160 
WILLIS AND GARDINER ; BOTANY 
quite different. Few individuals arrive, and probably only 
at long intervals of time apart, and there is no special Reason 
why further arrivals should necessarily be others of the same 
species. Finding a large vacant area, these forms may 
spread in great abundance over the interior of the island, 
and ultimately give rise to endemic species, whose chance 
of further distribution is small, and depends on the nearness 
of other islands. It has been elsewhere shown that wind 
distribution tends to carry more species, but fewer indivi- 
duals than animal distribution, and we may therefore expect 
perhaps to find more wind-carried species than bird-carried 
giving rise to endemics. Treub, in his work on the new flora 
of Krakatoa,^ has shown that we may expect to find many 
ferns in oceanic islands, due to their carriage by wind, and 
supported by the actual state of the floras of Juan Fernandez, 
Ascension, &c. We shall expect to find ferns and other 
cryptogams specially well represented, but may also look to 
find Compositæ (wind), Rubiaceæ (birds). Palms (birds), 
small seeded plants (birds’ feet), and others carried in 
similar ways. Examination of the actual floras of oceanic 
islands will show that a large number of their endemic 
species belong to orders in which means of carriage are well 
marked, but, on the other hand, there are many in which it 
is difficult to assume such ancestry. Thus, in the Admiralty 
Islands the few endemic species belong to the genera 
Medinilla, Hydnophytum, Dendrobium, Cyathea, Hymeno- 
phyllum. Polypodium, Acrostichum. In Diego Garcia 
occurs an endemic Asplénium, the only endemic species in 
the Laccadive-Chagos chain among the higher plants, with 
which alone we are at present dealing. In Rodriguez we 
have a small island isolated by a considerable distance from 
its nearest neighbours, the Seychelles and the Chagos. In 
former times it may have formed part of a large island 
lying east of the Seychelles-Mascarene Island, indicated on 
the map by the 2,000-fathom line, but any land connection 
with that region must have been at a vast distance of time, 
* Aim. Buitenz., VII. 
