CYRTANDREÆ (AUCT. C. B. CLARKE). 5 
The genera also are very continuous in their areas of distribution. 
and do not comprise outlying species. The most striking exception is 
that a species of the South-east Asian genus Epithema has been col- 
lected in West Tropical Africa by Mann. There has also been sent 
lately from Madagascar by the Rev? R. B. Baron a plant herein 
referred to Didymocarpus, but it forms a section by itself. The 
European genera are not found out of Europe nor the African out of 
Africa; Coronanthera has its 9 species confined to New Caledonia: 
while the numerous species of Polynesia all belong to one genus, 
Cyrtandra. 
Each group of Polynesian Islands has its own set of species, 
whieh do not extend to any neighbouring group : the Sandwich Isles 
have supplied 32 species, all endemic; Tahiti 10 species, Samoa 
10 species, Viti 20 species, all endemic. Moreover, in one group of 
islands, each isle generally possesses its own species; in the Sand- 
wich group, 16 species are confined to Oahu, 4 to Hawaï, 7 to 
Kauai, while only 2 species are as yet known to occur on more than 
one island; and, in these 2 cases, the species occur in different 
islands as different varieties. 
That the vegetation of oceanic islands should be of an endemic 
character is however a fact illustrated by many natural Orders 
besides the Tribe Cyrtandreæ. The extremely localised character of 
the distribution of the Cyrtandreæ is more strikingly shewn by the 
manner in which closely allied species of Didymocarpus are confined 
to single provinces of the Himalaya, although there appears no rea- 
son, either in soil or climate, why they should not spread to the 
adjacent valleys. We find Didymocarpus pedicellata confined to 
Gurwhal and Kumaon; D. villosa to Nepaul: D. Andersoni, D. auran- 
tiaca, D. albicalyx, D. Mortoni to Sikkim; D. Punduana, D. Hoo- 
keri, D. acuminata to the Khasi Hills. In these cases the ground has 
been examined sufficiently to render it very improbable that the 
area of the species will be greatly extended by future collectors. They 
all belong to the same section of Didymocarpus, and are so closely 
allied that several of them might be regarded as geographic subspe- 
