British Educational Research Journal, Vol. 13, No. 3, 1987 283 METHODOLOGICAL DEBATE: ETHNOGRAPHY Ethnography and the Cumulative Development of Theory: a discussion of Woods' proposal for 'phase two' research 1 MARTYN HAMMERSLEY, School of Education, The Open University The last twenty years have witnessed a sharp growth in the amount of ethnographic research on schools, in Britain and North America (Hammersley 1980; Delamont 1981; Woods 1983). However, most of this research has been relatively exploratory and descriptive, and diverse in focus, opening up new areas rather than systemati- cally investigating those where work has already begun. Reflecting on this, several writers have recently called for a concerted effort to develop and test theory (D. Hargreaves, 1981; Woods, 1985a; Hammersley, Scarth & Webb, 1985). However, they are not in agreement about what the cumulative development of theory involves and about how much has already taken place. In this paper I want to examine one of the most influential statements about this issue in the sociology of education: Peter Woods' argument for what he calls 'phase two' ethnographic research (Woods 1985a). I shall argue that his proposals do not go nearly far enough to bring about the cumulation of theory in this field. Woods argues that what is required now is phase two work which cultivates "leaps of (theoretical) imagination" (p. 52). It is necessary to move away from merely 'descriptive' to 'sensitizing' concepts, to develop and fill in existing theory, and to formulate new theories (p. 57). At the same time, though, Woods also sees phase two studies as involving "hypothesis formation and testing" and the process of analytic induction (p. 58). Woods does not regard phase two research as a complete break with the past, but rather as an extension of what has already occurred in some areas. As an illustration of an area where some theoretical cumulation has already taken place, he cites work on teachers' strategies: In some areas there has been cumulative work which illustrates the promise in maintaining the dialectic between theory and data collection. One of these areas is that of 'social' (Lacey, 1977), 'coping' (A. Har- greaves, 1978), or 'survival' (Woods, 1979) strategies. It is interesting that