The often-cited reproduction of a crab's lost claw or a lizard's tail, Mr. Darwin thought explicable by his hypothesis of diffused gemmules, representing all organs or their component cells. But though, after simple amputation, regrowth of the proximate part of the tail is conceivable as hence resulting, it is not easy to understand how the remoter part, the components of which are now absent from the organism, can arise afresh from gemmules no longer originated in due proportion. Prof. Weismann's hypothesis, again, implies that there must exist at the place of separation, a ready-provided supply of determinants, previously latent, able to reproduce the missing tail in all its details--nay, even to do this several times over: a strong supposition! The hypothesis of physiological units, as set forth in preceding chapters, appears less incompetent: reproduction of the lost part would seem to be a normal result of the proclivity towards the form of the entire organism. But now what are we to say when, instead of being cut off transversely, the tail is divided longitudinally and each half becomes a complete tail? What are we to say when, if these two tails are similarly dealt with, the halves again complete themselves; and so until as many as sixteen tails have been formed? Here the hypothesis of physiological units appears to fail utterly; for the tendency it implies is to complete the specific form, by reproducing a single tail only. Various annulose animals display anomalies of development difficult to explain on any hypothesis. We have creatures like the fresh-water _Nais_ which, though it has advanced structures, including a vascular system, branchiæ, and a nervous cord ending with cephalic ganglia, nevertheless shows us an ability like that of the _Hydra_ to reproduce the whole from a small part: nearly forty pieces into which a _Nais_ was cut having severally grown into complete animals. Again we have, in the order _Polychætæ_, types like _Myrianida_, in which by longitudinal budding a string of individuals, sometimes numbering even thirty, severally develop certain segments into heads, while increasing their segments in number. In yet other types there occurs not longitudinal gemmation only, but lateral gemmation: a segment will send out sideways a bud which presently becomes a complete worm. Once more, _Syllis ramosa_ is a species in which the individual worms growing from lateral buds, while remaining attached to the parent, themselves give origin to buds; and so produce a branched aggregate of worms. How shall we explain the reparative and reproductive powers thus exemplified? It seems undeniable that each portion has an ability to produce, according to circumstances, the whole creature or a missing part of the creature. When we read of Sir J. Dalyell that he "cut a _Dasychone_ into three pieces; the hindermost produced a head, the anterior piece developed an anus, and the middle portion formed both a head and a tail" we