HRIDGMAN. — THERMODYNAMIC PROPEUTIKS OF WATER. 335 



more uniform and consistent results when applied to the same set of 

 data than the fi;raphical method. 



The method of computation adopted was first to calcuhite inde- 

 pendently from the individual observations of each set of readinj^s the 

 thermal dilatation at six mean pressures between 2200 and 12,000 kgm. 

 Then smooth curves were drawn through these points for each set of 

 readings, the curves being spaced in tlie best way so as to give regular 

 variations with both pressure and temperature. The values given 

 by the smooth cur\-es of each set of readings were then combined into 

 the grand mean. In taking this grand mean, as already explained, 

 almost the entire weight was given to the last two sets of readings. 

 The agreement between the different sets was best at the higher 

 temperatures, 60° to 80°, and about equally good between 20° and 

 40° and 40° and 60°. All four sets of curves, while not agreeing very 

 well as to the numerical value of the coefficient, do agree as to the 

 general character of the results, which are, perhaps, not quite what 

 would be expected. The unexpected feature is the change in the 

 sign of the temperature derivative of the dilatation at the higher 

 pressures. At the low pressures the dilatation is greater at the higher 

 temperatures, but at the higher pressures the thermal dilatation 

 becomes less at the higher temperatures. This essential feature is 

 verified on all four sets of curves. There are indications that it may 

 be an essential characteristic of the behavior of any normal liquid at 

 high pressures, and that it is not peculiar to water alone. This is 

 shown by the work on kerosene, and is also indicated by the work at 

 present being done on still other liquids. This will be taken up in 

 greater detail later. The other feature not to be expected is the 

 increase in the value of the thermal expansion between 20° and 40° 

 at the higher pressures. It is to be distinctly expected that the ther- 

 mal dilatation will decrease with rise of pressure, as indeed it does for 

 all the other intervals of temperature, but this rise between 20° and 40° 

 is shown by all the sets of determinations and seems to be an un- 

 doubted fact. It is probably connected with some new abnormality 

 in the behavior of water at the higher pressures, which may be con- 

 nected in some way with the appearance of the new variety of ice. 



The values fuially taken as the best values for the thermal dilatation 

 are the mean of the results of the four determinations, much the greater 

 weight being given, as already explained, to the two latter determina- 

 tions. Figure 2 gives these results, as also those of the other methods 

 at the lower pressures. The agreement of the two best determina- 

 tions at the higher pressures is about 5% for the lower temperature 



