68 



brescen 



ARBORETUM [vol. n 



!ies rostraia. In 1868 Anders- 



son reduced his subspecies to varieties, and we find under var. /3 rostraia 



ifol 



ifol 



Gray in 1867 used the name S. livida occidenialis for S. rostrata. Bebb, 

 however, took up this name again in 1885 (in Coulter's Manual). Sargent, 

 in 1895, changed the name S. rostrata to S. Behhiana on account of the 

 older S. rostrata Thuillier (FI. Envir. Paris ed. 2, 517 [1797]). This new 

 name has been accepted by several later authors. Robinson and Fernald 

 (1909) keep Richardson's name, regarding S. rostrata Thuillier as a strict 

 synonym. This name, however, is not an unconditional synonym (as for 



longifolia 



A. & 



E.-G. Camus (Class. Saul. d'Europe i. 163 [1904]) mention S. rostrata 

 Thuillier in the synonymy of S. repens Linnaeus but again as a synonym of 

 their var. A vulgaris subvar. microphylla. Von Seemen (in Ascherson & 

 Graebener, Syn. Mittelcurop. Fl. iv. 127 [1909]) refers Thuillier's name as 

 a synonym to his S. repens B. rosmarini folia thus regarding it as identical 

 with a form different from that of Camus. It is, therefore, possible that 

 Thuillier's name may be used again as a specific name, and as the case 

 stands it seems not advisable to apply the name S. rostraia to an American 

 species. 



With regard to the variability of 5. Behhiana it ought to be said that the 

 floral characters as a whole seem to be very constant. The length of the 

 style varies to a certain degree but we should need the investigation of a 

 very large series of well-developed female specimens to decide whether 

 the forms with a more conspicuous style (almost equaling the length of the 

 stigmas) can be regarded as distinct. Fernald's var. projeda, a still very 

 little known variety, seems to be the only one of which the flowers differ 

 somewhat from those of the type, but here, too, the differences are not very 

 important. On the other hand the variability of the leaves is much greater 

 but it is extremely difficult to limit varieties. Andersson, apparently, did 

 not see copious material, and he was inclined to lay too much stress upon 

 certain variations which look very distinct as long as intermediate forms are 

 not seen. As I have already mentioned Andersson first considered 5. Bebbi- 

 a?ia as var. occidentab's of his 5. vagans 1, cinerascens. Later also he kept it 

 as a variety of S. vagans.^ I do not wish to-day to discuss the question 

 whether the American S. Behhiana is so closely related to any of the Euro- 

 pean-Asiatic forms which have been united under the name livida, depressa 

 or Starkeana that it, too, must be regarded as a variety or a subspecies of it. 

 I think it best to keep the American forms as a separate species. 



As already explained, Andersson, in 1867 and 1868, established quite a 

 number of forms of his subspecies or var. rostrata. It is strange that he 

 also has besides rostrata which only consists of American forms another 



1 This S. vagan.9 is the same as S. livida of Camus and S. depressa of Von Seemen. As I have 

 tried to show the oldest name for this species is S. Starkeana \Vil]denow, see Sargent V\ Wils 

 111. 151 (1916). 



