404 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1907. 
where they have been investigated, scarcely differed from the spo- 
rangia of certain Ferns. Here, so far as the evidence extends, there 
is a wide difference from any known Gymnosperms and a near ap- 
proach to the Filicinee. 
3. The anatomical structure. There is probably no constant dis- 
tinctive character in the structure either of stem or leaf. The anatomy 
of the stem in Lyginodendron does not differ esentially from that in 
Poroxylon, which appears to find its nearest allies in the Cordaitee, 
while other plants, such as Pitys antiqua and Dadoxylon Spenceri, 
which likewise possess primary centripetal wood in the stem, prob- 
ably also belong to the latter group. It would not always be possible 
to tell from the structure of the stem alone whether a given plant 
belonged to the Pteridospermez or the Cordaitex. So, too, with the 
leaf. The anatomy of the petiole and lamina in Medullosa is essen- 
tially that of a Cycadophyte, while in Lyginodendron it is that of a 
Fern. Taking the swm of anatomical characters, however, the Pteri- 
dosperms, so far as we know them, are much more Fern-like than any 
typical Gymnosperm. We might frame a provisional diagnosis of 
the Pteridospermez as follows: Male and female sporophylls little 
differentiated from the vegetative foliage; no cones formed. Anat- 
omy of either stem, or leaf, or both, of a Filicinean type, as was also 
the habit. 
The chief practical reason for keeping the Pteridosperms apart 
from the Gymnosperms is their manifestly more primitive character, 
shown in one respect or another throughout the group. Even in the 
seed, the most advanced of their organs, possible primitive indications 
are not wanting. In Physostoma, with its integument breaking up 
into a ring of free tentacles taking the place of the micropylar tube, 
we have a unique form of seed investment. The characters of the 
male fructification, if we may take Crossotheca as a fair example, 
appear to have been frankly Cryptogamic, and the same applies to 
the anatomy of such plants as Sutclifia and Heterangium, genera 
which show such evident relations to Medullosa and Lyginodendron, 
respectively, that we can not doubt their being Pteridosperms. It 
seems to me desirable to give full weight to primitive characters such 
as these and to keep the Pteridosperms distinct, rather than to merge 
them in the Gymnosperms, a group which has departed so much fur- 
ther from Cryptogamic traditions. At the same time I fully recog- 
nize that this is a matter of expediency rather than of principle, for 
further research will undoubtedly tend to fill up the gap between the 
two classes. 
A more fundamental question is that of the relation of the Pterido- 
spermez to the Cryptogams. All the characters in which the Pteri- 
dosperms show Cryptogamie affinities, whether in anatomical struc- 
ture, in the morphology of the sporophyll, or in the nature of the male 
