1 8 Humphrey. — The Development of the 
cells along the future course of the cavity begin quite early to 
show signs of degeneration, and a dark band extending from 
the base of the embryo-sac first marks the path of its 
extension (Fig. 29). Early in the development of the seed, 
two cylindrical differentiations of slightly denser-looking 
tissue may be seen in the chalazal region of the nucellus, 
curving slightly and lying parallel with each other on either 
side of the plane of symmetry of the seed, but evidently 
originating in the nucellar tissue itself (p. c., Fig. 29). They 
begin abruptly in the region of the base of the inner integu- 
ment. These two cylinders then follow in their development 
the extension of the embryo-sac, so that there are formed two 
parallel tissue-masses curving with this cavity, but just within 
its curve, and lying on either side of its median plane (/. c., 
Fig. 30). This tissue finally consists of loose, brown and 
dead cells, through which run branches of the vascular 
bundle which earlier reaches only to the base of the nu- 
cellus (v. b., Fig. 29). In the adult seed these two c perisperm- 
canals’ appear as branches from a simple projection into the 
perisperm from the chalazal region. This seems to be a mass 
of chalazal tissue which has been included by the growth of 
the nucellus, and strictly homologous with the similar included 
tissue-masses shown by Meunier (’ 90 ) to occur in the seeds of 
various Centrospermae. These latter seeds are developed by 
growth similar to that observed in the Marantaceae from 
similar ana-campylotropous ovules. Unfortunately, just the 
stages lacking in my material are the important ones between 
Figs. 29 and 30, during which the differentiation of the canals 
takes place. This remarkable development of the perisperm- 
canals in Thalia is probably their most extremely developed 
condition in the Marantaceae ; but they form a very character- 
istic feature of the family, and perhaps no strictly homologous 
structures are yet known in other plants. In most genera of 
the family there is but a single canal, extending inward be- 
tween the arms of the embryo-sac and in the same plane. The 
first speculations as to its morphological significance appear 
to have been those of Robert Brown (TO), who suggested that 
