158 Grace Wigglesworth . 
again a portion of the ground tissue in the central stele. The outer 
edge of the middle stele continues to bend inwards on the dorsal 
side, depressing the central siphonostele, which then separates from 
the middle siphonostele, in which a gap is now left (fig. 3); this 
gap widens, while the outermost stele first becomes flattened 
dorsally and then bends inwards between the two free ends of the 
middle stele, which have bent outwards as seen in fig. 4. Ultimately 
a connection is formed between the two steles. The two outer 
steles have now the form of a double horse-shoe connected by a bar 
across the narrow end of the inner one, which closes up the middle 
stele (fig. 5). The middle stele then becomes separated from the 
outer one, leaving the latter with a gap. 
The incurved ends of the outer stele are prolonged upwards 
and continue into the vascular system of the leaf stalk (fig. 6). 
Whilst the connection is taking place between the middle and 
outer steles the xylem of the central stele becomes interrupted on 
the dorsal side, but the endodermis and parenchymatous cells 
enclosed remain continuous. 
Immediately behind the leaf insertion the three siphonosteles 
will be found concentrically one within the other and quite 
independent one of the other. 
. The main differences between the structure described above 
and that of Mr. Seward’s material are at once apparent. The 
vascular system consists here of three complete siphonosteles. 
It is not easy to compare at once the steles described above 
with those found by Mr. Seward. The outermost of the siphono¬ 
steles is much the same in both cases. It has numerous protoxylem 
groups (generally about eighteen) lying in the depressions, where 
the inner margin of the stele is crenulated. The middle siphono¬ 
stele on the other hand, according to Mr. Seward’s figure and 
description, has only one protoxylem group, whereas in the specimen 
described above it has numerous protoxylem groups and is also 
slightly crenulated. The innermost of the three siphonosteles, as 
stated above, has only one group of protoxylem, which would agree 
with that described by Mr. Seward for the second siphonostele, 
except that it occupies the ventral and not a dorsal position in the 
stele." 
The occurrence in the same species of either a central siphono¬ 
stele or a single axial strand, as in the specimen described by 
Mr. Seward, is of considerable interest, and bears out Mr. Seward’s 
view that “the annular form of stele is not of great systematic 
