512 



THE AMERICAN NATURALIST [Vol. XL VIII 



ory's expectation would be 15:1 and our own 35:1. Moreover, 

 the individual records fluctuate in both directions about the 

 ratios to be expected upon our point of view, but practically all 

 vary in the same direction from the expectation of Gregory, 

 namely, in the direction of the other expectation. 



There is reason, then, to believe that in these Primulas the 

 factors derived from the same parents may segregate from each 

 other as allelomorphs, while allelomorphs derived from opposite 

 parents meanwhile assort at random. For although the allelo- 

 morphs exist in sets of four they must pair two by two for segre- 

 gation, as do the chromosomes, and two derived from the same 

 parent may happen to pair with one another. The chance that 

 this should occur is one third, since there are three possible 

 modes of pairing. Such a result is difficult to explain except on 

 the chromosome view of heredity. It would give ratios different 

 from those theoretically expected by Gregory, but more in 

 accord with his experimental data. The principle upon which 

 our own expectation is founded may be briefly summed up by 

 saying that where more than two factors which are normally 

 allelomorphic to each other are present, the pairing of these 

 ;ilh'lonior|»hs with each other preparatory to segregation usually 5 

 takes place at random. 



Hermann J. Muller 



s That this is not always true is shown by Bridges' case of "non-dis- 

 junctional" females of Drosophila, which contain one Y and two X chromo- 



other in the reduction division of the normal fly, which contains only two 



dom, for they oftener undergo the segregation X-XY than XX-Y, pre- 



Y, an/so more apt to act as homologues. 



