696 THE AMERICAN NATURALIST [Vol. LI 



remains green. The fertile spikelets occurring in a small 

 number on the otherwise sterile panicle appear on rip- 

 ening as yellow spots scattered among green spikelets; 

 the plants with both sterile and fertile panicles appear in 

 the fall also as mosaic forms with green and yellow 

 leaves. This feature of the sterile plants is in direct 

 contrast to the behavior of the mosaic plants with the 

 variegated and the entirely green leaves studied by De 

 Vries and Correns. 



The observations in the foregoing pages seem to paral- 

 lel those made by the authors cited at the beginning of 

 this paper. In the present investigation, however, there 

 was observed also the transformation of allelomorphs in 

 the opposite direction, that is, the transformation of the 

 dominant allelomorph into the recessive allelomorph, 

 something scarcely mentioned in the investigations re- 

 ferred to above. The observations in this regard were in 

 brief as follows. 



In the first place, the spontaneous occurrence of segre- 

 gating families was observed again among the descend- 

 ants of the families which had proved in the experiments 

 already described to be constantly fertile. This suggests, 

 just as did the occurrence in Family A and Family B in 

 1912, the probability of the AA cell changing into the Aa 

 cell. 



In the second place, a constant tendency of the dom- 

 inant allelomorph to be transformed into the recessive 

 allelomorph was observed in certain strains. In 1913, 

 special attention was paid to such segregating families in 

 which the excess of recessive segregates over the theo- 

 retical expectation was particularly high. Although, as 

 already noted, the variation among the segregating fam- 

 ilies in 1913 with regard to the deviations from the reces- 

 sive proportion might possibly have arisen from experi- 

 mental errors associated with a certain probability of alle- 

 lomorphic reversion from recessive to dominant, yet it 

 was deemed not impossible that the very considerable 

 excess of recessives exhibited by certain families might be 

 caused by other reasons. This point was seemingly de- 



