194 THE JOURNAL OF BOTANY 
thus free communication is established reece the central 
strand. The pits are very large and nu tee so that little 
owt 
separate individual course, continues this ‘se unbranched while 
traversing the node, and then, branching ey begins the 
formation of the subcortical and peripheral layers of a new in- 
jibe chad (joint). This is characteristic of H. macroloba and H, in- 
cra. 
‘.. Ta other species there is no general apa of pits, but 
communication is effected by fusion “of the filaments in groups of 
two or three. The i aaielag ee of the fusing filaments i is com- 
formation of a joint, shi fased portion continues its growth as a 
single filament for some little distance to form the node ; ulti- 
mately the filaments branch di- or trichotomously, and begin the 
formation of the i joint. This occurs in H. Tuna, H. gracilis, 
and at other specie 
. The third muita of communication peeomies the peagne | in 
so tas as it is limited to adjacent filaments. But in this case the 
number of fusing filaments is always two, and their identity i is not 
completely lost, for immediately after the junction they appear 
m m 
cation between the central filaments belonging to one of these three 
ee? and this character serves as a convenient basis for a syste- 
arrangement of the species. This arrangement does not run 
soentet to the old lines of classification, but it serves to show that 
several species formerly regarded as independent, on the ground of 
variety in external characters, are really nothing but forms of one 
or another Mer species. This classification is worked out fully 
n my paper ‘The Genus Halimeda”’ (Siboga Expeditie, Mono- 
graphs Ix. Leiden, 1901). 
Sporangia of H. gracilis. 
The oo ie pen in haar between she central strand of fila- 
ments and the angia is very close, since the sporangi = 
are a set i of Sailvsi filaments wie after fusion emerge 
from a joint, and bear sporangia instead of forming a oii joint. 
The exact position therefore of the sporangiophores on a joint ogee 
depend on the course followed by the central filaments inside 
joint, and this course varies in different species. In H. gracilis the 
filaments of the central strand are chiefly confined to the main axis 
which runs directly through a joint to the next joint above, and to 
the two lateral branches of that main axis, each leading to an appa 
angle of the joint. The fruiting filaments therefore of H. gracilis 
