358 Of the Propagation of Heat 



ance. Could it be at the upper surface of it ? That ap- 

 peared to me to be impossible ; for, the experiment being 

 made in a room warmed by a German stove, the tem- 

 perature of the air which reposed on that surface was 

 considerably above the point at which water freezes. 



Could it be at its lower surface, where it rested on the 

 upper surface of the mercury ? If that should happen, 

 it would show that, notwithstanding the diminution of 

 the specific gravity of the water in passing from the tem- 

 perature of 41° to that of 32°, and the tendency which 

 this diminution gave it to quit the service of the mer- 

 cury from the instant v/hen, in being cooled by a contact 

 with it, it had passed the point of 41°, yet there was 

 time sufficient for the congelation to be completed before 

 the -particle of water so cooled could jnake its escape. 



The reader will naturally conclude from what was said 

 in the preceding page, that it was merely with a view to 

 the determination of that single fact that this experiment 

 was contrived ; and he will perceive by the result of it 

 that my expectations with regard to it were fully an- 

 swered. 



Ice was not only formed at the bottom of the water, at 

 its under surface, where it was in contact with the cold 

 mercury, but I found on repeating the experiment, and 

 varying it, by previously cooling the mercury in the 

 tumbler to about 10°, that boiling hot water, poured 

 gently upon it, was instantly frozen, and gradually 

 formed a thick cake of ice, covering the mercury; 

 though almost the whole of the mass of the unfrozen 

 water which rested on this ice remained nearly boiling 

 hot. 



This experiment not only determines the point for 

 the decision of which it was undertaken, but also enables 



