Designs for experiments--parallel comparisons of treatment.