THE MECHANISM OF HEREDITY 71 



about 200, the grandparental sizes probaWy would have 

 been obtained. Ftirthennore, if one studies the results 

 obtained in the F3, F^ and F^ generations, considering 

 only the range of their variability, it is dear they differ in 

 both type and extent of variation. 



No assumptions unproved for the inheritance of quali- 

 tative characters are necessary for thus visualizing the 

 inheritance of quantitative characters, and no facts dis- 

 covered in tracing the inheritance of other characters — 

 such as those involving linkage — are overlooked. But in 

 order to picture the situation easily, let us assume that 

 dominance is usually absent (often the case), that two 

 doses {i.e., the homozygous condition) of a factor have 

 twice the effect of one dose (true for all practical pur- 

 poses), that independent factors cumulative in their oper- 

 ation are allelomorphio to their absence in the hybrid 

 (linkage though it complicates matters, does not change 

 our reasoning) . 



Let us assume a case of "blended" inheritance where 

 all fluctuations due to environment are eliminated. A 

 plant 12 inches tall is supposed to be crossed with a plant 

 28 inches tall. The difference between them is 16 inches. 

 If this difference is due to one allelomorphic pair in which 

 dominance is absent, the F^ generation is all intermediate 

 — about 20 inches — and the Fj generation falls into three 

 classes in! which two represent the grandparental forms 

 and one represents the F-^ form. Twenty-five per cent, 

 are 12 inches tall, fifty per cent, are 20 inches tall and 

 twenty-five per cent, are 28 inches tall. 



But suppose this 16-inch difference between the 

 parents is represented by two allelomorphic pairs instead 

 of one. The F^ generation is again 20 inches tall, but 



