We suggesT that a practice-based model is an important route to understanding Swahili urban spaces. It does not deny or replace the models derived from Shanga's plan, but instead attempts to take into account the historicity of Swahili sites, and the ways that urban spaces can be built up through daily practice and through the
Lifecycles of the inhabitants and their social spaces. Approaching the site plans through the buildings that make them up can also provide insights into the ways that sites can develop, even at those places where there does seem to have been some element of regularized patterning, such as Takwa (Horton anD Middleton 2000; Smith 2007; Wilson 1982). As Allen (1979:8) puts it: "[A]s settlements developed in size and complexity, the need for a town plan became more pressing, and the owners of interlocking stone houses had to take this into account." Although we might question whether the "need" for a town plan develops with size, this quotation does illustrate the notion that a plan is built up from the houses, rather than imposed on them.
SWAHILI URBAN SPACES Of THE EASTERN AFRICAN COAST
What is compelling abouT the notion of town planning through delimited space and through practices that were intertwined with architecture is that it allows us to start thinking beyond the interior spaces or monumental architecture. The delimited space of the site, which - although external - is defined through the built structures, could include those communal spaces and structures that are created through their placement. In particular, many Swahili sites seem often to have contained open areas commonly associated with the mosque area. At Mtwapa, Kusimba (1996) suggests several uses for the open areas near the mosque, assuming that elite members of society deliberately created and controlled them. He suggests that open areas were reserved for future expansion, causing less powerful residents to have to build outside the town walls. Similarly, at Takwa, Wilson explains the open spaces just inside the town wall gates as possible storage areas for market produce. Both of these explanations rely on the notion of centralized planning and elite controL of open space. As Horton cogently argues, open spaces surrounding the mosque assumed important roles. However, rather than seeing such spaces as atomistic parts of an idealized Swahili town plan, a practice-based approach to open spaces might explore them as the spatial effects of house development, which may themselves have structured the patterning of buildings, and as spaces that came to have meaning through their use rather than by design (M. L. SmIth 2008). Thus, for example, Horton assumed that a central open space was part of Shanga's plan from the start, a protected space with ritual structures; yet, there is little consideration that this space may have become important through the process of building a mosque, as well as other flanking domestic structures nearby. In
This alternative, rather than seeking out town plans from twentieth-century Mijikenda kayas or Swahili towns, we might rather examine the architectural development of the town, and the changing use of delimited spaces. At Shanga, we can ask questions about why the central space, imbued with such importance, was encroached upon in later centuries, sucH that by the settlement's abandonment the central space ceased to exist. A narrow focus on architecture and street layout does not give us the tools to approach such questions or to view the towns as products of the activities conducted within them. We therefore feel that the approach pioneered by Allen for the Swahili house, which has already been extremely influential in the ways that we have viewed stonehouses, can be usefully extended to the entire urban environment and give us the tools to better understand SwaHili urban space.
We have begun that work at the ifteenth - to sixteenth-century site of Songo Mnara, on the southern Tanzanian coast (Fleisher and Wynne-Jones 2010). There, a central open space seems to be a constant part of the town plan, albeit for a short period of occupation. However, even this open space emerges through practices that delimited it, and seems to be anchored more closely to houses in the southern part of the site rather than those in the north. Although we are only beginning to tease apart the developmental history of open spaces at Songo Mnara, the central space there provides a crucial place to examine the interplay between high/middle and low-level meanings. As yet, it is unclear whether this area was structured around the building of a small central mosque and cemetery or if it was the delimited result of domestic structures to the south, east, and north. In either case, the space itself evidently became the place of a set of practices related to burial and memorialization. These ongoing acts of commemoration were crucial to the way that the area was preserved and maintained, a striking contrast to the open space aT Shanga. In the case of Songo Mnara, then, we need not envision that the town was built with a predetermined idea of a memorial, open space. Through the construction of houses, their expansion, as well as the placement of a mosque and burials associated with it, the "open space" at Songo Mnara, may have emerged through a history of only loosely interlinked practices. This does not, in any way, take away from the prominence and importance that this space assumed in the daily life of the town.
What we argue is that the acts of powerful people and places need not have been predetermined; the emergence of a town plan does not need to be the result of elites directing the construction and maintenance of town buildings and spaces. The generational power of elite people at Swahili sites relied not so much on structuring a town arounD the high-level meanings of Islam, purity, and pedigree, but rather their ability to imbue places and spaces with meaning that emerged through the developmental life of the town. The construction of tombs in the central open space, and the conduct of memorial practices at these sites is one example of how this might have been manifest. In this way, high-level meaning was easily located in the form anD development of urban town plans, as el-Zein demonstrates. However, we need to be careful not to substitute elite ideological notions of town plans, ones that were probably part of their ongoing power and authority within towns, for an historical model of how towns actually grew into various forms. Although we cannot lose sight oF high - and middle-level meanings that contributed to the structuring of Swahili towns, we would be foolish to assume that these alone were the determinants of town plans. There is no doubt that structural power was a contributing factor in the organization of Swahili towns, but as we learn from the regional study of Swahili towns and the emergence of the Swahili house, the practices of people are foundationaL in understanding how they emerge.
Swahili urban spaces of the eastern African coast
World History
