THE MORPHOLOGY AND AFFINITIES OF GNETUM 1 75 
were observed even in the endosperm. Coulter's account follows 
the embryo-formation farther than I have done, showing that the 
multinucleate terminal cell becomes divided up by cleavage walls 
into uninucleate cells. 
The most important point about this process of embryo formation 
is that the free nuclear divisions characteristic of Gymnosperms are 
eliminated. Division of the fertilized egg in G. sp. jj at least, is 
accompanied by wall formation as in Angiosperms. In another im- 
portant respect therefore Gnetum takes its place with the Angio- 
sperms. This elimination of the free nuclear stage was foreshadowed 
in Welwitschia (Pearson, 21) in which only two free nuclei are formed 
and are then separated by a cleavage wall. 
12. General Conclusions 
The significance of the conditions found in Gnetum has been dis- 
cussed in connection with the descriptions of the various structures 
examined. It remains to point out certain general conclusions which 
may be deduced from a consideration of the evidence as a whole. 
The conclusions are in respect to the relationships of the Gnetales 
(a) to each other, (b) to the Gymnosperms, (c) to the Angiosperms 
and also in respect to the origin of Angiosperms. 
(a) The evidence shows that the three genera of the Gnetales 
are widely different in many morphological points and yet are really 
phylogenetically related. In habit, anatomy, gametophytic struc- 
ture, endosperm and embryo formation the genera are all very dif- 
ferent. Nevertheless in regard to each of these subjects there are 
essential points which indicate a common phylogeny. For example, 
the occurrence in the flowering axis of Gnetum of a type of vessel 
found elsewhere only in Ephedra can be explained only on the basis 
of a phylogenetic connection. Other examples are the similarity in 
flower structure and in the early stages of embryo formation, although 
the details vary greatly in the three genera. If all three are really 
related it follows from their present great differences that the Gnetales 
must formerly have been a very large and diversified group. While 
the three genera can by no means be arranged in a series according 
to the degree of specialization it is clear that Gnetum is in many 
respects the most advanced, that in a few respects Welwitschia has 
progressed furthest, and that Ephedra is nearest to the ancestral 
Gymnosperm stock. 
