J> 



1920] SCHNEIDER, NOTES ON AMERICAN WILLOWS. VH 169 



for the type. It is but a fruiting ament with its peduncle which bears two 

 leaflets. The second was found by Drummond '' between N.Y. and Cum- 

 berland House." It has male aments, and it is not mentioned by Bebb. 

 The third is a sterile branchlet collected in 1825 by Drummond at Cum- 

 berland House. This piece apparently is identical with the specimen at 

 Kew where the sheet bears still another fragment with a fruiting ament. 

 By a hand unknown to me it is marked *' cordata Muhl. fide Ands. 1857, 



In 1867, Andersson described his S. pyrifolia from specimens collected 

 " in America boreali ad lacum Winipeg (in paludibus prope Brochet) et in 

 regione fl. Saskatchavan (Bourgeau)." Neither in 1867 nor in 1868 does 

 Andersson make mention of the piece of Drummond which he in 1858 

 named S. cordata. In the Gray Herbarium there is a male piece of Bour- 

 geau from " Winipeg Valley " which Andersson himself marked "S. py- 

 rolaefolia Ledeb," He apparently made this determination before he de- 

 cided to describe a new species because in 1867 he states that S. pyrifolia 

 is ** sine dubio S. pyrolifoliae asiaticae in America analoga.'* 



The name pyrifolia has already been used by Schleicher (Cat, PL Helv. ed. 

 3, 26 [1815]) but this is anomen nudum, and according to the international 

 rules Andersson's name can stand. Nelson's name S. columbiae has to be 

 adopted by those who follow the Philadelphia Code, but this is a very un- 

 suitable name because the species is wanting or at least extremely rare in 

 British Columbia and not known from the region of the Columbia River 

 at alL Nelson probably has been misled by Ball's use of the name S. py- 

 rifolia (apud Coulter & Nelson, New Man, Rocky Mts. Bot, 133 [1909]) for 

 a species of sect. Cordatae for which he has since proposed a new name 

 which unfortunately has not yet been published. As Fernald already ex- 

 plained (in Rhodoraxvi. 116 [1914]) '' the highly appropriate and long used 

 name, S. balsamifera Barratt, was published as a specific name merely in 

 synonymy and was not validated as a specific name until 1879, when Bebb 

 brought it forward (Bot. Gaz. iv. 190), In the mean time S. pyrifolia An- 

 dersson was properly published , , , in 1867 and as the first valid specific 

 name must stand. 



Andersson . . in 1867, also proposed a S. pyrifolia *obscura citing as type 

 " in Rocky Mountains (Bourgeau)." This name is mentioned by Ball, 

 apud Coulter & Nelson, 1. c. 133 (1909) with the synonym S, rotundifolia 

 Nutt, Later, as stated above. Ball gave to his pyrifolia a new name, and 

 he at present regards Andersson's var. obscura as a form of his new species. 

 In my opinion Andersson's form is really an obscure one, which cannot be 

 properly interpreted without seeing the type specimen. In 1868, Andersson 

 gives the type locality as " vallee des Arcs (August, 1858: Bourgeau) " and 

 says: " — forma jS. obscura est pyrolaefolia adhuc similior, sed capsulis 

 longius pedicellatis, obtusiusculis, stylo multo productiori et foliis subovali- 

 bus differt. potius cum S. cordata comparanda; folia autem subtus viridula, 

 i.e. concoloria, breviora et latiora, et amenta affinitatem cum vera pyrifolia 

 demonstrant." I cannot find a mention of a locality " vallee des Arcs " in 

 Palliser's Papers Explor. Brit. N. Am. (1859) where Bourgeau's route is not 



5 J 



