Page:JOSA-Vol 06-06.djvu/22

 D. .—It is not legitimate to represent the psychological primaries, as Titchener does, as corners in a quadrilateral construction, since there is no correlation between psychological primacy and saturation, which such a construction implies.

The functions which link the attributes of color, expressed in threshold units (vide supra) with the characteristics of the stimulus have thus far been determined only imperfectly. Reference should be had to the Report of the Committee on Visual Sensitometry () for data on the relation between brilliance and stimulus intensity. Jones (), utilizing measurements made by Steindler, Nutting and himself, has determined the function connecting the chromaticity of spectral colors with wave-length. Since the spectral colors, even for conditions of pure cone vision, are by no means of equal saturation (vide infra) Jones’ so-called “hue scale”’ is in reality a resultant hue-saturation scale of a very special kind. However, as an index of the law of change of chromaticity with respect to wave-lengths of homogeneous radiation, his results are of fundamental importance. They are reproduced in. Each chromaticity unit corresponds to one just noticeable difference (both in chromaticity and in wave-length), and the reciprocal of the wave-length difference is the sensibility to change in wave-length. It will be noted that this sensibility has four distinct maxima in the spectrum, the two most important ones lying at 494 and 588 m&mu;, respectively, where the wave-length threshold is approximately 1.0 m&mu;. Jones finds 128 just noticeable chromaticity steps in the spectrum, and about 20 additional steps in the non-spectral purples and magentas, as determined by their complementaries.

Since the hues form a cyclic series, it would seem more appropriate to express the hue scale in angular than in linear notation. If there are H hue steps in the complete cycle, the angular unit will evidently be 2&pi;/H radians. The magnitude of this unit, however—if it is to correspond always to an integral step—must vary with the saturation, so that the linear unit is probably the more convenient. No determinations have yet been made of the