mer is terns of Ferns as a Phylogenetic Study . 3 1 1 
enclosed between two of these principal walls, and thus the 
whole meristem may be regarded as consisting of three wedge- 
shaped masses. Taking now those masses singly into con- 
sideration, it will be clearly seen that in Fig. 3 each is divided 
into two unequal parts by walls marked ( s ), which do not 
proceed to the centre, but on passing towards it curve 
gradually out of the radial plane, and insert themselves at 
right angles on the principal walls : these may be called the 
sextant-walls. A similar arrangement, but less clearly marked, 
may be seen in such examples as that represented in Fig. 4. 
In each case the error of regarding sections below the real 
initial cell or cells as including the initial cells, has been care- 
fully avoided by comparing the whole series of sections cut 
from each individual root. 
For comparison with this arrangement of the initial cells 
and their derivatives I have quoted as Fig. 5 one of the 
drawings of Naegeli and Leitgeb, which appeared in their 
well-known work, ‘ Entstehung und Wachsthum der Wurzeln,’ 
as Taf. XII, Fig. 8. This represents a view of the structure 
seen immediately below the apical cell in the root of Pteris 
hastata , and it will be seen that in the most essential points 
the structure of its central portion is similar to that in Figs. 3 
or 4 : and if we imagine these principal (p) and sextant- walls 
( s ) to be continued upwards in Pteris or any leptosporangiate 
Fern, till they reach the root-cap, a structure of the root-apex 
would result which would correspond to that observed in 
certain cases in Osmunda regalis : in other words, accepting 
the view of Sachs that the apical cell is a gap in the system of 
construction, such roots as these of Osmunda regalis show the 
gap less complete, and the three principal walls continued 
through it, so as to divide it into three parts. 
Since the characters of the meristem in the root of Osmunda 
have thus been shown to be subject to variation, and since it 
is hardly possible to determine with certainty from longi- 
tudinal sections what is the actual number and form of the 
initial cells, a full and exact interpretation of the structure in 
longitudinal sections cannot be expected. In the majority of 
