Cook : EMBRYOLOGY OF RHYTIDOPHYLLUM 18] 
was always present a single large pyriform cell ( figure 15) whose 
origin was difficult to determine, but it appeared to be the result 
of one of the early divisions of the endosperm. 
At the time of the primary division of the endosperm-nucleus 
some starch was present in the sac and also in the cells immedi- 
ately surrounding the micropyle, and in a very short time all the 
cells lining the sac developed starch. Before the disintegration of 
the endosperm was complete these nucellar cells were also under- 
going disintegration and this continued until the entire nucellus 
had disappeared. In brief, both the endosperm and the nucellus 
undergo disintegration to feed a rapidly growing embryo and 
eventually leave the embryo surrounded by the integuments only. 
When the embryo-sac is in the eight-nucleate stage (figure 1d) 
the nucellus consists of four to six layers of cells; as the ovule 
increases in size the cells divide, but the inner cells are continually 
disintegrating to feed the embryo. At about the time of the 
appearance of the cotyledons (figure 23a) the nucellus consists of 
four or five layers of cells (figure 16), of which the inner cells 
show the effects of disintegration. After this time no nucellar 
cells are formed, but those already existing are gradually used by 
the growing embryo (figure 25). 
The embryo. — The fertilized egg, by successive divisions, forms 
an embryo of from four to six cells in linear arrangement ( figures 
17, 18), with the basal cell very much elongated, but not enlarged 
as in the case of Capsella Bursa-pastoris as described by Coulter 
and Chamberlain (2) and Hanstein (3). The apical cell now 
divides by two longitudinal walls placed at right angles to each 
other, thus forming the quadrant stage (figure 77). The quad- 
rant is now divided by a cross-wall, thus forming the octant stage. 
Instead of producing the dermatogen immediately, as in the case of 
Capsella Bursa-pastoris and Alyssum macrocarpum, it now divides 
repeatedly by longitudinal walls (figure 78). At the same time 
the suspensor becomes very large and elongates, pushing itself 
back into the micropyle and forming an enlargement ( figure 18). 
It apparently functions for a short time as a simple haustorium, 
but very soon disintegrates. However, its old course can still be 
traced (figure 1g) in very late stages. In the meantime, the small 
suspensor cells next to the embryo divide by cross-walls. The 
