1898] THE GROUPS OF ANGIOSPERMS 345 
diverge from the Ranales in various ways, some following one 
- direction of development, some another. 
The Rheeadales have long been recognized to be closely 
lated to the Ranales, for the Papaveracee show analogies with 
the Nympheacez in their mostly numerous stamens, and in the 
weasional occurrence of several (though united) carpels in the 
gyneceum. 
The Sarraceniales furnish a parallel series to the Rhceadales. 
| This series shows much in common in the arrangement of floral 
parts with the Nymphezaceze and Papaveracee. The placenta 
- tion in the Sarraceniacez is the principal character distinguish- 
ig the series from the Rhceadales. 
I have considerably extended the Rosales, which very often 
have apocarpy and hypogyny or perigyny in common with the 
tales, but more often show syncarpy and epigyny. The 
= comprises the Saxifraginez, the Rosiflore, and the Legu- 
Mose of Eichler. We have never been able to discover a 
“prehensive distinction between the Rosacea and Saxifrag- 
«te; nor is there any dividing line between the Rosaceae and 
the Leguminosae sufficient to warrant putting them in separate 
oe It May seem strange that the Podostemonacee stand at 
ihe beginning of Rosales, but Warming has shown their connec- 
‘ion with the Saxifragacee on the basis of floral structure. The 
uminosae and the Connaracee are to be regarded sister- 
s of the Rosacez, since the families of Rosales are, as a 
© 80 nearly related that it is difficult to conceive of them 
Be ttions, The Saxifraginee, however,. form a center 
‘ig: °pment, from which the Podostemoninee branch in one 
10n, the 
In 
Men 
‘a 
on 
& Oxalidaceas are followed by the zygomorphous and oligom- 
“ting 
