170 
WALTER P. THOMPSON 
becomes reversed, the large end being below. The nucellus is grad- 
ually replaced. The originally expanded part of the sac enlarges 
very little, if any, and the nuclei and protoplasm within it disappear. 
The cells in this region become largely disorganized by the growth 
of the suspensors. 
It should be pointed out that considerable development has taken 
'place in this endosperm before fertilization occurs. While there is no 
cellular tissue in the sac at the entrance of the pollen tube, a con- 
siderable amount of it develops before the actual fusion of the sexual 
nuclei. In fact fertilization does not take place until the narrow part 
of the sac is filled with compartments and the group of cells have 
been formed in the expanded part of the sac. 
It is evident that the process of endosperm formation in G. sp. jj 
is more primitive than that of G. gnemon. In the first place the 
whole sac of the former species becomes divided into cells as in typical 
Gymnosperms and all the nuclei contribute to endosperm formation, 
while in G. gnemon only a few of the nuclei contribute to endosperm 
formation. In the second place the development of endosperm in 
G. sp. jj has gone much farther before fertilization than in G. gnemon. 
The important departures from the typical Gymnospermic method 
of endosperm formation are three in number: (i) The delay in cell 
formation until after the pollen tube has entered, (2) the fusion of 
nuclei in each cell, (3) the participation of only a few of the nuclei 
in endosperm formation {G. gnemon) . Now all these departures from 
the Gymnospermic method are approaches to the Angiospermic 
method. They will be discussed separately. 
I. In all other Gymnosperms a cellular endosperm is formed 
before the pollen tube enters the archegonium and the egg is fertilized. 
In Angiosperms the endosperm never forms until after the pollen 
tube enters the embryo-sac though it may begin to develop before 
actual fertilization takes place. In Gnetum one never finds endo- 
sperm before the entrance of the pollen tube. It is plain therefore 
that in this very important respect, Gnetum follows the Angiospermic 
method. But in G. sp. jj there is a reminiscence of the Gymnospermic 
method in that a considerable mass of endosperm is formed before 
the actual fusion of sexual nuclei occurs. In G. gnemon fertilization 
takes place sooner, though I have never seen a fertilized egg before a 
few cells were formed. In this connection it should be pointed out 
that in Casuarina, one of the lowest of the Angiosperms, Treub (31) 
