Deductions from the Curve. 39 



depths varying from 162 to 845 fathoms. From this it has 

 been inferred, that in a basin separated from t/ie adjoining ocean 

 by a submarine elevation, the temperature of the water decreases 

 from the surface down to the level of that elevation, and remains 

 stationary from that level down to the bottom. 



The subsequent proceedings of H.M.S. " Challenger," in the 

 Western Pacific and in the seas of the Indian Archipelago 

 (Plates 15 and 16), leave little room for doubting the correct- 

 ness of this conclusion. In the basin extending from the New 

 Hebrides to Torres Strait (which might appropriately be called 

 the Melanesian Sea), in the Banda Sea, Celebes, Sulu, and China 

 Sea, the temperature of the water was found to decrease from 

 the surface down to a certain level, and remain stationary 

 between that level and the bottom. The Sulu Sea furnishes 

 the most striking illustration of this hitherto unsuspected 

 phenomenon. At the time of the two visits of H.M.S. 

 " Challenger," in October, 1874, and January, 1875, tne tem ~ 

 perature was observed to fall from 2 8° and 27 C. at the 

 surface to io°.2 C. at a depth of 400 fathoms, and to remain 

 stationary at the latter temperature from 400 fathoms down to 

 the bottom, at depths of 2550 and 2225 fathoms, forming a 

 stratum of more than two miles in depth of the comparatively 

 high temperature of io°.2 C, or 50°.4 F. Although we possess 

 no complete surveys of the bottom of the seas above mentioned, 

 a glance at a chart will show that they all are more or less 

 land-locked basins, and, as^ the temperature-conditions seem to 

 prove, probably cut off from the colder strata of the Pacific and 

 Indian Oceans by submarine ridges rising to the level at which 

 the decrease of temperature is arrested. 



The second exception, observed both in the South Polar 

 and North Polar regions, may require further research before it 

 can rank as an established scientific fact. Besides the numerous 

 difficulties which beset thermometric observations uuder the 



