644 GRIGGS: SPECIES OF HELICONIA 
these subgenera the species are arranged according to an artificial 
key. This arrangement often separates widely species which are 
very similar except in some unimportant character, and thus ignores 
the existence of natural groups smaller than the subgenus. That 
such groups do exist is certain, ¢. g., that referred to under //. 
Champneiana and that to which 1. fortuwosa and H. latispatha be- 
long. It is believed that their recognition will be of material 
assistance in the study of the genus. The delimitation of the sub- 
genera has also been considerably emended. It is not to be 
denied that the shape of the branch-bracts is in a general way corre- 
lated with the relationships of the species, but it is only an acci- 
dental parallelism without much physiological importance, for 
there are many exceptions — species quite similar in all respects 
except that the branch-bracts are sufficiently different to place 
them in different subgenera as heretofore defined; for instance, Hi. 
tortuosa and H1. latispatha. It is an undesirable character to use 
in a key because of the difficulty of application and the liability to 
error. In the inflorescences of several species the lower branch- 
bracts are lanceolate, while the upper are ovate and the intermedi- 
ate neither one or the other. Besides, there are several species 
whose branch-bracts are ovate when fresh, but shrink so as to be 
lanceolate in the herbarium specimen. 
Two of the subgenera, Stenochlamys and Taentostrobus, ate 
natural groups, but the third, Platychlamys, the name of which - 
must of course go with the type, comprises simply the residue. 
Its present application can be considered only provisional. The 
extreme form of Stenochlamys (H. psittacorum) is very different 
from the members of the other two groups, but the difference 1S 
of only subgeneric importance, for these extremes shade through 
a series of intermediates into the other groups. This is true 
whatever characters are used to distinguish them, whether the 
width of the bracts or the habit, but the habit represents a natu- 
ral difference of some physiological significance ; and it seems to 
me to accord with the probable relationship of the plants. 
In the groups with the typical Hediconia habit, those spel 
with erect inflorescences have been separated from those whose 
inflorescences droop. The importance of this character seems 
not to have been appreciated heretofore because of the lack of 
