186 GENERAL DISCUSSION [ch. xiv 



may even be enormous, crossing the oceans, or even the equator. 

 Natural selection could not explain this by any destruction of 

 transitions, for the separations are of all sizes and in all directions. 

 The only simple explanation so far proposed is that the local 

 genera or species are due to direct mutations from the linking 

 large and widely distributed genera or species that cover the 

 places in which they occur. This of course involves the acceptance 

 of differentiation. In case xxix (p. 156) it is showTi how variety 

 in structure shows no necessary relation to variety in conditions, 

 as one would expect under natural selection. In case xxx (p. 158) 

 the difficulty is pointed out, of explaining, under natural selec- 

 tion, a very common type of distribution. Many genera show one 

 or more widely distributed species, usually very polymorphous, 

 accompanied by local endemic species of the same genus in 

 various parts of their range. The only simple explanation is that 

 put forward by Guppy and by the writer that these endemics 

 have been derived from the widely distributed species by one or 

 more mutations. The incomprehensibility of selection is further 

 developed in case xxxi (p. 161). Incasexxxii (p. 161) the incon- 

 sistency of the contention that characters are less constant the 

 less useful they are, is pointed out, and in case xxxiii (p. 162) the 

 bearings of Hooker's discovery of the constancy of the numerical 

 relation between Mono- and Di-cotyledons are pointed out, with 

 the fact that there is no monocotyledonous mode of life. Case 

 XXXIV treats of overlap of genera. 



It is clear that the tests give very strong evidence indeed in 

 favour of the theory of differentiation or divergent mutation, 

 according to which the course of evolution is in the opposite 

 direction to what has hitherto been supposed, and by mutations 

 which tend to diminish as time goes on, but go in the direction 

 family — genus — species. The organism that first represents the 

 family is, of course, at the same time its first genus and species, 

 but these are of different rank from genera and species in a 

 larger family. By further mutations this will then give rise to 

 further genera and species. The first new genus formed will 

 usually be widely divergent from the parent genus of the family, 

 even if the family be quite small, e.g. of two genera only. Later 

 formations will be less and less divergent on the whole, but will 

 show some of the characters of divergence of their first parents. 

 The main lines of divergence are therefore given by the latter, 

 and later genera fill them in, as shown by a good dichotomous 

 key. As time goes on, new genera will necessarily be evolved at a 

 continually increasing rate, and each, given time enough, may 



