140 ON A NEW CHARACTER IN THE FRUIT OF QUERCUS. 
This last section comes very near Lithocarpus, Blume, in whieh’ the 
acorn is said to be joined to the involuere, which covers it entirely. 
Next comes Castanopsis, Spach, with the inflorescence and flower of 
of those Oaks which are included in Pasania and the following sections, 
as well as the echinate fruit of Castanea, from which it differs by its 3- 
celled ovary. Castanea, with its 6—7-celled ovaries, and Fagus, are too 
well known to be mentioned here. so 
T have not admitted the genus Synedrys, Lindl., founded on the pre- 
sence of incomplete partitions, which penetrate the spermoderm and 
cotyledons. This character, remarkably enough, exists in some Oaks 
(Q. Skinneri, from Mexico, Q. cornea, Lour, Q. Korthalsii, Blume; 
from the Indian Archipelago) which have nothing else in common, 
is not found in those species which are most closely allied ; besides, 
there are transitions in other species in the form of slight folds which 
scarcely penetrate, or as undulations of the cotyledons, and even in the 
species indicated the folds are irregular. C: 
Q. virens, Ait. (Q. oleoides, Cham. et Schl.), a species of extensive 
range in the south of North America, offers a very singular character, 
but I do not yet understand either its value or its constancy. In the four 
seeds I have examined, the radiele is buried in the homogeneous firm 
substance, which represents either two combined eotyledoiis or a single 
cylindrical cotyledon. Its central position towards the upper part of 
the fruit indicates rather two intimately combined cotyledons. 1 have 
seen nothing like it’ either in Q. lez, the most nearly allied, or any 
other species. Tt will be interesting to examine the development of 
this seed, as I have been unable to learn anything further from the con- 
dition of the herbarium specimens at my command. 
The greatest difficulty is how to divide Lepidobalanus, that natural 
section of the genus Quercus which alone contains more than half the 
species, some of which appear at first sight to differ much; for in- 
stance, Quercus Robur, Cerris, Vullonea, Libani, rubra, Xalapensis, 
etc. I wish I could have formed natural groups round these species 
which seem to have very marked characters, in other words, subsections 
analogous to the numerous ones into which Spach divides the subgenus 
Lepidobalanus, Endl. Webb, Endlicher, and especially M. Gay, have 
already attempted this, but I must say they have only reached acertam 
point, passing over a crowd of species from Mexico and southern or 
western Asia, which a few years ago were little known. M. Gay has 
