On Gymnospermous Seeds. 15 1 
first place both show a bulky zone within the testa, broken up into 
longitudinal chambers by the intrusion of trabeculae or radial 
plates. If the trabeculae of Pachytesta instead of converging at the 
points rp. ran out radially from the bases of the grooves g, and if at 
the same time the inner containing membrane of the endotesta 
were indurated, with these slight changes the peripheral regions of 
the two seeds would be essentially identical in structure. The 
presence of the vascular strands v in the chambers of Lagenostoma 
is the most important difference. 
A second point of agreement resides in the grooving of the 
endotesta and canopy, and in the interlocking with these grooves of 
longitudinal ridges on the nucellus. It is hardly an exaggeration 
to say that both seeds possessed a grooved “ canopy ” which 
engaged with longitudinal ridges on the nucellar surface. 
What may have been the significance of this contrivance, 
whether it was of importance mechanically in properly centering 
the nucellus, or whether it had some other function, is a matter 
difficult to decide. It is moreover an open question whether the 
relations that may have existed between nucellus and endotesta at 
a later stage in the development of these two seeds can be fully 
realized from the fossils, for like all other palaeozoic seeds they 
are only known in a stage of development just prior to ferti¬ 
lization. In any case the peculiar relations of nucellus and 
integument in Pachytesta and Lagenostoma suggest that this condition, 
whatever its signification here, may have formed the starting point 
of those much more complicated foldings which exist between the 
central and peripheral portions of the seed in the little-known genus 
Torveya. The next section of this paper deals briefly with the nature 
and origin of these foldings in that genus. 
A minor point of agreement in Lagenostoma and Pachytesta is 
in the form of the tracheal strands which radiate from the base of the 
nucellus. These strands running in the nucellar wall of Pachytesta 
are shewn in Brongniart’s PI. xviii., fig. 2 , /, in transverse section. 
They are compact and tangentially elongated, having a lenticular 
section. In good specimens of Lagenostoma the characters of these 
nucellar bundles are essentially the same. Figures of the latter 
have not however been published. This particular form of strand 
does not appear to be very common in palaeozoic seeds. 
Torveya. 
The curious relations exhibited by the seeds described above 
suggested a comparison with Torveya, remarkable as being the only 
