ELECTRICAL RESISTANCE UNDER PRESSURE. 101 



and had l)een last used in 1911, when it had been exposed to the 

 freezing- pressure of mercury at 0°, and had since been resting in a 

 dust proof container. The measurements on the hquid recorded here 

 were made in 1917. The techniciue of the previous work has })een 

 considerably impro\'ed in the inter\^al between 1908 and 1917. The 

 transmitting hquid used in 1908 was a mixture of water and glycerine, 

 and elaborate precautions had to be taken to avoid short circuits. 

 The liquid is now a hydrocarbon, and the insulation properties are 

 perfect. The values now obtained for the change of resistance differ 

 by about 1.5% from the previous ones. Part of this difference is 

 perhaps to be ascribed to improvement in technique, part to change 

 in the beha^'ior of the glass, which has been resting ten years since 

 blowing and annealing, and perhaps part to difference in the standard 

 of pressure. ]My very earliest results, which reached to only 0500 kg., 

 depended on a pressure gauge of smaller range and somewhat different 

 construction from that used in all my subsequent work reaching to 

 12000 kg. or more. All this later work depends on the same gauge, 

 and assumes as the fundamental constant that the freezing pressure 

 of mercury at 0° is 7(340 kg. Any discrepancy now found with the 

 former values for the resistance of mercury cannot affect the Aalidity 

 of any of the later work up to a range of 12000 kg. 



The glass was seasoned before the measurements by an application 

 of 11300 kg. at room temperature, and then by raising temperature to 

 96° while still under pressure. Five runs were made on the liquid, in 

 addition to the seasoning runs; these were at 0°, 26°, 50°, 7()°, and 97°. 

 The maximum zero displacement after a run at any of the temperatures 

 was 0.46% of the total effect. The maximum departure from a smooth 

 curve of any of the observed points at any temperature was 0.6% of 

 the total eft'ect, and the points usually lay on a smooth curve within 

 the sensitiveness of the readings, which was about one part in 5000 

 of the maximum eft'ect. 



The observed results were smoothed for pressure and temperature 

 and a table constructed for the resistance at uniform interAals of pres- 

 sure and temperature in the regular way. The resistance of the liquid 

 is shown in Table IX, and the coefficients are shown later in Table XX. 

 The resistances given in this table are corrected for the compressi- 

 bility and thermal expansion of the glass, using the values already 

 given. The correction for temperature amounts to a change of 

 obbcrA-ed resistance of 0.0007 for the range of 100°, which is about 

 0.74% of the measured eft'ect, and the correction on the pressure effect 

 is 0.0007 for 1000 kg., which is initially 2.2% of the pressure change. 

 The values gWen in the Table are, therefore, relative values of the 



