The explorations of structuring principles in twentieth-century Lamu and discussions of space in more ancient Swahili settlements therefore have common themes. Both focus on the centrality of the mosque in discussing meaning in the built environment and the principles of planning that produced Swahili towns. In this, as well as in other details of the town plan, these approaches assume a relationship between social structure and physical plan. Horton has extendeD this to map out the social constitution of ancient Shanga through its spatial manifestation. As such, all of these studies demonstrate how high-level principles such as Islam are played ouT through practice and more worldly interactions. For all of the studies mentioned, these worldly interactions occur in the realm of what Rapoport (1988) has dubbed "mid-level meanings," referring to the ways that the town plan reflects the agency of elite groups within society. The category of "elite" in Swahili includes rulers of towns who were organized as ranked hierarchies (e. g., Chwaka; see LaViolette anD Fleisher 2009), but also groups of wealthy merchants that formed more oligarchic town organizations. Those who shared in these more horizontally differentiated power bases were called waung-wana and were themselves distinct from, and hierarchically ranked over, urban newcomers and other non-freeborn members of society (LaViolette and Fleisher 2005:340). Thus, claims to historicity, grand monumental statements, and piety, might all be expressed through the built environment in different ways. Space might also be controlled and manipulated within towns, to the beneit of elite groups. Although this is not always explicitly discussed, there is an assumption that architectural and spatial control exists, giving agency to certain groups and their concerns within the Swahili towns.
These assumptions have the effect of assigning agency only to a certain portion of the population: this is actually at odds with the ethnographies that see structuring principles being played out through the agency of all inhabitants. Thus, the town layout is seen as being "planned" according to the intentions of particular, elite inhabitants. Secondly, the dominance of a particular model of Swahili urban form, linked in a circular argument with a historical model of Swahili society, means that there is little room to account
For diversity. The vast majority of fourteenth - to fifteenth-century towns therefore appear "unplanned" as they do not match with this model, or elaborate arguments and chronological leaps are required to make the spaces matcH the supposedly universal model.
SWAHILI URBAN SPACES Of THE EASTERN AFRICAN COAST
Thus, while these studies have gone a long way toward socializing our understandings of Swahili urban spaces, we suggest that it is necessary to take seriously the insights of the ethnographies with regard to structure played out through practice. As well as exploring the actions of powerful people, and the communal spaces and structures of the town, this means entwining daily practice - Rapoport's "low-level meaning" - into our explorations.
World History
