FLOWERS OF IMP A TIENS FULVA 149 



These flowers are considerably larger than the typical cleistogamous 

 flowers. The cleistogamous flowers at the stage ol fertilization average 

 two millimeters. Their morphology is that of the chasmogamous 

 flowers. The four sepals are present in the same position and form. 

 The anterior sepals show the same double appearance; the posterior 

 has the well-developed spur; the two lobed petals are fully developed, 

 the whorl of stamens appears fused and each stamen bears four pollen 

 sacs like the chasmogamous, but unlike the cleistogamous flowers which 

 have only two pollen sacs. They could not be distinguished from chas- 

 mogamous buds of the same age. Apparently these flowers are devel- 

 oping to be chasmogamous flowers, but fertilization occurs before the 

 bud opens. The whole mass of calyx, corolla and stamens is then 

 loosened, and as the pistil expands this mass is forced off. Some of 

 these flowers were marked with tags at the stage when loosening of this 

 mass was first evident. Subsequent observations showed that the 

 majority of such marked flowers developed fruit and seed similar to 

 those of chasmogamous flowers. 



These intermediate flowers belong to the type which Hansgirg 

 (20) has designated pseudo-cleistogamous. Morphologically the flowers 

 are like the chasmogamous, but physiologically they behave as cleis- 

 togamous flowers. Adverse conditions of light, temperature, moisture, 

 or drought apparently lead to self-fertilization in buds. Hansgirg 

 distinguishes photo-, hydro- and thermo-cleistogamous flowers. These 

 do not include all the causes which lead to this phenomenon. Numerous 

 references occur through the literature to the occurrence of the phenome- 

 non under varying conditions and in numerous species. Partial lists 

 of such species and conditions are given by Hansgirg (20) and Knuth 

 (26 I pp. 55-58 English Translation). In the case of I.fulva and pallida 

 observed in late August and early September 1916, they were probably 

 induced by dry weather in August. None was observed during July 

 of that year before the dry spell began, but they were not rare toward 

 the close of the dry weather and for a short time succeeding. In none 

 of the experiments described below, however, in which plants were 

 subjected to various adverse conditions, were pseudocleistogamous 

 flowers detected. 



Still another type of flower which would fall within Hansgirg's group 

 of pseudocleistogamous flowers, was discovered in castration experiments 

 described below. Full grown buds just about to expand were cut open. 

 The pollen sacs were found already ruptured. The stamens were 

 removed, and the flower covered to prevent the entrance of insects. 



