114 jVickoh and Franklin — A Spectro-photometric, etc. 



in color and the spectro-photometric examination of their spec- 

 tra would give ns a set of curves corresponding in character to 

 those shown in figure 4. These spectra would be of the same 

 brightness in the yellow (\=6000), but at no other point 

 within the limits of the visible spectrum. The lamps would 

 differ in total energy of radiation, in the energy represented 

 by their light-producing rays, and in total luminosity. With 

 increasing incandescence the total energy of the regions lying 

 beyond wave-length 6000, and also their luminosity, increase 

 rapidly ; and the corresponding decrease in regions between 

 that wave-length and the red end of the spectrum is not suffi- 

 cient to keep the energy and luminosity of the whole visible 

 spectrum constant. It follows, therefore, that candle-power, 

 as determined by means of the Bunsen photometer, affords us 

 a correct measure neither of the light-giving energy nor of the 

 luminosity of the source of light ; the direction of the error 

 always being such as to favor sources of a low degree of incan- 

 descence when compared with those of higher temperature. 



The discrepancy between candle-power and luminosity, with 

 increasing incandescence, is less marked than that between 

 candle-power and energy of radiation, because luminosity de- 

 pends very largely upon the less refrangible rays of the spec- 

 trum. Schumann's admirable study of the color and bright- 

 ness of the incandescent lamp, to which reference has already 

 been made, affords data for a comparison of candle-power and 

 total luminosity of this source of light, luminosity being de- 

 termined according to the method of Mace de Lepinay. Cal- 

 culations based upon his measurements show that when the 

 incandescence of a glow lamp is raised from 10 candles to 50 

 candles, its luminosity is increased in the ratio of 6 : 1 instead 

 of 5 : 1. For light emanating from sources of higher tempera- 

 ture the discrepancy is still more marked. In the case of the 

 electric-arc light, for example, the ratio of luminosity to candle- 

 power is approximately 5 : 4 and in the case of day light it is 

 probably commonly in excess of 6 : 4. 



Every improvement, therefore in our methods of artificial 

 lighting which makes it possible to raise the temperature of 

 incandescence of the illuminant will be accompanied by a 

 double gain. On the one hand a greater candle-power per 

 unit of energy expended will be obtained, and on the other 

 hand, the light-giving value per candle-power will be increased. 



Every such step will render our present method of reference 

 to the standard candle as a unit less satisfactory, and the time 

 is probably not far distant when we shall have learned to mea- 

 sure the temperature of our illuminants and to express their 

 light-giving value in terms of the temperature of the glowing 

 material and the area of the illuminating surface. 



