Helioseismic inferences have demonstrated very clearly the importance of including element diffusion and settling in solar modelling: models incorporating these processes are in substantially better agreement with the inferred solar sound speed than are models that neglect them. The remaining discrepancy between the models and the Sun has been taken as evidence for mixing in the region just beneath the convection zone. However, rather more serious discrepancies have resulted from a revision of solar abundances, and no obvious solution to this problem has been found so far. This perhaps demonstrates the danger of complacency when dealing with so complex a thing as a star. Hydrodynamical instabilities are likely to play a more important role than acknowledged in standard stellar modelling. An interesting example, if not relevant to modelling up to the present solar age, is the possible onset of semiconvective instability just beneath the convection zone, as first emphasized by Bahcall et al. (2001).