I next examined two vacuum-tubes of an entirely different character. The one was a tube from Geissler of stout glass, some fifteen inches long, without electrodes, and an inch across. Within this tube was a second of uranium glass, with bulbs blown in it. In contact with both tubes a quantity of fluid mercury ran loose (Plate X. fig. 7). Upon shaking this tube with the hand brilliant flashes of blue-white light, like summer lightning, flashed out. These were discernible (though faintly) even in daylight. The fine terminal wires of the coil being wrapped round each end of this tube, when the current passed, a bright and white induced discharge, with a considerable amount of stratification, was seen in the tube. The other tube was that of a mercurial siphon-barometer. This being placed in a stand, one terminal wire was placed in the mercury in the short leg of the siphon, while the other terminal was made into a little coil and placed on the upper closed extremity of the barometer-tube. On passing the current, the entire short space above the mercury was filled with a grey-white light, not stratified, but showing a conspicuous bright ring just above the level of the mercury. [Sidenote: Spectrum of both these tubes described.] Both these tubes, when examined with the spectroscope, showed four bright rather uniform bands (the central one being the brightest), which I assigned to the carbon-spectra (see Plate XIV. spectra 5 and 6). The Geissler tube was probably filled designedly with coal-gas. In the case of the barometer-tube the spectrum must be assumed to be the result of some carbon impurity. No lines of mercury could be detected in either case. An effort was made to examine the light of the Geissler mercury-tube as excited by motion only, but the spectrum could not be kept in the field; the four lines were, however, seen to flash out as the light passed before the slit. [Illustration: Plate XV.] _Air-tubes._ [Sidenote: Air-tube illuminated.] The first tube I examined was an ordinary Geissler tube charged with rarefied air. The bulbs, on passing the discharge, were filled with the well-known rose-tinged light like to the Aurora-streams. This in the capillary part was condensed into a brighter and whiter thread, while the platinum wire of the negative pole was surrounded by its characteristic mauve or violet glow. [Sidenote: Spectrum described.] The spectrum, even with a weak current, was quite bright, and consisted mainly of the nitrogen-lines and bands, with the lines Hα, Hβ, and Hγ, and some of the intermediate lines of the H tube. The double line α was undoubtedly the brightest in the spectrum when taken in the capillary part of the tube. After this followed β, and then γ(H), δ, and ε. I was, however, uncertain as to the relative brightness of the last three, and marked their intensities with hesitation. I tested them several times independently with differing results, and suspected them of variability with the current.