Recently, an inquirer of great caution and calmness of judgment, Prof. C. Lloyd Morgan, has expressed the opinion that the Lamarckian principle must at least be admitted as a working hypothesis. But with this I cannot agree, at least as things stand at present. A working hypothesis may be false, and yet lead to further progress; that is, it may constitute an advance to the extent of being useful in formulating the problem and in illuminating paths that are likely to lead to results. But it seems to me that a hypothesis of this kind has performed its services and must be discarded the moment it is found to be at hopeless variance with the facts. If it can be proved that precisely the same degenerative processes also take place in such superfluous parts as have only _passive_ and not active functions, as is the case with the _chitinous parts of the skeleton of Arthropoda_, then it is a demonstrated fact, that the cessation of functional action is not the efficient cause of the process of degeneration. At once your legitimate working hypothesis is transformed into an illegitimate dogma--illegitimate because it no longer serves as a guide on the path to knowledge but {33} blocks that path. For the person who is convinced he has found the right explanation is not going to seek for it. I can understand perfectly well the hesitation that has prevailed on this point in many minds, from their having seen _one_ aspect of the facts more distinctly than the other. From this sceptical point of view Osborn has drawn the following perfectly correct conclusion: "If acquired variations are transmitted, there must be some unknown principle in heredity; if they are not transmitted, there must be some unknown factor in evolution."[12] Such in fact is the case and I shall attempt to point out to you what this factor is. My inference is a very simple one: if we are forced by the facts on all hands to the assumption that the useful variations which render selection possible are always present, then _some profound connection must exist between the utility of a variation and its actual appearance_, or, in other words, _the direction of the variation of a part must be determined by utility_, and we shall have to see whether facts exist that confirm our conjecture.