2 
162 J.D. Dana on Changes of Level in the Pacific Ocean. 
extensive barrier about the Gambier’s (see cut, vol. xiii, p. 130), 
that this subsidence, although less than farther north, was by no 
means of small amount. On page 371, vol. xi, we have estimated 
it at 1150 feet. These islands, therefore, although towards the 
limits of the subsiding area, were still far within it. The valley- 
bays of the Mangareva islets are of great depth, and afford addi- 
tional evidence of the subsidence. 
many parts, are three-fourths of a mile wide, but no inner chan- 
We have estimated the subsidence at one or two hundred 
would afford apparently the same evidence ; and it is quite poss 
ble that the facts indicate a sinking which either preceded the 
ejections that now cover the eastern and western extremities . 
Upolu, or accompanied this change of level. Savaii has smal 
reefs, from which we gather no certain facts bearing on this sub- 
ject. East of Tutuila is the coral island, Rose. It may bé 
therefore, that the greatest subsidence in the group was a oe 
eastern extremity. : | 
Feejee Islands.—We have already remarked upon this group: 
A large amount of subsidence is indicated by the reefs — 
portion of the group, but it was greatest beyond doubt in ' 
northeastern part. ' it 
Ladrones.—The Ladrones appear to have undergone thet 
greatest subsidence at the north extremity of the range, the part 
north have continued longest to burn, the islands are the sma % 
of the group, the whole having disappeared except the summits, 
which still eject cinders. The southern islands of the 
have wide reefs, but they afford no good evidence of any 8h 
extent of subsidence since the reefs began to form. 
