173 



the ala; the latter, again bifurcating, sends a branch to the 

 stamen, from which clearly proves that ala y and stamen 

 belong to the same organ, thus both being portions of an 

 antisepalous staminode. This important fact is already fore- 

 shadowed in the flowertnbe (fig. 6) which in the sepalous 

 sections has three vascular bundles, whereas in the petalous 

 ones there are only two on the same radius. Although in two 

 of the sepalous sections the very tiny bundle does not induce 

 a branching of the ala, it may be yet admitted that there is 

 at least a slight indication of a filament. Only in the section 

 in which / is placed a real filament with anther comes to 

 proper development. 



After having goiifi through a series of mere facts and obser- 

 vations I feel it incumbent upon me to recall to mind that 

 Lestiboudois (I, p. 113 et seq.), the clever French morphologist, 

 who understood so well the flower of the Zinyibenweae, as early 

 as in 1829 described labellum, stamen and x of Canna as ^'trois 

 etamines interieures, dont deux steriles et une antherilere .... 

 Tune, sterile, revolutee, libre, repond exactement ^ un sepale 

 interne . . . les deux autres, I'une sterile et I'autre antheriferique, 

 sont iwesqu' entierement soudees entr^elles . . . ." 



That Lestiboudois allotted the fertile stamen to the inner 

 whorl is but of little account, being led to his opinion by his 

 looking upon the three alae, [3, y and also « as outer staminodes. 



Also Dickie's "Note on the characters of the genus Canna" 

 (IV) deserves honourable mention on account of both the 

 accuracy of his observations and of his correct conclusions: 

 1° that the anther is two-celled, 2° that x is to be considered 

 as a separate, petaloid stamen. He quotes in favour of this a 

 statement by A. Richard (Diet, des sc. nat.) which says that 

 the petaloid filament ought to be held as being formed by the 

 "union of two stamens, one producing pollen, the other abor- 

 tive, which is represented by the petaloid lamina, on one of 

 the sides of w^hich the anther is inserted.''' 



An indirect confirmation of the genetic independence of the 

 stamen is afforded by cases such as represented in fig. 16, in 



