Theory of the Flicker Photometer. 719 



In fig. 7 are plotted a set of the straight lines connecting 



^ and log I. These are diagrammatic, all drawn to cross 



arbitrarily at a point of large log I, and all to cross again at 

 the origin of abscissae after the abrupt change of slope at low 

 intensity. The slopes and other constants were here arbitra- 

 rily chosen of convenient values to represent on one plot the 

 complete behaviour of the flicker photometer. The line S 

 represents the standard illumination, e. g. " white/' while R 

 represents red light, G green, and B blue light. The carves 

 between the straight lines are the values found from equation 

 (8), and show the values of R, G, and B for the condition of 

 no flicker, calculated from the values shown and a value of 

 10 for F. 



By dropping vertically down or up to the corresponding 

 points on the straight line and reading values of K/X 2 , it is 

 possible to ascertain the phase difference introduced if the 

 indicated relative amounts of the various colours are chosen. 

 It is so small for large values of I in using either the ex- 

 perimental values or the arbitrarily chosen exaggerated 

 values of fig. 7 that its effect on the results, if any, is negli- 

 gible *. For small values of I this lag may become of 

 significance. To it is possibly due the loss in sensibility of 

 the flicker photometer at low illuminations. 



* The colour phenomena exhibited by black and white disks when 

 slowly rotated may be interpreted as showing the existence of con- 

 siderably different lags for different colours. But the present line of 

 reasoning- is not affected even if this be the case, for the only result of 

 such a condition would be a loss of sensibility. In the case of 180° 

 difference of lag for two colours the sensibility would be diminished to 

 zero. Great caution, however, must be exercised in reasoning from 

 phenomena accompanying interrupted and flickering sensations to the 

 case of an established condition such as exists in the flicker photometer. 

 The "overshooting" effect shown by covering either one of the two 

 compared lights in the flicker phoxometer is an instance of this. ICither 

 compared colour when violently flickering looks much brighter than a 

 steady illumination of the same mean value, and can even look brighter 

 than the steady illumination of twice the mean value, which is obtained 

 M-h en the flicker apparatus is stopped. But when both colours are dove- 

 tailed the non-flickering field appears identical in brightness with one 

 obtained by ordinary superposition of the two colours of the same mean 

 value, showing this overshooting to be associated with the sensation of 

 flicker. Similar caution must be applied to reasoning from the appearance 

 of briefly exposed fields {vide the work of Broca and Sultzer, Comptes 

 Rendus, exxxiv. p. 831, 1902). These phenomena apply to the instant 

 the flicker photometer is started, but not to the condition of steady 

 running where the illumination is built up from an infinite series of 

 impressions. 



