OF ARTS AND SCIENCES. 239 



Mean Scale Readings of both Sets. 



10.9 13.0 12.2 7.7 



Relative Brilliancies. 



V 7.7 2 = 5929 " = 10800 " 285 



After measuring the brightness, I observed the limits of the spec* 

 truni under two different brilliancies, and the very curious effect was 

 noticed, that while the red end under the increased illumination 

 advanced considerably, — in the present instance 27', — the violet did 

 not move at all. The same effect is noticeable in all the lights to a 

 greater or less extent, the violet usually moving from 1' to 3'. This 

 is probably accounted for by the fact that the fluids of the eye absorb 

 nearly all the rays of short-wave length, thus cutting off all the 

 spectra at nearly the same place. The position of the red end, on the 

 other hand, depends merely on the intensity of the light. 



These figures do not represent the deviation of the ray, but merely 

 the numbering on my divided circle. Reducing them to wave lengths 

 we obtain : — 



Next the total brilliancy of the light in candle-powers is measured. 

 In the present case, two determinations were made, one at the end 

 of each set. These measurements were made with a Bunsen pho- 



