PHYSICAL PROPERTIES OF WATER, ETC. 51 



essentially unstable, can be approximately treated. But the up- or down-rushes which 

 result from instability are hopelessly beyond the powers of mathematics. 



One remark of a curious character may be added, viz. that in a very tall column 

 of water (salt or fresh), at the same temperature throughout, the equilibrium might be 

 rendered unstable in consequence of the heat developed by a sudden large increase of 

 pressure. For, as will be seen later, the expansibility of water is notably increased by 

 pressure ; and thus the lower parts of the column will become hotter, and less com- 

 pressible, than the upper. This effect is not produced in a tall column of air, for the 

 expansibility is practically unaltered by pressure. And the opposite effect is produced 

 in bodies like alcohol, &c, where the compressibility steadily increases with rise of 

 temperature. 



XII. Change of Temperature produced by Compression. 



The thermal effects of a sudden increase or relaxation of pressure formed an 

 important element in my examination of the Challenger thermometers, and were 

 practically the origin of this inquiry ; one of the most unexpected of the results I 

 obtained being the very considerable compression-change of temperature of the vulcanite 

 slabs on which the thermometers are mounted. Thomson's formula for this heating 

 effect, in terms of the pressure applied, and of the specific heat and expansibility of the 

 body compressed, is given in Appendix C to my former Report. My first direct 

 experiment on the subject was described as follows : ] — 



" When . . . the bulb of one of the thermometers was surrounded by a shell of 

 lard upwards of half an inch thick, the total effect produced by a pressure of 3j tons 

 weight was 5° F. ; while for the same pressure, without the lard, the effect was only 

 l° - 8 F. The temperature of the water in the compression apparatus was 43° F., so that 

 the temperature effect due to the compression of water was less than o- 2 F." 



On May 16 of the same year I read a second note on the subject, from which I 

 extract the following : 2 — 



"I have examined for a number of substances the rise of temperature produced 

 by a sudden application of great pressure, and the corresponding fall of temperature 

 when the pressure was very suddenly relaxed. The copper-iron circuit is, however, 

 too little sensitive for very accurate measurements ; as, from the nature of the 

 apparatus, the wires must be so thin as to have considerable resistance, and the 

 thermo-electric power of the combination is not large. ... I content myself, for 

 the present, with a general statement of the results for cork and for vulcanized india- 

 rubber, which are apparently typical of two classes of solids quite distinct from one 

 another in their behaviour. 



1 Proc. Roy. Soc. Edin., vol. xi. p. 51, 1881. ■ Proc. Roy. Soc. Edin., vol. xi. pp. 217, 218, 1881. 



