20 



The first thing to be noticed is the non-coincidence of the range of 

 variations of the anatomical and floral characters. When we put our new 

 genus Sasa under the subtribe Eubambuseas for the sake of its 6 stamens, 

 then the border-line drawn according to the difference of the root-structure 

 passes through between Sasa and Bambusa, cutting Eubambuseie into two 

 sections. Schwende n e r has pointed out a quite analogous case in his well- 

 known paper on the mcstom-sheath of the graminaceous leaves. 1} The 

 grouping of the tribes belonging to the Graminese into two subdivisions 

 according to the mestom -sheath, brings about the border-line passing through 

 the tribe Pa nice 03 ; a part of the tribe together with Maydese and Andro- 

 pogonea3 being sharply distinguished from the whole mass of the remaining 

 members of the grass-family by the absence of the mestom-sheath. 

 Schwendener has yet appropriately never regarded Paniceee as an un- 

 natural group, because the multitude of other anatomical characters proves 

 itself strongly in favor of the homogeneity of the tribe P What is then 

 the case with Eubambuseie ? It seems to us, however, decidedly unnatural 

 to put Sasa under Eubambusete, for not only the difference of the root- 

 structure, but the total sum of internal and external characters tends 

 evidently to outweigh the mere agreement in the number of stamens, which 

 is moreover not always free from variation. It must then be more reasonable 

 to exclude Sasa from Eubambusese and to introduce it into Arundinarieae. 

 It is a noteworthy fuct that Sasa borealis is very closely allied in its 

 vegetative characters to Arunduiaria japonicap an endemic species of that 

 phylogenetically old genu?. The relationship of the two is made more in- 

 timate by the circumstance that A. japonica has at times a greater number 

 of stamens than 3, and we have good reason to seek in this very point 

 the close connection between Sasa and Ammdinaria. We have schematised, 

 after careful observations of both floral and vegetative characters, the rela- 

 tionship among the spscies of our new genus in the following manner. 



1) Sell wend en er, Die Mestomseheide d. (xramineenblatter. Gesaimnelte Bot. Mittei* 

 lu 11 gen. Ed. II, p. 190. 



2) Schwendener, he. tit. p. 182. 



3)Makino, Bambusaceoe Japonica?. TKe Botanical Magazine, Tokyo, Vol. XIV, p. 21, 

 Arundinaria japonica—" Ya-dake." 



