Bibliography: pages 194-207. / This ethnography is the product of fourteen months of communication with residents of a squatter settlement near Somerset West in the Western Cape. The thesis explores the ways in which domestic relationships altered over the research period, locating these changing patterns in the contexts of informal settlement in the region. I show that in the context of the settlement the use of household as an analytic term was problematic because domestic relationships were fluid and ephemeral, making it difficult to establish patterns of 'belonging' over time. Network approaches are more effective than household in describing social relationships, but networks were also problematic in that they tend to assume patterns of reciprocity which were not always echoed in the behaviours of residents of Die Bos. The thesis concentrates on three main areas of social interaction. I explore labour relationships within and between households, showing that a focus solely on households obscures the processes of labour allocation within domestic units, and those which occur across their (permeable) boundaries. I examine changing patterns of commensality among some members of the population of Die Bos, showing how movement and labour were intimately linked with eating patterns. Here I show how the most effective way of describing these patterns is in terms of networks of informal interaction which are formalised briefly. I then discuss of how movements of certain sections of the population render the boundaries of domestic units extremely permeable. I conclude by showing that although the notion of household is useful in some contexts in describing interactions in Die Bos, it tends to assume too much homogeneity and constancy to describe accurately the fluidity of social relationships. Network approaches are possibly of greater use in such descriptions, but are shown to be problematic in that they assume constancy (although of a lesser degree than households do) in interaction.