444 AMERICAN JOURNAL OF BOTANY [Vol. 10, 



results as summarized in table I. The number of seeds per capsule ranged 

 to 121 ; twenty-eight capsules had more than 30 seeds each, but 202 capsules 

 contained less than 11 seeds per capsule, and 14 contained only rudiments 

 of seeds. The average number of seeds per pod for those that had good 

 seeds was 8.91. In 1918 the self-compatibility of this plant was quite the 

 same as in 1917. 



In making the counts of seeds in 300 capsules (191 7 crop) of this plant, 

 position was taken into account. Each branch involved in the counts 

 was divided into three sections of equal length, here designated as the 

 lower third, middle third, and upper third. In the lower third of all branches 

 there were 10 capsules with more than 30 seeds each. It is clear that the 

 average number of seeds per capsule was lower toward the top of the 

 branches of this plant, but smaller capsules and fewer seeds per capsule are, 

 as a rule, to be expected toward the apex. Otherwise the plant was rather 

 uniformly highly self -compatible, and capsules with seeds were produced 

 in a considerable proportion of the flowers subjected to open pollination 

 throughout the entire period of bloom. 



Another mid-styled plant (M no. j) was grown in my own home garden. 

 This plant made a vigorous growth and produced in 191 7 at least 6,000 

 flowers. At the close of the season it was found that relatively few of the 

 flowers produced pods during the first two thirds of the period of bloom, 

 but that later nearly every flower produced a pod. 



In 1918 it was planned to test experimentally the behavior of this plant, 

 and especially to determine if the marked difference in production of fruit 

 involved changes in the innate ability of the plant to produce fruit or indi- 

 cated merely some difference in insect-pollination. A large, well-developed 

 long-styled plant was planted by its side. The two began blooming only 

 one day apart. The long-styled plant was allowed to bloom, thus affording 

 opportunity for free cross-pollinations by insects between the two, until 

 the 17th of August, when the long-styled plant was cut down. During the 

 time that both were in bloom nearly all flowers that opened on the mid-styled 

 plant developed fine pods, showing that the feeble production of pods during 

 the early part of bloom as observed in the previous year when the plant was 

 in isolation was not due to an impotence of the pistils. During the rest of 

 the season its performance was quite as in the previous year. 



In 19 19 the long-styled plant was kept cut down so that no flowers were 

 produced by it, and a series of guarded self-pollinations were made on the 

 plant M no. 1. Branches were enclosed in glassine paper bags. When- 

 ever pollen from short stamens was used for pollinations, the flowers to be 

 thus pollinated were opened early in the morning and the long stamens 

 were removed, and then pollen from the short stamens was used later in the 

 day when the anthers were dehiscing. 



Legitimate cross-pollinations, using pollen of flowers brought in vials 

 from the experimental plots over a mile distant, were made on 22 flowers 



