180 



but in all other cases it was fully absent and the style thread- 

 like. On close examination it represented S in as many as 14 

 flowers, but in two of them S was rudimentary and s grown 

 out. A curious fact was that in one flower ,s'^ was of full growth 

 but united with an S which is also threadlike. In consequence 

 of unequal growth they had turned round one another in a 

 spiral, just like a Convolvulus and a thin thread. This was 

 the only case in which two styles came to development in the 

 same flower, in the other ones they took the shape of lacinia 

 as represented in the figures 19 a and b. By means of their 

 peculiar development they constitute a kind of coronula round 

 the style (fig. 20) which is connected to it but only neax the 

 bottom. The petaloid character which the (original) petals gave 

 up and which was transmitted to the stamens and the wmg 

 of the style reaches in our "third" flower even the upper portion 

 of the carpels. 



The hypothesis of the stipules (Stipulargebilde). 



As is generally known the hypothesis of stipular branching 

 has been devised by the famous author of the " Bliithendia- 

 gramme". According to his opinion he proclaims the alae as 

 produced through the fission of one primordium that in Diste- 

 mon only bifurcates to x and anther but in Canna proper 

 moreover yields one or more alae. Adhering to this hypothesis 

 many years ago I meant to advocate it by calling attention 

 to some appendages on the Inner^xdiQ of both labellum and x 

 (X). Both the hypothesis and its support now appear to be 

 worthless, Eichler's opinion because the alae /5, / and d take 

 the very place from which they must spring in relation to 

 the sepals and styles, their intermediate spaces being occupied 

 by a, labellum and x. There is no question of fission, at best 

 of notches at the tops and incisions of the margins. And as 

 to the support, one has only to cut up a flower vertically in 

 two symmetrical halves and once more to look at the append- 

 ages on the innerside of x and labellum, but now fully aware 



