498 Agnes Arber.—On the Structure of the Palaeozoic Seed 
ever, little reason to doubt that in Mitrospermum compressmn the space 
between integument and nucellus was natural and not artificial. 
In the upper part of the seed the nucellus, as seen in transverse section, 
is characterized by four salient angles, which lie in the principal plane and 
the plane perpendicular to it (PI. XXXVIII, Figs. 17 and 19). It has been 
already mentioned that near the micropyle the shell cavity becomes diamond¬ 
shaped in section. It is probable that this fact explains the four-angled 
form of the nucellus, since in the living seed it must have fitted the shell 
cavity fairly closely. The shrinking of the softer tissues, due to fossilization, 
has, however, caused the angled part of the nucellus to be drawn down to 
a level below that of the corresponding angled part of the shell-cavity. 
7 'here is, unfortunately, no case known in Mitrospermum compressum of 
a typical, well-preserved pollen-chamber at the apex of the nucellus, but a 
preparation in the University College Collection (PI. XXXVIII, Fig. 21) 
recalls one of Scott and Maslen’s figures of an oblique longitudinal section of 
Trigonocarpus Otiveri , l in which the pollen-chamber f is apparently cut 
through \ The marked ‘ shoulders ’ of the nucellus in this preparation of 
Mitrospermum compressum also recall certain figures given by Brongniart 2 
of Cardiocarpus sclerotesta and C. tenuis. In Mr. Watson’s Collection there 
is an interesting transverse section, which seems to pass through the pollen- 
chamber (PI. XXXVIII, Figs. 17 and 18). The section of the nucellus is 
very small and almost square. Elongated cells radiate from a tiny space 
in the centre. If we imagine that the pollen-chamber was of the type 
found in Cordaianthus Grand'Euryip a section such as the one under 
consideration might be explained as cutting through the pollen-chamber at 
the base of the beak. Renault describes the wall of the pollen-canal in this 
species as formed of cells elongated in the transverse direction, and radiating 
round the central conduit. 
The megaspore and the narrow nucellus are as a rule much shrivelled 
and carbonized, but the most favourable sections indicate that the megaspore 
was surrounded by a nucellar tapetum. This is often preserved only as 
a structureless line. On the outer side of the tapetum, there are a few 
layers of thin-walled cells, among which, as has already been stated, there 
are occasional traces of elements which may have been tracheal in nature. 
This delicate tissue is clothed externally by a carbonized nucellar epidermis 
(PI. XXXVIII, Figs. 22 and 23). 
A seed of which there are two sections in Dr. Scott’s Collection 4 has 
a tissue within the embryo-sac, which may be a prothallus (PI. XXXIX, 
Fig. 24). This consists of irregular roundish cells, and there is no indica¬ 
tion of the tubular method of ingrowth characteristic of the modern 
1 Scott and Maslen (’ 07 ), PL XIII, Fig. 20. 
2 Brongniart (’ 81 ), PI. II, Fig. 1, and PI. V, Fig. 3. 
3 Renault (’ 79 ), Pi. XVII, Fig. 15. 4 D. PI. S., 1802 and 1803. 
