N. C. Janke's criticism of my paper seems aimed more at the title than the paper itself. I acknowledge the importance of shape in determining the style and rate of movement of particles in fluids (Janke, 1966), but my paper attempted a different task, clarifying the basic concepts of shape, which I believe are spatial, not behavioural. It did occur to me at the time of writing that a complementary review of the relationships between the shape of particles and their behaviour in different situations would be useful, but I saw this as a separate problem. The main purpose of Janke's 1966 paper was not to provide a measure of some particular aspect of shape, but to find a formula using different predetermined aspects of shape to predict settling velocities of particles. Nevertheless, points do arise in his discussion that could usefully have been taken up in my review, notably his acceptance of Prandtl's (1952) view that shape, as it affects settling velocity at least, has two components, a form component and a surface area component. Although Janke had considerable success in finding an appropriate mix of each component to allow prediction of settling velocity from two shape measures, he does not argue a basis for the measures themselves. For example he replaces Corey's shape factor with his shape factor E because the latter ‘allows more discrimination’for shapes with equal axial lengths. Corey's factor assigns a value of 1 to spheres, equant cylinders, equal double cones and cubes, whereas Janke's E gives 1, 0·866, 1 and 0·775 respectively. However, I have difficulty imagining a useful concept of form in which spheres and double cones are in some sense equal and distinct from cylinders and cubes, and it seems to me important that measures like Janke's are conceptually clear as well as efficacious. I would very much like to see someone with experience like Dr Janke undertake the review I mentioned earlier in the hope of achieving such a clarification.