292 H. B. GUPPY, M.B., 
originally came direct from the Indian Archipelago: for in 
such cases it is necessary-that we should assume that the 
nearest land is the starting-point. Again, the Krakatoa 
pumice arrived at the Cocos Islands in great quantity a few 
weeks after the eruption in the Sunda Straits, a circumstance 
showing that it must have pursued a course fairly direct 
from its source. 
Hence we must conclude that most of the vegetable drift 
takes a course fairly direct to the Cocos Islands, whether from 
the Indian Archipelago or from the western shores of 
Australia. My experiments and those of Dr. Treub make it 
quite clear that several of the seeds of the trees already 
established on these islands can germinate after floating 
6 or 7 weeks in sea-water. During this period they might 
have been transported by a surface-current running only one 
knot an hour a distance of from 1,000 to 1,200 miles, which 
is all that we require to establish the possibility of their 
germinating after a passage from the islands of the Indian 
Archipelago or from the nearest coasts of Australia. 
Java and Sumatra, with the islands adjacent to them, 
probably supply a large proportion of the fruits and seeds 
stranded on the eastern shores of this small group. Accord- 
ing to Mr. Keating, as shea by Mr. Darwin, many of the 
familiar woods, seeds, and fruits of the region of the Indian 
Archipelago arrive at the Cocos Islands. A canoe has 
been thence drifted, and I may add that Java water-bottles, 
made of bamboo, and the large bamboo fishing-stakes em- 
ployed in that part of the archipelago, are now frequently 
beached on the Cocos Islands. However, I must leave to 
those who determine my collections the task of ascertaining 
more definite facts concerning the source of the vegetable 
drift. I may, however, add that, as shown in the note ap- 
pended to this paper nearly all the Cocos Island plants are to 
be found on the adjacent south coast of Java. 
The part taken by sea-fowl and migratory birds in stocking 
these islands with plants has yet to be investigated. It is 
obvious, however, that birds are less certain agents in the 
process. The frigate-bird, for instance, when it takes its 
flight over the inland regions of Java and returns to Christmas 
Island and to North Keeling Island, where it nests in great 
numbers, must sometimes aid in the distribution of plants in 
the manner [ have already instanced in the case of Cesalpinia 
' Bonducella. The species of Pisonia distributed in the Cocos 
Islands is evidently distributed in the fashion before remarked. 
I have also pointed out that the seeds of Pemphis acidula and 
