346 BOTANICAL GAZETTE [May 
laceze, which occupy mostly the same plane as the Geraniacez, 
but in the other direction they approach very near to the 
Cneoracee, Rutacee and Simarubacee. Since the Rutacez are 
intimately connected with the Simarubacee, Burseracez and 
Meliacez, it is possible to conceive of these families all together 
as Geraniinee. Morphological advance in another direction is 
seen in the Malpighiinee. It is expedient also to distinguish 
the Polygalinew, Dichapetalineae and Tricocce as subseries, 
which possess scarcely any surviving common features. 
In the Sapindales, considering the entire organization, foliar 
arrangement and anatomy, we are forced to establish a large 
number of subseries. At the first are the Buxinee, provided that 
the simple perianth is primitive rather than reduced. The 
Empetracee, Coriariacee, and Limnanthacee# occupy similar 
rank in the character of perianth and the number of ovules, but 
are withal so distinct that each family must be regarded as the 
representative of a subseries. The resinous Anacardiacet may 
be regarded as an independent subseries coordinate with and 
opposed to the Celastrinee and Icacinex. The pee 
comprise the closely related Aceracea, Hippocastanaces’ s 
Sapindacez, and the Sabiineze, Melianthinex and Balsamminine®- 
- The setting apart of so many subseries shows there 2.50 ; 
cient ground for the derivation of the more complex families ° 
the series from the simpler. 
The Rhamnales are now confined to t 
chlamydez with opposite stamens. 
As in the Geraniales and Sapindales, 
he tetracyclic Archi- 
latter case complete syncarpy prevails. 
Malvales follow the Geraniales, Sapindales, 
the close relationship of the Elaocarpace® and 
to the Parietales, within which the floral evolutl les cleatly 
reached very complicated floral types. The Ma is considera 
show how a family may reach in certain directions ape behind. 
ble stage of development, while in others it see? families 
Conditions are present which show that the near rela 
. 
