FLOWERS OF IMPA TIENS FULVA 163 



In view of the structure of the staminal whorl, the occurrence of 

 spontaneous self-pollination in the experiments with pallida and fulva 

 described above, and the occurrence of pseudocleistogamy present a 

 problem. The ligule-lLke staminal processes roof over the pistil in 

 such manner that no pollen from the anthers could lodge upon the stig- 

 ma. In the experiments referred to, the flowers before opening were 

 enclosed in paper envelopes so as to exclude insect visitors. A number 

 of these flowers set seed. In the pseudocleistogamous flowers fertiliza- 

 tion occurred before the opening of buds that morphologically were of 

 the chasmogamous type. Fertilization occurred at various stages 

 from that when the bud was five miUimeters in length to the mature 

 flower just about to open its sepals and petals. There are three possible 

 methods by which pollination might occur before opening of the flowers. 

 In the first place, the pistil might, as Gray figures, protrude through 

 the staminal ligules and pollen grains lodge upon it. The pistil is 

 figured protruding through the staminal ligules in /. balsamina by Engler 

 andPrantl (49) and in /. «o//-/awgere by Le Maout and Descaisne (27). 

 Knuth (26) describes the protrusion of the pistil in /. noli-tangere, parvi- 

 flora and balsamina. In /. latifolia, the stigmas are not covered (30a). 

 The detection of this protrusion in pseudocleistogamous flowers would 

 be a Utle difficult owing to the fact that the condition of pseudocleisto- 

 gamylis not likely to be recognized until the perianth cap is being pushed 

 off. At this stage the stamens are so far shrivelled and dried as to 

 place little value upon any appearance of opening through the staminal 

 flaps. Furthermore, the protrusion of the pistil in this manner is of 

 apparently rare occurrence in the chasmogamous flowers, yet the per- 

 centage of spontaneously self -fertilized flowers in fulva was surprisingly 

 high, three out of five cases. 



The second possible method by which self-pollination could occur is 

 by the growth of pollen tubes directly through the mass of the staminal 

 roof to the stigmatic sufaces below. No pollen tubes taking this course 

 were ever detected either in the dissection of the flowers or in micro- 

 scopic sections. Such negative evidence is, however, of little value 

 because a structure as dehcate as a pollen tube would hardly remain 

 intact through the drying and shrivelling of the tissue that occurs before 

 these pseudocleistogamous flowers are recognized. 



A third possibility is the detachment and dislocation of the staminal 

 whorl before the opening of the flower. In the smaller pseudocleisto- 

 gamous flowers this dislocation would be impossible because of the small 

 size of the flower and the compact condition of the floral parts. In the 



