T. N. Cornsweet, Visual Perception (Academic, New York, 1970), p. 189.

In this paper the long-wavelength cones (loosely, red-sensitive cones) are referred to as *R* cones, the midspectral (green-sensitive) cones as *G* cones and the short wavelength (blue-sensitive) cones as *B* cones. The quantities *R*, *G*, and *B* are here used to stand for the excitations of the respective cone types by any given stimulus. The corresponding spectral sensitivities are the values of *R*, *G*, and *B* for an equal energy spectrum. The estimated spectral sensitivities for the three cone types used in constructing Figs. 1 and 2, expressed as linear combinations of Judd’s 1951 color matching functions, *x*¯_{j}, *y*¯_{j}, *z*¯_{j}, are*R*_{λ} = 0.15514 *x*¯_{jλ} + 0.54312 *y*¯_{jλ} - 0.03286 *z*¯_{jλ}, *G*_{λ} = -0.15514 *x*¯_{jλ} + 0.45684 *y*¯_{jλ} + 0.03286 *z*¯_{jλ}, *B*_{λ} = 0.01608 *z*¯_{jλ}.These are the spectral sensitivities of V. C. Smith and J. Pokorny [Vision Res. 15, 161–172 (1975)]. The coefficient applied to *z*¯_{jλ} is chosen for convenience, as explained in the text in connection with Fig. 2. Diagrams like the one proposed here could be constructed for any set of spectral sensitivities having the property that the sensitivities of the *R* and *G* cones suminate to give the photopic luminosity function.

The first true chromaticity diagram, suggested by J. Clerk Maxwell, Trans. R. Soc. Edin. 21, 275–298 (1855), was of this form but the equilateral triangle has not been popular in recent times. A. Konig and C. Dieterici, Z. Psych. Physiol. Sinnesorg. 4, 241–347 (1893) and W. A. H. Rushton, Invest. Ophthalmol. 10, 311–322 (1971) have proposed equilateral triangle diagrams based on cone excitations.

For evidence that the *B* cones make no contribution to luminance, as measured by flicker photometry or by the approximately equivalent minimally distant border method, see Smith and Pokorny (Ref. 2), B. W. Tansley and R. M. Boynton, Vision Res. 18, 683–697 (1978), B. W. Tansley and R. J. Glushko, Vision Res. 18, 699–706 (1978).

This drawback is characteristic of the CIE chromaticity diagram and all other chromaticity diagrams that have previously been proposed.

A nonlinear transformation of a similar diagram, sharing this advantage, was introduced by R. W. Rockeck [The Vertebrate Retina (W. H. Freeman, San Francisco, 1973), p. 737] for the analysis of color discrimination.

There is in fact reason to expect otherwise [see, for example, Fig. 6.57 on p. 558 of G. Wyszecki and W. S. Stiles, Color Science, (Wiley, New York, 1967)].

Working copies of the chromaticity diagram, and tables of the chromaticity coordinates (*r*_{λ}, *b*_{λ}) for spectral lights, are available from the authors.