I have sometimes used the trick of writing the paper before the experiments were donenot publishing it, of course,but the discipline is very good for choosing which experiments should be done.