57 
united has been the course of evolution, thus A/agnolia precedes 
Dianthus. Further, it has long been recognized that from petals- 
or sepals-separate to petals- or sepals-united is the direction of 
change; Polypetalae precede Sympetalae. In other words, the 
union of parts is a very important principle in varied aspects of 
plant evolution. 
Separate placentae (that is, parietal placentation) to united 
placentae (that is, axile placentation) is then only a part of a 
broad general principle. It has been suggested that the phrase 
“ontogeny recapitulates phylogeny’? should be changed to 
“ontogeny throws light on phylogeny.” Similarly, understand- 
ing the development of placentation throws light on the classi- 
fication of flowering plants. Three stages are suggested: (1) 
Families having carpels separate or single: Magnolia, Rosa, Protea; 
(2) Carpels sy but placentae separate: Papaver, Cistus, 
Salix, Betula; (3) Carpels united and placentae united: Geranium, 
Hydrangea, Dianthus, and nearly all Sympetalae. In plants, as 
in animals, improved placentation means more effective nourish- 
ment of the next generation. 
The question arises as to what extent is it justifiable to apply 
placentation as a main principle in the classification of flowering 
plants. A minimum would be the bringing together the Crstus 
— 
and Papaver groups. This is in fact done in the Bentham and 
Hooker, Eichler, Warming, Wettstein, Rendle, and Hutchinson 
systems. Their separation in the Engler system must be con- 
sidered as a mistake. It is still a question along what different 
lines change in placentation may have occurred. Further, to 
what extent do these ideas conform with paleobotanical evidence? 
I believe they are not contradicted by them. 
SYSTEMATIC BOTANY 
Eleocharis: Galapagos Islands: Local Flora 
By Henry K. SVENSON 
In addition to executive work of the department during 1937, 
I continued studies in the Cyperaceae, chiefly in the genus 
Eleocharis. My visit to European herbaria during the summer 
