100 
P. 0. BOWER. 
Lastly, this peculiarity of structure of the roots is a cha- 
racter, among many others, which points out the Osmun- 
daceae as a family of Ferns having a close affinity with the 
Marattiaceae. I have elsewhere shown 1 that, in the develop- 
ment of the leaf, and especially of the apex of the leaf, and in 
the conformation of the base of the leaf, the Osmundaceae 
approach the Marattiaceae; also that Todea is in certain 
character’s nearer to them than Osmund a. The observa- 
tions above detailed bear out this conclusion, and though such 
details should not be pressed too nearly home, still it should 
be noted that while no clear example of four initial cells was 
observed in Osmunda regalis, that structure appeared to 
be most frequent in Todea barbara. Again, the recent 
observations of Goebel (“ Verleichende Entwickelungsge- 
schichte,” ‘ Schenks' Handbuch,’ Bd. iii, pp. 387 — 388) leave it 
still in doubt whether Osmunda be a Leptosporangiate Fern 
or not. 2 Thus in a number of characters, perhaps not very 
1 ‘On the Comparative Morphology of the Leaf in the Vascular Crypto- 
gams and Gymnosperms,’ communicated to the Royal Society. 
2 My observations on the Sporangia of Osmunda, as far as they go, confirm 
those of Goebel ; but better material was at hand for the investigation of the 
development of the sporangium in Todea barbara. Here the essential 
parts of the sporangium originate apparently from a single cell, which is, how- 
ever, deeply sunk in the tissue of the young pinnule, but is exposed at the outer 
surface. It has a square base, that is the cell is not conical ; divisions appear 
in it by walls perpendicular to the outer surface. When observed in surface 
view from the outside it is seen that four cells result from this division, three 
of them surrounding one central cell. This arrangement is not unlike that 
in the young sporangium of the Leptosporangiate Ferns ; the chief differ- 
ence is that up to this point the sporangium does not project far beyond the 
general surface of the leaf. Subsequently a periclinal wall separates a super- 
ficial cell from the archespore. The form of the archespore is a matter of 
some interest. Here, as in the case of the apical cell of the root, there is 
not exact uniformity ; it is sometimes conical, sometimes rectangular. This 
variation depends upon the varying inclination of the three anticlinal walls, 
one to another ; in some cases they are nearly parallel, and the result is that 
the archespore in these examples has a square base ; in other cases they are 
inclined, and may meet one another below ; the result is then a conical arche- 
spore. This appears to be the case also in Osmunda, judging from Goebel’s 
figures (1. c., figs. 103a and b), and is no doubt to be connected with the 
