NATURAL SELECTION 141 



ange of variation, the limits being set by the path of evolution 

 : ollowed hitherto, the organisation already attained. Within 

 :hese limits there may be an infinite number of variations just as 

 i botanical systematist may subdivide his families into almost 

 inlimited species and sub-species. This does not alter the fact 

 :hat heredity keeps a tight hand even on the most variable 

 groups limiting strictly their methods of adaptation. An 

 example will make this point clear. In the wings of birds, if 

 we consider size, structure and colour, there is no end to the 

 variations we may find. Even the legs show peculiarities 

 which are numberless if the minutest difference is considered 

 worthy of notice. But supposing that locomotion, for instance, 

 s the burning question, there are only two means possible for 

 Dirds : there is flight by means of wings (avian wings, that is), 

 md there is a biped gait. Wings involving the two pairs of 

 imbs, such as a bat possesses, or a quadruped gait, are out of the 

 question. In fact, if the conditions become more stringent, birds 

 nust adapt themselves by improving along the lines on which 

 :hey have already advanced so far. To specialise on different 

 lines they must first revert to a more primitive form. But such 

 reversion requires a slackening of the stress of competition : such 

 is seclusion on an island where there are few rivals. Not a few 

 Crustaceans have resigned their free-swimming life and have 

 affixed themselves to rocks, or as parasites have lost nearly all 

 their distinctive characters. But birds have advanced too far to 

 resort to such slums and back ways. For them, except under 

 very exceptional conditions, further adaptation must mean further 

 perfecting of their existing powers. Widely different forms may 

 length arise, but it can only be through long continuance along 



owly diverging lines. 

 All that I have said on this subject is only an amplification of 



hat has become a Darwinian commonplace : the offspring tend 

 be like their parents but not exactly like. The principle of 



eredity governs the variations no less than the resemblances. 

 a breed of horses tends to vary, the variations are equine and 



ot bovine. It would be strange indeed if the organism, after 



Dughly but truly recapitulating its phylogeny and going through 



