1896.] of Moonlight and Starlight compared loith Candle. 315 



giving it varying lengths of exposure to a source of light on different 

 small square areas of the plate; on development with the ferrous 

 oxalate developer a scale of graduated opacities was obtained which 

 were carefully measured by the method I have already described in 

 my paper " On the Transmission of Sunlight through the Earth's 

 Atmosphere" ('Phil. Trans.,' 1891). Visually, the measurements 

 of this scale were very good as circumstantial evidence proved, 

 but it remained to determine whether the silver opacities were more 

 or less opaque to the rays used in photography than they were to 

 the eye. Four scales of gradation had been made on the same plate, 

 by giving exposures for the same time to four distinct squares of the 

 plate, another exposure was made on another four distinct squares, 

 and so on ; so that eventually, on development, four complete scales 

 of similar gradation were prepared — all being developed together. 

 Each scale was measured and found to be practically the same. The 

 four scales were then separated. Platinum paper was placed behind 

 each, and they were severally placed accurately at such distances 

 from the positive pole of the electric arc light that the illumination of 

 the paper, after passing through a square of different density on each 

 scale, should, according to the eye measurements, be the same. If 

 the photographic and visual opacities were the same, then, on 

 illuminating the four scales which, of course, were fixed normally 

 to the line joining them and the carbon points, the amount of 

 platinum black deposited, after treatment with oxalate of potash, 

 should be the same on the part of papers covered by these squares 

 whose transparency was being tested. 



Several experiments carried out in this way showed conclusively 

 that the opacities, visual and photographic, were the same. In no 

 case was there a variation from the calculated opacity of 2 per cent. 

 It may therefore be taken that the scale of gradation used in the 

 moonlight and starlight measurements will allow intensities of photo- 

 graphic light to pass through in the same proportion that it does visual 

 light. For convenience, the transparencies of the different squares of 

 the scale were calculated out in powers of 2, for I have shown 

 elsewhere that, by using intensities of light acting on a plate in a 

 geometrical series for abscissae, and making the measured trans- 

 parencies the ordinates, a curve is produced which is, for a consider 

 able distance, practically a straight line. For checking the results of 

 measurement it thus becomes of great use. Moreover, this plan 

 enables us to use a much greater range of exposure for diagrammatic 

 purposes than is practicable if such range is expressed as an arith- 

 metical scale. Hence its adoption for the experiments to be described. 



It must be remembered that there are two ways of giving varying 

 exposure to light, one by using a constant source of illumination, and 

 varying the time of exposure, the other by exposing for a constant 



VOL- LIX. Z 



