18J 



of these " appendages " beiog in fact the more or less corollinic 

 tops of the carpels. In view of this the stipule-hypothesis must 

 be abandoned, for the variable number of alae is no longer of 

 any use, neither is fission, which has only been properly ob- 

 served in doubling /^e/rt/i' (XI, p. 64) nor the limited number of prim- 

 ordia. As to this point it seems doubtful whether Eichler's 

 observations have been sufficiently minute, when considering 

 Payer's observation upon the origin of x confronted with that 



of ElCHLER. 



Eichler's conception of the style originating from only one 

 carpel has been unjustly attacked by Schumann. Both Schumann 

 and Payer found three protuberances springing from the top 

 of the ovary and coalescing to a " chimney ". This is the style, 

 they say. No, it is the honey-containing flowertube on the 

 outerwall to which the labellum and ,v are to be attached and 

 also the inner ala a. It is actually the flowertube terminating 

 in three tips, one of which lengthens to a (winged) style whereas 

 the other two stiffen to rudiments. 



As to the pretended homodromy enough has already been 

 said, only it may here be added that Eichler, who also recog- 

 nised the Marantaceae as the nearest relatives of Canna, would 

 be still more convinced of this affinity, had he been able to 

 explain the apparent but not real deviation from antidromy. 

 I have as yet not had an opportunity to study the flowers of 

 the said Marantaceae but believe, from some points indicated 

 by Eichler and other authors, that much of what is said on 

 Canna will prove applicable to them also. 



Formula and diagram. 



After the above descriptions and considerations it hardly 

 need be demonstrated that the formula of the " quadri-alatae " 

 completely corresponds with that of Amaryllidaceae and other 

 typical Monocotyledons. There is no part wanting nor super- 

 numerary, only one, viz. y (or in the second flower ^?) doubled 

 to ala proper and stamen. 



With the diagram it is somewhat different. Taking fig. 18 



