Now it hardly needs pointing out to students of Darwinism that indefinite variation is the equivalent of "a variation to which no exact limits can be placed," and in this sense the assumption is a perfectly sound one. From one point of view the variations must be definite, that is, they can only occur within certain limits. An elephant will not vary in the direction of wings, nor will a bird in the direction of a rose bush. But so long as we cannot fix the exact limits of variation we are quite warranted in speaking of them as indefinite. That this is a methodological device no one denies, but so are most of the other distinctions that we frame. Scientific generalizations consist of abstractions, and Mr. Schiller himself of necessity employs the same device. Mr. Schiller argues, quite properly, that while Natural Selection states the conditions under which animal life evolves, it does not state any reason why it should evolve. Selection may keep a species stationary or it may even cause it to degenerate. Both are fairly common phenomena in the animal and plant world. Moreover, if there are an indefinite number of variations, and if they tend in an indefinite number of directions, then the variation in any one direction can never be more than an infinitesimal portion of the whole, and that this one should persist supplies a still further reason for belief in "a purposive direction of variations." Mr. Schiller overlooks an important point here, but a very simple one. It is true that any one variation is small in relation to the whole of the possible or actual number of variations. But it is not in relation to quantity but quality that survival takes place, and in proportion to the keenness of the struggle the variation that gives its possessor an advantage need only be of the smaller kind. In a struggle of endurance between two athletes it is the one capable of holding out for an extra minute who carries off the prize.