Comparison of Sources of Artificial Illumination. 101 



3. various illuminating-gas flames ; 



4. a lime-light. 



5. electric arc-lights of the two prevailing commercial types 

 (long-arc and short-arc) ; 



6. daylight by clear and by clouded sky ; 



7. an incandescent lamp of high resistance, at various tempera- 

 tures ; 



8. an incandescent lamp of low resistance, at normal candle- 

 power. 



The spectro-photonieter employed in these measurements was 

 of the type already described in the pages of this Journal.* 

 For the comparison-flame usually employed, we substituted 

 however a sixteen-candle Edison lamp, maintained constantly 

 at its normal voltage. 



The rays from this lamp, having been rendered parallel by 

 means of a lens of four inches aperture and short focal length, 

 were passed through two Nicol's prisms, the first of which was 

 free to revolve, before entering the total-reflection prisms at 

 the slit of the spectroscope. This comparison lamp was placed 

 at the observer's right hand, the axis of the condensing lens 

 and of the Nicol's prisms being at right angles with the colli- 

 mator tube and in the same vertical plane as the slit. The 

 light to be compared was introduced into the collimator tube 

 from the opposite side, by means of a similar pair of total 

 reflection prisms. The arrangement of these four prisms has 

 been described in the article last cited. 



One of the chief sources of error in the comparison of vari- 

 ous illuminants, lies in the great range of temperature in the 

 glowing material which constitutes the source of light. Each 

 element of the radiating surface adds its quota to the general 

 illumination, and it is impossible to find any single region, the 

 spectrum of which is identical with that of the entire source. 

 We met this difficulty by allowing the light under observation 

 to fall upon the face of a block of magnesium carbonate, tak- 

 ing the spectrum of the rays reflected by the latter. Magne- 

 sium carbonate is not a pure white, f but it reflects all the wave- 

 lengths of the visible spectrum in sufficient quantity for the 

 purpose in question. 



Ten regions of the visible spectrum were selected for mea- 

 surement, and the intensity of the spectrum of an Edison in- 

 candescent lamp, giving 16 candles at 100 volts, was taken as 

 the unit of comparison for each region. All other spectra 

 were reduced to unit intensity in the region of the D line, 



* This Journal, vol. xxxvi, p. 332. 



f See E. L. Nichols, "On Black and White," Transactions of the Kansas 

 Academy of Science, vol. x. 



