V. GRANT 49 



the ordinate the number of flowers in the inflorescence and the ab- 

 scissa the length of the corolla. The absence of a discontinuity 

 in the variation patterns of the different species is apparent. 

 Similar results are obtained when other characters are used. 



Relatively few biologists subscribe any longer to the old idea 

 that species can be defined by the sterility of hybrids. There are 

 too many cases of good species which can produce fertile hybrids 

 under laboratory or garden conditions although they remain 

 amply distinct in nature. Furthermore, all degrees of sterility 

 exist so that the distinction between "intersterile" and "inter- 

 fertile" is frequently an arbitrary one. The existence of intra- 

 specific sterility phenomena also makes any general application 

 of the sterility criterion of species hazardous. Nevertheless, there 

 can be no doubt that sterility barriers comprise some of the most 

 effective mechanisms which separate species, and their presence 

 must be weighed carefully in any attempt to devise a classifica- 

 tion of species which accords with the realities of nature. 



The most essential property of a species is its ability to main- 

 tain its distinctive features, or rather the gene patterns which 

 determine those features, from generation to generation in a 

 community of organisms. Sympatry is the final test of species 

 status. Two sympatric populations will normally be regarded as 

 belonging to separate species. It by no means follows as an auto- 

 matic rule, however, that related allopatric populations are neces- 

 sarily conspecific. 



We have several pairs of related allopatric species in Gilia. 

 Gilia tricolor and G. angelensis, for example, belong to the same 

 species group and occupy similar ecological habitats in contigu- 

 ous geographical areas ( Grant, 1952 ) . It was at first assumed as a 

 working hypothesis that they were members of the same poly- 

 typic species. This hypothesis had to be abandoned when it was 

 learned that the two forms are completely intersterile in the 

 experimental garden and probably possess different genomes as 

 well. The intersterility of G. tricolor and G. angelensis stands in 

 marked contrast to the absence of any known intraspecific steril- 

 ity within either entity. The facts are very similar in the case of 

 G. millefoliata and G. laciniata; G. splendens and G. leptalea; and 



