I need not here go further into detail, but I should still like to show that, in reality, as I assumed in regard to the regenerative capacity of a part, the root of the regeneration-idioplasm lies in the germ-plasm, that it is present there as an independent determinant-group, and, like every other bodily rudiment (_Anlage_), must be handed on from generation to generation. This assumption is necessary, as has been already indicated, on the ground that the faculty of regeneration is hereditary, and hereditarily variable, on the same ground, therefore, as that on which the whole determinant theory is based. The regeneration-determinants must be contained _as such_ in the germ-plasm, otherwise a twofold phyletic development could not have occurred, as it actually has, in many parts. The tail of the lizard is adapted for autotomy; it breaks off when it is held by the tip, and this depends on a special adaptation of the vertebræ, which are very brittle in a definite plane from the seventh onwards. This is thus a very effective adaptation to persecution by enemies. The tail which has been seized remains with the pursuer, but the lizard itself escapes, and the tail grows again. But this regeneration does not take place in the same way as in the embryo; no new vertebræ are formed, but only a 'cartilaginous-tube,' a new structure, a substitute for the vertebral column; the spinal cord with its nerves is not regenerated either, and the arrangement of the scales is somewhat different. This last point, in particular, indicates that the determinants of the regeneration-rudiment may pursue an independent phylogenetic path of their own, for this scale arrangement of the regenerated tail is an atavistic one, that is, it corresponds to a more primitive mode of scale arrangement in these Saurians. We know quite a number of cases similar to this. It not infrequently happens that cut-off parts regenerate, but that they do so not in the modern form, but in one that is in all probability phyletically older. Thus the legs of various Orthoptera, as of the cockroaches and grasshoppers, regenerate readily, but with a tarsus composed of four joints instead of five[3], and the long-fingered claws of a shrimp (_Atyoida potimirim_) is replaced by the older short-fingered type of claw, while in the Axolotl an atavistic five-fingered hand grows instead of the amputated four-fingered one. [3] New investigations, specially directed to this point, by R. Godelmann, have shown that 'in the great majority of cases' the regenerated legs of a Phasmid (_Bacillus rossii_) exhibit a four-jointed tarsus; but the regeneration of five joints also occurs, though only after autotomy, and only in seven out of fifty cases (_Archiv für Entwicklungsmechanik_, Bd. xii, Heft 2, July 1901). The regeneration-rudiment in this species seems to be in process of advancing slowly to the five-jointed type.