194 Phillips . — The Development of the Cystocarp 
filaments. The two unfertilized carpogonial branches of the 
opposite side of the axis can still be traced. Fig. 28 is an 
enlarged view of the essential parts of Fig. 27 . 
When the cystocarp has matured, the papillar outgrowth 
becomes tuberculated in N. Hilliae , owing to uneven growth 
of the vertical filaments. 
I desire now to discuss the mutual relationships of the 
seven species, the structure of whose cystocarps has been 
described, and then to consider the affinities of the family to 
which they belong. 
To consider first the four species, D. sanguined , D. alata , 
D. Hypoglossum , and D. ruscifolia. It is clear that the two 
plants most closely related are the two last. The leaf-like 
branches are equally vegetative and reproductive in both, and 
the branching is exclusively adventitious. In D. alata , while 
the ordinary forkings may give rise to the reproductive 
organs, there is a marked tendency to produce them ex- 
clusively on somewhat specialized adventitious branches. 
What is only a tendency in D. alata has become a fixed 
condition in D. sanguinea. The four species thus form a 
natural series of which D. ruscifolia is perhaps the least 
specialized, and D. sanguinea certainly the most specialized. 
As has been shown, the structure and the arrangement of the 
procarps and the development of the cystocarp in D. san- 
guinea present so many features in common with the other 
three, that there can bg. no doubt about its inclusion in the 
same family with them. On account of considerations arising 
chiefly from a study of the tetrasporiferous segments, Agardh 
(76) placed the species as Hydrolap a thu m sanguineum in the 
family Rhodymeniaceae. This decision he has recently (’97) 
discussed in some detail and confirmed. He regards the 
stichidia or tetrasporiferous branches as unlike those of any 
other Florideae, finds the closest analogy in the origin and 
