OF LUMINOUS IMPRESSIONS OF THE EYE. 39 
when the spot E disappears will give the ratio of the intensities of the lights 
transmitted by the aperture C. 
3dly, Adopting the arrangement in fig 3, we may compare the illumination 
of the screens by the following method :—Let the aperture D, whose light is re- 
flected at E, be made larger than the aperture ©, or let D be placed so that the 
distance D E F is considerably less than C E F. Then, by properly adjusting 
the position of the glass E, the aperture C will be seen projected on the reflected 
image of D. The distances of the lights A B may then be adjusted, so that the 
aperture C, seen through the reflected image of D, just disappears. The appa- 
rent brightness of C will then have a certain ratio to that of D, depending on 
the power of the observer’s eye to discriminate between lights of different 
intensity ; and, as before, the apparent brightnesses of C will be in the ratio of 
the squares of the distances of the flame B from the screen D when C disappears. 
From former experience I have found that it requires an almost painful effort 
of attention to compare the brightness of a surface illuminated by sudden 
flashes with that of another which is constantly illuminated. The method of 
observation now described seems to possess the great advantage of substituting 
mere watching for the disappearance of the flashes in the place of an estimation 
of equality of brightnesses. 
4thly, We may observe the brightness of impressions of short duration by means . 
of a single fiame in the following manner :—The disc G transmits flashes from 
the flame A to the aperture C in a screen, filled with translucent glass. The ob- 
server at E looks at U along the edge of another similar piece of translucent glass 
D. Since illuminated surfaces appear equally bright at all distances and at all 
inclinations of the visual ray, the ratio of the brightnesses of the aperture at C 
when the disc revolves, and when it is stationary or removed, will be that of the 
squares of the distances of the glass D from the flame A, at which the observer 
estimates C and D as equally bright. By this method of observation, the effect 
of any want of constancy in the illuminating power of the flame A will be entirely 
eliminated. 
