40 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY. 
island with a ridge having a height of about 1,100 feet, a western island 
with a height of about 800 feet, and a northwestern island with a sum- 
mit of the same height. These islands might thus be reduced to three 
separate ridges, giving no indication that they had formed part of the 
rim of the crater of an extinct volcano. 
The denudation and erosion could be carried still further, leaving only 
islets, the summits of the higher peaks, to indicate the former position 
of the rim, the islets being joined by coral patches connecting their 
extremities, much as the present opening between the horns of the rim 
of the crater is closed by the outer reef. We may still further imagine 
it to be so far cut down as to form reef flats upon which coral would 
grow, thus forming a nearly circular atoll with a depth of 35 fathoms, 
— an atoll with the formation of which subsidence has had nothing to 
do. But this is not an imaginary atoll I am reconstructing. A number 
of such atolls are found in Fiji, the formation of which can be satisfactorily 
explained on the theory that the ring of coral patches represents the 
rim of an extinct volcano which has been cut away to below low water 
mark. Such atolls in the Fijis are probably Thakau Momo, Thakau 
Lasemarawa, Thakau Lekaleka, Motua Levu, Motua lai lai, Pitman 
and Williamson Reefs, and perhaps others. 
The example of Thombia, one of the Ringgold Islands, in which there 
is only a distant outer reef, would also indicate the possibility of the rim 
of the crater of a small volcanic peak cut down to the surface and forming 
the circular flats upon which corals might grow. In the case of Thom- 
bia such a condition would result in forming a diminutive atoll not more 
than a third of a mile in diameter, enclosed within an encircling barrier 
reef. 
We might also consider the “ Boilers,” the diminutive ‘ Serpuline 
atolls” inside of the lagoon of the outer reef off the south shore of the 
main island of the Bermudas,! as a series of such interior atolls, though 
the mode of origin is very different from that of subordinate atolls, 
formed, as I have suggested, upon the rim of an extinct crater like 
Thombia. In either case, the explanation of the formation of such 
interior or subordinate atolls is radically different from that given by 
Darwin? for their growth in the Maldive atolls, an explanation also 
accepted by Dana. 
It is becoming more and more apparent that each locality must be 
1 Bull. Mus, Comp. Zool., XXVI. No. 2, 1895, Plates XXII.-XXVL., p. 253 
2 Darwin’s Coral Reefs, 3d ed., p. 44. 
