Size inheritance and the pure line theory. 233 



with the idea of definite modifiers or factors of uniform vahie assumed 

 in the current interpretation of size inheritance. What I mean to suggest 

 is this, that it is unnecessary to invoke a multiplicity of size factors 

 in order to explain the increased variability of F«, since a single factor 

 if it is subject to quantitative modification in the P\ zygote would 

 account for it equally well. 



But, it may be asked, how can we on such an hypothesis account 

 for the occasional case in which an Fa individual is as extreme in size 

 as the uncrossed grandparent, say as extreme as 3 or 6 in the illustration 

 given. We have only to suppose that in such a case no modification 

 occurs by contamination in the Fi zygote, so that giunetes aie formed, 

 not 4 and 5, but 3 and 6 respectively. 



But it may be objected further, occasionally in F« an individual 

 is obtained smaller than the small grandparent or larger than the large 

 one; how can such occurrences be explained? Before requiring an ex- 

 planation of such cases, it should first be established whether they 

 really fall beyond the actual range of the grandparental race, or 

 whether they merely fall beyond the empirical range as determined by 

 an insufficient number of observations. For example in Table 3, it will 

 be observed that in one of the F:\ families a variate is recorded (class 

 113) smaller than the smallest individual observed in the grandparental 

 race, B. Emeeson and East suggest that in view of such variation 

 it seems possible that from this F3 family selection might isolate "a stable 

 type with seeds even smaller" than those of race B. But it should he 

 observed that the empirical range of race B rests on 18 observations 

 only, while that of the F3 family includes 80 such observations, a number 

 sufficient to more than double the accuracy of the determination. It is 

 therefore not certain that the range of the F3 family actually extends 

 below that of race B; and if it does not, it would seem to afford much 

 poorer material for selection in the direction of small seed size than 

 the uncrossed race B, which has a nruch lower mean, and a variability 

 nearly three-fourths as great as that of the F3 family. (The coefficients 

 of variation are given as 6'05 + • 68 and 8'70 ± • 46 respectively; the 

 means are 134'39 and 141'31 respectively.) It is scarcely safe to assume 

 that genetic variation is wholly wanting in race B, though present in 

 Fs, wliich is only a little more variable. 



Is there any reason to think that a race of animals exists not 

 variable genetically as to size? Jennings alone on the basis of direct 

 observation seems to have answered this question in the affirmative. 

 He believed that he had isolated eight "pure hues" of Paramecium each 



