336 Bower . — The comparative examination of the 
assume in a secondary fashion, and for a short time, those 
qualities which mark the true initial cells. 
But even in young leaves the apex does not always show 
such regularity as in this first case ; thus in Fig. 37 the initial 
cell, though recognisable, is of less definite form ; in Fig. 38 
is seen the apex of a rather older leaf, and here the initial cell 
is still smaller, its segments less regular, and it is even 
difficult to see how the whole meristem can be referred as 
regards its origin to the one initial cell : secondary initials 
(0) are also present, but not in so marked a manner as in 
Fig. 36. Again, in a still older leaf, in which the pinnae 
were already formed, there was no single initial cell to be 
seen (Fig. 39) : here the meristem has a structure resembling 
that of the apex of some of the roots of Osmunda 1 : three 
principal walls are present, as in those roots, and the 
same theoretical interest attaches to that arrangement here 
as in the case of those roots : in fact, we may represent this 
condition to the mind as resulting from the continuation of 
the principal walls upwards, without deflection, so as to 
divide the apical cell itself, the result being, as in Fig. 39, 
that three initials would be produced (x, x) ; though c second- 
ary initials 5 are also present here, it is possible to refer the 
whole of this complicated meristem to three initials, each of 
which is formative of one-third of the meristem lying between 
two of the principal walls (/,/). The arrows in Figs. 36, 37, 
38, indicate the direction of the apex of the stem, and a 
comparison will show that the orientation of the apical cell 
is not constant as in Osmunda , but that sometimes one side, 
sometimes one angle, is directed towards the apex of the 
stem. 
The above results are borne out by a study of median 
longitudinal sections : in the young leaf of Marattia a single 
conical cell is found (Fig. 40), from which segments are cut 
off with considerable regularity ; here and there other conical 
cells are found, which may correspond to the secondary 
initials above noted in plan : a prominent case of this is 
1 Compare Figs. 3 and 4. 
