Notes on Trigonocarpus and Polylophospermum. 99 
The epidermis appears to be still in situ on the convex sides 
of the pollen-chamber, but from the level of the pollen-chamber 
floor and extending almost to the chalaza it is completely 
separated from the nucellus and in places lies in contact with the 
lining of the testa (photos. 1 and 2, E). It is quite evident that the 
gaps on either side of the prothallial region of the nucellus (photo. 
2, G) are the result of splitting, and the question arises as to what 
were the relations of this “ epidermis” to the inner surface of the 
testa, with which it is in contact over part of its course. So far as 
can be judged from a study of these preparations as well as those 
of other French seeds of like organization, there would not appear 
to be any positive evidence for supposing this epidermis to have 
been adherent to the testa. So that it may be supposed that the 
nucellus stood originally free within the testa from the chalaza 
upwards (as in Lcpidocarpon). 
The alternative view that the relations were as in recent 
Cycads, or, to go back to older types, as in Lagenostoma the seed of 
Lyginodendron, is unsupported by observational data and depends 
on certain theoretical considerations. On this view the collar or 
flange (photo. 3, C) might be regarded as marking the limits of the 
free surface of the nucellus, but one difficulty in adopting it rests 
in the absence of a return-epidermis running to the micropyle. 
We may pass now to a consideration of the main body of the 
nucellus. This is occupied by an extensive prothallium of which 
the preservation is fairly good, though no archegonia are evident in 
this specimen. The remaining hypoderm of the nucellus, stript of 
its surface layer, is found cut in tangential sections at places (as at 
T, photo. 1), and here are seen the tracheal strands, so marked a 
feature in French Pernio-carboniferous seeds. Elsewhere, too, 
where the wall is cut in a radial direction the tracheal elements are 
evident, but it is when they can be studied in surface view that 
they are seen to the best advantage. A portion of the tangentially cut 
area ( T , photo. 1) is shewn enlarged in photo 4. The tracheal 
strands follow a somewhat irregular course and at places shew 
anastomoses. The form of the tracheides is rather variable. 
Whilst some are long and relatively narrow, others, especially such 
as occur at the junctions or when a strand follows an undulating 
course, are short and broad. The type of thickening of the wall is 
a close scalariform easily confounded with spiral. So far as can 
be judged from the fragmentary specimens, the whole nucellus 
below its epidermis was ensheathed in a series of anastomosing 
