NATURAL SELECTION 291 



failure to show that all the species in a genus are distinguished 

 by adaptations. Usually a pair of species are picked out and 

 contrasted and the other species are left out of account. 



On the whole this type of evidence does not carry very much 

 weight. At the most one would say that two cases (Lutz, 1908 ; 

 Boettger, 1921) are suggestive that sections of a population 

 may be adaptively differentiated. Against this very incon- 

 clusive evidence one has to set an enormous array of instances 

 of species and subspecies which are tolerably well known and 

 for the structural differentia of which no adaptive explanation 

 is available. Particular attention is directed to those intensive 

 studies of racial diversity (Crampton, Gulick, etc.) in which a 

 high degree of local differentiation is found amid uniform 

 environmental and bionomic conditions. This is particularly 

 well seen in Crampton's Partulas of the Society Islands, where 

 we have ample evidence of the origin of clearly differentiated 

 local groups amid uniform conditions. Isolation coupled with 

 rapid mutation seems to have played the major part in pro- 

 moting divergence. 



The number of such examples could probably be consider- 

 ably extended, more especially by admitting less closely 

 allied pairs of species, though in the latter case many authors 

 would probably regard the characters as generic rather than 

 specific. Bat, even if the above list were multiplied many 

 times over, it would still be possible to compile a parallel 

 and much longer list of specific characters of no adaptive 

 significance. We may mention the careful study by Whedon 

 (1918) of the morphology and functions of the abdomen in 

 dragon-flies. In these insects some of the most remarkable 

 structural modifications are very difficult to explain on a 

 functional basis, and, in the genus Lestes, females with very 

 different abdomen-lengths all occur together and lay their 

 eggs in the same plants, so that the theory originally propounded 

 that length of abdomen was correlated with egg-laying habits 

 seems difficult to maintain. 



The establishment of a use for structural specific characters 

 advances our problem only one stage. We have still to show 

 that the change of function has been a real advantage. This 

 question we consider in section (3) (p. 300). 



(2) The problem of secondary sexual characters. — Secondary 

 sexual characters, more especially male characters, form a very 



