5 1 8 Ber ridge. — The Structure of the Flower of 
it is evident that the scales, i. e. the leaves, on these might naturally develop 
later than the floral leaves of the primary axis. 
All the evidence, in fact, which is brought forward by Prantl in support 
of his conclusion that the cupule is an intercalary growth of the axis — 
for instance, the leaf-like nature of the cupular scales in Fagus Sieboldii, or 
the appearance of bracts below and prior to the outgrowth of the valves in 
Castanea — is equally applicable in support of the view that the valves 
are modified secondary or tertiary axes, and not, as Eichler maintained, 
merely bracts covered with emergences. 
The Anatomy of the Flower. 
There is a great similarity in the structure of the flower of the four 
genera of the Fagaceae, and the following description, which applies 
primarily to Castanopsis , is in the main true of all. In Fagus a larger 
number of bundles is present than in Castanopsis , while in Quercus they 
show much irregularity both in number and degree of fusion, but there is no 
essential difference in their arrangement. 
In all the genera the receptacular cup is fused to the ovary wall right 
up to the base of the styles, and it is only at this level that the traces of the 
parts of the flower become separate from one another. There is, therefore, 
much more complete fusion than in most rosalian forms of inferior ovaries, 
and more resemblance to such advanced types as those of the Philadelpheae 
and Curcurbitaceae. In the ovary wall, therefore, we find only one series of 
bundles ; in Castanopsis they number twelve, and are large and fan-shaped 
and evidently compound. 
Six of these, occupying the angles and middle of the sides of the 
triangular ovary (Fig. 6, A , B , £ 7 , &c.), divide radially at the margin of 
the receptacle into three branches, which supply the perianth leaves, 
the outer whorl of six stamens, and the styles ; the six intermediate bundles 
(Fig. 6, a, b y c , &c.) divide into two branches only, these passing to the 
inner whorl of stamens and to the styles. 
In view of the fact that these intermediate bundles (a, b, c, & c.) originate 
in exactly the same manner as the whorl of six (A, B, C, &c.) from which the 
perianth bundles spring, it is not unreasonable to suppose that they primi- 
tively supplied a whorl of petals, and that the inner whorl of stamens is an 
antipetalous series, although we have no direct evidence of the previous 
existence of a corolla in these genera. 
In Castanea the number of bundles in the ovary wall is often increased 
by branches from the original twelve, while in Fagus a group of strands 
occurs in each angle of the nut and spreads out into the wing ; each seems 
to be equivalent to three or four wall-bundles with the component traces 
less closely associated together, and this brings the total number of these 
main bundles up to 18 or 20. 
