NATURAL SELECTION 259 



is visible only on microscopic examination of a considerable 

 series of specimens (Richards, 1928, p. 385). Somewhat 

 similar resemblances among Oriental Papilios are mentioned 

 by Jordan (1896). The remarkable convergence in colour 

 described by Buxton (1923) in many desert animals has been 

 already discussed (p. 239). The colour convergence may, 

 in some rodents, extend to the soles of the feet. According 

 to Buxton there are considerable difficulties in regarding this 

 convergence as due to protective coloration ; on the other 

 hand, Sumner (1932) has shown that some at least of the desert 

 forms are hereditary races, so that determination by the 

 environment would raise certain theoretical difficulties. 



(4) The same argument may be applied to the mimicry 

 theory as will be applied later (pp. 275-6) to the Natural 

 Selection theory in general — viz. if it can be shown that certain 

 cases of apparent mimicry are very unlikely to be the result of 

 selection, then mimicry must in those cases have other causes, 

 and it is therefore impossible, without better evidence than is 

 usually available, to say what cause has been active in a par- 

 ticular case. Such an admission would make it easy to main- 

 tain that colour resemblances are due to selection, where the 

 evidence for such selection is strongest, while allowing the less 

 well-established cases to be left sub judice. This argument 

 would appear to be applicable, even if no alternative to the 

 selective explanation can be directly demonstrated. 



One of the types of mimetic association least easy to explain 

 on the selection hypothesis is found amongst the Hymenoptera, 

 e.g. in the Hawaiian wasps (chiefly Eumeninae) described by 

 Perkins (191 2), and in the Vespids of S. America. In Hawaii, 

 Perkins shows that wasps fall into a number of very distinct 

 colour groups which cut right across groupings based on 

 structural characters. A few of the colour groups are more 

 or less confined to particular islands, but others are found on 

 several islands and most islands have representatives of more 

 than one group. At the present day no birds are known to 

 prey on these Hymenoptera, though admittedly man has 

 greatly altered the fauna in recent years. It has been main- 

 tained (Poulton, 1 91 2) that these colour groupings are Mul- 

 lerian associations ; but it is very difficult on this hypothesis 

 to see why so many different groups should be formed in 

 islands of relatively small size. This difficulty is accentuated 



