ABSTRACT
The excavation strategy was in part dictated by the presence or absence of Muslim graves which for obvious reasons could not be disturbed (see below). The methodology, however, was similar to that successfully used in comparable urban sites by this author in Gao and Timbuktu, Mali (Insoll 1996, 2000, 2002); namely the use of arbitrary levels where actual levels were difficult to recognise owing to, for example, colour changes in deposits as they dried out following exposure. The numbers assigned to the levels are placed in parentheses within the text. All deposits were sieved using a 2.5mm mesh sieve to allow the recovery of smaller classes of artefacts such as fish bones and beads.
