150 F. W, Oliver . 
and rp.~ The lowest plate (r/>. 3 ) is shown in its younger unsplit con¬ 
dition. To the inner edge oF each radial plate two divergent fibrous 
membranes or trabeculae ( tvab.) are attached, which, as shown in 
the diagram, end in close relation with the bottoms of six paired 
grooves (g) in an epidermis-like membrane. The general matrix of 
this region or envelope of the seed, though preserved in a fragmentary 
condition, may be safely regarded as having been a continuous tissue 
of delicate texture. With the grooved membrane what seem to 
have been ridges on the nucellus engage (nr.) ; whilst in the thick¬ 
ness of the wall of the nucellus there is a circle of vascular cords (v d ). 
Within, the nucellus is without organic structure. Figure 5 is based 
on Brongniart’s fig 2, pi. xviii., and Renault’s fig. 5 , pi. lxxxiii. 
These two drawings are evidently based on the same preparation, 
though the interpretation of the parts is different. The facts, as 
to which there seems no question, are interpreted on the supposi¬ 
tion that the region between e x p and e x p x in Brongniart’s pi. xviii., 
fig. 2, is a soft inner layer of the testa (endotesta) of the existence 
of which in other seeds adequate evidence is forthcoming. 
Briefly, Pachytesta seems to have possessed an extensive 
endotesta ( endt .) of spongy consistency, this layer was grooved, and 
with these grooves ridges from the nucellus engaged^ In this way, 
perhaps, the bulky nucellus was supported, and kept adjusted 
and properly centred. The whole arrangement was strengthened 
by sclerised trabeculae, which are portions of the endotesta 
specially differentiated. In essentials, however, there is no struc¬ 
ture present not readily derivable from a simple seed like Steplian- 
ospermum akenioides. That the filling tissue of the endotesta served 
as floating tissue and that Pachytesta was a drift seed, as Renault 
suggests, seems most plausible. The fact that the relations between 
testa and nucellus near the apex were more complex and not wholly 
intelligible as yet—perhaps owing to the forking of the longitu¬ 
dinal furrows and ridges so that they became twelve in place of six 1 
—is not material to the argument. We cannot hope for a full 
knowledge of the structure of this seed till a series of transverse 
sections of a specimen with unusually good preservation is obtained. 
If Pachytesta and Lagenostoma be compared with one another 
the points of resemblance are striking. That transverse sections 
from different heights in the two seeds are being compared, from 
the pollen-chamber region in the latter and across the middle of 
the seed in the former, is not a matter of great moment. In the 
Renault, loc. cit., p. 392, and pi. lxxxiv., fig. 1. 
1 
