60 Salisbury . — On the Structure and Relationships of 
of the surface, the longitudinal dimension can be estimated, it is about three 
times as great as the width (0-09 mm.). As seen in this view, the cells are more 
or less oblong in shape, whilst cut transversely they appear distinctly convex 
on the external face, a feature which seems to point to the absence of close 
contact with an inner flesh. 
Under the best conditions of preservation, the epidermal cells resemble 
very closely the secretory sacs and, like them, have the appearance of 
possessing a thickened wall. In several places the outer layer of the 
exterior walls, together with part of the middle lamellae from between each 
pair of the component cells, has split away, so that in transverse section it 
appears as a thin membrane with pegs projecting inwards. Evidently this 
outer layer was of a durable character, as it can occasionally be recognized 
even where the interior tissue has decayed away, and is much less subject 
to contraction, suggesting that it may probably have been a cuticularized 
layer such as we find with similar pegs amongst modern plants. 
Such a specialized epidermis we should hardly expect to find in 
a completely enclosed structure like the nucellus, if it were in close contact 
with a soft inner flesh. 
Between the epidermis and the tracheal sheath there intervened a zone 
of soft parenchymatous elements in which numerous secretory cells were 
present. The ground-tissue was without intercellular spaces and formed of 
polygonal cells with very thin walls (0*03 mm. to 0-08 mm. x 0*15 mm.). 
The secretory cells were embedded in this ground-tissue, and where they did 
not occur singly, formed radial plates of varying extent and usually consist- 
ing of a single row of secretory elements, though more rarely at the base of 
the seed of two such rows. The greatest number of secretory cells which 
were present in any one radial plane appears to have been five. The 
proximity of the radial plates and the number of elements in each decreased 
as the apex of the seed was approached, so that some two-thirds from the 
base of the pollen chamber the plates were about a third of a millimetre 
apart and reduced to one, two, or at the most three elements. The secre- 
tory sacs themselves (0-06-0*09 mm.) exhibit much the same structure as 
those of the sarcotesta, but with thinner walls. In the longitudinal direction 
the carbonized contents show segmentation into short lengths, but the 
septation thus produced probably bore no relation to transverse walls. 
In two of the seeds (viz. S. 31, U. C. Coll., and Man. R. 1161), besides 
the longitudinal ridges already referred to, others more pronounced are 
present, occupying positions opposite to the commissural ribs, and bear no 
relation to contained radial files of secretory elements. In the series S. 33, 
U. C. Coll., with its excellently preserved nucellus, no such ridges are how- 
ever to be seen, which may be correlated with the fact that this seed was 
probably in an immature condition, judging from its relatively small size, 
the thinness of the testa, and the slight development of the ribs. In S. 31 ,g, 
