174 E. Merritt — Light from Incandescent Lamps. 



the final deflection was reached. The needle always moved 

 in this way, no matter whether the final deflection was large or 

 small. Using a steady gas flame as a source of heat, and tak- 

 ing special precautions to avoid drafts of air, I determined the 

 ratio of this first deflection to the final deflection throughout 

 the range of deflections used in these experiments. The ratio 

 was constant and equal to 32-6 per cent for deflections between 

 20 and 100 scale divisions. For deflections less than 20 the 

 ratio rose rapidly to 50 per cent at 1*7 divisions. The curve 

 found by plotting this ratio, and the first deflection, resembled 

 an equilateral hyperbola with vertical and horizontal asymp- 

 totes. 



In the experiments with incandescent lamps the first deflec- 

 tion alone was observed, and the value of the corresponding 

 final deflection was calculated from the curve just described. 

 As the first maximum was reached in about five seconds the 

 error from change of zero point was very small. 



The proportionality of deflections to heating effects was 

 shown by plotting a curve for each lamp tested, in which ordi- 

 nates represented deflections and abscissae Watts expended in 

 the lamp. In every case this curve was very nearly a straight 

 line passing through the origin. What slight variation there 

 was might be accounted for by the convection from the glass 

 of the lamp. 



In the actual tests the alum cell C was first removed and the 

 deflection corresponding to total radiation was observed. Then 

 the cell was placed in position, and the deflection due to the light 

 was taken. Finally a cell containing an opaque iodine solution 

 was placed between the alum cell and the pile, and the deflec- 

 tion again observed. This gave the correction for dark heat 

 passing through the alum. In each case the correction due to 

 any difference in temperature between the pile and the appa- 

 ratus was determined by taking the deflections when the lamp 

 was turned off. The correction for light absorbed was found 

 in the same way as when the calorimeter was used. It varied 

 from 25 per cent to 30 per cent as before. Only about 10 per 

 cent, however, of the radiations passing through the alum, 

 were dark heat. For low candle powers the percentage was 

 much less even than this. 



The lamp current was measured by a Thomson ammeter, 

 calibrated from the large tangent galvanometer. With each 

 lamp tested the curve between current and Watts was deter- 

 mined once for all by measurements with the galvanometer. 

 The energy expended in the lamp could then be calculated 

 from its current. 



The results for the four lamps tested by this method are 

 given below. As in table A, W is the total energy expended, 



