20 PROCEEDINGS OF THE AMERICAN ACADEMY. 



the corresponding changes of volume were taken as negative. To 

 obtain the volume at lower pressures, the measurements at the lower 

 pressures were reduced in much the same way as has been described 

 for the higher pressures. One difference is that the measurements 

 at the lower pressures were at 20° instead of 40°, because several of 

 the liquids boil at less than 40° at atmospheric pressure. It was not 

 possible to reach entirely to the zero of pressure with the low-pressure 

 measurements, because of the slight friction of the piston even when 

 the pressure had not been pushed so high as to permanently upset 

 the piston. It was not possible to get much nearer to zero than 100 

 or 200 kgm. after the pressure had once been pushed to 2000 kgm. 

 In order to come still closer to zero, several measurements were made 

 during the very first application of pressure, before the cylinder has 

 been seasoned for hysteresis. During these first readings the pres- 

 sure was increased to about 1000 kgm. and then seasoned for hysteresis 

 as before. As a result the second set of readings between 200 and 

 2000 kgm. made after this seasoning process did not make close con- 

 nection with the first set. The discrepancy was due in part to hystere- 

 sis, but also in part to better adaptation of the packing to all the 

 crevices in the steel washers. The direction of the two curves was 

 usually the same, however, within the limits of error. This allowed 

 an extrapolation to zero by combining the results of the first set of 

 readings with those of the second. In this way the change of the unit 

 volume from atmospheric pressure up to 2000 kgm. was determined. 

 It must be remembered, however, that the readings at the very lowest 

 pressures are almost certainly in error, because the instrument is not 

 designed for low pressures. But with increasing pressure the readings 

 become more and more trustw'orth^^, until above 500 kgm. they seem 

 to merit entire confidence, if we can judge from self consistency. To 

 get the total change of volume from atmospheric pressure it is desir- 

 able, therefore, to supplement these readings with others made with 

 apparatus especially designed for low pressures. Such measure- 

 ments are afforded by the data of Amagat, and in some cases by 

 others. 



At the time that the computations of this paper were made, the 

 ■data of Amagat were the best that we had for the compressibility at 

 low pressures. Between the time of computing these results and 

 writing this account, however, there has appeared a paper by Richards,^ 

 in which the compressibility up to 500 kgm. of several of the liquids 



S Richards, Jour. Amer. Chem. Soc, 34, 971-993 (1912). 



