28 Humphrey . — The Development of the 
who describes only the structure of the ripe seed. I have 
been able to examine ovules and young seeds of about adult 
size of a species sent from Jamaica under the name M. rosea , 
but could not follow the development of the former into the 
latter. The ovule (Fig. 50) presents nothing noteworthy, 
except a dense felt of long simple trichomes arising from 
the sides of the funiculus. No trace of these remains in the 
seed. As compared with the ovule, the seed (Fig. 50 a) 
shows great lateral extension, so that the embryonal cavity 
is much shorter than broad. The trace of the micropylar 
opening can still be recognized (m., Fig. 50 a), and, as might 
be expected from the indehiscent fruit, no aril is developed. 
The outer cell-layer of the testa forms a deep palisade-layer, 
and its inner layer consists of large cuboidal cells. Between 
these two the cells become much elongated parallel to the 
surface, and the walls of a few of the outer layers become 
conspicuously thickened and pitted. Whether the remaining 
cells constitute a nutritive layer and finally become compressed 
could not be determined. At the chalazal end the inner in- 
tegument is bent upward, and my material points to the 
differentiation of a chalazal mass similar to that of Costus. 
Wittmack’s (’68) figure also indicates the presence of such 
a structure in M. Ensete . The micropylar collar so character- 
istic of most Scitamineae is here present also ; and its outer 
end is closed by a germinal lid differentiated from the inner 
layer of the outer integument, as in Amomum and other genera. 
The epidermis of the nucellus thickens into a conspicuous 
micropylar pad (;;/./.? Fig. 50 a). The inner nucellar tissue 
is largely obliterated by the growth of the embryo-sac, but in 
the oldest seeds examined there is still a broad band of 
perisperm present, though the deposit of starch in its cells has 
not yet begun. Within the cavity neither endosperm nor 
embryo could be found, but without more complete material 
it is impossible to say that this species forms only abortive 
seeds. The form of the cavity would determine for the embryo, 
if developed, the fungiform shape observed in other species of 
M usa. 
