174 CARROLL— ON DEVELOPMENT OF 



the two outer walls alone. By the time the cotyledons have been 

 differentiated the septa are represented only by tags adhering to the 

 placenta and ovarian wall (Fig. 35). 



Early in the history of the ovary, very soon after its appearance 

 and before the appearance of the ovule, there is a striking differentiation 

 in the hypodermal layer as in other members of the flower (Fig. 34). 

 This layer is composed of large, regular, squarish cells quite distinct 

 in shape and size from neighboring layers. The cells very early become 

 filled with tannin which adds to their prominence in a section. At 

 five points, opposite the centers of the septa, the large squarish cells 

 are replaced by two, or sometimes three or four small cells also filled 

 with tannin (Fig. 34). Below these small hypodermal cells a line of 

 small cells extends transversely through the ovarian wall to the inside. 

 These mark the lines of rupture. Some time after the total degenera- 

 tion of the septa, the innermost layer of the ovarian wall begins to show 

 signs of degeneration and shrinkage. This degeneration extends up 

 the small lines of cells that mark the lines of dehiscence. As growth 

 continues degeneration extends backward until it reaches the vascular 

 bundle in the middle of the valve and until it reaches the hypodermal 

 layer in the lines of dehiscence (Fig. 34). The epidermal cells by this 

 time have become filled with tannin, and are in a state of turgescence. 

 No mechanical thickenings appear. The bursting of the pod is con- 

 trolled apparently by turgescence of the hypodermal and epidermal 

 layers. 



Conclusions as to Cleistogamy 



Many of the attempts to explain the phenomenon of cleistogamy 

 have been based on teleological considerations. Knuth, (26) for example, 

 explains the appearance of cleistogamous flowers in Drosera rotundi- 

 folia by the assertion that insects are more attracted by the glistening 

 drops on the glandular leaves and do not visit the flowers. "Owing to 

 the continual capture of insect-prey open flowers are useless for the 

 sundew, and it therefore develops cleistogamous ones." In Oxalis 

 acetosella, the cleistogamous flowers are developed in June and July 

 because insects would not visit the rather inconspicuous flowers of 

 Oxalis when more showy flowers are in blooom, but in spring when the 

 plant does not have to compete with more showy forms the chasmo- 

 gamous flowers appear. Knuth explains the occurrence of cleistoga- 

 mous flowers in the Anonaceae on the authority of Loew as an adapta- 

 tion for protection against ants. Darwin (13, p. 338) found the explana- 



