Another theory might be based upon the supposition that natural death has been derived, in the course of time, from an artificial death which always appeared at the same stage of each individual life—as we have supposed to be the case in the Orthonectides. I cannot agree with this view, because it involves the transmission of acquired characters, which is at present unproved and must not be assumed to occur until it has been either directly or indirectly demonstrated[85]. I cannot imagine any way in which the somatic cells could communicate this assumed death by starvation to the reproductive cells in such a manner that the somatic cells of the resulting offspring would spontaneously die of hunger in the same manner and at a corresponding time as those of the parent. It would be as impossible to imagine a theoretical conception of such transmission as of the supposed instance of kittens being born without a tail after the parent’s tail had been docked; although to make the cases parallel the kittens’ tails ought to be lost at the same period of life as that at which the parent lost hers. And in my opinion we do not add to the intelligibility of such an idea by assuming the artificial removal of tails through hundreds of generations. Such changes, and indeed all changes, are, as I think, only conceivable and indeed possible when they arise from within, that is, when they arise from changes in the reproductive cells. But I find no difficulty in believing that variations in these cells took place during the transition from Homoplastids to Heteroplastids, variations which formed the material upon which the unceasing process of natural selection could operate, and thus led to the differentiation of the previously identical cells of the colony into dissimilar ones—some becoming perishable somatic cells, and others the immortal reproductive cells.