THE ORIGIN OF VARIATION 51 



of the evolution of the more highly organised animals can be 

 complete that does not explain the evolution of instinct as 

 fully as that of structure. Unfortunately, this is a matter of 

 which we are largely ignorant. Instincts are less fixed than 

 structures and their heredity and modifications are much more 

 difficult to study accurately. Undoubtedly in the vertebrates 

 it becomes difficult to distinguish inherited aptitudes from 

 traditions handed directly from one generation to the next. 

 Even in Arthropods there may well be a bigger element of 

 tradition (of rather special sort) than has usually been 

 allowed. 



As an introduction to the subject we will consider the 

 predacious habits of the wasps of the family Crabronidae 

 which have been discussed by Hamm and Richards (1926). 

 Most species, in the store of dead or paralysed prey laid up 

 for their offspring, include flies, but particular species capture 

 insects of most of the more important orders. In a few species 

 members of two or more orders are mixed, while in others 

 there is great specialisation, the prey being sometimes practi- 

 cally restricted to one sex of one genus. From the present 

 point of view the most interesting species are those which, 

 while tending to specialise on one kind of prey, always capture 

 some or many individuals of a widely different systematic 

 category. Such a species is Crabro leucostomus, which always 

 captures a high proportion of Stratiomyid flies, but includes 

 also Diptera belonging to a large number of other families. 

 This habit is independent of the habitat in which the wasp 

 is nesting. The behaviour of C. leucostomus suggests rather 

 a special form of ' larval memory,' the insect having a 

 tendency to capture the food that it received during its own 

 larval life, as has been suggested by Wheeler (1923, p. 57), who 

 points out that the central nervous system is almost the only 

 larval structure not radically modified at metamorphosis. 

 Such a larval memory would not at first be hereditary in 

 the ordinary sense, but may well have become so in those 

 species which are now strictly specialised. 



The general question of biological races in Arthropods is 

 reviewed elsewhere (p. 119) and the facts need only be briefly 

 dealt with here. The most important conclusions are the 

 following : 



(1) There are numerous instances of species which are 



