I have here our magic-lantern with its lime-light, in which the solid lime glows with a white heat, in consequence of the jets of oxygen and hydrogen burning round it. This was the light Kirchhoff used, and you know it will give a continuous bright band in the spectroscope. I put a cap with a narrow slit in it over the lantern tube, so as to get a narrow beam of light; in front of this I put a lens _l_, and in front of this again the prism _p_. The slit and the prism act exactly like your spectroscopes, and you can all see the continuous spectrum on the screen (_sp_, A, Fig. 48). Next I put a lighted lamp of very weak spirit in front of the slit, and find that it makes no difference, for whatever light it gives only strengthens the spectrum. But now notice carefully. I am going to put a little salt into the flame, and you would expect that the sodium in it, when turned to glowing vapour, causing it to look yellow, would strengthen the yellow part of the spectrum and give a bright line. This is what Kirchhoff expected, but to his intense surprise he saw as you do now a _dark line_ D start out where the bright line should have been.
What can have happened? It is this. The oxyhydrogen light is very hot indeed, the spirit flame with the sodium is comparatively weak and cool. So when those special coloured waves of the oxyhydrogen light which agree with those of the sodium light reached the flame, they spent all their energy in heating up those waves to their own temperature, and while all the other coloured rays travelled on and reached the screen, these waves were stopped or _absorbed_ on the way, and consequently there was a blank, black space in the spectrum where they should have been. If I could put a hydrogen flame cooler than the original light in the road, then there would be three dark lines where the bright hydrogen lines should be, and so with every other gas. _The cool vapour in front of the hot light cuts off from the white ray exactly those waves which it gives out itself when burning._
Thus each black line of the sun-spectrum (No. 2, Plate I.), tells us that some particular ray of sunlight has been absorbed by a cooler vapour _of its own kind_ somewhere between the sun and us, and it must be in the sun itself, for when we examine other stars we often find dark lines in their spectrum different from those in the sun, and this shows that the missing rays must have been stopped close at home, for if they were stopped in our atmosphere they would all be alike.