22 Messrs. Nichols and Snow on the Character of 



values which at one thousand degrees were more than one 

 hundred degrees apart. We think that the well-known 

 formula of Matthiesen, 



l=l (l + -00000851 1+ -0000000035 * 2 ), 



which we made use of in these experiments, may he applied 

 to any commercial specimen of platinum, of reasonable purity, 

 without the introduction of such serious errors, and that, 

 whatever may be the general correction which it may be 

 found necessary to apply to that formula, in order to obtain 

 temperatures in absolute measure, it will always be possible 

 for any one knowing the formula used to be sure of the 

 approximate temperatures, without reference to the character 

 of the individual piece of platinum employed. 



Table I. contains the relative intensities of the platinum 

 spectrum between 700° and 1000° C. The reference standard 

 was the incandescent lamp already referred to, and the 

 intensity of the region of the D line of the platinum spectrum 

 was taken as unity throughout. Table II. gives the intensities 

 of the spectrum of the film of zinc oxide for the same wave- 

 lengths and the same range of temperatures. The results are 

 presented graphically in figs. 1 to 6, six of the nine wave- 

 lengths being plotted. In these figures the isochromatic curves 

 of platinum and of the oxide are plotted side by side to facili- 

 tate comparison. The curves for platinum are smooth and 

 all of the same type. They agree in all essentials with those 

 obtained from the measurements described in the articles 

 already cited (Nichols, 1879, and Violle, 1881). The corre- 

 sponding curves for zinc oxide, however, are all broken at a 

 point in the neighbourhood of 880°, at which temperature the 

 radiating-power evidently undergoes sudden and marked in- 

 crease, continuing with further rise of temperature to follow 

 a law different from that which prevailed at lower tem- 

 peratures. It will be noticed, moreover, that in the extreme 

 red (A-^'713) the platinum is brighter throughout the entire 

 range of temperature than the oxide. The succeeding curves, 

 however, cut each other at 1000° (A, =• 638), at 920° (\ = -587), 

 at 755° (X=*511), and at some lower temperature than 700° 

 for the blue and violet. Throughout the entire range the 

 colour of the oxide is whiter, that is to say, it is relatively 

 richer in the shorter wave-lengths than the light from platinum 

 at the same temperature, There is in the case of the oxide a 

 region of comparative feebleness in the yellow and green, 

 corresponding to the region of maximum reflecting -power in 

 the heated solid. To every one who has watched the striking 

 behaviour of zinc oxide under the blowpipe-flame, the effects 



