The question then arises, do species differ from species _solely_ in points of a useful character, as they undoubtedly should do if natural selection has been the sole factor in their formation? Investigation shows that systematists recognise a species by a collection of characters, the value of a character depending not on its _utility_, but upon its _stability_; in fact, a large proportional number of specific characters, such as minute details of structure, form, and colour, are wholly without meaning from a utilitarian point of view. Investigation further shows that the most general of all the 'notes' of a true species is _cross-infertility_, that is, the infertility of the offspring of two individuals belonging to separate species: this, it was urged, could not be due to the action of natural selection. Lastly, apart from the primary distinction of cross-infertility, and the inutility of so many of the secondary specific distinctions, neither of which can be explained by the action of natural selection, Mr. Romanes was strongly of the opinion that even if a beneficial variation did arise, the swamping effects of free intercrossing would reabsorb it, and so render evolution of species in divergent lines, as distinguished from linear transmutation, impossible. This last difficulty can only be met by assuming that the same beneficial variation arises in a number of individuals simultaneously, for which assumption our present knowledge furnishes no warrant. If natural selection is brought forward as the sole factor in the guidance of organic evolution, then he considered that these difficulties remain insurmountable; if, however, it is regarded as a factor, even the chief factor, then these difficulties vanish, it being consistent, in the latter case, to hold the other factor, or factors, responsible for an explanation of the difficulties in question. It was the object of this paper to suggest another factor in the formation of species, which, although _independent_ of natural selection, was in no way _opposed_ to it, and might be called _supplementary_ to it, and was at the same time capable of explaining the facts, of the _inutility of many specific characters_, the _cross-infertility of allied species_, and the _non-occurrence of free intercrossing_. Very briefly indicated, Mr. Romanes' line of argument is as follows:--Every generation of every species presents an enormous number of variations, of which only the ones that happen to be useful are preserved by natural selection. The useless variations are allowed to die out immediately by intercrossing.