PHYSICAL PROPERTIES OF WATER, ETC. 55 



As. was to be expected from the fact that the getting up of pressure requires a 

 short time, while the relief is practically instantaneous, the heating effect is generally 

 a little smaller than the cooling effect for the same change of pressure. 



These experimenters thus completely confirmed my statements as to the curiously 

 exceptional behaviour of cork, but they found no other substance, in the long list of 

 those which they examined, which behaves in a similar manner. 



It is to be remarked that as, in all the experiments described or cited in this 

 section, the temperature-changes were measured by a thermo-electric junction which 

 was itself exposed to the high pressures employed, there may be error due to the com- 

 pression of the materials forming the junction. The wires were, for several reasons, 

 very thin ; so that the error, if any, is not due to changes of temperature in them, but 

 to (possible) change of relative thermo-electric position, due to pressure. This is a 

 very insidious source of error, and it is not easy to see how to avoid it. 



XIII. Effect of Pressure on the Maximum-Density Point. 



Though the lowering of the maximum-density point of water by pressure is an 

 immediate consequence of Canton's discovery, that the compressibility diminishes as 

 the temperature is raised, it seems to have been first pointed out, so lately as 1875, by 

 Puschl. 1 I was quite unaware of his work, and of that of Van der Waals, 2 when (as 

 shown in Section XII. above) I was led to the same conclusion by the differences 

 between theory and experiment, as to the heat 

 developed by compression of water. 



This can very easily be shown as follows. 

 Let the (vertical) ordinates of the curve ABC 

 represent the volume of water at 1 atm., the 

 abscissae the corresponding temperatures, B the 

 maximum-density point. Let the dotted curve 

 abc represent the same for a greater pressure, 



say two atmospheres. Then, by Canton's result, L?_ . t 



the vertical distance between these curves (the 



difference between corresponding ordinates) diminishes continuously from A to C ; 

 so long, .at least, as the temperature at C is under that of minimum compressibility. 

 Hence the inclination of abc to the axis of temperatures is everywhere greater than 

 that of the corresponding part of ABC. Thus the minimum, 6, of the dotted 

 curve (where its tangent is horizontal) must correspond to a point, /3, in the full curve, 

 where the inclination is negative — i.e. a point at a lower temperature than B. 



i Sitzungsb. d. maih.-naturw. CI. d. k. Akad. d. Wiss. Wien, Bd. lxxii. p. 283, 1875. 

 2 Archives Niierl., torn. xii. p. 457, Haarlem, 1877. 



