NATURAL SELECTION 251 



on and of investigating the extent of their discrimination. 

 Further, such examination must be made on a really large 

 scale to have any significance, in view of the great variation in 

 the habits of many predators. Again, the investigation of the 

 whole predacious fauna is very desirable ; if only a small part 

 of the fauna is studied, it may give quite a wrong idea of the 

 degree to which any particular group is attacked. 



The great extent to which certain groups usually supposed 

 to be distasteful are preyed upon is rather surprising, and can- 

 not but make one hesitate (without further evidence) to treat 

 them as specially protected. This is particularly the case in 

 the Hemiptera, where the malodorous Pentatomidae seem to 

 be much eaten. In other cases, as in the Hymenoptera, where 

 only one sex is protected by a sting, the data are not sufficiently 

 detailed to allow any conclusion to be drawn. The small 

 extent to which butterflies appear to be attacked is rather 

 remarkable, but may partly be due to the difficulty of identi- 

 fying their fragments. Even though a selective attack consti- 

 tuting a very small part of the total of predation might lead to 

 important evolutionary changes, we cannot but feel that the 

 degree of attack recorded (if it is not really deceptively low) 

 is minimal compared with the enormous changes that such 

 attacks are sometimes supposed to have brought about. 



(2) Mimicry. — The theory of mimicry is of high importance 

 in the selectionist argument, for two reasons : the large 

 amount and varied nature of the available data, and the 

 fact emphasised by Fisher (1930, p. 146) that if the theory of 

 mimicry is mainly true, then we appear to have a long series 

 of cases in which characters either actually specific or sub- 

 specific, or of the same status as characters specific in other 

 groups, are of adaptive value. 



Mimicry in its technical sense implies convergent resem- 

 blance in colour (and often in shape, habits and habitat) 

 between two animals, one of which (' Batesian mimicry ') or 

 both (' Miillerian mimicry ') are in some way protected or 

 advantaged by the resemblance. The number of established 

 cases of such convergence is now very large, and most of the 

 chief insect and arachnid groups contain typical examples of 

 the phenomenon. It is probably most plentifully seen in the 

 Lepidoptera, Hymenoptera and Diptera. The degree of 

 convergence and the number of species involved in the case of 



