For the present I need only draw the conclusion that one and the same caterpillar may exhibit the initial stages of both, and that it depends on the manner in which these marking elements are _intensified_ and _combined_ by natural selection whether whitish longitudinal or oblique stripes should result. In this case then the "useful variations" were actually "always there," and we see that in the same group of Lepidoptera, e.g. species of Sphingidae, evolution has occurred in both directions according to whether the form lived among grass or on broad leaves with oblique lateral veins, and we can observe even now that the species with oblique stripes have longitudinal stripes when young, that is to say, while the stripes have no biological significance. The white places in the skin which gave rise, probably first as small spots, to this protective marking could be combined in one way or another according to the requirements of the species. They must therefore either have possessed selection-value from the first, or, if this was not the case at their earliest occurrence, there must have been _some other factors_ which raised them to the point of selection-value. I shall return to this in discussing germinal selection. But the case may be followed still farther, and leads us to the same alternative on a still more secure basis. Many years ago I observed in caterpillars of _Smerinthus populi_ (the poplar hawk-moth), which also possess white oblique stripes, that certain individuals showed _red spots_ above these stripes; these spots occurred only on certain segments, and never flowed together to form continuous stripes. In another species (_Smerinthus tiliae_) similar blood-red spots unite to form a line-like coloured seam in the last stage of larval life, while in _S. ocellata_ rust-red spots appear in individual caterpillars, but more rarely than in _S. populi_, and they show no tendency to flow together.