Change pf State : Solid-Liquid. 43 



If we have p—p r , we have the melting-point ; and in this case, 

 by subtracting, we obtain 



PV 



— <-T- < 14 > 



Now the pressure required to lower the melting-point to the 

 same degree when both ice and water are compressed is given 



b y 



™-™= — y 5 (15) 



or P' __ v f — v 



= •087 (16) 



Or the fall in melting-point caused by a given pressure on the 

 ice alone is about 11-J times as great as when both ice and 

 water are compressed. That is, 1 atmosphere lowers the 

 melting-point about *0843° C, and 11'7 atmospheres lower it 

 1° C. This result may be obtained in the same way as Prof. 

 J. Thomson's formula, on the supposition that the process is 

 reversible ; but as I was led to the result by the above consi- 

 derations, I have given only this proof. 



This seems to have an important bearing on ordinary cases 

 of regelation, when two pieces of ice are brought into con- 

 tact at one or two points. About that point the ice will 

 be subjected to great pressure ; but the melted water is 

 not necessarily subject to the pressure, and accordingly the 

 melting-point may be lowered by 11^ times as much as has 

 been formerly supposed. I have made some experiments to 

 test this result; and the best arrangement I have yet devised 

 has been the following: — A block of ice, fitting in a hollow 

 iron cylinder with open ends, was laid on a bed of sand on 

 the top of which was placed one junction of a copper-iron 

 thermopile ; the other junction was placed in melting ice. 

 When the two junctions reached the same temperature, as 

 indicated by a galvanometer in the thermopile circuit, pressure 

 was applied to the ice by a hydraulic press. The water from 

 the melting of the ice was able to escape freely through the 

 sand, and was therefore only at atmospheric pressure. The 

 results so far have been very variable, sometimes indicating- 

 no greater lowering of the melting-point than that usually 

 assumed — "0073° per atmosphere. But in several cases the 

 lowering has been decidedly greater. The following experi- 

 ment gives the greatest value I have } ,- et obtained for the low- 

 ering of the melting-point. The galvanometer-deflection per 



