EMPIRICAL KNOWLEDGE, EXPLAXATIOX, dc. 179 



afterwards verified almost exactly by Sir W . Thomson . 

 In this very remarkable speculation, an entirely novel 

 physical phenomenon was predicted, in anticipation of 

 any direct experiments on the subject ; and the actual 

 observation of the phenomenon was pointed out as a 

 highly interesting object for experimental research/ Just 

 as liquids which expand in solidifying will have the tem 

 perature of solidification lowered by pressure, so liquids 

 which contract in solidifying will exhibit the reverse effect. 

 They will be assisted in solidifying, as it were, by pressure, 

 so as to become solid at a higher temperature, as the 

 pressure is greater. This latter result was verified by 

 Buusen and Hopkins, in the case of paraffin, spermaceti, 

 wax, and stearin. The effect upon water has more recently 

 been carried to such an extent by Mousson, that under 

 the vast pressure of 1300 atmospheres, water did not 

 freeze until cooled down to -18 Cent. Another remark 

 able prediction of Professor Thomson was to the effect 

 that, if a metallic spring be weakened by a rise of tem 

 perature, work done against the spring, by bending it, 

 must cause a cooling effect. Although the amount of 

 effect to be expected in a certain apparatus was only 

 about four-thousandths of a degree Centigrade, Dr. Joule a 

 succeeded in detecting and measuring the effect to the 

 extent of three-thousandths of a degree, such is the deli 

 cacy of modern methods of measurement. I cannot 

 refrain from quoting Dr. Joule s reflections upon this 

 fact 1 . Thus even in the above delicate case/ he says, 

 is the formula of Professor Thompson completely verified. 

 The mathematical investigation of the thermo-elastic 

 qualities of metals has enabled my illustrious friend to 



z Maxwell s Theory of Heat, p. 174. Philosophical Magazine/ 

 August, 1850. Third Series, vol. xxxvii. p. 123. 



a Philosophical Transactions, 1858, vol. cxlviii. p. 127. 

 b Ibid. p. 1 30. 



N 2 



