26 Humphrey , — The Development of the 
be seen a slight groove extending from the micropylar end 
on either side downward around the chalazal end. A section 
at right angles to the plane of this groove (Fig. 39) shows an 
ingrowth from the chalazal region, reaching to about the 
middle of the seed. At its apex this ingrowth broadens 
slightly. If the seed be cut in the plane of the groove first 
mentioned (Fig. 40), it is found that the ingrowth before seen 
is a section of a dense testa-like diaphragm, extending across 
the chalazal half of the seed, and that the groove marks the 
line of its union with the testa : above the upper limit of 
the diaphragm, it is continued to the micropylar end by 
a slight projection of the testa into the perisperm. The 
diaphragm is sharply bounded against the perisperm by 
a cell-layer continuous with, and similar to, that derived 
from the inner layer of the outer ovular integument. This 
layer stops abruptly just below the top of the thickened 
upper edge of the diaphragm, and it may be observed that 
the thickening of this edge is due to a sudden outward 
bending of this bounding layer (Fig. 39), strikingly like the 
upward bend which limits the chalazal mass of Costus 
(Fig. 46). It is to be regretted that my material contains 
no intermediate stages between the practically adult one just 
described and young ovules (Fig. 38). These latter show the 
usual structure, and give no hint of the origin of the dia- 
phragm. But, from the homology of Costus , it may be 
suspected that its formation is begun by the ingrowth of the 
lower end of the inner integument, with perhaps an infolding 
at the sides to meet the upward growth. The development 
of this interesting seed would probably prove well worth 
following on suitable material. The intrusion of this dense 
sheet of tissue into the region usually occupied by the 
embryo-sac causes the latter to spread out over the top and 
then to extend downward on either side of the diaphragm 
( e . Fig. 39). The embryo-sac becomes largely filled with 
endosperm, and the lower part remains so, as the embryo 
does not completely fill it. Tschirch ( 90 ) has remarked 
that the embryo of Alpinia nutans is two-lobed in a sickle- 
