This paper suggests that the first BaCoMET [Basic Components of Mathematics Education for Teachers] project (Christiansen, Howson and Otte 1986) can be seen as an important early attempt to sketch a "disciplinary matrix" (Kuhn 1962, 1970) for the field of mathematics education. This project brought together representatives of different national traditions of research in mathematics education, with the aim of identifying fundamental ideas which should be given high priority in any teacher education programme. My analysis of the project draws on Kuhn's (1970) prioritisation of different senses of "paradigm" in relation to the development of "normal science", and consequently draws out the central part played by clusters of exemplary problems (and solutions) in mediating between symbolic generalisations and practical action.