[8] See the colored figure on the frontispiece of this work, which clearly illustrates this method of color measurement. Larger disks of spectrum red, green, and violet are interlocked and adjusted so that they present, respectively, 32, 42, and 26 per cent. of the circumference; superimposed on these is a single smaller disk of neutral gray, and on this two still smaller disks of black and white, the former occupying 79, the latter 21, per cent. of the area. The result of this combination of colors, when the disks are rapidly revolved, is that the entire surface becomes a uniform neutral gray precisely like the middle disk, which blends so completely with the color inside and outside its limits that no trace of division can be detected. Hence, neutral gray equals a combination of red 32, green 42, and violet 26 per cent., and also equals a combination of black 79 and white 21 per cent. As further illustrating the point, it may be mentioned that not only does the above-mentioned combination of the three primary colors equal neutral gray but so also does the combination of any color ("secondary" or "tertiary" as well as primary) with its complementary, though the darkness or lightness of the gray varies somewhat, as the following table shows: Spectrum Color. Complementary Color. Equivalent Gray. Name. Per Cent. Per Cent. Composition. Black. White. Red 44 56 Blue 41 + Green 59. 72.5 27.5 Orange 28.5 71.5 Blue 51.5 + Green 48.5. 69 31 Yellow 33 67 Blue 60.5 + Violet 39.5. 64 36 Green 51 49 Red 57.5 + Violet 42.5. 73 27 Blue 64 36 Yellow 82 + Orange 18. 62 37 Violet 62.5 37.5 Yellow 69 + Green 31. 61.5 38.5 [9] The number is doubled so that every other one represents an intermediate hue not shown in color. [10] Owing to the circumstance that spectrum orange does not, at least when mixed with gray, fairly represent a medium hue between red and orange, being much nearer the former, a hue much near to yellow (yellow-orange, No. 15) has been selected. [11] For satisfactory color-wheel work it is necessary to discard practically all the so-called artists' colors, as being much too dull to even approximately represent the colors of the spectrum, and to substitute carefully selected aniline or coal-tar dyes, of which, fortunately, there is a very large number of remarkable purity of hue. Indeed, the work of most color-physicists is vitiated by their use of such crude colors as vermilion, carmine, scarlet-lake, chrome yellow, emerald green, Prussian blue, etc. (For a list of dyes and pigments used in preparing the Maxwell disks representing the thirty-six colors of the chromatic scale, see pages 26, 27.)