the Light emitted by Incandescent Zinc Oxide, 21 



at a constant voltage, the temperature being below normal to 

 secure permanence. The quality of the light was not very 

 different from that of an ordinary luminous gas-flame. 



Similar measurements to the above upon the spectrum of 

 the light radiated from a surface of zinc oxide, the wave- 

 lengths selected being the same as those in the platinum 

 spectrum, made it possible to compare the radiating-power of 

 the two materials, when subjected to the same conditions of 

 temperature. 



The apparatus and the method used have been described in 

 detail in the paper on the Colour of Pigments, just cited. In 

 our determination of colour at high temperatures, a film of the 

 oxide which had been prepared by smoking a strip of platinum 

 foil over the flame of burning zinc was heated to the desired 

 degree by the action of an electric current upon the foil. The 

 surface of the oxide was illuminated by means of an incandes- 

 cent lamp, and the spectrum of the reflected light was compared 

 with that of another lamp similar to the first. In the experi- 

 ments to be described in this paper, the first lamp was dispensed 

 with, and a study was made of the light from the glowing- 

 surface itself. Since the rays to be measured were for the 

 most part very feeble, it was necessary to carry on the 

 experiments in a perfectly dark room. Temperatures, as in 

 our study of colour, were determined from the expansion of 

 the foil. 



The measurement of the temperature of incandescent foil 

 by means of the expansion of the metal is somewhat less 

 sensitive than the method of change of electric resistance, but 

 it is much to be preferred to the latter whenever a direct 

 calibration of the metal by reference to the air-thermometer, 

 or to the melting-point of metals, is impracticable ; for the 

 reason that the coefficient of expansion is much less affected 

 by traces of impurity than the resistance coefficient is. A 

 comparison of the various formulas proposed for the calcu- 

 lation of the temperature of platinum from its electrical 

 conductivity, is in itself sufficient to show the importance of 

 the direct calibration. Two specimens of platinum, which 

 had been ordered for the purpose from two well-known 

 dealers, who had been informed of the use to which the wire 

 was to be put and requested to pay especial attention to the 

 matter of the purity of the samples, were thus tested in the 

 physical laboratory of Cornell University by Mr. Ernest 

 Merritt. The curves for resistance and temperatures obtained 

 were widely different, the one being convex and the other 

 concave to the base-line of temperatures. Any single formula 

 for temperature, applied to these two wires, would have given 



