1356 A TEXTBOOK OF THEORETICAL BOTANY 



pollinating insects arises, since they are wind-pollinated in the chasmogamic 

 state. 



Certain facts seem to be incompatible with the idea of arrested develop- 

 ment as the cause of cleistogamy. The wide difference between the two 

 types of flower in the Grasses has already been remarked. Chase pointed 

 out that the cleistogamous flowers were so different from the others that 

 they would not, by themselves, be classified in the same tribe of Gramineae. 

 In Amphicarpa (Papilionaceae) the chasmogamous fruits are falcate in 

 shape, sparingly hairy and contain 2-3 seeds. The fruits of the subterranean 

 cleistogenes are pyriform and one-seeded, while those of the aerial cleisto- 

 genes are densely hairy and 1-3 seeded. The anatomical structure of both 

 kinds of fruit is different. In Ononis alopecuroides, the cleistogamous 

 flowers are not reduced and show no marked difference in structure from 

 the chasmogamous flowers. Moreover, the obvious fact that pollen and 

 ovaries are precociously mature seems conclusively against the view that 

 arrested development is the cause of cleistogamy, though arrested develop- 

 ment of the later-formed structures may well be the result of it. More 

 probably the fate of the flower is determined at the time of initiation in the 

 inflorescence primordium, the change of physiological balance being per- 

 haps dependent on a threshold reaction. 



The cytology of cleistogamy has only been studied in detail in the 

 classic case of Viola (Fig. 1263). In F. odor at a var. praecox there are three 

 types of flower, chasmogamous, semi-cleistogamous and cleistogamous. The 

 first-named never set seed. In Viola riviniana there are the same three types, 

 but the chasmogamous flowers occasionally form seed (Fig. 1264). In the 

 anthers of all these types of flowers, two types of pollen grains are found: 

 ovoid or tetrahedral grains, and round, ridged grains. The former only 

 germinate if conveyed to the stigma, but the latter germinate in the anther 

 of all three types of flower. While the pollen tubes remain inside the anthers 

 of the chasmogamous flowers, in the cleistogamous flower they penetrate 

 the reduced anther wall and grow out on to the stigma, while the tetrahedral 

 grains degenerate. The semi-cleistogamous flowers show all stages of inter- 

 mediacy as regard flower size, development of the spur and nectary, bending 

 of the style towards the anthers and reduction of the anthers, but they behave 

 like cleistogamic flowers in respect of their pollination. Such transitional 

 forms are well known in other genera, e.g., Oxalis, where flowers which are 

 perfect in development, but of minute size, are pollinated cleistogamously. 

 The actual details of the fertilization in cleistogamous Viola show nothing 

 specially correlated with cleistogamy. 



A final point in support of the belief that cleistogamy involves changes 

 going deeper than simple arrest of development is to be found in the 

 curiously neglected observation of Darwin's, that in trimorphic species of 

 Oxalis, all three forms of flower occur in the cleistogamous as well as in the 

 normal flowers. Although self-pollination of the chasmogamous flowers is 

 almost totally sterile, yet the habitual self-pollination of the cleistogamous 

 flowers is fully fertile. 



