330 THE JOURNAL OF BOTANY 



l-3*5 cm. long. Sepals lanceolate to narrowly ovate, subacute to 

 acute, l-2-2'6 cm. long by 6-8 mm. broad; petals lanceolate to 

 rotund-ovate or -obovate, blunt, slightly shorter, subequal to, or 

 slightly longer than the sepals, white or purple, sometimes more 

 or less reduced in size or absent (T. Smallii Maxim.). Stamens 

 half or less than half the length of the petals (6-10 mm. long), 

 connective often shortly produced above the anther- cells ; stigmas 

 2 mm. long ; ovary pyramidal, longitudinally winged. Berry -8-2 cm. 

 in diameter ; seeds 2 mm. long. 



This approaches nearest to form (1) of T. erectum, which 

 resembles it in the small flowers, approach to equality between 

 filament and anther, short stigma, and relative length of stamens 

 and stigmas. The Asiatic specimens are generally much more 

 robust, with larger generally relatively broader leaves, and are 

 characterized by subequality between the petals and sepals and 

 the filaments and anthers. The petals are also more delicate in 

 texture. 



T. Smallii Maxim. (/. c. p. 862) seems to be a form having a 

 varying and unequal amount of reduction in the size of the petals, 

 which may even be altogether absent. 



I have seen the following specimens : — 



Himalaya. Sikkim, 10,000 ft. {Gamble, no. 643, and Pantling 

 in herb. Clarke, no. 46652) ; Bhotan {Grijfith, no. 5601). 



China. Hupeh {Henry, no. 6067 & 6067 B) ; West Szechuen 

 and Tibetan frontier {Pratt, no. 840) ; Szechwan, Mt. Omei, 8000 ft. 

 {Faber, no. 980). 



Japan. Hakodate {Maximou-icz) ; Nippon, Nikko {Tschonoski, 

 Bisset); Nagasaki (3/rt.i'i///c»u-/c5) ; Hakone (Z>/ss<'^) ; Chinsenji (Z>/s- 

 set); Miogisan {Bisset); Fujisan {Bisset); Central Mts., 2-7000 ft. 

 (Maries). 



10. T. GRANDiFLORUM Salisb. Parad. t. 1 (1805); Watson in Proc. 

 Amer. Acad. xiv. 274 (1879). 



T. rhomhoideum var. grandijiorum Mich. Fl. Bor. Amer. i. 216 

 (1803). 



We have no specimen from Salisbury, but among his drawings, 

 where also fragments of plants from which the sketches were made 

 are sometimes found, I find the originals of the dissections for the 

 plate in the Paradisus. There is evidence that the dissections on 

 the plates were generally copied from Salisbury's drawings, though 

 no intimation of this is given in the work. The plate is such an 

 excellent representation of the species, that there is no question as 

 to identity. 



Specimens show considerable variation in size of leaves and 

 flowers, length of pedicel, &c. The large petals are generally 

 obovate or oblauceolate, sometimes tending to broadly elliptical; 

 they are blunt, and longer than the sepals. The narrow filaments 

 are rather shorter than the linear anthers (from five-eighths to six- 

 sevenths of their length), and the anthers exceed the slender erect 

 or somewhat spreading stigmas, the latter, generally between 4 mm. 

 and 6 mm. long, but sometimes reaching 1 cm., are occasionally 

 coherent for a very short distance at the base. I find, as Salisbury 



