> 
CHAP. VIII. ] _ THE GULF-STREAM. 40] 
expected if the Gulf-stream came close to the western 
shore. 
While the communication between the North 
Atlantic, and the Arctic Sea—itself a second cul 
de sac—is thus restricted, limiting the interchange of 
warm and cold water in the normal direction of the 
flow of the Gulf-stream, and causing the diversion of 
a large part of the stream to the southwards, the 
communication with the Antarctic basin is as open as 
the day ;—a continuous and wide valley upwards of 
2,000 fathoms in depth stretching northwards along 
the western coasts of Africa and Europe. 
That the southern water wells up into this valley 
there could be little doubt from the form of the 
eround; but here again we have curious corroborative 
evidence on the map in the remarkable reversal of the 
curves of the isotherms. The temperature of the bot- 
tom water at 1,230 fathoms off Rockall is 3°22 C., 
exactly the same as that of water at the same depth in 
the serial sounding, lat. 47° 38’ N., long. 12° 08° W. in 
the Bay of Biscay, which affords a strong presumption 
that the water in both cases is derived from the same 
source; and the bottom water off Rockall is warmer 
than the bottom water in the Bay of Biscay (2°5 C.), 
while a cordon of temperature soundings drawn from 
the north-west of Scotland to a point on the Iceland 
shallow gives no temperature lower than 6°5 C. This 
makes it very improbable that the low temperature 
of the Bay of Biscay is due to any considerable por- 
tion of the Spitzbergen current passing down the west 
coast of Scotland; and as the cold current to the 
east of Iceland passes southwards considerably to the 
westward, as indicated on the map by the successive 
DD 
