go BOTANICAL GAZETTE [FEBRUARY 
in the same way that the protostele in it becomes a siphonostele, 
by the differentiation of a strand of phloem in the center of the 
xylem. This repetition of the process by which the first cylinder 
was made does not take place in Marattia, but the medullary sys- 
tem becomes cylindri- 
cal by the anastomosis 
of two central strands 
(text fig. 3) in one case, 
although there are sug- 
gestions that the cylin- 
der may also be formed 
by the branching of the 
strand as Kin (15) 
suggests. In the one 
clear case of the forma- 
tion of the second cylin- 
der in my material two 
medullary strands fuse 
to form a crescent. A 
branch from one horn 
of the crescent passes 
over to the other horn, 
then a strand from the 
first horn is cut off and 
runs out to a leaf gap. 
Further branching 
breaks up the second- 
ary solenostele. 
Fic. 2b.—The same as fig. 2a, with the side cut 
away to show the course of the medullary strand. On account of the 
spiral arrangement of 
the leaves, the medullary strand that shares in closing a leaf gap 
at one level makes part of the vascular tissue of the leaf above 
in that rank. When the crowding of the leaves breaks the 
regularity of the spiral the medullary strand is less definitely 
related to the leaf traces. The influence of the compact form 
of the stem is further emphasized in the stems described above 
whose upper half elongates. In these the medullary strand 
* 
