^%^!'' THE EXPLORATION OF CORAL REEFS. 51 



To take one case : at the east end of the Paumotu Archipelago 

 the coral islands are well-wooded, have broad reef grounds, and their 

 lagoons are filled up ; that is to say, they possess the characters of 

 stationary reefs. Going westward, the islands are seen to decrease 

 in size and elevation ; the lagoons persist, and the islands become 

 less and less wooded. At Bellingshausen Atoll the last stage is seen, 

 and beyond this there is nothing but open sea for nearly goo miles ; 

 then some low banks of tide-washed coral commence the Samoan 

 Archipelago, in which the order is reversed, till it culminates in the 

 great wooded coral island of Savaii. The inference is that the central 

 open sea represents an area where subsidence was too rapid for the 

 reefs to keep pace with it, and so they have been submerged, while 

 the gradual increase of the reefs on either side marks the decrease in 

 the rate of subsidence. If this deduction be correct, then, by the 

 application of this method, one can map out the whole of the Central 

 Pacific into subsiding or stationary districts. Broadly speaking, 

 there are two main lines across the Pacific, along which the land is 

 either rising or is at least stationary. The first runs as a continuation 

 of the longer axis of New Zealand, north-eastward through the 

 Kermadec and Tonga Islands, to the volcanoes of the Samoans, and 

 thence past the Phoenix group and the Kingman Shoals to the Sand- 

 wich Islands in the extreme north-east. Along the whole of this line 

 there occurs a chain of volcanoes, a submarine ridge, occasional 

 shoals, and islands with raised reefs, and some plutonic rocks. The 

 second line starts from the coast of South America, as the submerged 

 plateau that extends for a great distance north-west of Patagonia, and 

 runs thence through the Paumotu group, crossing the other line 

 almost at a right angle at the Samoans, and thence it continues west- 

 ward as the main chain of Malaysia, Parallel to these two lines, and 

 on both sides of them, are areas of deep and open sea, which certainly 

 seem to represent areas of subsidence. It is in one of these that the 

 greatest expanse of radiolarian deposits occurs, stretching as a great 

 elliptic sheet trending north-east and south-west ; where this crosses 

 the second line it is completely cut across by a band of a less abyssal 

 globigerina ooze, as is well shown in the beautiful map of the distribu- 

 tion of the deep-sea deposits which accompanies the last volume of 

 the " Challenger " Reports. 



It is manifest that any boring on one of the stationary or rising 

 lines would be quite valueless, as shallow reefs might occur anywhere 

 along them. The islands that are most likely to yield a conclusive 

 result are the low atolls on the very margins of the great coral archi- 

 pelagos, such as Enderbury Island at the northern end of the Phoenix, 

 Rose Atoll at the eastern end of the Samoans, or Bellingshausen 

 Island at the western end of the Society group. But these atolls 

 are so very inaccessible that some more convenient locality would 

 naturally be preferred. Diego Garcia, for example, or any of those 

 in the Indian Ocean, would be much more easily reached. But it is 



D 2 



