328 Belling. 



in the zygote. We usually regard the haploid generation as the bearer 

 of imperceptible factors destined to aid in the appearance of characters 

 in the diploid generation. Then, except for theoretical purposes, the 

 haploid generation is negligible. Now the tables are turned, and we 

 may regard the diploid generation of zygotes as negligible, except as 

 the bearer of imperceptible factors destined to affect the haploid gene- 

 ration, after separation of the pairs, by visible changes among its 

 members. It will be shown hereafter how this ratio of 1 : 1 is readily 

 intelligible in this way. 



To sum up: — The pollen-grains of half of the plants of 

 the second generation are half aborted. The other half of the 

 plants have normal pollen-grains. 



Ovules of the Second Generation. 



Velvet by Lyon. — Many of the plants had abundant gaps in the 

 seed-rows showing aborted ovules, but no count was made of anjthing 

 but good seeds. With regard to the average number of good seeds in 

 a pod, there were more plants below the average of the Velvet and 

 Lyon than above this. Some of these owed their position to a large 

 number of dead seeds, not of aborted ovules. 



Lyon by Velvet. — Of those plants with white pubescence on their 

 shoots, 113 showed many gaps in their dry pods; 116 plants had some 

 gaps; 169 had full pods; and 73 plants bore no pods, or were doubtful, 

 or were accidentally not noted for this character. One hundred and 

 two plants had black tomentum on their shoots, and moStly failed to 

 set pods; while those that did set a fair crop often showed gaps due 

 probably to imperfect pollination, for they often have an abnormally 

 short style. If, omitting the black plants, we add the first two classes, 

 we get 229 plants with some gaps, to 169 with full pods. But a count 

 of seeds and dead ovules showed that some of these 229 belonged to 

 the full -podded class. Hence, I consider that the two divisions were 

 probably approximately equal. 



The following table gives a list of plants in which the seeds and 

 aborted ovules were counted. This lot of second-generation plants is 

 not a random sample with respect to semi -sterility, for some regard 

 was paid to shape of pod when selection was made for pod pubescence, 

 earliness, and size of crop. Seeds of several of these plants were sown 

 for the next generation, and in these cases the pods (about twenty) 

 wliich were opened to obtain seeds are not included in the table. 

 LV-92 was a very early plant : and LV-486 was a heteroz3'gous lilack 



