are also compared with other lines in the violet light 4654 and 4601; and the Professor then concludes that it may be in general assumed that the feeble bands of the aurora-spectrum belong to the spectrum of the negative pole, possibly changed more or less by additions from the banded or the line air-spectrum.

4thly. That the only probable explanation of the bright line is that it owes its origin to fluorescence or phosphorescence, the Professor remarking on this point that "an electric discharge may easily be imagined which, though in itself of feeble light, may be rich in ultra-violet light, and therefore in a condition to cause a sufficiently strong fluorescence," remarking also that oxygen and some of its compounds are fluorescent.

spec

5thly. That there is no need, in order to account for the trum of the aurora, to have recourse to the "very great variability of gas-spectra according to the varying circumstances of pressure and temperature" (Dr. Vogel's theory). Professor Angström does not admit such variability, and does not admit that the way a gas may be brought to glow or burn can alter the nature of the

spectrum.

In order to test the Professor's conclusions in an experimental way, I have recently examined some tube and other spectra not only for line-positions, but also for general resemblance to an aurora-spectrum. These experiments, few at first, led to others; and the results, though not perhaps developing any striking or new features, may help to clear the way on some points, and to suggest further experiments bearing on an interesting subject, which up to the present time has been almost entirely in the hands of foreign spectroscopists. It did not seem desirable to use powerful currents: a 4-inch-spark coil worked by a quart bichromatecell was found sufficient to illuminate the tubes steadily. The spectroscope used was one made for me by Mr. Browning specially for auroral purposes; of the direct-vision form, with five large prisms, the collimator and telescope having clear 1-inch lenses.

The field of view extends at one glance from near C to near G, the extremities of the spectrum being obtained by traversing with the telescope; with a fine slit the D line is widely separated. In the eyepiece is inserted a reduced scale (photographed on microscopic glass and mounted in balsam) of bright lines or spaces upon an opaque ground. The faintest spectrum brought in part upon this scale shines through the spaces and renders the divisions visible, showing the spectrum-lines above and touching them. The drawings were made on enlarged photocopies of the microscopic scale itself, so as to avoid any chance of error as between the actual scale and the copy used for the drawings. I found this plan the most effective for rapid and correct sketching; and it gives close results as compared with

other recognized micrometer-measured spectra when they are compared. In most cases the central part of the spectrum only (corresponding to the central lines of the aurora) is mapped, the red line in the aurora not being found to correspond with any prominent line in the gas-spectra examined, and the auroral line near solar G being at present so indefinitely fixed as to render comparison almost valueless. I have selected Dr. Vogel's spectrum as given in his Memoir (English Mechanic,' vol. xviii. No. 461, p. 446) for comparison, it being, so far as I am aware, the most accurately mapped with regard to wave-length at one observation of any auroral spectrum, and it seeming an unsafe plan to attempt to obtain an average aurora by comparison of differing observations made at various times by different observers with all sorts of instruments-the difficulty, too, being increased by the suspicion that the spectrum itself at times varies in number and position, as well as intensity of its lines.

In illustration of the difficulty of constructing any thing like a general typical aurora-spectrum I append a Table of eight auroral spectra taken at hazard.

* Mr. Procter's and Lord Lindsay's lines had no wave-lengths given to them. No doubt some of the discrepancies are instrumental and personal; but there is, I think, still enough remaining of discrepancy to lead to the inference that the aurora prevails in distinct forms or types, and with the fainter lines or bands differing in position. From notes and drawings kindly supplied at times to me by Mr. Backhouse, there is also no doubt that the relative intensity of the lines frequently differs. With a sufficient number of observations these forms or types might be systematically arranged and compared, with a prospect of advancing our knowledge of the nature of aurora.

I shall feel much indebted to any observers who will furnish me, for this particular purpose, with charts of spectra and wave-lengths of lines: these last, however, I can, more or less perfectly, deduce from such drawings, if the positions of a few solar or other well-known lines are marked on the scale.

Dr. Vogel's spectrum does not comprise the line near G; but I have added this (in an approximate place only) in order to complete the set of lines. For drawing of Dr. Vogel's spectrum, with its scales attached, see Plate III.

Hydrogen-Tube.

This tube was one of Geissler's and of rather small calibre. On illumination the wide ends were easily lighted with a silver-grey glow, having a considerable amount of stratification. The capillary part glowed brilliantly with silver-white, bright green, and crimson light according to the intensity of the current; with the commutator slowly working, white running into green and bright green were the main features of the thread of light; on the current passing more rapidly the capillary thread became of an intense crimson, at the same time apparently increasing in diameter, an effect probably due to irradiation.

The spectrum was very brilliant, consisting of the three bright lines usually distinguished as Ha, Hẞ, and Hy, and a number of shaded bands and fainter lines between these, with a bright continuous spectrum as a background to the whole.

The lines Ha, HB, and Hy were found to vary in intensity with the current, and in accordance with the colour of the light as seen by the eye-a fact, as I think, not without bearing on the question of the aurora, the varying tints of which are so well known. The fainter lines or bands were mostly stripes of pretty equal intensity throughout, and all about the width of the HB line. I did not trace any marked degrading on either side of the lines, though the edges were not uniformly so sharp as Ha and HB; some of the lines are coincident in position with lines of the air-spectrum.

One line only actually coincides with the aurora-spectrum, this being that to which Dr. Vogel assigns an identical wavelength, viz. 5189. Other lines, however, fall somewhat near the aurora-lines 5569, 5390, 5233, and 5004, two faint lines also falling within the band 4694 to 4629.

The lines (adopting Dr. Vogel's wave-lengths for the H lines) are, when compared, as under:

{

4694

to 4629

5189 5008

4632

I remarked that a line 5596, described by Dr. Vogel as "very bright" in his H spectrum, does not appear in my tube, though in most other respects our H spectra agree.

I thought this tube afforded a good opportunity for testing the

effect of distance upon the spectrum. The slit was made rather fine. At 6 inches distance from it the line a (see drawing) in the bluegreen (F solar) was very bright. The lines marked B, 7, 8, e, and also survived, but were faint. At 12 inches from the slit α and γ were alone seen; and at 24 inches a stood by itself upou a dark ground. I also noticed that the red and yellow parts. of the spectrum first lost their light on the tube being withdrawn from the slit; and this appeared to account for B disappearing while y survived. For drawing of the hydrogen-spectrum see Plate IV. spectrum 1.

Carbon and Oxygen Tubes.

By way of introduction let me state that I do not here propose to enter into the question now vexed by Professor Piazzi Smyth as to the real character of the carbon-spectrum, and that when I speak of carbon-tubes and spectra I use the words in a general sense, and as meaning those tubes and spectra which have hitherto passed under that designation.

The following tube-observations have been taken together, because my friend, Mr. Henry R. Procter (to whom I am in debt for many profitable hints and suggestions in auroral work) is disposed to regard the spectra found in the carbon-tubes, and in those marked "O," as identical; considering that pure O with the ordinary non-intensified discharge gives only a continuous spectrum, and that the O tubes are in fact generally lighted up by a carbon-spectrum, the result of impurity from accidental causes. The tubes examined for the purpose of comparison were as follows:-A coal-gas tube, a tube marked "C.A.," three O tubes, two of (I believe) London make, and the third from Geissler, and an OH, tube, also from Geissler. The carbon-tubes were both brilliantly and steadily lighted by the current. The C.A. tube glowed with a peculiar silvery grey green light in the capillary part, and with a grey glow considerably stratified in the bulbs. The coal-gas-tube discharge was whiter and still more brilliant than the C.A., and with even finer stratification in the bulbs. The spectra of both tubes were conspicuous for the same three well-known principal bright lines or bands in the yellow, green, and blue (with one fainter in the violet), all shading off towards the violet, and in both cases with fainter intervening bands or lines; and these last bands or lines only partially coincided when the two tubes were compared.

The spectra in both cases were rich and glowing, with a certain amount of continuous spectrum between the lines; and the three principal bands or lines in both cases showed well and distinctly their respective place-colours.

Tested for distance.-In the case of the C.A. tube at 18 inches from the slit the continuous spectrum and fainter lines disappeared, while the four principal lines still shone out, that in the green being the strongest. At 24 inches the same lines were still visible, though somewhat faintly.

In the case of coal-gas, at 24 inches the whole spectrum was quite brilliant, the four principal lines being very bright, and even preserving their distinctive colours. The H line, near the line or band in the blue, was also plainly seen. The O tubes, when treated by the same current as the carbon-tubes, were found to be all three identical in general features. The discharge lighted up each of the tubes feebly and somewhat intermittently. Grey in the bulbs, and a faint but decidedly pinkish white in the capillary part, were the distinguishing light colours, while nothing could be more marked than the difference in brilliancy between these and the preceding carbon-tubes.

The OH, tube presented very much the same character; but the discharge occasionally varied from a pinkish white to a yellow colour, somewhat like what artists call brown pink, reminding one of the "golden rays" in certain auroræ. These O spectra presented, in common with the carbon-tubes, three principal bright lines or bands in the yellow, green, and blue, with a fainter one in the violet, all shading off towards the violet. The bands, however, showed but very little trace of local colour; and the whole spectrum had a faint and washed-out look, very different from the carbon-spectra. (I certainly by a little management subsequently succeeded in getting the same look to the C.A. spectrum; but it was only by removing the tube to some distance from the slit, and thus depriving the spectrum of very much of its brightness.) The hydrogen line (solar F) was bright, more so than any of the O lines. The intensity of the three principal lines seemed to me to run in the following order :

Between the lines y and a in the Geissler O tube I found a rather bright line, which I shall have occasion to refer to hereafter. At 12 inches distance from the slit the O spectrum lost nearly all its light; the H line and the three lines y, a, and ß alone faintly remaining, a being decidedly the brightest. At 24 inches no spectrum at all was to be seen.

I carefully compared together the three principal lines of the two spectra of coal-gas and O by means of :

1st, the photographed micrometer before described;