order to save an hypothesis at all costs, we make so violent a supposition as this, still we should do so in vain. For it would still remain undeniably certain that the reflex mechanism is _not_ of any selective value. Even now the mechanism in the dog is not sufficiently precise to subserve the only function which occasionally and abortively it attempts to perform. Thus it has all the appearance of being but an imitating shadow of certain neuro-muscular adjustments, which have been habitually performed in the canine phyla by a volitional response to cutaneous irritation. Were it necessary, this argument might be strengthened by observing that the reflex action is positively _improved_ by removal of the brain. The second example of a nascent reflex in dogs which I have to mention is as follows. Goltz found that his brainless dogs, when wetted with water, would shake themselves as dry as possible, in just the same way as normal dogs will do under similar circumstances. This, of course, proves that the shaking movements may be performed by a reflex mechanism, which can have no other function to perform in the organization of a dog, and which, besides being of a highly elaborate character, will respond only to a very special kind of stimulation. Now, here also I find that the mechanism is congenital, or not acquired by individual experience. For the puppies on which I experimented were kept indoors from the time of their birth--so as never to have had any experience of being wetted by rain, &c.--till they were old enough to run about with a full power of co-ordinating their general movements. If these young animals were suddenly plunged into water, the shock proved too great: they would merely lie and shiver. But if their feet alone were wetted, by being dipped in a basin of water, the puppies would soon afterwards shake their heads in the peculiar manner which is required for shaking water off the ears, and which in adult dogs constitutes the first phase of a general shaking of the whole body.