136 REDUCTION OF VARIABILITY. 



"centre of stability". This is Wagner's argument. Geographic 

 barriers may be effective means of species formation. A group 

 of animals on an island, or on the other side of a wide stream 

 must have a potential variability smaller than that of the mul- 

 titude of its species. We do not need to assume that a species 

 gradually, by a slow process of natural selection became adap- 

 ted to very special conditions, but that the peculiarity which 

 makes the group fit may have been the "accidental" result of a 

 recombination of genes, and the cause which drove the first in- 

 dividuals of this sort to establish themselves in the new condit- 

 ions or perish. A species may be pure, without potential varia- 

 bility, before some of its members colonize, in such a case no 

 new species will be produced. But if it has some potential var- 

 iability left, or if its variability happens to be heightened by a 

 recent cross, a colony derived from it may soon have its own 

 genotype, and therefore its own characters, its own species 

 type. Each little American desert has its own species of 

 ground-squirell, which is relatively pure, and somewhat differ- 

 ent from all the other species. 



Of course species-differentiation is possible in allogamous or- 

 ganisms in cases, where barriers of any kind are absent. In our 

 hypothetical examples we assumed, that there was a consider- 

 able interchange of individuals going on throughout the whole 

 range covered by the group. If the conditions are such that a 

 gene, introduced at one end of the range may be transmitted in 

 a few dozen generations to individuals at the other end of the 

 range, the whole group will be one species, it will tend all the 

 time to reduce its potential variability, it will tend to assume 

 one single genotype, and consequently one single phenotype. 

 Those plants and animals will not easily form local species, lo- 

 cal forms. Examples are the English sparrow, the Norway rat. 

 But we can see that in a slow-moving animal the chance for a 

 gene, introduced somewhere by crossing to become part of the 

 eventual genotype of the group is very much greater than in 

 roving animals, with a rapid dispersal 



If every individual has on the average only one descendant, 



