It remains that I should make a few remarks on the evidence which that wonderful instrument of research, the spectroscope, has afforded respecting the light of the aurora. Angström was the first to observe the spectrum of the aurora borealis. He found that the greater part of the auroral light, as observed in 1867, was of one colour, yellow, but three faint bands of green and greenish blue colour were also seen. The aurora of April 15, 1869, was seen under very favourable conditions in America. Prof. Winlock, observing it at New York, found its spectrum to consist of five bright lines, of which the brightest was the yellow line just mentioned. One of the others seems to agree very nearly, if not exactly, in position with a green line, which is the most conspicuous feature of the spectrum of the solar corona. During the aurora of October 6, 1869, Flögel noticed the strong yellow line and a faint green band. Schmidt, on April 5, 1870, made a similar observation. He saw the strong yellow line, and from it there extended towards the violet end of the spectrum a faint greenish band, which, however, at times showed three defined lines, fainter, than the yellow line. It was not till the magnificent aurora of October 24, 25, 1870, that any red lines were seen in the spectrum of an aurora. On that occasion the background of auroral light was ruddy, and on the ruddy background there were seen three deep red streamers very well defined. The ruddy streamers, on the night of October 25, converged towards the auroral crown, which was on that occasion singularly well seen. Förster of Berlin failed to see any red line or band despite the marked ruddiness of the auroral light. But Capron at Guildford saw a faint line in the red part of the spectrum; and Elger at Bedford observed a red band in the light of the red streamers, the band disappearing, however, when the spectroscope was directed on the white rays of the aurora. As yet the auroral spectrum has not been interpreted. It is not a spectrum which can be (at present) artificially produced. We understand the spectrum of the sun and stars, because spectra of the same order can be produced in our laboratories. The spectra of the planets, so far as they differ from the spectrum of reflected sunlight in showing signs of the absorptive action of the planetary atmosphere, have been similarly interpreted. So also the spectra of the coloured solar prominences are understood, while those of nebulæ and comets, though not as yet thoroughly explained, have been partly interpreted, because of their partial agreement with the known spectra of earthly elements. But as yet neither the spectrum of the aurora nor that of the solar corona has been explained. The reason probably is, that the conditions under which the light of the aurora as of the corona is formed are not such as have been or perhaps can be attained or even approached in laboratory experiments. [Illustration] VII. _THE LUNAR HALO._