the fact that the spark does not show the cyanogen bands when cyanides are submitted to its action. In this case it is possible that the temperature is too high, and that the cyanogen is decomposed, possibly by oxidation, for there can be no doubt that such condensed sparks are at a higher temperature than that of the arc. We know, too, that several metals are oxidised when volatilised in the spark, if not entirely at least partially.* But by using gold electrodes with the cyanides we do not obtain even a carbon spectrum.
Here again, possibly, the carbon is oxidised, and we know that carbon dioxide in carbonates yields no spectrum of carbon, nor any lines peculiar to carbon dioxide.
I have sought in every direction for a reasonable explanation of that which, up to the present, has proved inexplicable, in order that by working on some hypothesis one might devise a means of putties' the matter to experimental proof. This has now been accomplished in the following manner.
An almost saturated solution of pure crystallised potassium cyanide was put into a tube fitted with graphite electrodes in the manner described in a previous communication.f
The apparatus was fitted into a horizontal wooden tube with a window of quartz at one end, and carbon dioxide was passed into the tube until filled. The spark was then passed for five minutes, and again for ten minutes, a photograph being taken of the two spectra. The instrument used gave a dispersion equal to four quartz prisms. A glass tube with a similar window of quartz was fitted with g-old electrodes and filled with cyanogen gas, and another spectrum was photographed. A fourth spectrum was then obtained by passing cyanogen into the wooden tube containing the graphite electrodes; after the carbon dioxide had been expelled by air and replaced by cyanogen, the (J-tube was filled up with the solution of potassium cyanide. In all four cases the principal group of the cyanogen bands was obtained, but it was not very strong. A flame of cyanogen was then photographed with exposures varying from one to two, five, and ten minutes. A beautiful series of spectra was obtained, and the lines belonging to the edges of bands constituting the principal group were found to coincide exactly with those photographed from the potassium cyanide solution when the spark was passed in an atmosphere of carbon dioxide and in cyanogen, also when the spark was passed between gold electrodes in cyanogen. These appear to be the bands referred to by Eder and Valenta, which were described as carbon bands% when graphite electrodes were used with the spark
- ‘Roy. Soc. Proc.,’ vol. 49, p. 448, “ On the Physical Characters of the Lines in
the Spark Spectra of the Elements.” f ‘ Phil. Trans.,’ vol. 175, p. 49, 1884. X Hartley and Adeny, ‘ Phil. Trans.,’ vol. 175, p. 63, Part I, 1884.
