This paper presents a technique for analysing the processes of population change which occur within an existing stock of housing. The first signs of residential change appear in changing occupancy duration, with turnover rates increasing or decreasing as the new pattern becomes established. But every area, however stable, has some degree of population change; the problem therefore, is to distinguish the genuine signs of change.From cumulative duration data, five distinct types of 'sales curves' are derived-the hypothetical limiting cases of occupancy durationreflecting different probabilities of occupants' leaving or remaining after given lengths of stay. By comparing actual with 'predicted' household changes, the essential supports of a stable residential pattern, or the significant changes, can be isolated.Of all urban land uses, residential areas are at once the most extensive and the focus of public attention in the planning process. But rarely do they attract an appropriate degree of interest from urban research workers. As a result, the development of basic analytic techniques, sensitive enough to attack the complex problems raised by residential areas, has taken a laborious and circuitous course over the years. Even today, there are relatively few such techniques which are widely accepted. Whilst elaborate and welltried research frameworks are applied to the commercial and industrial structure of cities, there persists a tendency to regard the great extent of residential land as one, implicitly homogeneous, functional element, or at best to distinguish between rather loosely defined categories of residential quality.Of course, this is not to deny all progress, for the landmarks are obvious. Of the early work, that of Burgess and the Chicago 'school' of human ecologists 1 and of Hoyt 2 is outstanding. More recently, has come renewed interest in 'social area analysis', 3 the advance of multivariate and other statistical techniques, 4 and the development by Alonso and others, of some sort of theoretical rationale behind urban residential patterns. 5 In addition, there are many useful descriptive and historical accounts of residential areas, some broad enough in context to suggest quite general conclusions,6 and others which attempt to define and measure the occurrence of specific