142 Shaiv — Structure of the Flcnccrs in 



tion for any reason took place before the flower opened, the 

 corolla would cease developing. " I believe that cleistogamy 

 is caused by the physiological condition of great fertility 

 without crossing, coexisting with the morphological one, of 

 germination of the pollen while on the anther cell, or at least 

 before expansion of the perianth. The result of the latter 

 condition is the arrest of the floral envelopes." 



Since cleistogamic flowers are often buried in the earth, and 

 so produce fruits which are safe from the danger of being 

 nipped by herbivorous animals, the question of cleistogamy 

 has been identified with that of speciallj' protected forms of 

 fruit. Huth, ' speaking particularly of the case of Polygala 

 polygama, says : " One must look upon this case also as a 

 contrivance for protection against grazing animals." The 

 force of this explanation disappears when one takes into account 

 the aerial cleistogamic flowers. Polygala polygauia grows 

 largely where tender grasses are not abundant, and suffers 

 much from grazing animals. The green cleistogamic flowers, 

 growing among the foliage, are situated as unfavorably as 

 possible to escape these, and therefore Huth's statement 

 obviously fails to explain them. If any one chooses, he can 

 assume that the two kinds of minute flowers have two differ- 

 ent e.xplanations, that the subterranean ones are developed to 

 produce protected seeds, and the aerial ones for another pur- 

 pose. But if one believ-es, as does the writer, that one expla- 

 nation should chiefly cover the whole case in a single plant, 

 the idea of these flowers being developed for the sake of pro- 

 tection must be dropped, except as a secondary condition. 



Chodat,- also writing of Polygala polygauia in particular 

 speaks of it as a case of teratology, and differing from other 

 abnormalities only in the frequency of its occurrence. Is it 

 not difficult to suppose, either that this plant is essentially 



^ Loc. cit. 

 ' Loc. cit. 



