r^ 







376 



G. 0. Rosendahl. 



four genera and eighteen species exclusive of the Japanese ones. As this 

 latest work on Mitella adopts all the genera that taxonomists have at 

 different times proposed for the group, so also it utilizes in a similar 

 artificial manner the characters that have been employed for their sepa- 

 ration. An enumeration of these characters and the manner in which they 

 are used as the basis of classification will be taken up at this point before 

 proceeding to what we consider a more natural and consistent arrangement. 



The number and disposition of the stamens has been taken as of first 

 importance in the division and splitting-up of the genus. As the stamens 

 are either 10 or 5 in number it allows of three divisions being made; 

 one division with \ stamens, one with 5 stamens opposite the petals, and 

 one with 5 opposite the sepals. When it is borne in mind that plants 

 like M. nuda and M. caulescens^ which in every other way show the 

 closest possible natural relation, would be separated from each other by 

 this scheme, it is at once seen that such a character as stamen number 

 has no more significance here in defining affinities than in the Linnaean 

 artificial system. 



Furthermore, it often happens, as pointed out by Hooker i) and Piper 2), 

 that one or more stamens in the 5-stamen groups are transformed into 

 staminodia, a fact which further shows the inconstancy of stamen cha- 

 racters in this genus. 



The extent of fusion between the pistil and the axis has been con- 

 sidered as a character of sectional or generic merit In two of the most 

 closely related species of the genus, M. nuda and If. caulescens^ this 

 character does not hold good. In the former the ovary is free from the 

 axis to the base, whereas in the latter the ovary is half inferior. In the 

 section Eumitella the placentae have been regarded as more basal than 

 in the others. This is either an error in observation or a failure to re- 

 cognize the fact that modifications take place in the ripening fruit which 

 affect the apparent position of these structures. The placentae are in all 

 cases parietal but the ovary, varying in the extent of its fusion with the 

 axis, behaves differently in ripening into the capsule. Where it is free 

 or only slightly fused with the axis the enlargement of the capsule is 

 mainly above the placentae, making them appear relatively basal in the 

 fruit, and on the other hand where it is deeply sunk in the axis, the 

 growth of the capsule is more basal so that the ripening fruit evaginates 

 more or less completely causing the placentae to appear higher placed. 



Eumitella has also been characterized as having larger and fewer 

 ovules, but as this applies really only to if. diphylla it cannot be con- 

 sidered of any value in holding M. diphyUa and M. nuda together as 

 distinct from the others. 



^) Hooker, W. J., Fl. Bor. Am. vol. \ p. 241. 1840. 

 a) Piper, G. V., Erythea, vol. 7 p. 4 63. 1899. 





