BANUNGULAGEJE. 



55 



stamens are very numerous, free, and liypogynous ; each consists of 

 a filament with a club-shaped swelling below its summit, which 



Thalictrum aquilegifolium . 



Fig. 97. 

 Flower. 



Fig. 98. 



Longitudinal section of liowcr. 



tapers to a point' to support the two-celled basifixed anther that 

 dehisces marginally or somewhat internally by two longitudinal 

 clefts.- The gynseceum is formed of an indefinite number of free 

 carpels,-^ each borne on a slender sialk,' 

 and composed of an ovary which tapers 

 above into a beak grooved longitudinally 

 along the inner angle. The thickened and 

 everted edges of this groove are covered 

 with stigmatic papilla?. In the inner 

 angle of the cell of the ovary is a single 

 ovule, suspended, with its raphe dorsal, 

 and the micropyle looking upwards and 

 inwards.'^ The fruit consists of several 

 achenes (fig. 99), whose form varies ac- 

 cording to the species;" and in the suspended seed (fig. 100) we 

 find copious albumen, with a small embrvo near the summit. 



Thalictrum elattivi. 

 Fig. 99. Fig. 100. 



Fruit. Longitudinal section 



of flower. 



* The filaments are here erect and divergent. 

 In many species they are slender and capillary, 

 and yet very long ; so that after the flower has 

 opened they hang down the pedicel. 



^ The anthers are sometimes apiculate, as in 

 T. sylvaticum KoCH. When the dehiscence is 

 quite marginal, as in C. exallalum C. A. Met., 

 majiis MuKK., &c., each cell opens out into two 

 equal panels, which spread out and become plane, 

 so as to touch the walls of the neighbouring cell. 

 Each anther has then tlie form of a double flat 

 plate placed edgewise, and.as it were,in alinewitli 

 a radius of the flower; and all that is seen of it 

 is the inside of its cells, from which all the pollen 



falls oiF. Sometimes cultivation transforms some 

 of the outer stamens into petals, as in Clematis. 



3 When the carpels are four in number, like 

 the sepals, it may happen that the former seora 

 to alternate exactly with the latter. Eut this 

 relation is certainly not constant. 



•' This stalk is wanting in nil the species of 

 which De ('andolle makes his section iii.. 

 Euthalictnim {Syst., i. 172 ; Frodr., i. 12). 



•* This ovule has two coats. Above it may bo 

 seen a slight swelling of the placental lobes, but 

 not sterile ovules in two vertical rows. 



'■' The form of these aclieues is the i-bief foundii- 

 liou of the division of the genus into sections as 



