A. B. Stout 93 



might seem that the fertilities have been very much increased and that 

 further selection might yield a race all of which would be self-fertile, 

 at least to some degree. In eveiy case, however, a large number of the 

 ofifspring grown from highly self-fertile plants selected from such series 

 have been self-sterile, and on the whole the record for such progeny has 

 not been above that of the preceding generation. 



In certain cases also, high records of fertility seem to be correlated 

 with an ancestral record that is high. While the results in this respect 

 are not uniformly in agreement, a number of the most highly fertile 

 series do have an immediate parent of high self- fertility. Some evidence 

 on this point may be gained by grouping the results according to the 

 fertilities of the immediate parentage and without regard to the family 

 or generation. This has been done in the following table. 



The results show that in respect to the proportion of self-fertile 

 plants there were only slight differences in the progeny of parents 

 of high or of low fertility : the lowest percentage (31) was obtained in 

 the progeny of parents whose self-fertility ranged from •11-20. In 

 regard to the range of the self-fertilities, the offspring of parents of the 

 lowest fertilities (-Ol-lO) extended only to 23 7o- The highest range 

 is seen in offspring of the class 0*43-0'50. The number of offspring, 

 especially of the lowest class of parentage, is not as large as one would 

 wish. The evidence is quite conclusive, however, that there is strong 

 regression, especially in the offspring of plants of high fertility, and that 

 the various progenies do not differ in the degree that is decidedly 

 correlated with the performance of the respective parents. Still the 

 data are suggestive that the higher ranges and averages are to be 



