160 CARROLL— ON DEVELOPMENT OF 



of fulva. They show the same two lobed structure as the chasmogamous 

 petals. The four sepals are present in the same position as in the 

 chasmogamous flowers (Fig. 9). The spur sometimes appears as a 

 prominence, but sometimes no trace of it could be detected. Darwin, 

 quoted by Bennett,'* speaks of the "nectary in the cleistogamous flowers 

 as a mere rudiment." Five stamens are present but differ from the 

 chasmogamous stamens in a manner to be explained below. The ovary 

 is five-celled at first, but becomes one-celled at maturity through the 

 breaking down of the septa. 



The condition in Impatiens fulva must not be regarded as typical 

 for cleistogamous flowers in general. In many there is reduction in 

 either caH'x or corolla, or in both. Helene Ritzerow^^ lists a number of 

 species suffering such reduction. She states that in Specularia perfol- 

 iata the number of sepals is reduced; in Cardamine chenopodifolia and 

 Helianthemum glomeratum the corolla is wanting; in Viola sp., Poly gala 

 sp., and AmpJiicarpaea monoica the corolla is present but rudimentary; 

 in quite a number of species the corolla shows lesser degrees of reduc- 

 tion, among which she mentions Impatiens sp. but does not state which 

 species. In Impatiens noli-tangere the petals appear as whitish scales. (35) 



The calyx and corolla do not open, but the two whorls are cast off to- 

 gether as a cap, (Fig. 2)2)) which has been compared in appearance to the 

 calytra of a moss. This phenomenon appears also in /. noli-tangere (26) 

 and in /. pallida. Bennett (4) and Gray (18) both figure these caps. 

 Bennett suggests that the mechanism of expulsion may be in part the elas- 

 tic filaments. In the expelled caps the somewhat coiled position of the 

 filaments suggests this condition. The expansion of the ovary would 

 seem to play also an important part in the process. Gray's figures 

 suggest this explanation, and caps may frequently be found adhering 

 to the tips of enlarged ovaries as if forced up simply by the increase 

 in the size of the ovary. A similar expulsion of the united mass of sepals, 

 petals and stamens is described for /. noli-tangere (26). 



Of some interest is the occurrence of crystals and tannin throughout 

 the flowers and other parts of the plant. The occurrence of raphides 

 and tannin is characteristic of the genus (49). Sections even of very 

 young parts show the presence of an internal secretion. This tested with 

 ferric chloride proved to be tannin. It occurs in single cells, or scat- 

 tered, or in definite layers, and in vessels formed by the disappearance 

 of end walls of elongated cells. These tannin vessels are frequent 

 in the cortex of stem, petioles and peduncles. In hypodermal layers 

 of stem and leaves and through the mesophyll both isolated cells and 



