164 CARROLL— ON DEVELOPMENT OF 



older stages there is a possibility that it might occur. The failure to 

 detect it cannot be taken as evidence against the existence of this condi- 

 tion, as no definite search was made with its detection in mind. Fur- 

 ther investigation is necessary to explain the method of self-pollination 

 in these cases. 



The arrangement and behavior of the stamens in other members of 

 the genus do not always agree with fulva. In pallida, the staminal 

 structure cannot be distinguished from that of fulva. In sultani the 

 structure is similar, but the relative sizes slightly different. In the com- 

 plete covering of the pistil by the Hgular flaps, in the dehiscence of the 

 anthers before opening of the flower and in the subsequent fall of the 

 whole staminal whorl, sultani resembles fulva and pallida. In noli- 

 tangere, (Knuth 26, II p. 236) dehiscence occurs before opening of the 

 flower, subsequently the anthers separate and the stigma matures. 

 In balsamina (Knuth 26, II p. 236) the mechanism is the same as in 

 noli-tangere. Knuth reports the mechanism of parviflora as being 

 similar to that of balsamina, but in many of these forms the whorl of 

 stamens is easily detached and would readily fall away owing to the 

 resupinate position of the flowers. 



The Stamens in the cleistogamous Flowers 

 The early stages of staminal development in the cleistogamous 

 flowers are exactly like those of the chasmogamous, but the mature 

 condition is strikingly different. Not infrequently one o^- two of the 

 stamens are shorter than the others. The mature stamen is narrow and 

 strap shaped. There is no tendency to broaden and unite with adja- 

 cent stamens as in the chasmogamous. The flap-like ligular appen- 

 dages on the inner face are absent. The stamens meet above the pistil, 

 but there is no tendency toward fusion. Bennett has published good 

 figures of these in their mature condition (4 fig. 10 and 11). The pollen 

 sacs are greatly reduced in capacity and only two, instead of four as in 

 the chasmogamous flowers (Fig. 15). Each sac is surrounded by an 

 endothecium which is exactly similar to that of the chasmogamous 

 anthers. The great extent of the endothecium in the cleistogamous 

 flower is striking. Relatively it is much greater in extent than in the 

 chasmogamous flower, being well developed on the connective side 

 of the pollen sac, as well as on the outer sides. The endothecium pos- 

 sesses similar thickenings and opens by a stomium on its inner face 

 (fig. 15). The quahty of pollen is very greatly reduced, frequently 

 only two or three grains appearing in a microscopic section, and these 



