552 Robinson: Polycodium 



such as V. leucanthum Schlecht., V. stenophyllum Steud., and V. 

 cubense Griseb., none at all closely allied to Polycodium. This is 

 not all, for in the United States no single character is as certain 

 to ensure the instant identification of Polycodium as the long- 

 exserted anthers; moreover, the tubes forming the prolongation 

 of the anther cells are unusually long, both absolutely and rela- 

 tively to the anther cells. But there is a Mexican species, Vaccin- 

 ium Kunthianum Klotzsch, so closely related to Polycodium 

 stamineum that neither Kunth nor Dunal* thought it worthy of 

 specific rank. Its stamens were described as half-exserted, and 

 figured as well exserted, but in no collection that I have seen, 

 so identified by others or by myself, can they be considered as more 

 than barely exserted, the anthers are shorter than in the other 

 species, and the anther tubes only about one and a half times the 

 length of the anther cells. In all other respects it is a perfectly 

 good Polycodium, and if the genus is to be maintained, must be 

 transferred to it, forming a section by itself, on the basis of the 

 characters just stated. 



Finally, the anther awns are often revolute, but too much 

 emphasis should not be placed on this, as it is not always constant 

 within a single flower; at least, however, they are divaricate, but 

 so they are in species which no one has suggested separating from 

 Vaccinium, such as V. caespitosum Michx. 



The flowers of Polycodium are articulated with the pedicel; 

 this is also true of certain species placed in Vaccinium by most 

 authors, notably of the Disterigma species. That group, which 

 does not come north of Mexico, was until recently treated by all 

 authors as a section of Vaccinium, but Niedenzuf and HoroldJ 

 so far separate it from that genus that they place it in the Thibaud- 

 ieae. On the basis of floral characters there seems to be no 

 reason for so wide separation; indeed, unless Vaccinium is to be 

 radically divided, I at least believe that there is as much reason for 

 placing V. Myrtillus L. and its American allies in a different 

 genus from V. corymbosum L. and its allies, as there is for so segre- 

 gating Disterigma from the latter. Drude§ has placed considerable 



*H. B. K. Nov. Gen. & Sp. 3: 267. pi. 253. 1819; DC. Prodr. 7: 568. 1839. 



t Engl. Bot. Jahrb. 11: 209. 1889. 



X Engl. Bot. Jahrb. 42: 282. 1909. 



§ Engler & Prantl, Die Nat. Prlanzenfara. 4 1 : 32. 1889. 



