It will also be seen that some of his curves are negative curves lying below the base. This shows that the three standard colours he took are somewhat wrong. The dotted curve gives the combination of his three sensations at every point, and should be the luminosity curve; but owing to his having taken empirically certain standards of luminosity for his three colours, it does not represent the truth, as may be seen on comparison with Fig. 11, page 79. It must be recollected that since Maxwell's observations the subject has been largely experimented upon, and naturally improved appliances and greater knowledge have enabled more nearly correct views to be entertained regarding it. CHAPTER XIII. Match of Compound Colours with Simple Colours--All Colours reduced to Numbers--Method of matching a Colour with a Spectrum Colour and White Light. If we place the solution of bichromate of potassium in front of the slit of the collimator, we shall see that on producing a spectrum on the screen, all rays from the red to the yellow-green pass; hence bichromate of potash transmits a colour which is a compound colour. It has been shown that this orange colour and the spectral yellow can be matched by mixing the simple colours of red and green together; but it will be instructive to see if a simple colour in the spectrum itself can be found which can match such a compound colour as that of the bichromate. If we place the bichromate in the reflected beam of the colour patch apparatus and illuminate one shadow cast by the rod with the light transmitted by it, and pass a slit along the spectrum, to produce monochromatic light, with which the other shadow of the rod is illuminated, a position will be found near the orange sodium line "D," where the two colours apparently match in every respect; when the intensities of the two illuminated shadows are equalized as before by the rotating sectors. In the same way by filling the part of the square with the pigment on which the shadow illuminated by the reflected beam falls, we can see if we can match emerald green, cyanine blue, and other coloured pigments. It will often be--more often than not--necessary, however, to dilute the spectrum colour thrown on the white half of the patch with a trace of white light. By reference to our previous experiments we arrive at what may appear an unlooked-for result, that _no matter what the colour_ may be, we can refer it to one ray of the spectrum, together with a percentage of added white light. It is worthy of remark, that the place in the spectrum where the simple and the compound colours match, varies according to the kind of light with which the pigment is illuminated. This we can show in a very simple way.