HI. epipactidea Reichb.f., but the column and lip struc- 
ture is very different. Both this and AH. obovata Sum- 
merh., which is clearly its closest ally, have some of the 
characters of the genus Roeperocharis, notably the broad 
anther connective. In H. coeloglossoides, indeed, there 
is a papillose cushion near the base of each stigmatic arm, 
which may be looked upon as a reduced form of the upper 
stigmatic appendage in Roeperocharis. The rostellum, 
however, is quite different from that in the latter genus, 
while there are quite long anther canals and correspond- 
ing caudicles, features which are lacking in true Roeper- 
ocharis species. 
The specific epithet is given on account of the strong 
superficial resemblance to Cocloglossum viride (1.) 
Hartm., the British Frog Orchid. 
Habenaria gabonensis Reichenbach filius var. psi- 
loceras (Welw. ex Reichb.f.) Summerhayes comb. nov. 
Habenaria psiloceras Welwitsch ex Reichenbach filius 
in Flora 50 (1867) 99. 
In a previous note dealing with HZ. gabonensis (Bull. 
Mise. Inform. Kew (1988) 144) I considered H. psiloce- 
ras as conspecific. Since then I have seen material col- 
lected in eastern Angola by Mr. KE. Milne-Redhead which 
agrees very closely with Welwitsch’s original specimens. 
Both gatherings have spurs from 13-16 em. in length, 
whereas those in all the specimens of HI. gabonensis 
which I have seen never exceed 9 cm. In view of the 
identity in the other details of the floral structure it seems 
most suitable to consider HZ. psiloceras as a long-spurred 
variety of H. gabonensis and I am therefore treating it 
as such. The variety is restricted to Angola so far as is 
known at present. H. gabonensis occurs in the southern 
Cameroons, Gabon, Princes Island, and possibly also in 
Angola from which I have seen a plant, collected by 
[ 263 | 
