354 SIR W. DE W. ABNEY : MODIFIED APPAKATUS FOR MEASUREMENT OF COLOUR 



the wrong colours being used. In order to take the three negatives from which the 

 prints are produced, it is necessary to place screens of different colours (reddish, 

 greenish, and blue) in front of the sensitive plate in order to get distinctive images 

 which will represent the three sensations in the three printings. As to the printing 

 itself, nothing need be said in this communication, but I shall confine myself to the 

 negatives alone. If the negatives are correct, three transparencies from them should 

 give three images, which, if illuminated by the three colours which represent best 

 the three sensations, and superposed, should give the true colours of nature. 



Where the three positives are each devoid of deposit at the same part of the image, 

 the mixture of colours should give white, which means that in the negatives the 

 deposits should be equally opaque. This is the starting point of the process. 



The deposit being without colour, the different parts of the three component 

 negatives have to be such that the transparencies, when projected on a screen, allow 

 so much of each coloured beam to pass as will give the natural colour by mixture. 

 [It may be remarked that the negatives themselves, if illuminated with the three 

 colours, and the images superposed, should show the complementary colours.] If 

 there were a perfect photographic plate, there would not be much difficulty in 

 calculating directly the colours for the three screens which should be used. As, 

 however, no photographic plate is perfect in one sense, the proper exposing screens 

 have to be ascertained by trial. It is useless to make such trials with the spectrum, and 

 I have adopted a system which allows an accurate determination to be made by trial. 



(23.) The Pritidple on which a Colour Sensitometcr is Made. 



The principle I have employed, and which has been outlined before, is as follows : 

 If we have to find a screen to take what we may call the red negative (i.e., one in 

 which the opacities of deposit are proportional to the red components of the objects 

 photographed), we may take a variety of pigments, each of which contains red, and 

 utilize them for the purpose. Such pigments may show a diversity of luminosities, 

 and the relative proportions of red, green, and blue will also be very different in each. 

 If (say) squares of paper are covered with the pigments of different colours and 

 photographed through almost any coloured screen we should be unable to say without 

 measuring the different opacities of deposit whether the screen was correct or not. 

 If, however, by some artifice we are able to make all the red components in each 

 of the pigments identical, and then photograph them, it is evident that the only 

 screen which would be correct would be that which would make the opacities of all 

 the images of the different squares of colour the same. The mode I have adopted of 

 reducing the intensities of pigments and making all the luminosities of red, green, or 

 blue the same, is by making annuluses of the different pigments and filling up parts 

 of them with black pigment (the amount of white light reflected from such black 

 being measured and taken into account), and then rotating them round the centre of 

 the disc on which they are fixed. 



