also as wings for carrying the animal a short distance through air. Again, in the case of the so-called flying squirrels we find the limbs united to the body by means of large extensions of the skin, so-that when jumping from one tree to another the animal is able to sustain itself through a long distance in the air by merely spreading out its limbs, and thus allowing the skin-extensions to act after the manner of a parachute. Here, of course, we have not yet got a wing, any more than we have in the case of the flying-fish; but we have the foundations laid for the possible development of a future wing, upon a somewhat similar plan as that which has been so wonderfully perfected in the case of bats. And through all the stages of progressive expansion which the skin of the squirrel has undergone, the expansion has been of use, even though it has not yet so much as begun to acquire the distinctive functions of a wing. Here, then, there is obviously nothing "prophetic" in the matter, any more than there was in the case of the swim-bladder and the lung, or in that of the nerve-ending and the eye. In short, it is the business of natural selection to secure the highest available degree of adaptation for the time being; and, in doing this, it not unfrequently happens that an extreme development of a structure in one direction (produced by natural selection for the sake of better and better adapting the structure to perform some particular function) ends by beginning to adapt it to the performance of some other function. And, whenever this happens to be the case, natural selection forthwith begins to act upon the structure, so to speak, from a new point of departure. So much, then, for the Duke's premiss--namely, that "every modification of structure _must_ have been functionless _at first_, when it began to appear." This premiss is clearly opposed to observable fact. But now, the second position is that, even if this were not so, the Duke's conclusion would not follow. This conclusion, it will be remembered, is, that if incipient structures are useless, it necessarily follows that natural selection can have had no part whatever in their inception. Now, this is a conclusion which does not "necessarily" follow. Even if it be granted that there are structures which in their first beginnings are not of any use at all for any purpose, it is still possible that they may owe their origin to natural selection--not indeed directly, but indirectly. This possibility arises from the occurrence in nature of a principle which has been called the Correlation of Growth.