52 THE VARIATION OF ANIMALS IN NATURE 



divided into one or more strains differing little, if at all, 

 morphologically, but with different habits, e.g. different larval 

 food, host of parasite, etc. 



(2) Experiment has shown that, if one race is, for instance, 

 forcibly maintained on the food of another, there is at first 

 little oviposition or breeding and a heavy mortality. Often 

 a few individuals manage to perpetuate the race on the new 

 food, to which it eventually becomes adapted. This would 

 suggest at first sight that there has been a selection of a suitable 

 stock, but this interpretation breaks down (cf. (3) ). 



(3) It has been possible in several cases to take a race A, 

 normally feeding on a food a, and adapt it to the food b of 

 race B. In these circumstances it may be as difficult to make 

 the survivors of A (on b) return to a as it was to make the 

 original change. This does not fit in with the theory of strain- 

 selection, and Thorpe (1930) definitely postulates a process of 

 more or less permanent habit-modification. It is by no means 

 necessary that this change should at first be incorporated into 

 the normal hereditary mechanism as claimed by Harrison 



(1927)- 



(4) There is some evidence that there is a tendency for these 



biological strains to mate within the race and therefore to 

 stabilise their constitution. 



From the evolutionary point of view, instinct is, as we have 

 said, a particularly important subject, but unfortunately we 

 do not know how new instincts arise nor how they are in- 

 herited. When an instinct is very firmly established it is 

 naturally handed from parent to offspring like any somatic 

 character, but this is by no means necessarily the case in the 

 early stages of instinct-acquirement. In the songs of birds, 

 for instance, while there is a large hereditary element, there is 

 also much that is local and individual. Yet it is very plausible 

 to suggest that the former element was originally built up from 

 the latter without, in all cases, the actual selection of individuals 

 with a particular song-type. 



A further phenomenon which may be considered under 

 the heading of habit-formation is that of voltinism in insects. 

 The subject has recently been discussed in an interesting 

 paper by Dawson (1931), who summarises the main theories 

 and presents some very valuable experimental data (see also 

 Baumberger, 19 17). The problem is seen at its clearest in 



