175 



taken from the basis, nothing peculiar is to be observed, Imt 

 in B which corresponds with the middle, the dissepiments are 

 no longer radial but give the impression as though the axis 

 had been turned about in the direction of the little arrows 

 and dragged along the inner margins of the dissepiments. 

 Moreover these dissepiments are often arched, viz. with the 

 convexity in the direction of the arrows. From a number of 

 observations I am inclined to gather that the turning of the 

 axis takes place in the same direction as the succession of the 

 sepals. The latter may be to the right or to the left, whether 

 imbricate or in such a way that one covers at both sides, 

 one covers and is covered, one is covered on both sides. This 

 difference in aestivation characterises distinct flowers of the 

 same specimens but not distinct specimens. Comparing the two 

 flowers of one cicinnus in this respect it appears that in general 

 they show antidromy (with exceptions however) apart from the 

 fact that the first flower has an imbricate calyx and the second 

 may show one sepal overlapped at both sides (or in inverse 

 proportion). Schumann (Vll) has dealt with this matter elabor- 

 ately and supported his arguments with numeration, especially 

 with a view to prove that Eichler was wrong in his statement 

 that homodroiiiy should characterise the flower of Canna. In 

 how far this contest is well-founded will be apparent from 

 what follows. 



As to the corolla it is sure that in the first flower of a cicinnus 

 the parts show a right spiral line but in the second flower are turn- 

 ed to the left (apart in both flowers fi-om deviations in inverse 

 direction). As to the calyx and the corolla of the same flower 

 it can be stated that they turn their components in opposite 

 directions, but again with exceptions. 



The staminodes seem constantly to follow the same direction 

 and in the bud to overlap each other as the breadths of a 

 rolled-up umbrella, whether the flower be the first or the 

 second of the cicinnus. This direction is from right to left, 

 always supposing the observator in the axis. In keeping with 

 this arrangement is the anther which always springs from the 



