Variation and Natural Selection. 109 



advance in the two groups, though analogous, would not be 

 identical, and divergence would thus be possible under 

 practically similar conditions of life. 



In his observations on the terrestrial molluscs of the 

 Sandwich Islands, Mr. Gulick notes that different forms 

 are found in districts which present essentially the same 

 environment, and that there is no greater divergence when 

 the climatic conditions are dissimilar than there is when 

 those conditions are similar. As before noticed, the degree 

 of divergence is, roughly speaking, directly as the distance 

 the varietal forms are apart. Again, Darwin notes that 

 the climate and environment in the several islands of the 

 Galapagos group are much the same, though each island 

 has a somewhat divergent fauna and flora. These facts 

 lend countenance to the view that divergence can and does 

 occur under similar conditions of life, if there be isolation. 

 They seem, also, so far as they go, to negative the view - 

 that the species is moulded directly by the external con- 

 ditions. For, if this factor were powerful, it would over- 

 ride the effects of experimental combination of characters 

 when the conditions were similar, and would give rise 

 to well-marked varietal forms when the conditions were 

 diverse. 



If we admit preferential breeding as a segregation-factor 

 (and arising out of it sexual selection, in a modified form, 

 as a determining one in the evolution of the plumage of 

 male birds), it is evident that the standard of recognition- 

 marks can only be maintained by a uniformity of preference 

 or taste. Still, the uniformity is not likely to be absolute. 

 In this matter, as in others, variations will occur, and 

 after the lapse of a thousand generations, in which elimina- 

 tion has been steadily at work, it is hardly probable that 

 the recognition standard would remain absolutely un- 

 changed. For, though there may not be any direct 

 elimination in this particular respect, there might well be 

 colour-eliminations in other (e.g. protective) respects, and 

 the mental nature would not remain quite unchanged. 

 Moreover, we know that secondary sexual characters are 



