Fagaceae , and its Bearing on the A jfinities of the Group. 5 1 1 
angiospermous forms possessing bisexual flowers suggest affinity with any 
modern group of Archichlamydeae, but seem, rather, to incline to Coulter’s 
( 6 ) conclusion that, ‘ whether they represent a single genetic stock or several, 
they appear to be isolated from the higher alliances.’ 
Hallier has attempted to demonstrate an affinity between the Cupuli- 
ferae and one or other of the higher groups of Dicotyledons. In his 
earlier scheme for the classification of the Angiosperms (12), he derived the 
Cupuliferae from the Hamamelidaceae, and through this order from the 
Laurineae, thus connecting them indirectly with rosalian forms allied to 
the Combretaceae ; in later papers, however, he discarded this view in 
favour of the theory that they were descended from the Anacardiaceae and 
Burseraceae, chiefly on account of the strong anatomical likeness of Juliania 
to these orders on the one hand, and to Juglans on the other (13). 
But this view, based on the characteristics of some of the most reduced 
members of long reduction series, does not seem to take sufficiently into 
account the more complex members of the group, such as the Fagaceae, 
and the facts brought forward in the following pages seem to a certain 
extent to confirm the earlier theory, that the ancestors of the Cupuliferae 
were nearly allied to the Rosaceae . 1 
Morphological Characters of Castanopsis chrysophylla. 
Castanopsis chrysophylla , the Golden-leaved Chestnut, a Californian 
shrub sometimes grown in English gardens, was selected for special study, 
as being one of the less well-known members of the Fagaceae. The fact, 
moreover, that it possesses flowers which, though functionally female, are 
structurally bisexual, with an androecium of twelve stamens constantly 
present, seemed to promise that its investigation would throw light on the 
organization of the more reduced forms in the group. 
The genus is very closely allied to Castanea, the chief points in which 
it differs from the latter being the trilocular form of the ovary, and the 
irregularity of the cupule ; the latter is sometimes absent, and sometimes 
encloses, according to Prantl (20), only one flower. This writer regards the 
genus Castanea as a branch of the large tropical family Castanopsis , which 
has become adapted to northern temperate climates. 
The flowering branches of Castanopsis chrysophylla bear numerous 
catkins ; the lower ones, growing in the axils of the leaves, are entirely 
staminate, and though closely resembling those of Castanea vulgaris , are 
shorter and smaller. Only two or three small catkins in the axils of bracts 
1 In a more recent paper, L’Origine et le Systeme phyletique des Angiospermes (Archives 
Neerlandaises des Sci. Nat., s6r. 3 B., t. i, 1912, p. 146), Hallier, while still adhering to his later view, 
admits the possibility of the derivation of the Amentaceae from the Hamamelidaceae or Rosaceae, 
particularly in view of certain likenesses in the anatomical structure of the wood. 
