A natomy of the Cone and Fertile Stem of Equisetnm . 699 
by the endodermis, which in these species surrounds the stele. In the stem 
of E. limosum (and presumably in all cases in which the vascular strands are 
surrounded by separate endodermes) even this barrier is broken down ; for 
at the breaking up of the nodal ring of wood the external and internal endo- 
dermes found in this region become involuted round each bundle, so that 
pith and cortex are in direct communication (Pfitzer, pp. 336-7). Definite 
endodermal markings are, however, very difficult to distinguish in the cone 
of this species. It would seem that, phylogenetically speaking, parenchyma- 
tous tissue first arose in the internodal part of the siphonostelic xylem of 
the cone of Equisetnm. It is not surprising that this parenchyma should 
have arisen vertically above the traces, since the current of water, being 
deflected into the sporangiophores, would be diminished above them. From 
the dimensions of the fossil E guise tites we have good reason to believe that 
existing species of Equisetnm are reduced, and reduction of the xylem would 
tend to make the mesh arise closer to the trace. This is exemplified in the 
cone of E. limosum , in which there is very little xylem ; here the traces are 
often only surmounted by a very few lignified elements, though even in this 
species the parenchymatous mesh sometimes originates at a considerable 
distance above the trace. 
In the preceding pages we have traced the phylogeny of the irregular 
network of strands of the cone of E. limosum from the more regular stele of 
E. arvense, which is imperfectly siphonostelic at the nodes and dictyostelic in 
the internodes. Considered separately, the irregular network appears to 
baffle interpretation ; considered in the light of a comparative study of the 
cones of E . palustre and E. arvense , we see that it has arisen in the phylogeny 
by the reduction of xylem, this reduction being manifested by extension of 
the parenchymatous meshes upwards, downwards, and laterally. 
In 1899 Jeffrey published a photograph of a cone of E. arvense cut 
longitudinally in two ; this photograph was intended to show that the 
majority of the bundles of the cone do not alternate at the nodes (Jeffrey ( 1 ), 
PI. XXX, Fig. 3). This feature was, he said, more or less marked in the 
cones of all the species of Equisetnm that he had examined. Some years 
later he returned to this point, and after commenting on the greater con- 
servatism of reproductive axes as compared with vegetative ones he pro- 
ceeded : c It appears to have been shown above and beyond any doubt that 
the equisetaceous strobilus perpetuates both the alternating strands and the 
complete absence of foliar gaps of the oldest Calamitean forms’ (Jeffrey (2), 
p. 254). While, as Jeffrey remarks, it seems obvious from the data of Stur, 
which have never been called in question, that the oldest Calamariae were 
without alternation of the strands at the nodes, and while, according to 
Hick ling, the bundles of the cone of Calamostachys did not alternate at the 
nodes (Hickling, 2 , p. 10), and those of Palaeostachya vera seem to show no 
regular pectination, but probably occasional and irregular communication 
