THE MORPHOLOGY AND AFFINITIES OF GNETUM I5I 
both leaf and stem. There is no reason why they cannot occur on the 
base as well as on the sides of the sporangial cavity. Therefore the 
obviously cauline character of the ovules of Gnetum cannot be used as 
an argument against its being related to the type from which Angio- 
sperms were derived. Indeed it rather supports that view because 
cauline ovules are found chiefly among primitive Angiosperms. 
The stamen of Gnetum has already been compared with that of 
Angiosperms. 
The perianth of the Gnetales according to our interpretation 
consists of two pairs of connate bracts. Among the Amentales the 
perianth consists of similar small colorless bracts which are either 
distinct {e. g., Myricaceae) , or fused {e. g., Juglandaceae, Betulaceae). 
Therefore it appears that there is more than mere analogy between 
the perianth of primitive Angiosperms and that which we have called 
perianth in Gnetum. 
We conclude that in regard to every part the flower of Gnetum 
closely resembles that of primitive Angiosperms. And we have 
already seen that the arrangement of the flowers themselves particu- 
larly in the abnormal spiral strobili is just the same as in those primi- 
tive Angiosperms. The flower of Gnetum is, therefore, not far re- 
moved from the type from which that of Angiosperms was derived. 
This view implies that the Amentalean flower is the most primitive 
type found in Angiosperms. Another admittedly primitive type is 
that of the Ranales with numerous carpels, stamens and floral leaves 
all arranged in spirals. But if the Gnetalean derivation of the Angio- 
sperms is the correct one this type must be considered more specialized 
than the Amentalean one. In this connection it is interesting to 
recall the structure of those abnormal strobili previously referred to 
in which there is a mass of ovulate flowers above the usual set of 
staminate ones. If the Gnetalean flowers be considered as simple 
ovaries and stamens and not as reduced from a hermaphrodite con- 
dition, the resemblance of these abnormal strobili to the Ranalean 
flower is marked. But all the evidence indicates that they are really 
reduced and therefore we must conclude that this resemblance is 
only superficial. 
(b) Nucellus and Archesporium. — As stated by Strasburger (25) 
the archesporium always consists of two or more hypodermal cells. 
I have examined an abundance of material in all the early stages and 
have always found that as soon as an archesporium is recognizable 
