This paper deals in the first place with the effect of minute gaseous impurities which have apparently an important effect in promoting the formation of active nitrogen by the electric discharge. Oxygen is one of them. It is shown that the action of oxygen must be on the walls of the tube: for when a minute oxygen tributary is added to the nitrogen stream, it takes much longer to assert its action than the time required to establish the changed composition of the gas stream: and similarly when the oxygen tributary is checked its activity persists for a time. This is interpreted to mean that the action of the oxygen or other impurity is to modify the glass wall in a way favourable to the accumulation of active nitrogen. The restoration of the afterglow by addition of tributary stream of oxygen can be observed even better with the electrodeless discharge at low pressure (0.3 mm. say) than with high-pressure electrical discharges at say 3 mm. pressure. The intensity in some experiments has been raised 32-fold by the admission of oxygen. Under certain conditions of the glass the cutting off of the oxygen tributary causes a temporary fall of intensity to considerably less than the static value, followed by a rise to that static value. This shows that the phenomena are complicated and not likely to admit of any very simple analysis. The paper goes on to examine how various treatments of a glass vessel affect its action on the afterglow. As found by Herzberg strong preliminary heating in vacuo makes it destructive of the glow. It is now shown that heating in nitrogen at atmospheric pressure does the same. Heating in oxygen, even at 1 mm. pressure, restores it. It seems clear that these effects cannot easily be explained by the formation or removal of gas layers. The behaviour of glass is therefore very complex, and not likely to be easily unravelled. The behaviour of the gas itself away from any surface is more fundamental. By a special device it is shown that in this case the purest nitrogen is the best, and that the addition of a trace of oxygen has no favourable effect in promoting the active nitrogen phenomena.
THE REACTIONS OF ACTIVE NITROGEN WITH CHLOROMETHANES