Optimum experimental designs with a lack of a priori information. II. Designs for the estimation of the whole response function.