The regrowth of an amputated part, as in the salamander or snail, is explained by the presence and development of gemmules previously thrown off from the part. The difficulty that a limb is produced of the same age as that which was lost, and not a larval limb, and that the cells with which the gemmules must unite at first are not those which precede them in the course of growth, but mature cells, is met by the consideration that this power is a special one adapted to meet special dangers to certain parts of certain animals, and that it is therefore probable that appropriate provision has been made by natural selection: it may be in the form of “a stock of nascent cells or of partially developed gemmules.” The existence of these latter in buds, and their absence from sexual cells, may account for bud development being the more direct and brief of the two. The much greater tendency to repair lost parts in lower and younger forms may be due to the same cause. The occasional tendency of hybrids to resemble one parent in one part and the other in another may be due to superabundance of gemmules in the fertilised germ, those from one parent having “some advantage in number, affinity, or vigour over those derived from the other parent.” The general preponderance of one parent over the other may be similarly explained. The cases in which “the colour or other characters of either parent tend to appear in stripes or blotches” are to be understood by the gemmules having an affinity for others of the same kind. _The sterility of hybrids_ is entirely due to the reproductive organs being affected; in the case of plants they continue to propagate freely by buds. The hybrid cells throw off hybrid gemmules which collect in the buds but cannot do so in the reproductive organs. _Development and metamorphosis._--The remarkable facts of development and metamorphosis are well explained by the hypothesis. Allied forms may pass to a similar end through very dissimilar stages or conversely. Parts may appear to develop within previously existing corresponding parts, or they may appear within parts which are quite distinct. These divergent facts are explained by the hypothesis, each part during each stage being formed independently from the gemmules of the same part in previous generations, and not, although it may appear to do so, from the corresponding parts of earlier stages. In the process of time certain parts during certain stages may be affected by use or disuse or surroundings, and the parts of subsequent generations will be similarly affected, because formed from correspondingly altered gemmules; but this need not affect the other stages of the same parts.