5 22 
Ber ridge . — The Structure of the Flower of 
originally arranged in multiples of five, and the possibility of a relationship 
with families prevailingly pentamerous must be taken into account. 
It would appear to be a point of some significance that the vascular 
structure of the inferior ovary is practically identical in Castanea and 
Castanopsis. That structure corresponds rather with the multilocular con- 
dition in the former than with the trilocular form of the ovary in the latter. 
In Castanea six loculi are most commonly present, situated opposite the 
perianth leaves, and therefore with the six largest main bundles taking the 
position of dorsal carpellary bundles. From such an arrangement it would 
appear that the trilocular ovary of Castanopsis and Qucrcus is derived by 
the obliteration of the loculi opposite the smaller perianth leaves, for when 
a fourth is present it occupies that position. It is very probable also that 
the 6-locular ovary most usual in Castanea is derived in the same manner 
from ovaries containing many loculi, for 12-locular forms are not infrequent, 
i. e. forms in which the number of loculi, and therefore of carpels, corresponds 
with the number of carpellary bundles. The variations, therefore, in the 
ovary appear to be due to the disappearance of septa only, not of complete 
whorls of carpels, as Eichler (8) suggests, and the frequent occurrence of 
partial septa in Quercus and Castanea support this view. 
Such a mode of reduction seems to indicate that the Fagaceae are 
derived from ancestors which possessed syncarpous and multilocular 
ovaries ; if the latter were also inferior, the union of the outer walls with 
the receptacular cup would also have tended to preserve the vascular supply 
of the carpellary wall from complete disappearance, since it becomes merged 
in the general supply of the whole flower. 
Comparison with Juglans regia. 
No attempt has been made at present by the writer to work out the 
floral anatomy of other amentiferous forms with a view to a comparison with 
the Fagaceae. The vascular structure of the flower of Juglans regia has, 
however, been very fully described by Van Tieghem ( 24 ) and Nicoloff ( 19 ), 
and also by Benson and Welsford ( 5 ), and has been the basis of much dis- 
cussion concerning the affinities of the group ; a comparison, therefore, can 
be readily made, and, moreover, seems to furnish an explanation of one of 
the features which led Nicoloff to insist so strongly on the axial nature of 
its ovule. 
His Fig. 15, which represents a transverse section of the pedicel in 
Juglans , and which is here reproduced (Fig. 9), differs from a section taken 
through the same region in Quercus Robur mainly in the fact that, while in 
the Oak the outer circle of bundles supplies the cupule only, and the inner 
supplies both perianth and carpels, here the cupule and perianth bundles 
are associated together in the outer circle, while the inner circle consists of 
