430 PEOCEEDINGS OF THE AMERICAN ACADEMY. 



the alcohol in the capillary tube afforded an opportunity for evapo- 

 ration, and as considerable time elapsed between observations on the 

 coefficient of expansion, the loss due to evaporation would introduce 

 some error in the measurement of the change of volume. Further- 

 more, the liquid retreating down the capillary tube with decreasing 

 temperature, leaves a film on the walls of the tube. This makes the 

 observed change of volume greater than the true change of volume. 

 This film will not disappear for a long time, and a similar cause of 

 error enters when the temperature is again increased. Both of these 

 sources of error were avoided by having mercury in the lower part 

 of the dilatometer. 



It was hoped that a piezometer of the same form as the dilatometer 

 might be used in the work on compressibility ; for the error due to the 

 adhesion of the liquid on the walls of the capillary tube enters into 

 the data on compressibility as it did into the data on expansion. The 

 error due to the evaporation from the surface of the liquid was in 

 the work on compressibility not large enough to be of importance ; for 

 the observations on compressibility were made very rapidly. It was, 

 however, found that since it was impracticable to boil the mercury 

 into such a piezometer, the slight traces of air from which the bulb 

 and stem could not be entirely free, introduced serious error in 

 the compressibility at low pressures. It was therefore decided to 

 dispense with the mercury in the piezometer, and to use an instru- 

 ment of the kind shown in Figure 5. The liquid was boiled into this 

 piezometer in the same way in which it was boiled into the dila- 

 tometer. After the filling, the liquid in the bulb was heated 2° or 3° C. 

 above the temperature at which observations were to be made, ^yhen 

 che liquid again cooled down to the desired temperature, the exposed 

 surface was somewhere near the middle of the capillary tube P. 



It was possible to get a rough estimate of the error due to the 

 adhesion of the liquid on the walls of the tube. A glass tube of 

 about the same bore as the capillary tube of the piezometer was 

 filled with ether or alcohol and the liquid was then allowed to flow 

 out of it, leaving the inner walls of the tube wet. The weight of the 

 film of liquid adhering to the tube was determined at once. Knowing 

 the capacity of the tube, one found that about 2.5 per cent of the 

 liquid was left on the walls of the tube. In all the work on compres- 

 sibility the necessary correction was made for this source of error. 



Measurement of Change of Volume. 



Since the cross section of the capillary tube of the piezometer and 

 that of the dilatometer were accurately known, in order to measure 



