330 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS COLOR THEORY The primary specification of the color of a given sample is based on the spectrophotometric curve. The solid line in Fig. 1 is the spectro- photometric curve of a beige powder in the form of pressed cake. Plotted horizontally on the graph is the wavelength of light. The vertical dimension of the graph represents the relative amount of light reflected by the sample. Figure 1, for example, shows that less than 30% of the blue light is reflected and about 60% of the red light is re- flected. The exact reflectance of the sample for any wavelength of light may be read directly from the curve. The spectrophotometric curve of a colored product has several useful properties. If two samples have identical spectrophotometric curves, one will be a color match for the other, assuming no differences between them in texture or gloss. Furthermore, these two samples will appear to be a match regardless of what light is used for illumination and re- gardless of whether the observer has normal color vision. Unfortu- nately, however, the converse is not true. One cannot assume that if two samples are a color match under one particular illumination they have identical spectrophotometric curves and will match in all illumi- nants. The samples may be metameric that is, they may match under one illuminant but not under another. In this case the spectrophoto- metric curves will not be identical. This problem of metamerism is responsible for many of the color matching problems in industry. Another important property of the spectrophotometric curve of a mixture of pigments is that it exhibits the spectral characteristics of the individual pigments used in the mix. The color represented by Fig. 1, for example, was made with an iron oxide red, Mapico ©* yellow, ultra- marine blue, and white (titanium dioxide). The spectrophotometric curves of these pigments are shown in Figs. 2 and 3. It should be noted that the characteristic valleys which appear in the curves of the mix- tures of the colored pigments with white also show up in Fig. 1, i.e., when all three pigments are used in the mixture. The Mapico yellow shows a characteristic dip in the 400 to 450 nm range in both Figs. 1 and 3. The red shows the same slope in the 550 and 580 nm range in both Figs. 1 and 2. The characteristic slope of the ultramarine blue in the region 650 to 700 nm shown in Fig. 3 is repeated in Fig. 1, although somewhat obscured by the red and yellow. Because these pigment characteristics are retained in mixtures, appropriate pigments may be * Mapico is a trade name of Columbian Carbon Co., Trenton, N.J.

INSTRUMENTATION IN COSMETIC COLOR CONTROL IOO •'• COMPUTED•T R 40 UE 20 O•oo ,,•'o •6o •'o 60'0 6•'o 7oo WAVELENGTH, nm Figure 1. Spectrophotometric curves of a pressed cake sample ("True") containing red iron oxide, Mapico yellow, ultramarine blue, and tita- nium dioxide and the predicted curve ("computed") based on the known pigment quantities •00 450 500 550 600 650 700 WAVELENGTH, rim Figure 2. Spectrophotometric curves of pressed cake samples containing white only, a mixture of 1% red iron oxide and 99% white, and a mixture of 1% black and 99% white 331 iooj i i i i i 80 60 - . • J• YELLOW 0400 WAVELENGTH, nm Figure 3. Spectrophotometric curves of pressed cake samples containing 1% ultramarine blue and 99% white, and 1% Mapico yellow and 99% white I00 I [ I •60 ,oo ,•'o •[o do •o ' 650 700 WAVELENGTH, nm Figure 4. Spectrophotometric curve of a pressed cake sample containing Mapico yellow, red iron oxide, black and titanium dioxide