14 PROCEEDINGS OF THE AMERICAN ACADEMY. 



only difference being that no seasoning precautions were taken. At 

 high pressures the thermostat was first adjusted to 40°, and then the 

 cyhnder seasoned over the entire pressure range by ad^'ancing the pres- 

 sure to the maximum and releasing it three times. The details of the 

 seasoning process are described in the former paper. The compressi- 

 bility measurements at 40° were then made. Readings were taken 

 both at increasing and decreasing pressures, so as to correct for hy- 

 steresis. The range of these measurements was from 2000 to 12000 

 kgm., returning after the complete set of readings to the initial pres- 

 sure of 2000 kgm. Next, the cylinder was seasoned for the thermal 

 expansion readings by advancing the pressure to 11500 kgm. raising 

 the temperature to 80° at constant volume, releasing the pressure 

 to 2000 kgm. at 80°, and finally reducing the temperature to 20°. The 

 thermal expansion measurements were then made by the method 

 already described. 



In the later form of procedure, in which the effects at high and low 

 pressures were measured with the same set-up, all the measurements 

 at the low pressures and the isothermal compressibility at 40° were 

 made on the same day; on the next day the thermal dilation at six 

 mean pressures was measured, and the apparatus taken down and set 

 up again with a new liquid ready for another run on the next day. 



By the use of large reservoirs of hot water and by nearly emptying 

 the thermostat at every change of temperature, it was possible to 

 make the readings with changing temperature fairly rapidly, the 

 elapsed time between readings at successive temperatures being 

 about fifteen minutes. A much longer time than this would have 

 been necessary to secure temperature equilibrium throughout the 

 mass of the cylinder if a special device had not been adopted. This 

 consisted in running the temperature past the final value, and then 

 returning to it. Thus, let us suppose that the temperature was to 

 be changed from 20° to 40°. Water was drawn from the thermostat 

 and enough hot water poured in to raise the temperature to 45°, and 

 not until the lapse of several minutes, the exact time to be determined 

 by experiment, was the temperature reduced to 40° and the regulator 

 set at this final value. After some practise it was possible to reach 

 temperature equilibrium in little more time than was necessary for 

 the manipulations of drawing water and putting in fresh. Readings 

 were never made, however, until at least three minutes had elapsed 

 without change of pressure. 



The actual experiments, after the preliminary work, occupied four 

 months, from February through May, 1912. Many of the early 



