446 AMERICAN JOURNAL OF BOTANY [Vol. 10, 



hence due to a change in the physiological relations between pistils and the 

 pollen of long stamens. Pollen from both short stamens and long stamens 

 was examined at frequent intervals throughout the entire period of bloom; 

 there was very little abortion, the pollen was successfully germinated in 

 artificial media, and the use of such pollen in legitimate crosses on several 

 dates during the first 30 days of bloom was almost invariably highly effective 

 in pod- and seed-production. There were in this plant no noticeable 

 evidences of impotence of stamens and anthers, such as are to be seen in 

 some plants of this species. 



The branches left to open-pollination produced pods quite as in the year 

 191 7. Several of the main branches were selected and divided into thirds, 

 and seeds in all pods in the lower two thirds were counted; then an equal 

 number of pods from the many pods in the uppermost third were taken at 

 random for counts, the entire number, 142, being as near the number of 

 pods secured from the selfing by hand in which the pollen of long stamens 

 was used as was possible. As shown in table 1, the range for number of 

 seeds per pod and the average were higher for the lower two thirds than in 

 the last 'third, showing that the change in compatibility involves number of 

 pods formed rather than number of seeds per pod. Comparison shows that 

 the controlled pollinations in which pollen from long stamens was used were 

 somewhat more successful than free open-pollinations, when judged by the 

 average number of seeds produced. 



A third mid-styled plant (M no. 3) was grown in isolation in the New 

 York Botanical Garden at a distance of about one mile from the location of 

 the plant M no. 1. This plant proved to be decidedly less self-compatible to 

 open self-pollination than were M no. 1 and M no. 2, but, as in M no. 2, 

 there was quite the same proportion of pods produced throughout the entire 

 season of bloom, no change in compatibility being evident as observed in 

 the plant M no. 1. 



Summary. In these tests seeds were obtained to illegitimate self- 

 pollination of plants of all three forms. Mid-styled plants were most 

 highly self-compatible in respect to the number of pods produced. The 

 pods found were distributed irregularly over the entire period of bloom 

 except for one plant (M no. 1), in which it was found that there was an actual 

 change in the degree of self-compatibility to pollen from its own stamens. 

 The results from year to year have been very uniform for all plants, except 

 for the plant S no. 2. Its feeble production of pods in 1918, followed by 

 the production of 161 pods in 19 19, suggests that the plant is rather strongly 

 self-compatible provided insects make the proper pollinations. It may 

 readily be conceded that the kinds of insects that can most efficiently self- 

 pollinate long-styled and mid-styled plants are not the ones which best self 

 the short-styled plants. 



Self-pollination does not appear to be uniformly as efficient in seed- 

 production as are certain legitimate cross-pollinations, at least for the one 



