CH. XX] THE ORIGIN OF SPECIES 213 



being somewhat isolated, have comparatively small floras, which 

 have not, therefore, been able as yet to form very elaborate 

 plant societies suited to their various conditions, and into which, 

 therefore, a newcomer ma}^ more easily enter. In the same way, 

 the frequent, and apparently quite casual, appearance of young 

 and localised species in the great forests may be due simply to 

 the fact that the fall of a great tree has for the time so changed 

 the conditions as to give the newcomer a better chance of estab- 

 lisliing itself before the old conditions are completely restored. 

 The further out one goes, the smaller on the average does the 

 number of species j^er genus become, aiid perhaps therefore the 

 plant societies may tend to be more open. 



Whether a new form upon its aj^pearance will or will not 

 survive, will depend chiefly upon natural selection, for it will 

 at once have a struggle for existence of the most remorseless 

 kind. It will also depend appreciably upon mere chance (cf. 

 Didymocarpus and Christisonia on p. 151). A fire or a flood may 

 easily kill it out, howe\'er perfectly suited to its environment it 

 may be. 



As our object in the present work is simply to criticise some 

 of the directions in which existing theories do not seem properly 

 to meet the facts, and to suggest some directions in which it is 

 conceivable that they may be improved, there is no need to go 

 into an}^ discussion of possible causes of mutation. If, as is not 

 impossible, they depend immediately upon some chemical change 

 that has somewhere taken place, one can understand why 

 changes should be mutational, for chemical change docs not 

 usually take place by continuous variation. 



On account of the insuperable difficulties in the way of evo- 

 lution by means of the natural selection of infinitesimal variations, 

 opinion has for a long time been steadily coming round to favour 

 the idea of change by mutation. Even the most enthusiastic 

 supporters of infinitesimal variation now generally begin with a 

 measurable change, impro\4ng it afterwards by the old method. 

 A recent writer of this school, for exam])le, cites a change from 

 inches to feet as an infinitesimal ^-ariation. 



The A\'ork upon Age and Area outlined in Part I provides, as 

 has already been indicated, strong arguments against infinitesimal 

 variation, and the further work gwaw in the last few chapters, 

 which seems to show that when one deals with large numbers 

 and the long run cA-olution of new genera and species, and their 

 distribution about the world, is very much a process which has 



