ADAMS AND JOHNSTON: STANDARD TEMPERATURE SCALE 283 



In this connection one point remains to be noted — namely the 

 accuracy of the Reichsanstalt scale in the region 1000-1100? 

 Holborn and Valentiner state in one place 16 that the uncertainty 

 at 1000° amounts to 2 to 3°; in another place, 17 in discussing the 

 reliability of their newer measurements at high temperatures, 

 they state that there is a difference amounting to 5° between the 

 older (1900) and the newer (1906) Reichsanstalt determinations 

 at 1100°, and continue: "The deviation from the mean would 

 still fall within the limits of error of the earlier determinations. 

 We consider it better however to attach greater weight to the for- 

 mer measurements, because the temperature gradient in the gas 

 thermometer bulb was much smaller in the earlier measurements/' 

 This may well be, for they give figures 18 which show that in the 1906 

 determinations at 1124° there were differences of temperature 

 from one point of the bulb to another of as much as 346 micro- 

 volts, or about 29°. 



Summary. In this note a new calibration curve for copper- 

 constantan thermo-elements, extending from 0° up to 360°, is 

 given, together with a series of independent measurements of 

 the temperature differences between the boiling points of naphtha- 

 lene (217?95) and benzophenone (305? 9) on the one hand, and 

 the freezing points of tin, bismuth, cadmium and lead on the other. 

 These measurements lead to the following values of the freezing 

 points: Sn, 231?8; Bi, 271?0; Cd, 320?9; Pb, 327?3. The con- 

 cordance of these values with those obtained by other measure- 

 ments show that the thermoelement is not inferior to the resist- 

 ance thermometer within this range of temperature (0 to 360°). 

 Moreover, a comparison of the results obtained with these inter- 

 polation instruments (thermoelement, resistance thermometer, 

 etc.) which measure not temperature independently but a well- 

 defined physical property which changes continuously with the 

 temperature, affords an excellent opportunity, through this con- 

 tinuity, for the discovery of inconsistencies in the gas thermometer 

 measurements. The remarkable concordance of the present 



16 Sitzungsber. Akad. Wiss. Berlin 44: 414. 1906. 



17 Ann. Physik. 22 : 19. 1907. 



18 loc. cit. p. 8. 



