This is precisely analogous to the behaviour of our *Tubularia*. *Tubularia* also may behave in three different ways, if, as I described to you, the terminal one of its two newly arisen rings of tentacle primordia is removed again. It may complete what is left, say the basal tentacle ring, then put forth from the horny skeleton (the “perisarc”) the new head as far as it is ready, and finally complete this head by a regular process of budding regeneration. But it also may behave differently. It may “erase” by a process of retro-differentiation all that has been left of what had already been formed, and then may form *de novo* the totality of the primordia of a new head. Or, lastly, it may remove a part of the middle of the one ring of tentacle rudiments which was left, and may use this one ring for the formation of two, which, of course, will not be quite in the normal relations of place with regard to each other and to the whole, but will be regulated afterwards by processes of growth. Thus, indeed, there is a sort of equifinality of restitution: one starting-point, one end, but three different means and ways. It would, of course, contradict the principle of univocality, as we shall see more fully later on, to assume that there actually are different ways of regulation whilst all the conditions and stimuli are the same. We are obliged to assume, on the contrary, that this is not the case, that there are certain differences in the constellation, say of the general conditions of age or of metabolism, which are responsible for any given individual choosing one process of restitution instead of another; but even then the phenomenon of equifinality remains very striking. It has long been known that restitution in general does not always follow the same lines of morphogenesis as are taken by ontogeny, and it was this feature that once led Roux to point out that the adult forms of organisms seem to be more constant than their modes of origin. But, comparing ontogeny with restitution in general, we see that only the ends are the same, not the points of starting; the latter are normal or non-typical in ontogeny, atypical in restitution. In the new discoveries of an equifinality of restitutions we have the *same* starting-point, which is decidedly non-typical but atypical, *i.e.* dependent on our arbitrary choice, leading by *different* ways always to the *same* end. There may be many who will regard the fact of equifinality as a proof of vitalism. I should not like to argue in this easy way; I indeed prefer to include part of the phenomena of equifinality in our first proof of autonomy, and part in the second one, which is to follow.