at its Maximum Density. 267 



As I had found by some of my experiments made in 

 the year 1797 (of which an account is given in my 

 seventh Essay, Part I.) that water at the temperature of 

 about 42 F., and .consequently what we should call very 

 cold, melted considerably more ice, when standing on it, 

 than an equal quantity of boiling-hot water in the same 

 situation, I was very curious to see whether the ther- 

 mometer, the bulb of which lay in the cork cup, would 

 not also be less heated by the ball when it should be 

 applied very hot to the surface of the water, than when 

 its temperature was much lower. 



Seeing that this research ought to throw great light on 

 the mysterious operations of the distribution of heat 

 in liquids, I hastened to make the following experiment. 



Experiment No. 3. The cylindrical vessel with its 

 contents having been once more reduced to the uniform 

 temperature of freezing water, the metallic ball was heated 

 in boiling water, and being as expeditiously as possible 

 taken out of that hot liquid, its projecting conical point 

 was suddenly submerged in the ice-cold water, as in the 

 former experiments. 



The result of this experiment was very interesting. 

 It was not till 50 seconds had elapsed that the ther- 

 mometer began to show any signs of rising, and at the 

 end of i minute and 7 seconds it had risen only 2 

 degrees. 



In the foregoing experiment, when the metallic ball 

 was so much colder, the thermometer began to rise in 

 10 seconds, and at the end of I minute and 3 seconds it 

 had risen 5 degrees. 



This difference is very remarkable, and if it does not 

 prove the existence and great efficacy of currents in con- 

 veying heat in fluids, I must confess that I do not see 



