82 THE VARIATION OF ANIMALS IN NATURE 



facility. A great deal of work has been done on Protozoa 

 and insects, a less amount on Mollusca, and still less on 

 birds and mammals (wild), fishes and Crustacea. Among the 

 other groups our knowledge is defective. No general inferences 

 can be made from these results as to what characters are 

 especially prone to be heritable nor as to the likely incidence 

 of such variation in the vast number of described species. As 

 regards the heritability of the characters which distinguish 

 local races it is still more difficult to generalise. From the 

 work of Sumner (mammals), Schmidt (fishes), Harrison, 

 Tower, Goldschmidt (insects) and Woltereck (Crustacea) it is 

 evident that some races tend to breed true, though the racial 

 complex is dissociated and broken up on crossing. 



4. There are certain special types of local variation which 

 are more properly considered in relation to the causes which 

 are presumed to have encouraged or given rise to them. 

 Prominent amongst these is the occurrence of special insular 

 forms. These include not only normal divergences from the 

 adjacent continental forms, but also certain abnormalities, 

 such as melanic, dwarf and giant types, which have repeatedly 

 been noted as characteristic of insular faunas (see Chapter V). 



5. In the intensive study of local variation involving the 

 comparison of distinct races or subspecies there is sometimes 

 available data for estimating the relative size of local groups. 

 Such data have often given us the impression that in a group 

 of closely related groups (races or subspecies) one particular 

 group will tend to occupy a larger area or otherwise tend to 

 predominate over the others. This is usually recognised in 

 taxonomy as the typical form. The means for judging how 

 frequent this predominance of one or more forms within a 

 species may be, are not very extensive, as the appropriate 

 data are not often given. If it is, as we suspect, of general 

 occurrence, it is a phenomenon of some consequence and 

 might conceivably be adduced as evidence for the operation 

 of selection. Instances are seen in the distribution of the 

 subspecies of American marmots (Howell, 191 5) and Glaucomys 

 {id. 1 91 8) and also in the races ofPartula (Crampton, 1 916-1932). 



It may be pointed out here that the suggestion put 

 forward by Willis (1922) that the size of the area occupied by 

 a species is an index of its age (more recent species occupy- 

 ing smaller areas) has been in some measure confirmed for 



