For the following reasons, that birds and mammals seem to furnish the best field for testing the question by direct observation. First, these classes present many genera which have been more carefully worked out than is usually the case with genera of invertebrates, or even of cold-blooded vertebrates. Secondly, they comprise many genera each including a large number of species, whose habits and conditions of life are better known than is the case with species belonging to large genera of other classes. Thirdly, as birds and mammals represent the highest products of evolution in respect of organization, a more severe test is imposed than could be imposed elsewhere, when the question is as to the utility of specific characters; for if these highest products of organization fail to reveal, in a large proportional number of cases, the utility of their specific characters, much more is this likely to be the case among organic beings which stand lower in the scale of organization, and therefore, _ex hypothesi_, are less elaborate products of natural selection. Fourthly, and lastly, birds and mammals are the classes which Mr. Wallace has expressly chosen to constitute his ground of argument with regard to the issue on which we are now engaged. It would take far too long to show, even in epitome, the results of this inquiry. Therefore I will only state the general upshot. Choosing genera of birds and mammals which contain a large number of species whose diagnostic characters have been worked out with most completeness, I restricted the inquiry to specific distinctions of colour, not only for the sake of having a uniform basis for comparisons, but still more because it seemed that the argument from our ignorance of possibly unknown uses could be more successfully met in the case of slight differences of colour or of shading, than in that of any differences of structure or of form. Finally, after tabulating all the differences of colour which are given as diagnostic of each species in a genus, and placing in one column those which may conceivably be useful, while placing in another column those of which it appeared inconceivable that any use could be suggested, I added up the figures in the two columns, and thus obtained a grand total of all the specific characters of the genus in respect of colours, separated into the two classes of conceivably useful and apparently useless. Now, in all cases the apparently useless characters largely preponderated over the conceivably useful ones; and therefore I abundantly satisfied myself regarding the accuracy of my previous statement, that a large proportional number--if not an actual majority--of specific characters belong to the latter category. The following is a brief abstract of these results.