With respect to reflex actions, therefore, it is desirable to begin by pointing out how widely the adaptations which they involve differ from those where no manufacture, so to speak, of special machinery is required. Thus, it is easy to understand how natural selection alone is capable of gradually accumulating congenital variations in the direction of protective colouring; of mimicry; of general size, form, mutual correlation of parts as connected with superior strength, fleetness, agility, &c.; of greater or less development of particular parts, such as legs, wings, tails, &c. For in all such cases the adaptation which is in process of accumulation is from its very commencement and throughout each of its subsequent stages, of _use_ in the struggle for existence. And inasmuch as all the individuals of each successive generation vary round the specific mean which characterized the preceding generation, there will always be a sufficient number of individuals which present congenital variations of the kind required for natural selection to seize upon, without danger of their being swamped by free intercrossing--as Mr. Wallace has very ably shown in his _Darwinism_. But this law of averages can apply only to cases where single structures--or a single group of correlated structures--are already present, and already varying round a specific mean. The case is quite different where a _co-ordination_ of structures is required for the performance of a _previously non-existent_ reflex action. For some, at least, of these structures must be _new_, as must also be the function which all of them first conspire to perform. Therefore, neither the new elements of structure, nor the new combination of structures, can have been previously given as varying round a specific mean. On the contrary, a very definite piece of machinery, consisting of many co-ordinated parts, must somehow or other be originated in a high degree of working efficiency, before it can be capable of answering its purpose in the prompt performance of a particular action under particular circumstances of stimulation. Lastly, such pieces of machinery are always of a highly delicate character, and usually involve so immensely complex a co-ordination of mutually dependent parts, that it is only a physiologist who can fully appreciate the magnitude of the distinction between "adaptations" of this kind, and "adaptations" of the kind which arise through natural selection seizing upon congenital variations as these oscillate round a specific mean. Or the whole argument may be presented in another form, under three different headings, thus:--