would carry on the normal or abnormal unmixed character into another generation. But if the possession of six fingers conveyed an undoubted advantage, and if natural selection did really act in civilised man as regards the transmission of morphological characters, then a stable variety (_Homo sapiens hexadactylus_, let us say) might be produced by its agency. The mutations which we consider in the investigation of the inheritance of alternating characters are therefore just as much the material for natural selections as the mutations which occur among the ordinary variations displayed by organisms in general: but since only one or two characters appear to be subject to this mode of transmission, the process would be so slow as to be inadmissible as an exclusive cause of evolution. If we assume that de Vries’ mutations are the material on which selection works, this difficulty is immediately removed, for we now have to deal with _groups_ of stable deviations: not one or two, but _all_ the characters of the organism appear to share in the mutability. But another difficulty now arises. A species of plant or animal may have got along very well with its ordinary structural endowment, and then a number of individuals begin to mutate. Some of the deviations from the specific type may be of real advantage, but others may not: we can, indeed, imagine an in-co-ordination between the mutating parts or organs which would be fatal to the animal; on the other hand, there might be complete co-ordination, with the result that great advantage might be conferred upon the individual. It is easy to see how co-ordination of mutating parts is absolutely essential. An animal which preserves its existence by successful avoidance of its enemies would not be greatly benefited by a more transparent crystalline lens if the vitreous humour of its eye were slightly opaque; and even if all the parts of the eye were perfectly co-ordinated, increased acuity of vision would not greatly help it if its limbs were not able to respond all the more quickly to the more acute sensation. Un-co-ordinated mutations would therefore tend to become eliminated, while co-ordinated ones would become selected and would become the characters of new species.