188 THE MENDEL JOURNAL 



thing which has size possesses length, height, and width 

 ( = thickness) ; it has three dimensions. What evidence 

 has Dr. Pearl that each one of these dimensions is not a 

 biologically independent variable ? In other words, the 

 length, height and width of a single comb may be separate 

 gametic entities, each segregable from the other. Or, on 

 the other hand, they may of course be coupled together. 

 Before we can be assured that this material is the " most 

 homogeneous possible," we must not only consider the 

 comb, but each one of its separate characters. Points like 

 these have to be considered and the homogeneity of the 

 material proved, not by massing heterogeneities together, 

 but by geiietic experiment, before we have any real basis 

 upon which our judgement as to the homogeneity or 

 otherwise of the material can be formed. 



We must now consider another source of error in this 

 biometrical treatment of the subject. It is the implica- 

 tion that the ordinary or older method of selection by 

 inspection of the visible or somatic characters will enable 

 us to fix a homogeneous race. This older method of 

 selection is based upon an old conception. In pre- 

 Mendelian days it was generally believed that the external 

 or visible body characters were a true and trustworthy 

 criterion of the gametic constitution of an individual. 

 We now know that in the majority of cases which have been 

 investigated this may not be so. We know, for instance, 

 that two races of white sweet pea, which so long as each 

 race is bred within itself will produce only white-flowered 

 individuals, but yet when the two races are crossed 

 purple-flowered individuals m ay result . It is clear that the 

 visible and somatic white character of the flowers of these 

 two races is not a reliable criterion of the hidden factors 

 which each race carries. Selection of these tAvo races 

 because of their visible white character does not produce 

 a race homogeneous for whiteness. Again, we know it 

 is not possible to fix the jonquil character of canaries by 

 selecting the best jonquil canaries to breed from. And 

 we know also that we cannot fix the blue Andalusian 

 character of the Andalusian hen by selecting the best 

 blues and rejecting the black and splashed- white indivi- 

 duals. For many years blue Andalusians have been 

 selected by such a mode of selection, namely, that based 



