379 
The Embryo-Sac of Angiosperms. 
free nuclei function as eggs. This view would therefore entail 
giving up all connection between the Gnetales and the Angiosperms 
—a connection which recent work has shown to be probable. 
In the above brief summary no account has been made of 
certain exceptions to the general rule that the Angiosperm embryo- 
sac is bipolar and contains, at the end of free nuclear division, no 
more than eight nuclei. These exceptions will be discussed later ; 
it suffices at present to note that they have been used as instances 
in support of most of the theories mentioned above. It is both 
interesting and instructive to note how they have been used by 
various investigators to uphold the most fundamentally different 
hypotheses ; and the existence of such widely divergent explanations 
of the same set of facts shows how much some new data were needed 
to confirm or correct these theories. Such data have recently been 
furnished by Pearson’s researches on Welwitschia ,’ in which he has 
found a process of endosperm formation which can be closely 
paralleled with that seen in the Angiosperms. In the embryo-sac 
of Welwitschia, all the nuclei at the end of free nuclear division 
are potential gametes. Some of them remain free, and can function 
as gametes, but the majority fuse in groups of six to twenty to form 
a number of primary endosperm nuclei, by whose division the 
cellular endosperm is produced. Such a method of endosperm 
formation is unknown in any of the lower Gymnosperms, but there 
are indications that more detailed investigation will show that 
similar processes take place in Gneturn. 
An endosperm which originates from fusion nuclei at once 
suggests a comparison with that of the Angiosperms, and Pearson 
has suggested that the endosperm of the primitive Angiosperm 
was homologous with that of Welwitschia —that “in a broad sense, 
the Angiosperms and Gnetoideae ” ( Gneturn and Welwitschia) “are 
derived from the same stock,” the Gnetoidete being probably 
“ the last representatives of a race which separated from the main 
Angiosperm line.” 2 This implies that the embryo-sac of the 
primitive Angiosperm contained many free nuclei, all potential 
gametes, and a large number of primary endosperm nuclei, formed 
by the fusion of these gametes. In Welwitschia, free nuclear 
division stops at the eleventh generation from the megaspore, and 
in a similarly constituted sac, as a result of the progressive 
reduction that lias always marked the gametophyte, it is easy 
enough to imagine free nuclear division stopping at successively 
earlier stages, until it now ends as a rule at the third generation 
from the megaspore. With this reduction in the number of nuclei, 
the number of those available for fusion would decrease, until now 
there are only sufficient to form one primary endosperm nucleus. 
If this hypothesis is correct, the eight nuclei which now, in 
normal cases, represent this reduced gametophyte, are all to be 
regarded as potential or reduced gametes. In the great majority of 
cases, the egg is the only one now functional; nevertheless there 
are instances in which others of the eight nuclei have shown 
that they may possess similar potentialities. Thus several cases 
have been described in which the synergids have been fertilised and 
produced embryos 3 , and fertilisation of an antipodal has been seen 
1 Pearson, 1909. 2 Pearson, 1909, p. 386. 
* Dodel, 1891, ex Coulter and Chamberlain, 1903 p. 217 ; Overton, 
1891, ex Mann, 1892, p. 373 ; Guignard, 1901. 
