NATURAL SELECTION 279 



build mud cells and stop their entrances with spider's web, it 

 may need certain specialisations of structure, but we cannot 

 see what advantage there was in beginning to build this type 

 of nest. 



We will now proceed to a more general consideration of 

 the problem. One important preliminary reservation is neces- 

 sary. We have spoken of structures or habits of no known use. 

 Our knowledge of the details of the lives of most animals is 

 still so small that it is quite legitimate to assume that a good 

 many apparently useless characters will be found to have 

 some function. Again, it is a well-known principle of genetics 

 that many hereditary units have multiple effects, and it 

 is possible that some of the useless structural differences 

 employed in the separation of species are merely ' indicators ' 

 of important physiological differences which may be highly 

 adaptive (cf. p. 208). But there is a point beyond which it is 

 unprofitable to go in assuming that either a use or a corre- 

 lation with an adaptation will be discovered, and, when we 

 find that probably more than half the characters defining 

 families and probably at least 90 per cent, of the characters 

 defining genera and species not only are not proved to be 

 adaptive but have no known use at all, the assumption that 

 Natural Selection has been the main agent in the evolution 

 of natural populations is too comprehensive to help us very 

 far. To be valuable as a working hypothesis a theory 

 should ' work ' in not less than half the cases to which it is 

 applied. 



The next point for consideration is the number of in- 

 stances known in which characters separating species are 

 related to differences in the life-history. This raises a question 

 not very easy to answer, chiefly because the limits of many 

 genera are still uncertain, and what one author would call a 

 generic, another would call a specific character. This difficulty 

 is to some extent avoided if we consider only species which are 

 evidently quite closely allied. Even when we have shown 

 that a use is made of a structure, we have to prove that the 

 use is adaptive. For convenience we shall consider the subject 

 under two headings : (a) Differences in colour ; (b) Differ- 

 ences in structure. 



(a) Differences in colour. — Much of the matter relevant here 

 has already been discussed (p. 232 and foil.) in connection 



