336 PROCEEDINGS OF THE AMERICAN ACADEMY 



A perfect agreement of opinion respectinj^ the roses of America is 

 therefore not to be expected. In the following revision eighteen 

 species are recognized. Some of these are very clearly defined, while 

 others are more nearly related and are united more or less closely by 

 intermediate forms, or diverge into extremes in one or more directions. 

 Tliere has seemed to be no necessity for proposing a single new spe- 

 cies, though it would have been easy to increase the number largely. 

 But if farther subdivision were attempted, it would be difficult to say 

 where it should stof). 



In the general grouping I have followed nearly that of Crepin. 

 The species divide naturally into two series, one having the sepals 

 persistent or breaking away at length by an irregular rupture above 

 the base, in the other the sepals deciduous from the receptacle by a 

 clean circumscission at the base. The first divides again into two groups, 

 the one without and the other with infrastipular spines. The unarmed 

 group ranges from extreme Northern Alaska to Hudson's Bay, New- 

 foundland, and the northern border of the Atlantic States, and south- 

 ward along the Rocky Mountains to Colorado, as also westward 

 through Northern Asia and Europe to Scandinavia. This may be con- 

 sidered all one species (R. blanda), or it can be divided into four fairly 

 well-defined species, which may well be believed to be derivatives 

 from one common stock. The second group ranges from Alaska along 

 the coast to Lower California, and through the interior to Western 

 Texas and the plains east of the Rocky Mountains. The extreme 

 northwestern and southwestern forms (R. Nuthana and R. minutifolia) 

 are well defined, the latter remarkably so. The four remaining spe- 

 cies are closely related and run into each other more or less, the most 

 eastern one showing in its lobed sepals a gradation toward the Atlantic 

 Carolina group. 



The second series, with deciduous sepals, has two strongly charac- 

 terized outlying members, the eastern climbing species with connate 

 styles {R. setigera), nnd the western R. gymnocarpa, with its recepta- 

 cle becoming perfectly naked and closed in fruit. Two other related 

 but apparently distinct species occur in the extreme Southwest and 

 in adjacent Mexico ; but the four remaining species, occupying the 

 region east of the Mississippi, are again very close allies. 



The extreme of possible reduction would seem therefore to be to 

 nine species, viz. : — 



1. R. blanda. (/?. acicularis, R. Sayi, R. Arlcansana.) 



2. R. Nutkana. 



3. R. "Woodsii. {R. Californica, R. Fendleri, R. pisocarpa.) 



