The intensity of the different rays of the spectrum reflected from metallic surfaces can also be measured, if for the first or second right-angled prism a small piece of the metal is substituted, using it as a reflecting surface, as can also the rays reflected from any surface which is bright and polished. In Fig. 18 the dotted curves show the _luminosity_ of the spectrum reflected from the different metals, curve V being that of white light. These curves are derived by reducing the ordinates of curve V proportionately to the intensity curves. Thus at 49 brass reflects 77% of the light, and the luminosity of the white is 80. The luminosity of the light from the brass is therefore 77/100 of 80, or 61. This shows the method which is adopted, of deducing luminosities from intensities. Fig. 21.--1. Emerald Green. 2. Chromous Oxide. 3. Terre Verte. The light reflected from pigments can also be measured by the same plan. The procedure adopted is that carried out when measuring their luminosities, viz. to cause the ray from one spectrum to fall on a strip of a white surface, and that from the other on a strip of the coloured surface (see page 82). This is a more convenient method than that just described, when the coloured surface is small. The annexed figures (Figs. 19, 20, 21, 22) show the results obtained from various pigments. Fig. 22.--1. Indigo. 2. Antwerp Blue. 3. Cobalt. 4. French Ultramarine. Fig. 23.--Method of obtaining a Colour Template. From curves such as these we are able to produce the colour of the pigment on the screen from the spectrum itself. This is a useful proof of the truth of the measurements made. To do this we must mark off on a card (Fig. 23) the absolute scale of the spectrum along the radius of a circle, and draw circles at the various points of the scale from its centre. From the same centre we must draw lines at angles to the fixed radius corresponding to the various apertures of the sectors required at the various points of the scale to measure the light reflected from a pigment. Where each radial line cuts the circle drawn through the particular point of the scale to which its angle has reference, gives us points which joined give a curved figure. Such a figure, when cut out and rotated in front of the spectrum in the proper position (as for instance by making the D sodium line correspond with that on the scale), will cut off exactly the same proportion of each colour that the pigment absorbs. The spectrum, when recombined, should give a patch of the exact colour of that measured. The spectrum must be made narrow, as the template is only theoretically correct for a spectrum of the width of a line, as can be readily seen.