mégalithes dans le Nord-Ouest de la République Centrafricaine. BAR Intema-tional Séries 768. Oxford.

Zangato, E. 2007. Les Ateliers d'Oboui:premières communautés métallurgistes dans le Nord-Ouest du Centrafrique. Recher­ches sur les Civilizations, Paris.

68