The spectrum of auroras had attracted the attention of many observers; however, when in 1910, in connection with an investigation of the properties of cosmic rays, I had occasion to acquaint myself with the literature on the spectrum of auroras, I found that almost all the measurements were too inaccurate to be used for the identification of lines and, moreover, that there were as many interpretations of the spectrum as there were observers, if not more. The analysis and interpretation of the auroral spectrum were therefore, at that time, an open question.

In 1912–13 I began observations of the auroral spectrum at Bossekop in Finnmarken1. On this first expedition I used a spectrograph combining high luminosity with considerable dispersion, and in addition I also had a spectroscope at my disposal. With the spectroscope I observed the brightest green line, and a series of the best observations gave a wavelength of \(\lambda = 5577\ \text{Å}\). With the spectrograph I obtained the green line and six lines in the blue and violet part of the spectrum; the latter proved to coincide with strong lines of the negative band spectrum of nitrogen2. Recently this result has been confirmed by Lord Rayleigh (jun.). Various determinations of the green line showed some fluctuations, and therefore I undertook new measurements in Christiania with a spectroscope of greater dispersion. The first series of observations gave \(\lambda = 5578.4\ \text{Å}\); the second, most extensive series gave \(\lambda = 5577.6\ \text{Å}\). These values agree well with the data obtained by Slipher and, recently, by Babcock at the Mount Wilson Observatory in photographing the spectrum of the night sky3.

Thus the wavelength of the green line was established with considerable accuracy; however, its origin remained as mysterious as before. In order to attack the problem of the origin of the green line, I again began systematic investigations of the auroral spectrum. Suitable spectrographs were constructed and mounted on the roof of the Geophysical Institute in Tromsø. Observations were carried out in the winter of 1922–23 and in the last winter.

The work of the first winter yielded many good spectrograms; 35 lines and bands were measured in the visible and ultraviolet part of the spectrum. With the exception of the green line and three weak lines or bands, the remaining lines could be identified with known nitrogen lines. The auroral spectrum is remarkable not only for its green line; the rest of the spectrum is also of an extremely distinctive type.

With the exception of a huge number of lines and bands of the nitrogen spectrum, it is possible to distinguish only a very small number of lines or band limits (band-heads) of other origin.

Although, in photographing, the strong lines of the auroral spectrum were considerably overexposed, nevertheless no traces whatsoever of hydrogen or helium were found. Therefore one must abandon the previously generally accepted view that, at altitudes above 100 km, the atmosphere consists chiefly of light gases. It may, to be sure, be supposed that hydrogen, helium, or the hypothetical “geocoronium” produces the green line; however, it is surprising that neither hydrogen nor helium reveals its ordinary lines. A comparison of the intensity of the green line with the intensity of the known nitrogen lines of the aurora on spectrograms obtained from the lower and upper edges of the aurora showed that the green line cannot arise from the indicated light gases; instead of increasing from below upward, the green line at the bottom proved to be relatively somewhat brighter than at the upper edge of the aurora. This circumstance, in connection with the fact that almost the entire spectrum consists of nitrogen lines, gives strong support to the view, expressed by me as early as 1910, that the green line also belongs to nitrogen and is emitted under the special conditions existing in the region where auroras arise.

Independently of the explanation of the origin of the green line, we must regard nitrogen as the principal constituent of the atmosphere at its upper boundary. To explain this, one must suppose either an increase of temperature with height, or else assume that nitrogen is electrically charged and is drawn upward by electrical forces. The character of the spectrum and other reasons compelled me to reject the first supposition. Analyzing the second supposition, I found that an electrified atmosphere in an extremely ionized state cannot exist in the form of an ordinary gas. Thus I arrived at the supposition that nitrogen at very low temperatures condenses into large or small crystals.

This hypothesis, somewhat astonishing at first glance and contradicting generally accepted conceptions, proved, however, to be very fruitful. First of all, it assumes for the auroral zone such a physical state as explains why the auroral spectrum has hitherto not been successfully reproduced in laboratory experiments; it also makes it possible to explain the abrupt changes in the color of auroras and other cosmic phenomena—the twinkling of stars, the blue color of the sky, and the zodiacal light¹).

¹ Cf. L. Vegard, Zeitschr. f. Physik 10, p. 367. 1923. This work, containing the substantiation of Vegard’s hypothesis, is reported in detail in the following note by L. I. Vasil’skaya (p. 210). —Ed.

From this hypothesis it follows that the spectrum of the aurora is produced when an atmosphere consisting of nitrogen crystals is bombarded by electric rays. These conditions are very difficult to reproduce exactly; however, they can be obtained approximately by bombarding solid nitrogen with electric rays produced in the laboratory. Thanks to the kindness of Prof. Kamerlingh-Onnes, I was able to carry out experiments of this kind in the cryogenic laboratory at Leiden. I succeeded in carrying out the first series of experiments with the light obtained by bombarding solid nitrogen with cathode rays. These experiments gave the very clear effect which I had expected1.

The rays from a Wehnelt cathode fell upon a layer of solid nitrogen formed on a copper surface cooled by liquid hydrogen. I took five spectra (Fig. 1), corresponding to different velocities of the cathode rays. At a voltage of 75 volts only the brightest edges of the negative band spectrum were observed. At 200 volts (Fig. 1, No. 1), a green line near the blue part (\(N_2\)) also appeared, as well as a blurred band near the yellow part.

When the potential was raised to 500 volts, the layer of nitrogen became brightly shining green and, in addition to \(N_2\), there now appeared a broad line or narrow band \(N_1\) of great intensity (Fig. 1, No. 2). When the potential was lowered to 350 volts, \(N_1\) disappeared (Fig. 1, No. 3), but when it was raised to 700 volts (Fig. 1, No. 4) or 750 volts (No. 5), the intensity of the glow, and especially of the line \(N_1\), increased extraordinarily2. The line \(N_1\) now dominated the spectrum just as the green line (5577) dominates the spectrum of the aurora.

In order to form an idea of the relative intensity of the glow at different potentials, it should be noted that in Fig. 1 the exposure times of the spectra are respectively 30, 15, 10, 5, and 5 minutes.

When the bombardment by cathode rays was stopped, the solid nitrogen remained luminous for more than five minutes; in this connection it is interesting to point out that an afterglow of approximately the same duration is also observed in auroras. Measurements showed that the green line of the aurora (5577) is situated in the broad line \(N_1\). The line \(N_2\), whose wavelength proved to be about 5230 Å, coincides with the second weak green line of the auroral spectrum.

TABLE 1.

Fig. 1.

Visible labels in the figure:

Visible wavelength labels above the spectra: 5577, 5330, 4708, 4276, 4058, 3998, 3914, 3705, 3156.

Fig. 2.

Visible labels at right, from top to bottom:

Visible wavelength labels above the spectra: 5577, 5330, 4708, 4344, 4278, 4058, 3998, 3914, 3890, 3750, 3577, 3536, 3371, 3180, 3155.

In Fig. 2 are reproduced three spectra of auroras (Nos. 1, 2, 3) and one spectrum of solid hydrogen (No. 4). No. 1 was obtained with a large quartz spectrograph. The others are enlarged copies of spectrograms taken with a small glass spectrograph; Nos. 1 and 2 were taken on Imperial Eclipse plates, No. 3 on a panchromatic plate, and No. 4 on an orthochromatic plate.

The coincidence of the spectra of auroras and of solid nitrogen is not confined to the green part; in the blue and violet regions crystalline nitrogen reduces the number of lines, which is highly characteristic of the auroral spectrum.

The comparison shows that the typical spectrum of auroras is emitted by solid nitrogen, and thus my hypothesis concerning the structure of the upper layers of the atmosphere has been confirmed.

The considerable velocity of the cathode rays needed to obtain the lines \(N_1\) and \(N_2\), and the very great variability of their intensity, are facts very remarkable from the physical point of view; this variability may help in understanding many changes of color in auroras. The afterglow shows that the new luminous effect is a kind of phosphorescence, but to elucidate its physical nature more experimental material is needed.

The new effect opens up a new field of research, one interesting both from the cosmic point of view and from the purely physical one. A more detailed study of the light emitted by solid nitrogen at various temperatures, down to the temperature of liquid helium inclusive, may make it possible to determine more accurately the temperature interval in the auroral zone; and it is possible that at an extremely low temperature the broad line \(N_1\) will become narrower. Experiments with rays of various velocities and with various carriers may provide information on the velocities and the physical nature of the cosmic rays that produce auroras.

If other gases as well—i.e., hydrogen, oxygen, argon, neon, ammonia, oxides of nitrogen, carbon monoxide, cyanogen, etc.—when brought into the solid state, give under the action of electric rays a luminous effect similar to that of nitrogen, then we may hope in this way to explain the lines of nebulae and to obtain definite information concerning the structure of nebulae. Perhaps “nebulium” awaits the fate that befell “geocoronium.”

I am continuing experiments in the directions indicated here in the Kamerlingh Onnes laboratory in Leiden.

Translated by S. Vasiliev.

1. A fuller description of the experiment will appear in Proc. of the Royal Acad. of Sciences. Amsterdam. ↩↩

2. To identify the lines, a spectrum of ionized neon has been placed in the figure beneath the spectra of solid nitrogen. Translator’s note. ↩↩

3. By means of the Fabry and Perot interferometer. (Astrophys. Journ., 1923.) Translator’s note. ↩