EVOLUTION 415 



Given any special form of life, it may spread or be 

 transferred in some manner so that groups of it become 

 subject to different environments, and this may lead to a 

 differentiation in type. The two branches would then by 

 geographical position be isolated from each other, so that 

 the two types could not cross. Such isolation may well 

 give rise to local races, but can it give rise to species ? 

 We see a progressive differential change of a complex- 

 of organs taking place — not necessarily the reproductive 

 organs — owing to selection, but this does not in itself 

 connote absolute or relative infertility. We must first 

 show that the reproductive organs or their modes of 

 functioning are correlated with the organs selected, and 

 that the differentiation here is so great that physiologically 

 or mechanically mutual fertility of the two types is 

 impossible. Isolation may account for the origin of local 

 races, but never for the origin of species unless it is 

 accompanied by a differential fertility. 



In the same manner a form of life may be differentiated 

 even in the same environment if the modal type ceases to 

 be suited to its surroundings. Thus a short femur or a 

 long radius might be suitable alternatives, but neither 

 correspond to the modal man, and the combination of the 

 two may be so infrequent as to be negligible as a material 

 for variation. If selection were so intense as to act 

 effectively in one generation, we should have a differentia- 

 tion of type in that generation, but unless this differentia- 

 tion in type were accompanied by some form of restraint 

 on intercrossing, the selection could only be periodic, i.e. 

 recurring in each generation. If crossing took place at 

 random there would be, with equal numbers in the two 

 differentiated types, 25 per cent of the first, 25 per cent 

 of the second, and 50 per cent of the mixed type. If 

 selection did not again take place, the mixed type would 

 in the second generation, with random mating and equal 

 fertility, have six times as many members as either pure 

 type, and there would further be types intermediate 

 between the mixed and either pure type four times 

 as numerous as the latter. To maintain the types 



