1 4 o THE VARIATION OF ANIMALS IN NATURE 



extinct, and will not contribute to the evolution of the group. 

 It seems doubtful whether this principle is very helpful, except 

 in comparing fairly similar forms, and it can scarcely explain 

 the anomalies of differentiation in archipelagoes, etc. 



Apparently much more importance must be ascribed to 

 innate differences in species, which we have to allow for but 

 cannot at present explain. When once we admit that some 

 species may have an innate tendency to unusual variability, we 

 make it very difficult to study the effects of isolation. A high 

 degree of innate variability will increase the chance that any 

 isolated parts of a population will have a composition differing 

 from the norm of the species. If permanent isolation depends 

 on the cumulative effect of various small accidental dishar- 

 monies, then geographically isolated populations of a variable 

 species may be expected to reach a state of permanent isolation 

 more quickly. 



Later in this chapter there is a discussion of whether 

 permanent isolation is most often gained by the accumulation 

 of numerous small differences rather than by one substantial 

 change. It can be shown that relatively slight differences 

 sometimes maintain a significant degree of isolation, and it is 

 much easier to imagine the evolution of the isolatory mechanism 

 by several small steps than by one big step. In larger animals, 

 on the other hand, geographical isolation may be very im- 

 portant, but, as body-size is reduced, it becomes progressively 

 less significant. This is probably a natural result of large 

 animals x wandering over extensive areas, which often include 

 numerous types of habitat, while smaller animals can maintain 

 themselves in a population of efficient size, within the much 

 smaller limits of perhaps a single restricted habitat. 



Our argument, then, runs as follows : in large animals 

 geographical isolation is probably an important factor, though 

 the degree to which the inherent variability of the species is 

 developed is no less important. Unless a population changes 

 enough to become permanently isolated, it will be liable to be 

 recombined with the parent stock by subsequent topographical 

 changes. We do not know how long it takes to evolve per- 

 manent isolation, but, at least in some species, evolution is 



1 von Schweppenburg (1924, p. 143) says he knows of no clear case in birds 

 in which subspecies are in the least likely to have arisen in one place. Where 

 there is considerable overlap it is likely to have arisen by spread since the races 

 originated. 



