229 



man* added a third, Ivadescantia pilosa. With the exception of 

 Waker all these authors used the same two specific names, apply- 

 ing Ventenat's Iradescantia rosea properly and making Tradescantia 

 Virguiiana elastic enough to embrace everything else savoring of 

 Tradescantia that existed in their respective regions. Walter ap- 

 plied the name Virginica not to the Linnaean type, but to the form 

 that Ventenat later described as Tradescantia rosea and proposed 

 the name cristata for one of the larger forms which most authors 

 thought was the real TradescantiaVirginiana of Linnaeus, but judg- 

 ing from Walter's description I am inclined to think he had in 

 mind a later described species. 



Rafinesque's work on the southern Tradescantias must needs be 

 mentioned. This eccentric author described no less than twenty- 

 si-K species and varieties in eastern North America, thus treating 

 the genus from the standpoint of extreme segregation just as the 

 authors mentioned above treated it from the standpoint of ex- 

 treme aggregation. Rafinesque apparently founded a species on 

 nearly every specimen he collected and of course his work needs 

 extensive reduction, but to what previously described species to 

 refer many of the Rafinesquian names is a difficult task. How- 

 ever, several of the forms he described, prove to be excellent 

 species, for example, Tradescantia brevicanlis which Dr. Morong 

 restored several years ago f and Tradescantia refiexa which I re- 

 store in this paper. 



An attempt to segregate the material in an herbarium on the 

 lines laid down in the several different works above referred to 

 must at once prove futile and not until we recognize the several 

 segregates into which the Viigijiiana type naturally separates it- 

 self can we hope for a clear or scientific interpretation of the 

 group from a specific standpoint. 



Key to the Species. 



Umbel like cymes peduncled, subtended by small or minute bracts. 



Leaves ovate or ovate-lanceolate; sepals 2-3 mm. long. i. T. Floridana. 



Leaves linear or almost filiform ; sepals 5-6 mm. long. 2. T. rosea. 



Umbel like cymes sessile, subtended by large leaf like bracts. 



Leaves linear or linear-lanceolate, 12-50 times longer than broad, more or less 

 involute ; plants glabrous, villous, hirsute or glandular. 



* Fl. S. States, 498. 



f Bull. Torr. Club, 20: 470. 



