STRUCTURE OF CORAL ISLANDS. 195 
of this character, speaks of the rock as compact, and having 
the fracture of a secondary limestone. 
The surface of Metia is singularly rough, owing to ero- 
sion by the rains. The paths that cross it wind through nar- 
row passages among ragged needles and ridges of rock as high 
as the head, the peaks and narrow defiles forming a miniature 
model of the grandest Alpine scenery. ‘There is but little 
soil, yet the island is covered with trees and shrubbery. 
The shores at the first elevation of the island, must have 
been worn away to a large extent by the sea; and the cliff 
and some isolated pinnacles of coral rock still standing on the 
coast are evidence of the degradation. But at present there 
is a wide shore-platform of coral reef, two hundred or two 
hundred and fifty feet wide, resembling that of the low coral 
islands, and having growing coral, as usual, about its margin 
and in the shallow depths beyond. 
In the face of the cliff there are two horizontal lines, 
along which cavities or caverns are most frequent, which con- 
sequently give an appearance of stratification to the rock, 
dividing it into three nearly equal layers. 
We might continue this account of coral reefs and islands, 
by particular descriptions of others in the Pacific. But the 
similarity among them is so great, and their peculiarities are 
already so fully detailed, that this would amount only to a 
succession of repetitions. The characters of a few, briefly 
stated, will suffice in this place. The first eight mentioned 
beyond ‘are small islands situated within ten degrees of the 
equator; Birnie’s, Enderbury’s, and Hull’s belonging to the 
Pheenix Group, and Oatafu and Fakaafo to the Union Group. 
Their positions are shown on Plate VIILf. The remainder are “7Y~ e 
9° 
a few islands of the Paumotu Archipelago, in latitudes 12 
