iqiq] SCHNEIDER— AMERICAN WILLOWS 345 



Ofv. K. Vet.-Akad. Forh. 15:116. 1858, ex parte; Macoun, Cat. 



Canad. PI. 450. 1883, ex parte; Sargent, Rep. For. Trees N. Am. 



loth Census U.S. 9:168. 1884, ex parte.— 5. fluviatilis Sargent 



in Gard. and For. 8:463. 1895, ex parte, non Nutt.; Rowlee 



in Bull. Torr. Bot. Club 27:254. 1900, ex parte; Henry, Fl. S. 



Br. Col. 97. 1915. — 5. interior Rowlee, I.e., 253, ex parte; Britt. 



and Br., 111. Fl. 1:595.. i9i3» ex parte. — S. linearifolia Rydbg. in 



Britton, Man. 316. 1901; Fl. Color. 94. 1906, ex parte; Fl. R. 



Mts. 192. 1917, ex parte; Small, Fl. S.E.U.S. 342. 1903, ex 



parte. — S. longifolia var. interior Jones, Willow Fam. 25. 1908, 



ex parte. — I have seen a photograph and fragments of the 



type of var. pedicellata, collected by E. Bourgeau, "Saskatchewan 



bords des Lacs, abondant, 21 Juin 1858" and preserved in 



Herb. K., and also of the type of 5. rubra Rich, from the 



''Mackenzie River." This specimen of Richardson's represents 



the same form as the material from "Cumberland House" 



in Saskatchewan, which is a syntype of S. linearifolia Rydbg. 



in Herb. N. This variety differs from typical S. longifolia 



chiefly in its narrower, linear leaves, and its glabrous ovaries, which 



are more or less sessile when young but usually distinctly pediceled 



when in fruit, the pedicels often being twice as long as the ventral 



gland. As previously stated, var. pedicellata is the prevailing form 



in the northwestern part of the range of 5. longifolia, but there are 



also forms near the southern limit of its habitat which can hardly be 



distinguished from var. pedicellata (for instance Munson's specimens 



from the Red River near Colbert's Ferry, north of Denison, Texas, 



April 19, 191 1, f., fr.; A.). 



As previously stated, the most southern form of S. longifolia 



is represented by var. angustissima And. (1858'^) with which I have 



dealt in Bot. Gaz. 65:26. 1918. Besides the Mexican specimens 



here cited, I refer the following to this variety, which seems too 



closely connected with the typical S. longifolia to be kept as a 



distinct species. 



Specimens examined. — Texas: without exact locality and date, Berlandier 

 (nos. 911, 2341, 2368, 3019, cotypes; G., M.; 1938, f.; M.; nos. 2341 and 



'3 Later, in Monogr. 1867 and in Prodr. 1868, Andersson used this name for dif- 

 ferent forms, partly belonging to S. sessilifolia var. Hindsiana, partly to 5. exigua 

 (probably var. stenophylla). 



