NATURAL SELECTION 225 



value may have a more definite meaning when applied to 

 the population inhabiting a small part of the range, but when 

 the problem is numerically reduced to this extent the actual 

 values of mutation-rate (as distinct from survival value) and 

 population density become highly relevant. 



The small positive or negative survival values which have 

 to be arbitrarily assigned to mutants for the purpose of mathe- 

 matical calculation can have little relation to the facts of 

 nature, and we may doubt whether the predictions based on 

 them are very likely to be fulfilled. The actual course of 

 evolution appears too much determined by special circum- 

 stances to be very amenable to generalised mathematical 

 treatment. 



(c) Random mating. — Practically all speculation as to the 

 spread of mutants has been based on the assumption of random 

 mating. It is evident that nothing approaching real random 

 mating actually occurs — i.e. it is not true that within a species 

 any male is equally likely to mate with any female. On the 

 other hand, if we attempted to allow for selective mating, 

 our ignorance of the facts would force us to make very large 

 assumptions which would detract from the otherwise convincing 

 argument. It might be possible, for instance, to introduce a 

 factor relating the likelihood of mating to the distance apart 

 at which the individuals live, but of course it cannot really 

 be held that the degree of isolation would be a linear function 

 of the distance. 



In Chapter V we considered this subject and were forced 

 to conclude that permanent isolation of species depended on 

 a variety of factors working in conjunction, and in any one 

 section of the population one of the factors may have a 

 potency which it lacks elsewhere. The species itself must 

 be expected to be broken up into minor populations, 

 and much of the evidence presented in Chapter IV supports 

 this. 



If mating is not strictly at random, this will reduce the 

 effective size of the population in which any one evolutionary 

 step is proceeding. It may not diminish the power of selection 

 to spread beneficial variants, but it will make the process of 

 spread irregular and very difficult to predict, and once more 

 it is suggested that the numerical values of the mutation-rate 

 may not be so unimportant as has been supposed. 



Q 



