254 Heredity and Environment 



2. It is possible that environmental changes acting upon germ 

 cells at a sensitive period of their development may produce ger- 

 minal variations or mutations and thus give rise to new races. 

 But while this is possible there is little evidence of a satisfactory 

 nature that it actually occurs and there is no evidence that such 

 changes in hereditary constitution are reversible and that the 

 race reverts to its former state when the old environment is re- 

 stored. Such reversible changes undoubtedly occur in somatic 

 characters but they are not inherited; they are modifications of 

 development, not of heredity; they are personal fluctuations, not 

 racial mutations. Not infrequently reversion of cultivated races 

 to wild stock is due to hybridization, as in the sweet peas de- 

 scribed on pages 101-104, or in hybrids of some races of domestic 

 pigeons (p. 81). 



II. ARTIFICIAL SELECTION 



Since the beginning of historic times, and probably through 

 long prehistoric ages, breeders have followed the method of se- 

 lecting desirable individuals for propagating their stock. There 

 can be no doubt that almost all that man has done in the produc- 

 tion of domestic animals and cultivated plants has been accom- 

 plished by this process of selection. 



Hozv Has Selection Acted? Darzvin's Views. Until very 

 recently it was generally believed that continued selection of in- 

 dividuals which showed desirable characters gradually led to the 

 improvement of those characters and thus to the production of 

 new races; it was supposed that the character in question was 

 "built up" by continued selection in one direction, and that the 

 average development of the character in all the offspring was thus 

 increased in successive generations to an indefinite extent. It 

 was this view as to the supposed action of artificial selection 

 which formed the basis of Darwin's theory of natural selection. 



Non-effect of Selection on Pure Lines. On the other hand it 

 has been known for a long time that the limits of the possible im- 

 provement of any character by artificial selection are soon reached 



