CHANGES OF LEVEL. 395 
Samoa, and the Salomon Islands to the Pelews, it will form nearly a 
straight boundary trending N.70° W., between the atolls and the high 
islands of the Pacific, the former lying to the north of the line, and 
the latter to the south. 
Between this boundary line and the Hawaiian Islands, an area 
nearly two thousand miles wide and six thousand long, there are two 
hundred and four islands, of which only three are high exclusive of the 
eight Marquesas. These three are Ualan, Banabe (Ascension or 
Pounypet) and Hogoleu, all in the Caroline Archipelago. South of 
the same line, within three degrees of it, there is an occasional atoll ; 
but beyond this distance, there are none excepting the few in the 
Friendly Group, and one or two in the Feejees. The colouring of the 
map indicates their positions. 
If each coral island scattered over this wide area indicates a subsi- 
dence of an island, we may believe that the subsidence was general 
throughout the area. Moreover, each atoll, could we measure the 
thickness of the coral constituting it, would inform us nearly of the 
extent of the subsidence where it stands; for they are actually so 
many registers placed over the ocean, marking out not only the site 
of a buried island, but also the depth at which it lies covered. We 
have not the means of applying the evidence; but there are facts at 
hand, which may give at least comparative results. 
a. We observe, first, that barrier reefs are in general evidence of 
less subsidence than atoll reefs, (p. 130.) Consequently, the great 
preponderance of the former just below the southern boundary line of 
the coral island area, and farther south the entire absence of atolls 
while atolls prevail so universally north of this line, are evidence of 
little depression just below the line; of less farther south ; and of the 
greatest amount, north of the line or over the coral area. 
b. The subsidence producing an atoll, when continued, gradually 
reduces its size, and finally it becomes so small that the lagoon is obli- 
terated ; and consequently a prevalence of these small islands is pre- 
sumptive evidence of the greater subsidence. We observe, in appli- 
cation of this principle, that the coral islands about the equator, 
five or ten degrees south, between the Paumotus and the Tarawan 
Islands, are the smallest of the ocean: several of them are with- 
out lagoons, and some not a mile in diameter. At the same time, 
in the Paumotus, and among the Tarawan and Marshall Islands, 
there are atolls twenty to fifty miles in length, and rarely one less 
than three miles. It is probable, therefore, that the subsidence indi- 
