Such a cocoon is comparable to a work of art in which every part harmonizes with the rest, and all together are adapted as well as possible to their purpose. And yet it is all accomplished without the caterpillar having the remotest conception of what it is aiming at when it winds the endless silken thread about itself in the artistic and precisely prescribed coils. Nor has it any time for trying experiments or for learning; it must make all the complex bendings and turnings of the head which spins the thread, and of the anterior part of the body which guides the thread, quite exactly and correctly the first time if a good cocoon is to be produced. Here every possibility of interpreting this instinct as 'an inherited habit' is excluded, for each caterpillar becomes a pupa only once; and it is just as impossible to suppose that it can be directed by intelligence, since it can neither know that it is about to become a pupa, nor that, in the pupa stage, it will be in danger from enemies which will attempt to force their way into the cocoon, nor that the hedge of bristles will protect it from such enemies. Our only clue to an interpretation is in the slow process by which minute useful variations in the primitive instinct of spinning are accumulated through selection; and it is wonderful to see how exactly these spinning powers are adapted to the particular life-conditions of individual species. Thus there are several of the Saturnides whose enormous caterpillars live on large-leaved trees, and these make use of the large leaves to form a shelter for the pupa stage, spinning them together so that the cocoon is for the most part surrounded by leaf. But as the leaf might easily fall off with the weight of the pupa, they make the leaf-stalk fast to the twig from which it grows by binding the two firmly together with a broad, strong, closely-apposed silken band. Seitz relates of the largest of all these spinners, the Chinese _Attacus atlas_, that this silk sheath 'is continued to the nearest strong branch, so that it is impossible with the hand to detach the leaves that conceal an Atlas-pupa from the tree.' To be sure, this pupa weighs about eleven grammes! Since instincts vary, as well as the visible parts of an animal, a fulcrum is afforded by means of which selection can bring about all these very special adaptations to given conditions, since it always preserves for breeding the best suited variations of an already existing instinct. Any other interpretation is once more excluded.