THE MIXTURE OF PURE-COLOUR STIMULI 39 



From the experimental point of view a constant spectrum is chosen 

 and three colours are selected, e.g. a red, a green and a blue. Each 

 part of the spectrum is then matched by mixing different quantities 

 of the three together. This process is called " gauging the spectrum " 

 [Aiclinng cles Spektrums, v. Kries). There are many technical difficulties 

 and the results are not wholly free from objection. The best published 

 results are those of Konig and Dieterici^ and Abney and Watson. The 

 colour tables deduced from these results are seen in Figs. 8 and 9. 



Green 



Red G,n Blue 



Fig. 8. Colour triangle. A, B, C, etc., Fraunhofer lines. (Konig.) 



In Konig's curve the sudden bend at the extreme end of the violet 

 is probably due to fluorescence. 



As already mentioned colour mixtures often produce unsaturated 

 colour sensations. Consequently the match has then to be made 

 by a suitable addition of white light to the comparison spectral 

 colour. 



No three spectral colours can be chosen which when mixed will 

 accurately match in hue and saturation all spectral colours. There is 

 good physiological evidence of colour-sensations of much greater 

 saturation than the spectral colours {vide infra, Section V, Chap. ii). 

 Such colours, being less mixed with white, must lie outside the colour 

 table, in some such positions as shown at the angles of the circum- 

 scribed triangles in Figs. 8 and 9. 



1 Ztsch.f. Psychol, u. Physiol, d. Sinnesorg. iv. 241, 1893. 



