

TV 



392 



C. 0.-Rosendahl. . 



to those in the other series. The 



stigmas are 



either small, entire and 



pointed or else more or less prominently 2 — 4-lobed (fig. 8). 



To these characters should also be added the one of short, thick, 

 nearly always ebracteolate flower-pedicels of the section Eumitella^ and 

 the slender, more or less elongated, bi-bracteolate flower-pedicels and fre- 

 quent occurrence of 2-flowered cymes of the section Mitellastra. 



h 



Fig. 



J. 



Stamens in the Sect. Etmiitella. a M. diphylla, b M. stauropetala^ c and d 



M, stauropetala var. stenopetala^ e and f M. irifida, g M, trifula var. violacea, 



h M. diver si folia, X ^0. 



It would seem from all this that when as many as four distinct 

 floral structures coincide in the remarkable manner in which they are 

 shown to do in these two series, there can be no doubt about the genetic 

 development following the two main lines indicated in our diagram (fig. D). 

 It is on the ground of the combination of so many fundamental cha- 

 racters as this that a new alignment of the species of the genus Mitella is 

 proposed. As already pointed out, the division of the genus has been made 

 purely upon the number and position of the stamens, with the result that 

 the sections have run squarely across the lines of genetic development 

 instead of paralleling them. There are, therefore, not only no adequate 

 reasons for dividing the genus up into five distinct sections, but such 



Fig. 6. Stamens in the Sect. Mitellastra. a M, mida, h M. cauleacens, e M. Breivert: 



d M. ovalis, e M. peiitandra^ f M paticifloray y M, japonica. X ^t*- 



divisions also violate or ignore the true relationship existing between the 

 species. To us it appears most natural to regard M. diphylla und M. 

 nuda as two present-day basic types that have already diverged from 

 some common ancestral type which has disappeared. They are about on 

 the same level of development. They approach each other in structure 

 of the floral axis and in the stamen number, but in other respects they 

 diverge. From them as starting points have been developed in connected 

 sequence all the living species. 



In one direction M, diphylla has given rise to the M. trifida group 

 m about the following manner. A deepening of the floral axis has taken 



t 1 









