the yellow, and violet where it overlies the blue, the yellow and blue giving a bright green, while the central space, under the action of all the colors, is white (Fig. 1). As I said, the diagram now on the screen is the typical expression of the old theory, and is constructed so as to humor as much as possible the ideas of its supporters. If I had selected three pigments, and honestly worked the diagram out by their mixture, the result would have been much less brilliant and attractive.
Let me make an actual experiment on this point: I throw upon the screen the image of three plates of stained glass; their colors are red, yellow, and blue; they are also rich and intense. These pieces are arranged so as to correspond to our three colored circles, and, in fact, where the blue crosses the yellow, a green hue is actually produced, but it is darker than either of its constituents; the violet is much darker than the red or blue, and, where all the plates cross at the centre, we have, instead of white light, complete darkness (Fig. 2). These peculiar strides in each case toward blackness would have been observed, if a corresponding experiment had been made with any three pigments, but this at present is a minor matter, and I leave it for the consideration of a vastly greater difficulty under which the old theory labors.
Let us inquire how the superimposed yellow and blue glasses came to produce green. The white light of the lantern contains all the different luminous waves, and it so happens that the yellow and blue glasses both agree mainly in transmitting only waves of a medium length, or, what is the same thing, green light. This can be proved by an examination with the spectroscope, which also reveals the fact that their agreement in this respect is by no means perfect, and that the green rays are also compelled to pay toll for their passage, though in a less proportion than the others. Exactly the same reasoning applies to blue and yellow pigments, and, from the effects produced by their mixture, it does not in the least follow that yellow and blue light make green light. This important point I now propose to test by what may be considered a fundamental experiment. For this pur-