meris terns of Ferns as a Phylogenetic Study. 345 
from the half-disc-shaped marginal cells, successive segments 
are cut off by transverse walls, and these are subdivided in 
such a manner that the transverse section of the margin of 
the frond presents an almost identical appearance with that of 
a median longitudinal section through the apex of the midrib 
of other species 1 : thus Prantl’s conjecture is correct, and 
though the ultimate bulk and complexity of the leaf in this 
species is as great as that in many of the Polypodiaceae, the 
marginal cells retain the form and simple segmentation charac- 
teristic of the Hymenophyllaceae. 
The parallel case of Hymenophyllum dilatatum was also exa- 
mined : exclusive of the nerves, the wings consist in this species 
of three to four layers of cells, usually three : of these the cells 
of the central layer are larger, while the superficial ones are 
relatively small. The wings extend over all the branches of 
the leaf, and may be traced down the petiole. In the young 
state a marginal series of half-disc-shaped cells is to be seen, 
from which successive segments are cut off by transverse 
walls, as in other Hymenophyllaceae (Fig. 52) : the sub- 
division of these segments is singular, and seems to follow 
an almost regular rule : the segment first divides into two 
unequal parts (Fig. 52, the two youngest segments), by a 
periclinal wall, the larger cell then again divides, and the three 
resulting cells contribute to the three layers of which the 
mature wing is usually composed, and may undergo further 
divisions by anticlinal walls as in Fig. 54. When the marginal 
cell is about to cease its activity, irregular divisions appear in it, 
the walls being variously inclined (Figs. 52, 54), and the result 
is an appearance which at first sight resembles the segmentation 
of the marginal cells in the Polypodiaceae ; but the same 
regularity does not hold here, and it is possible in all cases 
which I have observed to see (as in Figs. 53, 54) that these 
irregular and oblique segmentations succeed the more regular 
transverse ones : in fact they are not the type for this plant, 
any more than the oblique segmentation in the wing of 
1 Compare Prantl’s Fig. 39, Plate III of Trichomanes speciosum ; also Sachs’ 
Lectures, Fig. 295 A. 
