154 
WALTER P. THOMPSON 
The relationships of these envelopes to those of the typical flower 
were not understood. I have observed that every abortive ovulate 
flower at a certain stage in its development possesses a rudiment of a 
third envelope between the two well-developed ones (see fig. 26). 
It seems evident, therefore, that the envelopes present are the inner 
one (ovary) and the outer which we call perianth and that the middle 
envelope is represented onlygby the vestige present during development. 
The invariable presence of the middle envelope although in a rudi- 
mentary condition, together with the frequent presence of a typical 
embryo-sac and the rare development of fruit, shows that these 
flowers are quite homologous with the functional ovulate flowers. 
And this homology suggests that the immediate ancestors of Gnetum 
had bisporangiate strobili. In view of the conclusion that the flowers 
themselves were originally bisporangiate (page 18) and have been 
reduced to the monosporangiate condition, it is difficult to understand 
why the bisporangiate strobili should have been developed and then 
reduced. 
7. Male Gametophyte 
Until the publication of Pearson's (22) incomplete account of 
conditions in G. africanum our only knowledge of the male gameto- 
phyte of the genus was the statement of Lotsy (19) that the tube 
nucleus and two male cells of the ordinary type are in the pollen tube 
just before fertilization. The writer has been able to observe almost 
all stages in the development of the male gametophyte in G. gnemon, 
G. latifolium and G. sp. jj and several stages in other species. The 
different species are quite alike in all essential points. 
The young microspore on being freed from the cavity of the 
mother cell has a very thin wall in which only one layer can be distin- 
guished. Later a second layer appears and the typical exine and 
intine are present. The exine becomes covered with very small 
protuberances. 
The nucleus of the young microspore is large and rather dense 
and has a prominent nucleolus. This nucleus divides very soon after 
the microspore is freed. The first division does not give rise to a 
prothallial cell as in so many Gymnosperms but to a tube nucleus and 
generative cell (see fig. 29). No prothallial cells are formed. In 
this important respect, therefore, Gnetum has completely departed 
from the Gymnosperm condition and has arrived at the Angiosperm 
