in such a tumour, then, of course, the inoculating instinct of these animals could not have been developed by natural selection. But, given these two conditions, and it appears to me there is nothing very much more remarkable about an accidental correlation between the effects of a parasitic larva on a plant and the needs of that parasite, than there is between the similarly accidental correlation between a hydated parasite and the nutrition furnished to it by the tissues of a warm-blooded animal. Doubtless the case of galls is somewhat more remarkable, inasmuch as the morbid growth of the plant has more concern in the correlation--being, in many instances, a more specialized structure on the part of a host than occurs anywhere else, either in the animal or vegetable world. But here I may suggest that although natural selection cannot have acted upon the plant directly, so as to have produced galls ever better and better adapted to the needs of the insect, it may have so acted upon the plants indirectly _though the insects_. For it may very well have been that natural selection would ever tend to preserve those individual insects, the quality of whose emanations tended to produce the form of galls best suited to nourish the insect progeny; and thus the character of these pathological growths may have become ever better and better adapted to the needs of the insects. Lastly, looking to the enormous number of relations and inter-relations between all organic species, it is scarcely to be wondered at that even so extraordinary an instance of correlation as this should have arisen thus by accident, and then have been perfected by such an _indirect_ agency of natural selection as is here suggested[37]. [37] Note B. * * * * *