Since the time of Newton, the mixture of colored stimuli has been an important means for studying color-vision. As Newton pointed out, when two colors are combined in various proportions, two laws of color-mixture are demonstrated. Sometimes the mixture resembles both of the components in some degree. When complementary colors are combined, the resultant may resemble either of the component hues, and, with careful adjustment of the amounts of each stimulus, resemblance to both can be eliminated. Both laws of color-mixture may operate simultaneously. Cancellation of a hue by known amounts of its complement may be used to measure the amount of that hue produced by a given stimulus. This principle has been used by Hurvich and Jameson to establish the quantitative basis for their paired component theory of vision.1 In the present study, it is proposed to use the cancellation of green produced by a stimulus at 5G2 mp to determine the relative redness of stimuli in the long-wave end of the spectrum. The basic assumption is made that if the same amount, in ft.L., of green is completely cancelled by red bands of different dominant wavelengths, the amounts of red will be the same. Wavelengths beyond 615 m,u, give substantially matching reds though requiring illcreasing amounts of energy as the wavelength gets longer. Wide-band red Slters have been used, because the total energy transmitted is much greater than that transmitted by a monochromatic filter. Consequently, much greater eiciency is obtained in many applications. Although two different band filters with transmissions beyond 600 mp may give reds that look alike, it cannot be assumed that they will act alike in all applications. Recent experiments have raised questions on this point with respect to mixture,2 and to adaptation.3