The spectrum of highly rarefied atmospheric air is chiefly that of nitrogen, for the oxygen combines with the platinum of the negative terminal, and is in too small a quantity to transmit the electricity through the tube. The rarefied vapours of chlorine, bromine, and iodine are so rapidly combined with the platinum of the negative terminal, that it is difficult to determine their spectra; but they have peculiarities in common, which distinguish them from all other spectra. The bright lines that cross them are first at rest, but soon become flickering. In the iodine spectrum, five of those lines of flickering light of great beauty are in the green, two of them close together. The bromine spectrum shows a greater number, which extend across the colours of its middle part, accompanied by dark lines; and in the chlorine spectrum there are many lines, both of flickering light and darkness. New lines are brought out in the iodine spectrum by increase of temperature. At a low heat it is crossed by a number of dark lines, but with a higher temperature the vapour has a greenish hue, which is resolved by the prism into green lines at some distance from one another, and fainter blue light, crossed by groups of luminous bands. Rarefied compound gases are resolved by the electricity into their component parts, and the result is superposed spectra, one belonging to each element. M. Seguin considers the aspect of the electric spark to be a sure indication of chemical action, for while the decomposition is in progress, the electric spark is encompassed by a halo, and the bright lines of the double spectrum are less distinct; but when the reaction is finished, the spark becomes slender, and the spectrum bands distinct. In the decomposition of highly carburetted and attenuated hydrogen gas, the spark resembles a flame, and the spectrum is like that of white light. When the gas is decomposed, the hydrogen is disengaged, and the carbon deposited on the extremities of the conducting wires; the spark becomes slender, and then the lines of the hydrogen, the lines belonging to the hydro-carbon and to carbon itself may be seen on the spectrum.[17] The bright and coloured lines on the spectra of the gases, and the vapours of a great number of the metals and metallic salts, were known before MM. Bunsen and Kirchhoff began their systematic researches, during which they added many more, some so difficult and analogous, that it required all their skill and experience to make them out.