Those who have examined long series of these cases of resemblance among butterflies find it hard to believe that there is not some connection between them apart from climatic influence. One feels that they are too numerous and too striking to be all explained away as mere coincidences engendered by like conditions. Nor is it improbable that natural selection in the form of the discriminating enemy may have played a part in connection with them, though a different one from that advocated on the current theory of mimicry. If we assume that sudden and readily appreciable variations of the nature of "sports" turn up from time to time, and if these variations happen to resemble a form protected by distastefulness so closely that the two can be confused by an enemy which has learned to avoid the latter, then there would appear to be good grounds for the mimicking sport becoming established as the type form of the species. For it has already been seen that a rare sport is not swamped by intercrossing with the normal form, but that on the contrary if it possess even a slight advantage, it must rapidly displace the form from which it sprang (cf. Chap. VIII). On this view natural selection in the form of the discriminating enemy will have played its part, but now with a difference. Instead of building up a {144} mimetic likeness bit by bit it will merely have conserved and rendered numerically preponderant a likeness which had turned up quite independently. The function of natural selection in respect of a mimetic likeness lies not in its formation but in its conservation. It does not bring about the likeness, neither does it accentuate it: it brings about the survival of those forms which happen to shew the likeness. Why variations on the part of one species should bear a strong resemblance to other, and often distantly related, species is another question altogether.