Xo. 504] 



THE FIXATION OF CHARACTER 



709 



distribution and is much more successful than most seed 

 plants. The great majority of fundamental distinctions 

 and conservative characters seem to be of as little sur- 

 vival value as is pentamery to the echinoderms or the 

 presence or absence of stipules to any family of the 

 dicotyledons. 



The theory of natural selection, at least in its extreme 

 form, can not therefore well be regarded as a satisfactory 

 explanation for structural conservatism. Darwin himself 

 frequently called attention to the fact that "the physio- 

 logical importance of an organ does not determine its 

 classificatory value'' 3 and cited many examples of organs 

 or characters obviously insignificant or useless which are 

 nevertheless very constant and of great value in deter- 

 mining relationships. Darwin voices the general defence 

 of selectionists on this point, however, when he states 

 that "the importance, for classification, of trifling char- 

 acters mainly depends on their being correlated with 

 other characters of more or less importance." 4 



If the conservatism of a useless character depends on 

 its correlation with one of great functional value which is 

 continually preserved through the action of natural selec- 

 tion, it ought to be possible to discover this essential 

 character and to use it in classification. A search for 

 such universal and vitally important distinctions, how- 

 ever, is strangely fruitless, for in most families the only 

 characters which we can definitely point out as common 

 to all the individuals are precisely those which seem 

 utterly insignificant for survival. This fact becomes in- 

 creasingly obvious as we consider still broader groups 

 where the number of common characters becomes smaller 

 and smaller until there are but one or two features of 

 absolute diagnostic value. The two great divisions of the 

 amniotes, for example, the Sauropsida on the one hand 

 and the Mammalia on the other, can be rigidly distin- 



