262 CORALS AND CORAL ISLANDS. 
In sustaining the theory, the fact of the subsidence re- 
quires proof, and secondly, its sufficiency for the result 
claimed. 
Darwin gives as evidence of the subsidence the near 
identity of barrier-girt islands and atolls. He compares 
the two, points out the fact that a slight change in the 
former by submergence is all that is required to convert 
it into an atoll, and enforces the argument by pointing to 
transitions between the two states. 
The facts from the Feejee Archipelago illustrate the sub- 
ject well. On the map, Plate XII., let the reader glance 
successively at the islands Goro, Angau, Nairai, Lakemba, 
Argo Reef, Exploring Isles, and Nanuku. It will be ob- 
served that in Goro the reef closely encircles the land upon 
whose submarine shores it was built up. In the island next 
mentioned, the reef has the same character, but 1s more dis- 
tant from the shores, forming what has been termed a bar- 
rier reef; the name implying a difference in position, but 
none in mode of formation. In the last of the islands 
enumerated, the barrier reef includes a large sea, and the 
island it encloses is but a rocky peak within this sea. 
If, now, the island Angau were sinking slowly, at a rate 
not more rapid than that of the upward growth of the reef, 
there would be a gradual disappearance of the land beneath 
the waters, while the reef might keep its level unchanged. 
Should the sinking go on until the land had mostly gone, 
the condition would be like that of the Exploring Isles, in 
which only a single ridge and a few isolated summits stand 
above the waters; and, at a stage beyond when only a single 
peak was left, the reef-girt island would have become a 
Nanuku. The subsidence of Goro, on the same principle, 
would produce an Angau. 

