THE ORIGIN OF SPECIES 377 



evolution and some are not. All the evidence indicates that the 

 effective variations are germinal and not somatic. Changes 

 arising in the soma acquired characters are unable 

 so specifically to modify the germ that they are 'born again' 

 and evolution must be brought about by the evolution of the 

 germ itself. Accordingly selection must operate to eliminate 

 the 'unfit' germ plasm rather than the unfit soma, though as 

 a matter of fact the fitness of an individual is determined 

 largely by its somatic characters. This apparently is the 

 crux of the matter and presents a complication of the mental 

 picture of the operations of selection which did not exist 

 when we thought of soma producing germ and germ produc- 

 ing soma again. Since individuals frequently belie their 

 germinal condition what they will pass on to their progeny 

 selection has, so to speak, a more devious though not less 

 sure path. 



Secondly, how does the germ plasm change? Developed 

 characters are the result of the activities of one or more genes. 

 Of course, a gene is not a character. It is not even an unde- 

 veloped character. Characters in many cases arise from the 

 interaction of several genes, though, since one gene deter- 

 mines whether the gene complex will give rise to a certain 

 character, it is really the determining factor for example, 

 the so-called sex gene on the so-called sex chromosome. Such 

 being the case, characters may be changed by alterations of 

 the gene complex. (Cf. p. 298.) This may be from the 

 influence of changes within the soma itself or from the en- 

 vironment of the organism, but here particularly we are 

 on debated ground. On the whole, it may be said that muta- 

 tions germinal changes other than those arising from 

 recombinations seem to be infrequent compared with 

 non-heritable changes of somatic origin. 



These facts from genetics, taken in connection with 



