158 THE VARIATION OF ANIMALS IN NATURE 



is the only way in which new species arise. The 

 permanent isolation of geographical races must be 

 established in much the same way as permanent 

 isolation between species inhabiting the same area. 

 (3) The establishment of isolation is probably due to the 

 interaction of a number of different factors, none 

 of which would be effective by itself. 



The third generalisation is the one which appears to be 

 most useful. The study of geographical races is not likely to 

 be helpful, except in the narrow zone where two races meet, 

 and not here if, as often happens, the races interbreed at this 

 point. It is rather in the study of biological races of animals 

 that our hope lies. These closely allied taxonomic groups, 

 differing more in habits than in structure, show us where the 

 fission of species is just beginning. Since the races often occur 

 together without much intercrossing, isolation must have been 

 developed and may be analysed with some likelihood of 

 reaching definite conclusions (cf. Thorpe, 1929, 1930). In 

 1896 Jordan was able to make out a case for the theory that 

 permanent isolation would be developed only between species 

 already geographically isolated. It seemed at that date 

 that a difference sufficient to isolate two forms could not arise 

 at one step without a new species also arising suddenly, and 

 this appeared to contradict the widely accepted generalisation 

 that specific change was gradual. The more recent study of 

 biological races demonstrates that these a priori arguments 

 are unsound. Whether it seems probable or not, biological 

 races more or less isolated from one another do appear to 

 arise from an originally homogeneous species. 



The occurrence of local breakdowns in a normally effective 

 isolatory mechanism also suggests the complex nature of the 

 process. Delcourt's (1909) study of Notonecta shows that 

 species isolated in part of their range may interbreed in a small 

 area, von Schweppenburg (1924) records the same thing 

 in Passer domesticas and P. hispaniolensis, and Tutt (1909, 19 10) 

 in Agriades thetis and A. coridon. If we take an imaginary 

 example in which two species are separated almost completely 

 by the time of their breeding season, and if we suppose that 

 the onset of the breeding season partly depends on climate, 

 but that the two species do not react to climate in precisely 



