Roux explicitly adopted mechanical causality as a necessary explanatory principle in embryology, while Driesch introduced the complementary idea of the morphogenetic field. The present status of these concepts in developmental biology is described, together with recent developments in the formal description of field behaviour. From this, two general properties of developing tissue are derived which suggest that an important locus of decision-making in pattern formation is the cell membrane or cortex, and that the basis for this logical switching may lie in molecular interactions of the type which generate phase transitions. The role of statistical mechanics in a developmental context, both as a general explanatory model and as an explicit analytical tool, is discussed.