Few zoologists realize that the usefulness to the systematist of any characteristic of an animal depends on the end in view. Since speciesare the units with which our work is chiefly done, their most usable characters are those which will readily distinguish them from each other. But when we try to arrange them in genera and show the natural interrelationships of any group, the characters in which they are alike are of far more importance than those in which they differ. For defining species then we are often justified in using characters which seem to us trivial and of no significance. Not infrequently newly proposed species are condemned because the characters on which they are based are said to be "so trivial." When this criticism is examined, however, it will be found that it really confuses two very different things which may be illustrated as follows. A newly described bird is said to differ from another species in having the iris white instead of brown and the tail feathers tipped with cream-color instead of yellow. Such differences are certainly trivial, but experience has taught us that the color of the iris rarely, if ever, shows such a degree of diversity within a single species, and if the examination of a fair amount of material shows the two characters given to be constant, the validity of the species can hardly be questioned. If, however, the new bird is said to be characterized by a longer bill, a more markedly yellow tinge below and a greater amount of yellow on the tail feathers, these characters would not only be considered trivial but we should be justified in being skeptical of the validity of the species because experience has taught us that such characters are subject to very great diversity. If new species are to be condemned, then accuracy demands that it shall not be because the characters assigned are trivial, but because they are inconstant and hence unreliable. When we come to genera, however, while it is of course desirable that the characters should not be trivial in any sense, the essential point is that they should have some significance, either historical or physiological. These two terms are not synonymous, for such a structure as the vermiform appendix in man has a historical, though apparently not a physiological, significance. Usually, significant characters are more or less-conspicuous, and I am strongly inclined to believe that where they are internal, or otherwise difficult to ascertain, they are associated with external, or in some way obvious, characters. The formation of genera based on larval characters, or those of some other special stage of development, is greatly to be deplored, and I have no doubt that if such groups really represent natural relationships, differential characters will be found in the adults. If they are not, it seems to me clear that, notwithstanding the possibilities of what we call "parallelism" and "convergence," the characters of the special stage are temporary adaptations of variable significance. It is impossible to designate any
