baccata , with Remarks on the A ntiquity of the Taxineae . 1 2 7 
for we are yet wholly ignorant as to the nature of the plant that bore the 
Palaeozoic seed, but the resemblances between the seeds certainly appear to 
be due to more than a chance. Such points of agreement as those in 
general shape and size may be ignored as being superficial. But it is at 
least a fact of some significance that, so far as known, of all the fossil seeds 
which show evidence of a Cordaitean affinity, it is Taxospermum that shows 
the greatest amount of fusion between nucellus and integument : an 
important advance in the direction of Torreya and Taxus. (See Bertrand, 
1907 , p. 216.) 
Apart from the difference in the position of the ‘ outer ’ vascular 
system — which I have just attempted to explain — the most important 
difference between the two seeds is that in Taxus there are two basal 
supply bundles instead of the single one in Taxospermum . But there is 
every justification for the view that the two basal ovular strands, lying in 
the principal plane, which are a striking characteristic of the Ginkgoales, 
Taxales, and Coniferales, have been derived by the splitting of a single 
median bundle ; for it is an observed fact that a centralized main supply 
prevails in the overwhelming majority of known Palaeozoic Gymnospermous 
seeds, whether platyspermic or radiospermic. If we imagine the two 
incurrent canals of Taxus to approach each other and fuse into a single 
median canal, the two contained bundles at the same time fusing into one, 
the thickened angles in these bundles would coalesce into a vascular mass 
exactly corresponding in position to the subnucellar tracheal pad charac- 
teristic of most Palaeozoic seeds with a centralized main supply. It is now 
easy to see the significance of the fact, to which attention was drawn in the 
descriptive part of the paper, that these thickened angles in the ovular, 
strands of Taxus are situated nearer to the nucellus than any other part of 
the vascular supply. 
The absence of a definite ‘ internal ’ (nucellar) system of strands in 
Taxus will hardly be taken as a serious objection to the comparison here 
instituted, for it is doubtless related to the absence of fertilization by motile 
sperms. I have already (p. 120) offered some reason for the view that the 
thickened angles in the ovular strands in Taxus probably mark the points 
of origin of the nucellar bundles in the ancestral type, although these 
bundles have become extinct in the modern genus. 
Having considered whatever evidence there is for deriving the seed of 
Taxus from a type like Taxospermum , we shall now attempt to see in what 
relation the seeds of Torreya and Cephalotaxtis stand towards either of the 
former seeds. Nos. 7-1 1 on Fig. 7 illustrate the view which I consider to 
be the most plausible. It will be noticed that I have attempted to derive 
the seed of Torreya also from the same source ( Taxospermum ) but along 
a line distinct from that of Taxus . At first sight the seed of Torreya 
appears so different from Taxospermum that it is difficult to see how a 
