"The eyes of these fish are curiously distorted in order that both eyes may be upon the upper side, where alone they would be of any use. . . . Now if we suppose this process, which in the young is completed in a few days or weeks, to have been spread over thousands of generations during the development of these fish, those usually surviving _whose eyes retained more and more of the position into which the young fish tried to twist them_ [italics mine], the change becomes intelligible." {24} When it was said by Professor Ray Lankester--who knows as well as most people what Lamarck taught--that this was "flat Lamarckism," Mr. Wallace rejoined that it was the survival of the modified individuals that did it all, not the efforts of the young fish to twist their eyes, and the transmission to descendants of the effects of those efforts. But this, as I said in my book, "Evolution, Old and New," {25} is like saying that horses are swift runners, not by reason of the causes, whatever they were, that occasioned the direct line of their progenitors to vary towards ever greater and greater swiftness, but because their more slow- going uncles and aunts go away. Plain people will prefer to say that the main cause of any accumulation of favourable modifications consists rather in that which brings about the initial variations, and in the fact that these can be inherited at all, than in the fact that the unmodified individuals were not successful. People do not become rich because the poor in large numbers go away, but because they have been lucky, or provident, or more commonly both. If they would keep their wealth when they have made it they must exclude luck thenceforth to the utmost of their power, and their children must follow their example, or they will soon lose their money. The fact that the weaker go to the wall does not bring about the greater strength of the stronger; it is the consequence of this last and not the cause--unless, indeed, it be contended that a knowledge that the weak go to the wall stimulates the strong to exertions which they would not otherwise so make, and that these exertions produce inheritable modifications. Even in this case, however, it would be the exertions, or use and disuse, that would be the main agents in the modification. But it is not often that Mr. Wallace thus backslides. His present position is that acquired (as distinguished from congenital) modifications are not inherited at all. He does not indeed put his faith prominently forward and pin himself to it as plainly as could be wished, but under the heading, "The Non-Heredity of Acquired Characters," he writes as follows on p. 440 of his recent work in reference to Professor Weismann's Theory of Heredity:--