In the first place, it will be evident from what has just been said, that such a piece of machinery as is concerned in even the simplest reflex action cannot have occurred in any considerable number of individuals of a species, _when it first began to be constructed_. On the contrary, if its _origin_ were dependent on congenital variations alone, the needful co-adaptation of parts which it requires can scarcely have happened to occur in more than a very small percentage of cases--even if it be held conceivable that by such means alone it should ever have occurred at all. Hence, instead of preservation and subsequent improvement having taken place _in consequence of_ free intercrossing among all individuals of the species (as in the cases of protective colouring, &c., where adaptation has no reference to any mechanical co-adaptation of parts), they must have taken place _in spite of_ such intercrossing. In the second place, adaptations due to organic machineries of this kind differ in another all-important respect from those due to a summation of adaptive characters which are already present and already varying round a specific mean. The latter depend for their summation upon the fact--not merely, as just stated, that they are already present, already varying round a specific mean, and therefore owe their progressive evolution to free intercrossing, but also--_that they admit of very different degrees of adaptation_. It is only because the degree of adaptation in generation B is superior to that in generation A that _gradual improvement_ in respect of adaptation is here possible. In the case of protective resemblance, for example, a very imperfect and merely accidental resemblance to a leaf, to another insect, &c., may at the first start have conferred a sufficient degree of adaptive imitation to count for something in the struggle for life; and, if so, the basis would be given for a progressive building up by natural selection of structures and colours in ever-advancing degrees of adaptive resemblance. There is here no necessity to suppose--nor in point of fact is it ever supposed, since the supposition would involve nothing short of a miracle--that such extreme perfection in this respect as we now so frequently admire has originated suddenly in a single generation, as a collective variation of a congenital kind affecting simultaneously a large proportional number of individuals. But in the case of a reflex mechanism--which may involve even greater marvels of adaptive adjustment, and _all_ the parts of which must occur in the same _individuals_ to be of any use--it _is_ necessary to suppose some such sudden and collective origin in some very high degree of efficiency, if natural selection has been the only principle concerned in afterwards perfecting the mechanism. For it is self-evident that a reflex action, from its very nature, cannot admit of any great differences in its degrees of adaptation: if it is to work at all, so as to count for