Primitive Angiosperms. 135 
equally capable of fertilization, would represent an archegonium initial 
in some ancestor, more or less remote. As Miss Berridge remarks, this 
hypothesis fully justifies Dr. Lotsy’s assertion that the archegonium in 
Gnetum is reduced to a naked nucleus, and that reduction has in this respect 
proceeded further than in Angiosperms, where the ovum is a complete cell 
(55, p. 103 ). 
At first sight the interruption of prothallial formation by the act 
of fertilization, as described by Karsten (50) in other species of Gnetum , 
recalls the Angiospermous history. But there is no real resemblance. 
When the pollen-tube enters the embryo-sac of Gnetum , the endosperm 
consists of numerous free nuclei embedded in its parietal cytoplasm. After 
the epoch of fertilization these nuclei appear to divide ; cell-walls are formed 
round them ; and development proceeds in the usual way until the embryo- 
sac is filled with tissue. The homology with the endosperm of other 
Gymnosperms would be clear even without the link supplied by Lotsy’s 
description of Gnetum Gnemon (55). In the embryo-sac of Angiosperms, 
however, the whole endosperm begins with the mitosis of a fusion nucleus 
which contains a male element from the pollen-tube. Thus the endosperm 
is of necessity developed after fertilization, because the nucleus which gives 
rise to it is not complete until that epoch. No trace of such a process 
is found in Gnetum. The primary nucleus of the endosperm is not a 
fusion product. It has given rise .to a number of equivalent nuclei before 
fertilization occurs, and growth is resumed afterwards by the simultaneous 
activity of these nuclei, or the greater part of them. 
No tissue then in Gnetum , nor, so far as we know, in Welwitschia, can 
be considered as the direct representative of the Angiospermous endo- 
sperm. 
To sum up, the germination of the embryo-sac and the history of the 
endosperm isolate Monocotyledons and Dicotyledons from all other plants. 
The only adequate explanation of the identity of two processes so com- 
plicated in two separate races is inheritance of these features from an 
ancestor common to both. The alternative explanation is independent 
evolution of both members along distinct lines of descent, and to attain 
identity in that way would require a series of coincidences so improbable 
as to be inconceivable. 
The argument from these two features is very much stronger than the 
similar argument founded on the identity of the carpels in both classes, 
since the coincidences in structure are more numerous and more striking. 
The value of both arguments is much increased when they are considered 
together. 
Monocotyledons and Dicotyledons have another member of great 
importance in common — their flower. But whether a morphologist considers 
