II GERANIACE^ 141 



flowers ; indeed, in the upland valleys of the Tyrol it is 

 said to produce cleistogamous flowers only, in which the 

 pollen grains are separate, not tied together by threads 

 as in the open flowers. In the ordinary ones such a 

 provision would be useless. According to Snyder and 

 Meehan, in the United States /. fulva is visited by 

 humming-birds. 



The capsule is five-chambered. The dividing walls 

 are thin, and eventually separate themselves from the 

 centre, which thus becomes a pillar standing in the 

 middle of the fruit. As the fruit dries the cells imme- 

 diately below the epidermis are in a state of gradually 

 increasing tension, more so than the layers below. 

 Moreover, while the carpels of Geranium are straight, 

 and thus assume a position like that of a watch-spring, 

 those of Impatiens turn slightly to one side (the right), 

 the result of which is that in contracting they resemble 

 a corkscrew. Finally, the fruit bursts, the valves roll 

 up suddenly like a watch-spring, and fly oS", carrying 

 the seeds with them. In this case, therefore, the 

 elastic tissue is part of the ovary — not, as in the pre- 

 ceding genus, the outer coating of the seed itself^ 



Moreover, while in Geranium the inner cells swell 

 and the valves curve outwards, in this case it is the outer 

 layer which is elastic and the valves curve inwards. It 

 is interesting and suggestive that our genera of this 

 family should present such different modes of arriving 

 at the same object. 



The leaves are thin, flat, and delicate, as in other 

 species which inhabit damp and shady localities (see 

 Dentaria, p. 80). 



' Zimmermann explained the deliiscence by the tension of the woody layer ; 

 Steinbrinck, by the difference between the tension of the woody layer and of the 

 outer epidermis, which is also Eichholz's view (Pringsheim's Jahrh. Wiss. Bot. 

 vol. xvii. (1886)). 



