298 BOTANICAL GAZETTE [NOVEMBER 
27). The early divisions of the endosperm nuclei occur more 
rapidly than those of the embryo, and very soon cell walls make 
their appearance. Even after the formation of cell walls, the 
endosperm continues to form so rapidly that when the seed is 
mature the small, insignificant, yet massive embryo (figs. 28-30) 
is practically inclosed in rich endosperm tissue. In its development 
the embryo does not always divide into regular octants (figs. 28-30), 
but at times is quite irregular. Very frequently in its early develop- 
ment it looks more like a pteridophyte embryo (fig. 28), and is 
characterized by a long suspensor. 
Relationships 
The Umbelliferae are very closely related to the Araliaceae 
and Cornaceae, with which families they form an alliance or order 
known as the Umbellales, or the ‘‘Umbelliflorae” of ENGLER. 
Although several workers, notably HALLIER (30), WETTSTEIN (83), 
and WERNHAM (82), have elaborated various tables showing the 
probable phylogenetic relationships, the scheme of ENGLER is still 
followed by most workers. In ENGLER’s classification the dicoty- 
ledons are divided into two great divisions, the Archichlamydeae 
and Sympetalae, which, according to COULTER and CHAMBERLAIN 
(12), however, show no sharp distinction, for ‘“‘sympetalous forms 
occur among the former, and polypetalous forms among the latter.” 
The distinction is laid principally upon a single character, namely, 
apetaly or polypetaly for the Archichlamydeae, and sympetaly for 
the Sympetalae. Without doubt this is pressing a single character 
too far, and as a result the Umbellales “‘stand so stiffly apart from 
other Archichlamydeae as to raise the question whether they do 
not really belong among the higher Sympetalae”’ (12). 
It will not be out of place to indicate such evidence as there is 
for such an assumption from the taxonomists’ point of view. For 
this purpose the recent extensive work of VIGUIER (78) may be cited. 
In his chapter dealing with the relationship of the Araliaceae with 
other families, he definitely proves the close relationship of the three 
families forming the order Umbellales, claiming, for example, that 
the only character which separates the Araliaceae from the Umbel- 
liferae is the drupaceous fruit of the former, whereas the two 
