72 JOURNAL OF THE ARNOLD ARBORETUM [vol. u 



Santa Fe, September 18, 1916, A. Rehder (Nos. 604. 609. st.; A.; "shrub 10 feet"); 

 same place, September 4, 1894, C. S, Sargent (st., A.; "tree 20 X 1 ft., pendulous 

 branches, rough trunk, deeply furrowed dark brown bark"). 



forms 



seen 



the types the following may be said. The first variety is var. luxurians 

 (in Rhodoni, ix. 223 [1907]) w^hich "is clearly an extreme variation of the 

 common S. rostrata" It chiefly differs from it by its longer capsules (9 to 

 12 mm. long) and longer pedicels (5 to 8.5 mm. long). So far it has been 

 only seen "on banks of the St. Lawrence from Rimouski Co. to Gaspe Co." 

 in Quebec. It should be looked for in other localities of the type, and it ap- 

 parently is nothing but a forma luxurians. 



Fernald's 



ifolia 



"presents the most extreme development of pubescence in the species . . . 

 while var. perrostrata shows the opposite tendency." According to the 



material before me var. capreifolia is closely connected w^ith the type by 

 frequent intermediate forms. Fernald thinks that it **may prove to be 

 the same as S. vagaiis 1, rostrala forma laiifolia Andersson'* of which I have 

 already spoken. This is, however, in my opinion an uncertain form, and 

 may even belong to another species. To var. capreifolia Fernald refers 

 specimens from Newfoundland, eastern Quebec and Nova Scotia, and it 

 apparently has a wider distribution. 



Fernald states "in typical S, rostraia Richardson (S. Bebbiana Sargent) 

 the new branchlets are pubescent at tip, but the pubescence is early de- 

 ciduous/' In the specimen of Richardson which I regard as the co-type 

 (see p. 66) the branchlets of the season are distinctly pubescent, and only 

 somewhat glabrcscent toward the base, but even the branchlets of the 

 preceding year are at least partly pubescent. Very rarely these branchlets 

 are glabrous or almost so. The pubescence of the leaves, too, of typical 

 rostrala is hardly very different from that of var. capreifolia of which the 

 lower leaves also are partly glabrcscent, A closer study of the western 

 form may, ho^yever, lead to the hypothesis that there are glabrcscent and 

 pubescent forms of the type as well as of var, perrostrata. 



The most striking of Fernald's forms is var. projecta (in Rhodora, xvi. 178 

 [1914]) which so far is very incompletely known. It has only been found in 

 Newfoundland, Wild Cove, south of Bay of Islands, June 11, 1896, by 

 A. C. Waghorne (fr.; G.). According to Fernald it differs from all varieties 

 of S. Bebbiana "in the slender elongate ament, the long scales, the short 



capsules and pedicels shorter than the scales.*' Unfortunately there are 

 neither mature leaves nor well-ripened capsules. The male plant, too, is 

 unknown. Fernald says: "when better 

 tinct species.'* 



known 



nana 



(1858), quoad spccim. fem.; in Proc. Am. Acad. iv. 63 (Sal. Bor.-Am. 17) 



An 



VI. 86 (Monog. Salic.) (1867), excl. t. 5, fig. 50; in De CandoUe, Prodr. 

 XVI. 2 226 (1868). — Rydberg in Mem. N.Y. Bot. Gard. i. 114 (Cat. Fl. 



