VARIATION AND MENDELIAN INHERITANCE 379 



As is well known, the exponents of this latter hypothesis lay 

 stress upon those undoubted cases in which the second hybrid 

 generation, while showing an intermediate condition, like the 

 first, nevertheless displays a higher variability than the latter. 

 The necessity for such an increase in variability, as a result of 

 segregation, is obvious where a single pair of unit factors is con- 

 cerned, as in ordinary Mendelian inheritance. That it would 

 be equally necessary if a given character difference were con- 

 ditioned by two or more pairs of independently segregating 

 factors may readily be proved." 



Such a general increase in variability, in later hybrid genera- 

 tions, was, it is worth while noting, well recognized by the early 

 hybridists before the work of Mendel became known. To one 

 who is not committed to the doctrine of the immutability of the 

 'gene,' such an increased variability is intelligible upon the 

 assumption of a tendency for the parental contributions to seg- 

 regate from one another during the formation of the germ cells. 

 This tendency may be completely realized, as in the case of typi- 

 cal Mendelian inheritance. It may be overcome, wholly or 

 partially, by a tendency toward fusion, in those numerous cases 

 in which we have a permanent blending of characters, whether 

 or not an increased variability is shown in later hybrid gen- 

 erations. 



As between these two theories, the case is by no means as 

 definitely closed as the confident assertions of various recent 

 Mendelian writers would lead one to suppose. Both are still 

 legitimate scientific hypotheses. ^^ x .personally lean toward the 

 view which seems to me to involve the fewest unproved assump- 

 tions — the view, namely, that characters, genetic as well as 

 somatic, may and do actually blend with one another per- 

 manently. 



I have thus far reared Fi and r2 generations from three dif- 

 ferent crosses among my various races of mice. I am able to 



'1 This point has been treated satisfactorily in various recent works (e.g., 

 Babcock and Clausen's 'Genetics,' pages 183-186) and need not receive further 

 consideration here. 



12 This I feel warranted in asserting, despite Castle's recent defection from the 

 ranks of those who uphold the view-point here advocated. 



