constituting independent individuals; and in other cases doing it so that the segments spontaneously separated are but a small part of the string. Thus a _Syllis_, Fig. 168, after reaching a certain length, begins to transform itself into two individuals: one of the posterior segments develops into an imperfect head, and simultaneously narrows its connexion with the preceding segments, from which it eventually separates. Still more remarkable is the extent to which this process is carried in certain kindred types; which exhibit to us several individuals thus being simultaneously formed out of groups of segments. Fig. 169, copied (omitting the appendages) from one contained in a memoir by M. Milne-Edwards, represents six worms of different ages in course of development: the terminal one being the eldest, the one having the greatest number of segments, and the one that will first detach itself; and the successively anterior ones, with their successively smaller numbers of segments, being successively less advanced towards fitness for separation and independence. Here among groups of segments we see repeated what in the previous cases occurs with single segments. And then in other annelids we find that the string of segments arising by gemmation from a single germ becomes a permanently united whole: the tendency to any more complete fission than that which marks out the segments, being lost; or, in other words, the integration having become relatively complete. Leaving out of sight the question of alliance among the types above grouped together, that which it here concerns us to notice is, that longitudinal gemmation does go on; that it is displayed in that primitive form in which the gemmæ separate as soon as produced; that we have types in which such gemmæ hang together in groups of four, or in groups of eight and ten, from which however the gemmæ successively separate as individuals; that among higher types we have long strings of similarly-formed gemmæ which do not become individually independent, but separate into organized groups; and that from these we advance to forms in which all the gemmæ remain parts of a single individual. One other significant fact must be added. There are cases in which annelids multiply by lateral gemmation.[25] That the longitudinally-produced gemmæ which compose an annelid, should thus have, one of them or several of them, the power of laterally budding-off gemmæ, from which other annelids arise, gives further support to the hypothesis that, primordially, the segments were independent individuals. And it suggests this belief the more strongly because, in certain types of _Cœlenterata_, we see that longitudinal and lateral gemmation _do_ occur together, where the longitudinally-united gemmæ are demonstrably independent individuals.