152 CARROLL— ON DEVELOPMENT OF 



glanduligera, Landl. ( = 1. Roylei Walp) Loew (30a) has described a dif- 

 ferent mechanism. A cleft between the anterior stamens forms a passage 

 through which the hairs on the back of the bumble-bee are thrust as 

 the bee makes its way below the stamens. The staminal processes aid 

 in the function of collecting the pollen forming a "pseudo-stigma." 

 According to Knuth, (26) Hildebrand and Delpino (cited by Knuth), 

 however, the mechanism is in this species much as in the preceding 

 species described. 



Accounts differ as to the fertility of the various species of Impatiens 

 with their own pollen. In the Gardener's Chronicle for 1868 a quo- 

 tation from "American Gardening" states that the "Balsams, {Im- 

 patiens pallida and fulva) cannot be fertilized with their own pollen." 

 This does not agree with Darwin's account or with my own experiments. 

 In a personal communication from Darwin, quoted by Bennett, (4, 

 152), Darwin gives the following account — "Eleven such pods from 

 perfect flowers, spontaneously self -fertilized yielded on an average 

 3.45 seeds. I carefully brushed away the pollen from some of the 

 perfect flowers, and fertilized them with pollen from a distant plant, 

 but got only three pods, containing to my surprise only 2, 2 and 1 seed. " 

 Darwin (12, p. 366) states that Impatiens barbigera set seed abundantly 

 when covered with a net, and that plants of /. noli-tangere covered with 

 a net produced from perfect flowers eleven spontaneously self-fertilized 

 capsules which contained on an average 3.45 seeds (12, p. 367). 



My own experiments in self-pollination included I. fulva, pallida 

 and sultani. Each proved fertile to its own pollen. Each was tested 

 by two methods, the first to determine the occurrence of spontaneous 

 self-pollination, the second to determine the results of the direct applica- 

 tion of pollen. In the first method the flowers were sealed in paper 

 envelopes to exclude insect visitors. In five experiments with sultani 

 no seed was set. In the greenhouse where were some two dozen pots 

 of /. sultani no seed occurred during the winter except as a result of 

 hand pollination. When cross-pollinated by hand they set abundant 

 seed. In five flowers of fulva sealed to allow spontaneous selfing, three 

 set seed, and in five flowers of pallida, one set seed. In the second 

 method to test the effects of direct application of the pollen, a loosened 

 whorl of stamens was removed, or one was taken that had fallen from 

 position but lay on the petals of the flower, care being observed to 

 exclude possibility of cross pollination, and the pollen was applied to 

 the stigma. Five experiments on each of the species resulted as follows: 

 in pallida four set seed; in fulva five; in sultani, five. In pallida and 



