512 Johnston and Adams — Influence of Pressure on the 



cases which have heen investigated : namely, for couples com- 

 posed of platinum or other metals with mercury the difference 

 of reading produced by a change of pressure of 1000 atmos- 

 pheres is only about 0"2 microvolts.* It is therefore practi- 

 cally certain that the error introduced by neglecting this 

 variation does not exceed one or two microvolts, i; e., it is less, 

 and probably much less, than 0'05°. 



It may be stated that we were able to get very sharp points ; 

 it is obvious that the temperature did not stay long at the 

 melting point, on account of the smallness of the charge and 

 of the rapid rate of cooling, but the break in the curve was 

 perfectly definite and reproducible. This is shown by the 

 results, which have been brought together in chronological 

 order in Table II, for the four metals investigated, namely, 

 tin, lead, cadmium, and bismuth. 



In the second column is given the pressure in the system in 

 atmospheres at the instant of freezing. The third column 

 shows the observed reading of the thermoelectric E.M.F. at the 

 freezing point ; while in the last column these latter values are 

 converted into degrees (uncorrected, that is, taken directly 

 from Table I). Freezing points rather than melting points 

 were taken since the former were with our apparatus much the 

 sharper. Melting points were taken in a few cases, however, 

 and found to agree well with the freezing point under the 

 same pressure. 



A study of the results in Table II shows conclusively (1) that 

 the melting point of a given metal under a given pressure is 

 reproducible, since measurements made on different days, with 

 different charges are in good agreement ; and (2) that since the 

 same results were obtained with increasing and decreasing 

 pressure, the hysteresis effect of the gage was not large enough 

 to interfere seriously with the accuracy of the measurements. 



Relation Between Melting Points and Pressures. 



It was expected that the change in melting point, At, would 

 be a linear function of the pressure P. Accordingly, the 

 straight line, t — a + bP, which best fits the data, was calcu- 

 lated by the method of least squares, from the observations 

 with each of the four metals used. The results of these calcu- 

 lations are shown in Table III. In the first column of each 

 table is given the pressure arranged in ascending order of 

 magnitude. Columns 2 and 3 show respectively the observed 

 change of melting point and that calculated from the 

 least square curve. The differences between observed and 



*E. Wagner, Ann. d. Physik., xxvii, 955, 1908 ; HOrig, ibid., xxviii, 371, 

 1909. 



