1910] SMITH—ERIOCAULON 285 
but the analogy stops here. In Eriocaulon the endosperm cells are 
never binucleate and the first-formed walls are permanent. 
The cells continue to extend centripetally by vacuolization, and 
then undergo periclinal division, resulting in an outer layer of flat 
cells which do not divide further, and an inner layer of larger cells 
(7g. 25). At this time no starch is present. The inner layer by 
further divisions obliterates the central cavity of the embryo sac and 
then rapidly fills with starch grains. In the ripe seed three regions 
of the endosperm are recognizable: (1) the outer layer, still more 
flattened and containing little starch; (2) the large thin-walled cells 
of the interior, with abundant starch; (3) a deeply staining group 
of starch-containing and elongated cells near the base. 
Embryo 
It is in the development of the embryo that this plant deviates 
more widely from ordinary angiosperms. The first division is, as 
usual, transverse (fig. 27). Each of the two resulting cells then 
divides in a longitudinal direction, the one toward the base of the 
sac dividing first as a rule (figs. 28, 29). The next divisions are also 
longitudinal, and at right angles to the preceding (fig. 30). Thus 
the embryo has no suspensor whatever, and passes through regular 
quadrant and octant stages. While these divisions are taking place, 
the embryo becomes globular in form and its cytoplasm beautifully 
vacuolated. The octant division is followed by the cutting off of 
adermatogen. The process is not simultaneous in the two halves. 
-Thus in figs. 31 and 32 it will be seen that the micropylar half has 
divided most rapidly, while in fig. 35 it is the basal half. The embryo 
of the ripe seed is shown in fig. 36. It is bell-shaped with flaring 
edges, and quite frequently the edges are upturned, owing to the 
Pressure of the endosperm. There is no differentiation of the embry- 
onic organs, nor any indication where these shall have their origin. 
It is doubtful if a comparison of the early segmentation of the 
embryo is of value in determining relationship. ‘There is too much 
variation within certain orders, and often the peculiarities are too 
manifestly adaptive. The nature of the endosperm and the embry- 
onic organs is much more significant. 
I have found no detailed study of plants nearly related to Erio- 
