MODIFICATION OF SPECIES 359 



ter of the mutations with which it starts. Some students of evohition 

 have assumed that mutations are of every conceivable sort, just as a 

 needle thrown on the floor may eventually, if thrown often enough, 

 point in every horizontal direction. This seems an unreasonable assump- 

 tion because, if genes are chemical in their nature, they should be no more 

 free to enter into unlimited reactions than other substances are. Chem- 

 ical substances are restricted to a certain range of reactions by the 

 structure of their molecules. Furthermore, in organisms which, like 

 Drosophila, have produced the greatest numbers of observed mutations, 

 there is not so much variety among the mutations as a purely random 

 determination of them should produce. They are too much alike, and 

 some of them occur too often, to be the result of chance alone. It is 

 more likely that each gene is capable of mutating in certain ways, and 

 only in those ways. If this is correct, a species can evolve along any 

 line which its possible mutations provide, but along no other. From 

 these possible lines something has to choose. 



With respect to the combinations of genes that result from hybrid- 

 ization within the species, chance probably plays an important role in 

 the early stages of differentiation. When certain genes are present in a 

 population in given numbers of individuals, certain combinations of 

 genes are expected to occur in calculable proportions of individuals. 

 Almost certainly, however, the expected proportion is never exactly 

 realized. The accidental meeting and pairing of individuals will usually 

 result in some small deviation from the expected result. A gene that 

 ought theoretically to occur in 25 per cent of the individuals may easily 

 happen to be in 28 per cent or only 22 per cent solely through chance. 

 Should the deviation from expectation in the next generation happen to 

 be in the same direction, the difference is accentuated. Different parts 

 of a range may thus come to be inhabited by groups of individuals which, 

 while still belonging to the same species, nevertheless have their genes in 

 different proportions. These groups may look essentially alike, especially 

 if the genes in which they differ are recessive and exist mostly in heter- 

 ozygotes; but their potentialities for the future are distinctly different. 

 Such differences tend to be preserved by lack of random mating. No 

 individual travels the whole range of its species, so that it mates with one 

 of its neighbors. When the genes become numerous enough to produce 

 many homozygotes, or if they are or become dominant, the two groups of 

 individuals show noticeable differences. 



It is believed that varieties of a species may arise and come to occupy 

 different parts of the range, entirely as a result of random wandering and 

 the accidental union and fortuitous survival of certain gene combinations. 

 Possibly even a divergence great enough to mark two separate species 

 may take place in this purely random manner. Beyond this degree of 



