270 



THE AMERICAN NATURALIST 



[Vol. LI 



ural selection. It proposes an entirely new explanation 

 of mimicry in the following terms : 



liMvc playcMl It-. |inil. hut vww with ;i <liffererice. Instead of building 



iiiacpeiideiilly. . '. ' Wliv vnrintiniis on the pnrt of one species^should 

 hem- a stron- rrseu.bhmce to oilier, an.l often distantly related, species 

 is another (|iiesiioii. . . . The oeenrrenee ol ii)iiiiotic resemblances is the 

 expression of the fact that color pallei'ii is (lei)cndent upon definite 



is to be expected tliat certain of the color patterns exhibited by one 



I'lioii t'xniiiitiiit ion these statements appear to embody 

 a runiial cxplntiatioii of tlie facts to wliich it is difficuU 

 to tak<' ex. •option, llonce it must be admitted that this is 

 l'oi li;ip- tiio u-oal lowai-d which with regard to this prob- 

 Iriii iiatiit ali-t> lia\ (' lieen working for more than half a 

 ('( iitufy. r,nl while his hypothesis may bo correct, its 

 author's reasons for deeming it so seem quite insutlicient. 



The chief points of support on which it rests are the 

 following, which are not arranged in the order of im- 

 portance assigned to them: 



1. The difficulty of finding the appropriate enemy 

 which shall exercise the discrimination postulated by cur- 

 rent hypotheses. 



•1. Tlie alleged fact, matliematically demonstrated, that 

 i-critD'ocal miiiiiofx- lioiweeii two species can not be estab- 

 lish,. d <elcetioii of a lot!-- >eries of slight variations. 



."!. The t heoreti<-al (liliii'iilt> of the initial variation in 

 cn^cv otlii r lhati that aitox-c, for it seems reasonable that 

 if the aiice>lral Ixpe- I'foiii whicli mimic and model are 

 (h'tivcMl wet-e in the heginning veiy unlike in appearance 



