34 Humphrey . — The Development of the 
Since the evident purpose of the lid is to facilitate germination, 
and that of the collar is to ensure the efficient connexion of 
the plantlet with the food-supply in the seed, and as the testa 
of Strelitzia is especially firm and shows no other special 
adaptation for these purposes, a study of the germination of 
this seed may prove of considerable interest. 
While arillar structures are very common in those members 
of the group which have dehiscent fruits, they are commonly 
not at all developed in species with indehiscent ones. When- 
ever present, the aril arises from funiculus and integument in 
the micropylar region. It may form a mass at this end or an 
envelope about the seed. 
Greater or less penetration of chalazal tissue into the 
nucellus appears to occur in Musa , in various Zingiberaceae, 
and probably in all Marantaceae. In Costus and Musa the 
chalazal differentiation has merely the form of a rounded 
mass, while in Alpinia there is formed a diaphragm pene- 
trating to the middle of the seed ; and in the Marantaceae the 
included chalazal tissue appears to be continued far into the 
seed by a differentiation (in Thalia , two) in the nucellar 
tissue. In how far the homologies here suggested are really 
valid must be determined by studies of the development of 
a variety of forms. 
The formation of the true testa and of all the special 
structures associated with it in other forms is suppressed in 
Heliconia by the development of a stony endocarp. Here, 
apparently, the mouth of the endocarp replaces the micro- 
pylar collar, and the sclerotic plug replaces the germinal lid. 
Even in Canna the pollen-tube must penetrate the nucellar 
epidermis to reach the embryo-sac, and in most of the species 
examined this portion of the epidermis is thickened into a firm 
micropylar pad, which reaches its greatest thickness after 
fertilization, and becomes most conspicuously developed in 
the Zingiberaceae and Musaceae. 
In the plants examined, the starch-bearing tissue of the 
seed is entirely perisperm, except in the Musaceous genera. 
In Heliconia there is a narrow layer of functional perisperm 
