157 
it is often somewhat irregular, the longitudinal walls being inclined. I have not been 
able to decide if real cell-walls are formed between the spores in the sporangium. 
The latter is surrounded with a distinct wall consisting of more than one layer; at 
the lines of separation between the spores the inner layer is seen to be continuous, 
without penetrating between the spores. The sporangia develop in the main axis 
as well as in the branches down to 1 to 2 cm. from the base. 
The antheridia (spermatangia) form a continuous layer on almost the whole 
surface of the male individuals. They are cut off by inclined, often upwards convex, 
intersecting walls of the upper end of the antheridia-bearing cells (SVEDELIUS’ sper- 
matangial mother-cells), at two (or perhaps more than two) sides (fig. 74, F). The 
form of the antheridia-bearing cells is rather variable according to the varying 
length and breadth; they contain a single nucleus but seem to be destitute of chrom- 
atophores. Beneath a fully developed antheridium a new one can arise, a little cell 
very rich in contents being cut off in the same direction as the former. In the 
middle of fig. 74 F above, the oldest antheridium is seen to be connected through 
a pit with the youngest one formed right under it. The continued formation of 
antheridia thus takes place by intercalary divisions, and the antheridia are placed 
in two (or more) series, but owing to the evacuation of the spermatia, at most two 
antheridia are to be seen at the same time in the same series. The antheridial 
development in this plant thus does not correspond with any of the types set up 
by SvEDELIUS (Martensia, K. Sy. Vet. Ak. Handl. Band 43. No. 7. 1908 p. 76). 
The development of the cystocarp was found to agree with what Okamura found 
in examining Japanese specimens. The carpogonial branches arise from the inner 
part of the wall of the hollow frond, frequently from a cell in a longitudinal fila- 
ment or from a cell given off from it. They are 5-celled and curved, in particular 
at the'upper end, where the carpogonium is cut off by an oblique wall intersecting 
the underlying wall (comp. Scumirz |. c. fig. 22, Okamura |. c. fig. 4). 
The auxiliary-cell filaments, being very numerous, as the carpogonial filaments 
as well, have a similar position to these. They are somewhat curved, and consist 
of 4 to 5, more rarely 6, rather low cells with rich contents. They are frequently 
placed quite near the carpogonial filaments; it may even happen that a carpogonial 
filament arises from the base of an auxiliary-cell filament (fig. 75 A). After fecunda- 
tion, fusions take place between the carpogonium and one or more cells in the 
carpogonial filament, resulting in the formation of a great fusion-cell of very ir- 
regular form, giving off sporogenous filaments in various directions; in fig. 75 E 4 
such filaments are present. The auxiliary cells with which they become connected 
are usually the second cell from the base of the auxiliary-cell filaments, sometimes 
the third or even the fourth cell. After fusion, the auxiliary-cell, when giving rise 
to a cystocarp, produces, at the convex side of the filament, a number of cells which 
after several divisions form a group of carpospores placed around a placentar cell 
originating from the auxiliary cell (or the fusion cell). A curious anomaly is shown 
in fig. 75 F. In the ventral part of the carpogonium no nucleus was visible, but 
