NATURAL SELECTION 235 



more broken it is rarer. That we should find close 

 resemblance mainly when the background is very 

 homogeneous is somewhat important. 



To what extent animals fail to develop this resemblance 

 is very hard to estimate. Roosevelt (191 1, p. 171) states that 

 half the mammals in the United States either are not protec- 

 tively coloured or owe their safety to particular habits. This 

 estimate must be largely guess-work. The question is compli- 

 cated by our lack of knowledge as to whether the habits and 

 postures of animals are appropriate to the situations in which 

 their colours might be advantageous {cf. Roosevelt and Heller, 

 19 15). Moreover, an animal may seem to be ' protectively' 

 coloured or modelled vis-a-vis a particular landscape and yet 

 range over a variety of backgrounds. Thus di Cesnola and 

 Poulton and Saunders (p. 202) claim to have shown that 

 certain insects are protected by their colour when on a given 

 type of background. As we have pointed out, the colours, etc., 

 could be regarded as adaptive only if it could be shown that 

 they are correlated with the habit of keeping to a particular 

 background. 



We have introduced this subject here because it is one of 

 the standard cases adduced in favour of Natural Selection. 

 We are not unmindful that in many cases an alternative 

 explanation is possible. A great deal of the homochromatic 

 resemblances might be due to individual accommodation, or 

 even to the inherited effects of such accommodation. We know, 

 however, of no evidence that such accommodation ever occurs 

 in the higher vertebrates, and this explanation ought to be 

 sought only in particular cases (insects) in which there is definite 

 experimental evidence. Finally, we cannot believe that such 

 causes play any part in producing assimilative resemblances. 



Nevertheless, while we incline a priori to a selective explana- 

 tion, we cannot but admit that the difficulty of establishing a 

 solid proof of this is very considerable. The mere citation of 

 innumerable cases of resemblance is plainly not enough. What 

 we need is direct evidence as to how the resemblances have 

 arisen, and that is very inadequate. It is for this reason indeed 

 that we are obliged to neglect the bulk of the remarkable cases 

 of assimilation and some classical cases of homochromy such as 

 that of the flatfishes, and fall back on certain closely studied 

 instances of simple homochromy. 



