Inconstancy of Improved Races 787 



continued the culture during several genera- 

 tions, choosing ears with less than the average 

 number of rows. The excellence of the 

 race at once disappeared, and the ordinary 

 average of the variety from which I had started 

 seven years before, returned within two or three 

 seasons. This shows that the attained improve- 

 ment is neither fixed nor assured and is depend- 

 ent on continued selection. This result only 

 confirms the universal experience of breeders, 

 which teaches the general dependency of im- 

 proved races on continued selection. Here a 

 striking contrast with elementary species or 

 true varieties is obvious. The strains which na- 

 ture affords are true to their type; their aver- 

 age condition remains the same during all the 

 succeeding generations, and even if it should 

 be slightly altered by changes in the external 

 conditions, it returns to the type, as soon as 

 these changes come to an end. It is a real aver- 

 age, being the sum of the contribution of all 

 the members of the strain. Improved races 

 have only an apparent average, which is in fact 

 biased by the exclusion of whole groups of in- 

 dividuals. If left to themselves, their appear- 

 ance changes, and the real average soon re- 

 turns. This is the common experience of breed- 

 ers. 



A third point is to be discussed in con- 



