I90I-2.] Photography in Natural Colours. 379 



luminosity or visual intensity. It is the compensation of these 

 irregularities by the filters that renders their adjustments so difficult. 



The spectrum method of adjusting filters, which are almost universally 

 composed of stained films or liquid solutions, consists in so altering 

 their absorptions that the opacity patches produced on negatives of the 

 spectrum obtained through them agree with the colour-mixture curves 

 (Fig. i). Photometric measurements of the densities of the negatives 

 are required, and the tedious nature of this process renders it impractic- 

 able for commercial use. It would I am satisfied give results un- 

 excelled by any other means, and this statement I hope to prove at 

 some future date by constructing a set of filters by such a method. 



The ideal method has, so far, only been considered ; the use of pure 

 red, green, and blue-violet spectrum light for the primary or reproduc- 

 tion colours, and the theoretical colour curves, based on measurements 

 made with such light. In practice, however, such conditions can not 

 prevail ; the nearest approach to the pure colours obtainable by 

 selective absorption must necessarily be used for the reproduction 

 colours, and the colour curves must be changed slightly to compensate 

 for the errors thus introduced. The stained films that were used for 

 backing the transparencies already shown are examples of such 

 monochromatic colours. Their analysis in the spectrum shows that all 

 but a fairly narrow band of the spectrum is absorbed [s/iown], but they 

 do not approach the purity of colour obtained by the three slits. These 

 reproduction colours, which must, first of all, fulfil the same condition as 

 the three spectrum primaries that, when mixed positively, white light 

 shall be produced, can be used in the same way as the three pure 

 colours to match any colour. This is effected by backing with these 

 colours three circular openings one in each of the three lanterns and 

 causing the three coloured patches thereby produced to overlap on the 

 screen [s/iown]. The intensity of the colours can be diminished, not, as 

 in the spectrum experiment, by narrowing the aperture but by inter- 

 posing patches of developed grey of different densities. By suitably 

 varying the intensities of these reproduction colours, in a similar degree 

 to the spectrum primaries, the whole range of spectral and extra- 

 spectral colours can be produced [s/wzvn] exactly as in the former 

 experiment. 



The use of these reproductions, instead of the pure colours naturally 

 leads to the use of an artificial instead of a pure spectrum as a test 

 object in the adjustment of filters. A number of small squares of 

 coloured glass, representing the principal colours of the spectrum as 

 red, yellow, green, blue, violet, and also including purple and white, are 



