96 
F. 0. BOWER. 
structure with three initial cells is closely related to that with 
four. If we remember that a four-sided apical cell has been 
observed in Osmunda, and, according to Holle Bot. Ztg./ 
1875, p. 301), also in Marattia cicutsefolia, a change similar 
to that above suggested in the case of the meristem with a 
three-sided cell, would in the case of a four-sided cell produce 
a structure corresponding to that observed by Schwendener in 
the Marattiacese, and by myself in Todea barbara. Here, 
instead of three radial walls meeting at a central point, there 
would be four such walls. Observation shows, however, that 
where this is the case the four walls do not meet exactly at 
one point, but slight irregularities are usually found. It is 
well known that in the roots of the type investigated by 
Naegeli and Leitgeb, the cap-segments divide by radial walls 
disposed crosswise into four cells ; a similar arrangement has 
been shown to be the result of divisions in the root-cap of 
Osmunda (figs. 5, 7), even where the initial cells are of very 
different character. It can hardly be a matter of surprise that 
this incongruity between the radial walls of the root cap and 
those of the body of the root is removed in that type of struc- 
ture which appears to be the more permanent. In the Marat- 
tiaceous type the first formed radial walls, both in the root- 
cap and in the body of the root, are assimilated to one regular 
and uniform system, viz. two planes cutting one another at 
right angles. The incongruity of the radial walls in those 
examples of Osmunda in which there are three initial cells 
may be regarded as characteristic of a transitional structure, 
and it is not perpetuated in the more constant Marattiaceous 
type. 
Before passing on to other points, a comparison is to be 
drawn between those roots of Osmunda with three initial 
cells and the structure of the apex of the stem of Lycopo- 
dium Selago, as shown by Strasburger (‘ Das Botanische 
Practicum,’ figs. 93 and 94). Here there are also three initial 
cells disposed in a manner similar to that observed in the 
figures of Osmunda. Thus there are at least two examples 
now known of a type of grouping of initial cells, which is 
