CHAP. VIII.] THE GULF-STREAM. 371 



local accidents as those which produce the Gulf- 

 stream proper, and which gives movement to a much 

 larger and deeper hody of water than the latter can 

 affect. The evidence of such an interchange is two- 

 fold — that of physical theory, and that of actual 

 observation. Such a movement must take place, as 

 was long since pointed out by Professor Buflf, when- 

 ever an extended body of water is heated at one part 

 and cooled at another ; it is made use of in the warm- 

 ing of buildings by the hot-water apparatus, and it 

 was admirably displayed at the Royal Institution a 

 few months since in the following experiment kindly 

 prepared for me by Dr. Odling." Dr. Carpenter 

 then repeats Professor Buff's convection experiment, 

 the heat being applied by a steam jet introduced 

 vertically at one end of a narrow glass trough while 

 a block of ice was wedged into the other end. 

 "Thus a circulation was shown to be maintained 

 in the trough by the application of heat at one of its 

 extremities and of cold at the other, the heated water 

 flowing along the surface from the warm to the cold 

 end, and the cooled water flowing along the bottom 

 from the cold to the warm end ; just as it has been 

 maintained that equatorial water streams on the 

 surface towards the poles, and that polar water 

 returns along the bottom towards the equator, if 

 the movement be not interfered with by interposed 

 obstacles, or prevented by antagonistic currents 

 arising from local peculiarities." ^ 



That such a movement cannot take place on this 

 hypothesis has been already shown ; and Dr. Car- 



' The Gulf-stream. A letter from Dr. Carpenter to the Editor of 

 JVfriwf, dated Gibraltar, August 1 1th, 1870. {Nature, mo\. ii. p. 334.) 



BB 2 



