106 WRIGHT: THE GENUS DIOSPYROS 
leave the question of affinity until detailed information for 
the remaining species is at hand. 
The fossil species described by TJnger, Ettingshausen, 
Massalongo, and others, the present distribution through 
tropical regions of the eastern and western hemispheres, in 
sub-tropical South Africa and Madagascar, and the occurrence 
in temperate regions are suggestive of great age. The fossil 
forms, the conspicuous differences between existing species, 
especially between those in Ceylon, together with the pro¬ 
bable early extinction by natural causes of many Ceylon 
species, suggests the likelihood that the Ebenaceae had in the 
past a much wider distribution and a larger number of 
species than at the present day. 
The great point of interest in connection with the Ceylon 
species is the fact that a good artificial key for working out 
the species can be constructed on separate characters, e.g., the 
leaf, flower, or seedlings. The differences in leaf characters 
are very marked. It is impossible to confuse the tough 
large leaf of D. Moonii or D. acuta with the thin leaves of 
D. Thwaitesii andD. montana, or with the thin hairy leaves 
of D. pruriens. The pellucid character of the veins in 
leaves of D. Ebenum, D. Embryopteris, D. Toposia, and D. 
Gardneri separates these species from one another and from 
all other species, and questions of size and form provide 
sufficient differences for separating the remaining species. 
Then again we have seen that it is very convenient to group 
the species according to their sex characters* 
spresentatives can be 
one, the classification 
r cotyledon, abortion 
or persistence of the median cotyledonary traces, characters 
of cotyledons, epicotyledonary leaves, and hypocotyl. 
The secondary xylem is the most constant structure 
throughout the genus, and in the respective species shows 
differences only in percentage number of the elements. 
* Sex in Dioepyroe. Brit. Assoc., Belfast, 1902. 
be continued in next number. 
A good artificial key for the Ceylon r< 
being based on the number of traces pei 
