356 BOTANICAL GAZETTE [OCTOBER 
little before the beginning of nuclear division in the embryo cell 
proper. The extent of tissue development in B. Championtt may 
be judged by Ernst and BERNARD’s statement that there are 6-8 
cells in the median longitudinal line in the mature sac, and that 
B. coelestis has about 30 endosperm cells at maturity. 
The antipodal cells, never conspicuous, usually appear in a 
little V-shaped region below the haustorium region, sometimes as 
a row of cells, more frequently as two cells above one. 
The cell giving rise to the embryo, whether after fertilization 
or not, goes through at least one nuclear division, and usually more. 
Gonyanthes candida, as reported by TREUB (13), develops a 2-celled 
embryo; as reported by Jonow, and again by Ernst and BERNARD 
(as B. candida), it has a 3-celled embryo. Jonow (8) found in 
Gymnosiphon tenellus a 3-celled situation similar to B. candida, and 
in Dictyostegia orobanchioides and A pteria setacea a 4-celled embryo, 
comparable to that found in B. javanica by TREUB (13). Gymmno- 
siphon trinitatis (8) and Thismia javanica (3) show slightly greater 
development in a 6 or more-celled embryo, whereas Thismia clan- 
destina (4) shows the greatest differentiation in a structure con- 
sisting of a 3-celled suspensor and a spherical body in which a single 
outermost layer of cells is differentiated from the inner mass. There 
is a striking similarity to Orchidaceae (12) so far as extent of 
development of the embryo is concerned. The contrast in the 
mature seed, on the other hand, due to failure of endosperm devel- 
opment in Orchidaceae, is equally noticeable. Jouow, and later 
Ernst and BERNARD, have described the development of a small 
“nucellus polster’”’ above the embryo sac, and an even more con- 
spicuous tissue at the chalazal end. The possibility of the func- 
tioning of the latter at the time of germination of the seed as a 
region of water transfer (the rest of the tissue shows great cutiniza- 
tion) has been suggested, although no evidence of experimental 
character has been forthcoming. In contrast to the striking 
nucellus tissue at the ends, there is very evident degeneration of the 
cells in the middle zone or ring, as in Gymnosiphon, Burmannia 
candida, and Thismia clandestina. 
In comparison with the thorough work done on embryo sacs, 
the scant attention paid to the pollen situation brings forth prac- 
