Most biological students are familiar with the similarity of the so-called eye of the mollusc Pecten and that of the vertebrate. The resemblance is one of general structure: in each of these organs there is a _camera obscura_, a transparent cornea, and behind that a crystalline lens. On the posterior wall of the camera there is a receptor organ, or retina, and this is composed of several layers of nervous elements. The actual nerve-endings are on the surface of the retina, which is turned away from the light, that is, the optic nerve runs towards the anterior surface of the retina, and then its fibres turn backwards. This “inversion of the retinal layers” occurs in all vertebrate animals, but it is exceptional in the invertebrates. The above general description applies equally well to the eye of the vertebrate and to that of Pecten. Let us admit that these mantle organs in Pecten _are_ eyes, for there is no conclusive experimental evidence that they really are visual organs, and plausible reasoning suggests that they may subserve other functions. Let us assume that the minute structure of the Pecten eye is similar to that of the vertebrate, and that its development is also similar: as a matter of fact both histology and embryology are different. Then we have to explain, on the principles of natural selection, the parallel evolution of similar structures along independent lines of descent; for mollusc and vertebrate have certainly been evolved from some very remote common ancestor in which the eye could not have been more than a simple pigment spot with a special nerve termination behind it. In each case the organ was formed by a very great number of serially occurring variations, yet these two sets of variations must have been the same at each stage in two independently occurring processes. On any reasonable assumption as to the number of co-ordinated variations required, and their chances of occurrence, the mathematical improbability that these two series of variations did occur is so great as to amount to impossibility so far as our theory of transformism is concerned. Natural selection could not, therefore, have produced these two organs. This argument of Bergson’s fails, of course, in the particular instance chosen by him, but this is because the case is an unfortunate one. Probably a morphologist could find a very much better case of convergent evolution--the parallelism between the teeth of some Marsupials and some Rodents, for instance. If detailed histological and embryological investigation should show a similarity of structure and development, in such compared organs Bergson’s argument would retain all its force. We should then have to assume that there was a directing agency, or tendency in the organism, co-ordinating, or perhaps actually producing, variations.