408 159. anAMINE RF. 
1. Arundo Donax, Linn. Sp. Pl. p. 81; Lour. Fl. Cochinch. 
p. 45; Steud. Syn. Gram. p. 193; Hack. in Bull. Herb. Boiss. 
vil. (1899) p. 705; Hook. f. Fi. Brit. Ind. vii. p. 302. 
Arundo bengalensis, Retz. Obs. v. p. 905 A. Henry, in Trans, Asiat, 
Soc. Jap. xxiv., Suppl. (1896) p. 100. 
Arundo bifaria, Retz. Obs. iv. p. 21. 
Amphidonax bifaria, Nees, ev Steud, Syn, Gram. p. 197; Miq. in 
Ann. Mus. Bot. Lugd.-Bat. ii. p. 278; Franch, et Savat. Enum. Pl. Jap. 
ii. p. 171. 
Yunnan: 8-10 ft. high, Mengtze at 4600 ft. (A. Henry, 
9660 A !); Kwanaruna: by North River, Canton (Nevin and 
Graves, in herb. Hance, 19303 in part!), Amoy (Hance, 1405 !), 
Whampoa (Hance, 5501!). Mus. Brit.; Herb. Kew. 
Mountains of India; Europe, Northern Africa, Temperate 
Asia. 
Var. coleotricha, Jack. in. Bull. Herb. Boiss. vii. (1899) 
p. 724. 
Formosa: Tamsui (Makino, fide Hackel). 
I have seen no specimen. 
2. Arundo formosana, Hack. in Bull. Herb. Boiss. vii. (1899) 
p. 724. 
Formosa: Takow, Apes’ Hill (4. Henry, 1153, 1153 A! 
Playfair, 20901, Bankinsing mountains (A. Henry, 1153 B!) 
Tamsui, sandhills (Hancock, SI, Mus. Brit. ; Herb. Kew. 
Var. gracilis, Hack. l e. 
Formosa: Tamsui (A. Henry, 1462 !). Herb. Kew. 
I have not seen Professor Hackel’s specimens and have 
referred the above to his species añd variety on account of 
a general agreement with the descriptions. There is, however, 
one serious difference. Hackel describes the sterile glumes in 
his species as one-nerved, a remarkable character, whereas in 
the specimens which L have seen the sterile glumes are three- 
nerved as in other species of the genus, The spikelets are 
moreover two- to three-flowered, and show a wider range in size 
than is indicated in Hackel’s description, which was based on a 
single specimen. The spikelets in the species are 2—4-flowered, and 
21-4 lines long, the lower sterile glume ranges from 21-21 lines, 
the upper from 23-3 lines, and the lowest fertile glume from 
2-23 lines long, with an awn 1{-13 lines long. The spikelets 
