DISCONTINUOUS VARIATION. 41 



nearly corresponded to the lower mean value of Wei- 

 don's crabs (viz., 630.32 as against 630.62). Giard 

 thinks that the dimorphism in the length of the forceps 

 of the earwig observed by Bateson can be similarly ex- 

 plained, for the short individuals appear to be infested 

 with Gregarines, and the longer ones not. He does 

 not wish to insist, however, that all dimorphism is the 

 result of parasitic influence, but merely that it may be 

 so in certain instances. It is obvious, indeed, that be- 

 tween two absolutely distinct varieties or species, and 

 between pure monomorphic forms, all intermediate 

 stages may exist. But how . are these intermediate 

 stages to be classified? When is one justified in assum- 

 ing the existence of two distinct species, and when of 

 only one species with an increased range of variation 

 and perhaps a tendency to split up? To overcome this 

 difficulty Davenport and Blankinship * have suggested 

 that in order to decide in any given case whether we are 

 dealing with two or more confluent species, or only with 

 varieties, the following procedure should be adopted: 

 First of all one should determine the most distinctive 

 character of the members of the group, and after mak- 

 ing a series of measurements in respect of this char- 

 acter, plot out a curve showing the relative frequency 

 of occurrence of each measurement. Supposing that 

 in this way a double humped curve is obtained, then 

 the degree, of isolation of the constituent races is esti- 

 mated by measuring the depth of the depression be- 

 tween the two humps, from the level of the maximum 

 of the lower hump. This depth may be expressed as a 

 percentage on the length of the maximum ordinate, or 

 * Science, N. S. vol. vii. p. 685, 1898. 



