APPENDIX. 237 



Amirantes, certainly are not atoll-formed, and they differ 

 as a group from every other group with which I am 

 acquainted ; I have not coloured them ; but probably the 

 reefs belong to the fringing class. Their formation is 

 attributed, both by Dr. Allan and Capt. F. Moresby, to the 

 action of the currents, here exceedingly violent, on banks, 

 which no doubt have had an independent geological origin. 

 They resemble in many respects some islands and banks 

 in the West Indies, which owe their origin to a similar 

 agency, in conjunction with an elevation of the entire area. 

 In close vicinity to the several islands, there are three 

 others of an apparently different nature : first, Juan de 

 JVova, which appears from some plans and accounts to be 

 an atoll ; but from others does not appear to be so ; not 

 coloured. Secondly, Cosmoledo ; "this group consists of a 

 ring of coral, ten leagues in circumference, and a quarter 

 of a mile broad in some places, enclosing a magnificent 

 lagoon, into which there did not appear a single opening " 

 (Horsburgh, vol. i. p. 151); coloured blue. Thirdly, 

 Aldabra; it consists of three islets, about 25 feet in height, 

 with red cliffs (Horsburgh, vol. i. p. 176) surrounding a very 

 shallow basin or lagoon. The sea is profoundly deep close 

 to the shore. Viewing thi? island in a chart, it would be 

 thought an atoll ; but the foregoing description shows that 

 there is something different in its nature; Dr. Allan also 

 states that it is cavernous, and that the coral-rock has a 

 vitrified appearance. Is it an upheaved atoll, or the crater 

 of a volcano ? — uncoloured. 



Comoro Group. — Mayotta, according to Horsburgh 

 (vol. i. p. 216, 4th edit), is completely surrounded by a 

 reef, which runs at the distance of three, four, and in some 

 places even five miles from the land; in an old chart, 

 published by Dalrymple, a depth in many places of 36 and 



