Archaeological investigations of settlement and emerging complexity in the Middle Senegal Valley
McIntosh, Susan Keech
Doctor of Philosophy
This thesis reports on three months of excavation and a month of site survey undertaken in 1999--2000 in the central sector of the Middle Senegal Valley that was associated with the historical polity of Takrur. The focus of the research was two-fold: to investigate the earliest permanent or semi-permanent settlement along the Middle Senegal Valley floodplain; and to examine the emergence of larger-scale, more complex settlements in the region. Excavation of the five hectare site of Walalde, provisionally dated to the first century A.D., revealed a much earlier occupation by iron-using cattle herders that began c. 800--550 B.C., and continued until c. 200 B.C. The sequence appears to document the transition from stone- to iron-based technology, with the use of iron objects and stone initially, and evidence for iron production (smelting and forging) from 550--200 B.C. Copper with the distinctive chemical signature of the Akjoujt mines in Mauritania was also present after 550 B.C., attesting to trade and interaction over long distances. The iron and copper at Walalde are among the earliest metals recovered from excavation contexts in West Africa. Other important aspects of the Walalde sequence include ceramic materials and a series of red ochre burials. Excavations at a large cluster of sites at Kaskas revealed a continuation of the Walalde sequence for the period 150 B.C.--A.D. 100, linking the settlement history documented by this field work to the existing archaeological sequence for the region. Iron production was a significant activity at some of these, and fishing was intensively practiced at others. Kaskas may represent the spatial integration of specialist economies (herding, fishing) to achieve greater productivity in a situation of high environmental stress produced by an extreme drought c. 1900 BP. The survey found and documented 22 sites on the transitional zone from floodplain to upland. This adds to our understanding of settlement distributions formerly known primarily from survey on the floodplain.