286 Of the Propagation of Heat 



tion, in the form of a shallow basin, was formed in the 

 ice by this descending warm current. 



The warm water contained in this basin overflowed its 

 banks as soon as the basin began to be formed, and, issu- 

 ing out on that side which happened to be the lowest, 

 opened itself a passage under the tin plate to the edge 

 of the ice, over which it was precipitated and fell down 

 to the bottom of the jar. The water of this rivulet be- 

 ing warm, it soon formed for itself a dejsp channel in the 

 ice, and at the end of the experiment it was found to be 

 everywhere deeper than the bottom of the basin where it 

 took its rise. 



This manner of accounting for the appearances in 

 question seemed to me to be quite satisfactory ; and the 

 more I meditated on the subject, the more I was con- 

 firmed in my suspicions that all liquids must necessarily 

 be perfect non-conductors of Heat. 



On these principles I was now enabled to account for 

 the melting of the ice at the bottom of the hot water in 

 the experiment No. 16, as also for the slowness with 

 which that process went on ; and, encouraged by this 

 success, I now proceeded with confidence to plan and to 

 execute still more decisive experiments ; from the results 

 of which, I may venture to say it, the important facts in 

 question have been put beyond all possibility of doubt. 



If water be in fact a perfect non-conductor of Heat, 

 that is to say, if there be no communication whatever of 

 Heat between neighbouring particles or molecules of that 

 fluid (which is what I suppose), then, as Heat cannot be 

 propagated in it but only in consequence of the motions 

 occasioned in the fluid by the changes in the specific 

 gravity of those particles which are occasioned by the 

 changes of their temperature, it follows that Heat can- 



