No. 594] BLENDING AND MENDELIAN INHERITANCE 327 



out the further assumption of gametic contamination 

 (which Hoshino makes) and the latter must invoke con- 

 tamination of different degrees in order to explain the 

 pseudo-early and pseudo-late classes. The former scheme 

 involves two explanatory principles, the latter only one. 

 Other things being equal, the simpler hypothesis is to be 

 preferred. 



TABLE III 



Experiments decisive between the two hypotheses are 

 difficult to devise, but certain tests are possible. On the 

 hypothesis of Hoshino one would not expect to obtain a 

 class splitting into homozygous early intermediate and 

 homozygous late intermediate. On the alternative hy- 

 pothesis such a class (40-50) should be obtainable. A 

 "constant" class exactly intermediate between the parent 

 varieties (say 45) would be impossible on Hoshino 1 's 

 hypothesis, unless he is willing to admit an indefinite 

 amount of contamination, which, however, would render 

 the two-factor hypothesis superfluous. On the alterna- 

 tive hypothesis such a constant intermediate class should 

 be obtainable after a sufficient number of inbred genera- 

 tions. In reality Hoshino 's observations show that it is 

 obtained in F 4 and is then more abundant than any other 

 "constant" class. The largest group of hybrid offspring 

 in Hoshino 's experiments and that belonging to the latest 

 inbred generation is his F 4 generation raised in 1914 (see 



