1898 ] THE ORIGIN OF GYMNOSPERMS 157 
are organized into a strobilus, a character common to pterido- 
phytes and gymnosperms. But such characters can be used 
only as cumulative testimony. In such evidences as we have of 
the structure of the male gametophyte, however, we obtain 
-some valuable suggestions. Within the mature microspore 
there appears a considerable group of polygonal cells. In liv- 
ing groups of gymnosperms, so far as investigated, there is no 
such structure; and if we look to pteridophytes for suggestion, 
we are constrained to believe that this group of cells is either 
prothallial or sperm mother cells. In either event, it would 
represent a condition of things much nearer pteridophytes than 
is shown by any living seed plant. In view of the discovery of 
spermatozoids in Cycas, Zamia, and Ginkgo, taken in connection 
with the peculiar structure of the male gametophyte just 
described, I am of the opinion that the Cordaites also devel- 
oped spermatozoids. With either hypothesis as to the nature 
of the cells developed within the microspore of Cordaites, in 
seeking for the pteridophyte origin of the group, we are led 
away from such heterosporous pteridophytes as now exist, for 
in them the male gametophyte is much more reduced than in 
Cordaites, in fact, more reduced than in most living cycads and 
conifers. 
Additional testimony to the same effect is furnished by sec- 
tions of the seeds of Cordaites. In addition to the remarkable 
nucellus beak, which probably has no phylogenetic significance, 
the large pollen chamber is the most conspicuous feature. This 
is sometimes so extraordinarily large that it occupies the whole 
upper portion of the nucellus, and has been observed to contain 
numerous pollen grains. The pollen chamber is a well-known 
Cycad feature, and seems to be associated with the early devel- 
opment of siphonogamy. By means of it, the tubular outgrowth 
from the antheridium wall is reduced to a minimum, and may 
coexist with spermatozoid development, as shown by Hirase, 
Ikeno, and Webber. 
The testimony all indicates that in Cordaites we have the 
beginnings of a siphonogamic line, brought about by the reten- 
