(96) In Fig. 15 the vertical curve describes the values of the spectrum as they grade from red through yellow, green, blue, and purple. The horizontal curve describes the chromas of the spectrum in the same sequence; while the third curve leaning outward is obtained by uniting the first two by two planes at right angles to one another, and sums up the three qualities by a single descriptive line. Now the red and purple ends are far apart, and science forbids their junction because of their great difference in wave length. But the mind is prone to unite them in order to produce the red-purples which we see in clouds at sunset, in flowers and grapes and the amethyst. Indeed, it has been done unhesitatingly in most color schemes in order to supply the opposite of green. (97) This gives a slanting circuit joining all the branch ends of the color tree, and has been likened to the rings of Saturn in Chapter I., paragraph 17. +A prismatic color sphere.+ (98) With a little effort of the imagination we can picture a prismatic color sphere, using only the colors of light. In a cylindrical chamber is hung a diaphanous ball similar to a huge soap bubble, which can display color on its surface without obscuring its interior. Then, at the proper points of the surrounding wall, three pure beams of colored light are admitted,--one red, another green, and the third violet-blue. (99) They fall at proper levels on three sides of the sphere, while their intermediate gradations encircle the sphere with a complete spectrum plus the needed purple. As they penetrate the sphere, they unite to balance each other in neutrality. Pure whiteness is at the top, and, by some imaginary means their light gradually diminishes until they disappear in darkness below. (100) This ideal color system is impossible in the present state of our knowledge and implements. Even were it possible, its immaterial hues could not serve to dye materials or paint pictures. Pigments are, and will in all probability continue to be, the practical agents of coloristic productions, however reluctant the scientist may be to accept them as the basis of a color system. It is true that they are chemically impure and imperfectly represent the colors of light. Some of them fade rapidly and undergo chemical change, as in the notable case of a green pigment tested by this measured system, which in a few weeks lost four steps of chroma, gained two steps of value, and swung into a bluer hue. (101) But the color sphere to be next described is worked out with a few reliable pigments, mostly natural earths, whose fading is a matter of years and so slight as to be almost imperceptible. Besides, its principal hues are preserved in safe keeping by imperishable enamels, which can be used to correct any tendency of the pigments to distort the measured intervals of the color sphere.