The Embryo-Sac of Angiosperms. 385 
“ it is quite impossible sometimes to be certain whether the 
structures present at the time of fertilisation are the products of a 
single embryo-sac, or of two.” 1 These facts, in conjunction with 
Campbell’s figures (notably those in which several young embryo- 
sacs form a linear row), strongly suggest that the “ sporogenous 
cells ” are the products of the division of one or more megaspore- 
mother-cells, and that cases where an increased number of nuclei 
are seen, result from the germination of more than one spore to 
form a single embryo-sac. 
There remains to be considered one more set of results, which 
all writers fortunately agree in regarding as revealing reduction- 
phenomena. Two cases have been described in which there are 
only two free nuclear divisions within the embryo-sac, which thus 
contains only four nuclei. One is that of the orchid Cypripediuni 1 , 
where two of the nuclei form cells round themselves and function 
as the egg and an attendant synergid, while the remaining two form 
the primary endosperm nucleus. (In this case the mother-cell 
divides only once; so according to Pace, who holds Coulter’s view, 
two megaspore-nuclei are included in the sac, each of which divides 
once). The other case is that of three genera of the order 
Onagraceas 3 , where three of the nuclei group themselves as an egg- 
apparatus, while the fourth functions as the primary endosperm 
nucleus. Here the embryo-sac is one of a row of four megaspores, 
and it may be remarked that if each of these megaspores were to 
germinate thus and become included in the one embryo-sac, the 
result would probably be a sac such as is seen in the Penasaceas. 
Another case which may be noted here is that described in two 
genera of the Podostemaceae by Went. 4 In these, the mother-cell 
divides once; the upper daughter-cell then degenerates, and the 
lower divides again. Those who hold Coulter’s view will consider 
that the two nuclei formed by the latter division are those of two 
megaspores, and this explanation is supported by their behaviour, 
for the lower one shrinks to a shapeless chromatin mass, while the 
upper divides to form a four-nucleate embryo-sac. Went, however, 
considers that both these nuclei belong to the embryo-sac, and that 
the division which forms them is the first in the sac. Of the four 
nuclei produced by the division of the upper nucleus, three form an 
egg-apparatus and the fourth degenerates so that no endosperm is 
formed. In Limnocharis 5 and Helosis n , (in both of which, it may be 
significant to note, no row of spores is formed), after the first 
division in the-mother cell the upper nucleus divides twice to form 
the egg-apparatus and single primary endosperm nucleus, while the 
lower nucleus quickly degenerates. 
It is apparent that all the Angiosperm embryo-sacs which have 
hitherto been regarded as primitive are now under suspicion. The 
discovery of an undoubtedly primitive type of embryo-sac — a type 
which, to satisfy all workers, would need to show more than five 
successive nuclear divisions from mother-cell to egg — would be 
especially welcome at the present time, when an important new 
hypothesis waits to be tested. It is difficult to know where to look 
for such a sac, as most of the genera which on other grounds 
might be considered as primitive have already been investigated. 
1 Campbell,1903, p. 671. 2 Pace, 1907. 3 Modilewski, 1909. 
* Went, 1908. 6 Hall, 1902. 0 Chodat and Bernard, 1900. 
