the Seed in the Alsinoideae . 
35 
cell formation subsequently follows, resulting in the formation of a single 
layer of small quadrate cells with dense protoplasmic contents (Figs, 20 , 21 , 
end. c.). This layer of cells keeps pace with growth by means of constant 
anticlinal divisions at the apex of the embryo-sac, but it gradually thins out 
into large uninucleated cells with vacuolated contents over the remaining 
portion (Fig. 21 , end). In the mature seed the endosperm-cap invests the 
root-cap of the embryo, and the rest of the embryo-sac is lined by a thin 
film of very large cells. The embryo is therefore enclosed by a single 
continuous peripheral layer of endosperm-cells of diverse character in the 
hypocotyledonary and cotyledonary regions respectively. In sections the 
cells composing the larger portion of the layer are difficult to see, owing to 
their size and extreme thinness, but they are easily differentiated by careful 
staining, and this portion can be dissected out in its entirety ; but it is very 
difficult to get any part of the cap off on account of its intimate relation 
with the root-cap of the embryo on the one side and the nucellus on the 
other. 
In consideration of these facts Hegelmaier (12) is hardly correct 
in describing the embryo as being enclosed in transitory endosperm in 
the Curvembryae, a group in which he had previously stated the endosperm 
to be limited entirely to the micropylar end ( 10 ). 
Cell-wall formation in the endosperm is recorded for several families 
at a stage similar to that at which it occurs in Stellaria media of the 
Alsinoideae, that is, when the cotyledons first become differentiated. 
Guignard (17) records it for the endosperm of some Leguminosae, and 
concludes it becomes definitely organized at this stage to meet the in- 
creasing requirements of the embryo, the suspensor, it is assumed, being 
now no longer capable of doing so. 
Strasburger (30), for the Eualchemillas, states that the endosperm 
forms walls as the embryo becomes heart-shaped. He makes the apposite 
suggestion that, as the embryo-sac is then full-sized, the stoppage of growth 
causes the endosperm-nuclei to remain in contact and so start cell division 
(30, p. 124 ). That this observation applies in the present case also is 
borne out by the regular arrangement of the nuclei of the endosperm- 
cap (Fig. 19 , end. c), which obtains just before cell division takes place. 
Pechoutre also (28) arrives at a similar result in the case of the Rosaceae. 
In Stellaria media we see that the endosperm is differentiated in 
its apical portion into a compact layer of cells with dense and homogeneous 
contents, which in organization and appearance strongly suggest ferment- 
cells. This cap invests the apex of the embryo with its inner surface, 
whilst externally it is in direct contact with the axile rows of the nucellus 
(Fig. 22 ; end. c). The cells of these rows adjoining the endosperm-cells 
show paucity of contents and very slight starch reaction, but they abut 
directly on the perisperm tissue of the nucellus, the latter appearing as 
