BLAKE: NEOMILLSPAUGHIA 83 
shrub without tendrils; but Dammer has found what he considers 
a young and undeveloped tendril in a specimen of this species, 
and states that all the specimens known are merely pieces from 
the region of the ‘inflorescence, so that it is probable that this 
species agrees with the other better known members of the genus 
in the possession of tendrils. 
The descriptions of the embryo given by Bentham and Hooker 
(‘embryo in uno lobo leviter incumbenti-incurvus”’) and by Dam- 
mer, in the Pflanzenfamilien (‘‘E. incumbent, einwarts gekriimmt 
Oll’”’), do not entirely agree with what I have found in the dissec- 
tion of numerous seeds of B. cirrhosa. The albumen is six-sulcate 
by the intruded testa, otherwise not ruminate; the embryo is 
marginal in one of the lobes, straightish or very slightly incurved, 
and the radicle erect and in no way applied to the cotyledons. 
In all the specimens of the African species so far described the 
achenes, although in some apparently mature, have been empty of 
seed and filled with a fungal growth. 
2. ANTIGONON Endl. 
The perianth segments are given as five by Bentham and 
Hooker and by Baillon. Dammer, in the Pflanzenfamilien, gives 
them as five or six, and figures a flower of A. leptopus with six 
segments. I have seen no flower with six perianth parts, and 
believe this number must be very rare if not abnormal. The 
embryo, in the seeds examined, was subcentral, flat, and straight. 
3. GYMNOPODIUM Rolfe, Hook, Icon. 27: pl. 2699. 1901 
Millspaughia Robinson, Bot. Jahrb. 36 (Beibl. 80): 13. 1905. 
The filaments in this genus are normally nine, as described 
by Rolfe, the six outer inserted at the outer edge of a short thick- 
ened somehat knobby disk at the base of the ovary, the three 
inner borne on this disk opposite the sulcate faces of the ovary. 
They are free, in which the genus differs from Antigonon, where 
an annulus of basally united filaments is always developed. The 
other characters by which Gymnopodium differs from Antigonon 
have already been indicated above. The ovule in G. antigonoides 
is erect on a funicle several times its length, not subsessile as 
described by Rolfe in G. floribundum. The three species now 
known may be distinguished by the following key: 
