serious confusion. In view of all the evidence then which must be taken into account, it is clear that the final decision on the validity of any group must be rendered by the systematic workers in the larger group to which it belongs. It is not to be expected that physiologists or experimental zoologists or even entomologists can decide as to either the validity of a species of sea-urchin or the desirability of a genus of birds. And this leads to a third principle often overlooked.
While genera and larger groups in our systems of classification ought to be based on relationship, their delimitation is often of necessity artificial and is purely a matter of expediency or convenience.
The English zoologist Duncan maintained that it is "impossible to admit genera which are not differentiated by characters which have a decided and important physiological value." Other zoologists have ignored or distinctly repudiated this view. The difference of opinion seems to be based on a difference in the approach to the question and it is to this same difference of approach that all discussion as to the relative merits of large and small genera is due. The zoologist who is particularly occupied with the diversity of species and who has examined large numbers of individuals in the attempt to establish specific limits becomes impressed with the great importance of constancy in any given character and, when he finds a group of species which possess in common a character, or certain characters, constantly maintained, he finds it desirable to designate them by a common name and the group is to him a genus. Such a process naturally leads to numerous small genera. On the other hand, the zoologist who studies, from the morphologist's view point, relatively few specimens, representing perhaps many species, is naturally most impressed by the resemblances, and he finds that as regards characters which to him seem of physiological importance, his material divides into comparatively few groups. These he designates as genera and since the minor characters reveal a more or less marked differentiation into species his genera are naturally large. It seems to me that the difficulty is the same as would arise were one trying to decide what is a branchlet on a tree. In certain old, nearly dead trees, or in very young ones, there might be little difficulty, main branches, secondary branches, branchlets and twigs would all be sufficiently few to be distinguishable and there would be little disagreement as to the limits of the branchlets. But in most trees this would not be the case and much would depend on whether one began at the twigs and worked downward or at the branch and worked upward. In other words, while genera are, or at least ought to be, natural groups, their limits are often necessarily artificial and arbitrary. We recognize them by name for convenience and their limitation is largely to be determined by expediency. To say, as Duncan does, that it is "impossible" to recognize certain