180 Gibson . — Contributions towards a Knowledge 
thirdly, with the inner marginal strand of the median stele of 
the branch with which it runs down as an anastomotic cord 
for some distance, fusing ultimately with the adjacent proto- 
xylem in the usual way. The outer marginal protoxylem of 
the ventral stele of the chief axis remains distinct as the outer 
marginal protoxylem of the ventral stele of the axis beneath ; 
the inner strand, however, forms a short temporary union, as 
above stated, with the inner margin of the median stele, and 
then fuses with the ventral cord of the median stele of the 
branch axis. In short, there is a temporary union of the 
different steles just at the point of origin of a branch, but on 
that side only on which the branch arises. Finally, at the 
forking, three strands become inserted on the stelic system, 
two derived from the abortive leafy shoots (which so frequently 
in this species replace the rhizophores) and one from the 
axillary leaf. The cord from the dorsal abortive shoot is 
inserted on the dorsal stele, that from the ventral on the 
median stele, whilst the axillary leaf-trace is inserted on the 
ventral stele. The three primary steles may be traced at 
least in procambial form right up to the merismatic region 
in the primary shoot. 
In larger and older stems the dorsal and ventral steles may 
each become divided, so that a small distinct stele comes to 
lie on the inner side of each. These steles are usually more 
or less united to the parent steles by pericyclar tissue. This 
rather complicated arrangement of the vascular cords may 
perhaps be made clear by a comparison of the above descrip- 
tion with the figures. 
If now the creeping shoots be examined, it will be found that 
all transitions are to be obtained between the tristelic con- 
dition of the erect axis and a true monostelic condition. 
Fig. 67, PI. XI, shows five sections of the vascular core at dif- 
ferent and successive levels. Fig. 67, V, represents the ordinary 
tristelic arrangement ; in IV the ventral stele has not yet 
become isolated, although it exists as a distinct ventrally- 
placed xylem-mass ; in III the dorsal stele has become almost 
isolated, whilst the ventral stele is merely a prominent xylem- 
