269 



for the mean value of a division. From a rough calibration of the 

 tube which was made, I am convinced that the values of the divisions 

 at no part of the tube differ by more than gV'-h of this amount, 

 from the true mean value ; and, taking into account all the sources 

 of uncertainty, I think it probable that each of the divisions on the 

 tube of the ether thermometer corresponds to something between ^V 

 and t'j of a degree Fahrenheit. 



With this thermometer in its glass envelope, and with a strong 

 glass cylinder (CErsted's apparatus for the compression of water), 

 an experiment was made in the following manner : — 



The compression vessel was partly filled with pieces of clean ice, 

 and water : a glass tube about a foot long and yV^'h of an inch in- 

 ternal diameter, closed at one end, was inserted with its open end 

 downwards, to indicate the fluid pressure by the compression of the 

 air which it contained : and the ether thermometer was let down and 

 allowed to rest with the lower end of its glass envelope pressing on 

 the bottom of the vessel. A lead ring was let down so as to keep 

 free from ice the water in the compression cylinder round that 

 part of the thermometer tube where readings were expected. More 

 ice was added above, so that both above and below the clear space, 

 which was only about two inches deep, the compression cylinder 

 was full of pieces of ice. Water was then poured in by a tube with 

 a stopcock fitted in the neck of the vessel, till the vessel was full up 

 to the piston, after which the stopcock was shut. 



After it was observed that the column of ether in the thermo- 

 meter stood at about 67°, with reference to the"divisionson the tube, 

 a pressure of from 12 to 15 atmospheres was applied, by forcing 

 the piston down with the screw. Immediately the column of ether 

 descended very rapidly, and in a very few minutes it was below 61^. 

 The pressure was then suddenly removed, and immediately the 

 column in the thermometer began to rise rapidly. Several times 

 pressure was again suddenly applied, and again suddenly removed, 

 and the effects upon the thermometer were most marked. 



The fact that the freezing-point of water is sensibly lowered by a 

 few atmospheres of pressure, was thus established beyond all doubt. 

 After that, I attempted, in a more deliberate experiment, to deter- 

 mine as accurately as my means of observation allowed me to do, 

 the actual extent to which the temperature of freezing is affected by 

 determinate applications of pressure. 



