COBLENTZ: CONSTANTS OF SPECTKAL RADIATION 13 



tubes of the design used by Lummer and Pringsheim. Energy 

 curves were obtained from these tubes before and after they 

 were blackened with chromium and cobalt oxides. It was found 

 that the percentage of the energy curves fitting the Planck equa- 

 tion was greatly increased when the radiators were blackened. 

 There are still a few outstanding discrepancies which need further 

 investigation. For example, the average value of Co for a given 

 set of observations may be 0.2 to 0.5 per cent higher than a simi- 

 lar series made, under slightly different conditions, but using the 

 same prism, which was flawless. Usually these fluctuations coin- 

 cide with variations in humiditj'', but this is not sufficient to fully 

 explain the matter. Again, when using a fluorite prism con- 

 taining numerous flaws, which caused a conspicuous scattering 

 of light, the energy curves appear to be distorted so that only 

 25 to 30 per cent of them fit the Planck equation, and the value 

 of C2 is extraordinarily high, being of the order of C2 = 14,700. 

 The data obtained in 1910 and in 1911, using a flawless prism, 

 give an average value of Ca = 14,540. Using the same prism, in 

 1912, the value of the constant is slightly lower, being of the 

 order of C2 = 14,490. Altho this difference of 0.3 to 0.4 per 

 cent in the values of Ca is systematic it is within the experimental 

 errors of observation. It therefore appears that the weighted 

 value of this constant will be found to be close to Co = 14,500; 

 and \mT = 2920. This value of C2 = 14,500 appears to be about 

 1 per cent higher than the prehminary results published by 

 Warburg and his associates at the Reichsanstalt. 



The constant, C2, is of great importance in optical pyrometry, 

 and heretofore owing to the absence of concordant data experi- 

 menters have used values ranging from C2 = 14,200 to C2 = 14,600. 

 The value of the melting point of platinum, on the basis of the 

 optical temperature scale, using C2 = 14,500, was observed by 

 Waidner and Burgess to be 1753°. The latest work on the gas 

 temperature scale by Day and Sosman, extrapolating the thermo- 

 electric scale from palladium, m. p. = 1549?2, gives a value of 

 1750° to 1755°, with a mean value of 1752°, for the melting point 

 of platinum. A melting point of 1755° is equivalent to a value 

 of Co = 14,475 which is close to the predominating values of this 



