24 



Pearl and Bartlett. 



characters not obviously connected with those which determine the 

 visible external appearance of the grains. But the Fo classes show 

 throughout differences in chemical composition from the corresponding 

 pure parent races. At first glance these differences appear to be 

 entirely irregular and without meaning. But in the present section 

 of the paper it has been shown that the actually observed 

 mean results of the analysis of the several F.j classes with 

 respect to protein, crude fat, and ash content are in very 

 close agreement with what they would be expected to be 

 if each of these characters segregated separately as unit 

 characters independently of all others. With a somewhat 

 smaller degree of precision the same thing is true of the carbohydrate 

 complexes "crude fiber" and "pentosans". The only assumption made 

 in deducing these results is that the determiners of the "sweet" 

 character (i. e., high sugar content), when present in homozygous 

 condition, tend to inhibit the expression of the dominant condition 

 of the other chemical characters named. This is by no means an 

 unreasonable assumption a priori. Recent Mendelian literature fur- 

 nishes a number of examples of one character inhibiting the expression 

 of another. Furthermore, quite apart from the a priori reasonableness 

 of the assumption m the present case, it possesses the most fundamental 

 of pragmatic attributes in that it "works". That is to say, let it be 

 considered pure hypothesis (which it is) : it brings into accord under 

 a simple scheme of relationship a whole set of observational data 

 apparently irregular and unrelated. The success with which a hypo- 

 thesis does this thing is at once the test of its usefulness and its 

 "truth" (i. e., its permanency until a more useful one has been found). 

 We do not wish to be understood as maintaining that the evidence 

 presented in this section conclusively proves the independent 

 segregation as unit characters of the chemical characters dealt with. 

 We are of opinion, however, that the evidence so far as it goes is 

 cogent and definite, and supports such a conclusion, though it does 

 not prove it. Whether there is or is not an independent segregation 

 of each of these chemical characters as a separate unit, it is quite 

 certain, as shown earlier in the paper, that there is segregation of 

 these characters as a class. 



Discussion and Summary of Results. 



The results set forth in this paper appear to be of significance 

 in two directions. In the first place they demonstrate in a direct 



