So much for the general principle of selective value as applied to this particular case. And similarly, of course, we might here repeat the application of all the other general principles, which have just been applied in the two preceding cases. But it is only one of these other general principles which I desire in the present case specially to consider, for the purpose of considering more closely than hitherto the difficulty which this principle presents to ultra-Darwinian theory. The difficulty to which I allude is that of understanding how all the stages in the _development_ of a reflex action can have been due to natural selection, seeing that, before the reflex mechanism has been sufficiently elaborated to perform its function, it cannot have presented any degree of utility. Now the particular force of the present example, the action of scratching--as also of the one to follow--consists in the fact that it is a case where a reflex action is not yet completely organized. It appears to be only in course of construction, so that it is neither invariably present, nor, when it is present, is it ever fully adapted to the performance of its function. That it is not invariably present (when the brain is so) may be proved by trying the simple experiment on a number of puppies--and also of full-grown dogs. Again, that even when it is present it is far from being fully adapted to the performance of its function, may be proved by observing that only in rare instances does the scratching leg succeed in scratching the place which is being irritated. The movements are made more or less at random, and as often as not the foot fails to touch the body at any place at all. Hence, although we have a "prophecy" of a reflex action well designed for the discharge of a particular function, at present the machinery is not sufficiently perfected for the adequate discharge of that function. In this important respect it differs from the otherwise closely analogous reflex action of the frog, whereby the foot of the hind leg is enabled to localize with precision a seat of irritation on the side of the body. But this beautiful mechanism in the frog cannot have sprung into existence ready formed at any historical moment in the past history of the phyla. It must have been the subject of a more or less prolonged evolution, in some stage of which it must presumably have resembled the now nascent scratching reflex of the dog, in making merely abortive attempts at localizing the seat of irritation--supposing, of course, that some physiologist had been there to try the experiment by first removing the brain. Now, even if one could imagine it to be, either in the frog or in the dog, a matter of selective importance that so exceedingly refined a mechanism should have been developed for the sole purpose of inhibiting the bites of parasites--which in every normal animal would certainly be discharged by an _intentional_ performance of the movements in question,--even if, in