Temperature on the Colour of Pigments. 



405 



sideration, the spectra are separated by a much more clearly 

 defined median line than can be secured with the ordinary 

 comparison-prism at the slit. The sharpness of the inner 

 boundary between the spectra to be compared has great 

 influence upon the accuracy of the spectro-photometric de- 

 termination ; indeed, where close measurements are desired, 

 definition in this region may be regarded as essential. There 

 is another condition which is met by the use of the four 

 prisms. The selective absorption of the glass through which 

 the light passes in a spectro-photometer is by no means 

 negligible. Even in the clearest of optical glass the absorp- 

 tion varies in a marked degree with the wave-length of the 

 ray ; and it is therefore important that the two sets of rays 

 which' are to furnish the spectra to be compared should be 

 subjected to the same treatment, passing through the same or 

 similar lenses and prisms and traversing the same total thick- 

 ness of glass *. In the instrument used in our experiments 

 this condition was fulfilled in every respect but one ; viz. the 

 light from the comparison-lamp L passed through the con- 

 densing-lens A and the two Nicol prisms which were not in 

 the path of the rays from I/. The method by which the 

 absorption due to these was eliminated will be indicated later. 

 After having traversed the collimator-tube, the rays from 

 the pigment and those from the lamp L were dispersed by the 

 prism p and passed on to the eyepiece. The field of view 

 consisted of two spectra situated one above the other. By 

 means of an adjustable diaphragm in the eyepiece any desired 

 region of the spectrum could be isolated. When thus restricted 

 the field consisted of two narrow vertical strips of colour, 

 separated by a black line which was the image of the adjacent 

 edges of the reflecting-prisms before the slit. The two 

 strips of colour were identical in wave-length and differed only 

 in intensity. The brightness of the lower depended upon that 

 of the comparison-lamp L, and upon the angle between the 

 polarizing-planes of the Nicol prisms. Of these last the 

 polarizer N (fig. 1) had freedom of rotation, and was provided 

 with a circle reading to five minutes of arc. The brightness 

 of the upper field depended upon the amount of light, of the 

 wave-length in question, reflected by the pigment under 

 investigation. The spectrum of the comparison-lamp, although 

 reduced by polarization, and by absorption and reflexion 

 suffered in traversing the lens A and the calcite prisms, was 



* Abney and Festing, in their paper on Colour Photometry, already 

 cited, have pointed out the importance of this precaution (see p. 549 

 of their monograph). 



