in Fluids. _ 297 



Having found, as well from the results of the experi- 

 ments made with cold water as from those made with hot 

 water, that a considerable quantity of ice was melted in 

 the act of pouring the water into the jar, and in conse- 

 quence of those undulatory motions into which the wa- 

 ter was thrown in that operation, notwithstanding all the 

 pains I had taken to diminish those motions and prevent 

 their effects, I now doubled my precautions in guarding 

 against those sources of error and uncertainty. 



Before I poured the water into the jar I covered the 

 surface of the ice to the height of 0.956 of an inch with 

 ice-cold water, and this I did when water at the temperature 

 of 41° was used, as well as in those experiments in which 

 ]3oiling-hot water was employed. In the former experi- 

 ments I had covered the surface of the ice with ice-cold 

 water only in those experiments in which hot water was 

 used, and even in those I used only half as much ice- 

 cold water as I now employed for that purpose. 



I also now poured the water into the jar in a smaller 

 stream, employing no less than three minutes in filling 

 it up to the height of eight inches above the surface of 

 the ice ; and I endeavoured to ascertain how far the re- 

 sults of the experiments were influenced by the temper- 

 ature of the air, and also by wrapping up the jar in warm 

 covering. 



The same jar was used in all the experiments, and it 

 was always placed in the same earthen dish, and sur- 

 rounded, to the level of the top of the ice, with melting 

 snow. This jar is very regular in its form, being very 

 nearly a perfect cylinder, and is on that account pecu- 

 liarly well calculated for the use for which I selected it. 



In each of the three first experiments which are en- 

 tered in the following table, the jar was well covered up 



