A Method of Testing Classification Relatives [Discussion]

In the introduction to my paper I emphasized that the new method of testing classification relativities proposed therein was undoubtedly subject to faults and susceptible to improvement and I asked for criticisms and suggestions. The discussions presented by Messrs. Bailey and Marshall show that this prediction was correct and that the method proposed can be substantially improved. I am very grateful for the criticisms offered and hope that they will help to develop a workable and conclusive method of testing classification relativities. The criticisms and suggestions fall into two groups; Mr. Marshall is chiefly concerned with the object and application of the method, while Mr. Bailey deals mainly with its technical aspects.