perfectly pure colours and pass our strip of scarlet paper along that variously coloured band of light, we shall be able to test the truth of several statements I have made as to the nature of colour. I have said colour is only an impression, and not a reality; and that it does not exist apart from light. Now, I can show you more, namely, that the colour of the so-called coloured object is entirely dependent on the existence in the light of the special coloured rays which it radiates, and that this scarlet paper depends on the red light of the spectrum for the existence of its redness. On passing the piece of scarlet paper along the coloured band of light, it appears red only when in the red portion of the spectrum, whilst in the other portions, though it is illumined, yet it has no colour, in fact it looks black. Hence what I have said is true, and, moreover, that red paper looks red because, as you see, it absorbs and extinguishes all the rays of the spectrum but the red ones, and these it radiates. A bright green strip of paper placed in the red has no colour, and looks black, but transferred to the pure green portion it radiates that at once, does not absorb it as it did the red, and so the green shines out finely. I have told you that sodium salts give to a colourless flame a fine monochromatic or pure yellow colour. Now, if this be so, and if all the light available in this world were of such a character, then such a colour as blue would be unknown. We will now ask ourselves another question, "We have a new blue colouring matter, and we desire to know if we may expect it to be one of the greatest possible brilliancy, what spectroscopic conditions ought it to fulfil?" On examining a solution of it, or rather the light passing through a solution of it, with the spectroscope, we ought to find that all the rays of the spectrum lying between and nearly to H and b (Fig. 16), _i.e._ all the bluish-violet, blue, and blue-green rays pass through it unchanged, unabsorbed, whilst all the rest should be completely absorbed. In like manner a pure yellow colour would allow all the rays lying between orange-red and greenish-yellow (Fig. 16) to pass through unchanged, but would absorb all the other colours of the spectrum.