694 



THE AMERICAN NATURALIST [Vol. LI 



The fertile spikelets of sterile plants are generally scat- 

 tered at random over the panicle, and each fertile spikelet 

 may be regarded as representing a separate case of re- 

 version ; but in mosaic forms which show higher fertility 

 and are of rarer occurrence, the reversion may have 

 taken place in earlier stages of plant development, result- 

 ing in larger fertile sections. Consequently, when the 

 count of fertile spikelets is made with only the first type 

 of sterile plants, a more correct value for the frequency 

 of reversion may be obtained. The result of such a count 

 on 902 panicles containing 93,635 spikelets is 1,858 fertile 

 spikelets, i. e., 1.98 per cent, of the total number of 

 spikelets. 



The mosaic forms appear in several different grades of 

 partial fertility. In a panicle either one or more branches 

 or even one half of the panicle can be highly or entirely 

 fertile, the remaining part being absolutely or nearly ab- 

 solutely sterile. Similarly, in a single plant some whole 

 panicles can be entirely or highly fertile while others are 

 of the ordinary grade of partial fertility. Furthermore, 

 similar mosaic conditions were also observed in single 

 flowers of sterile spikelets. While all six anthers of a 

 sterile spikelet generally bear none or but few pollen 

 grains, occasionally flowers appear in which certain 

 anthers contain a considerable number of pollen grains 

 of normal appearance and others show the ordinary state 

 of sterility. Hence it may be assumed that the reversion 

 can take place at any stage of plant development. 



The partial homozygosity of heterozygotes, correspond- 

 ing to the partial fertility of sterile plants, may be esti- 

 mated in the following way. Assuming that the possi- 

 bility of reversion at any stage of a plant's life, similar 

 to that observed above, may also occur in heterozygotic 

 cells, then we may distinguish for convenience two differ- 

 ent types of reversions; there is the reversion which will 

 cause partial homozygosity within a single flower, and the 

 reversion which will produce an entirely homozygotic 

 spikelet or larger homozygotic sectant. Suppose then 

 that the latter reversion will give to the heterozygote 



