Ångström and Herschel suggest its presence in the Aurora in connexion with phosphorescence or fluorescence. With a spark-discharge in air at ordinary pressure, a mixed spectrum of bright lines of N and O is found; while in the case of Geissler vacuum-tubes (representing a glow-discharge in a much more rarefied atmosphere) the N lines appear mainly to usurp the spectrum. [Sidenote: H₂O tube referred to.] It must, however, be borne in mind that a Geissler tube, as to temperature at least, in no way represents the conditions of the Aurora; and when we remember the association of oxygen and ozone, and the way in which the latter is affected by heat, it may well be that temperature plays an important part in the matter. In proof of this conduct of oxygen, it may be cited that, in the case of a H₂O tube, the H lines come out sharp and brilliant in the spectrum, while what is seen of the O lines is comparatively weak, misty, and ill-defined. Vogel, it will be remembered, makes 5189 of the Aurora coincident with an O line. [Sidenote: Residual phosphorescence in Geissler tubes. Garland tube.] Professor Herschel has pointed out, and I have confirmed, that the residual phosphorescence in Geissler tubes, after the spark has passed, is probably associated with oxygen. He also alludes to the fact that when one of the globes of a “Garland” tube was heated, it did not shine after the spark had passed, apparently because of the destruction of the ozone by heat. [Some experiments with a tube of this description will be found detailed in Part III. Oxygen was not, I think, the gas it was filled with.] [Sidenote: Dr. Schuster’s tubes described.] Subsequently to my examination and comparison of the O and CO₂ spectra before detailed, Dr. Arthur Schuster was good enough to send me three vacuum-tubes of his own preparation, showing an oxygen-spectrum. One, with large disk-shaped brass electrodes, was unfortunately broken in transit. Dr. Schuster informed me it showed the carbonic-oxide spectrum as well as that of oxygen. The other two tubes had aluminium electrodes. They were similar in shape to ordinary Geissler tubes, but had attached to each a supplemental bulb containing dry oxide of manganese. Illuminated by the larger coil, one of these tubes (which had a slight crack in the manganese bulb) lighted up faintly; the other was fairly bright, and the glow had a somewhat reddish tint. Plate XVIII. fig. 15 represents as the upper spectrum Vogel’s Aurora, with W.L. numbers, as the middle spectrum the capillary part of Dr. Schuster’s O tube, and as the lower spectrum the negative (violet) pole of the same tube. [Sidenote: Spectra described.] The tube-spectra were mapped out with the aid of the diaphragm micrometer before described. [Sidenote: Capillary.]