ROXBURGH'S HORTUS BENGALENSIS. 



415 



Specter Pttrttkhed 



On Page 







69 





P^nH^n iiq fiirpatiic 



F CSHUCIII U9 1 Ul UC< IU3 



71 



H M 2 t R 



1— r Ct I IO OUI II VOCl 



71 



HA 5 f /5 / £ 



Calamus latifolius 



73 



H. M. 12. t. 65. 



Lycopodium aristatum 



75 



Dill. Muse. £. 65. /. 7. 



Ophioglossum filiforme 



75 



H. M. 12. t. 3Jt. 



Polypodium excavatum 



75 



H. A. 6. t. 35. f. 1. 



Pteris scandens 



75 



H. M. 12. t. 35. 



Eranthemum montanum 



80 



Justicia montana Cor. PI. 



Piper arborescens 



80 



H. A. 5. t. 28. f. 1. 



The following are at first sight in the same class, but must be 

 held unpublished, except possibly Pandanus Millore. 



. Name. Page. 

 Eteusine stricta 8 



Cerbera quaternifolia 19 



Crinum brevifolium 23 

 Hedysarum lagenaria 57 



Leucacephala graminifolia 68 



Leucacephala spathacea 68 

 Pandanus Millore 71 



Remarks. 



"Cynosurus Linn." There is no 

 Cynosurus strictus Linn. 



"H. A. 2. t. 363." There is no such 

 plate, and no other means of 

 identifying the species. 



"See asiaticum. Curt. Mag. 1073." 



"Aeschynomene L." There is no 

 Aeschynomene lagenaria Linn., 

 though there is one of Loureiro. 

 Roxburgh's name, at the utmost, 

 is a synonym. 



"Eriocaulon Linn." There is no 

 Eriocaulon graminifolium Linn., 

 nor was the generic name Leuca- 

 cephala ever published. 



Same as preceding. 



"Nicobar bread fruit tree, As. Res. 

 3. 161. seems only a variety of 

 odoratissima." 



A short supplementary list might be made of species where 

 Roxburgh's spelling differs somewhat strikingly from that of 

 preceding authors, but where he apparently had no intention of 

 forming new names. Perhaps the best cases of this are Jasmi- 

 num Zamfoac, Pettospermum Tohira, Laurus Culitlaban, Vitex 

 trifoliata, and Piper Malmaris. 



This list is based on the assumption that a published binomial 

 can never be used for any species other than that to which it 

 was originally applied, whether or not the name be valid in the 

 genus to which it was attributed or in any other to which it 

 may be transferred. There are about 185 additional names 

 in the Hortus Bengalensis which would be considered as published 

 therein, were it not that they had already been used. In nearly 

 all of these cases, Roxburgh was correctly using names which 



