I9IS-] STERILITY IN CERTAIN PLANTS. 71 



pollen found is morphologically perfect. The maturation difficulty 

 in spermatogenesis is largely at the first spermatocyte division. 



Fi plants are as fertile inter se as in back crosses with either 

 parent. 



Segregation of determiners for fertility occurs in F^, so that by 

 recombination some perfectly fertile plants are obtained in Fo. 



Nearly all fertile Fo plants selfed give only fertile progeny. Oc- 

 casionally a fertile F. plant selfed may give a slightly non-fertile 

 daughter. 



Numerous combinations that should be possible in Fo are omitted 

 in the population obtained. Combinations approaching N. rustica 

 seem to be more frequent than those approaching N. paniciilata. 

 Many more homozygous combinations occur in F, than might be 

 expected. 



Perfectly fertile plants giving perfectly fertile progeny, hetero- 

 zygous for many allelomorphs, do occur in Fo. 



No more than a very general formal interpretation of these facts 

 can be made at present, but assuming that the chromosomes carry 

 the hereditary character determiners, and that these react with the 

 cytoplasm under proper environmental conditions to build up the 

 soma, attention is called to the following possibilities of satisfying 

 the conditions imposed by the data. 



1. There is selective elimination of F, zygotes. 



2. There is no evidence of selective fertilization. (I infer this 

 from the fact that F.^ plants are as fertile inter se as in backcrosses.) 



3. The selective elimination of non- functional gametes that must 

 occur in F^ and the recombinations of functional gametes that give 

 different grades of fertility in F, cannot be interpreted by a Mende- 

 lian factorial notation without subsidiary assumptions, but possibly 

 may be the result of one of the two following hypotheses : 



{A) Through multipolar spindles, mating of non-homologous 

 chromosome pairs at synapsis, or other mitotic aberrations at the 

 reduction division, the 24 chromosomes characterizing each of the 

 two species may be irregularly distributed at gametogenesis. If 

 some of these irregular gametes may function, the majority of the 

 experimental data are satisfied, but there are reasons which there is 

 not time to consider which make this scheme improbable. 



