412 ROBINSON. 
There is clearly no sufficient description for publication of the 
name Carina glauca, but this had already been accomplished by 
Linnaeus.^ Neither is Phrynium spicatum published, but this is 
its first appearance, and Roxburgh himself subsequently ^ printed 
for it a sufficient diagnosis. 
The "3" after Phrynium dichotomum refers to a footnote, 
which is *'S. Seetwla-pwtrika. H. A. 4. t. 7." This enables the 
insufficient information given in the text to be supplemented by 
the figure cited in Rumphius' Herbarium Amboinense, which is 
accompanied by text, and is an ample description of the species. 
But Roxburgh had already * published this name. 
Similarly, the "5" after Gratiola amara, points to the footnote 
"Karanga amara. Vahl, H. A. 5. t, 170. f. 1." This is the first 
appearance of Gratiola amara, and it is here published. Vahl 
actually wrote the generic name as Caranga,^ and Jussieu as 
Curanga;^ both cited the figure of Herbarium Amboinense. 
There are ninety-one cases, approximately of the same nature 
as the last of these, where a new binomial is published, by cita- 
tion either of a previous binomial, or of a sufficient description 
with or without a plate. Admittedly, some of those referring 
to a previous binomial do so merely by inference, the generic 
name only being given. However, even these would be con- 
sidered to be published, by almost all botanists, when the specific 
name was used by the author cited under the given generic 
name. As it happens, this is not always true of the new com- 
binations made in such fashion by Roxburgh; some, therefore, 
must be rejected as unpubhshed; and no other of the names 
in the Hortus Bengalensis based on a binomial of earlier authors 
has stood the test of time. Oldenlandia herbacea may prove to 
be an exception. 
It may fairly be questioned whether there was any intention 
of publishing species in the Hortus Bengalensis, but Carey, at 
least, thought that this was being done. "The number of de- 
scribed species now in the garden amounts to about Three Thou- 
sand Five Hundred; for the knowledge of no fewer than One 
Thousand Five Hundred and Ten of which, as named and de- 
scribed in this catalogue, we are indebted to his (Roxburgh's) 
mdefatigable and discriminating researches." There is much 
reason to believe, if from nothing else than the title of the second 
' Sp. PI. (1753) 1. 
•Fl. Ind. 1 (1820) 5. 
*As. Res. 11 (1810) 324. 
■Enum. PI. 1 (1805) 100. 
•Ann. Mus. Hist. Nat. 9 (1807) 319. 
