736 



ANNUAL REPORT SMITHSONIAN INSTITUTION, 1912. 



The explanation of this effect which I gave at my Friday evening discourse at the 

 Royal Institution last May was founded on Pettersson's theory of ice melting in salt 

 water. By this theory, which can easily be verified by a simple experiment, ice 

 melting in salt water produces three currents: (1) A current of salt water cooled by 

 the ice which sinks downward by gravity, (2) a current of warm salt water flowing 

 toward the ice, and (3) a current of light fresh water from the ice rising and spreading 

 out over the surface of the salt water. 



I at fii-st thought that it was this surface current of fresh water that influenced the 

 microthermometer. The fringe of this lighter water would be warmer than the sea 



/0am //am /z. hopn. / pm Z-pm 3pm A-pm Spm 



Fig. 10.— Miceo-thermogkau taken from the Strait of Belle Isle showing effect of iceberg 



MEETI.VG. 



water on account of the action of the sun and scattered radiation, which is very strong 

 at sea. The lighter water would retain the heat because it could not mix readily with 

 the sea water. Near the iceberg I considered that a fall of temperature would result 

 from the cooling influence of the surface current of fresher water. 



My recent tests have shown, however, that an iceberg melts so slowly that little 

 effect of the dilution ran be detected, even right beside the berg. I took a number of 

 samples of sea water at different distances from icebergs as well as samples taken far from 

 ice. These samples I carefully bottled and brought home to the laboratory, where 

 they were most accurately tested by the electric conductivity method in the physico- 



