128 Sahni. — On Certain Archaic Features in the Seed of Taxus 
comparison is possible. But if we agree with the conclusion reached by 
Professor Oliver, that the part of the seed lying below the level of the 
‘ chalazal foramina ’ is a phylogenetically recent formation, the difference 
from Taxospermum does not appear so great. For the sake of clearness, I 
have interposed between these two seeds a hypothetical intermediate 
condition (Fig. 7, No. 9) which may for convenience be referred to as the 
p ve-Torreya stage. Between Taxospermum and pr ^.-Torreya there are only 
two points of difference. Firstly, pre-Torreya has at the base two strands 
which enter the stone through two incurrent canals, and between which an 
appreciable portion of the sclerotesta has been newly intercalated, just as in 
Taxus. Secondly, the excurrent canals in pre-Torreya are nearer to the 
micropyle than they are in Taxospermum — it has already been suggested 
that they have a tendency to move towards the micropyle. The dotted 
portion of the stone in Fig. 7, No. 9, may thus be described as having grown 
at the expense of the part shaded with lines, to the extent that the excurrent 
canals have moved forward. At the same time, however, the newly inter- 
calated portion of the stone (shown cross-hatched) must be regarded as 
having increased in size at the expense of the dotted portion of the stone, 
for, ever since the splitting of the main supply bundle of the ancestral type, 
the two incurrent canals (conveying the resulting bundles) have also steadily 
travelled towards the micropyle. 
Pr e-Torreya is thus a strange kind of seed, in which the stone is pierced 
in the principal plane by two pairs of canals (one excurrent, the other 
incurrent), both of which are, in a phylogenetic sense, moving towards the 
micropyle. From this stage Torreya is only one step farther. We have 
only to imagine that the hinder pair of canals, having moved faster than the 
front pair, has overtaken the latter and become confluent with it. A glance 
at the figures will show that the obvious result of this process is that the 
dotted region of the stone is gradually squeezed out of existence. It will 
thus be seen that the incurrent canals of Taxus are not strictly comparable 
to those which convey the nucellar bundles of Torreya into the seed-cavity 
— a want of correspondence which is in keeping with the fact that the 
strands conveyed by the canals in the two genera are themselves not homo- 
logous, being the main supply bundles in the case of Taxus and nucellar 
bundles in Torreya. 
It is of some interest to find that although we have derived the seed of 
Torreya from the Cardiocarpus type through a long succession of varying 
forms, our interpretation of this peculiar seed entirely agrees with that 
which Professor Oliver arrived at from a direct comparison between Torreya 
and Cardiocarpus. If we look back at the series of figures from Torreya to 
Cardiocarpus , it will be seen, from the conventional method of shading that 
we have throughout adopted, that in Torreya the small apical region of the 
stone is the representative of the entire stone of the ancestral type, which 
