26 



PROCEEDINGS OF THE AMERICAN ACADEMY. 



volume. These changes were of the order of 200 kgm. for 20° intervals 

 and could be read on the bridge with an accuracy of about |%. The 

 agreement between the two independent sets of readings was not in 

 general as good as this, averaging about 2%. This is better than was 

 expected at first could be obtained with the method, and is certainly 



.160 



.120 



20° 30° 



Temperature 



Figure 3.^ A sample set of observations on the change of volume with 

 temperature 'at constant pressure. The ordinates are piston displacements 

 in inches. _ Two independent sets of readings are shown on the diagram, those 

 with the circles are the repeated set. The liquid shown here is carbon bisul- 

 phide. The accuracy for this is almost exactly the average accuracy for all 

 twelve liquids. 



much better than could have been obtained with the alternative method 

 of determining the difference between isothermal lines for different 

 temperatures. Figure 3 for carbon bisulphide shows a fair average 

 of the order of agreement. A more detailed account of the order of 

 accuracy will be given under the description of the individual liquids. 

 The material was now at hand for the construction of the table of 

 volumes. Up to 5000 kgm. the volume was to be tabulated at inter- 

 vals of 500 kgm., and above 5000 kgm. at intervals of 1000 kgm. The 

 difference in the length of the pressure steps is desirable because at 

 low pressures the volume changes more rapidly than at high pressures. 

 The tabulation at each pressure was to be made at temperature inter- 

 vals of 10°. The volume as a function of temperature at atmospheric 

 pressure was first tabulated. This was taken from the formulas of 

 Landolt and Bornstein, or from other sources to be described in detail 

 under the separate liquids. The agreement between different observ- 

 ers, even for atmospheric pressure, is not always as close as could be 



