340 BOTANICAL GAZETTE [May 
But it isa striking fact that these departures occur in several 
of those families which were formerly known as apetalous. This 
seems the more reason why all these families should occupy a 
lower place in the system of dicotyledons. The manner of fer- 
tilization and the development of the embryo sac is here not so 
fixed as in the other dicotyledons. The Proteacez, also, which 
I have placed much lower than Eichler, show a striking departure 
from other dicotyledons in that the number of cotyledons may 
reach even eight. 
MONOCOTYLEDONS. 
Among the nonocotyledons those series in which typical 
achlamydeous flowers occur represent the earliest stages of 
development. These are the Pandanales, Helobia, and Glumi- 
flore. Of these the Pandanales are most primitive, because of 
indefiniteness of floral parts, and the spiral arrangement of se 
mens. Whether the Helobie or the Glumiflore stand next i 
order is difficult to decide. Owing to the great instability pre 
vailing in the Helobiz, I have preferred to let them precede the 
Glumiflore, and have left the Gramineze and Cyperacee together 
in Glumiflore. 
Common to the three series Principes, Synanthes and Spathi- 
floras is the appearance of a floral envelope, which in some ae 
is suppressed by the more vigorous development of the spat P 
Several subseries under the Farinose must be set ra 
notably the Flagellariineee, which do not affiliate closely ae 
any other family. The subseries Enantioblasta is sams ne 
by its orthotropous ovules, while the remaining Farinos® "4 
anatropous ovules. : 
Most of the Liliiflore belong to the subseries Li a 
have regarded: the Juncacez as a separate subseries, ge ; 
being intermediate in the structure of their albumen iia the 
the Farinose and the Lilliiflore. Likewise I have ee 
Iridaceze as subseries Iridinez, basing it upon the leafy . n the 
liner. | 
. . . iy 
ment and division of the style branches, as wellias 
position of the leaves. 
