STUDY OF INFRARED RADIATION OF THE NIGHT SKY*

V. I. Krasovskii

Nothing characterizes the properties and processes of the upper layers of the Earth’s atmosphere so well as their own radiation. In the daytime, however, it is masked by scattered sunlight and is detected so far only at night, for which reason it has received the name “radiation of the night sky.” Until recently, spectra of such radiation were known from 3000 to 7000–8000 Å. They were exhausted by a small number of low-intensity lines or bands of atoms and molecules of oxygen and nitrogen. In recent years a more powerful infrared radiation of the night sky has also been discovered and studied. With its discovery people began to speak of “weak” and “strong” radiation, but even the latter is in essence very low in intensity. High-speed spectrographs make it possible to obtain the spectrum of the night sky in the ultraviolet and visible regions of the spectrum, at relatively small dispersion (1000 Å/mm), only as the result of many-hour exposures. Here a brief summary is presented of studies of night-sky radiation in the near infrared region of the spectrum up to the long-wavelength limit of 12,000 Å. Although these studies concerned mainly the testing of new methods of spectroscopy with the aid of an electro-optical converter ¹, ², ⁶, ⁸, ¹⁰ and did not have as their aim a complete and regular observation of the infrared radiation of the night sky, they nevertheless are of interest for judgments about this radiation, especially beyond the long-wavelength limit 8000–9000 Å, generally not surpassed abroad. The accumulated material is not uniform in quality. Each year the apparatus underwent substantial improvements. In 1948 the first studies were carried out with ground-based instruments with a dispersion of about

* Report delivered at a meeting of the Division of Physical and Mathematical Sciences of the Academy of Sciences of the USSR in Moscow on March 19, 1952.

7500 Å/mm and a resolving power of up to 300 Å. In 1950 spectrographs with diffraction gratings were used. They gave a dispersion of up to 150 Å/mm and a resolving power of up to 5 Å. This apparatus surpassed the best foreign equipment even beyond the long-wavelength limit of 8000–9000 Å[^18]. At first, when spectra were photographed with low dispersion, the exposures reached several hours. But even in the transition in the latest investigations to high dispersion, the exposures did not increase, and at 800 Å/mm even decreased to ten minutes. It should also be mentioned that the improvement of the apparatus ensured an increase in the blackening of the photographs.

The new means were first tested on the radiation of the night sky in 1948, at a time when the supposition of the American investigators Stebbins, Whitford, and Swings was widely accepted: that the infrared radiation of the night sky is more or less monochromatic and concentrated around 10,440 Å. At that time there was debate as to whether it belonged to molecular or to atomic nitrogen[^14]. If such radiation belonged to atomic nitrogen, it would have to be most nearly monochromatic. In that case, when photographing the night sky in infrared rays through a Fabry–Perot etalon, interference rings analogous to those observed in the radiation of the well-known green line of the night-sky glow at 5577 Å were expected. Such photographing was carried out. But no interference rings in the radiation of the night sky near 10,440 Å were found. It became clear that the infrared radiation is not strictly monochromatic and therefore cannot belong in significant part to atomic nitrogen. Soon spectra of this radiation were obtained with a dispersion of about 7500 Å/mm. They showed that the radiation is distributed over the entire spectrum from 7000 to 11,000 Å. At 10,440 Å no maximum of radiation whatsoever was found. Thus another supposition of the American investigators was disproved: that the infrared radiation of the night sky belonged wholly to the 0–0 band of the first positive group of molecular nitrogen.

A general idea of the intensity distribution of the infrared radiation of the night sky over the spectrum is given by Fig. 1 (see insert). It reproduces a reproduction from photographs of spectra obtained as early as 1949 with a dispersion of 1200 Å/mm and a resolving power of up to 60 Å near 8400 Å, and with a dispersion of 2400 Å/mm and a resolving power of up to 120 Å near 10,800 Å[^3][^4]. This reproduction gives a clear idea of the most intense parts of the spectrum. In Fig. 2 (see insert) the true distribution of the radiation over the spectrum is shown in

Fig. 1.

Fig. 2.

Fig. 3.

Fig. 4a.

Fig. 4b.

Fig. 4c.

To the article by V. I. Krasovskii

Fig. 5.

Fig. 6.

to the article by V. I. Krasovskii

relative units, obtained as a result of processing the upper photograph just shown, with allowance for the spectral sensitivity of the apparatus. The intensity curve is reproduced without taking secondary details into account. The part of the curve indicated by a dashed line belongs to the region of reduced sensitivity of the apparatus, and its high accuracy cannot be vouched for. The intensity-distribution curve shows that the radiation practically fills the whole portion of the spectrum under consideration. The observed maxima are relatively small. It should be noted that a quite distinct maximum is present even in the region of intense absorption by water vapor near 9400 Å. Earlier fears that the continuous filling of the whole spectrum with radiation was a consequence of insufficient resolving power of the apparatus are thus removed. Spectrograms have now been obtained with considerably greater dispersion and resolving power, on which radiation has indeed been found in the regions corresponding to the minima of the intensity curve just shown.

In Fig. 3 (see insert) reproductions are shown from photographs of spectra with somewhat greater dispersion[^10]. The spectrograph used to obtain them, with a diffraction grating, had a dispersion of 850 Å/mm and a resolving power up to 30 Å. Various reproductions are mounted in parallel. Of these, a and v correspond to one and the same original, obtained with an exposure of 15 min. Section a was printed on photographic paper with overexposure, and section v with underexposure. Reproduction b was made from another original, also exposed for 15 min. However, in order to print a broader portion of the spectrum, the print on photographic paper was made with illumination smoothly decreasing toward the long waves. It should be noted that on the reproduction on the short-wave side there is a diffuse band absent from photograph a. Reproduction g was made from a negative obtained with an exposure of 45 min. The first dark band on the short-wave side may be caused by absorption by water vapor. The spectrum with dispersion 850 Å/mm is in satisfactory agreement with the spectrum of smaller dispersion previously obtained in 1949. Study of the new spectrum, as of the one shown above, made it possible to establish that the intensity minima are not deep and are filled with diffuse radiation, although uneven over the spectrum.

In Fig. 4 (see insert) reproductions are presented from several other photographs of the infrared spectra of the radiation of the night sky[^11],[^12]. They were obtained by means of a spectrograph with a diffraction grating, at a dispersion of 175 Å/mm and a resolving power up to 5 Å. The upper and lower reproductions are duplicated by prints of different density. In Fig. 5 (see insert)

shown are reproductions from various photographs of the same portion of the spectrum[^10][^11][^12]. They were taken with the same spectrograph, but at different times. The two middle reproductions are prints of different density from one and the same original. There are grounds to suppose that the spectral composition of the radiation of the night sky is subject to certain changes. It should also be emphasized that even with large dispersion all the minima of the radiation are shallow. The diffuse radiation is relatively intense, although uneven across the spectrum. Whether, at still greater dispersion, it can be resolved into individual details is unknown. This diffuse radiation and the closely spaced lines or bands that form blends make it difficult to determine the true intensity of the individual elements of the infrared spectrum of the night sky.

From the exposure times used in photographing, as a result of three to four years of observations, it has been established that quite often the intensity of the infrared radiation at midnight is about twice as great as in the evening or in the morning. This result confirms an analogous one obtained earlier and independently by S. F. Rodionov with the aid of an electrophotometer[^15]. From the exposure times used in photographing it was also found that the intensity of the radiation under study is higher at the beginning of the year than in summer. There are indications that in the summer of 1950 this radiation became two to three times weaker than in the summer of 1948. At times the intensity is so weak that photographing the spectra becomes impossible. Such conditions, for example, occurred in the Crimea in August 1950. Although all the investigations mentioned here were deliberately carried out in clear weather, it is nevertheless difficult to assert that meteorological conditions, not subject to exhaustive accounting at the time of photographing at night, have absolutely no bearing on the fluctuations of intensity.

A rough estimate of the absolute intensity of all the infrared radiation of the night sky recorded at the Earth’s surface in the region from 7000 to 11,000 Å exceeds, by approximately one hundred to two hundred times, the intensity of the well-known green line of the night-sky glow at 5577 Å. Such an estimate is in good agreement with the result of S. F. Rodionov, obtained independently with the aid of an electrophotometer[^15].

In the ultraviolet and visible regions of the night-sky spectrum, radiation of atomic nitrogen has not been reliably detected, while molecular nitrogen is represented by the Vegard–Kaplan band system of low intensity and by very weak bands of the first positive group, whose presence is often disputed[^14]. In the infrared radiation of the night sky it has not been possible to detect emission of atomic and molecular nitrogen. This, however, does not mean that radiation of atomic and molecular nitrogen is absent—

exists completely. It is possible that it is only masked by considerably more powerful radiation of another origin.

In the infrared part of the spectrum, radiation from atomic oxygen has also not yet been detected; in the shorter-wavelength region it is represented by its most intense known green and red lines.

In the ultraviolet region of the spectrum, relatively weak Herzberg bands belong to molecular oxygen[^17]. A characteristic feature of this band system is the distribution of maximum intensities within it. The most intense bands correspond to transitions from high vibrational levels close to the dissociation threshold. Bands corresponding to transitions from the zero and nearby vibrational levels are either weak or absent.

In the new infrared region of the spectrum, the molecular oxygen band 0—1 from the transition from the state \({}^{1}\Sigma\) to the ground state \({}^{3}\Sigma\) has been identified with complete certainty. At our largest dispersions this band splits into two separate, but structureless, branches. Apparently it is quite significant that the indicated band does not have constant intensity. In Fig. 1 are assembled reproductions from spectra of the night sky, taken in the same way with one and the same apparatus (dispersion about \(1200\ \text{\AA}/\mathrm{mm}\)). The photographs, however, were obtained at different geographical locations and at different times of the year. The \(O_2({}^{1}\Sigma—{}^{3}\Sigma)\) band is found only on our summer photographs at Simeiz (upper reproduction) and is absent from winter photographs near Moscow (lower reproduction). In Fig. 6 are assembled reproductions of spectra: at the top—from a photograph obtained by Meinel[^18] at the Lick Observatory in the winter of 1949/50, and at the bottom—from one of our photographs taken in winter near Moscow. The \(O_2({}^{1}\Sigma—{}^{3}\Sigma)\) band present on Meinel’s photograph is absent from ours. The relative intensity of this band is especially large on the spectrogram obtained by Dufay[^21] in the south of France in 1950. It is also of interest to note that the molecular oxygen band \({}^{1}\Sigma—{}^{3}\Sigma\) is the first detected emission band of the night sky with a varying contour, apparently indicating a nonconstant rotational temperature of the emitting molecules.

Although the mentioned band of molecular oxygen is absent in a number of cases, nevertheless its maximum intensity sometimes reaches 10—20 intensities of the known red emission line of the night sky at \(6300\ \text{\AA}\). From the intensity of this band one can calculate the intensity of another, still stronger 0—0 band of the same system, formed in transitions from the same initial level. Recent laboratory investigations by Kwiatkowski[^23] have established that the intensity of the 0—0 band is fifty times greater than that of the 0—1 band. Thus

Thus, the intensity of the 0—0 band may exceed the intensity of the well-known red emission line of the night sky at 6300 Å by 500–1000 times. This conclusion is very significant, since it indicates that the true emission of molecular oxygen in the upper layers of the Earth’s atmosphere belongs to the most powerful radiations in the near infrared region of the spectrum, up to 12,000 Å. The non-observability of this radiation, concentrated at 7600 Å, is very simply explained by its complete absorption by the lower layers of the atmosphere. As is known, the 0—0 band of molecular oxygen in the transition from the ground state \({}^{3}\Sigma\) to the excited \({}^{1}\Sigma\) is the most intense absorption band of the atmosphere.

Identifying the infrared spectrum of the night sky obtained by us with small dispersion, Dufay\(^{23}\) assumed that the powerful radiation at 10,000 Å was due to the 0—2 band of the same band system of molecular oxygen. However, in our spectra taken with large dispersion, at the place indicated by Dufay, among the most intense details no branch of this band is found. Its absence agrees well with the laboratory investigations of Kwiat, who did not detect even traces of such a band\(^{23}\).

It is very interesting that, in the radiation of the night sky, bands of molecular oxygen \({}^{1}\Sigma - {}^{3}\Sigma\), which would correspond to transitions from the first and other higher vibrational levels, are not observed. It should also be recalled that in the laboratory Kwiat observed only one band, 1—1, with an intensity two hundred times smaller than that of the 0—0 band\(^{23}\).

Somewhat earlier we put forward the supposition\(^{5}\) that molecular oxygen in the upper layers of the Earth’s atmosphere may also emit bands corresponding to transitions from the \({}^{1}\Delta\) state to the ground state \({}^{3}\Sigma\). The presumed position of these bands falls within the spectral region under investigation. Among the many intense details of the infrared radiation of the night sky which as yet cannot be identified, or can be identified only with difficulty, one might have suspected some to be the bands of \(\mathrm{O}_{2}({}^{1}\Delta - {}^{3}\Sigma)\), corresponding to the most probable, according to the Franck–Condon principle, transitions from high vibrational levels. However, the latest experimental data on the exceptional population of the zero vibrational level of the \({}^{1}\Sigma\) state of molecular oxygen compel one to expect a similar population also in its more easily excited and more metastable \({}^{1}\Delta\) state. If this latter supposition is indeed correct, then in the spectral region studied only one band of this system, 1—0, near 10,700 Å, has a chance of being detected. Unfortunately, clear photographs of the night-sky spectrum in this region are not yet available. But the existing photographs (see Figs. 3 and 4) show that near

10 700 Å there is an intense broad defocused band, which we had earlier taken to be the \(R\)-branch of the hydroxyl band \(5—2\). Even then, however, independently of the supposition about the bands of \(O_2({}^1\Delta—{}^3\Sigma)\), it was already noted that this band is too intense and broad for it to be attributable only to a single \(R\)-branch of the hydroxyl band \(^{13}\). Observing the utmost caution, one may say that the available factual material does not yet contradict the supposition of the presence of the band \(1—0\ O_2({}^1\Delta—{}^3\Sigma)\). If such a supposition were to prove correct, the existence of another, much more intense band of the same \(0—0\) system, situated near 12 500 Å, would thereby be demonstrated. Thus it is possible that the bands of molecular oxygen may predominate in the infrared radiation of the upper layers of the Earth’s atmosphere up to 13 000 Å.

In spectra of the night sky down to 8500 Å obtained abroad by the ordinary photographic method, Meinel \(^{18}\) identified rotational-vibrational bands of hydroxyl in the ground state. This identification was developed further by I. S. Shklovskii \(^{16}\). It was assumed that practically all the infrared radiation of the night sky is exhausted by hydroxyl radiation. However, our material, it is true covering only the near infrared part of the spectrum, apparently still gives no grounds for such unambiguous conclusions. In Fig. 6 Meinel’s spectrum is compared with ours. In the center are placed prints of different density from our original. At the edges are mounted reproductions from Meinel’s spectrum. As has already been noted \(^{12}\), a number of details of Meinel’s spectrum and ours coincide, some things are absent in ours, but, most importantly, there is much that is absent in Meinel’s, as well as considerable nonuniform diffuse radiation. According to Meinel’s materials, the intensity of the \(Q\)-branch of the hydroxyl band amounts to about one quarter of the intensity of the whole band. Proceeding from this, our photograph makes it possible to estimate the intensity of the hydroxyl bands in the comparable portion of the spectrum. If all the radiation in the region under consideration is taken into account, including the diffuse radiation not resolved into details, it turns out that the total intensity of the hydroxyl bands \(6—2\) and \(6—3\) not only by no means exhausts all this radiation, but is not even its predominant part. Such is the state of affairs in that region of the spectrum where it is possible to compare our results with the most popular foreign ones.

Turning to the longer-wavelength region extending beyond 10 000 Å, one is inevitably compelled to confine oneself to a presentation of our results alone \(^{11}\). In the middle part of the diagram of Fig. 7 only some of the clearest details of the night-sky spectrum in this region are shown. The diagram is compiled on a scale of \(\mathrm{cm}^{-1}\).

The designations of lines or bands indicate the width and the reliability. Pairs of lines with hatched intervals are, possibly, a single formation, since separately they are distinguished with great difficulty. The broad diffuse band, located approximately between 9650–9750 Å, is denoted by one line corresponding to the place of maximum intensity at 9712 Å. In the upper part of the diagram in Fig. 7 some characteristic positions of the rotational-vibrational bands of hydroxyl in the ground state, calculated from Meinel’s data, are indicated. The marked lines of the \(R\)- and \(P\)-branches correspond to the imaginary state \({}^{1}\Pi\). For simplicity the doublet structure has not been taken into account. The positions of the true, most intense \(P\)-lines of OH \({}^{3}\Pi_{3/2}\) are found, on the long-wave side, from the marks (\(\sim\) by \(10\ \mathrm{cm}^{-1}\)).

Fig. 7.

So far a satisfactory agreement, apparently, is found only for the bands 4—1 and 9—5, but even this, with the exception of the \(Q\)-branches, leaves much to be desired. The identification of the band 3—0 is completely unsatisfactory. Thus, for example, its obligatory intense \(Q\)-branch is absent. If the line 10288 Å is taken as the \(Q\)-branch of the band 4—1, and the line 10013 Å as the \(Q\)-branch of the band 9—5, then the total intensity of the two indicated branches, instead of the necessary \(1/4\), will turn out to be approximately equal to no more than \(1/10\) of the full radiation intensity at the positions of the OH bands 4—1 and 9—5. Evidently, radiation not connected with hydroxyl is superposed on these bands. The intensity of the OH band 3—0, most likely, is substantially less than the intensity of the bands 4—1 and 9—5 and does not exceed the intensity of the background.

Although the rotational-vibrational bands of hydroxyl practically do not exhaust all the infrared radiation of the night sky, nevertheless the presence of some of them has now undoubtedly been established also in the longer-wave region, not investigated by foreign authors. All discovered hydroxyl bands belong only to the —4 and —3 sequences. This means that the presence of all the remaining more intense-

of the bands of the smaller sequences. I. S. Shklovsky, using Scholz’s formula, calculated the relative intensities of the bands of various sequences for many initial vibrational levels of hydroxyl^16. Thus it is now possible to estimate the intensity of all the remaining, longer-wavelength bands from the intensities of known ones. A check of Scholz’s formula on the intensities of these bands indicates a discrepancy, by factors of several, between some calculated and observed intensities^9. It is at present difficult to judge whether this is due to the approximate nature of Scholz’s formula, or to inaccuracy in determining the intensities of the known hydroxyl bands, or, finally, to both together. Nevertheless, this does not cast doubt on the order of magnitude of the final result. The total intensity of all the longer-wavelength hydroxyl bands, inaccessible to the means of modern investigation, proves to exceed the intensity of the known red emission line of the night sky at 6300 Å by thousands or tens of thousands of times^7,^16. This means that the absolute intensity of the long-wavelength hydroxyl radiation is estimated to be of the order of \(1 \div 10\) erg cm\(^{-2}\) sec\(^{-1}\). It is hard to imagine that, in the region of wavelengths longer than 15,000 Å, there could be any electronic spectra of atoms or molecules of comparable power.

Many details of the infrared spectrum of the night sky and the discrepancies in wavelengths remain unexplained. This may be due to radiation of non-hydroxyl origin which has not yet been identified. Blending of elements of the hydroxyl spectrum with extraneous ones may be the cause of the disagreement between the effective wavelengths and intensities and those expected. The absence of an exhaustive identification of the night-sky spectrum makes it advisable to search for other interpretations. But it goes without saying that all this can so far be done only in a working fashion, without any claim to absolute reliability. We shall therefore confine ourselves to the briefest remarks on this subject.

Are there not present in the infrared radiation of the night sky, alongside the hydroxyl bands, bands of another hydride, NH? The rotational-vibrational spectrum of the NH molecule has not yet been observed under laboratory conditions, just as the analogous spectrum of hydroxyl has not. Calculation of the NH spectrum is hampered by the absence of accurate molecular constants for this molecule. However, as a trial it was assumed that they are close to the mean values of the corresponding constants of the CH and OH molecules^11. In the lower part of the diagram in Fig. 7 are indicated some characteristic positions of the rotational-vibrational bands of the NH molecule obtained in this way. The heads of the \(R\)-branches with a red shade are denoted by \(R_k\). Their disagreement with details of the night-sky spectrum is not so great that the presence of NH radiation can be completely neglected. It should

It should be noted that the positions of the assumed bands indicated in the diagram of Fig. 3 belong to sequence —4 and correspond to transitions from high vibrational levels. Other bands of this sequence, in transitions from lower levels, are situated in the shorter-wavelength region of the spectrum down to 8300 Å. However, by analogy with the hydroxyl of the upper layers of the earth’s atmosphere, one may suppose that, within one and the same high sequence, the intensities of the bands rapidly weaken with transition to lower initial levels.

As has already been reported, the entire infrared region of the night-sky spectrum investigated by us, apart from lines and bands, is filled with radiation which is not yet resolved into separate details. It is possible that this radiation is due to some extent to the NO$_2$ molecule. It is of interest to note that similar radiation was recently obtained by Kaplan in the artificial reproduction, under laboratory conditions, of the night-sky spectrum$^{24}$. This author obtained diffuse NO$_2$ bands beginning in the visible region and extending to the long-wavelength limit of 9000 Å, determined by the sensitivity threshold of the photographic material used in photographing the spectrum.

A broad diffuse band with an intensity maximum near 9712 Å, which in the long-wavelength region may partly be attributed to NH radiation, may also belong to solar radiation scattered by the earth’s atmosphere. Spectra of the daytime sky obtained with our spectrograph indicate the existence near 9700 Å of a slight maximum of intensity, monotonically decreasing toward long waves. If this supposition should subsequently prove correct, it will have to be admitted that our spectra were obtained at a time when the infrared radiation of the upper layers of the earth’s atmosphere did not appreciably exceed the solar radiation scattered by the atmosphere.

With this we conclude a very brief review of the present factual material on the infrared radiation of the night sky. There can be no doubt that it characterizes only the initial stage of the assault on the secrets of the radiation of the upper layers of the earth’s atmosphere. Various assumptions require confirmation and refinement, while the factual material, which continues to accumulate without interruption, still continues to pose new problems for solution. Apparently, the infrared radiation of the night sky is not constant in absolute intensity or in spectral composition. It is difficult, however, at present to assert anything more definite about the scale of this variability.

Both here and abroad, the new rich factual material could not but stimulate attempts to explain the mechanism

excitation of the radiation of the night sky. The old hypotheses have been considerably improved, and new ones have appeared. Unfortunately, they are still far from an exhaustive completion. We shall confine ourselves to an account of the most popular assumptions.

The old point of view seems to us the most just, according to which the source of the energy of the radiation of the night sky is the energy of dissociation of molecules\(^{7, 11}\). In the daytime short-wave solar radiation dissociates into atoms the molecules of the upper layers of the earth’s atmosphere. The predominant share of this energy is spent, above all, on the dissociation of the most readily dissociated molecular oxygen. Molecular nitrogen is dissociated only slightly, since its dissociation requires radiation that is shorter-wave and therefore less intense in the solar spectrum. The oxygen atoms accumulated during the day provide the radiation of the night sky at night. The formation of molecules from atoms is most effective in triple collisions. At altitudes above one hundred kilometers, i.e. in the zone of the most effective dissociation of oxygen molecules, each atom or molecule experiences a triple collision only once in several hours. As a result, the concentration of atomic oxygen during the night does not decrease appreciably, and this ensures a more or less constant intensity of the radiation of the night sky. One may attempt to give a rough estimate of the upper limit of the total intensity of this radiation. The dissociation occurring in the daytime is, on the average, always balanced by the association process, which proceeds continuously throughout the whole day. Molecular oxygen is substantially dissociated only by radiation shorter than 1700 Å. A rough estimate of solar radiation shorter than this limit can be obtained if the Sun is taken to be an absolutely black body with a temperature of \(6000^\circ\) K. It is self-evident, of course, that such an assumption can give only the order of magnitude of the quantity of interest to us, since solar radiation may be richer or poorer in ultraviolet than an absolutely black body. The solar constant for radiation shorter than 1700 Å thus proves to be equal to \(700\ \text{erg}\ \text{cm}^{-2}\ \text{sec}^{-1}\). This energy can be expended on the dissociation of oxygen only in part. Some part of the oxygen always associates in the daytime, the rest at night. Thus, the total intensity of the radiation of the night sky may reach several hundreds of \(\text{erg}\ \text{cm}^{-2}\ \text{sec}^{-1}\).

The power of all known and presumed radiation of molecular oxygen and even of hydroxyl is much less than this limiting value. Where, then, does the predominant part of the energy of dissociation disappear? By what channels is it removed from the upper layers of the earth’s atmosphere? It has always seemed strange to us,

why the question of the sources of energy of the night-sky radiation is posed so sharply, and not, conversely, the question of the causes of the excessive weakness of the known night-sky radiation.

According to generally accepted ideas, atomic oxygen can recombine into molecules only as a result of three-body collisions involving a pair of atoms or an atom and an oxygen molecule. As a result of the latter combination ozone is formed, while oxygen molecules appear only as a consequence of the bimolecular reaction of ozone with ozone or with atomic oxygen. One may also indicate another path for the recombination of atomic oxygen into molecules. First, in three-body collisions, nitric oxide appears. Then it enters into a bimolecular reaction with nitric oxide itself or with atomic oxygen, as a result of which molecular oxygen also arises. In the final analysis it is immaterial by what path the oxygen molecules appear. What is essential is only that they are born in excited states, the limit of which is the dissociation threshold, equal to 117 kilocalories per mole. It goes without saying that part of this energy may be transferred to a third body, for example to an oxygen atom. At present we still cannot predict in which excited states the new oxygen molecules are formed. In principle such states may be the metastable electronic states \(F\,{}^{3}\Sigma\), \({}^{1}\Sigma\), and \({}^{1}\Delta\). However, the intensity of the observed bands corresponding to transitions from these states indicates that they do not account for a significant part of the possible energy of the night-sky radiation. It follows from this that the predominant part of the molecular oxygen arises mainly in the ground electronic state \({}^{3}\Sigma\) with intense vibrational excitation. Oxygen molecules in such a state cannot be deactivated by radiation. As for the newly formed molecules of ozone and nitric oxide, then on energetic grounds they are possible only in ground electronic states at high vibrational levels. Fortunately, for what follows it is practically immaterial what the electronic state of the newly formed excited molecules is.

Previously it was usually assumed that the excited oxygen molecules newly formed in the upper layers of the Earth’s atmosphere are chemically stable and lose their excitation only through radiation or in repeated collisions. In the new point of view, the essential assertion is that atoms and molecules appearing in an excited metastable or nearly metastable state are so chemically active that in most cases they disappear by reacting with unexcited atoms and molecules, and do not have time to radiate or to lose, in collisions, a significant part of the excitation energy. An analogous

participation befalls all the newly formed excited products. A large number of diverse, rapidly disappearing and reappearing compounds arise. The continuous transformations continue until the excitation energy has been given up by the reaction products through radiation. The inevitability of such a process is clear from comparing the lifetime of metastable states of molecular oxygen, the time of preservation of excitation in collisions, and the time between successive reactions into which atoms and molecules enter with excitation exceeding the activation energy. We give several examples. The mean lifetime of excited molecular oxygen upon deactivation by radiation in the state \({}^{1}\Sigma\) is about 10 seconds; in the state \({}^{1}\Delta\) it is considerably greater than this value, and in the ground electronic state it is practically equal to infinity. In collisions, excited molecules usually transfer only one vibrational quantum, corresponding to a transition between neighboring levels. The probability of this process is the greater the higher the excitation. Nevertheless, on average, its value is characterized by a quantity several orders of magnitude less than unity.

At the same time, the probability of a reaction of excited oxygen molecules, for example, with nitrogen and hydrogen atoms is greater than 0.1. It goes without saying that at an altitude of about 100 km, where the number of binary collisions experienced by an excited molecule has values close to \(10^{5}\), a reaction of this molecule with the few nitrogen and hydrogen atoms, even when their relative concentrations are of the order \(10^{-4}\), will occur sooner than its complete deactivation by collisions or radiation. The resulting chain of chemical reactions is most rapidly broken off by deactivation through radiation of the least metastable product.

The number of intermediate chemical reactions and compounds possible in the upper layers of the earth’s atmosphere is large. As an example we shall consider only some of them, restricting ourselves to those which, in our view, are most essential.

We shall agree to denote an excited product by a prime, and the energy of its excitation by kilocalories per mole, or by the vibrational level of the ground state, if the product is assumed to be in such a state. When writing chemical formulas, these values will be indicated in parentheses after the symbols of atoms and molecules.

As has already been stated, the principal primary excited products can only be molecules of oxygen, ozone, and nitrous oxide. The first and most probable reactions of these excited molecules are their reactions with the most widespread, in the upper layers of the earth’s atmosphere, unexcited

by atoms and molecules of oxygen and molecules of nitrogen:

\[ \mathrm{O}'_2\;(>95)+\mathrm{O}_2\longrightarrow \mathrm{O}'_3+\mathrm{O}, \]

\[ \mathrm{O}'_2\;(>80)+\mathrm{N}_2\longrightarrow \mathrm{N}_2\mathrm{O}'+\mathrm{O}, \]

\[ \mathrm{O}'_3\;(v>1)+\mathrm{O}\longrightarrow \mathrm{O}'_2+\mathrm{O}'_2, \]

\[ \mathrm{N}_2\mathrm{O}'\;(v>1)+\mathrm{O}\longrightarrow \mathrm{N}'_2+\mathrm{O}'_2. \]

All these reactions become possible when the excitation energy of the primary product exceeds the value indicated in parentheses. Even the most cursory examination of the last two reactions shows that excited molecules of ozone and nitrous oxide disappear in their first collisions with oxygen atoms. A characteristic feature of these reactions is the inevitability of the distribution of excitation energy between both final products. In contrast to simple collisions, in this case there occurs a transfer of vibrational energy in amounts exceeding the energy difference of two neighboring vibrational levels. The fate of excited oxygen molecules is different. They become chemically stable with respect to molecular oxygen at excitation energies less than 95 kilocalories per mole, and with respect to molecular nitrogen at excitation energies less than 80 kilocalories per mole.

Thus, conditions are created for the effective accumulation of chemically inert oxygen molecules with excitation energy less than 80 kilocalories per mole and, to a lesser degree, with excitation energy less than 95 kilocalories per mole. Excitation energies exceeding these limits are rapidly redistributed, as a result of the third and fourth reactions, between their two final products.

If the composition of the upper layers of the earth’s atmosphere were exhausted exclusively by oxygen molecules and atoms and nitrogen molecules, then the vibrationally excited oxygen molecules in the ground electronic state could be deactivated only by elastic collisions. For complete deactivation of such molecules, a number of collisions exceeding \(10^5\) would be required. Their fate, however, changes substantially in the presence in the upper layers of the earth’s atmosphere of such atoms and molecules as \(\mathrm{N}\), \(\mathrm{H}\), \(\mathrm{NO}\), and \(\mathrm{CO}\). If the relative concentration of these elements exceeds the value \(10^{-5}\), then the oxygen molecules that have not had time to be deactivated enter with them into any one of the reactions listed below, which are very effective also in the reverse direction. The effectiveness or ineffectiveness of the reverse reaction depends above all on the ability of the molecules \(\mathrm{NO}'\), \(\mathrm{OH}'\), \(\mathrm{NO}'_2\), and \(\mathrm{CO}'_2\) to rapidly

be deactivated by radiation:

\[ \mathrm{O_2'}(\leqslant 95;\ \leqslant 80)+\mathrm{N}\rightleftarrows \mathrm{NO'}+\mathrm{O}, \]

\[ \mathrm{O_2'}(\leqslant 95;\ \leqslant 80)+\mathrm{H}\rightleftarrows \mathrm{OH'}(v\leqslant 9;\ v\leqslant 7)+\mathrm{O}, \]

\[ \mathrm{O_2'}(\leqslant 95;\ \leqslant 80)+\mathrm{NO}\rightleftarrows \mathrm{NO_2'}+\mathrm{O}, \]

\[ \mathrm{O_2'}(\leqslant 95;\ \leqslant 80)+\mathrm{CO}\rightleftarrows \mathrm{CO_2'}+\mathrm{O}. \]

As a result of the indicated reactions, excited molecules of \(\mathrm{NO}\), \(\mathrm{OH}\), \(\mathrm{NO_2}\), and \(\mathrm{CO_2}\) are formed. It is essential that, to one degree or another, these molecules are capable of being deactivated by means of radiation. Let us also note that, as a result of such reactions, not only unexcited but also excited oxygen atoms may appear. The molecules \(\mathrm{NO}\), \(\mathrm{NO_2}\), and \(\mathrm{CO_2}\), apparently, arise, like the already known \(\mathrm{OH}\) molecules, chiefly in an electronic state with vibrational excitation. Unfortunately, the \(\mathrm{NO}\) molecules are insufficiently anharmonic. Their most intense radiation falls in the far long-wave region, inaccessible to the means of modern investigation. Vibrationally excited \(\mathrm{NO_2}\) and \(\mathrm{CO_2}\) molecules can radiate effectively in the same long-wave region. The situation is different for hydroxyl molecules. Owing to the considerable anharmonicity of these molecules, in the spectral region investigated by us it is possible to detect rotational-vibrational bands of high sequences. It is probable that in the high layers of the Earth’s atmosphere the total concentration of \(\mathrm{N}\), \(\mathrm{NO}\), and \(\mathrm{CO}\) is greater than the concentration of \(\mathrm{H}\). In reactions with these particles, a large part of the energy of the primary excitation is intercepted. In such a case the observed and assumed hydroxyl radiations cannot exhaust all the infrared radiation of the night sky. It is very characteristic that the limiting excitation of hydroxyl corresponds to the 9th vibrational level if it is formed from oxygen molecules with a limiting excitation of 95 kilocalories per mole, and to the 7th vibrational level if the limiting excitation of the oxygen molecules is equal to 80 kilocalories per mole. Indeed, the hydroxyl bands observed in the radiation of the night sky are most intense in transitions from the 7th and lower vibrational levels and are practically not observed in transitions from levels exceeding the 9th. Thus, the observed hydroxyl radiations reflect the population of the chemically most stable states of molecular oxygen.

According to the hypothesis we are developing, the intensity of the total hydroxyl radiation depends most substantially on the relative concentration of atomic hydrogen in the mixture of all atoms and molecules capable of entering into reaction with excited oxygen molecules, i.e. \(\mathrm{N}\), \(\mathrm{H}\), \(\mathrm{NO}\), and \(\mathrm{CO}\). The absolute concentration of these elements determines chiefly only the rate

emission of the excitation energy initially associated with oxygen molecules. However, even at insignificant concentrations of N, H, NO, and CO this process takes place in a time negligible in comparison with the duration of the night. A decrease in the absolute concentration of N, H, NO, and CO only increases the role of deactivation by inelastic collisions. In this case the mean vibrational excitation of oxygen molecules decreases owing to an increase in the population of lower levels. As a consequence, the hydroxyl bands due to transitions from lower vibrational levels prove to be more intense than the limiting 9th and 7th levels.

The variations in the intensity of night-sky radiation can now be explained by changes in the relative and absolute concentrations of N, H, NO, and CO. One may also suppose that if, during auroras, the relative concentration of atomic nitrogen increases owing to dissociation by an electric discharge, then the intensity of hydroxyl radiation will weaken. B. A. Bagaryatskii has apparently already observed such a phenomenon[^13].

The predominant mechanism of hydroxyl excitation described by us is possibly not the only one. Thus, hydroxyl may be excited up to the 9th vibrational level in dissociating water vapor by the reaction:

$$ \mathrm{H_2O}+\mathrm{O}({}^{1}S)\to \mathrm{OH}' + \mathrm{OH}'. $$

The state \(\mathrm{O}({}^{1}S)\) is the initial one in the emission of the well-known green night-sky line 5577 Å.

It is not uninteresting to note that, from the energy point of view, a reaction of strongly excited hydroxyl with oxygen and nitrogen molecules is possible:

$$ \mathrm{OH}'(v>9)+\mathrm{O_2}\to \mathrm{O_3}' + \mathrm{H}, $$

$$ \mathrm{OH}'(v>7)+\mathrm{N_2}\to \mathrm{N_2O}' + \mathrm{H}. $$

These reactions make impossible the stable existence of vibrational levels of OH higher than the 7th and, in particular, the 9th.

Thus, the observed weakening of hydroxyl radiation in bands due to transitions from vibrational levels higher than the 7th, and the practically complete absence of bands due to transitions from levels higher than the 9th, is not mysterious and can be satisfactorily explained by several quite different causes.

The obvious presence in the upper layers of the earth’s atmosphere of excited molecules \(\mathrm{N_2}\), NO, and OH makes it appropriate to suppose that they undergo further reactions with atomic hydrogen and nitrogen:

$$ \mathrm{N_2}' + \mathrm{H}\to \mathrm{NH}' + \mathrm{N}, $$

$$ \mathrm{NO}' + \mathrm{H}\to \mathrm{NH}' + \mathrm{O}, $$

$$ \mathrm{OH}' + \mathrm{N}\to \mathrm{NH}' + \mathrm{O}. $$

As a result of these reactions, the appearance of excited NH molecules is inevitable. The assumed intensities of the bands of this molecule are connected first of all with the possible concentrations of hydrogen, nitrogen, and oxygen atoms. However, sufficient data for judgments on this question are at present lacking. Nevertheless, it seems appropriate to draw attention to the possibility of the reaction:

\[ :\mathrm{NH}' + \mathrm{N} \to \mathrm{N}_2' + \mathrm{H}.] \]

If the dissociation energy of molecular nitrogen is taken to be 225 kilocalories per mole, and the dissociation energy of NH to be 90 kilocalories per mole (the mean value of the dissociation energies of CH and OH), then, taking into account the activation energy of this reaction (12 kilocalories per mole), the minimum excitation of the newly formed nitrogen molecule will be equal to 147 kilocalories per mole. In the case of molecular nitrogen in the \(A^3\Sigma\) state, such excitation corresponds, as the lowest, to the 1st vibrational level. May not, therefore, the absence\({}^{14}\) in the Vegard–Kaplan system of intense bands from transitions from the zero vibrational level of the \(A^3\Sigma\) state serve as evidence of the predominant excitation of them at the expense of the reaction of nitrogen atoms with NH? Molecular nitrogen in the \(A^3\Sigma\) state is very metastable. But the zero vibrational level of this state cannot become populated, since the lifetime is short because of the rapid disappearance of excited molecular nitrogen in the reaction

\[ \mathrm{N}_2'(A^3\Sigma) + \mathrm{O} \to \mathrm{NO}' + \mathrm{N}. \]

Something similar, apparently, also takes place for \(\mathrm{HO}_2'(F^3\Sigma)\) because of reaction with N.

As has already been indicated, atomic and molecular oxygen can be excited at the moment of formation of molecules in triple collisions. Excitation is also possible as a result of intermediate reactions involving excited products. It is very important, however, that excited oxygen atoms in the states \({}'D\) and \({}'S\) are chemically unstable. They can above all enter into reaction with oxygen molecules, and the final products will be an excited molecule and an unexcited oxygen atom:

\[ \mathrm{O}_2({}^3\Sigma) + \mathrm{O}'({}'D;\ {}'S) \rightleftarrows \mathrm{O}_2'({}'\Delta;\ {}'\Sigma) + \mathrm{O}({}^3P). \]

Such a direct reaction provides a more effective quenching of the excited states of atomic oxygen than an ordinary inelastic collision. The reverse reaction can be effective only with considerable dissociation of molecular oxygen. The absence in the radiation of the night sky of the bands of molecular oxygen, \({}'\Sigma - {}^3\Sigma\), from vibrational levels higher than the zero one,

is amenable to explanation only if one assumes that the indicated radiation originates in layers of the earth’s atmosphere in which the content of atoms N and H and of molecules NO and CO, reacting with \(\mathrm{O_2}({}^{1}\Sigma)\), is so small that deactivation of the oxygen molecules is practically accomplished through numerous inelastic collisions. In this case the mean lifetime of \(\mathrm{O_2}({}^{1}\Sigma)\) at the zero vibrational level considerably exceeds the mean lifetime in the same state at higher vibrational levels, since an inelastic collision of \(\mathrm{O_2}({}^{1}\Sigma)\) at the zero vibrational level is associated with a considerably greater and, consequently, less probable change in energy than in the loss of part of the vibrational excitation. Such conditions, apparently, exist only in the lower dense layers of the earth’s atmosphere. It may therefore be assumed that the rotational temperature of the bands of molecular oxygen \({}^{1}\Sigma — {}^{3}\Sigma\) reflects the temperature of the surrounding medium. The rotational temperature of the bands \(\mathrm{O_2}({}^{1}\Sigma — {}^{3}\Sigma)\) is very low \((150 \div 200^\circ \mathrm{K})^{20,\,21}\). Such a temperature is possible only at an altitude of about \(80\) km in the region of the temperature minimum. The change in the rotational temperature of the bands \(\mathrm{O_2}({}^{1}\Sigma — {}^{3}\Sigma)\) may be explained by a change in the altitude at which this radiation arises. The cause of the disappearance of these bands may be, for example, the penetration into the lower layers of the atmosphere of large quantities of N, H, NO, and CO.

The point of view set forth characterizes only one direction in attempts to unravel the nature of the radiation of the night sky. The first information on hydroxyl for a time diverted attention from the question of where the energy of the ultraviolet solar radiation, absorbed by the upper layers of the earth’s atmosphere and dissociating its molecules, goes. Instead of explaining the radiation of the night sky as a process removing the energy released in the formation of molecules, searches were undertaken for an entirely new powerful source of energy for the excitation of hydroxyl.

It is therefore impossible not to mention the ozone–hydrogen hypothesis, first proposed by Bates and Nicolet and developed in detail in this country by I. S. Shklovskii\({}^{16}\). From the point of view of Shklovskii, Bates, and Nicolet, the excitation of hydroxyl is due to the reaction

\[ \mathrm{O_3 + H \to OH' + O_2}. \]

The heat effect of this reaction provides maximum excitation of hydroxyl only up to the 9th vibrational level of the ground state. The authors of the ozone–hydrogen hypothesis identify this circumstance with the uniqueness of the ozone–hydrogen reaction as a source of hydroxyl excitation. The increase in the population of the lower vibrational levels is explained by partial excitation of the oxygen molecule, which is one of the final products of the ozone–hydrogen reaction. It is indicated that molecules

oxygen are formed in the states \(^1\Delta\) and \(^1\Sigma\), which, however, from our point of view, is not an exhaustive assumption, since these molecules may also arise in the ground state with high vibrational excitation. The rotational temperature of the hydroxyl bands, equal to \(260^\circ\mathrm{K}\), is taken as the temperature of the surrounding medium. The height of the emitting layer of the earth’s atmosphere is determined as one of the heights at which such a temperature occurs, i.e., 40, 75, or 80 km. Greatest preference is given to the height of 75 km. I. S. Shklovsky assumes that at this height there is a concentration of ozone molecules equal in the daytime, according to Wulf and Demming, to \(10^7\ \mathrm{cm}^{-3}\).

Proceeding from such an ozone concentration and a fairly correct estimate of the rate of the ozone–hydrogen reaction \((3\cdot10^{-12})\), it is calculated that, in order to maintain the intensity of the hydroxyl radiation at the observed level, a concentration of hydrogen atoms equal to \(10^{10}\ \mathrm{cm}^{-3}\) is necessary. At this point, however, the development of the ozone–hydrogen hypothesis breaks off.

In light of the foregoing, the absence in the spectrum of the night sky of hydroxyl bands from transitions from vibrational levels higher than the 9th cannot serve as an unambiguous indication of the ozone–hydrogen reaction alone. More significant is the fact that the authors of the new hypothesis neglect the activation energy of the ozone–hydrogen reaction or use doubtful values of this quantity. Taking it together with the thermal effect leaves no doubt that in this reaction the limiting 10th vibrational level of hydroxyl in the ground state is excited.

Thus, the argument concerning the 9th vibrational level not only is not consistent with, but in general contradicts, the ozone–hydrogen hypothesis.

The determination of the height of the layer emitting the hydroxyl bands from the rotational temperature of these bands also does not appear to be entirely justified. Considerations have already been published to the effect that the rotational temperature of the emission bands of the night sky cannot accurately reflect the temperature of the surrounding medium if the lifetime of the molecule in the excited state is small in comparison with the intervals between collisions of this molecule, or is commensurate with them. From our point of view, such conditions exist for many excited molecules in the upper layers of the earth’s atmosphere, since they rapidly disappear in chemical reactions, without even having time to be significantly deactivated by radiation. Thus, complete thermodynamic equilibrium of the excited particles with the surrounding medium is doubtful. As an example of the absence of such equilibrium, it may be pointed out that the hydroxyl bands emitted by the earth’s atmosphere have a temperature of \(172^\circ\mathrm{K}\), if it is determined from the ratio of the intensities of the doublets, whereas their rotational temperature is \(260^\circ\mathrm{K}\) [19]. It is still not clear what the distribution

of excitation among all degrees of freedom of the molecules at the moment of their formation, it is also impossible to judge whether the newly formed molecules are heated or cooled, i.e., to indicate whether the temperature of the medium is higher or lower than the rotational and kinetic temperature of the emitting molecules. B. I. Stepanov expressed the supposition that the relative population of the doublet levels cannot be changed as a result of collisions and that the temperature of hydroxyl, determined from the ratio of the intensities of the doublets, therefore reflects the temperature of the molecules at the moment of their formation. This means that, once formed, the hydroxyl molecules gradually heat up. If the emitting hydroxyl molecules are not in equilibrium with the surrounding medium, then the consideration put forward by B. I. Stepanov gives grounds for assuming that the temperature of the medium where the excited hydroxyl molecules arise somewhat exceeds the rotational temperature of their bands, equal to \(260^\circ\mathrm{K}\).

The critical remarks we have made concerning certain details of the ozone–hydrogen hypothesis cannot, however, serve as its complete rejection in essence. The most serious other objections are connected with the localization of hydroxyl radiation at an altitude of \(75\ \mathrm{km}\). The rate of the ozone–hydrogen reaction at this altitude can maintain hydroxyl radiation at the observed level only for several seconds. The assumed reserve of ozone will be consumed within several tens of seconds. Maintaining the ozone concentration during the night at an unchanged level is in principle possible with continuous formation of ozone as a result of triple collisions of oxygen atoms.

However, the mean lifetime of free oxygen atoms at an altitude of \(75\ \mathrm{km}\) is considerably shorter than the duration of the night. Therefore, in order to maintain a more or less constant concentration of atomic oxygen, a continuous production of oxygen atoms is necessary. Until a powerful source of energy is found for the dissociation of oxygen molecules at night at an altitude of \(75\ \mathrm{km}\), the ozone–hydrogen hypothesis does not appear unambiguously convincing. It may seem that an upward displacement will save the ozone–hydrogen hypothesis from the shortage and rapid disappearance of atomic oxygen. But we have already shown that newly formed excited ozone disappears immediately, reacting with oxygen atoms and forming excited, but relatively chemically stable, oxygen molecules. The direct reaction of hydrogen with excited ozone, as with excited nitrous oxide, is possible in principle, but quantitatively it is insignificant in comparison with the reaction of atomic hydrogen with excited oxygen molecules.

In conclusion, it should also be recalled that the proposed localization of the emitting hydroxyl layer at an altitude of \(75\ \mathrm{km}\) agrees poorly ...

agrees with the localization at the same altitude of the emission \(O_2(^{1}\Sigma - {}^{3}\Sigma)\), indicating a very small concentration of atomic hydrogen. But this has already been reported above. Such localization is also not consistent with the quenching of hydroxyl emission in the auroral zone, discovered by B. A. Bagaryatskii and undoubtedly occurring at great altitudes.

The foregoing shows that the composition, including weak admixtures, the degree of dissociation of molecules, and the temperature and density of the upper layers of the earth’s atmosphere in the region of the ionosphere can be established by studying and analyzing the radiation of the night sky. The successful application for this purpose of a new method of investigating low-intensity emissions is the best proof of its high effectiveness.

CITED LITERATURE

1. V. I. Krassovsky, DAN 66, No. 1 (1949).
2. A. A. Kalinyak, V. I. Krassovsky and V. B. Nikonov, DAN 66, No. 1 (1949).
3. V. I. Krassovsky, DAN 70, No. 6 (1950).
4. V. I. Krassovsky, Izvestiya of the Crimean Observatory 5, 100 (1950).
5. V. I. Krassovsky, DAN 73, No. 4 (1950).
6. A. A. Kalinyak, V. I. Krassovsky and V. B. Nikonov, Izvestiya of the Crimean Observatory 6, 119 (1950).
7. V. I. Krassovsky, DAN 77, No. 3 (1951).
8. V. I. Krassovsky, DAN 78, No. 3 (1951).
9. V. I. Krassovsky, DAN 78, No. 4 (1951).
10. V. T. Lukashenya and V. I. Krassovsky, DAN 79, No. 2 (1951).
11. V. I. Krassovsky and V. T. Lukashenya, DAN 80, No. 5 (1951).
12. V. T. Lukashenya and V. I. Krassovsky, DAN 81, No. 5 (1951).
13. B. A. Bagaryatskii, V. I. Krassovsky and M. I. Mordukhovich, DAN 82, No. 4 (1952).
14. I. A. Khvostikov, The Glow of the Night Sky (1948).
15. S. F. Rodionov, Izvestiya of the Academy of Sciences of the USSR, physical series 14, No. 3 (1950).
16. I. S. Shklovskii, Izvestiya of the Crimean Observatory 7, 34 (1951).
17. P. Kuiper, Atmosphere of the Earth and Planets (1951).
18. A. B. Meinel, Astrophys. J. 111, 555 (1950).
19. A. B. Meinel, Astrophys. J. 112, 120 (1950).
20. A. B. Meinel, Astrophys. J. 112, 464 (1950).
21. J. Dufay and M. Dufay, Comptes Rendus 232, 426 (1951).
22. J. Dufay, Comptes Rendus 231, 1531 (1950).
23. G. Kvifte, Nature 168, 741 (1951).
24. J. Kaplan, Phys. Rev. 78, 82 (1950).

NOTE ADDED IN PROOF

In his article “Spectra of Atmospheric and Polar-Radiation Emission,” in Reports on Progress in Physics XIV, 121 (1952), Meinel calls all the infrared bands in the night-sky spectrum Meinel bands. Such terminology raises serious objections. In fact, Meinel did not discover any new bands in the night-sky emission. In 1950 he published a high-quality reproduction of the night-sky spectrum up to 8500 Å. However, it reproduces all the bands in this region that had been discovered by Slipher as early as 1933, although they had not been identified by him at that time [see Mon. Not. R. A. S. 93, 657 (1933)]. Unfortunately, because of the wide publicity in the USA given to the incorrect works of Stebbins, Whitford, and Swings [see Phys. Rev. 66, 255 (1944); Astrophys. Journ. 101, 39 (1945)], Slipher’s work was consigned to oblivion. Meinel’s spectrum differs from Slipher’s spectrum by its greater dispersion. Meinel’s contribution is the identification by him of some bands of this emission as the vibration-rotation spectrum of hydroxyl, which was also noted at the time in the Soviet press. However, the study of the infrared emission of the night sky, begun in the USSR as early as 1939 [see UFN 33, 572 (1947)], was carried out independently of Meinel and was not a continuation of his investigations. The night-sky spectra up to 12,000 Å obtained as a result of Soviet work not only fully confirmed Slipher’s result, but also led to the discovery of previously quite unknown and unexpected bands. It is highly characteristic that Meinel, in the article cited at the beginning, notes that in order to obtain a photograph of the night-sky spectrum near 10,000 Å with the aid of the best modern photographic materials, an exposure of a thousand hours would be required. In the work carried out in the USSR, however, the exposures did not exceed several hours.

The priority, uniqueness, and significance of the Soviet investigations of the infrared emission spectrum of the night sky beyond 8500 Å are acknowledged by Dufay [see Ann. de Geophys. 7, 1 (1951)], Pendorff [see Landolt-Börnstein, Zahlenwerte und Funktionen III, 774 (Springer-Verlag, 1952)], and even by Meinel, although the latter reports this only in a note added in proof to the article cited above. From this note it follows that he is acquainted only with the first Soviet papers in DAN 66, No. 1 (1949); 70, No. 6 (1950). Dufay and Pendorff also call all the infrared bands of night-sky emission Meinel bands and state that the work carried out in the USSR is a good confirmation of Meinel’s work. In the article cited above, it is also stated about the constancy of the intensity of the oxygen bands in night-sky emission at 8600 Å, also called by Meinel the Meinel-Kaplan band, although its existence had been assumed by Slipher as early as 1933. In contrast to his previously published works, Meinel reports that, compared with neighboring emissions, this oxygen band sometimes reaches a predominant intensity. However, this had been reported in the Soviet press before Meinel [see DAN 81, No. 5 (1951)], and only a few of Meinel’s words are new: that the band at 8600 Å is enhanced in the early-dawn spectrum. There Meinel also suggests a hypothesis about the nature of the diffuse emission in the infrared spectrum of the night sky, not described anywhere except in Soviet work [see DAN 79, No. 2 (1951); 80, No. 5 (1951); 81, No. 5 (1951)]. It is assumed that it is produced during recombination of hydroxyl molecules from oxygen and hydrogen atoms in three-body collisions. In works published in the USSR it was reported that the infrared emission of the night sky (diffuse or in the form of bands) is not exhausted by the vibration-rotation bands of hydroxyl alone [see DAN 79, No. 2 (1951); 80, No. 5 (1951); 81, No. 5 (1951)]. Evidently, Meinel cannot fail to take such factual data into account.