NYAME AKUMA No. 41 June 1994

In terminating my five-year editorial responsibility for Nyame Akuma, I also want to extend heartfelt thanks to the many contributors without whose assiduous-and often recurrent-efforts the bulletin would This will be my final issue of Nyame not exist. Generally during my tenure, but Akuma. Having served as editor for five particularly the past couple of years, there years, I feel the time has come for an has been a very gratifying (and even infusion of "new blood" in shepherding this reasonably steady!) flow of articles. Please bulletin. Dr. Pamela Willoughby continue the good work. (Department of Archaeology, University of closing, should mention it has lately of Alberta, Edmonton, Alberta, Canada T6G In I 2H4), has graciously agreed to assume the been confirmed that the forthcoming editorship of Nyame Akuma, and all Panafrican Congress, originally scheduled correspondence regarding publication in the for autumn, 1995, is now going to be held bulletin should be addressed to her, effective June 18-23, 1995, at the University of immediately. For my part, I would like to Zimbabwe, Harare. For details, please express a vote of thanks and my very best contact the organizing secretary, Professor wishes fur success to Dr. Willoughby. Gilbert Pwiti, History Department, University of Zimbabwe, P.O. Box MP167, Mount Pleasant, Harare, Zimbabwe. Good luck, Pam. NYAME AKUMA No. 41 June 1994

and conducting excavations to recover indications of architectural or other structural remains. Artifacts recovered from the excavations were analyzed in order to gain a better understanding of the local BENIN material culture in this archamlogically little known region of West Africa. Once a database of local material traditions was Recent Excavations at Savi: established, it was contrasted with imported trade items to determine the extent to which An Eighteenth-Century the indigenous culture had embraced non- West African Trade Town African goods. A further avenue of investigation has attempted an archae- Kenneth G. Kelly ological understanding of the social and Department of Anthropology, UCLA demographic effects of the slave trade upon 405 Hilgard Avenue trading societies (Fig. 1). Los Angeles, CA 90024 The results of the survey program at U.S.A. Savi showed the ancient town site to have been much larger than was anticipated. Archaeological materials recovered in This paper reports on archaeological controlled surface collections and shovel research conducted during 1992 and 1993 at tests indicated that the habitation area the site of Savi, in the Republic of B6nin, extended over an area about 5 km in West Africa (Fig. 1).Savi was the capital of diameter (Fig. 2). This large size demands the Hueda kingdom, a small coastal state reinterpretation of past town life, suggesting located in the vicinity of the modern town of the conclusion that the town may have had a Ouidah, about 40 km from Cotonou, the larger population than contemporary principal city of Benin. The Hueda state was observers implied, or that the town was actively trading with Europeans for slaves considerably more dispersed than modern and other commodities from about 1660 and historic towns in the area. This change until its conquest and destruction by in settlement density may indicate Dahomey in 1727. However, prior to its fall, signhcant transformative impacts resulting the capital of Savi was host to a considerable from the slave trade. European trading presence. Indeed, the Next, excavations at seven hypothesized Hueda kingdom was the port of primary house areas and two features were begun importance on the historic Slave Coast for (Fig. 31, with excavations ranging from a European traders seeking to obtain slaves to single 1 x 4 m trench to four 1 x 4 m and transport to the Western Hemisphere. I x 3 m trenches. Five of the seven house When Savi was destroyed completely by areas, although yielding substantial Dahomey in 1727, the historic site was not numbers of artifacts, contained inconclusive reoccupied. This provides an excellent evidence of structures, and the remaining opportunity to investigate town life in one two yielded clearly defined features that region of West Africa at a time before the were pWyexcavated. The two features, a European industrial revolution began trash pit apparently dating to the late significantly impacting African material seventeenth and early eighteenth centuries, culture. Research goals at Savi have and a 2 m deep pit resembling a well, included an archaeological survey although of insufficient depth, were also delimiting the boundary of the ancient town, excavated. attempting to identify the commercial and administrative center of the capital town, NYAME AKUMA No. 41 June 1994

(BIGHT OF BENIN)

L i 0 1 00 200 km

Fig. 1. Map of project area with project location, Savi, in bold. Hueda Kingdom shown in open letters, and modern nation names depicted in open capital letters. NYAME AKUMA No. 41 June 1994

Fig. 2. Area of ancient Savi as determined by transect survey.

4 NYAME AKUMA No. 41 June 1994

MAP OF CENTRAL SITE AREA OF SAVl House Areas: HA Feature: Fea Depressions: Q Footpaths: - ,

-300 meters

Fig. 3. Central area of excavations at Savi.

5 NYAME AKUMA No. 41 June 1994

Artifacts Recovered from cylinder. Both stone beads were biconically Survey and Excavations perforated. Artifacts of local, African-trade, and European origin were recovered during the Imported Materials course of investigations. Locally produced European imports make up the other materials were most common, and included main category of artifact finds from Savi. Of local ceramics, locally manufactured the range of imported materials found, smoking pipes, and a variety of less beads and tobacco pipes of European common local products such as terra cotta manufacture are by far the most numerous. figurines and copper ornaments. African Beads number in excess of 800 and reflect a products of more distant origin included wide range of types, including large (3 an) items as utilitarian as milling stones, and as clear-modeled faceted examples, "chevron" rare as several carnelian beads similar to the beads, and various drawn and wire-wound products of workshops in northern Mali beads. The largest assemblage of beads was (Gaussen n.d.) Local ceramics numbered in excavated from among the burned material excess of 100,000 sherds, and were found in a feature identified as a trash pit. Large over the entire ancient town, although the numbers of clear, faceted beads in this vast majority were broken into relatively deposit may have originated from a store or small sherds. The ceramics were minimally other stockpile of beads, perhaps belonging decorated; roughly 90% were plain wares, to a trader. Beads recovered from other and the decorated sherds were minimally locations at Savi were more generally intact, elaborated with impressed or painted possibly reflecting their accidental loss designs. Rims suggest the majority of during daily activities, as opposed to their ceramics fell into one of several categories of entering the archaeological record as a result slightly restricted cooking pots, storage jars, of a building's destruction. and small, rather flat bowls or plates. The other commonly recovered items of Locally manufactured artifacts of particular European trade are pipe stem and bowl importance and interest include tobacco fragments. Intact pipe bowls and marked pipes and terra cotta figurines. The pipes are heels were found at many locations among the most interesting artifacts throughout ancient Savi, both in excavated recovered from the site, firstly for the wealth house areas and during the boundary of decoration present, but more importantly survey. Stem bores were measured for all for their ability to complement and pieces collected, and all marked pieces were contribute to the refinement and illustrated. Significantly, the marked pieces improvement of the local pipe chronologies were almost exclusively of Dutch origin and typologies already well established in (Duco 19821, indicating that despite the Ghana (Afeku 1976, Ozanne 1976). Pipe permanent presence of English and French fragments number in excess of 250 and traders at Savi, the French and British pipe largely complete pipes are fairly common. industries were not getting their product Several baked clay figurines were excavated exported to all trading spheres. at Savi. These anthropomorphic figures, no more than 10 cm tall, were excavated from Imported ceramics and glass were the refuse pit that yielded such a diversity of recovered in considerably smaller quantities. artifacts. Similar figurines have not been The limited numbers of ceramics can be reported in reference to Southern Bknin. contrasted with the wide variety and great Other local materials include beads, of numbers of imported ceramics recovered by which only four locally manufactured beads DeCorse at Elmina, Ghana (DeCorse 19921, were recovered from excavations, two of which implies a very different intercultural which were made of fired clay, the other two dynamic at Savi than that prevailing on the of carnelian. Of the stone beads, one was a Gold Coast. Those recovered were all very flaked and ground carnelian disc, and the fragmentary, but despite their condition, a other was a ground and polished carnelian number of wares and forms were recognizable. Most numerous were tin- NYAME AKUMA No. 41 June 1994

glazed earthenwares and coarse yellow- states where before there had been none. By green lead-glazed storage vessels. Tin investigating the archaeological record of glazed wares of indeterminate origin this early trade period, we learn more about exhibited blue floral motifs, although the social, economic, and political processes occasionally purple or other colors were also that caused, and were resistant to, change. present. Forms represented included coffee Additionally, the role of trade in cups, plates, bowls, and other larger vessels, precipitating social and political change can possibly chamber or storage pots. Coarse be evaluated. ware storage vessels were present, as were a few examples of Chinese porcelain teacups Acknowledgments and bowls. The historic archaeological research at Glass was limited to a very few bottle Savi and Ouidah could never have taken bases and several stemware fragments. place without the permission and Archaeological indications of architecture or encouragement of individuals including structural remains were present, although Merrick Posnansky, Alexis Adand, Francois limited. The mud wall construction Dannou, Maire of Savi, Madame Nodie practiced in the region leaves few Apithy, Chef du Circonscription, Ouidah, archaeologically recoverable traces; and the people of Savi and Ouidah. The however, other structural elements may be research has been supported in part by the preserved or indicated. Excavation at one Department of History and Archaeology of location provisionally identified as a Dutch the Universite Nationale du Benin; the trading establishment revealed a threshold University Research Expeditions Program constructed of imported brick and a brick (UREP), University of California; the doorway arch supported with an iron plate. International Institute of Education The other primary indicator of a structure's Fulbright Research Abroad program; the presence was the occurrence of burned clay UCLA Department of Anthropology; the daub that preserved the shape of the palm- Fowler Museum of Culture History, UCLA; frond ceilings. the UCLA Friends of Archaeoloav; the International Studies and Overseas

Ozanne, P. of the vessel were collected. The data 1976 Tobacco-pipes of Accra and Shai. Accra, collected will be studied to determine the Ghana. amount of variability existing in the types, sizes, and uses of modem ceramics among the people of GrCa. The results will be used in a comparative analysis to gain CAMEROON information about the ceramics collected in excavations at the Iron Age sites at Grea. The Grea 'Wandala" are of interest because their oral traditions recount the An Ethnoarchaeological times before they became Wandala. In Approach to Understanding contrast to traditions among neighboring Archaeological Ceramic Wandala groups, the people at GrCa have no Assemblages from GrGa, memory of discord having existed between Northern Cameroon the Wandala and the Mafa. The GrCa "Wandala" have maintained oral traditions and "pagan" practices that recall pre- Claire Bourges Wandala times. The people of GrCa claim to Department of Archaeology be, not Wandala, but like the Wandala. University of Calgary Through time they have managed to Calgary, A Alberta, T2N 1N4 maintain a certain autonomy, never Canada becoming fully assimilated. This is exemplified by the fact that the Tligrba, the Wandala chief ruling over Grea, is not The community of GrCa is located along permitted to set foot there, but rather, lives the foot of GrCa inselberg, 25 km to the in Kolofata, a small village seven kilometers northwest of the town site of Mora (see to the west of GrCa. In addition, despite their MacEachem and Garba, this volume, Fig. 1). adoption of Islam, the Gda "Wandala" have During the 1992 field season of the Projet maintained a magico-religious tradition Maya-Wandala, 29 archaeological sites were involving sacrifices to bring rain. During my found and recorded on and around Grba 1993 fieldwork I was assured this ritual was inselberg and preliminary test excavations no longer practiced, and had not occurred were carried out at two of these sites for at least eight to ten years. I was told that (MacEachem 1993) (Fig. 1). GrCa was found the people of GrCa are now travelling more to be the location of a number of important and coming into contact with "others" (that Iron Age sites, with the materials from the is "other Muslims" or "other Wandala") two excavated areas producing very who warn the GrCa 'Wandala" that they different assemblages. Today the hamlet of should not be practicing the customs of their GrCa is home primarily to people who call ancestors; by doing so they are "forgetting themselves "Wandala," although they do the Koran." Shortly thereafter I was share their village with Shuwa and Kanuri introduced to the man who, I was told, had peoples (Islamic groups widespread in the formerly performed the rain rituals. When research area), as well as Mafa and other asked when he had last done these rituals, animist, or non-Muslim, groups. he answered that they had been performed Between July 5 and August 1, 1993, as only a few days before. part of my M.A. thesis research I carried out Further evidence of this maintained a ceramic census in the hamlet of GrCa autonomy is seen in the way people identify (Bourges 1993). The census resulted in the themselves. Previous ethnographic work recording of information about almost 400 done among the Wandala claim that no clan ceramic vessels, from 26 households. Each system exists among the Wandala. At GrCa, vessel was measured, drawn, and clan names are known and used by most photographed. In addition, data about people. When asked about ethnic affiliation, production, provenience, life span, and use most GrCa 'Wandala" informants gave their NYAME AKUMA No. 41 June 1994

Fig. 1. Gda sites. NYAME AKUMA No. 47 June 7994

clan name as their ethnic group, rather than since the Shuwa are traditionally cattle calling themselves Wandala. One informant pastoralists. listed seven clans at Gr6a and proceeded to Analysis of ceramics from an tell me who or what each clan represented. archaeological site within the study area has During a similar survey at Kirawa, the revealed that ceramic traditions in the original Wandala capital, similar clans' Mandara mountains area have tended to be names were noted and used, suggesting that conservative (David and Sterner 1987, the Wandala at Kirawa are also not fully Wahome 1989). Wahome (1989) studied the integrated into the Wandala kingdom, ceramic assemblages from MehC Djiddere, a whose seat of power is presently at Mom. site where excavations suggested 1000 years I had expected to find that the Gr6a of continuous occupation. Many of the 'Wandala" had maintained some autonomy decorative motifs and other ceramic in their material culture (particularly the attributes found in the Mehe Djiddere ceramic styles), just as there remains a excavations used similar techniques to those certain autonomy in their myths and used today. Chippendale (1986: 450) magico-religious practices. Contrary to my summarizes the work of several expectations, the Gr6a 'Wandala" ceramic archaeologists, who have demonstrated that, style has fully incorporated the ceramic contrary to comfortable assumptions, traditions of the Kanuri (Wandala do not at ceramic sequences do not always reflect present make ceramic pots, but use Kanuri political, religious, social, or cultural made vessels). changes. He notes that the changes in Preliminary results from the ceramic ceramic assemblages that do occur often inventory indicate that there are two readily happen much later than other cultural distinguishable ceramic traditions presently changes. I think that the adoption of Kanuri used at Gr6a: (1)Kanuri pottery used by the ceramics by the Gr6a 'Wandala" reflects a GrCa 'Wandala," Kanuri, and Shuwas (that desire to communicate to other Wandala is, the Muslim peoples), and (2) kirdi (the that they subscribe to Wandala ideology of local designation for pagans) ceramics used today, despite the fact that they have by non-Muslim/animist groups, mainly preciously maintained ties with their pre- represented by Mafa people at Gr6a. The Wandala ancestors in their nonmaterial Kanuri pottery is represented by large- culture. necked jars with inverted rims for storage of If the Gr6a Wandala are indeed grains and water. These are generally descendants of a non-Muslim people who decorated with knotted plait roulettes. The lived at the Gr6a inselberg prior to being Mafa frequently brew millet beer, requiring 'Wandalized," there is little in their material the use of many large jars. These are culture, most notably in their ceramic generally neckless jars, decorated with assemblage, that would attest to this pre- twisted cord roulettes. The Mafa pot Muslim past. On the other hand, oral assemblages also have a high incidence of traditions and cultural practices bottles associated with beer brewing, and (nonmaterial/nonvisual ethnic markers) ritualistic vessels represented by jar and have maintained this separateness. There bottle categories. seems to be a clear dichotomy between what Within the village of Grba, there exists a the 'Wandala" at Gr6a communicate with clear distinction in the ceramic assemblages their material culture and what they of Muslim vs. non-Muslim groups. Within verbalize about their culture and beliefs. the different Muslim groups, this ethnic Oral traditions and cultural practices, such distinction is not highly visible in the vessels as sacrifices, have a limited audience and are themselves, but rather in the ratios of certain of a limited duration. Only the witnesses of vessel types, and their arrangement within these events are aware of their existence. the households. Thus, the Gr6a 'Wandala," Who these witnesses will be can be carefully who are primarily agriculturalists, own controlled. Material culture will many large grain storage jars; these are not communicate messages about ethnicity and generally present in Shuwa households, cultural beliefs for an extended period, as NYAME AKUMA No. 41 June 1994

well as to a potentially much wider audience. By communicating their Histoire des Peuplements et "Wandala-ness" through their material de la Transformation des culture, the 'Wandala" at Gda are able to Paysages: Etat des articulate within the greater Wandala Recherches Arch6ologiques population, while maintaining a strong dans la Province du Sud connection with their past at Grea, a fact that (Cameroun M6ridional) is not readily communicated to outsider Wandala. Through an in-depth analysis of the J. Paul OSSH Mvondo modem ceramics at ma, I hope to gain an E.N.S. Universitd Yaoundd I understanding about the way that different BP 47 groups, living side by side, reflect their Yaound4, Cameroon ethnic affiliation through pottery. This understanding can then be used to interpret how the ceramics from excavations at Gr4a, Ce theme de recherche entre dans le and from surrounding villages in the study cadre du programme gh6ral de YORSTOM- area, have been used for signalling ethnic URSA associe au pupeBcofit dont je pris affiliation and the types of changes that have part en partenariat avec l'ORSTOM, dans occurred in the perceptions of ethnic groups. l'espace limite forst savane du sud At GrCa, the material culture has been Cameroun. manipulated in such a way as to reflect J'engageai par condquent les premiikes desired ethnicities, despite the fact that the recherches archblogiques dans la province ethniaties being reflected may not directly du sud Cameroun, dgion forestibe peupl6e correspond to the ethnicities that are de Bantou. La premiere phase de la verbally expressed. recherche consistait au reperage et identification des sites d'une part et d'autre References part aux sondages et debut de fouille des C sites les plus accessibles. Ce travail prdliminaire permet de situer les sites et les 1993 Unpublished Field Notes; Field Survey zones de recherche prCcis en vue de la of Modem Ceramic Traditions. Grha, r4solution arch4ologique de la problemat- Provence Extreme Nord, Cameroon, ique du programme dans le sud forestier. July 1993. La problematique generale du grand programme &ant de savoir comment s'est Chippendale, C faite la symbiose for& et sod6t4s historiques 1986 Archaeology, design theory, and the de chasseurs collecteurs ou de producteurs reconstruction of prehistoric design en relation avec les migrations Bantu d'une systems. Enuironment and Planning B: part et d'autre part au contact des Planning and Design 1986: 445-85. fluctuations des reds et des avancks de la for?t? Il revient B l'archblogue de dhire David, N., and Sterner, J. les anciens paysages forestiers, leur 1987 The Mandara Archaeological Project Cvolution, et de determiner les facteurs 1984-87. Nyame Akuma 29: 2-8. humains de leur transformation dans l'histoire. Wahome, E. Wachira Le partenariat avec I'ORSTOM vient 1989 Cnomics and History in the Iron Age of aussi soutenir les premiers efforts North Cameroon. Master's thesis, individuels de recherche archhlogique ddjB Department of Archaeology, University engagde dans cette province depuis 1989. of Calgary. Ce partenariat s'est manifest4 par deux missions sur le terrain A Ebolowa, sur le site d'habitat de BITON (Oton Melen). NYAME AKUMA No. 41 June 1994

La premiere mission eut lieu en fbvrier Les Sites Rdhistoriques 1992 avec le Pr. Jacque Evin du Centre de LfAbrissous Roche de Nkooveng datation par le radiocarbone, Universitk Claude Bernard Pr Essomba, et la deuxi&me Le site se localise B 1 km sur la route le 24 fhvrier 1993 sur le mike site et B Alcan d Edjom eta 12 km. du village de Fiebot. Les 11, dans le cadre du programme ECOFIT, coordonnks sont 3 28' 02" de latitude nord avec le Dr. Jean Maley, le Dr, Henry Robin, et 11 57'06" de longitude est. La facade a et Achundong. La mission avait des objectifs 140 m de long et la profondeur est de 21 m. cette fois pedologiques, botaniques, et Pas de vestige sur la surface peut-&re B archblogiques. cause des activitbs culturelles de populations qui s y dQoulent. Techniques de recherche Le Rocher de Nkolmegong Ce site est situ6 B 3 km 4 de Rebot sur la La premiPre 6tape a kt15 la prospection. route d Eboman. Les coordonn6e sont Elle avait pour but de localiser les sites 326'03" de latitude nord et 11 35' de anciens de peuplement dam la for& du sud longitude est. Cameroun. Cette prospection a 6t4 une couverture des trois departements que C est un rocher de 690 m d altitude. Ce comprend la province du sud Cameroun. Le rocher comprend les traces d occupation rkphge des sites permet donc d engager les humaine constituks d un ensemble de trous travaux de fouille pour repondre aux circulaires ovales, dont 14 trous drculaires et objectifs du programme. La technique de 6 creux ovales. Les trous ont un diametre prospection fut les enquetes orales qui variant entre 10 cm et 15 cm et une consistaient P interroger les anaens sur les profondeur de 3 cm B 12 an. Les trous anciennes zones de peuplement dans la for& ovales varient de 32 an B 90 cm de longueur, les points de fluctuations de la for& par pour 12 cm B 25 cm de largeur. Ces 1 action anthropique. La seconde mCthode structures circulaires organisks A la surface etait la recherche des anciens villages dans du rocher temoignent d une occupation la for& en se ri3Qant sur le changement et anaenne du rocher. les ruptures de continuit6 observk dans la L'Abris sow Roche de Akooveng for&. L utilisation des toponymes a permis 11 est situe au NE de Nkolmebong et se aussi d identifier les zones de peuplements compose de trois cavites qui lui donnent une anaens. Enfin 1 utilisation de la carte au forme de demi-polygone. La premike cavit6 200.00e a permis de r6p&er sur une grande a 2 m de haut et 9 m 50 de long, la deuxieme echelle le sens de 1occupation des sols et de cavite a une profondeur 6 m 02 et 8 m de la forst par les populations, et bvolution de haut. Louverture est de 19 m 7. Elle 1habitat et de 1 espace forestier dans la comprend des d+ts de 15 an d bpaisseur. province du sud Caneroun. Le rnatckiel se compose d un tesson decore, Certains sites dkouverts ont fait 1 objet ne presentant pas de trace d usure, ni de de sondage et de fouille pour verifier dbgraissant de quartz. Deux autres tessons 1 importance du site et du materiel furent collectes B -10 an de profondeur. archkologique. La fouille fut effeduk sur les La 3e avid a une ouverture de 13 m 40 sites les plus accessibles. de long et une profondeur maximale de 7 m 45. Le sol est incline et aucun vestige ne fut RBpartition des Sites Archdologiques trouv6 sur le sol. On distingue les sites des dkpartements Les Sites de age du Fer du Dja-et-Lobo, de la vallk du Ntem, et de la Mvilla. Le Site de Zoetele' Browse (2B1-Locus A) Le site est situe B 30 m B 1ouest, soit Djaat-Lobo 12 NW de la ville de Zoetelk. C est une butte anthropique de 52 an de haut et 43 an de Il a livr6 les sites pr4historiques et ceux diametre. Le sondage effectue a livr6 la terre de 1age du fer. bdlk et du charbon de his. Sur 90 cm de NYAME AKUMA No. 41 June 1994

profondeur le site @sente des morceaux de m4tallurgiques. L enclume mesure 36 an de charbon, et un fragment de tuyhe B 90 an large et une Bpaisseur de 25 an. La partie de profondeur. Ces indices permettent droite a une largeur de 15 cm et une d avancer 1 hypothese d un atelier de 4paisseu.r de 10 cm. La longueur totale est de rkduction du fer. 68 cm. L enclume presente des traces Le Site & Zoetele' Brousse (2BI) d usure B plusieurs points. Ce site fut r4p6rti 31 partir des scories Le Site & Ze (ZI) reparties en surface. Les scories sont Dans 1 arrondissement de Mintom, le dispersks sur 4 m de long et 1 m 50, pour site se localise dans le village Ze B 1 km de une surface de 6 m 2. Les scories sont de Mintom ville, sur la route Mintom-Dja. Les petite taille. La proximitB de la route permet sondages ont permis de dCcouvrir les de penser qu il a Ot6 remanib par endroits. thoins d une occupation ancienne de Ze. Le Site d'Otetek (OTI) Quatre types de vestige composent le site: une fosse, des morceaux de c#ramique, une ll est situ6 B 3 14 '06" de latitude Nord et perle, et un Mat. La fosse de forme 11 54 '05" de longitude est, soit B 3 km au rectangulaire a un profondeur de 1m 20,1 m sud-est de Zoeteld ville. Le site se trouvant B quelques metres du carrefour Minkoumou- 50 de long, et 1 m 10 de large. Les tessons au Nkoumadzap dans la cour de la mission nombre de dix-huit comprennent 14 du chretienne. Le materiel est varie et se corps du vase, et quatre de 1ouverture. compose des scories de petite taille, des La perle est bleue ciel perforbe par un fragments de tuyere, des poteries non trou de 2 mm. d6cor&s, du charbon de his sur pres de 100 Le Site d'AIat Mukay (AMZ) m 2. La taille petite des scories indiquerait Il s agit d un site mCtallurgique situe sur peut4tre un atelier de forge. la rive du Dja. Il se localise dans le village Le Site de Kumu (KMI) du meme nom sur la piste reliant Bi i Il se trouve dans la cacaoyere de Jean Lomie, B la cour de 1eglise EPC du village. Thomas Akoa. Le site selon 1informateur fut Les pluies fortes ont provoqu6 une Brosion occupe par les ancestres. Il se localise B du sol de 50 cm de profondeur, ce qui en 100 m de la rivikre Kumu. Le site quelque lessivant la couche d argile a mis en surface peu remanib a en surface un materiel de nombreuses scories tBmoins de la compos6 de briques briilks, des scories de metallurgie de fer. Le site a une structure toutes les tailles, des morceaux de tuyere et spatide circulaire, avec un diametre de 4 m. de la chmique. Les objets de surface se composent de scories de petites dimensions, du charbon de Les prospections, sondages, et fouilles en furent men6 aussi au sud de la grande for& his, d un outil silex, aucun morceau de dans les arrondissements de Djoum et tuyhe ne fut ramasd. Mintom du m6me departement. La particularit6 de ces arrondissements est Dipartement de la Val160 du Ntem qu ils possedent 1hsysteme forestier qui se Ce departement a livre deux sites: prolonge dans les pays voisins. En effet ils Mekomo et Biyan. sont frontalierset voisins du nord du Congo et du nord Gabon. Cet ecosyst&meforestier Le Site de Biyan est aussi apple la reserve du Dja. La prospection fut menee B Biyan, Le Site d'Ekom (EK1) village situd sur la route Meyo-Centre- Le site est actuellement colonid par les Ma an en passant par Ebolowa. Les arbres, les palmiers, et les cacaoyhes. Le ramassages de surface ont permis mat&el recueilli apres un sondage de 1 m 2 d identifier trois secteurs. Le premier secteur se compose de quelques morceaux de est Alsoatan, il s agit d un abris sous roche ceramique, des scories et surtout d une de 3 cavitbs. La cavik5 ouest est profonde, la enclume en place. L enclume et les scories cavit4 centrale comprend des dbmes attestent la pratique des activites internes, alors que la cavite est a une forme NYAME AKUMA No. 47 June 7994

de cuvette. Il n a kt6 trouv6 aucun vestige tessons. Le carre D ne livra que 31 tessons et dans 1abri, par contre B quelques metres de du charbon de his. 1abris sous roche au bord du cours d eau, Le sondage C1comprend un coGteau en deux outils tailles et polis furent ramasds. metal, une coquille d escargot, un fond plat Ces pierres sont h mettre en relation avec la d objet en fer perfod, 6 noix de palme, et grotte. La premibe hlche polie est rectiligne sept tessons. et plate, de forme triangulaire, mesurant Enfin le sondage C' a liduniquement 14 cm de long, et 1 cm d bpaisseur. La du charbon de his. Plusieurs morceaux de seconde hlche est ovoide avec les deux charbon furent prblevds pour datation. A extrhitds arrondies. Le deuxieme secteur 30 m au nord-ouest de la fouille furent Zookom a livr6 en surface des scories de identifibs deux pierres ii aiguiser, plies B la fonnes amondies, le demier secteur Messeng surface. Au total la fouille du site de 17 a lid en surface fragments de Mekomo fut riche, les donnk quan titatives c6ramique. Dans 1 ensemble le magriel de indiquent 144 pieces reparties en 4 surface du site comprend des hlches, des morceaux de verre, 4 objets en metal, 1 scories, et de la ceramique. Les fouilles coquille d escargot, 27 noix de palme, et 108 restent B entreprendre. tessons; la ckamique et les tessons sont les Le Site de Mekomo plus abondants d site. L etude de ce mahkiel permettra d eclairer les relations entre I1 se localise B 1 arrondissement 1 homme et 1 environnement dans cette d Olamze B quelques kilometres de la partie de la province du sud. frontiere gabonaise et equato-guinknne. La repartition des tesson sur le sol et 1existence Diparternent do la Mvilla de la petite monticule indiquant le secteur B fouiller. Deux sondages furent organids au Le Site de Biton sud-est pour verifier 1 &endue du site et sa composition. La fouille elle-meme consista B Ce site se situe B 50 km d Ebolowa sur la installer un carroyage de 4 m 2 de surface route Ebolowa-Kribi. Les vestiges ont etb avec quatre carres de 1 m 2 chacun. La r6p&Cs au lieu dit Oton-Melen. C est un site division du carroyage comprend les carrks riche au regard de la variete des vestiges. A, B, C, D. Le cam6 A etait localise au nord- L observation de surface indique une terre ouest, le B au nordest, le D au sud-ouest, et noire qui se retrouve dans les termitieres le C au sud-est. indicatrices de sites anthropiques. Les Les sondages C'1 et C' furent effedub B vestiges importants se composent de 1est du carroyage au prolongement de BC. nombreuses scories et de grandes tuyeres, L analyse stratigraphique indique une une enclume dans le secteur sud, et des premii?re couche A1 B 14 an de profondeur amas de charbon de bois de grande de coloration noire, contenant les vestiges, dimension. Le secteur nord comprend des une deuxieme couche entre 15 cm et 30 an palmiers antiques Btranglb par les grands de profondeur, stQile, et argileuse de arbres. Ce secteur permet d observer couleur jaune. 1holution de la for& avec la for& primaire, la for& secondaire, et la for& taillie Le materiel recolt6 se compose de plusieurs fois rernanik par 1homme. Entre cbramique, de metal, de noix de palme, du ces deux secteurs se trouve une meule en charbon de bois, des pieces de verre, des place. Le site est entour6 par de petites coquilles d escargot. Le carre A a lid 13 fosses et une source dans un rocher. Le coques de noix de palme, 27 tessons, 3 pike ramassage de surface a aussi lidun pot en de verre, et du charbon de bois. Le carre B ceramique B moitie casse. Les charbons livra 2 noix de palme, du charbon de bois, 13 furent preleves pour tenter les premieres une piece de verre, tessons. Le carre C datations avant la fouille ei 1 etude comporte une bague en metal, un objet systhmatique du site, puis des analyses perfore B plusieurs endroits, du charbon de antluacologiques. bois, six coques de noix de palme, et 30 NYAME AKUMA No. 41 June 1994

Le site de Mvamyetom W ETHIOPIA Il se situe dans le village du meme nom 3I 20 km d Ebolowa sur la route Eholowa- Ma an. Le site fut decouvert lors du The B.I.E.A. Aksum aeusement d une fosse 2 m de profondeur Excavations, 1993 pour le prelevement de la terre 3I brique. Dans la fosse, quatre morceaux de ceramique furent ramasds, parmi lesquels, David W. Phillipson 1/3 de vase, deux tessons noirs, et un tesson Museum of Archaeology & Anthropology marron. Ces dhents sont des indicateurs Cambridge University interessants pour la fouille du site. Downing Street Apres trois ans de recherche sur le Cambridge CB2 302 terrain dans le cadre du programme, England quatorze sites ont 6t6 decouverts. Certains ont fait 1objet de fouille et de sondage. Le mattkid est abondant et varie. Le materiel This is a preliminary account of the obtenu jusque-l3I necessite un debut de research undertaken under my direction in traitement et d analyse, notamment les the first year of renewed archaeological grandes quantites de charbon reparties sur investigation of the ancient Ethiopian capital les sites metallurgiques et dans la fouille, les at Aksum. It follows campaigns directed in sequences pedologiques des sites, et les 19-74 by the late Dr. Neville Chittick on tessons varies. behalf of The British Institute in Eastern Au terme de quelques annees de Africa. prospection, le bilan en terme de site et Work was conducted in Ethiopia during materiel archeologique est encourageant. the period 21 October2=3 December 1993. Pour resoudre la problematique du The project is sponsored by The British programme il convient d engager des Institute in Eastern Africa, with additional analyses de laboratoire sur le materiel d6j2 support from The Society of Antiquaries of recolt6, 31 travers la caracterisation physique London, The McDonald Institute for et mindralogique des tessons, celle des Archaeological Research in the University of sequences pedologiques, puis la compo- Cambridge, The British Academy, and The sition, caractkisation et determination des British Museum. dkgraissants vCg6taux et identification des The 1993 excavation was planned as the vegetaux contenus dans les tessons. La first of five annual field seasons. The overall phase suivante verra la fouille systhatique research design envisages continuing d un site et 1 analyse des vestiges des investigation of the main tomb-and-stelae couches profondes en vue de saisir les complex (begun in 1973~744);research on the palbenvironnements, ces fluctuation de la domestic economy and residential areas of foret et les facteurs humains de leurs ancient Aksum; and on the town s transformations. surroundings, including quarry sites, workshop areas, and peripheral settlements. In 1993 detailed work was concentrated on the first of these objectives, with reconnaissance towards the second and third. The professional team totalled 12 persons, 5 of whom were Ethiopians. The locally employed labour force was steadily increased through the season to a total of 65 men. Highlights of the season were the elucidation of the original setting of the NYAME AKUMA No. 4 1 June 1994

largest of the Aksumite stelae and its walls of the Mausoleum, constructed of associated monumental tombs, together rough stones set in mud-mortar, had with the first archaeological demonstration originally been thickly plastered with a of the importance of ivory in the ancient coarse gritty render: careful examination of Aksumite economy. the substantial surviving areas failed to reveal any traces of painted or other The Mausoleum and East Tomb decoration. This render is a previously unrecorded feature of Aksumite These tombs, located but not excavated architecture. during Chittick's investigations of 1974, are The Mausoleum covers an area at least both associated with the largest of the 16.65 m from east to west by 14.95 m from Aksumite stelae, Stela 1. The original plan north to south. Its total known floor area is for this stela is now apparent: it was approximately 143.3 sq m and, when first intended to stand at the rear of a deep entered, archaeological deposits extended to walled inlet or embayment set back from the within 10-35 an of the roof. It may thus be front terrace wall of the stelae area, with calculated that its total contained deposits entrances to a monumental tomb set in each comprised some 300 cu m, of which side of the embayment (Fig. 1). There is approximately one quarter was excavated in increasing evidence to support the view that 1993, the whole of the central passage being the stela, which would have been 30 m high cleared, together with the outer 1.0 m of two and over 500 tonnes in weight, was never side-chambers. The greater part of the successfully set upright, but that it fell and deposits in the central passage had clearly broke whilst in process of erection. been introduced through the vertical shafts, The tomb on the west side of the and Dr. Jake Phillips reports the presence embayment, designated the "Mausoleum" of much late Aksumite or post-Aksumite by Chittick, was entered through a material, including abundant faunal remains monumental portal leading into a passage and (in the shaft-fill) some charred seeds. 16.65 m long and 1.90 m wide, roofed with The lower levels overlying the flagged floor, massive dressed granite slabs in which were however, showed less disturbance and the cut three square apertures connected to pottery that they contained was of classic stone-lined shafts that originally led to the "Red Aksumite" type. The excavations have ground surface. Its excavation in 1993 was so far revealed no incontrovertible evidence supervised by Mr. Michael Harlow. The for primary burial deposits: it is possible passage has a height of 2.28 m, and its stone- that such will prove to be preserved in the flagged floor is 5.90 m below the modem side-chambers far from the vertical shafts, surface. At its western end, the passage but the possibility must also be borne in leads to a brick arch set upon massive slabs mind that the tomb was never put to its of dressed granite: no attempt was made to intended purpose. remove the earth blocking this arch, and Exactly opposite the Mausoleum portal, what lies beyond it remains completely in the east side of the embayment, is the unknown. It is likely that a similar arch was entrance to the East Tomb (Fig. 41, originally located immediately inside the excavation of which was supervised by Mr. entry-portal; the brickwork has not Gavin Rees. This tomb lacks a monumental survived, but the stone slabs from which it portal, but the two entrances are otherwise sprang have suffered only minor remarkably similar in style, construction, displacement. On each side of the passage is and position. It was only practicable to a series of five chambers, making ten in all, excavate the East Tomb to a distance of 2.4 each 6.57 m long by 1.70 m wide (Fig. 2). At m from the entrance: the main lintel and the the entrance to each side-chamber, first roof slab were both broken, the second bordering the central passage, had originally roof slab appeared to be missing or seriously been a brick arch springing from dressed displaced, and the south wall had slumped. stone slabs: none remains intact although It was concluded that the tomb could only the slabs and substantial remains of the be excavated with safety by working down brickwork survive (Fig. 3). The internal NYAME AKUMA No. 41 June 1994

Fig. 1. Reconstruction of the possible original design for Stele I, Mausoleum and East Tomb. The stele may have been intended to stand at a higher level, and the front of the embayment may have been walled in. NYAME AKUMA No. 41 June 1994 NYAME AKUMA No. 41 June 7994

Mausoleum, Aksum - Detail of sidechambers E and G

f-rl gypsum plaster [' drassed stone brlck

c 0 1 2 3 4 5ill

Fig. 3. Part of the Mausoleum's central passage, showing entrances to two side chambers not yet excavated.

19 NYAME AKUMA No. 41 June 1994

Mausoleum and East Tomb, Aksum - Cross Section A-A

Stele No. 1

~rchruppod rtwa

Mausoleum East Tomb

Dressed Stone Gypsum Plaster

Fig. 4. Section through Mausoleum, Stele 1 and East Tomb.

20 NYAME AKUMA No. 41 June 1994

from the modern surface. This will be a estimated as falling in the late third or early difficult and costly operation, because the fourth centuries A.D.: it now seems possible ancient floor lies at a depth of 6.4 m and the that the true age is somewhat greater. The extent of the East Tomb remains totally excavation was continued in 1993 under the unknown. The tomb contents appear to be supervision of Mr. Jess Tipper, with the significantly less disturbed than those in the assistance of Ato Getu Degefa. Mausoleum, and the pottery from most The tomb is approached by means of an levels was exclusively of classic "Red adit with 18 surviving stone steps (Fig. 5). Aksumite" type. Some doubt must remain, The adit, 12.0 m long by 1.5 m wide, attains however, as to whether the East Tomb was a depth of 6.2 m and was originally roofed ever completed: it appears to lack a stone horizontally with a series of rough stone flagged floor analogous to that in the lintels of which 7 survive in situ. The east Mausoleum. Furthermore, if the East Tomb side of the adit retains the original Aksumite entrance was originally designed to receive stone wall to its full height in a remarkably a portal, it seems more likely that this was fine state of preservation. At the foot of the never inserted than that it was subsequently steps, which are likewise well preserved, a removed. Clearly much further work will be horseshoe-shaped brick arch gives access to needed before the East Tomb can be fully an antechamber, most of the contents of understood. which were removed in 1974. A pit in the The complex of Stela 1, Mausoleum and floor of the antechamber, not noted by East Tomb must now be viewed as a unitary previous excavators, was found to contain a design. As noted above, there is evidence very large quantity of bronze derived from that the stela fell and broke before it was one or more receptacles decorated with successfully erected, and that the East Tomb inlaid plaques. may never have been completed. It is The tomb itself comprises four rock-cut possible, therefore, that the complex was chambers, their floor being approximately never put to its originally intended use. The 9.5 m below the modern ground surface complex's date of construction was (Fig. 6). It appears that the complex was estimated in 1974 as probably falling in the roughly carved out as a whole, then fourth century A.D.; further excavation will subdivided by the insertion of cross-walls be necessary before any attempt can be incorporating brick arches. Recesses in the made to refine this. It appears increasingly walls of two chambers, not yet excavated, likely that the fall of Stela I, broadly are provisionally interpreted as burial loncli. coinciding as it did with the adoption of Excavation in 1993 was concentrated in Christianity at Aksum, led to the the second chamber, where a mass of abandonment of the use of stelae as grave- archaeological material was preserved, markers. Interesting evidence for a change 1 despite having suffered some disturbance in in symbolic meaning of Stela following its ancient times. Complete and broken pots fall is provided by the observation that the were mixed with disarticulated human handle from the false door on its upper bones, fragmented glass vessels, remains of surface had been carefully chipped away in elaborately decorated wood and metal boxes ancient times. Later elite tombs were of which had originally contained quantities of different types, as represented by the 'Tomb glass beads, and other metal items. of the False Door" and those traditionally Particular interest attaches to a substantial attributed to Kaleb and Gebra Maskal, quantity of ivory, including two large plaques with fine carving of vines and The Tomb of the Brick Arches animals (Fig. 7): the lifting and conservation This tomb is located some 25 m east of of these unique but poorly preserved objects Stela 3, but does not appear to have been were accomplished through the skill of Ms associated with any particular stela. It was Noel Siver, the project's Object Conservator. entered in 1974 when its general plan was The complexity of this excavation recorded and the first of its four chambers necessitated meticulous work and slow largely excavated. Its date was then progress: further conclusions about the NYAME AKUMA No. 41 June 1994

Fig. 5. The adit to the Tomb of the Brick Arches. No. 4 7 Jurt. 1994

Plan of the Tomb of the Brick Arches, Aksum

Fig. 6. The Tomb of tho Brick Arches.

23 NYAME AKUMA No. 4 1 June 1994

Fig. 7. Cawed ivory plaque from the Tomb of the Brick Arches.

24 NYAME AKUM No. 41 June 1994

Tomb of the Brick Arches must clearly await British Council, Addis Ababa, Michael the completion of further research. Sargent Esq.; Ato Mabratu, Culture and Sports Bureau, Aksum; Dr. Merid Aregay, Reconnaissance Addis Ababa University; Professor Taddesse Tamrat, Addis Ababa University; Dr. Bahru Three areas of ancient quarrying were Zewde, Director, Institute of Ethiopian noted on Gobedra Hill, 3-4 km west of Studies, Addis Ababa University; and Aksum. Preliminary photographic recording numerous other people, from all walks of was undertaken, but much more remains to lifeI in Aksum, Makelle, and Addis Ababa. be done when the vegetation can be cleared. Their contributions will be fully Preliminary reconnaissance was also acknowledged in future publications. conducted to locate areas suitable for the investigation of Aksumite domestic occupation, planned for 1994. GHANA

Monument Conservation At the conclusion of the 1993 Before the Flood: excavations, all areas investigated were left The Golden Volta Basin securely protected, the details being agreed in advance with officials from the Ministry of Culture. The intention is that these E. Kofi Agorsah measures can, if it is decided at some future Black Studies Program date when excavation has been completed to Portland State University open the monuments to public view, be Portland, OR 97207 incorporated into a long-term conservation U.S.A. strategy. The entrances both to the ~ Mausoleum and to the East Tomb were consolidated with retaining walls built in The Northern Volta basin has engaged traditional Aksurnite style, and roofed in the attention of major research projects in such a way as not to impede general views the last ten years with greater intensity than of the area. The adit to the Tomb of the Brick has been known in the history of research in Arches was likewise consolidated and the area (Agorsah 1983, 1985, 1986, 1988, roofed, its entrance and that of the tomb 1990; Kense 1983, 1985; Stahl 1983, 1989). itself being securely sealed. Explanation of the cultural development as well as connections or relationships between Acknowledgements ethnic groups, their geographical The research was supported at every distribution and settlement patterning, and stage, both in Addis Ababa and in Tigrai by the nature and mechanism of functional the Centre for Research and Conservation of adaptation in the area are issues that appear the Cultural Heritage (CRCCH) in the to have been the main foci of investigation. Ethiopian Ministry of Culture and Sports. Between 1981 and 1985 the Volta Basin The interest, cooperation and assistance of Archaeological Research Project (VBARP) very many people is warmly acknowledged. has focused investigations on the Among these are: H. E. The Regional geographical area currently inhabited by the President of Tigrai, Ato Gebru Asrat; H. E. eastern Gonja, Nawuri, Nchumuru, and the The Minister for Culture and Sports Affairs, Krachi-speaking people. The initial focus, Ato Leule Selassie Temamo; His Holiness during the period, was on the Nchumuru The Patriarch; Her Britannic Majesty's settlements. Conclusions based on the data Ambassador to Ethiopia, James Glaze Esq., available emphasized the northern Volta CMG; Ato Masele Zeleke, Head of Culture Basin as an important area for the and Sports Bureau, Makelle; Dr. Kassaye development of early cultural traditions in Begashaw, Head of CRCCH; the Director, that part of West Africa. It was also NYAME AKUMA No. 41 June 1994

generalized that ecological, historical, and located on the western bank of the Volta cultural variables of the study indicate two Lake opposite the village of Boafri (Fig. 1). waves of human movements: a northsouth To the north, east, and west of the village of one occurring several thousand years ago in Bubuakro, also referred to as Tokuenya by the stone age; and a more recent and immigrant fishermen, is located the site of influential one in the opposite direction Kononaye, which forms part of the vast beginning in the second millennium A.D. open land that gently rises from the valley of These and other generalizations continue to the Volta Lake for some 10 m to just about require supporting evidence from other 100 m above sea level. Its northeast and parts of the basin. south boundaries are marked by the big Several field trips undertaken between meander of the marshlands of the Volta 1985 and 1990 to examine the location, Lake. Although there are slight elevation distribution, and patterning of settlements differences here and there, the topography is as well as traditional building construction generally uniform. Large mounds with lots and their related decay patterns in the of pottery, stone circles, and termite mounds Krachi area are discussed in this are some of the main features of the site, presentation, in an attempt to give further which appears to extend over 3 km along support to some of the generalizations made the banks and over 500 m inland. Shea in the earlier phase of the research of the butter and baobab trees are most common VBARP and to continue the study of mud although there are several other plants that wall decay patterns in the southern section are of medicinal and other uses to the of the northern Volta basin and among a inhabitants of Bubuakro. different ethnic group. The 1985-90 period of study, therefore, Ethnography Study shifted from the Nchumuru ethnic group Two aspects of ethnographic data area occupying the basin of the Volta and collection were very important for the the Daka rivers to the basins of the Oti, Sene, excavation conducted at the Kononaye site and the Volta rivers inhabited solely by the but can be only briefly discussed. Krachi who are one of the major Guang- speaking groups in the basin. Dadiase, The Krachi Krenkuase Yabin, Kadengben, Ahinkro, Akroso Beposo, Aworoso (Aweresi), Gharee, Unlike the Nchumuru who only recently Banka, and Begyamso are some of the well- introduced the paramount chief system, the known sites in the areas around Krachi. Krachi have since ancient times been Unfortunately these and many other sites centrally controlled by a paramount chief are now inundated by the expanding who must come from one of three clans, and artificially built Volta Lake. More recently the Krachi Dente (an oracle that had both the VBARP has located the sites of Abokono, spiritual as well as political powers). While Akwankwakwae, Old Kajaji, Old Mokraye, the paramount chief controlled the stool Kononketakpan (Kononaye), Kumpo, and which, as is known, is the embodiment of Takragya among others. As indicated in an the ancestral and other powers of state, earlier report of the survey, many of these Krachi Dente through the Dente Bosomfuo sites appear to have developed in valleys, controls the land. The implication of this flood plains, fans, confluences, and system for patterning and spatial behavior pediments, with the soils that are within Krachi settlements has not yet been periodically revived by alluvial and clearly identified but observations so far colluvial deposits. The streams, springs, and indicate that locational behavior patterns are seepage or groundwater provided water for only controlled by clan affiliations based on those located in areas where rainfall was social relationships similar to what has been inconsistent and river water was not readily observed to operate among the Nchumuru. available. However, it appears from the study of a The main excavation was conducted at typical Krachi village so far that the the site of Kononketakpan (Kononaye) locational or distribution pattern of NYAME AKUMA IVo. 47 June 9994

KONONICETAKPON (KONONAYE)

Fig. 1. Monanketakpon (Kononaye). NYAME AKUMA No. 47 June 7994

households is not physically the same as indicate the location of a particular type of among the Nchumuru. It is also clear that trees or plants or rubbish dumps or burials although the local rule (LR)model applied or an area of a specific activity of some kind? to Nchumuru settlements (Agorsah 1983, The termite hills are of economic use to 1986, 1990) does not generate the same Bubuakro as they are a source of feed for pattern, the trend of locational decision- fowls and the'clay obtained from them is making among the Krachi follows the same very good for making fish-smoking ovens, rule. hearths, and similar structures.

House Construction and Decay Patterns Archeological Evidence among the Krachi An eight-week excavation of the site of Two basic structural systems of house Kononaye, was conducted with the support construction are observed among the people of Earthwatch (Center for Field Research) of Bubuakro: wattle and daub in which the volunteers, graduate and undergraduate load of the roof and the floor are carried by a student volunteers from the University of framework until they were redistributed Ghana, High and Secondary school and after the spaces in the frame have been fied Training College students. with earth, and secondly the mass On the basis of results of soil chemical construction in which the weight of the roof analysis and surface distribution of is carried completely by an earth wall. Only structural features and artifacts, the site was the latter is observed among the Nchumuru divided into five areas named KY1 and the decay and collapse patterns of such (Kononaye I), KY2, KY3, KY4, and KY5 buildings differ from those using an initial (Fig. 2), indicating the sequence in which framework. In both types of construction, they are to be investigated, but by no means however, stability is not usually a serious representing the order in which they are problem as the buildings are small and short most threatened by inundation of the Volta or low. The lining of the bases of walls Lake. KYI was selected as the first to be appears to be a common precaution to avoid investigated because it was the most erosion from rain splash. The collapse of threatened of the five divisions at the time. walls by outward buckling as noted at Wiae KYl marks the northern-most part of the in the Nchumuru area also appears to be Kononaye site and consists of a series of common with the mass earth construction at mounds overlooking the lake, one of which Bubuakro. The wattle and daub house was completely, and two others partially, appears to collapse in a haphazard manner excavated during the eight-week season. although sometimes outward and its life is From the seven 1.5 to 2.0 x 6 m trenches and more easily threatened by termite attack. four 1 x 2 m test pits, approximately 30,000 A general observation was that the potsherds and over 200 other artifacts strength of a building in Bubuakro depends (mainly small finds) were recovered. on the number, size, and proportions of windows or openings. This appears to have Stratigraphy been the reason why such openings are kept at a minimum. As was noted at Old Wiae of Generally, all the trenches reached level the Nchumuru, there appears to be a shift 1 at a depth of 7 m on average, but three of from circular buildings to rectangular those which were closer to the lake touched because the floor areas of houses observed at water table at that level. The other four the archaeological site of Kononaye are trenches farther away from the lake reached predominantly circular while in the modern level five, about 2.8 m on the average at village very few circular buildings are bedrock. Levels 2,3, and 4 were the main observed. An issue that is being examined is cultural level, level 4 probably of a late stone the origins and distribution of termite hills age tradition indicated by polished stone in the Bubuakro area. Can it be assumed that axes (Fig. 31, a stone and bone beads, as well the location of termite hills represent the as the fewer, thicker, cruder, and past location of collapsed houses or do they undecorated potsherds. NYAME AKUMA No. 41 June 9994

Fig. 2. Kononaye. NYAME AKUMA No. 41 June 1994

Fig. 3. NYAME AKUMA No. 47 June 7994

Thin, and in some cases very thick, pieces, measuring approximately 2.5 crn lenses of shell signalled the appearance of a diameter with thicknesses varying between new level. The stratigraphy seems to .5 and 13an; an ivory pendant from level 2, indicate that there was a fairly long break which is conical and grooved around between levels 3 and 2, a break that toward both ends containing a hole at the probably supports the view that the site may wider end, which is approximately 5.1 cm have been temporarily abandoned. When long (Fig. 6a); bone awls or needles this break would have occurred is not yet generally 4.5 cm long, level 2 (Fig. 6 b, c); certainly known. Could this have been the one stone and one bone bead from level 4 time of the first movement of the Krachi (Fig. 6f, g); a few glass beads from surface people southward as generalized from the and level 3; one piece of a "Kintampo" cigar Nchumuru material? How much do we tell (surface collected by a local farmer near KY from the C14 dates so far obtained? 2, Fig. 6h); what appears to be a bone fish mending needle (Fig. 64; and a fragment of Stone Circles an ivory bracelet (Fig. 6d). A large quantity of shell was collected An important feature of levels 1 to 3 from the three top levels, clearly indicating a consists of circles of stone of approximately considerable reliance on shellfish as food 1.5 to 2 m in diameter placed around one resources of the River Volta. very large stone. The smallness of the size of the circles clearly suggest that they did not mark the bases of houses. Their location in Some Observations relation to two house floors identified at The Kononaye site coincides geograph - KY1 points to some other use: bath houses? ically with the area of the greatest or bases of fish-smoking ovens? or storage concentration of stone age sites, material of facilities? Circular fish-smoking ovens of the which type has been observed at the site. same diameter are still used in the modem The site is in the center of the area observed village of Bubuakro and appear to represent to have the greatest diversity of Guang a continuity from previous practice at languages, a criterion linguists associate Kononaye. with the location of the ancestral language of the Guang. It is difficult to associate the Pottery various movements from and into the Kononaye area with any period of the split Preliminary study of potsherds suggests within the Guang linguistic group. Evidence predominance of large pots such as water is clear from Kononaye that the site was storage pots in levels 4 and 5, while in the abandoned, even if briefly, and that would top three levels the main vessels are bowls be considered as the period of some kind of and cooking pots. There is some kind of movement out of the area. Radiocarbon uniformity in the quantity, types, and dates obtained so far are as follows: decoration of pottery material throughout the first three levels, although slight 1. Beta-59895 (KYY90/T2/L2) 140 f 50 differences have been noted. Pottery BP for the most recent level when manufacture continues to be a practice in the the site was temporarily abandoned modem village of Bubuakro where it is by the Krachi until it was resettled claimed to be an ancient tradition that in the mid-1960s; (mid nineteenth produced pots of various forms and with century) different decorations (Fig. 4). 2 Beta-59896 (KY1/90/TZ/L3),480 f 50 BP for the earliest occupation of the Other Finds site by the Krachi. This date could extend half or one century earlier as Other finds include several local ceramic level 3 overlay an earlier cultural smoking pipes mainly from levels I, none level consisting mainly of Late Stone from levels 3 and 4 (Fig. 5); spindle whorls AgeiIron Age tradition, which in mainly from level 2 (Fig. 6i,), ceramic game NYAME AKUMA No. 41 June 1994

Fig. 4. NYAME AKUMA No. 47 June 7994

Fig. 5. NYAME AKUMA No. 41 June 1994

Fig. 6. No. 41 June 1994

the general Ghana area dates to the Ghana Teaching Service, Brong Ahafo, ninth century A.D. at New Buipe. Ghana as well as my sons Patrick and These dates appear to support the Francis. To all these individuals and speculation that the Krachi ethnic group had organizations and also particularly to the moved into that part of the basin and Chief and elders of Bubuakro and Buafore established viable traditional political and specially my Research Assistant, Mr. groupings consisting of family groups or Seth Kumah of the Sociology Department, clans possibly as early as the early part of University of Ghana, Legon Accra, I give my the fifteenth century. The dates also indicate, very sincere thanks. if we are to assume that they represent the true picture of the period of occupation of References the basin by the two main groups in that portion of the basin, that the Krachi were in Agorsah, K the basin more than a century before the An ethnoarchaeological study of Nchumuru on the basis of mid seventeenth settlement and behavior patterns of a century dates from the site of Old Wiae, and West African traditional society: the almost a century before the earliest Nchumuru of Banda-Wiae in Ghana. Europeans set foot on the coast of Ghana. Of Ph.D. dissertation, UCLA, Univ. course these suggestions take the dates over Microfilms Int. Ann Arbor Mich (83- its widest range A.D. 1470 f 50 BP and 1810 21,948). f 50. New dates may require the revision of Social behavior and spatial context. this conclusion and at this point it is unlikely African Study Monographs, 4: 119-28, that this conclusion can be used for any University of Kyoto, Japan serious generalizations on the sequence of Archaeological implications of occupation of the basin. It is necessary to traditional house construction among obtain many more dates not only for the two the Nchumuru of Northern Ghana. ethnic groups, but for as many other groups Cumnt Anthropology 26 (1): 103-15. as share the basin today. The Volta basin undoubtedly is an area that needs to be Excavations in the Northern Volta much more closely examined to determine Basin. West African Journal of the colonization of much of what is Ghana Ardracology, 15: 11-40. and adjoining areas. This research is only The internal spatial organization of one of several, and it is hoped that the traditional houses in Northern Volta analysis, when completed, will enable more Basin of Ghana. Research Review, 2 (2): conaete generalizations to be made for a 101-34. Inst. African Studies, Legon better understanding of the relationships Settlement Pattern Analysis in the between settlements of the Krachi and other Northrm Vdta Basin. Archaeological ethnic groups who shared the Volta basin Research Report for 1983-85, presented and its resources. to the National Geographic Society and the University of Ghana, Legon Acknowledgments Evaluating spatial behavior patterns of This study was supported by the prehistoric societies. Journal of Research Committees of the University of Anthropological Archaeology, 7 (3): 231- Ghana Legon, Earthwatch and Center For 47. Field Research, Watertown, USA., The Settlement history of the northern Volta Ghana Museums and Monuments Board, Basin of Ghana. National Geographic Acaa, and the University of the West Indies, RrPcanh 4 (3): 371-85. Kingston, Jamaica. Participating in the Ethnoarchaeology: the search for a self- project were students and volunteers from corrective approach to the study of past Krachi Secondary School in the Volta human behavior. African Archaeological Region, Ghana, my brothers Felix Agorsah Review 8: 189-207. of the National Nursery Teachers Training Center, Acaa, and Francis Agorsah of the NYAME AKUMA No. 41 June 1994

Kense, E preliminary results of the surveys carried 1983a Daboya Project update. Nyame Akuma out in 1992 and 1993. 22: 10-12. 1983b Field report on Daboya, Ghana. Nyame Survey Procedure Aktan 23: 8-11. The initial focus of survey work at 1983~ Western Gonja Project (preliminary Sehonghong is the area within a two-hour report). Nyame Akuma 26: 18-20. walk of the main site. This contour is chosen as there is reason to think that the bulk of Stahl, A. the plant resources exploited from Sehong- 1983 Ghana. Nyame Aktan 22 11-12 hong, as well as some of the animal resources, will have fallen within it, but I do 1989 Protohistoric archaeology in the Banda not expect it to have acted as a leash to area. Nymne Akuma 32: 12-17. which people were inflexibly tethered; resources lying outside it will also have been LESOTHO exploited. Geology is an important variable in affecting the local archaeology and rock- shelters are almost entirely restricted to the First Results of the Survey of lower-lying Clarens sandstone. Within a Hunter-Gatherer Landscape Sehonghong's two-hour site territory some in the Lesotho Highlands 80-90% of the Clarens Formation has now been inspected. Above and on top of it, survey coverage is not as complete and has Peter J. Mitchell taken the form of linear transects. We have Department of Archaeology attempted to space these approximately I University of Wales km apart and to keep them straight, but the Lampeter location of villages, fields, and some SA48 7ED topographic features (e.g., steep diff-faces) U.K. acts as a constraint on our ability to do this. So far our coverage is more complete to the north of the Sehonghong River than to its Sehonghong rock-shelter in the eastern south, but there is no indication in the local highlands of Lesotho has a long sequence of ecology or topography to indicate that the Middle and Later Stone Age deposits, two areas should exhibit any differences in associated with good quality preservation of the kind or number of sites present. For the organics, that extend back well into the moment, we have not yet analysed Upper Pleistocene (Carter and Vogel 1974, assemblages in the field, and we have also Carter et al. 1988). Excavations here were refrained from collecting them. renewed in 1992 as part of a long-term investigation of late Quaternary hunter- Results gatherer adaptations in this part of southern In less than three weeks of work we Africa. As previously reported, (Mitchell have located 54 new Stone Age sites within 1993,1994a; Mitchell & Vogel in press), these our survey area. Together with sites have already shown that the occupation previously recorded by P. Carter (1978) and history of the Sehonghong site is much more P. Vinnicombe (1976) in the 1970s and by complex than previously suspected. Single Smit's (1973) ARAL Project in 1985, some of sites, no matter how rich their archaeology, which we have relocated; this brings the do not, however, exist on their own, total of sites known in the area to 91. Two of although they may well, at times, have been these can be attributed to the Early Stone the focal points of regional settlement Age (ESA), 30 to the Middle Stone Age systems. Consequently, an additional part of (MSA) and 24 to the Later Stone Age (LSA). work at Sehonghong is field-survey of the Five sites have evidence of both MSA and surrounding area. I report here on LSA occupation, five have undiagnostic NYAME AKUMA No. 41 June 1994

lithic scatters and six contain only used in identifying sites, but more isolated grindstones. In addition there are 19 painted "smears" of artefacts have been located sites for which associated lithic scatters are elsewhere in the survey area. Quany sites either absent or have not yet been reported. I remain elusive, although small outcrops of now discuss some of the key points that are opalines, baked sandstone, and dolerite beginning to emerge from the analysis of associated with some flaking activity have these sites and their patterning across the been located within the lavas of the Lesotho landscape. Formation; to date we have not identified It is abundantly clear that Sehonghong stone-working at the principal dolerite is the only major rock-shelter within the dykes within our survey area, while many survey area; no other site comes close to it in opalines may have been obtained as river- size, depth of deposit or richness of rock art. borne nodules. Of particular interest, However, there are at least three other however, has been the recognition, so far on shelters that have some archaeological a nonquantitative basis, of what appear to be deposit present and surface material of four different kinds of MSA assemblage. In probable post-classic Wilton Industry addition to those that seem to emphasize affiliation. Dates from Sehonghong (Mitchell small blades and flakes, we have found and Vogel in press) suggest that these sites others where flakes are large and exclusively will have dates of < 2000 BP; one of them in coarse-grained rocks. Assemblages that was observed in 1985 to contain the base of a are mostly made up of flakes in either pot at a depth of some 70 an below the opalines and/or coarser rocks, and surface and, although this has since been occurrences that have a strong blade/point lost to river erosion, sherds that appear to be component form the other two kinds of identical to those found in the uppermost MSA open site. The latter seem to occur in layers at Sehonghong were noted there in particularly dense scatters, for example at 1992. This pottery is, in its colour, crudeness the two collapsed shelters already of manufacture, mean thickness, etc., very mentioned. There is obvious scope here for different from local nineteenth-/twentieth- investigating the roles of chronology and century Basotho pottery. function in understanding this intersite The existence of rock-shelters in the variability. Sehonghong area that contain lithic Two LSA open air sites are perhaps the assemblages, but that lack paintings, has not most spectacular to have been found so far. previously been recorded. Nine such sites The first of these received brief mention in are now known, as well as two extensive an earlier report (Mitchell 1993) and lies on MSA open sites a little downstream of the the banks of the Orange River about I km Sehonghong site that lie immediately in upstream from the latter's confluence with front of and below what are probably the Sehonghong River. Five phases of collapsed rock-shelters. Representing occupation, all apparently LSA, are visible in another kind of shelter use are sites, the exposed section, separated from each sometimes no more than large upstanding other by sterile silts. Bone, including fish boulders, that contain one or a few vertebrae and the remains of dassie (Procavia grindstones, and nothing more. It seems capensis) and small bovids, is well-preserved likely that these artefacts have been cached and charcoals are also present. This site, at these locations, all within the river which is, to my knowledge, unique within valleys, for repeated use in on-thespot plant Lesotho, holds great potential for processing, although one has traces of red understanding how activities at an open-air ochre. site differed from those undertaken within Slightly more open air sites are known rock-shelters. More importantly, it may help than rock-shelters, and this preponderance establish if the marked pulsing evident at will ahnost certainly increase greatly as Sehonghong and in other southern African survey of the plateaux above the river rock-shelter sequences (Mitchell and Vogel valleys is intensified. An arbitrary cut-off in press) is truly a signature of regional point of 2 3 artefacts/square metre has been settlement history, or rather only of the NYAME AKUMA No. 41 June 1994

periods when rock-shelters were preferred exposed. A tendency for painted shelters to as habitation sites. Excavations at this site be located in the main river valley or its are planned for 1995. principal tributaries is also evident in both The second site lies on the plateau a few the Senqunyane and Sehonghong surveys. kilometres north of Sehonghong. Located in Furthermore, as in the PTB Basin rich open 1993, it consists of a series of at least five air MSA scatters are often in prominent piles of rock, no more than half a metre positions that afford excellent vantage high, strung out along a line on the southern points for monitoring the landscape, for side of a steep, tributary valley. Although example on knolls overlooking the close to a modern village, it is clearly confluence of the Orange and Sehonghong unrelated to either farming activity or house Rivers. ESA sites have only been recorded at construction; nor is it associated with a Sehonghong and cannot therefore be cemetery. Further investigation is required, compared with the other two areas. but I tentatively suggest that it may An important preliminary conclusion is represent the remains of a driveline used in that the archaeological sites within the the hunting of antelope. The rock piles Sehonghong site territory seem to exhibit a would have appeared more impressive in pronounced clumping. This needs testing by height when looked at from below, but may appropriate spatial analytical techniques also originally have been larger, or have (e.g., nearest-neighbour analysis, Hodder been extended by using branches, skins, etc. and Orton 1978), as well as by further The opposite slope is far too steep for survey. However, it appears for now that animals to have climbed up and the location both MSA and LSA sites are concentrated of these piles on the gentler incline would within a 2-3 km radius of Sehonghong, and thus have helped funnel game along the particularly along the Sehonghong Valley stream, and probably into a trap or ambush. and around its confluence with the Orange Both Arbousset (1852) and Barrow (1801) River. Beyond this area, a scatter of shelters, recorded the use of such drivelines by some painted, some with attendant artefact Bushmen in the early nineteenth century scatters and others with both, is found along and scenes showing the use of nets strung the Orange River, but open air artefact across narrow ravines to trap small bovids scatters are rare and typically consist of no are portrayed in the rock art of both the more than a few dozen artefacts. The western Cape (Manhire et al. 1985) and implication is that activities requiring Lesotho Winnicombe 1976: 292). extensive stone-working were limited spatially and that the factors affecting this Discussion (perhaps principally plant food and firewood availability?) affected both MSA Bousman (1988) has analysed the and LSA populations. distribution of sites by age and location in the Senqunyane Valley of south-central Such clumping of painted shelters close Lesotho, and I have carried out a similar to presumed major occupation sites (e.g., study for the rather different landscape of Sehonghong) was detected by Carter (19781, the Phuthiatsana-ea-Thaba Bosiu (PTB) who suggested that it might relate to the Basin in Lesotho's western lowlands employment of art as a form of territorial (Mitchell 1994b). The distribution of sites marking. Making full use of the fuller, around Sehonghong is in agreement with ethnographically grounded interpretations these studies in suggesting that MSA sites of San art that have been developed since are most typically found in open locations, then (e.g., Lewis-Williams and Dowson and LSA sites within rock-shelters; ESA sites 1989), I plan to examine the spatial are known only at Sehonghong and are distribution of particular themes within the restricted to the flats lying above the Orange art (such as explicit trance imagery, conflict River (Carter 1978). As in the Senqunyane scenes, domestic livestock). The analysis, Valley, rock-shelters are overwhelmingly which will benefit from the use of GIs concentrated in the river valleys because this technology, will hopefully add a cognitive is where the Clarens Formation is typically dimension to the study of hunter-gatherer NYAME AKUM No. 41 June 1994

landscapes at Sehonghong and perhaps help help of D. Turner, M. Clark, J-J. White, in identifying particular landscape locations P. Mann, R. Charles, G. Hall, or features as being of special symbolic T. MacManmon, and D. Watt. To all of them, sigruficance, as discussed by Deacon (1988) for their companionship, enthusiasm, and for the /Xam Bushmen of South Africa's keen sight, I am extremely grateful. Northern Cape Province. Fieldwork at Sehonghong has been Another striking feature of our survey is undertaken with funding from the the absence of flaked stone artefacts above University of Cape Town, the Swan Fund, about 2,050 metres in altitude. This is not the Boise Fund, the Society of Antiquaries, unexpected, since these higher altitude areas the University of Oxford, the Harry totally lack natural shelter and firewood and Oppenheimer Institute for African Studies, were almost certainly poor in both animal and the Prehistoric Society. I am also and plant resources relative to those below grateful to the Swan, Boise, and Wilkin them. It nevertheless reinforces the view Funds, Rhodes House, New College, (Mitchell 1994b) that the higher parts of the Oxford, and St. Anne's College, Oxford for Maluti Mountains are essentially lacking in grants to members of the 1992 team. archaeological evidence. Sites here will be restricted to locations where particular References resources (such as high quality stone) can be found (e.g., Parkington and Mitchell 1993) or Arbousset, T. to lines of movement between major valley 1852 Narrative of an Exploratoy Four to tke systems, such as the Matebeng Pass North-east of the Cape of Good Hope. (P. Carter, pers. comm.). London: J.C. Bishop. Both these features, which have the effect of creating archaeological "hotspots" Barrow, J. with a much lower density of sites in- 1801 Travels into the Interior of Southern between, are, I would suggest, likely to be Africa, 1797-1798. London: Cadell and repeated elsewhere. Indeed, it is already Davies. possible to suggest that the areas around the sites of Matebeng and Melikane, the only Bousman, B. other two large rock-shelters along this 1988 Prehistoric settlement in the portion of the Orange River (Carter 1978) Senqunyane Valley, Lesotho. South and each centred on another of its major African A~haedogicalBulletin 43: 33-37. west-flowing tributaries, will prove to host additional site concentrations. Directed Carter, P. L towards understanding the use made of the 1978 The Prehisto y of Eastern Lesotho. Ph.D. regional landscape by Stone Age people, the thesis, University of Cambridge. comparison of intersite variability within and between all three of these areas will be Carter, P. L, and Vogel, J. C. one of the long-term goals of continued field-survey in the Lesotho highlands. 1974 The dating of industrial assemblages from stratified sites in eastern Lesotho. Mmr 9: 557-70. Acknowledgments I am grateful to Mrs. N. Khitsane, Carter, P. L, Mitchell, P. J., and Vinnicombe, P. Chairman of the Protection and Preservation 1988 Sehonghong: the Middle and Later Stone Commission of Lesotho, for permission to Age Industrinl Sequence at a Lesotho Rock- conduct fieldwork in the Sehonghong area, shelter. Oxford: British Archaeological as well as to the Chiefs of Sehonghong and Reports. Khomo-ea-Mollo. The assistance of Mr. M. Phutsoe and the pilots of Senqu Air Services was also invaluable. Field survey would have been impossible without the No. 41 June 1994

Deacon, J. Smits, L G. A. 1988 The power of a place in understanding 1973 Rock-painting sites in the upper Senqu southern San rock engravings. Wd Valley, Lesotho. South African Archlogy 20: 129-40. Archnmlogical Bullctin 28: 32-38.

Hodder, I, and Orton, C Vinnicombe, P. 1976 Spatial Analysis in Archaeology. 1976 Pwple of the Elnnd. Pietermaritzburg: Cambridge: Cambridge University University of Natal Press. Press.

Lewis-Wiiams, J. D., and Dowson, T. A. 1989 Images of Power. Johannesburg: Southern Books.

Mitchell, P. J. 1993 Late Pleistocene hunter-gatherers in Lesotho: report on excavations at Sehonghong rock-shelter, July to September 1992. Nyame Akuma 39: 43- 49.

Mitchell, P. J. 1994a Understanding the MSA/LSA transition: the pre-20 000 BP assem- blages from new excavations at Sehonghong rock-shelter, Lesotho. Southern African Field Archaeology 3: 15-25.

Mitchell, P. J. 1994b The archaeology of the Phuthiatsana- ea-Thaba Bosiu Basin, western Lesotho, southern Africa: changes in Later Stone Age regional demography. Antiquity 68-83-96.

Mitchell, P. J., and Vogel, J. C in #em radiocarbon dates from Sehonghong rock-shelter, Lesotho. South Afican Journal of Science.

Parkington, J. E., and Mitchell, P. J. 1993 Archaeological Swey of the Phase IB Area of the Lesotho Highlands Water Project. Report submitted to the Lesotho Highlands Development Authority. NYAME AKUMA No. 47 June 1994

MADAGASCAR The capital of the royal clan of the Antandroy flourished between the seventeenth and nineteenth centuries. Its position and former significance were Finding Fenoarivo: largely forgotten. It was found by a Fieldwork in Central Androy combination of written sources, place names, oral tradition, and field swey. Mike Parker Pearson Department of Archaeology and Fenoarivo Prehistory In 1729 Robert Drury's journal referred to University of She ffield the capital of the Antandroy, "Fenno-arivo," She ffield S10 2TN located three days walk from the south coast U.K. and over 4 days walk from the Mandrare River to the east. It was here that the aew of Karen Godden the wrecked East Indiaman Degrave were Royal Anthropological Institute taken before their abortive kidnap attempt, 50 Fitzroy Street escape, and massacre in 1703. Fenoarivo was London W7 P 5HS described in the journal as standing in a U.K. wood, defended by a planted wall of thorny, living trees. It had two narrow entrances, on the south and north sides, and was Ramilisonina estimated to be a mile in circumference lnstitut de Civilization (Drury [I7291 1890,59). A few years later, it Musee dirt et d'Archeologie was sacked by the neighbouring Mahafaly B. P. 564 and subsequently rebuilt ([I7291 1890, 89- Antananarivo 10 1 90). Drury's document is of particular Madagascar interest because opinion is divided amongst scholars as to whether it is the accurate Retsihisatse account of a survivor or a work of Analamahery semifiction composed by Daniel Defoe (see FIR Andalatanosy Furbank and Owens 1988: 109-13, Molet- Ambovombe 604 Sauvaget 1992). Madagascar Other clues for locating Fenoarivo were derived from placename evidence and the present day distribution of the former royal Jean-Luc Schwenninger clan. Within the area of the main Department of Geography concentration of this clan is the dried-up Royal Holloway College lake of Fenoarivo ("full of a thousand"), University of London named after the small palace of the last king. Egham Intensive fieldwalking of an area of 15 Middlesex square kilometres in its vicinity revealed a U. K. sequence of single large sites from the mid seventeenth century until the mid nineteenth century. Since all timber Fieldwork in 1993 resulted in the buildings had rotted and the defences had discovery of the remains of the lost capital of left no trace, the only surface indications of the Antandroy, between the modem villages these sites were scatters of pottery and dark of Ambaro and Laparoy south of soil. The largest site of about 21 hectares was Antanimora (Fig. 1, 2). Further work was the last in the sequence of five. The earlier also continued in the coastal dunes to ones were much smaller at 1.5,5,12, and 15 investigate the relationship between human hectares, respectively. Their shifting habitations and Aepyornis (elephant bird) locations (over 4 km.in 300 years) indicates eggshell scatters. that they were abandoned at fairly regular NYAME AKUMA No. 41 June 1994

Fig. 1. Locations of areas surveyad in Androy, southern Madagascar (1988 area surveyed by Georges Heurtebize [Heurtebize 19861). NYAME AKUMA No. 4 1 June 1994

Fig. 2. The ambaro-Laparoy survey area. Sites were recognized as field scatters in this large sandy basin surrounded by forest.

intervals. The ceramics indicate a sequence The Regional Context of Fenoarivo of Antandroy styles beginning in the The 1993 survey area shows some seventeenth century. Some of the earlier pronounced differences from survey results ceramic styles with graphite surfaces and twenty to fifty kilometres further north impressed triangles can be matched with (Heurtebize 1986: 77-83, Parker Pearson later seventeenthcentury assemblages from 1992, Radimilahy 1988: 122-62). In the Ehoala in south-east Madagascar (Wright, upland and stony landscape a short distance Rakotoarisoa, Heurtebize, and V&in 1993), to the north, the stone-walled enclosures are but no overseas ceramic imports were found associated with pottery dating from the in pre-nineteenthcentury contexts. eleventh century until the seven teenth Prior to the seventeenth century, there century. The area was apparently deserted were two small sixteenth-seventeenth- until the mid or later nineteenth century century settlements and a larger thirteenth- when small, dispersed Antandroy fifteenthcentury site (1.3 ha. in extent). settlements were established here. We may After the mid-ninteenth century the speculate that the people living in the hills settlement pattern dispersed into 37 small were replaced, driven out, or conquered by settlements, some of which were loosely the newly emergent Antandroy kingdom in grouped into three clusters, the antecedents the seventeenth century, and that power and of the modem villages. We condude that population were subsequently centralized in Antandroy political centralization occurred the capital and regional subcentres until during the later seventeenth century and after 1850. It was with this dispersal that that the capital fragmented in the mid nonroyal clans initiated the fashion of ninteenth century. This latter observation monumental stone tombs so characteristic of tallies with oral traditions of the collapse of the Antandroy landscape today. the monarchy and the dispersal of the clans. NYAME AKUMA No. 41 June 1994

Limited survey in two other areas Laparoy, Ankiliabo, and Analamahery and revealed a further six archaeological sites. the town council of Antanimora must be One of these areas was on top of the heights thanked for permissions and hospitality. of Angavo, where oral traditions indicate This season's fieldwork was supported by that past kings were buried in a cave. On the grants from the Nuffield Foundation, the summit we located a large stone-walled British Academy, the Society of Antiquaries, enclosure of uncertain date and a small and the Department of Archaeology and sixteenth-seventeenth century settlement. Prehistory, University of Sheffield.

Aepyornis Remains References Research in the coastal dunes at Berger, R, Ducote, K, Robinson, K, and Maroaloke and at Kotoala located some Walter, H. large and dense scatters of Aepyomis 1975 Radiocarbon date for the largest extinct eggshells. These flightless birds were the bird. Nature 258: 709. largest ever recorded and are also known as el&hant birds on account of their huge Dewar, R legbones. Their extinction is thought to have Extinctions in Madagascar: the loss of been due to human predation on their eggs, 1984 the sub-fossil fauna. In P. Martin and although climatic change, ecological R competition by humans, and large-scale Klein (eds.), Quafenwry Extinctions, culling might also have been responsible pp. 574-93. Tucson: University of (Vkrin 1965). Existing radiocarbon Arizona Pmss. determinations for eggshell fragments and seventeenthcentury documentary sources Dw,R show that the bird survived until fewer than [lZ3]1890 Madagascac or Robert Dncy's Journal a thousand years ago (Berger, Ducote, during Fifteen Years Captimty on that Robinson, and Walter 1975; Dewar 1984). Island. Londorc Fisher Unwin. Despite our discovery of four 1000-year-old settlement sites in close proximity to the Furbank, P. N., and Owens, W. R eggshell scatters, we were unable to 1988 The Cnnonization of Daniel Defoe. New demonstrate with certainty that Aepyomis Haven: Yale University Press. shell fragments were directly associated with human activity. Conversely, the Heurtebii, G. absence of stone, burnt stone, or other 1986 Histoirr ahAfomnTolahy (Extr&me-Sud de cultural debris amongst the shell scatters Madagascar). Paris: CNRS. suggests that the scatters were nesting sites. We are embarking on a programme of Molet-Sauvaget, A. radiocarbon and optically stimulated le luminescence dating of stratified dune 1992 Madagascar ou Jmrnul de Robert Dmy sequences to establish the chronological par Daniel +. Paris: Harmattan. contexts of human and Aepyornis occupation. Parker Pearson, M. 1992 Tombs and monumentality in southern Madagascar: preliminary results of the Acknowledgments central Androy project. Antiquity 66: Many people contributed to the success 941-8. of the 1993 season. Victor Razanatovo looked after the cooking, driving, and Land Radimilahy, C. Rover maintenance. In particular we would 1988 L'Ancienne M&tallurgie du Fer d like to thank Georges Heurtebize, Chris Madagascar. British Archaeological Poole, Chantal Radimilahy, Jean-Aim6 Reports, International Series 422, Rakotoarisoa, Pierre Vkrin, and Jeremy Oxford. Viewing. The communities of Ambaro, NYAME AKUlMA No. 41 June 1994

V&n, P. It should be noted that the findings 1965 Pmspection des sites andens du littoral outlined here are by necessity brief and du Sud-Ouest de Madagascar. preliminary as the analysis of the material Mndagascar Raw& Geographic 6: 133-7. generated both by the excavations in Gao and the survey of the site of Koyma located Wright, H. T., Rakotoarisoa, J.-A., Heurtebize, G., on the opposite bank of the Niger to Gao is and V6riq P. not yet complete. 1993 The evolution of settlement systems in the Efaho River Valley, Anosy: a The Mosque of Mansa Musa preliminary report on archaeological Three possible zones for excavation in reconnaissances of 1988-1986. Indo- the mosque area were isolated in January Pacific Prehistory Associntim Bulletin 12: 1993 (Insoll 1993a: 42). The first zone 2-20. described as being that of the mosque is a misidentification as the area which MALI Raymond Mauny (1951: 843) excavated in 1950 is in reality 100 m. to the northeast of the area stated previously as being the location of the mosque. This mistake became Preliminary Results of apparent upon viewing the site after the Excavations at Gao, effects of the year's rains had been felt and September and October 1993 after discrepancies between different informants' accounts were noticed. The Timothy Insoll trench was in the end laid out in the area Department of Archaeology corresponding to the third zone of the so- University of Cambridge called Mosque of Mansa Musa sector. Downing Street The excavations revealed the presence in Cambridge CB2 3DZ the western half of the trench of the remains U.K. of a floor and what appears to be a column base constructed out of baked bricks of fairly uniform dimensions. The bricks which were used to build the column base are mould Excavations were conducted at two sites cm in the vicinity of the city of Gao in the formed and range in size from 21 length cm cm en Republic of Mali in September and October by 21.5 width by 6 depth to 26 cm cm This 1993. A 6 x 65 m. trench was excavated in length by 25 width by 7 depth. old Gao or Gao Ancien in the area known building of as yet unknown function was locally as the place where the Malian abutted by the remains of three further walls constructed respectively out of reused baked emperor Mansa Musa built his mosque on bricks. his return from pilgrimage in the fourteenth brick fragments, banco, and banco At century AD.The second 2 x 2 m. trench was present it would appear that these more excavated at the northwestern end of the crudely constructed walls represent a later habitation mound at Saney, located 4 km to period of squatter occupation. the east of the modern city of Gao. The The top 40 an of the deposits in part of potential of both these sites for the author's the western half of the site was subject to doctoral research into the spread and some contamination that was probably acceptance of Islam in the Western Sahel caused by the activity of robbers looking for between 800-1200 A.D. was revealed by a construction materials. They had left survey undertaken at Gao in January 1993, indications of their activities in the form of a when it was recognized that excavations narrow trench that neatly cut through part were urgently needed before these of the baked brick floor to a depth of 60 cm important sites were totally destroyed below the surface level. At a further 20 cm (Ins011 1993a, 1993b). depth a dense layer of glass, pottery, metal fragments, and beads was found. This layer NYAME AKUMA No. 47 June 1994

was encountered on the eastern half of the wares described by S. and R. McIntosh site, but the removal of four of the baked (1984: 33-35). brick flooring slabs showed that this layer The crude percentage figures for the also continued into the western half of the whole site are as follows (all sherds were site and was sealed by the baked brick floor. recorded). Undecorated red slipped wares This layer is of importance as it were the most commonly found, making up provided a sample of ceramics imported 37.4% of the 4883 identifiable sherds; from the wider Islamic world from which a roulette and twine decorated wares, 37.2%; preliminary date for this site could be deeply channelled wares, 11.9%; painted obtained. C14 samples have also been taken wares, 6.4%; geometric comb-impressed but at the time of writing these have not yet wares, 1.5%; undecorated black and brown been processed. The imported ceramics and burnished wares, 2.2%; miscellaneous wares glass from this layer were examined by the (drainage spout fragments, large coarse Department of Oriental Antiquities, the wares, etc), 3.4%. There is a slight change British Museum, London; and the noticeable in the types of wares present assemblage has been dated provisionally to midway in the site sequence; however, the the eleventh and twelfth centuries A.D., with results of the C14 analysis are needed before some elements of a possible tenth century the significance of this can be explained. A.D. date also present. In total 147 sherds of Certainly the most spectacular find imported glass and 36 sherds of imported made during the course of the excavations at pottery were recovered from the site, with the "Mosque of Mansa Musa" site was a pile 34 sherds of glass and 143 sherds of of approximately 53 hippopotamus tusks imported pottery coming from this level (The final number will not be known until alone. The origins of this material show that cleaning and conservation are complete). Gao was receiving material from more than The pile of tusks was found at a depth of one source, as lustred glass from Egypt, cut, 1.35 metres below the surface and could scratched, and trailed glass from Kairouan represent either a consignment of ivory and glazed pottery from Ifriqiya in North waiting to be shipped to North Africa that Africa were all found. was forgotten for some reason, a horde, or A large variety of beads of stone, bone, some form of trophy. metal, and glass were recovered from the At this stage in the proceedings it is not site along with a total of 206 pieces of iron possible to offer too many interpretations. and copper, with again the majority of this However, the evidence of the well- material coming from the level rich in constructed baked brick building and the imported ceramics and glass. The presence of the material imported from metalwork found included the remains of a North Africa would appear to indicate that finely worked bronze chain and what in Gao there were individuals who had appear to be fragments of two iron developed tastes for material of good pendants, several knives of different sizes, a quality, perhaps developed through contact hoe blade, 3 copper coins, and a variety of with North Africa or North African traders. nails and hooks. The excavations at the "Mansa Musa" site The locally made pottery recovered have also provided proof that Gao was from the excavations at the "Mosque of linked into the trans-Saharan trade Mansa Musa" site has some similarities with networks, during the period provisionally the pottery collected during January 1993 at dated to the eleventh and twelfth centuries Saney and in Gao Anaen. The assemblage A. D. again contains certain elements that are, as has already been noted (Insoll1993a: 42431, Saney similar to the material collected by S. and R. McIntosh (1984) at Gourma Rharous, to the Due to a lack of time it was deaded not northwest of Gao on the Niger Bend. These to excavate in the Royal cemetery at Saney; include the red slipped wares, deeply instead attention was focussed upon the channelled wares, and black burnished habitation mound. However, the results NYAME AKUMA No. 41 June 1994

obtained from these excavations proved to The site of Koyma is surmounted by a be slightly disappointing. Excavation large red sand dune and would appear to be revealed that the deposits were largely multiperiod in date. The remains of six composed of a fine grey ash, making it skeletons were found eroding out of the difficult to keep the sides of the excavation lower slopes of the dune in an area directly unit straight, impairing visibility of the fronting the river. These skeletons were stratigraphic profile and reducing the oriented with the head to the East and the working area. Additionally the threat of the feet to the west, as opposed to the Muslim trench collapsing was ever present, posing North-South orientation. No grave goods dangers for those working in the trench. For were found in association with the reasons of safety, excavations were stopped skeletons. at a depth of 2.30 m. below surface level, A single cowrie shell was found along before sterile deposits were reached. with a variety of different types of pottery. Owing to the nature of the deposits, a Although only a preliminary investigation midden, no structural remains were has been completed, it is still possible to encountered. A sizeable ceramic assemblage note a definite difference between this was recovered along with a good collection ceramic assemblage and those of Saney and of faunal material and a number of C14 "Mansa Musa." samples, all of which are in the process of being analysed and processed. The pottery Conclusions types encountered are similar to those found at the "Mansa Musa" site, though their The results achieved from the overall percentages do differ (50%sample of excavations and the survey work in the Gao sherds recorded). Deeply channelled wares region were much better than expected. A are the most commonly found at 35% of the good body of material has now been 654 identifiable sherds; red slipped wares, recovered and it is hoped that the detailed 18.2%; roulette and twine decorated wares, analysis of this material will begin to give us 33.3%; geometric comb-impressed wares, an insight into the development of Gao, its 1.1 %; painted wares, 7.6%; and trading partners, and the trade routes that miscellaneous wares, 4.8%. operated during the early second millenium A.D. It is hoped to continue excavations in A surface collection of archaeological 1995. material was again made aaoss the whole of Gao in late the habitation mound. Several sherds of abraded glazed pottery were found along Acknowledgments with a variety of beads, further mould I am very grateful to Dr. Sanogo, the fragments, and slag. A single sherd of pale director, and Dr. Dembele, the assistant green glazed pottery, similar to celadon director, of the Institut des Sciences ware, and probably of South East Asian Humaines, and Dr. Iam, the director of the origin was also found. Centre Nationale de la Recherche Scientifique et Technologique in Bamako for Koyma allowing me to conduct my research. I am A third site, Koyma, was briefly visited, also grateful to Monsieur Sekou Berthe and and a surface collection was undertaken. Nafogo Coulibaly for help at the "Mansa Koyma is located 4 km north of Gao on the Musa" site, and to Dr. Tereba Togola for opposite bank of the River Niger. Koyma help at Koyma and Saney. Thanks are also has been suggested as being a likely location due to Monsieur Toure, Chef de Division du for the site of the first Songhai settlement in Patrimoine Culturel, for practical assistance the Gao region, and according to oral in Gao. Funding for this research was kindly tradition was used as a port by the Songhai provided by the Emslie Horniman trust of Askiya rulers of the fifteenth and sixteenth the Royal Anthropological hstitutc, the British Academy, and the University of centuries A.D. (Cisse 1975: 166; Toure 1992: 11). Cambridge. NYAME AKUMA No. 47 June 7994

References H NIGERIA Cisse, B. 1975 Intervention. Actes du Colloque. Histoire et Tradition Omle. Ikc annee: L'Empiredu Preliminary Results of Mi.Paris: Copedith, pp. 155-66. Research by the Projet Maya- Wandala, Nigeria, 1993 Insoll, T. 1993a A preliminary reconnaissance and survey at Gao. Nyame Akuma 39: 40-43. Scott MacEachern Department of Archaeology 1993b Looting the antiquities of Mali: the University of Calgary story continues at Gao. Antiquity 67: Calgary, Alberta, T2N 1N4 628-32 Canada

1951 Notes d'arch6ologie au sujet de Gao. Abbakar Garba Bulletin de l'lnstitut Fondamental Centre for Trans-Saharan Studies d'Afrique Noire (B) 13: 83742. University of Maiduguri P.M. B. 1069 McIntosh, S. K, and R J. Maiduguri, Nigeria 1984 Archaeological reconnaissance in the region of Timbuktu, Mali. Final report to the National Geographic Society. Houston: The second field season of the Projet Texas. Maya-Wandala took place between 19 May and 31 August, 1992, in Borno State of Toure, M E. Nigeria. The crew for this season was made up of researchers from the University of 1992 Report on the antiquities and history of Calgary, the University of Maiduguri, and the Gao region. Untitled and the National Commission for Museums and unpublished paper. Gao: Division du Monuments, Nigeria. Research in 1993 was Patrimoine Culturel, Direction conducted from Gwoza, the centre of Gwoza Jeunesse. Local Government Area, and focused upon archaeological survey and excavation around the Mandara Mountains, which have their northernmost extension in this region (Fig. 1, 2). In addition, ethnohistorical enquiries were an important component of this season's research, as they have been in previous years. The objective of this work is to examine the historical relationships between populations found in this area today and in the recent past (i.e., over the last 150-200 years) and earlier populations, now remembered only as the semi- legendary 'Sao' and 'Maya' peoples. We wished to study the processes by which these earlier Iron Age plains populations had been supplanted by later groups, and especially by the Wandala, inhabitants of a small state of the same name who on linguistic and cultural grounds seem to have resided in the region for a long time. No. 47 June 1998

Fig. 1. The Projet Maya-Wandala survey area in Nigeria and Cameroon.

49 NYAME AKUMA No. 41 June 1994 NYAME AKUMA No. 41 June 1994

Archaeological Survey and [Ghwa Masogol and PMW 767 [Gwalagwa Site Inventory Ngarel) located potentially so significant Fieldwork over the period 4 June-22 that we decided to conduct excavations on August, 1993, resulted in the discovery of 58 this latter site. archaeological sites. (At least two of the sites Survey design was directed toward located were first found by Dr. Graham maximizing the number and variety of sites Connah [Connah 19841 in his 1981 survey of found in a limited time, and made use of the the area. These are our sites PMW 744 survey experience that we have [Ghwa Kival/Connah's B123 [Ndufal and accumulated in the course of previous our site PMW 761 [Dugje Gagal/Connah's research. We concentrated on examining B121 [Ugvake Thuktsel. It should also be massif- and inselberg-edge areas, but also note that Connah's 11984: Fig. 1, Table 41 site conducted survey work in the Mandara B117, which yielded a convincing stone massif itself and on the plains at some industry, appears to have been destroyed by kilometres distance from the mountains. We rock quarrying.) In addition, surface quickly focused on three different areas that collections were made in 12 localities where showed exceptional promise during our significant numbers of artefacts were found research: (a) the borders of the Kirawa but artefact densities did not warrant inselberg, on the Cameroon-Nigeria border, designation as a site (Fig. 2). The vast the first generally accepted Wandala capital; majority of Projet Maya-Wandala sites and (b) the northwest tip of the massif around the components that make them up are and to the southwest of the village of Puke; datable to the Iron Age, and probably to the and (c) the inselbergs in the plains to the last 1000-1200 years. However, in 1993 we west of Gwoza town, traditionally part of found two important sites dating to the the Wandala province of Kambwa (Fig. 2). Neolithic period (tentatively before ca 1 A.D. All three areas are closely linked to Wandala in this area). Neolithic material is so rare in occupation in oral and written histories. this area, and the Neolithic sites (PMW 756

Tablo 1. Radiocarbon dates from Projet Maya-Wandal surveys in 1992 and 1993.

Site bt Unit Level Designator Date* PMW602 1101 Pit 2 SW 5 Beta-61583 modern PMW602 1122 Pit 2 SW 8 Beta41584 150 *SO bp PMW602 1237 Pit 2 SE 24 Beta41586 1700 f 90 bp PMW744 93-566 Pit I NW 10 Beta49019 I050 f 70 bp (CAMS-10919) PMW744 93-1207 Pit 1NW 23 Beta49021 1630 f 60 bp 1-74 1 I----pit4~~i 10 ~e6-69018 2740 f 60 bp

+ Dates are 13C corrected, but not otherwise calibrated. NYAME AKUMA No. 47 June 7994

Most sites discovered, and all of the stands in the sorghum fields of the modem large sites, were located along the periphery community of Kiva. The surface of this of the Mandara massif or near inselbergs. mound is densely covered with Iron Age This parallels the results of our 1992 potsherds, and its cultural significance is research in Cameroon. Site densities are obvious at first glance. The area considerably lower in plains environments surrounding the mound is covered by a than along slope edges, while intense moderately dense scatter of Iron Age montagnard occupation in the massif itself artefacts, to a depth of ca 1 m. makes detection and discrimination of sites We eventually opened three units on the in that environment difficult. There is an site; Pit 1 was the most significant of these. obvious concentration of sites around Cultural material was found through the top Kirawa, Puke, Gwoza, and the inselbergs 260 an of the unit, indicating that almost all west of the massif. The number of sites at the mound's height was artificial deposits, Puke and along the massif just to the possibly on top of a slight natural rise. Only southwest is quite striking. Sites are more preliminary analysis of the ceramics from numerous and larger along the northern and any of the units excavated at Ghwa Kiva has eastern edge of the Kirawa inselberg than been done, but there seems to be broad along the rest of its periphery, which continuity in decoration and morphology, parallels the situation at the nearby and broad resemblance to present-day inselberg of Grba, in Cameroon ceramics, to ca 230 an bd. Below that, many (MacEachern 1993a). It has been argued sherds are decorated with incision and (MacEachern 1990: 178, 1993b) that one comb-stamping, rather than the typical Iron possible reason for the success of the Age rouletted material above. An iron Wandala state vis-a-vis local competitors fragment was found in a probably sealed like the Marghi was its advantageous context at 240 cm below datum. We may position for the control of an iron trade here have an industry transitional between between the northwestern edge of the the Neolithic and Iron Age, with ceramics Mandara Mountains and the Kanuri capital similar to the former but with some use of of Birni Ngazargamo. This site patterning iron. might provide support for such an The radiocarbon dates available hypothesis. We found a significant number (Table 1) and the lack of evidence for of sites near Kirawa inselberg, including interruptions in the occupation sequence some (Ngoye Kirawa, Dugje Gaga [PMW indicate that Ghwa Kiva has been occupied 7611, Gakara I-Iill/Kirawa N [PMW 7531, with no obvious hiatuses for much of the Thuliva Kwacha [PMW 7681 and the Mayo last 1700 years. The radiocarbon date of 1630 Kirawa Burials [PMW 7691) that would f 60 bp (Beta-69021) refers to one of the certainly repay further work. The mound earliest Iron Age (as usually defined) levels site of Dugje Gaga, for example, would in the site; similar dates have been derived require some months of excavation at least from early Iron Age levels at Grba- to make work on it worthwhile; it appears to Manaouatchi (PMW 602) and Mehe Djiddere be ca seven metres high. MAP 523) in Cameroon, making it possible that this was the period of establishment of Excavated Sites the local Iron Age industries that seem to be Ghwa Kiva (PMW 744) ancestral to those of present peoples in the area. The Ghwa Kiva site is located 200 m from the western side of the Mandara Ghwa Mesogo (PMW 756) Mountains, and 3 km southwest of the village of Puke (Fig. 2); it was first located Ghwa Masogo is the remains of a by Dr. Graham Connah (1984). The most Neolithic occupation-the first large, well- obvious sign of ancient human habitation on defined surface Neolithic site to have been this site is a single artificial mound, about found near the northern massif. It is located 3 m high and 45 m in total diameter, which in the plains to the west of the town of NYAME AKUMA No. 41 June 7334

Gwoza, at the base of a small hillock of a granary or similar structure. Our other primarily made of quartz. Neolithic units at this site located a probable midden populations in the region seem, in many and pit feature (Pit 4) or were completely cases, to have lived out on the plains away sterile (Pit 2). As stated above, Ghwa from the immediate environs of the Masogo is a complicated, as well as mountains, as at Blabli (MAP 506) and potentially an important, Neolithic site. It is Gwalagwa Ngare (PMW 767). The site itself noteworthy that the three units opened (Pits consists of a low-/mediumdensity scatter of 1/3,2, and 4) each revealed quite different lithic artefacts and potsherds over an area of stratigraphies and cultural remains. This site ca 300 x 150 m, just to the southeast of the would benefit from large-scale excavation. Ghwa Masogo hillock. It is probable that Ngoye Kirawa (PMW 755) some site deflation has taken place; we would roughly estimate that between 30 an Ngoye Kirawa is a large (ca 750 m x and 50 an of soil has been removed from the 300 m) complex of terraces, walls, and house top of the stratigraphic profile. At least structures just to the northwest of Kirawa twenty features (generally between 3 m and town. In layout and complexity it seems to 5 m in diameter) consist of irregular surface have been a community of substantial size, concentrations of rock (sometimes with while its position at the base of the inselberg evidence of heating), daub, grindstones and somewhat resembles that of Manaouatchi- grindstone fragments, and/or lithic GA(PMW 602) in Cameroon (MacEachern artefacts. Surface collection recovered 1993a). We decided to conduct a test significant amounts of Neolithic pottery, excavation on the terrace system at the along with ground and flaked stone former site in order to determine site depth axes/bifaces, axe fragments, grindstones and (hopefully) to recover organic material and grindstone fragments, and other lithic for radiocarbon dating. Tenace systems are artefacts. A large number of quartz flakes ubiquitous in the northern Mandara area, and fragments were also recovered; it is very but we have no direct data on the possible difficult to tell whether these pieces were ages of such systems-this must be a intentionally produced for use as knives, priority for future research. The two 2 m x scrapers and so on, or whether they are 1 m units excavated were disturbed by hoe simply fragments from the quartz deposits farming, and consisted of very soft, loamy that make up Ghwa Masogo. A probable soil with some Iron Age pottery occurring to MSA biface (evidence of an early occupation ca 100 an below the surface. A sample of of this region by humans [Soper 19651) was charcoal from Pit 2, Level 8 has been recovered from the first surface collection on submitted to IsoTrace of Canada for AMS this site, along the southern extremity of the radiocarbon dating. site, close to the small stream that bounds PMW 756 to the south. Excavation in Pits 1 and 3 uncovered a The concentration of sites around circular structure of stone, daub, and massif- and inselberg-edges detected in grindstone fragments from about 5 cm research in Cameroon appears to also hold below ground surface. This feature was for Nigeria. This probably is accounted for cone-shaped in profile, about 150 an in by considerations of defense and of access to diameter at ground surface and narrowing other resources--stone, iron, possibly good to ca 20 an at 32 an below datum, feature soils and water-during the Iron Age. It will bottom. It consisted of rocks with an average be necessary to determine if, and how, the size of about 10 an, probably laid into a relative weight of these considerations daub matrix and then fired. Approximately changed during the last two millennia. thirty grindstone fragments were recovered Neolithic populations appear to have been from the feature, as were small numbers of rather less tied to such environments, potsherds and some small bone fragments. possibly because of the lack of encroachment No direct present-day analogues for this by state societies during that period. The feature exist, although it might be the base distribution of sites, and especially the No. 41 June 1994

concentration around Pulke, may also lend Commission for Museums and Monuments, support to the hypothesis that trade with in both Lagos and Maiduguri, were Bornuan communities to the north was one uniformly helpful in providing the project of the engines that drove the differentiation crew with advice and assistance; We and development of the Wandala state. particularly want to thank Dr. L. I. Izuakor, We have located a good deal more Director of Research and Training for the evidence of Neolithic occupation in the area NCMM, for his kindness to MacEachern west of the Mandara Mountains than to the while in Lagos. east, in Cameroon. We discovered two We would also like to thank the staff of Neolithic sites during the 1993 research, as Gwoza Local Government Area, and well as traces of Neolithic (and probably particularly the Chairman and Vice- earlier) occupation from a number of other Chairman, for their help and for the localities, while only one (quite small) site welcome to Gwoza that they provided to us. (Blabli [PMW 5061) has been located over an Mr. Buba Karou, of Gwoza, ably provided extended period of research in Cameroon. translation services and many other sorts of This is probably due to the vagaries of assistance, and made our stay in Gwoza sampling, since there is no evidence that the much easier. We would particularly like to area west of the massif was favoured in any thank the people of Kiva, Loku-Disa, and way. One priority for further research will Kirawa districts for the help that they be examination of the nature of the provided to us in our survey and Neolithic-Iron Age transition in this area. excavation. The research conducted in Results from a number of sites, including Nigeria in 1993 was paid for by a Research Ghwa Kiva (PMW 7441, indicate that a Grant (number 410-92-1860) from the Social transition period did exist, and that it Sciences and Humanities Research Council involved iron use but general continuity in of Canada. ceramic traditions from the Neolithic. The transition to a "developed" Iron Age in the References region appears to occur around or just before A.D. 300. This is also a period that Brooks, G. Brooks (1993: 7-9) identifies as one of drastic 1993 Landlords and Strangers: Ecology, Society environmental change in West Africa in and Trade in Western Afnca, 1000-1630. general, which may have some relevance to Boulder, CO: Westview Press. these events. Projet Maya-Wandala research during Connah, G. the 1993 field season was outstandingly 1984 An archaeological exploration in successful in terms of both survey and southern Bornu. The African excavation. We look forward to continued Archaeological Review 2: 153-71. work in these areas in years to come. MacEachern, S. Acknowledgments Du Kunde: Processes of Montagnard Research in Nigeria in 1993 would not Ethnogenesis in the Northern Mandnra have been possible without the work of a Mountains of Cameroon. PhD. thesis, large number of people. We would University of Calgary, Calgary. particularly like to thank the people who Selling the iron for their shackles: were involved in the actual fieldwork, and Wandala-montagnard interactions in who worked extremely hard and sometimes northern Cameroon. ]ournal of Afican in rather difficult conditions of survey and Histoy 33(2): 241-70. excavation: Musa Hambolu, Joseph Ameje, The Projet Maya-Wandala: preliminary Claire Bourges, Sylvia Abonyi, Judith results of the 1992 field season. Nymne Wheeler, and Francois Vigneault. Without Akumn 39: 7-13. their help, the project would not have been possible. The staff of the National No. 47 June 7994

Soper, R. absolute values in weight percent or ppm, 1965 The Stone Age in Northern Nigeria. thus preventing comparisons between experiments carried out at different times Journal of the Historical Society-- of Nigeria - 3 (2): 175-94. under different experimental conditions. This also meant that a database of results could not effectively be built up and added to over the years. These individual SOUTH experiments, however, provided valuable AFRICA insights into both intra- and interassemblage variability (Gihwala et al. 1984,1985af 1985b; Jacobson et al. 1991,1994; Peisach et al. 1990, Recent Archaeological XRF 1991; Pineda et al. 1990). XRF results, on the Research in Southern Africa other hand, are expressed in weight per cent of oxides for the major elements or ppm for trace elements and can therefore be used to L. Jacobson build up a database that can eventually be McGregor Museum accessed by other workers. In addition, P.O. Box316 duplicate powder samples of most analysed 8300 Kimberley, South Africa specimens are being stored at the McGregor Museum so that additional or replicate W. A. van der Westhuken & H. de Bruiyn analyses or interlaboratory comparisons can Geology Department be made in the future. In preparing and University of the Orange Free State analyzing samples, all work is carried out 9300 Bloemfontein, South Afnka after the method of Nomsh and Hutton (1969). International geological standards are used for calibration. Over the past three years a series of XRF Secondly, by using the whole sample analyses have been run on a variety of W0 g), XRF is not susceptible to a potential materials including pottery, clays, hornfels sampling problem to which PIXE can be (both artefacts and geological samples), and liable. The scanning of a small surface area archaeological cave and open site sediments can make PIXE sensitive to inhomogeneous from southern African sites. The aim of this elemental concentrations and produce research is to investigate archaeological skewed results. problems to which chemical analysis can be Thirdly, additional elements determined applied such that "visually invisible" by XRF can give insight into manufacturing aspects of human behaviour can be techniques, composition or postdepositional reconstructed. For pottery and artefacts this processes (Bollong et al. 1993). means provenance studies as well as, in the One potential disadvantage of using case of pottery, technological investigations. XRF compared to PEE or NAA is that the In addition, microprobe analyses across more esoteric types of trace elements are sherd bodies are being made to obtain currently not able to be determined by us. concentration profiles of phosphorus and There can thus be a loss of sensitivity when other elements in order to investigate any charaderising pots or clay sources in that post-depositional changes to the chemistry the limits of variability for major or more of sherds that could complicate the above common trace elements can be more provenance studies. restricted so that a greater overlap between sources occurs. This problem, however, still Methodology and Assumptions needs detailed analytical testing. Another disadvantage of XRF is that the Much of this is an extension of similar sample for analysis is destroyed by crushing work carried out by PEE analysis but with whereas PIXE can analyse a sample intact three advantages. Firstly, most of the early with little or no surface damage. Neverthe- PIXE work was expressed in raw counts, not less, using these techniques in a com- NYAME AKUMA No. 41 June 1994

plementary and common-sense way can Current Results yield valuable insights into past processes. Sample Homogeneity The investigation of archaeological problems rests upon the assumption that As an example of the chemical variabil- one can chemically "fingerprinf' or separate ity in one vessel, we provide data taken different clay sources and that this unique from the multiple analysis of a thick, pattern will be replicated in pots made from pointed base Khoi vessel from Namibia that those sources. This chemical fingerprint can shows a different set of results for the rim be altered from clay to pot or from pot to and the base (Table I). Whether this is the archaeological sherd by a number of factors. result of different manufacturing techniques These include the addition of temper to the in that the base may have included small raw clay either to make the clay more amounts of temper not added to other parts suitable for firing or else for a specific of the vessel or differential contamination functional purpose of the finished vessel. through useage is not known at present. The increase in Ca and Na could suggest the The fabric of the vessel can also be latter and although further work such as contaminated by use or by post depositional thin sectioning could provide an answer, it factors. It is also possible that only parts of does serve as a warning to keep a record of an individual vessel (either before or after it the type of sherd analysed (e.g., rim, body, gets broken and becomes part of the lug, base, thick, thin). archaeological record as a number of sherds) are affected by the aforementioned factors which further complicates matters. Table I. Data for the analysis of a Microprobe analyses currently underway vessel from the Erongo, could offer further insight into this problem. Namibia. Two rim samples and a single base sample were Models and Problems analysed. Major elements have been normalised to 100%. The ideal problem for investigation is one that deals with samples from two or more sites each from quite different geological substrates. Intrusive vessels will stand out quite clearly and the degree of interaction between such sites can be quantified. Samples coming from similar substrates could be more difficult to separate from each other. If one can locate clay beds, this is a bonus as it can indicate the range of variability of the clays available to local potters. Otherwise, one has to rely on compositional groupings defined by one or other multivariate statistical technique. More than one cluster from one site does not necessarily mean imports but rather the variability of local clays. Archaeologists are earnestly requested to obtain samples of local clays (from 100 g minimum to 1 kg if possible) preferably in consultation with local people if they still have, or recently had, a pottery making tradition. Oral traditions often retain knowledge of the location of suitable clay beds. This information should be documented for further research. NYAME AKUMA No. 47 June 7994

Table 2 shows data from a sherd split types of horizons as well as from areas longitudinally and the individual sections outside the site for future analysis. analysed. There is the hint that some elements could be enriched, i.e., K on the Provenance studies outer surface, but obviously much more work remains to be done. Nevertheless, it Our current pottery research has does indicate that where possible additional focussed on Namibia (western Darnaraland parameters relating to mode of burial and the Kavango/Caprivi), northern should be recorded such as whether Botswana, the northern and northeastern completely buried, proximity to hearths, Cape Province, and the northern Transvaal. partially exposed, lying with the outer or These include specimens from ceramic LSA, inner surface exposed, etc. In addition, Herder, and Iron Age contexts. Where sediment samples for chemical analysis possible, clays are also analysed. should also be taken from various places in Data is still being finalised but as an the deposit as well as outside. These should example we briefly describe initial results be carefully curated with excavated from a project (jointly undertaken with T. N. materials for future analysis. Huffman of Wits) investigating the social and economic interactions of K2 and Mapungubwe sites with each other and the Table 2. Data for sherd (A71 from the surrounding areas within and beyond their site Omdraai in the Northern influence (Jacobson et al. 1994). This will Cape that had been split also include Great Zimbabwe. Some of the longitudinally such that the inner (IN) and outer (OU) problems that will receive attention will be surfaces, the center (MI, and a whether imported vessels at major sites whole (TI sample were increase in number, perhaps through the separately analysed. Major payment of tribute; whethcr any special elements have been classes of vessels which are only normalised to 100%. occasionally found at lower court or ward AEIA7-INA7-OU levels originate from a higher level ward and whether any aaft specialisation took sQ2 6527 65.01 64.37 place, i.e., were all or some vessels only Ti9 1.16 1.15 1.19 made from a fewer or a greater number of A12Q 20.80 20.76 20.76 FqQ 8.91 9-07 9.30 clay sources? MnO 0.08 0.11 0.12 As a first step, a suite of potsherds was w3“ 1. 1.10 127 analysed in order to determine, firstly, GO 0.93 0.95 0.93 whether pottery from these sites could be m20 0.25 0.28 0.31 differentiated chemically and, secondly, K20 1.37 1.32 1.53 whether XRF was sensitive enough for this p2 9 0.17 OZ 023 purpose. Twenty-five samples of pottery Total 100.00 100.00 100.00 were obtained from four sites. Twelve HzO- 0.31 0 0.11 samples (coded "k") came from K2 and a LO1 5.77 4.57 5.04 single sample from Mapungubwe (coded "kl"). Of the former, two examples, coded separately as "t" were made in the Archaeological Sediments distinctive style of Tautswe, a major Iron Cave and open station sediments from Age site to the west in Botswana. Nine selected sites are also being analysed. This sherds came from Great Zimbabwe ("p") will be used initially as baseline data to and three from Chibvumani ("j"), also a assess the degree of contamination of Great Zimbabwe period site. potsherds, bone or ostrich eggshell by their Inspection of the data for P205 (Fig. 1) surrounding sediments (i.e., phosphate showed that systematically higher values enrichment). Archaeologists should keep were obtained for the samples from Great samples of the sediments from different Zimbabwe (coded p). As these were both NYAME AKUMA No. 41 June 1994

higher than the Chibvurnani sherds (coded j geological background, separate well from and which come from the same geological K2 but show a great deal of variability. background) and any naturally occurring Whether a greater variety of local clay P205 that we are aware of, and following sources were used or they reflect trade items Freestone et al. (19851, it was deaded to omit cannot be answered as yet. There is also the this element from the statistical analysis as it possibility that the burial conditions which most probably reflects the burial conditions produced the high P205 levels in the Great of the sherds and not the source material Zimbabwe sherds could have introduced from which they were derived. other elemental contamination. This should A correspondence analysis (Greenacre be further investigated through miaoprobe 1986) of the data was then undertaken analysis and the analysis of the (Fig. 2). The K2 samples formed a very tight archaeological sediments themselves. cluster with the exception of 5 samples that form their own groups. The first, consisting Conclusion of t7, t8, and k3 form a group that is essentially maintained when the third axis is Archaeologists are earnestly requested considered. Of interest here is the fact that t7 to follow some of the guidelines above with respect to recording information when and t8 are sherds from K2 that are stylistically similar to Tautswe in Botswana. excavating pottery and the collection of archaeological sediments and clays whilst in This is, of course, not to say that they are the field. from that area. Of the second group, kl and ko, kl is the single sample from To conclude, provenance studies can be Mapungubwe. As such, not too much most accurately summed up as the invisible should be read into this as yet until a larger dimension to stylistic studies. Carefully sample from this site is available. The j and used, it has vast potential for adding to our p samples, both apparently from the same knowledge of the past.

Phosphate content relative to Silica: N.TvI and Zimbabwe Pottery

1 Log P205

Fig. 1. Phosphate content of vessels. Codes: k = K2, p = Great Zimbabwe, t = style of Tautswe, [ = Great Zimbabwe, j = Chibvumani. NYAME AKUMA No. 41 June 1994

Fig. 2. Correspondence analysis map of Northern Transvaal and Zimbabwe pottery. The horizontal axis is dimension 1 with inertia equal to 58.9%; the vertical axis is dimension 2 with inertia equal to 23.5%. A total inertia of 82.4% is represented in the above map.

References Boulle, G. J., and Peisach, h4. These include a listing of all publications 1977 Characterisation of South West African relating to PIXE and XRF work in Southern potsherds by neutron activation Africa. (Editor: This creates problems in analysis of trace elements. Journal of matching text references with items in this list. Radioanalytical Chemisty 39: 33-44. Readers needing clarification should contact the 1979 Trace element analysis of senior author.) archaeological materials and the use of Bollong, C. A., Vogel, J. C., Jacobson, L., van der pattern recognition methods to Westhuizen, W. A., and Sampson, C. G. establish identity. Journal of 1993 Direct dating and identity of fibre Radioanalytical Chemisty 50: 205-15. temper in pre-contact Bushman (Basarwa) pottery. Journal of Boulle, G. J., Peisach, M., and Jacobson, L. Archaeological Science 20: 41-55. 1979 Archaeological significance of trace element analysis of South West African NYAME AKUMA No. 41 June 1994

potsherds. South Afn'can Journal of Jacobson, L, Gihwala, D., Peisach, M,Pia, Sch75: 215-17. C. A., Woods, A., and Otto, A. 1985 On the trail of the pottery pedlars. Freestone, L C., Meeks, N. D., and NuclePr Actitne 33: 25-8. Middleton, A. P. 1985 Retention of phosphate in buried Jacobson, L., Gihwala, D., Pineda, C. A., and ceramics: An electron microbeam Peisach, M. approach Adaeom&y 27: 161-77. 1983 Gold and glass. NuclePr Active 29: 24-7.

Gihwala, D., Jacobson, L, Peisach, M., and Jacobson, L., Huffman, T., and van der Pineda, C A. Westhuizen, W. A. 1984 Determining the origin of 18th and 19th 1994 A preliminary XRF analysis of pottery century pottery and glasses using PIXE from the Northern Transvaal, South and PIPPS. Nuclear Instruments and Africa. Poster presented to the 29th Methods in Physics Research B 3: 408-11. International Symposium on 1986 Proton-induced techniques for the Archaeometry, Ankara, Turkey, 9-14 determination of some trace impurities May 1994. in gold objects. South African Journal of Scisrce 82: 258-60. Jacobson, L, Loubser, J. H. N., Peisach, M., Pida, C A., and Gihwala, D., Jacobson, L, Peisach, M., Pineda, van der Westhuizen, W. C. A., and Vos, H. N. 1991 PEE analysis of pre-European pottery 1985a Analysis of Chinese porcelains and from the northern Transvaal and its ceramics. South African Archaeological relevance to the distribution of ceramic BuMn 40: 9649. styles, social interaction and change. South Afican Archaeological Bulletin 46. Gihwala, D., Jacobson, L, Peisach, M., Pia, 19-24. C. A., Woods, A., and Otto, A. 198% Preliminary report on the trace element Jacobson, L, Pineda, C. A., Peisach, M., analysis of clays and pottery from the Morris, D., and Pillay, A. E. Kavango and Caprivi areas of northern 1994a PIXE analysis of herder and hunter- Namibia. South Afican Jouml of Science gatherer pottery from the southern 81: 38-41. Kalahari, South Africa. Poster presented to the 29th International Greenaae, M J. Symposium on Archaeometry, Ankara, 1986 SIMCA. A program to perform simple Turkey, 9-14 May 1994. correspondence analysis. The American 1994b A preliminary report on the PIXE Statistician 40: 230-1. analysis of ostrich eggshell and its potential for provenance studies in Jacobson, L Southern Africa. Poster presented to 1985 X-ray analytical techniques and the 29th International Symposium on archaeological materials. South African Archaeometry, Ankara, Turkey, 9-14 Archaeological Bulletin 40.110. May 1994.

Jacobson, L Jacobson, L, Pineda, CA., Moms, D., and Peisach, M. 1989 Forward into the past. Nuclear Active 41: 21-4. 1994 PEE analysis of pottery from the northern Cape Province of South Jacobson, L, and Peisach, M. Africa. Nuclear Instnunents and Methods in Physics Research B 85: 901-3. 1982 Prehistory's clay footsteps. Nuclear Actim 27: 8-12 NYAME AKUMA No. 41 June 7994

Miller, D. Pia,C A., Jacobson, L., Sampson, C. G., and 1991 Materials analysis of archaeological Peisach, M. ceramics in southern Africa. South 1990 PIXE analysis of Seacow River Valley Afiicnn A~haeologicalBulletin 46: 12-18. pottery. National Accelerator Centre Annual Report 118-9. Norrish, K,and Hutton, J. T. 1969 An accurate x-ray spectrographic Pineda, C. A., Peisach, M., and Jacobson, L. method for the analysis of a wide range 1988 Ion beam analysis for the of geological samples. Gaochimica determination of cation ratios as a Cosmochimica Acta 33: 43153. means of dating southern African rock varnishes. Nuclear Instruments and Peisach, M., Jacobson, L, Boulle, G. J., Methods in Physics Research B 35: 463-6. Gihwala, D., and Underhill, L G. 1989 The time-clock of aged patinas. Nuclear 198% PIXE analysis of archaeological Actiw 41: 17-20. material and the use of multivariate analysis for characterisation Nuclear Pineda, C. A., Peisach, M., Jacobson, L., and Instnmrents and Methods 193: 33741. Sampson, C. G. 1982b Multivariate analysis of trace elements 1990 Cation-ratio diierences in rock patinas determined in archaeological materials on hornfels and chalcedony using thick and its use for characterisation Jouml target PIXE. Nuclear instruments and of RadioanalppcalChemistry 69: 349-64. Methods in Physics Research B 49: 332-5.

Peisach, M., Pineda, C. A., and Jacobson, L 1990 Thick target PD(E analysis of coastal and inland Namibian pottery. Nuclear SUDAN Instruments and Methods in Physics . Research B 49: 309-12. 1991 Nuclear analytical study of rock Archaeological Mission of paintings. Journal of Radioanalytical and the University of Geneva NUClplr Chemishy, Articles 151: 221-7. at Kerma (Sudan): Report on Peisach, M., Pineda, C.A., Jacobson, L., and the 1992-1993 Campaign Loubser, J. H. N. 1991 Analytical study of pottery from Charles Bonnet Soutpansberg. Journal of Radioanalytical Geneva University Excavations in the and Nuclear Ckemistry, Articles 151: Sudan 229-37. 17 Chemin Bornalet 1242 Satigny Pineda, C. A., and Jacobson, L Switzerland 1992 PEE analysis of pottery from the western Cape coastal belt. National Accelerator Centre Annual Report Once again, the excavations led by the NAC/AR/92-01: 36. Swiss Mission at Kerma from December 12, 1992 to February 4,1993, took place under Pineda, C. A., Jacobson, L., Morris, D., and favourable conditions. More than 120 Peisach, M. workmen whose work was directed by the 1993 PIXE analysis of ancient pottery from rais from Tabo, Gad Abdallah, and Saleh the northern Cape, South Africa. Melieh, attended the working sites. The National Accelerator Centre Annual restoration of certain monuments, and more Report NAC/AR/93-01: 38-9. particularly that of the warehouses of the large roundhouse of the palace, occupied a NYAME AKUMA No. 47 June 7994

group of twenty builders and their 29 m square. At least three dwellings were assistants. Many other specialists built a erected in the housing perimeter. Round 300 m long surrounding wall so as to protect silos enabled us to situate the enclosed land the eastern side of the site where the wire to the north destined for food storage. As for fence had disappeared. the long courtyard, the southern entrance of A considerable effort was made towards which was protected by a low winding wall, excavating the southwestern part of the one may presume that it had a religious ancient city. Outside the protected site, the function. Traces of big pillars could belong spread of population and traffic threatened to the first outlay of a chapel. the archaeological remains. Thus, after During the Classical Kerma Period having come upon the remains of the palace (1750-1500 KC), the whole dwelling area during the previous campaign, the was reshaped, disregarding the older foundations of many peripheral buildings developments; the chapel was enlarged to a were brought to light. A true secondary square shape. The neighbouring courtyards built-up area appeared; therefore works on with their silos were still in use but stronger this area will have to go on for some time. walls marked the bounds of the dwelling Nearer the royal residence, a large dwelling unit; two blocks of buildings were added. completes the image of the city with its Other houses in the same area or further western quarter. This urban area is limited to the west are known to us. The collapse of to the south by fortification ditches that the pilaster-supported side walls enabled us could be traced on a length of over 140 m. to check the original height of the buildings, The study of the fortified western entrance which were at least 3 m high. resumed together with a stratigraphic Further to the south, the dried-up sounding near the deffufa. A specific ditches that surrounded the city gave us a investigation was undertaken in the eastern first piece of intelligence regarding the necropolis where several tombs were defence works. As a consequence of the superficially cleaned so as to study funerary dumping of debris outside the city walls, rites. these ditches filled up and made a new space. After a while, the slope of the ditch The Ancient City was reinforced with a layer of hardened The city grew progressively; thus one mud. The close study of these layers brought sometimes comes upon the layout of old out some of the layouts. On a portion of the ditches under more recent houses. The newly gained space, three minor buildings western fortified entrance was razed to the undoubtedly point to the location of the ground when the city expanded. Repeated casemates. cleaning made the understanding of its organisation possible. The access route ran The Secondary Suburban Settlement pe-FdiculG to the general east-west axis, The boundary of the ancient city of the between two enormous round bastions. southwest seems to skirt an area near the Those massive mounds, several times centre of which there is a mound, refaced, were crowned with rectangular or understood to be a tumulus. As a matter of originally circular towers. On the inside, fact, a few soundings proved that a religious facing the entrance, an impressive bastion neighbourhood had developed on this site. blocked the entrance once more. So far, about ten cult buildings have been Near the gate, a large dwelling Ru1122) traced, lined up along both sides of the was unearthed. Before it were two houses street. Their outlay varies from a simple (M126-M127), the orientation of which square room to complex units made of a corresponds to quite a different axis from sanctuary and two side annexes, preceded the neighbouring area. This spread towards by a peristyle courtyard. Former the west is associated with the Middle construction phases often point to the Kerma Period (2050-1750 KC.). House 122 is permanence of a place of cult that was identified with several plots of land within a generally enlarged. NYAME AKUMA No. 41 June 1994

The walls of the Holy of Holies were The Eastern Necropolis carefully built and may be quite thick (up to The thorough cleaning of several tumuli 0.70 m). On the other hand, the courtyard enabled us to understand how the buaanes walls had shallow foundations and must were placed to the south of the tombs. Close have been lower. They had wooden to these were some jars that were used for supports, two carbonized posts acknow- libations. Two ceramic plates placed on the ledged as doum palm tree having been well tumulus seem to represent a snake game; preserved underground. These wooden the model of a bird was also noted. column bases showed an unusual feature; they were circular structures made of brick The overall shape of the tumulus, with and stone clustered together with mud. its black stone ring, may be completed by an While these buildings seem to have been almost flat cover of white quartzite pebbles. destroyed by fire, the traces of many Only a burial place was excavated; it had fireplaces must also be linked with functions been totally plundered. or activities that took place within or around the monuments. Archaeological Material Little archaeological material has been An investigation of the ditches is recognized in the layers of destruction (the focussed on the search for seal impressions. only ones that have been excavated up to Some twenty-or-so of these objects were now); three fragmentary jars were still in found and will complement other findings situ, and some dozens of sherds were also fo the kind. They demonstrate the close links recovered. These elements mostly belong to between Kerma and the Egyptian the Middle Kerma Period, though some are administration of the fortresses of the of the Classical Kerma Period. Second Cataract. A building complex borders the Large fragments of an ostrich shell religious structures to the west and to the skilfully engraved show an interesting south; the buildings may have been iconography. Giraffes, a crocodile, and a designed as workshops, places of storage, or bovine-type animal belong to the Nubian sometimes even dwellings. Three polished traditional bestiary. A character placed stone axes were found in one of the units under a geometric garland seems to stand consisting of two rooms side by side. out as the centre of the composition, A general outlook of this whereas other human figures give a precise neighbourhood is still unavailable but an and detailed account of their clothing. The entry is indicated by rounded walls at the prow of a ship is also represented. far end of the main street, and a large elongated structure closing the area to the east defining a 70 m square for the central part. Discovery of Middle Stone A definite interpretation of functions of Age Sites near El Ghaddar this religious neighbourhood would of (Sudan) course be premature. It seems obvious that an institution based on an Egyptian model ruled its activities and that providing Michal Kobusiewicl offerings for the chapels kept a great Institute of Archeology and Ethnology number of people busy. This may have Polish Academy of Sciences involved a Nubian adaptation of the Zwierzyniecka 20 Egyptian worship dedicated to the memory 60-8 14 Poznan, Poland of deceased kings or governors, in some way a funerary complex close to the city of the lviing. A necropolis which remains to be During the first three days of December studied is included in the ruins of this 1993 the author, accompanied by religious complex at the end of the Classical J. Kabacinski, and invited by Dr. Krzysztof Kerma Period. NYAME AKUMA No. 4 7 June 7994

Grzymski from the Royal Ontario Museum a ridge running toward the SE. It consists of (Toronto), carried out an archaeological a broad spread of Levallois cores, flakes, and reconnaissance within the area of the dbbitage, accompanied by a chopping tool Canadian concession (Grzymski 1987) near (Fig. 3), but also includes some regular El Ghaddar, Sudan. Besides some late single-platform cores and a side scraper on prehistoric sites in the Letti Plain, several an elongated flake. All artefacts are made of interesting Middle Palaeolithic sites were ferriaete sandstone. discovered in the vicinity of El Ghaddar Apart from these sites, other similar village near Old Dongola. In the course of sites were discovered on most of the our fieldwork three sites were examined in sandstone jebels in this area. The artefact some detail. concentrations found on the flat tops of the inselbergs seem to be pretty much in situ. Jebel Sheik Wahab This is indicated by articulating pieces very close to each other, as well as by the good The site is located on the eastern edge of preservation of stone wind-shelters, which the top of the jebel (inselberg). Thousands of are probably connected to the Middle stone artefacts occur in a concentration Palaeolithic settlement. about 20 m in diameter. The assemblage is composed of Levallois cores in various It is impossible to estimate the exact age stages of exploitation, Levallois flakes, some of the concentrations, but typological and of which are partially retouched and a large technological criteria point strongly to the amount of dbbitage. All of these artefacts are Middle Stone Age. Some concentrations made of ferricrete sandstone, of which the composed of rolled and worn artefacts with top of the jebel is composed. A number of angular cores seem to be older than the such artefacts also cover the southern slope relatively fresh, purely Levallois of the jebel. assemblages. The Middle Palaeolithic MSA sites occur Jebel Kobkaba exclusively in very close proximity to outcrops of ferricrete sandstone. It seems This is a medium sized jebel situated that the sites represent a concern for good ca 800 m SE from Jebel Sheik Wahab, ca raw material to produce tools. 250 m E from site ROM 13/5 (a house The concentration of Middle Stone Age construction). A large concentration of reported here is the latest discovery of its artefacts made in the Levallois technique kind after those of similar sites from the occur on the top of the jebel, as well as on its vicinity of the Second Cataract slopes. In the western part of the (Clunielewski 1968, Guichard and Guichard concentration, these are typical Levallois 1968, Marks 19681, and the sites from the cores, flakes (Fig. 1,2) and points, (some of Dongola Reach near Debba on the west bank them very large), as well as abundant of the Nile (Marks et al. 196&, Marks et al. dbbitage and angular cores. Here the 196813). artefacts are very worn and rolled. In the eastern part the artefacts are clearly more fresh. Also in the eastern area the remnants of wall-like constructions were discovered, built of cobbles and slabs and still preserved to a height of 20-30 an. They form half circles or even ovals or circles ca 15-2 m in diameter. They may be the remains of shelters or windbreakers.

Jebel Ghaddar The site is located about 100 m E from the foot of Jebel Ghaddar at the beginning of NYAME AKUMA No. 41 June 1994

Fig. 1. O- 2 ern

Fig. 3. NYAME AKUMA No. 41 June 1994

References (CeRDO) in the Nubian Desert of the Sudan between latitudes 20 and 22 and longitudes Clunielewski, W. 32 and 36. Over 180 sites have so far been 1968 Early and Middle Paleolithic Sites near recorded, but many more are needed to Arkin, Sudan. In F. Wendorf (ed.), complete the chronological sequence of the Nubia 1: 11047. area's Holocene history. Among the sites discovered to date, there are Mesolithic Gnymski, K camps, possibly Neolithic graves, an A- 1987 Archaeological Reconnaissance in Group or predynastic related burial Upper Nubia. The SSEA Publications, tradition of the fifth millennium B .C., V. 16: 1-58 (Benben Publications, scattered evidence for a fourth millennium Toronto). KC. burial tradition, dynastic hieroglyphic inscriptions, as well as some Ptolemaic and Guichard, J. and Guichard, G. Napatan/Meroitic remains. More recent 1968 Contributions to the study of the Early remains consist of what appears to be a and Middle Paleotlithic of Nubia. In Blemmyean (local Beja) burial tradition of F. Wendorf (ed.), Nubin 1: 148-93. the first few centuries A.D., and a host of Medieval Islamic remains foremost among Marks, A. E. which are many mining villages and cemeteries. Various other rock engravings 1968 The Mousterian industries of Nubia. In and some forts and mines from this century F. Wendorf (ed.), Nubia 1: 194-314. complete the catalogue of finds. Marks, A. E., Hays, T. R, and Heinzelin, J. The jewel of Nubian Desert archaeology is the site of Deraheib, first reported in print 1968a Preliminary report of the Southern by Mohammed Ali Pasha's prospector Methodist University Expedition in the during the nineteenth century (Linant de Dongola Reack Kush 15: 165-92 Bellefonds 1868). This vast village of stone walled apartment style houses, avenues, and Marks, A. E., Shiner, J. L, and Hays, T. R castles was almost certainly the city of 196813 Survey and excavations in the Dongola Allaqi, the hub of Medieval Arab occupation Reach, Sudan. Current Anthropology 9: in the Nubian Desert. It was here during the ninth century that the rebel Rabia clan declared their independence from the caliphate in Baghdad. Arab writers describe it as a wealthy market town, a meeting place for all who seek gold in the desert, and a Preliminary Results of way station for pilgrims to Mecca via the CeRDO's Research in the Red Sea port of Aidhab (cf. various sources Nubian Desert in Vantini 1975). Before the Arabs arrived, Deraheib was Karim Sadr probably the seat of the Beja King 01 Bab. A Alfredo Castigilioni structure resembling a church (Manlio Angelo Castiglioni Sozzani, personal communication) suggests Centro Ricerche sul Deserto Orientale a Christian period occupation. Before that, a Via Gradisca 22 coin suggests that the site was inhabited 2 1 100 Varese during the Ptolemaic period. Archival Italy research by Giancarlo Negro suggests that at that time Deraheib may have been Berenice Panchrisia, the "all golden" outpost of the Hellenic pharaohs. History at this site may Since 1989 four seasons of survey and prove to extend even farther back in time. test excavations have been undertaken by Similarities in the site's layout to that shown the Centro Ricerche sul Deserto Orientale on the Turin papyrus suggest that Deraheib NYAME AKUlMA No. 41 June 1994

may have been the gold mine of Seti I of the One of these types is a simple conical New Kingdom period. Planned excavations tumulus with a diameter of about 3-5 m. at this fascinating site will, we hope, reveal Two examples were excavated, one of which these layers and much more. contained three beads similar to examples Beyond Deraheib there are hundreds of found in a Neolithic context in the Kerma other mining villages large and small. Basin (Charles Bonnet, personnal Although almost all of these contain at least communication). Both these graves showed some Medieval Aswani pottery, most were a burial tradition of secondary inhumations; probably established before the arrival of the that is to say the bones were already Arabs. Ancient Egyptian records tell us that disarticulated prior to burial. Secondary mining in the Nubian Desert began at least burials such as these are uncommon in the as early as the Middle Kingdom period Nile Valley, and the large numbers of this (Save-Soderbergh 1941). A gold bracelet type of tumulus suggests they were built by found in a fifth millennium B.C. grave, an indigenous population. however, suggests that gold may have been The other common grave type, and by extracted here even in predynastic times. far the most spectacular of the tumuli in the Partially destroyed by looters, this grave Nubian Desert, is a circular stone platform, contained aside from the golden bracelet, with some examples reaching 15 m in fine burnished ceramics, and a pendant diameter. Five of these tumuli were similar to predynastic wares (Jean excavated, and two were radiocarbon dated Vercoutter, personal communication). to the seventh and eighth centuries A.D. If Within the tumulus there was also an the dates are correct, these must be Beja offering area marked by small stelae and graves of the immediately pre-Islamic two animal horn cores (sheep and cattle; period. Two types of burial chambers-a Louis Chaix, personal communication), as domed vault and a slab covered chamber- well as masses of charcoal. Submitted to the are associated with this style of CSIR laboratory in Pretoria, the carbon superstructure. In both, the body is lain on dated to the mid-fifth millennium KC the right side in a flexed position. In two of Excavation of the two other known tumuli the graves the body had been lain on a of this type in coming seasons should leather blanket. Other burial goods include provide more information on this exciting the occasional bead, pendant, or other such new discovery. decorative items. Ceramic vessels were Among the other graves excavated, a apparently left as offerings outside the late-fourth millennium B .C. grave also tumuli. included cattle bones but was otherwise too Besides the mines and the graves, thoroughly looted to provide information surface scatters of artefacts-remains of about the tradition involved. The burial campsites-have also been found. Scattered structure consisted of a 25 m stone circle ceramics can be found all over the desert, with the looted burial chamber on the evidence of continuous, though light traffic eastern rim of the circle. In the center of the throughout the last six or seven millennia. circle a large fireplace was surrounded by Some artefact scatters were denser and more scattered bones of sheep and cattle. The clearly definable as sites. Among these were radiocarbon date was obtained from this a group of sites with Khartoum Horizon fireplace. style sherds, among which many were Architecturally, both this and the fifth identified as Mesolithic by Isabella Caneva. millennium B.C. tumulus are rare types, The richest of these sites are found around a suggesting that the graves may have vast depression which in the early Holocene belonged to outsiders. Two other types of may have fonned a playa. tumuli were excavated, both of which are Among the many rock engravings, those common enough to have been graves of showing cattle presumably date to the third locals. millennium B.C. and earlier since the climate after that would have been too dry for these beasts (Muzzolini 1982). A series of rock art NYAME AKUMA No. 41 June 1994

showing crude stick figures in association de Prehistoire, et d'Ethnologie des pays de with camels and horses, often shown in la Mediterranec Occidentale 2: 1-38. combat most likely date from Roman or Medieval times. Inscriptions abound in the Save-Soderbergh, T. Nubian Desert as well. Hieroglyphics found 1941 Agypten und Nubien: ein Beitrag zur at various localities name prospectors who Geschichte Altagyptischer Aussen-politik. traversed the desert (Damiano-Appia 1992), Lund Hake Ohlssons Boktryckeri while later Arabic texts and some writings in Coptic appear to be from military Vantini, G. expeditions in recent centuries. 1975 Oriental Sources concerning- Nubia. CeRDO's four seasons of research have Heidelberg and Warsaw. only scratched the surface of a complex historical sequence in a vast area. Much remains to be done and many of the sites deserve intensive excavations. Even so, what W TANZANIA has been unearthed so far has amply repaid the herculean efforts of mounting these difficult expeditions. The Nubian Desert is no longer the archaeological blank it used to Archaeological Research in be, and with luck it won't be long before we Karagwe District know enough about the desert's prehistory to contribute to a better understanding of the past in the Nile Basin as a whole. Andrew Reid Archaeology Unit University of Botswana Acknowledgments Private Bag 0022 The 1993 campaign of the Centro Gaborone Ricerche sul Deserto Orientale was Botswana sponsored by Sector Sport Watches, Italiana Petroli (IP), Iveco, Pirelli, and Banca J. E. K. Njau Briantea. The radiocarbon dates were National Museums of Tanzania processed by Dr. John Vogel at the CSIR P. 0.Box 51 1 laboratory, Pretoria. Dar es Salaam Tanzania References

In 1993 a programme of archaeological 1992 Inscriptions along the tracks from research was completed in Karagwe District, Kubban, Buhen, and Kumma to in the northwestern comer of Tanzania. The "Berenice Panchrysos" and to the lack of work in Karagwe contrasts with South. Paper presented at the neighbouring areas-Burundi, Rwanda, International Meroitic Conference, Uganda and Buhaya-where much has been Berlin. done, in certain places for more than fifty years. As a consequence, it was of Linant de Bellefonds, L considerable importance to begin to 1868 L'Etbaye. Pays Habite par 2es Arabs establish the cultural sequence for the Bicharieh. Gwgmphie, Ethnologic, Mines district: both to further academic interests d'Or. Paris. and to make a contribution to Cultural Resource Management in what continues to be an archaeologically poorly documented 1982 Les climates sahariens durant country. l'Holocene et la fin du Pleistocene. First an archaeological survey was Travaux du Laboratoire d'Anthropologie, camed out so as to commence the process of No. 41 June 1994

archaeological documentstion of the district. Headquarters. On much doser inspection, In the short time available it was impossible however, it was found that the to hope to cover the entirety of Karagwe (an archaeological material was isolated and area of over 7200 km2)or even to investigate was located in high erosional contexts. a statistically significant sample of the Artefacts were thus dispersed and very district. Furthermore, the district ranges in badly abraded. altitude from the floodplain of the Kagera at Subsequent investigations spread out 1200 m to the highest parts of the Karagwe over the Kayanga hilltop to attempt to look escarpment at 1700 m and ecozones range for primary contexts for these artefacts. from papyrus swamp through thick valley Despite being able to examine several forest to hilltop grassland. Seven locations recently cut sections, particularly new pit were chosen for survey: Kayanga, latrines, no primary contexts could be Rwarnbaizi, Nyabiyonza, Kaisho-Isingiro, located on the hilltop. On the less-exposed Murongo, Bugara, and Kitengule (Fig. 1). western flanks of the hill more ESA artefacts These covered a wide range of topography, were encountered. These were less abraded but by no means should be considered to than those from the eastern flank, being have covered the district's entire range of located in soil rather than gravel deposits. ecological diversity. However, as a consequence of being on A total of 84 sites or locations of agricultural land, many had suffered the archaeological material were encountered by ravages of the hoe and were badly fractured. this survey, ranging from ESA bifaces to Despite this disappointing distribution, iron from the nineteenth-century kingdom surface collections of material were made so of Karagwe hot to mention an abandoned as to begin the process of aeating an ESA tin-mining complex from the 1930s near assemblage for the district. The material Murongo with most of the iron machinery collected from Kayanga included large still in sib). Details of sites, including their bifaces. Apart from these large cutting tools, six figure map coordinates, are contained the heavy duty tools, such as picks, and within a report presented to the Department diminutive bifaces that were encountered of Antiquities, Dar es Salaam. suggest that most of the early lithic material encountered was from the late Acheulian Early Stone Age early Middle Stone Age interface. The number of such tools encountered is, Occurrences of Early Stone Age material however, small. were confined to isolated encounters, in exposed, frequently erosional deposits. All of this material was encountered on the Middle Stone Age eastern escarpment of Karagwe. The only Three sites characterised by MSA exception to this was a single large biace material were located. All three of these are seen embedded in a sediment cliff of the situated in lowland areas. one on the Kagera Kagera River, west of Murongo. Particular river itself and two underneath Rwambaizi concentration of material was recognised at and the eastern escarpment. All of these sites and around Kayanga, but isolated artefacts featured large quantities of material which were also encountered at Rwambaizi, were observed to be eroding out of Igurwa, and Bwera. Tool forms encountered sedimentary deposits. Collection was carried included bifaces, picks, core scrapers, and out at one of these sites, a low hill situated large flakes. Diminutive bifaces were near Kibwera, some 5 km northeast of recovered at Bwera and at Kibwera. Rwambaizi. Preliminary analysis of this Initial encounters with ESA material at material suggests long-term utilisation of Kayanga suggested a location both rich in this location. The assemblage included small artefacts and under threat from the activities bifaces, which may be considered of the local population producing building transitional between the ESA and MSA, rubble for a presently burgeoning characteristic MSA diminutive bifaces, construction industry at the District points, and core saapers, and tools with NYAME AKUMA No. 41 June 1994

10 km 1 O- roids district boundary

Fig. 1. Karagwe District. NYAME AKUMA No. 41 June 1994

very fine retouch, suggesting a transition system. During the wet season the plains are with the LSA. (There was also a low density prone to flooding and animal populations of fully LSA material present at this site.) disperse: human settlement during these The remaining site, located in sediments seasons is likely to have- been equally of the Kagera river, is consistent with the dispersed and as a consequence very regular use of the Kagera throughout the difficult to locate archaeologically. Therefore human career, documented on the north by examining the Mwisa sites and the bank in Uganda. As well as these three Bugara sites we can compare the concentrations of MSA material, a small assemblages used in different ecosystems, percentage of the material encountered at by broadly similar populations, at different predominantly LSA sites, discussed below, times of the year. could be categorised as MSA. Their location Work was concentrated at Byenturege. along the lower Mwisa river, as it Here there are large erosion scars, up to approaches the Kagera, and on the Kagera 50 m in length, on the lip of the river terrace, itself is consistent with floodplain which contain large quantities of lithic exploitation. material. Rather than use some arbitrary grid system to investigate these scars, each Late Stone Age scar itself was identified as a unit and therefore collections were kept separate for Artefacts of the LSA were also largely each unit. A total of 13 exposures were confined to lowland floodplain areas. Single examined in this way, covering an area artefacts were occasionally encountered on along the river of around 1.5 km. A smaller hills or ridges overlooking these lowlands collection was made at one location on the but never further into the Karagwe Kagera river below Bugara. On an Highlands. Particular concentrations of sites impressionistic level these LSA assemblages were found along the Kagera river below largely consisted of core tools (e.g., blade Bugara and along the Mwisa river south of cores, bipolar cores, bipolar blade cores, and Kitengule ranch, at a location called platform cores). There was also a low level Byenturege. Indeed in this latter area the of MSA material on these sites. surface scatters of material in places appear A notable absence in the survey to be continuous for more than a kilometre. conducted was the lack of evidence for There are also several LSA sites around Kansyore Ware. Sherds from this pottery Kibwera, below the eastern escarpment. tradition, in association with LSA artefacts, The main concentration of LSA sites is were encountered at Kansyore Island on the located on terraces above permanent river Ugandan side of the Kagera (Chapman courses. Further work at these sites used as a 1967). It would have been predicted that base Nelson's comprehensive analysis of the sites would have been encountered on the LSA assemblages from Nsongezi rock- southern side of the Kagera around shelter (Nelson and Posnansky 1970). In Murongo, Bugara, and Kitengule. addition to the detailed quantitative analysis of the assemblage Nelson was able to recognise differences in the assemblages Early Iron Age from different stratigraphic units. Having A still more significant absence in the recognised extensive surface scatters of cultural record was the lack of evidence for material on the Mwisa river at Byenturege the Early Iron Age. Not one single sherd of and on the Kagera below Bugara, it was Urewe ware was encountered in Karagwe. decided to make extensive collections of This contrasts with the well-known locations surface material so as to conduct a of early ironworking in Buhaya, Rwanda, comparative study with the Nsongezi and Burundi, and specifically at Kansyore assemblages. The assumption upon which Island and Nsongezi rock-shelter on the this work relied was that Nsongezi, the north side of the Kagera river. This latter Mwisa sites, and the Bugara sites were all occurrence of EIA material is intriguing part of a single or related exploitation given its apparent absence in Karagwe. It NYAME AKUMA No. 41 June 1994

illustrates that the earliest ironworkers were kingdom was renowned for its cattle and familiar with at least the margins of evidence for pastoralism had been readily Karagwe, but that they did not consider it to found 100 km or so further north in be worthy of settlement, Karagwe being Mawogola, southern Uganda (Nyame much dryer than neighbouring areas. Akuma 33). The bulk of the LIA material, the Later Iron Age ironworking, is noticeably concentrated As might have been expected the b& of along the top of the eastern escarpment of the archaeological material encountered can Karagwe. A few very small sites were be dated to the last one thousand years. Less encountered on the western side of predictable was the fact that most of this Karagwe, around Kaisho, several small material consisted of evidence for iron ironworking sites were encountered at the smelting. The Karagwe Kingdom, which foot of the eastern escarpment and appeared in the last few centuries, is known occasional slag and roulette decorated to have been characterised by a combination pottery was observed along the Kagera river of agriculture and specialised pastoralism. at Bugara and Kitengule. The lack of archaeological evidence for these Typically the sites on the eastern other activities may be explained in several escarpment consist of several piles of large ways. Those areas which were used for blocks of slag, each block measuring up to agriculture in the past are still likely to be 80 an in maximum dimension. Less cultivated today. Since roulette decorated frequently they also include tuyere pottery has continued in use well into this fragments and very occasionally burned century, it is very difficult to distinguish mud fragments. Not all sites are associated recent domestic debris lying on the surface with roulette decorated pottery, but since from older archaeological material. One none are associated with Early Iron Age indication of past occupation, however, may pottery it is assumed that they are all of later be the number of quernstones encountered date. It was noted with some concern that in within modern fields. There is some areas of high building activity, Kayanga in evidence to suggest that banana agriculture, particular, builders are actively collecting which is today predominant, is only a recent slag blocks and incorporating them into phenomenon in Karagwe appearing in the their new structures. last 100 years and replacing grain (sorghum Follow-up work sought to gain an and finger millet) agriculture. The presence understanding of the ironworking of significant numbers of quernstones may conducted in Karagwe, both in terms of its therefore indicate the older grain based technological characteristics and the way in agriculture. which the production of iron was organised The roulette-decorated pottery that was within the state. These questions were encountered is unexceptional. It is either recognised as being of importance in Knotted Strip roulette (KPR) or Twisted helping to understand the concentration of String roulette (TGR: after Soper 1985). ironworking sites at certain locations along There are no carved wooden roulettes and the eastern escarpment. There was also a no rolled over rims, which are so need to explain the unusually large size of characteristic of %go Ware," from southern slag blocks encountered at these sites. Uganda. This may suggest that the Karagwe Accordingly work was based at Igurwa, a area was merely a southern fringe of the location with numerous iron-smelting sites. roulette decorated world. It is also possible, A series of interviews were conducted with on the basis of the generally poor a number of informants by Rachel Insoll. As application of the roulette, that the pottery a result of these interviews several old men was made in recent times. In Uganda poorly agreed to reenact a smelt. Subsequently applied roulette is associated with recent preparations were made and the smelt was manufacture. As for pastoralism, no attempted. evidence whatsoever could be found. This Unfortunately no iron was produced was especially surprising since the Karagwe and the sole product of the furnace was a NYAME AKUMA No. 41 June 7994

form of modified ore, in the process of cultural resources have yet to be being, but not yet, slagged. As a result of documented. this failure to produce iron, there followed a highly productive debate amongst the old Acknowledgments men about why they had failed. This debate helped to clarify many questions concerning This project was funded by the link technology and the social relations of agreement between the University of Dar es production. Salaam and the University of Bergen. Assistance was also provided by the British Institute in Eastern Africa. Thanks are due to Discussion Mr. Mugangala, Mr. Kaitaba, The results of the survey work are Mr. Lwamtoga, and the people of Karagwe significant in several ways. Firstly they for their friendship and cooperation and to achieve the very basic goal of outlining the Ms Kiyembe of the Department of cultural sequence within Karagwe. A body Antiquities. of some 84 sites is therefore added to the register of known archaeological sites of References Tanzania. Secondly there are marked patterns within the distribution that invite Chapman, S. questions and further investigation. Why are 1%7 Kantsyore Island. Arania 2: 165-91. ESA and LIA phenomena largely confined to the Karagwe highlands, whilst MSA and Nelson C, and Posnansky, h4. LSA occurrences are found only in the 1970 The stone tools from the reexcavation lowland floodplains? The answers to such of Nsongezi Rock Shelter. Azania 5: questions are no doubt to be found partly in 119-72. geomorphological, environmental, and cultural locational factors, all of which may Soper, R be focussed on by future work. Thirdly the total absence of both Kansyore and Early 1985 Roulette decoration on African pottery: Iron Age sites runs counter to the predicted technical considerations, dating and cultural sequence for the district and distributions. The AFjcun Archaeological highlights the importance of actual RCDiRU 3: 3-31. fieldwork evidence. The LSA material and the ironworking in particular indicate that Karagwe District has great archaeological potential for future research. The work shows that through the adoption of relevant strategies for fieldwork, researchers can both make a direct contribution to the archaeological resource base of the country, whilst at the same time developing specific research interests. The survey work also demonstrated that it is hazardous to merely predict the archaeological record of such a large area on the basis of knowledge of neighbouring areas. Rather, actual archaeological studies are needed to document this record. This has important implications for both archaeologists keen to summarise large uninvestigated territories and for those responsible for creating cultural resource management strategies in cases where NYAME AKUMA No. 47 June 7994

African Contemporary Art Exhibition

Beginning January 1996 African Anthropological Archaeology Position The International Centre for Bantu Civilkations

Bowdoin College seeks an assistant professor for a new tenure-track position, The International Centre for Bantu beginning fall, 1995. The position requires Civilizations (CICIBA), in cooperation with an anthropological archaeologist the University of South Florida, is preparing specializing in the later archaeology of a travelling exhibition of African Africa. We prefer candidates with Ph.D. in contemporary art. It will open in Tampa, hand but will consider candidates very close Florida, in January 1996, and then travel to to completing their dissertations. The person selected venues in the Unites States, Canada, holding this position will teach four courses and Europe on a two- to three-year tour per year on a rotating basis, including schedule. Introduction to World Prehistory, Essentials of Archaeology, and Peoples and Cultures of The exhibition will present art work Africa, as well as courses in the individual's from seven painters and sculptors from the CICIBA member states. It will thus be area of specialization, such as the rise of six complex societies, the Iron Age, or possible to exhibit from to ten pieces per ethnoarchaeology. artist to the public. The overall theme of the exhibition is "African Artists in the Bantu- Please send a letter of application, c.v., Speaking Area: Traditons in Contemporary and the names and addresses of three African Art," with three subthemes: references to Susan E. Bell, Department of "Retaining Traditions," 'me African Female Sociology and Anthropology, Bowdoin Iconography," and "Social and Political College, Brunswick, Me 04011, U.S.A. Commentaries." Priority will be given to applications received by November 1, 1994. Bowdoin A color catalog will be produced College is committed to equal opportunity covering all the pieces exhibited. The seven through affirmative action. Women and artists will be invited to the Unites States for minority candidates are especially the opening ceremony and will participate encouraged to apply and identify in workshops, teaching, and interviews. themselves as such. For more information, contact Bernard Clist, "USA Exhibition," Departement d'Arch6ologie et de Mudologie, CICIBA, B. P. 770, Libreville, Gabon (Fax 7034.49). NYAME AKUMA No. 41 June 7994

who accepted many new members of various and mixed backgrounds and who connected closely with the European colonial centers. Some of the Kommando groups, particularly the Griquas in Morris's study area, formed relatively stable Book Review communities. Morris begins his study with the Alan G. Morris, The Skeletons of Contact: expectation that archaeologically derived A study of Protohistoric Burials from the human remains will reflect the complex Lower Orange River Valley, South Africa situation of the Orange River frontier. He analyzes four burial populations. Kit W. Wesler The Riet River group was largely Wickliffe Mounds Research Center recovered by amateur excavators in the P. 0.Box 155 1920s and 1930s. Uncorrected 14C dates on Murray State University these materials range from A.D. 1060 + 50 to Wickliffe, Kentucky 42087 A.D. 1570 + 50. Some of the burials are fairly U.S.A. well associated with Type-R settlements, characterized by round stone enclosures and a hunting economy, possibly adopting some pastoralism. The dates, and the general Alan G. Morris, The Skeletons of Contact: geographic range, of the Type-R settlements A study of Protohistoric Burinls from the Lower and the Riet River burial sample suggest a Orange River Valley, South Africa. close relationship. Literate travelers reported Witwatersrand University Press, Johannes- "Bushmen" with domestic animals in this burg, 1992.228 pp. R69.00. area in 1811, who were gone by the 1820s. Alan Moms has chosen a study area The Kakamas and Abrahamsdam along the Orange and Riet Rivers, in the burials were excavated in the 1930s. The former northern frontier territory of South latter group is small and not well preserved, Africa. He offers an analysis of four sets of and their analysis is less informative than late prehistoric to early historic burials, with that of the other groups. The Kakamas the goal of showing that human bioarchae- burials have been discussed widely in the ology can provide insights into the literature, but Morris shows that previous biological and cultural interactions among accounts were inconsistent and incomplete, the varied ethnic groups that came into prompting thorough reanalysis. Three 14C contact in this region. dates among this group range tightly in the As Morris desaibes it, the area is one of mid eighteenth to early nineteenth centuries, complex cultural interactions. Although the and one grave whose extended position history of the Orange River frontier, like that betrays Christian influences suggests a post- of many frontiers, is often presented simply 1811 funeral. Historic sources place a Khoi as an African-European encounter, Morris group called the Einiqua in this area, but no notes that the contact situation is much more domestic sites are associated with the complicated. Before European arrival, burials. Africans of Khoisan and Bantu derivation The Griqua burial sample is well met, maintaining or blending hunter- provenienced from an early historic gatherer (San), pastoral (Khoi), and cemetery, with documentary and oral agricultural (Bantu) adaptations. When testimony strongly supporting the Griqua Europeans arrived, each ethnic group and identity. These burials are extended and lifestyle interacted with the Europeans and supine, 28% in coffins, evidence of European their market-agricultural background. cultural affinities. By the nineteenth century, one result of Morris reviews ethnographic and this multilateral interaction was the groups historic data regarding burial practices, and broadly called Kommandos, displaced Khoi NYAME AKUMA No. 47 June 1994

is able to delineate four general patterns. system with its agriculture and refined Khoi burials rarely contain grave goods, and foods. The differences in caries between Riet the burial pits are characterized by niches River and Kakamas burials, the latter having and covered by large cairns. San graves better dental health, is related to higher often have accompanying artifacts, but levels of natural fluoridation in the drinking rarely niches or large cairns. Bantu burials water, a neat illustration that osteo- often have niches, low cairns, and abundant archaeological data reflect both the cultural grave goods. The Christian pattern is most and natural environments. marked in the extended, supine position of Chapter 8, presenting scenarios of gene- the body. The African patterns evidence flow between groups of equal and unequal considerable overlap, but can generally be social status, is a bit mare theoretical than distinguished. Of the archaeological the previous discussions, but raises samples, the Riet River style shows both San interesting and testable propositions while and Bantu affinities; the Kakamas pattern adding another perspective to the analysis. most resembles the Khoi style; and the Morris's book is based on his Ph.D. Griqua graves are characteristically work, and reflects something of a Christian extended burials, some with dissertation style. This is, on the whole, niches, which suggest Khoi and/or Bantu positive: the summaries are systematic and antecedents. clear throughout, the methodological The central chapters of Morris's sections, though tedious to a nonspecialist, monograph are the methodological basis for provide a solid explanation of how the comparing the archaeological samples to analysis was accomplished, and the great modem Khoisan, South African Negro, and masses of data are available in appendix. Caucasoid genetic groups. Through The production of the volume is good: univariate, bivariate, and multivariate typographical errors are rare or lacking, and statistical tests, he shows first that each although the few photos are a bit fuzzy, the archaeological sample is a relatively drawings are well done. In particular, the homogeneous group and may be treated as graphic presentations of the multivariate a single population. (But it would also have statistical analyses in chapter 6 are very been interesting to test whether all of the clear. archaeological samples could have been The various lines of evidence prove to treated as a homogeneous population, or at be complementary, and yield a more least the Riet River and Kakamas samples.) complex and complete picture of the Morphologically, the Riet River people archaeological populations than are most like Khoisan. The Kakarnas people documentary, archaeological, or osteological resemble both Khoisan and South African data could do in isolation. Morris has made Negro populations. The Griqua archae- a very useful contribution to the ological sample, although surprisingly archaeology of the region, and to the homogeneous given their historically synthesis of human bioarchaeology and diverse origins, show the most variability ethnohistorical studies. The only under multivariate analysis: they evidence a disappointment of the work is the lack of "predominantly Negro morphology" with connection of the burial populations with some Khoisan characteristics, and one or their archaeological domestic context: this is two specimens with Caucasoid affinities. not a fault of Morris or his study, but a From this reviewer's perspective, challenge to regional archaeologists. chapter 7, on osteological indications of lifestyle, is perhaps the most interesting. The heavy wear on the Riet River and Kakamas teeth is consistent with a traditional lifestyle, hunting and gathering with perhaps some pastoralism. The Griqua, in contrast, show much less wear but a high rate of caries, betraying their connections to the colonial NYAME AKUMA No. 41 June 1994

Reviewing the systematic context of stone Book Review artifacts, these extend from raw material acquisition, to tool production and use, and continue up to the point when artifacts are Marie-Louise Inizan, Hdl&ne Roche, and discarded into what becomes their Jacques fixier, Technology of Knapped archaeological context. All of the methods Stone currently used in lithic analysis are reviewed here (except for the behavioral Pamela R. Willoughby analysis which is used to interpret site Department of Anthropology, formation processes). University of Alberta Chapter 1 deals with raw material Edmonton, Alberta characteristics; why were certain materials T6G 2H4 selected for knapping and not others? Lithic Canada materials are classified here by their physical properties and suitability for working, rather than by how the rock was Marie-Louise Inizan, Hbl2ne Roche, and formed or its mineralogy. In this way, the Jacques Tixier, Technology of Knapped Stone. focus is on the decisions the prehistoric Pr6histoire de la Pierre Taillk 3. Meudon: knapper would have made, rather than with Cercle de Recherches et df8tudes the concerns of modern geologists. It is Prbhistoriques, 1992. ISBN 2-903516-03-0. pointed out that certain rocks are good for 127 pages. 130 FF plus 45 FF postage. some methods of reduction, but not for others. Experimental analysis has shown This volume is the third in a series on that raw material availability and variability lithic technology produced by the Cercle de are critically important for the interpretation Recherches et dj~tudesPr6historiques of prehistoric materials, even in regions (CREP) based in Meudon, France. It is where abundant lithic resources existed. available from the publishers (CREP, CNRS, Issues of heat treatment (why and how) are 1, place Aristide Briand, 92195, Meudon, addressed, as well as procurement. Is the France). This laboratory concentrates on the archaeological material from a local source, technological analysis of Palaeolithic or has it been transported over a long assemblages. The first (1980) volume distance? What does this tell us about the produced in this series was in French and actions and mobility patterns of the people dealt with general lithic terminology and who produced the assemblage? technology; the second, also in French, discussed blade technology (economic du The authors review methods that can be de'bitage hminaire). This third publication is a used to understand a lithic assemblage. revised and updated English translation of These include conjoining and refitting, the 1980 one. New sections deal with experimental replication of knapping specific research methods and analytical techniques, usewear analysis, and techniques developed or expanded since the examining the cognitive (including decision first volume was produced. making) and psychomotor processes used to produce tools. Inizan and her colleagues In this work, technology is defined as "a distinguish between conjoining and conceptual approach to prehistoric material refitting. Conjoining is matching pieces or culture, based on the reasoned study of fragments together after determination of techniques, including those of human positive or negative knapping surfaces. physical actions" (p. 11). This is more a Refitting involves a complete series of definition of the approaches one can use to conjoining sets that belong to the same block study technology rather than the term itself. of raw material. Both techniques are These approaches are the focus of this necessary in order to reconstruct the chaine volume and are grouped into sections d'opiratoire or reduction sequence used to examining the stages of artifact production, produce specific artifacts and assemblages. resulting in a sort of life history analysis. NYAME AKUMA No. 41 June 1994

Chapter 2 reviews (and distinguishes publishes The South Afrtcan Archaeological between) the natural and cultural processes Bulletin, which is the primary publication for that can alter the surface appearance of current archaeological research in southern stone. Initial happing techniques and their Africa. Periodically, thematic collections of products are reviewed in chapter 3, and final papers appear in the society's Goodwin tool shaping in chapter 4. Chapter 5 reviews Series, and the society also produces a cores and other debitage products, while popular newsletter, The Digging Stick. chapter 6 offers a way to identify and Membership of the society includes classify the full range of retouch techniques. subscriptions to these publications. Chapter 7 provides a terminological lexicon Authors of manuscripts on topics relating to lithic technology, summarizing relevant to African archaeology, and who much of the same information provided in wish to have them considered for Michel Brezillon's 1977 publication, la publication in The South African Dknomination des Objets de Pierre TailLie. The Archaeological Bulletin, should send three text is followed by a multilingual copies to The South African Archaeological vocabulary, with translations from English Society, P.O. Box 15700, Vlaeberg 8018, into Arabic, French, German, Greek, Italian, South Africa. For 1994, individual Russian, and Spanish. subscription rates are R40 for South This book provides a good introduction Africa/Namibia, R50 or WlOor US$16 for to the terminology of lithic analysis and is the rest of Africa, and R72 or UKE14 or primarily directed at Palaeolithic US$23 elsewhere in the world. Institutional archaeologists. Those of us who teach lithic rates are available on request. typology and/or technology are always on the lookout for good background information. Unfortunately, the perfect textbook for such a course does not yet exist. This volume would be a good supple- mentary text for a course in lithic analysis, or for those interested in studying the replication of tool manufacture and use. However, it should be compared to current publications, such as those dealing with North American or indeed African stone age research, in order to achieve an adequate coverage of the current state of this field.

I Journals I

South African Archaeological Society Judy Sealy, the Honorary Secretary of the South African Archaeological Society, has requested that Nyame Akuma call the attention of its readership to the Society's publications, particularly the South African Archaeological Bulletin. The following material was submitted by Dr. Sealy: The South African Archaeological Society was founded in 1945 to promote archaeology through education and publication. Bi-annually, the society NYAME AKUMA No. 47 June 7994

Amine d'Arddologie, mieux connue sous le sigle W.A.J.A. (West African Journal of Archueology) fetera, 1 ce moment-la, so quart de siMe d'existence. Un nouveau bureau executif de dix membres a 4td constitue, avec 1 sa ete pour Compte-Rendu: Vle Colloque la premiere fois dans l'histoire de cette de I'Association Ouest- organisation, une femme, Madam le Africaine d'Arch6ologie Professeur M. A. SOWUNMI, de (A.O.A.A.) 1'Universite d'Ibadan (Nig6ria). Par ailleurs l'Assemblk Gh6rale a adopt4 des mesures Cotonou, 28 mars-2 avril 1994 bnergiques de nature 1 assainir la gestion financike de l'association et 1 developper en son sein une ethique professionnelle plus Pendant pres d'une semaine, se sont grande. A cet sujet, 1'A.O.A.A. collabore reunis dans la viJle universitaire principal avec 1' ICMA 1 l'elaboration d'un Code de centre Cconomique de la Republique du diontologie en mat2re de pratique archhlogique Benin, 1 Cotonou, une cinquantaine de el Alnque. Avec son VIe colloque, 1'A.O.A.A. chercheurs, de conservateurs de musks, et prend un nouveau depart; il etait temps, vu d'etudiants archbologie, Venus de douze l'ampleur des tgches qui attendent les pays (Benin, Burkina Faso, Cameroun, archblogues de I'Ouest africain. France, Guine Mali, Maroc, Niger, Nigeria, On notera, avec inter&, l'identification Togo, U.S.A. et Zimbabwe). Le thhe de ce de trois themes majeurs d'int&& rdgional ou colloque aArcheologie et sauvegarde du sous dgional retenus 1 Cotonou: patrimoine* est d'actualit4, particulikement archblogie funQaire en Afrique de 180uest oh la rdce~te habitats perch& d'Afrique de 1'Ouest devaluation qui a frappe plusieurs Etat archblogie de l'esclavage membres de la zone C.F.A., aggrave les Des reseaux de chercheurs se problhes Cconomiques auxquels sont dbja constitueront autour de ces themes, places confrontiks les populations. chacun, sous la responsabilite d'un Le lien entre la d4gmdation des milieux coordinateur-respectivement MM Boube de vie (environnement) et celle du GADO, J. B. KIETHEGA, et B. W. ANDAH. patrimoine archeologique, d'une part et, la L'ssodation de dotera d'un bulletin semestriel crise konomique gtWralisk, d'autre part, de liaison et deploiera ses efforts vers la n'a pas Cchappe 1 la perspicacite des mobilisation des chercheurs et de fonds et participants au colloque de l'A.O.A.A. Afin aussi, pour la sensibilisation des pouvoirs de poser des diagnostics assures, une suite publics, des administrations, et des de trent communications a dot6 les dbbats communautes de base afin de reduire les d'un large eventail #etudes de cas ou de agents de destruction qui entament definitions de concepts opkatoires. Des dangereusement le patrimoine en surface ou travaux en commissions ont permi enfoui. d'approfondir la +flexion et de degager des Les participants au VIe colloque ont eu resolutions sous forme d'enoncb de bgalement 1 entendre les messages programme d'activitks pour les deux annks d'organisationssoeurs comme l'Asso$ation 1 venir. Un bilan des actions menb dans ce panafricaine de Rehistoire et des Etudes ddai, sera fait 1 l'occasion du VIIe colloque associks (P.A.A.) repdsentk par M. Gilbert qui correspondra au XX anniversaire de PWITI qui a fait part 1 l'assistance de 18Association.En hommage 1 la Republique l'avancement de l'organisation de la Fbderale du Nigeri qui l'a vue naitre en conference prdvue 1 Harare en septembre 1976, l'Assemblee GenQale reunie le 30 1995; la Socikte des Archfiologues mars 1994, dkidh que le prochain colloque Africanistes (SAfA) et le Conseil des Arts de et la commemoration auront lieu dans ce llAssociation des ~tudesAfricaines pays. 11 se trouve aussi que la Revue Ouest (A.C. A.S.A.) represen tks par le kofesseur NYAME AKUMA No. 41 June 7994

Merrick POSNANSKY et le Congres Enfin, signalons la parution (en Mondia. d'Ar&ologie (W.A.C.) reprCsent6 prbtirage) des Actes du Ve colloque-Di. ans par le Professeur Bassey W. ANDAH. de recherches archt!ologiques en Afrique de Le colloque s'est achev4 par une I'Ouest: perspectives de coopination rdgionale- excursion mgWevers lfuneou l'autre des (Ouagadougou, 27 juillet-ler aout 1992) qui vieilles ci&s de Porto-Novo et de Ouidah oh sont disponibles awc conditions suivantes: l'attention des participants a 4t4 pode tant France et Europe Occidentale : 125 FF sur les questions du patrimoine culturel (tous frais compris) (jardin botanique d'un c8t6, for& relicte de fitats-~nis et Canada: US $ 25 l'autre, Qosion cdtiere, etc.) que sur les (emballage et +dition compris) probQmes de maintien et de conservation Commande Zt passer aupres de du patrimoine culturel physique M. Obar* B. BAGODO (architectures locales, afro-bresilienne, Trknier de 1'A.OA.A. coloniale; musks; collections historiques ou B.P.-& arch40logiques~ etc.). Au total, le VIe Porto-Novo @&in) colloque s'est dQoul4 dans un atmosphbe mode de r&lement, de pdfbrence par stimulante d'dchanges et de franchise dans mandat postal; en cas de paiement par les propos. Malgr4 le moyens assez cheque, prbvoir les his de transfert modestes dont a disposk le Cornit6 national bancaire en sus. pr4paratoiref de conditions d'accueil et de Autres documents disponibles: travail d4centes ont 4t6 offertes Zt chacun de participants @ce awc efforts conjugubs du Ripertoire provisoire des archt?ologues gouvernement bbninois, d lfUniversit6 travaiUant en Afrique de 180uest, 1993, Nationale du B4nin et de quelques Porto-Novo, AOAA/WAAA, 25 pp. op4rateurs hnomiques de place. (25 FF; US%) Documents de travail du VIe colloque (Cotonou, 28 mars-2 avril 1994), AOAA/WAAA, 29 pp. (versions frangaise et anglaise disponibles: 25 FF; UW).