Relationships of the Gnetales . 1083 
is composed do not unite before entering the stele, but are inserted separately. 
In this respect it resembles the Cycads and Ginkgo. It is doubtful, how- 
ever, whether any importance can be attributed to this point, for even 
in many of those ancient forms of which it is characteristic the two parts 
unite before entering the stele (. Lyginodendron ), while in others they remain 
separate ( Poroxylon ). On this basis the primitive Conifers can claim just 
as close an affinity to Ephedra as can the Cycads, for they also have 
a double trace. Furthermore, the Bennettitales, which on account of their 
floral organization have a much stronger claim than their modern repre- 
sentatives on the Gnetalean ancestry, had only a single trace, as Wieland 1 
has shown. This fact has been overlooked in much recent work. For 
example, Miss Sykes, 2 in her work on W elwitschia , considered its double 
leaf-trace indicative of Cycadean affinities. It is obvious that the argument 
which involves the floral organization is quite contradictory to that which 
involves the double leaf-trace, and one or the other must be abandoned. 
The bundle intercalated between the strands of the leaf-trace is 
a feature not found in other Gymnosperms. This fact and its absence 
in certain cases in Ephedra indicate that it is a specialization. It is 
possibly another condition in which Ephedra approaches the Dicotyledones 
in which the several traces to a single leaf arise at various points in the 
circumference of the young stele and are separated by masses of wood. 
The girdle of tracheides at the node is a special feature with no 
significance in respect to relationship. It recalls strikingly the nodal wood 
of Equisetum , but of course this can be only a case of parallel development. 
This feature and the medullary diaphragm of both forms are probably 
correlated in some way with their similarity of habit. 
2. Structure. 
Each primary vascular bundle is roughly triangular, the innermost 
angle being occupied by the smallest protoxylem elements (PI. XCIV, Figs. 
1 and 2). The succession of elements from this point outward to the 
secondary wood is of the usual type. There is no centripetally developed 
wood, so that the bundles are of the endarch type. As has been noted above, 
a band of lignified elements which might easily be mistaken for centripetal 
wood often extends from the protoxylem for some distance into the pith 
(Fig. 2), but in no case was true centripetal wood found in the internode. 
At the node, however, centripetally developed tracheides are occa- 
sionally to be found. They occur in the pith about the level of the exit of 
the leaf-traces. In this position they are not of the usual type, but bear 
a close resemblance to transfusion tracheides of the leaf. Fig. 8 shows one 
1 American Fossil Cycads. Carnegie Institute, 1906. 
2 Anatomy and Morphology of the Leaves and Inflorescences of W elwitschia mirabilis . Phil. 
Trans, Roy. Soc., B., 1910. 
4 B 2 
