1898 } THE ORIGIN OF GYMNOSPERMS 165 
feature demanded by the retention of the megaspore, therefore, 
was not the scattering of the microspores, but the development 
of siphonogamy. That the first retained megaspores were 
exposed to the microspores can hardly be doubted, and in such 
cases we now know that the spermatozoid habit must have been 
retained, and that no tube, or a very small protuberance of the 
antheridium wall, was needed to discharge the spermatozoids 
sufficiently near the oosphere. If chemotropism can explain 
the guidance of a pollen tube through much intervening 
tissue, it would certainly be sufficient to cause the protrusion 
of an elastic antheridial wall. In the very few illustrations 
of Cordaites obtained, the megaspore is but slightly covered by 
sterile tissue at the bottom of a deep pollen chamber, and a very 
slight development of tube is necessary. The same condition is 
continued in the cycads, and thus the habit of siphonogamy 
may have been gradually built up. As siphonogamy developed, 
the gradual failure of the sperm mother cells to organize sper- 
matozoids followed, and presently, almost exclusively now in 
gymnosperms, sperm mother cells are found to function directly 
as male gametes, without further organization. 
The secondary results which followed the retention of the 
megaspore were numerous. The well-known effect of fertiliza- 
tion upon adjacent tissues necessarily involved at least the 
sporangium, and the seed resulted. The presence of abundant 
available nutrition and favorable conditions induced the imme- 
diate germination of the oospore, which the development of a 
resistant tissue about the sporangium checked. As a conse- 
quence, the development of the embryo was thrown into two 
stages, the intra-seminal and the extra-seminal. 
In the case of the angiosperms, however, another tendency 
was connected with the retention of the megaspore, namely, the 
tendency of the sporophyll to enclose the megasporangium, a 
tendency so evident in such pteridophytes as Isoetes and Mar- 
silea, that the direct pteridophyte origin of the group seems 
more natural than an origin from so specialized a type as the 
gymnosperms. Given the reduction of spore production toa 
