In this case the beginning of the action cannot lie in a voluntary action, for the insect cannot know what it means to be caught and killed, and the same is true of innumerable still lower animal forms, the hermit-crabs and the Serpulids, which withdraw with lightning speed into their houses, and so forth. It seems to me important theoretically, that the same action can be liberated at one time by the will, at another by the inborn instinct-mechanism. In both cases quite similar association-changes in the nerve-centres must lie at the root of the animal's action, but in the first case these are developed only in the course of the individual life by exercise, while in the second they are inborn. In the former, they are confined to the individual, and must be acquired in each generation by imitation of older individuals (tradition) and by inference from experience, in the latter they are inherited as a stable character of the species. It has been maintained by many that the origin of instincts through processes of selection is not conceivable, because it is improbable that the appropriate variations in the nervous system, which are necessary for the selective establishment of the relevant brain-mechanism, should occur fortuitously. But this is an objection directed against the principle of selection itself, and one which points, I think, to an incompleteness in it, as it was understood by Darwin and Wallace. The same objection can be made to every adaptation of an organ through natural selection; it is always doubtful whether the useful variations will present themselves, as long as they are due solely to chance, as the discoverers of the selective principle assumed. We shall attempt later to fill up this gap in the theory, but, in the meantime, I should like to point out that the process of selection offers the only possible explanation of the origin of instincts, since their origin through modifications of voluntary actions into instinctive actions, with subsequent transmission of the instinct-mechanism due to exercise in the individual life, has been shown to be untenable. If any one is still unconvinced of this, I can only refer to the cases we have already discussed of instincts which are only exercised once in a lifetime, since, in these, the only factor that can transform a voluntary action into an instinctive one is absent, namely, the frequent repetition of the action. In this case, if any explanation is to be attempted at all, it can only be through natural selection, and as we have assumed once for all that our world does admit of explanation, we may say, _these instincts have arisen through natural selection_.