Takeda— Old World Species of the Genus Mahonia. 213 



As regards the classification of the genus mention must first 

 be made of the subdivisions proposed by Fedde. He divided the 

 genus Mahonia into four groups, three of which occur in America, 

 while the fourth, Longibracteatae, is represented mainly in the 

 Old World, and with it only are we concerned at present. Fedde 

 distinguishes the Asiatic species in the first place by the nature 

 of the leaflets and particularly by the teeth— whether these occur 

 along the whole length of the margin or are confined to the upper 

 region of the leaflet.* 



Gagnepain, who paid special attention to the Asiatic species, 

 points out that the size, length, consistency, teeth, etc., of a 

 leaflet are subject to variation and are not definite, so that these 

 characters cannot be made use of in an analytical key of the 

 species. f He further mentions the important features exhibited 

 by the sepals, petals, stamens, and ovary. The same author 

 gives a key to the Asiatic species based upon these characters 

 combined with certain others. In the first place he distinguishes 

 two subdivisions by means of the stamens 4 



Schneider, on the other hand, distinguishes two primary 

 groups by the ovary— whether the ovary is provided with a 

 distinct style or not.§ Other organs made use of by this botanist 

 are the leaves, pedicels, bracts, and occasionally the number of 

 ovules contained in the ovary. 



The writer is of opinion that most of the characters above 

 mentioned, except perhaps the consistency of the leaflet and the 

 number of ovules, are relatively reliable, and may be regarded 

 as good diagnostic characters. It is however difficult to decide 

 which of these organs should be reckoned of primary importance ' 

 as showing actual relationship between each species. It appears 

 that Schneider's arrangement based on the presence or absence 

 of style is more natural than Gagnepain's system. It is, as is 

 pointed out above, not always easy to make out whether a 

 stigma is actually sessile or shortly stalked. For instance, 

 Schneider himself places his M. Zemanii {=M. confusa, Sprague), 

 which possesses a practically sessile stigma, in his first division, 

 the members of which are characterised by having a distinct style. 

 This method is therefore not very easy in practice especially when 

 fruit is lacking. It must however be admittted that according 

 to this classification closely related species, such as M. Fortunei 

 and M. confusa, would fall into the same group, whereas they 

 would be distantly separated if the stamens were made the 

 primarv principle for subdivision. 



The writer regrets not being in a position to express more 



* Fedde, I.e. pp. 78, 79. t Gagnepain, I.e. p. 133. 



J Gagnepain, I.e. p. 134. 



§ Schneider, in Sargent, PI. Wilsonianae, i, p. 380. 191 3- 



