CORAL-REEFS. 141 



separated into distinct portions ; and these portions would 

 tend to assume an atoll-like structure, from the coral growing 

 with vigour round their entire circumferences, when freely 

 exposed to an open sea. As we have some large islands 

 partly submerged with barrier-reefs marking their former 

 limits, such as New Caledonia, so our theory makes it 

 probable that there should be other large islands wholly 

 submerged; and these, we may now infer, would be sur- 

 mounted, not by one enormous atoll, but by several large 

 elongated ones, like the atolls in the Maldiva group ; 

 and these again, during long periods of subsidence, would 

 sometimes become dissevered into smaller atolls. I may 

 add, that both in the Marshall and Caroline Archipel- 

 agoes, there are atolls standing close together, which have 

 an evident relationship in form : we may suppose, in such 

 cases, either that two or more encircled islands originally 

 stood close together, and afforded bases for two or more 

 atolls, or that one atoll has been dissevered. From the 

 position, as well as form, of three atolls in the Caroline 

 Archipelago (the Namourrek and Elato group), which 

 are placed in an irregular circle, I am strongly tempted 

 to believe that they have originated by the process of 

 disseverment. 1 



Irregularly-formed atolls. — In the Marshall group, Mus- 

 quillo atoll consists of two loops united in one point ; and 



1 The same remark is, perhaps, applicable to the islands of Ollap, 

 Fanadik, and Tamatam in the Caroline Archipelago, of which charts 

 are given in the atlas of Duperrey's voyage : a line drawn through the 

 linear reefs and lagoons of these three islands forms a semicircle. 

 Consult also, the atlas of Lutke's voyage ; and for the Marshall group 

 that of Kotzebue ; for the Gilbert group consult the atlas of Duperrey's 

 voyage. Most of the points here referred to may, however, be seen 

 in Krusenstern's general Atlas of the Pacific. 



