174 
WALTER P. THOMPSON 
even beyond the endosperm into the tissue at the base of the nucellus. 
Figure 97 represents such a condition. I have never seen an embyro 
developing in this position, they were always within the endosperm. 
Before the embryo proper is formed the end of the suspensor en- 
larges and in the next stage that I have observed this swollen end con- 
tained four small nuclei (see figure 78). There was no wall separating 
the swollen end from the rest of the tube. From this structure the 
embryo is organized in a manner not determined. A young embryo 
consisting of a small group of cells irregularly arranged is shown in 
figure 80. 
In G. gnemon the pro-embryo differs from that of the other species 
in that only two cells are formed before the development of suspensors. 
The two-celled stage is shown in figure 70. I cannot state whether 
or not the dividing wall develops in connection with the division of 
the nucleus or is in the nature of a cleavage wall. Both cells develop 
into suspensors (figure 71), which grow down through the empty upper 
part of the sac to the developing endosperm which they penetrate. 
Figure 72 represents the suspensors growing towards the endosperm 
and figure 73 the same thing under greater magnification. I have 
not observed the later stages in this species. 
This account of the embryogeny of G. gnemon differs from that 
of Lotsy in regard to the division of the fertilized egg. Lotsy states 
that the fertilized egg itself elongates without division to form the 
suspensor. The explanation of the discrepancy in the two accounts 
is likely to be found in Lotsy's statement that both male cells always 
function in fertilization. Apparently what Lotsy considered to be 
two fertilized eggs, I have called a two-celled proembryo. In support 
of my view I would call attention to three points: (i) I have more 
than once observed one fertilized egg alone in a sac (this may, however, 
be interpreted as the failure of one of the male cells to function; but 
in that case, the statement that both male cells always function is 
incorrect) ; (2) the two cells are always in close contact and have 
every appearance of having resulted from the division of a single 
cell ; (3) in G. sp. jj several cells are undoubtedly formed before the 
suspensors elongate. 
Coulter (7) states that the suspensor within the endosperm is 
divided by transverse cleavage walls. While this may be true within 
the endosperm I have observed no such walls before the suspensor 
reaches that tissue. Furthermore in the other species no such walls 
