of Bennettites gibsonianus , Carr. 439 
oblique direction, so that the apex and the base are not 
touched, the testa therefore seems to be of uniform structure 
all round. It is only in preparations in which the section 
passes quite close beneath the apex of the spadix that we 
are able to see some of the seeds in exact transverse section. 
They are, in fact, so disposed that the longitudinal axis is 
everywhere perpendicular to the convex surface of the 
spadix 1 . Exact longitudinal sections are not easy to obtain, 
but they show that each seed, which is about three milli- 
meters in length and two in breadth, is borne on a long stalk 
which is the structure hitherto termed the ‘ cord 5 by 
Carruthers ; and moreover that the seed is not, as a matter 
of fact, lodged, as it appears to be, in the tissue of the 
external layer, but fills a flask-shaped pit or depression, and 
has a slender process passing through the narrow open canal 
of the pit, and terminating at the surface of the spadix. 
This surface has a peculiar appearance from the number of 
seed-containing pits with their narrow orifices, and reminds 
us of Dorstenia , or, as Carruthers 2 aptly suggests, of Tam - 
bourissa. Carruthers’ figure 3 is not readily intelligible with- 
out a knowledge of the object, and even then is not clear. 
The exactness with which the seed fills the pit which 
contains it is peculiar : there is nowhere the smallest interval 
between it and the wall of the pit, so that it is difficult to 
be quite sure whether or not there is continuity between the 
two, or how far it may extend ; I am obliged to suppose that 
at the base such continuity exists. 
In many seeds the contents are entirely destroyed, and then 
we find crystals and structureless remains of organic substance 
in their place. In other and not very few cases, they are 
very well preserved, and may be seen to be an embryo, in 
which radicle, hypocotyl, and the two fleshy cotyledons lying 
one on the other as in the bean, can be distinguished, if the 
1 In Carruthers’ work they are represented in the transverse section in pi. 59, 
f. 5 ; in the tangential section in pi. 59, figs. 1, 2, 4 ; in the oblique radial section in 
pi. 59, fig. 3 ; and in Plate XXV, fig. 9 of the present paper. 
2 Carruthers, loc. cit., p. 698. 
3 Carruthers, loc. cit., pi. 59, fig. 6. 
