42 PROCEEDINGS OF THE AMERICAN ACADEMY. 



These values are in the units of this paper. Tt will be seen that the 

 results of different observers do not agree within 5%. The results of 

 von Reis, however, do justify us in assuming that Cp increases with 

 rising temperature. 



Ethyl Alcohol. More measurements were made on this than 

 on most of the other substances, because it was the liquid with 

 which the preliminary tests of the apparatus were made, but several 

 of the early runs were not carried to completion because of accident. 

 Measurements were made with five fillings of the apparatus. The 

 first of the five fillings was made with the alcohol enclosed in a glass 

 bulb, instead of in a steel one, as in the final experiments. This 

 filling gave all the information desired at the low pressures, and also 

 the thermal dilatation over nearly the entire high pressure range, 

 but was terminated by polarization effects in the manganin. The 

 polarization was found to be due to the breaking of the glass bulb, 

 allowing the alcohol to diffuse to the coil. The readings before the 

 break appeared should be trustworthy. The second of the five sets 

 of measurements was of the compressibility at high pressures, and was 

 completed without accident, but had to be discarded, for reasons that 

 will appear later. The third set was of the dilatation and compres- 

 sibility at high pressures. This also showed polarization, but not 

 until the very end of the compressibility run. For the second and 

 third runs a steel bulb was used, but the top was put on with soft 

 solder. This soft solder gave way under pressure, allowing the 

 kerosene to mix with the alcohol. The polarization probably did not 

 occur as soon as the solder cracked, because it takes time for the 

 alcohol to diffuse through the kerosene to the coil. The compres- 

 sibility measurements of the third run are, therefore, more likely to 

 be in error than the dilatation measurements, which were made a 

 considerable time before the polarization appeared. Because the 

 apparatus w^as the same, the second run is likely to be in error just as 

 the compressibility measurements of the third, the polarization not 

 having time to appear in the second run before the apparatus was 

 taken apart. The early dilatation measurements were retained, 

 therefore, and the early compressibility measurements discarded. 

 The agreement of the early compressibility measurements, which 

 presumably were made on a mixture of kerosene and ethyl alcohol, 

 was good, 0.3%, but they were about 4% higher than the results of 

 the final successful run. The last two of the five runs were carried 

 through without accident, one being of the compressibilitj^ and dila- 

 tation at low pressures, and the other the corresponding measure- 



