672 Browne.— Contributions to our Knowledge of ihe 
words, when two neighbouring strands give off median traces and one forks 
a little higher up and the other does not. In this case one of the branches 
of the forking strand may fuse with the strand in which no mesh appeared. 
This particular variation, fairly common in E. limosum , seems to be very 
rare in E. palustre , where I only observed it twice. Another variation 
which may occur, though rarely, in E. palustre , 
associated with a reduction in number of the 
strands and traces of the next node and internode, 
consists in the fusion of two strands, from which 
median traces not subtending meshes have de- 
parted, to form a strand giving rise to but one 
trace at the next node (e. g. the strands giving 
rise to the sixth and seventh traces of the ninth 
whorl of Cone A). In all these variations the 
absence of a mesh is due to poor development of 
xylem, and is associated with the persistence of 
meshes of the internode below. In the cases 
mentioned first the absence of a mesh is due to 
the unusually strong development of xylem in the 
upper internode, and is associated with closure of 
meshes of the internode below. 
In E. pahistre no fresh meshes originate above 
the departure of a lateral or slightly internal trace ; 
for though in many cases the xylem-elements 
superposed to the trace soon die out, the paren- 
chyma which replaces them is laterally in contact 
with an adjacent parenchymatous mesh originat- 
ing lower down. The relatively wide band of 
parenchyma resulting from the congenital fusion 
of a mesh with one originating lower down may, 
from a phylogenetic point of view, be considered 
as biseriate. When the meshes in the cone of 
l ext-fig. 4. Longitudinal E. palustre persist through numerous internodes 
reconstruction of the xylem 01 * r 0 
Cone D of E. palustre (var. they appear in longitudinal reconstruction to be 
X a„ a d pteT widened first on one side and then on the other ; 
chyma white. Magnification this sudden widening usually occurs above a node 
cvca 24 ‘ by the dying out of the xylem superposed to a trace 
at or near the edge of one of the strands bordering on the mesh. This 
gives to the parenchymatous meshes of the higher orders a very sinuous 
outline. As in E. arvense , it is sometimes easy to see that the persis- 
tence of a mesh into two or more internodes is due to poor development 
of xylem at the nodes. For instance, the lignification of a small group 
of cells between the first and second traces of the third whorl and the 
