This was determined by means of a Gaertner spectrometer, an EG and G spectroradiometer, and by visual observation.
A movable ground–glass–mirror combination was placed between M3 and BS2 to divert chromatic light from channel 2 to the photometer for measurement of the chromatic light. However, this was installed only in the later phases of this research.
G. Wyszecki and W. S. Stiles, Color Science (Wiley, New York, 1967).
The one exception is Jones and Lowry11 (Fig. 2, No. 3).
For a discussion of direct and indirect methods of psychophysical scaling, see T. Engen in Woodworth and Schlosberg’s Experimental Psychology, edited by J. W. Kling and L. A. Riggs (Holt, Rinehart, and Winston, New York, 1971).
In these experiments, the two halves of the bipartite field were always kept equally bright by appropriate adjustment of the spectral or white component.
We are grateful to a reader whose comments prompted us to describe the precautions taken against such confounding variables. Although DMB is a coauthor, she was naive with regard to the theoretical issues and purpose of this research while she was an observer. Her major contribution to this paper is in the discussion of the Hurvich and Jameson model.
These data were often collected in conjunction with other experiments. Consequently, we did not use a constant illuminance level. The initial level was 50 td with later data collected at 100 and 150 td. Some observers were not available to be run under all conditions.
There is an assymmetry of the variability bars about the mean. This is related both to the use of a logarithmic ordinate scale and to our normalization procedure. In this procedure, we divided the mean value for 570 nm by the mean value plus or minus one standard deviation for each wavelength used in the experiment. As a result, large indices of saturation are at the top of the graph. Had we performed this normalization on the raw data and then calculated means and standard deviations, the variability bars would have been symmetrical on a linear scale. Our purpose in presenting variability data is for interobserver and intermethod comparison. Because the plotted values were all computed in the same way, they are useful in making these comparisons.
A clear description of the Hurvich and Jameson model as applied to saturation is given in Ref. 31. We have added the ability of this method to account for the chromatic response resulting from white light. Hurvich and Jameson assumed that the chromatic component from broadband radiation is zero. (See Ref. 31 p. 435). The transformation from the CIE color-matching functions to (y–b), (r–g), and (w–bk) are presented in Refs. 26 and 31.
C. H. Graham, in Vision and Visual Perception (Wiley, New York, 1965).
S. L. Guth, Color 69 Association Internationale de la Couleur Proceedings, V. 1, p. 172, (Muster-Schmidt, Gottingen, 1970). P. K. Kaiser, G. Wyszecki, and G. Fielder, Paper presented at Annual Meeting of Association for Research in Vision and Ophthalmology, April, 1975.
E. HeringOutlines of a Theory of the Light Sense, translated by L. M. Hurvich and D. Jameson, (Harvard, Cambridge, Mass., 1964); P. K. Kaiser, P. A. Herzberg, and R. M. Boynton, Vision Res. 11, 953 (1971); P. K. Kaiser and J. P. Comerford, Vision Res. 15, 1399 (1975).