THE MORPHOLOGY AND AFFINITIES OF GNETUM 1 49 
That which is called the integument of the Welwitschia ovule 
is, according to the theory, really an ovary. Lignier and Tison justly 
lay much emphasis on the style-like character of its projecting tip. 
Furthermore we have already called attention to the fact that its 
development and anatomy both in Ephedra and Welwitschia indi- 
cate that it is an ovary of two fused members. The stamen cycle of 
Welwitschia is fused at the base and divided above into six parts. 
But its anatomy shows that there are really two members (as in the 
other cycles) which branch above. This is assumed to be the an- 
cestral condition for the group. The perianth consists of two pairs 
of bracts. In Welwitschia the first pair are connately used and the 
second pair represented by bractlets. In Ephedra both pairs are 
connate. The ancestral flower of the Gnetales therefore consisted 
of an ovary of two fused members, tw^o stamens and two pairs of 
decussate bracts. 
It must be admitted that one finds difficulty in imagining how 
the staminate flower of Gnetum can have been reduced from such a 
type. This flower consists of a single stamen surrounded by a single 
envelope. It must be assumed that the ovary, one stamen, and one 
pair of perianth bracts have disappeared and that the other stamen 
has taken a position at the top of the axis. Although this seems a 
big assumption to make I believe it is justified particularly in view 
of the certain abnormalities which will be described in full elsewhere. 
They consist in brief of male flowers which had grown out into axes 
of considerable length and complexity. 
We are now in a position to apply the general conception of the 
Gnetalean flower to the structures in the female flower of Gnetum. 
It is plain that the inner envelope of Gnetum is even more like an 
ovary than that of Welwitschia for it not only resembles an ovary in 
form and anatomy but it also bears a style on which the pollen is 
caught and in which some of it germinates. Accordingly we conclude 
that the first envelope is really an ovary whether or not it is the 
homologue of the Angiosperm ovary. The second and third envelopes 
then represent fused bracts which may or may not constitute a true 
perianth. The stamen cycle has disappeared. The whole view 
receives strong support from the discovery of abnormal flowers bearing 
stamens within the second envelope. 
This body of evidence seems to demonstrate that we have at last 
obtained the proper interpretation of the envelopes of Gnetum. 
