haccata, with Remarks on the Antiquity of the Taxineae. 129 
has steadily undergone reduction, having been encroached upon in succes- 
sion by two phylogenetically younger portions of the sclerotesta, intercalated 
one after the other at the base of the seed. Each of these newly formed 
portions of the stone has shown a persistent tendency to grow in the 
direction of the micropyle. The modern genus Taxus illustrates the 
triumph of the older of these, which forms almost the whole of the sclero- 
testa in this genus ; in Torreya it is the younger of these newly created 
regions of the stone that protects nearly the whole of the seed. 
Turning now to the third genus, Cephalotaxus , the question arises in 
what light we are to regard the stone of this seed. We have seen that in 
the ancestral history of Torreya the ‘ chalazal foramina ’ have probably 
shown a continual tendency to shift nearer and nearer to the micropyle. 
In Cephalotaxus , however, the most careful search .reveals no trace of any 
perforations in the stone. May we conclude that in this genus the ‘ chalazal 
foramina 5 have actually passed beyond the micropyle — that the entire 
stone of Cephalotaxus corresponds to the newly intercalated portion of the 
stone of Torreya and to the small basal disc of Taxus ? This, at any rate, 
seems to me the most natural conclusion, strange as the suggestion may 
appear. Of course, the alternative suggestion may be made that the seed of 
Cephalotaxus differs essentially from that of Torreya only in having lost 
its nucellar bundles by reduction in situ , and that consequently the 
‘chalazal foramina’ which once admitted these strands have also become 
obliterated. If it were shown that the young seed contains vestiges of the 
nucellar bundles, or of foramina in the stone which become blocked up as 
development proceeds, this view would be indisputable. But in the absence 
of positive evidence on either side the conjecture first put forward is prefer- 
able, being more in conformity with the general line of argument previously 
adopted. The fact that there are in Cephalotaxus , as in Torreya , two 
separate supply bundles at the base would alone suggest that the part of 
the stone lying between them is a comparatively recent formation ; and 
since the bundles run up as far as the micropyle without branching, this 
conclusion must evidently apply to the entire stone. 
If the tendencies outlined above have had any significance in the 
evolution of the seeds in question, we have further evidence for the view — 
which is already well established on other grounds — that the Yews are 
descended from Cordaitalean ancestors. For the fact that it is possible to 
arrange these seeds into a more or less continuous series, helps us further to 
bridge the structural gap between the two groups, wide though the gap in 
time may be. The hope may therefore be well founded that rocks of an 
intermediate age will reveal evidence of the existence of seeds transitional 
between Taxospermum on the one hand, and Torreya and Taxus on the 
other. 
The comparison of the seeds of the Taxales with the older seeds 
K 
