338 Bower. — The comparative examination of the 
present, such as is found sometimes in the leaf of Marattia , 
and constantly in Osmunda. It may further be noted that 
there is usually present a clearly defined longitudinal wall 
(/,/ in the Figs. 42, 43), which runs at right angles to the 
median plane of the leaf: this divides the whole meristem 
unequally into adaxial and abaxial halves, and there appear 
to be as a rule two initials on the adaxial or posterior side of 
it, but only one on the abaxial or anterior side. Whether 
such cells as those marked (o') are actually initial cells for the 
whole of the areas indicated, or not (and such an arrangement 
has been seen in a number of leaves), I am not in a position 
to state definitely, but would point out that the arrangement 
of the cells about them is such as to lend countenance to 
such a view. Be this as it may, it is evident that the apex of 
this leaf is more complex than that of Marattia , and that 
there are at least three initial cells present. Inspection of 
these meristems as seen in plan also show that there is no 
marginal series of cells present here, having any such relation 
to the initial cells as that already noted in the leptospo- 
rangiate Ferns. 
Longitudinal sections of so complex a meristem are 
naturally difficult of interpretation ; the principal wall (p, p) 
is usually well marked, and cuts the meristem into anterior 
and posterior halves; this was recognised in my previous 
paper 1 . In a single median section (Fig. 44) several conical 
cells may be found, but, as already pointed out, the conical 
form does not necessarily imply that these are initial cells : 
even in sections which are believed to be exactly median it 
is very difficult to recognise the initial cells with certainty : 
the cells marked (x, x) in Figs. 44 and 45 are believed to be 
the initials, but I should not wish to state this point dog- 
matically. As regards the branching of the leaves of the 
Marattiaceae, it is exclusively monopodial, and on this point 
I have nothing to add to observations already published 2 . 
1 1 . c. p. 580, Fig. 12. 
2 1. c. p. 580, 586. 
