Review of Studies on Short-Wave Ultraviolet Radiation of the Sun

S. L. Mandelstam

I. Introduction

This review presents the results of studies carried out in recent years on the investigation of the short-wave ultraviolet radiation of the Sun by means of V-2 type rockets. These results are incomplete and unsystematic and in part deserve serious criticism; nevertheless, in some respects they are of undoubted interest.

As is well known, the study of the ultraviolet radiation of the Sun has in recent years been attracting ever greater attention from physicists and astrophysicists. This is due to many different reasons.

Obtaining the spectrum of the Sun throughout the entire ultraviolet region will, first of all, make it possible to supplement our information on the chemical composition of the solar atmosphere. At present the presence of 63 chemical elements has been reliably established on the Sun,\(^{1}\) of which the presence of Au and Th has been established only recently;\(^{2}\) the presence of three elements (Sn, Tb, Ta) is doubtful. Ne, A*), Kr, X, Cl, Br, J, As, Se, Te, Cs, Hg, Re, Tl, Bi, Po, Ra, Rn, Al, Pa, U have not been detected. Most of these elements have spectra that are difficult to excite; their resonance lines lie in the far ultraviolet region of the spectrum. In the near ultraviolet and visible regions of the spectrum there are lines that require large energies for their excitation.

However, this aspect of the matter is not the most important. Of considerably greater theoretical and practical interest is the study of ultraviolet radiation because it is closely connected with the physical nature of processes in the solar atmosphere, with solar activity, and with processes in the Earth’s atmosphere. The strengthening of the activity of active formations—spots, in particular—

*) Lines of highly ionized argon have recently been detected in the spectrum of the solar corona.

...the occurrence of chromospheric flares, faculae, flocculi, filaments, prominences—is closely connected with an intensification of solar radiation in the ultraviolet region of the spectrum. Solar activity, in turn, is connected with processes taking place in the Earth’s atmosphere—the ionosphere and troposphere. An understanding of these phenomena requires detailed knowledge of the magnitude of the ultraviolet radiation, its distribution over the spectrum, and the processes of its absorption in the Earth’s atmosphere. Meanwhile, this knowledge is still to a considerable degree incomplete, unsystematic, and often contradictory. Even with respect to such a thoroughly studied region at present as the ionosphere, comparatively little is yet known in essence. The opinions of various investigators concerning the wavelength of the ultraviolet radiation that causes ionization of the different layers of the ionosphere, and also concerning the depth of penetration of different regions of the radiation, diverge very greatly.

There is every reason to suppose that these phenomena are connected mainly with short-wave ultraviolet radiation, beginning with \(\lambda = 1215\ \text{\AA}\) (\(L_\alpha\)) and lying chiefly beyond the Lyman series limit \((\lambda = 912\ \text{\AA})\)³; in particular, only this radiation is capable of ionizing the Earth’s atmosphere.

A fundamental circumstance is that, in order to explain the processes occurring in the Sun’s atmosphere and in the Earth’s atmosphere, it is necessary to assume the existence of a considerable excess (by \(10\)—\(10^3\) times) of short-wave radiation in comparison with the radiation of the Sun regarded as a black body. In the explanation of this phenomenon, substantial changes have occurred in recent years. On the one hand, on the basis of a number of considerations, the temperature of the Sun in the ultraviolet region of the spectrum should evidently be estimated more correctly not by the value \(T = 5700^\circ\), as is customary for the visible region of the spectrum, but by \(T = 4800^\circ\), which lowers by about 400 times the flux of radiation with \(\lambda < 912\ \text{\AA}\). On the other hand, on the basis of radio measurements and spectroscopic data, the estimate of the electron temperature of the chromosphere has been raised to \(T \simeq 20\,000\)—\(25\,000^\circ\), and, finally, it has been established that the solar corona has an electron temperature of the order of \(10^6\) degrees. A detailed analysis of the short-wave ultraviolet radiation of the Sun in the light of these data has been carried out by I. S. Shklovskii³. According to Shklovskii’s theoretical calculations, the total radiation in the region \(\lambda < 912\ \text{\AA}\), incident in 1 sec on \(1\ \text{cm}^2\) of surface beyond the limits of the Earth’s atmosphere, is composed: 1) from the photosphere of the Sun—\(1.49 \cdot 10^{-5}\) erg, if \(T = 4800^\circ\), and \(6.0 \cdot 10^{-3}\) erg, if \(T = 5700^\circ\); 2) from the chromosphere—\(0.1\) erg; 3) from the continuous radiation of the corona—\(5.6 \cdot 10^{-2}\) erg; 4) from the most intense emission lines of the corona (\(\mathrm{Ne}\ VII/\lambda = 768/776\ \text{\AA}\) and \(\mathrm{Mg}\ XI/\lambda = 610/625\ \text{\AA}\))—\(1.15\) erg. According to Shklovskii’s calculations, the radiation in the region \(\lambda < 400\ \text{\AA}\) is determined by the corona (without taking monochromatic radiation into account), and in the region \(\lambda > 400\ \text{\AA}\)—by the chromosphere.

Moreover, a spectrophotometric analysis of the spectra of chromospheric flares, carried out by E. R. Mustel and A. B. Severny, showed that the radiation in the \(L_{\alpha}\) line during flares is sufficient not only for large disturbances of the ionosphere (the Dellinger effect), but also for the ejection of hydrogen atoms from the surface of the Sun\({}^{8}\).

What has been said is sufficient to explain the primary importance of studying the short-wave radiation of the Sun both for solar physics and for terrestrial physics.

As is known, the Earth’s atmosphere absorbs ultraviolet radiation. The spectrum of the Sun obtained on Earth is sharply cut off on the short-wavelength side near \(\lambda = 2900\) Å. Goetz\({}^{4}\), working at an altitude of 1300 m above sea level, reached \(\lambda = 2863\) Å; Fabry and Buisson\({}^{5}\), \(\lambda = 2885\) Å; however, up to the present time the spectrum of the Sun has been well studied\({}^{6}\) only down to \(\lambda > 2949\) Å.

Absorption of ultraviolet radiation by the Earth’s atmosphere is mainly due to absorption by molecules. In the interval from \(\lambda \simeq 2900\) Å to \(\lambda \simeq 2100\) Å there is a strongly absorbing ozone band with a maximum near \(\lambda = 2500\) Å. Adjacent to this band is the \(O_{2}\) band, extending from \(\lambda \simeq 1950\) Å to \(\lambda \simeq 1760\) Å; \(\lambda = 1760\) Å corresponds to the dissociation of the \(O_{2}\) molecule. Adjacent to this boundary is a region of continuous absorption extending to \(\lambda \simeq 1300\) Å, with a maximum at 1450 Å. From \(\lambda \simeq 1300\) Å to \(\lambda \simeq 740\) Å there are strongly absorbing \(O_{2}\) bands, passing at \(\lambda = 740\) Å into a region of strong continuous absorption by \(O_{2}^{+}\), with a maximum at \(\lambda \simeq 450\) Å. \(N_{2}\) is transparent down to \(\lambda \simeq 1450\) Å. In the region \(\lambda \simeq 1450\)—1000 Å the absorption is not very strong; at \(\lambda \simeq 1000\) Å there adjoins a very intense series of bands extending to \(\lambda \simeq 674\) Å, etc.

Figure 1 shows the general qualitative picture of the absorption of air in the region \(\lambda \simeq 2000\)—600 Å at different pressures, obtained under laboratory conditions, and gives the calculated heights \(H\), above which the corresponding equivalent thickness of the atmosphere is located, under the assumption of an unchanged composition of the atmosphere\({}^{7}\). In the region 2200—2000 Å there is a “window” corresponding to the joining of the absorption bands of ozone and \(O_{2}\). A second “window” is located in the region 1300—1000 Å, and the \(L_{\alpha}\) line of hydrogen falls within it.

Using the available laboratory data on the absorption coefficients of ozone, \(O_{2}\), \(O\), and \(N_{2}\), one can calculate the effective thicknesses of the air layer for various \(\lambda\), at which the atmosphere still has a specified value of transmission. Knowing, further, the distribution of pressure with height, one can calculate the depth of penetration of radiation of different wavelengths.

Fig. 1. Absorption spectrum of air in the short-wave ultraviolet region[^7]. The spectra were obtained with a thickness of the absorbing column of 100 cm. The air pressure—$P$ in cm Hg; $L$—the equivalent thickness of the air layer in cm at a pressure of 76 cm Hg; and $H$—the calculated height (in km) above which this equivalent thickness of the air layer is located, have the following values:

For Fig. 2 the results of such calculations are given for atmospheric transmissions, respectively, of 10% and 1% of the incident energy of different wavelengths.^8 It should be borne in mind, however, that such calculations are essentially qualitative in character. This is due, first of all, to insufficiently accurate knowledge of the absorption coefficients—the data of different investigators differ from one another by several times. Further, the results of laboratory measurements apply to not very low pressures; for calculating the absorption of the upper layers of the atmosphere

Fig. 2. Calculated heights to which solar radiation penetrates in the ultraviolet region of the spectrum.^8 Curve \(A\)—for a transmission of 10%, curve \(B\)—for a transmission of 1%. The solid line is under the assumption of complete dissociation of oxygen at an altitude \(>100\) km, the dashed line under the assumption of the absence of oxygen dissociation.

these data have to be extrapolated. At the same time it remains unknown whether the Bouguer–Beer law holds with respect to the constancy of the specific absorption coefficient, and, if this law is not fulfilled, what the magnitude of the deviations from it is. In particular, with respect to \(O_2\), according to the measurements of Warburg and Hailpern, the Bouguer–Beer law is not fulfilled,^9 whereas Weisler found no such deviations.^10 *) Finally, it is unknown whether the composition of the atmosphere remains unchanged at different heights; in particular, there is no sufficiently precise information as to the heights at which

*) Direct measurement of atmospheric absorption at very low pressures is impossible, since to obtain measurable absorption an enormous thickness of the absorbing layer is required.

dissociation of \(O_2\) and especially \(N_2\). The uncertainty of all these data, taking into account the extraordinarily large effective thickness of the atmosphere, leads to the fact that the differences in the magnitude of atmospheric transmittance in the calculations of different authors amount to several orders of magnitude\(^{11}\).

Despite the considerable uncertainty of all these data, they indicate that, in order to attain appreciable atmospheric transmittance in the region of the spectrum shorter than 2900 Å, the measuring apparatus must be raised several tens of kilometers above sea level*). These investigations have been carried out in recent years in the USA by raising optical apparatus on V-2 rocket projectiles; in this way an altitude of 150 km was reached.

Two groups of investigations were carried out. In the first group the spectrum of the Sun was studied with the aid of spectrographs in the region 3000—2000 Å. In the second group of work, the flux of short-wave radiation lying in the region 2000—0 Å was measured without spectral decomposition, by isolating narrow spectral intervals.

Fig. 3. Approximate trajectory of a rocket\(^{15}\).

The approximate trajectory of the rocket flight is shown in Fig. 3\(^{15}\). The axis of the rocket at launch is vertical, and the rocket preserves this position in flight during the operation of the engine. After the engine has ceased operating, the rocket acquires a more or less random rotation about its own axis with an angular velocity reaching 50 rev/min. In addition, the rocket undergoes precession, with an angular velocity reaching 10 rev/min, about an axis whose deviation from the vertical may reach 20°. This greatly hinders the registration of solar radiation; the effective exposure of the registration proves to be very small. The optical apparatus was placed in the nose part of the rocket or in its tail part and, at a certain point of the descending branch of the trajectory, was separated from the rocket and lowered by parachute.

*) Attempts have repeatedly been made to use the “window” near \(\lambda \simeq 2100\) Å for measurements at altitudes of several kilometers (mountains, balloon sondes), but apparently these attempts did not yield positive results. We shall return to their discussion at the end of the review.

II. SPECTROGRAPHIC INVESTIGATIONS

In the works of Baum, Johnson, Oberly, Rockwood, Strean, and Tousey1 and of Durand, Oberly, and Tousey23 results are given that were obtained in three flights: October 10, 1946, March 7, 1947, and October 9, 1947. The first two reports are preliminary in character, the third is the most detailed.

In all three flights a spectrograph with a diffraction grating was used; its schematic drawing is shown in Fig. 4. The spectrograph was placed in the tail section of the rocket.

To increase the aperture of the beam within whose limits the light still entered the spectrograph, a slitless spectrograph was used. The source was an image of the Sun 0.03 mm in diameter, obtained from a LiF sphere 2 mm in diameter.

To ensure that light entered the spectrograph during the rotation of the rocket, two such spheres were used. At launch, the optical axes of the illuminating systems were directed to the south and north and inclined at an angle of 45° to the zenith. This arrangement ensured the obtaining of a spectrum for angles between each optical axis and the line directed toward the Sun lying within limits up to 70°. In this case, however, for large angles some deterioration in the focusing of the spectrum was observed and, when the image of the Sun shifted during the exposure, a blurring of the spectrum. However, the gain in light, in comparison with the use of a slit

Fig. 4. Schematic drawing of the spectrograph.3

and of the scattering plate in front of the slit proved to be very large.

The light beam that had passed through each ball was directed, by means of rotating mirrors, onto a concave diffraction grating with a radius of curvature of 40 cm and having 6000 lines per centimeter. The spectrum was photographed on film 35 mm wide, placed along the Rowland circle. The spectra from both illumination systems were shifted relative to one another in height by a small rotation of the mirrors. When plane mirrors were used, owing to the astigmatism introduced by the grating, the height of the lines was 1 mm; in the case of cylindrical mirrors, the astigmatism could be compensated quite fully for one wavelength and greatly reduced for the entire spectrum.

The resulting decrease in the height of the spectrum increased the blackening of the lines; however, photometry of the spectrum was hindered by the different lengths of the lines in different regions of the spectrum. An automatically operating shutter ensured a repeating cycle of exposures consisting of exposures of 0.1, 0.6, and 3.0 sec. The film was also advanced automatically. Eastman 103—O film, sensitized for the ultraviolet region of the spectrum, was used.

Owing to the rotation of the rocket, most of the spectra were underexposed or badly blurred. Photograph I (see the insert to p. 156) shows some spectra obtained during the flight of 10.X 1946 at altitudes up to 88 km1. After this the rocket turned so that light ceased to enter the spectrograph; at the next turn (at an altitude of 107 km) the entire film had already been used up.

Up to an altitude of 44 km the rocket was stabilized. After this it rotated, and spectra $F$ and $G$ were obtained when the optical axis was not directed through the Sun. The spectra have been retouched in reproduction.

The spectrograms show an ever increasing extension into the ultraviolet region with increasing altitude. Spectrum $D$ (corresponding to an altitude of 25 km) extends to $\lambda \simeq 2925\,\text{\AA}$, spectrum $E$ (34 km) to $\lambda \simeq 2650\,\text{\AA}$ and shows noticeable blackening in the region 2100—2260 Å. In spectrum $F$ (55 km) complete exposure is observed in the region of the ozone absorption band.

Subsequent flights yielded more successful spectrograms. In particular, in one of the spectrograms obtained on 7.III 1947 at an altitude of 75 km, the resolving power was not inferior to the best resolving power obtained in the laboratory. This spectrogram, shown in photograph II (see insert), was used as the basis for a careful analysis of the spectrum3. A microphotogram of this spectrum is given in Fig. 5. In Fig. 6,a a micropho-

togram of the spectrum obtained on 10.X.1946 at an altitude of 35 km, showing the existence of a transmission “window” near 2100–2200 Å. For analysis of the spectrum in the region shorter than 2415 Å, a spectrogram obtained on 10.X.1946 at an altitude of 55 km was used, which in this region of the spectrum gave better resolution owing to its lower density.

Fig. 5. Microphotograms of the spectrum shown in photograph II 14.
* marks places where the film was damaged.

owing to its lower density. The corresponding microphotogram is shown in Fig. 6, б.

The analysis of the spectra was carried out as follows:

1. The wavelengths of the lines present in the spectrum were determined. For most of the lines these determinations were based on visual measurements on a comparator by several observers. For a small number of weak and unresolved lines, data from the microphotometric record were used. The constants entering into the interpolation formulas were determined so as to satisfy in the best possible way the measurements of 14 intense Fe I and Fe II lines distributed over the whole spectrum. The possible error in determining the wavelength of the other lines lies between 0.2–1.0 Å, depending on the sharpness of the lines and freedom from a “blend.”

1. The intensities of the lines were determined visually on a scale analogous to Rowland’s scale, i.e., with an estimate of the central intensity of the line and its width; the weakest observable lines correspond to an intensity of 3 on Rowland’s scale, and the strongest to an intensity of 1000.

2. The lines were identified. For this purpose, on the basis of Moore’s tables of multiplets[^6], a list of 1200 lines was compiled for which it could be assumed that they might be detected at the given resolving power. This list was compared with the spectral-measurement data, and for each line the possible elements whose spectra the line may belong to were indicated.

Fig. 6. Microphotograms of two spectra3:
a — at an altitude of 35 km, b — at an altitude of 55 km.

On the basis of an analysis of the corresponding multiplets and laboratory intensities, the most probable element is indicated for many lines. The results of these investigations are presented in a detailed table and cover about 300 lines. Examination of this table shows that the principal number of lines in the region 3000–2200 Å belongs to Fe I and Fe II, similarly to the previously accessible regions of the solar spectrum. In the regions 2750, 2630, 2550, and 2490 Å there is a very large concentration of iron lines. Developed multiplets have been found, identified as belonging to Si I, V I, V II, Cr II, Co I, Mn II, and lines identified as belonging to Al II, Mg I, Mg II. Further, the lines may possibly belong to Mn I, C I, Ti I, P I, Al I, Cu I, Ni I. Finally, a weak line \(\lambda \approx 2456.1\) Å, in Rosseland’s opinion, could belong to As I (laboratory value \(\lambda = 2456.53\) Å).

The lines located in the spectral region 2280–2290 Å could not be identified as atomic lines, and it is assumed that they belong to a molecular spectrum. In particular, this may be NO, which has a band near 2264 Å, whose absorption at layer thicknesses of 0.5 mm is very large.

The structure of the lines Mg II \(\lambda = 2803\) and \(\lambda = 2796\) Å proved to be very interesting. The wings of these lines extend 25 Å on both sides, and at their center there are bright emission lines (see Fig. 7), which had previously been observed in the \(H\) and \(K\) lines of Ca II. The great width of the Mg II lines probably indicates that absorption takes place in deep layers of the reversing layer, where the effect of pressure broadening of the lines is large. The appearance of the emission lines is interpreted by Menzel as the result of a strong eruption of hydrogen that had taken place an hour before the flight.

Fig. 7. Distribution of energy in the spectrum of the Sun in the ultraviolet region, expressed as percentages of black-body radiation at \(T = 6000^\circ\).¹⁴

Finally, by means of intensity marks applied to the films before the flight, the distribution of energy in the spectrum was measured. This distribution, shown in Fig. 7, is expressed as percentages of the intensity of a black body at \(T = 6000^\circ\) K.

In the work of Hopfield and Clearman¹⁶ the results of two flights are given: 1.IV 1947 at 13 h 10 m and 29.VII 1947 at 5 h 55 m. In both flights identical spectrographs were used; only the illuminating systems differed somewhat. A schematic drawing of the spectrograph is given in Fig. 8. The spectrograph was placed in the head section of the rocket. The spectrograph has two slits, located on both sides of the normal to the grating. Both slits are displaced somewhat in height, in opposite directions from the central plane, so that the spectra are likewise displaced vertically. The grating is concave, 6000 lines per centimeter, with a radius of curvature of 50 cm. The slit width is 0.025 mm, the height 2 mm.

In the first flight, in order to capture sunlight over as large a solid angle as possible, illumination of the slit was produced by light scattered on frosted glass and directed onto

slit by a flat mirror. In the second flight, concave cylindrical mirrors with slight corrugation were used, stretching the image of the Sun into a strip 5 mm wide and 60° long in the plane of the drawing. The mirrors were rotated about an axis lying in the plane of the drawing, and a tracking system controlled by two photocells ensured that the image of the Sun was kept on the slit of the spectral apparatus as the position of the rocket relative to the Sun changed.

Fig. 8. Schematic drawing of spectrograph 16.

The spectrum was photographed on 35-mm Eastman 103a O film, sensitive to the ultraviolet region of the spectrum. The exposure was regulated by an automatically acting shutter, which ensured the obtaining of 14 photographs with an exposure of 5 sec each during the ascent of the rocket in its stabilized position, and 5 photographs with exposures of 55 sec each during the remaining time of the rocket flight. The film was automatically rewound into an armored cassette. Additional photocells located near the slit, as well as the film-winding mechanism, produced electrical pulses transmitted to Earth, from which it was possible to determine the effective exposure of each photograph.

With the aid of this apparatus a series of spectrograms was obtained at different altitudes. In photograph III, as an example, a portion of the spectrum (together with the comparison spectrum) obtained at a mean altitude of 135 km is shown; photograph IV gives a microphotogram of this spectrum.

I. General view of the spectrum of the Sun, obtained at different altitudes1.

II. Spectrogram of the solar spectrum, obtained at an altitude of 75 km3.

To the article by S. L. Mandelstam.

III. Section of the spectrum of the Sun obtained at an altitude of 135 km^16.

IV. Microphotogram of the spectrum shown in photograph III^16.

For preliminary processing of the spectrograms, an iron spectrum was obtained with the aid of the same spectrograph before it was installed in the rocket. Enlarged images of the spectrum under study and of the iron spectrum were then superposed. More accurate processing was carried out using, as comparison standards, several of the least blended iron lines in the solar spectrum.

Owing to blending, the scatter of the results proved to be rather large; the wavelengths of the lines are therefore given with an accuracy of only 0.1 Å.

For the photometric processing of the spectrograms, blackening marks were applied to three pieces of the same film. One of these pieces was placed in the spectrograph cassette, the second (in a strong cylinder) was placed inside the nose section of the rocket, and the third was kept in the laboratory. Thus the first two pieces underwent all the same thermal, mechanical, and chemical effects as the film on which the spectrum was obtained. All three pieces of film gave identical results.

The paper gives a preliminary table of identified lines. All unidentified lines and lines identified as blends of weak lines are omitted. The table covers the spectral region 3000–2500 Å. The main part of the lines belongs to Fe I and Fe II. There are also lines belonging to Si I, P I, Na I, V I, V II, Mg I, Mg II, Mn II, Tl II, Cr I.

As in the preceding work, the great width of the Mg II 2796 and 2803 Å lines in all spectra is noted, leading to their partial overlap, as well as the presence at the centers of these lines of an emission line whose intensity reaches about 10% of the intensity of the surrounding continuous spectrum.

Determination of the curve of energy distribution over the spectrum is very difficult because of the strong decrease in intensity at the locations of the concentration of iron lines. Attention is drawn to two bright “windows”: \(\lambda \simeq 2638\) and \(\lambda \simeq 2643\) Å. The most natural supposition is that this is the true intensity of the continuous spectrum of the Sun, undistorted by absorption lines. However, the intensity in these “windows” is less than it should be for \(T = 5500^\circ\) K. The authors consider it possible that this is radiation of iron, calcium, and other elements existing in a highly ionized state in the solar corona.

III. STUDIES WITHOUT SPECTRAL DECOMPOSITION

The main difficulty of such studies lies in finding techniques for sufficiently steep cutting off of the long-wavelength side of the spectrum. It has not been possible to create liquid or solid filters for the short-wavelength ultraviolet region with a narrow transmission band, since all known materials, transpar—

... for this region of the spectrum, have absorption on the short-wavelength side.

In the work of Tousey, Watanabe, and Purcell ^17, the phosphor CaSO$_4$, activated with MnSO$_4$, was used as the radiation receiver.

Fig. 9. Spectral sensitivity of the phosphor CaSO$_4$ : MnSO$_4$ ^18.

This phosphor possesses the remarkable property that the region of its sensitivity is rather sharply bounded on the long-wavelength side of the spectrum near $\lambda = 1350$ Å. If it is irradiated with ultraviolet radiation $\lambda < 1350$ Å and then heated, it emits the stored light energy in the region 4480—5790 Å. This phosphor, whose principal properties had already been observed by Wiedemann and Schmidt in 1895, was studied in detail in the work of Watanabe ^18.

The phosphor was prepared by mixing the sulfates of Ca and Mn in weak sulfuric acid*); after several hours of standing, the solution was evaporated to dryness and the precipitate was heated to red heat. The powder was then pressed into a nickel mesh of 100 mesh and fixed with a 5% solution of cement in acetone, which reduced the sensitivity of the phosphor in the region 1100—1300 Å by 10%.

The energy stored in the phosphor when illuminated with the corresponding ultraviolet radiation was released by heating the phosphor by passing a current through the nickel mesh. The luminescence was measured with a photomultiplier, whose photocurrent was amplified and fed to a self-recording instrument. The light sum was obtained by integrating the recorded curve over time.

The principal properties of the phosphor are as follows: the spectral sensitivity of the phosphor (Fig. 9) has a maximum near 1030 Å; in the region of short waves the phosphor is sensitive to $\gamma$-rays; in the region of long waves it still has a small sensitivity near

*) Changing the amount of MnSO$_4$ within the limits from 0.2 to 17% did not affect the properties of the phosphor.

1600 Å (the sensitivity in this region is about \(10^{-3}\) of the sensitivity in the region 1100—1300 Å). The quantum yield in the region of maximum sensitivity is about 5—10%. The difference in sensitivity in the region 1050 Å for five phosphor samples lay within 13%. The luminescence spectrum of the phosphor upon heating is shown in Fig. 10. The light sum of the luminescence proved, over wide limits, to be independent of the phosphor-heating regime and linearly related to the magnitude of the exposure of the phosphor illumination (energy-time) over a range of exposure variation by a factor of \(10^4\).

Table I

Fig. 10. Microphotogram of the luminescence spectrum of the phosphor \(\mathrm{CaSO_4:MnSO_4}\) after excitation by ultraviolet radiation.\(^{18}\)

The magnitude of the light sum does not depend on the temperature of the phosphor at which it is maintained during exposure to ultraviolet radiation, within the range of temperature variation from \(-30^\circ\)C to \(+70^\circ\)C.

The law of luminescence decay is well approximated by the expression

\[ I=\frac{E_0F(T)}{a+t}, \]

where \(I\) is the luminescence intensity, \(E_0\) is the total light sum, \(F(T)\) is an unknown function of the temperature during emission, \(t\) is the time in minutes, and \(a\) is a constant of order 0.22 min. The rate of fading depends substantially on the heating temperature of the phosphor. At \(0^\circ\)C there is practically no fading; at \(20^\circ\)C it is noticeable, and at \(280^\circ\)C the fading is practically completed in 2—3 min. It was further found that fading is stimulated by the near-ultraviolet region of the spectrum \(\lambda > 2200\) Å. Finally, it was established that the presence of impurities in the phosphor (Mg, alkali elements) shifts the maximum of sensitivity and the long-wavelength boundary toward longer waves.

To obtain the intensity of ultraviolet radiation in various regions of the spectrum, phosphors of the indicated type were combined with filters cutting off the short-wavelength part of the spectrum. Plates of \(\mathrm{CaF_2}\) and LiF and beryllium foil were used as filters. The recorded portions of the spectrum with different

combinations of filters are given in Table I and are shown for CaF$_2$ and LiF in Fig. 11; the transmission of the Be filter at 10 Å is very small, and therefore the filter probably in fact transmits radiation shorter than 8 Å.

The values of curve $A$ without a filter were obtained by measuring the phosphor samples that took part in the flights. The values for the last two regions of the spectrum listed in Table I were obtained by subtracting the energy obtained with different filters. This method requires knowledge of the distribution of energy in the radiation spectrum. The data for these regions of the spectrum are therefore approximate. The principal

Fig. 11. Regions of the spectrum recorded by the phosphor with various filters$^{17}$:
$A$ — without filter,
$B$ — LiF filter, $C$ — CaF$_2$ filter.

difficulty lay in eliminating uncontrolled heating of the measuring phosphor samples between exposures and measurements. The experiments were therefore carried out only in the winter months, and the exposed phosphor samples were transported to the laboratory for measurement in thermos flasks with ice. To check for the absence of heating, in each flight control phosphor samples were used; these were irradiated 4–12 hours before the flight with a standard exposure from a hydrogen tube with a LiF window and were no longer exposed during the flight. Some of these control samples did not take part in the flight, and after the flight the light sum of these samples was compared with that of the control samples that had taken part in the flight. Such a comparison made it possible in some cases to introduce corrections into the measurement results. In all, during the period 1948–1950, six flights of V-2 rockets equipped with phosphors were carried out.

Two flights proved unsuccessful; data on the four successful flights are given in Tables II and III.

Table II

Table III

Flight No. 4, 17.II 1950, 11 h 00 min. Maximum altitude 150 km above sea level. Solar activity: during the flight no flares or sudden ionospheric disturbances were observed. Phosphor area 0.65 cm².

In the first flight, four sets were used, each of which consisted of six samples of phosphor and filter, placed in a steel casing located at a depth of about 12 mm from the surface of the rocket shell. This provided only partial protection against heating as the projectile passed through the lower layers of the atmosphere. The control samples showed the presence of substantial heating during the flight, as a result of which the results of this flight are only qualitative.

In the second and third flights the phosphors were protected from heating by the air flow. They were placed on a cylinder that rotated inside a cylindrical cassette, so that illumination of the phosphors began immediately after the rocket engine stopped operating (60 sec.) and continued until the moment preceding the jettisoning of the rocket nose section. The total exposure time was 200 sec.

In the fourth flight a phosphor with approximately fivefold increased sensitivity was used, which made it possible to use five sets of phosphors, exposed successively at different altitudes, beginning at 19 km, over the course of 50 sec. As a result of the strong impact of the apparatus against the ground, only 10 samples out of 50 survived.

The actual exposure time in each flight depended on the rotation of the rocket during flight. In flight No. 1 this time was estimated, while in flights No. 2 and No. 3 it was measured from the blackening of photographic film covered by a filter with low transmission. In the fourth flight the exposure time was determined from data on the position of the rocket with an accuracy of about 30%.

As examination of the tables shows, in flight No. 2, in which an altitude of 128 km was reached, the existence of X-radiation was recorded. At the same time, a sudden ionospheric disturbance was observed during the flight. In flight No. 1, although a greater altitude was reached, and also in flight No. 3, X-radiation was not recorded.

Some indication of the existence of X-radiation was obtained in flight No. 4; however here, as the authors note, the results are contradictory in the sense that the radiation intensity is greater for the altitude interval 19–82 km than for 82–127 km. The authors consider it possible that this is explained by the fact that the radiation in the X-ray region has the character of flashes.

Radiation in the region 1050–1340 Å was recorded beginning at altitudes of 80–90 km, as is proved by the registration of this radiation in flight No. 3 (altitude reached 88 km) and in flight No. 4 during exposure No. 1.

Radiation in the region 1240–1340 Å with considerable intensity was recorded in flight No. 2 in the altitude interval

49–128–86 km and in flight No. 4 in the interval 82–127 km. For altitudes below 88 km (flight No. 3), radiation in this region was not recorded.

The intensity of radiation in the region 795–1050 Å can, as indicated above, be roughly estimated by subtracting from the intensity obtained for the phosphor without a filter (0–1350 Å) the intensity transmitted through the Be filter (0–10 Å) and LiF (1050–1350 Å), with the corresponding corrections for the transmission of the filters. Radiation shorter than 795 Å can probably be regarded in this case as completely absorbed by \(N_2\). In flight No. 4, exposure No. 2, the energy measured by the phosphor without a filter was approximately 50 times greater than the energy measured by the phosphor with the LiF filter, while the radiation recorded in the 0–10 Å region was very small. This permits one to say that radiation in the region 795–1050 Å reaches, with appreciable intensity, the altitude interval 88–127 km. Exposure No. 1 of this flight gives some indication that this radiation penetrates deeper than 82 km. Here, however, the radiation intensity recorded without a filter is only 6 times greater than the intensity recorded by the phosphor with the LiF filter, which is very close to the threefold excess caused by absorption by the filter itself. Therefore, in the authors’ opinion, penetration of the 795–1050 Å radiation deeper than 82 km cannot be considered proven.

The intensity of this radiation at an altitude of 127–148 km is 50% greater than at an altitude of 88–127 km; this suggests that the 795–1050 Å radiation passes down to an altitude of 127 km without appreciable attenuation and is absorbed in the altitude interval 127–88 km.

Flight No. 4 made it possible to obtain some quantitative estimates of the magnitude of the radiation. The total radiation obtained by the phosphor in the interval 1050–1240 Å is estimated by the authors as \(0.04\ \mu\mathrm{W}/\mathrm{cm}^2\) for altitudes of 82–127 km, with a possible error of up to a factor of two. The value of the total radiation for altitudes located above the absorbing region of the atmosphere is probably not much greater. The total radiation for the region 1230–1340 Å is estimated as \(0.02\ \mu\mathrm{W}/\mathrm{cm}^2\). The accuracy of this estimate is considerably lower. Analyzing these results, the authors come to the conclusion that the experimental data agree best with the assumption that, in the region 1040–1340 Å, there is radiation corresponding to black-body radiation, on which is superimposed an intense and broad absorption line \(L_\alpha\) with a narrow and intense emission line at its center.

An estimate of the radiation flux for the regions 795–1030 Å was obtained by the method described above, taking into account that the quantum efficiency of the phosphor was from 5 to 10% for

\(\lambda \simeq 1050\) Å and approximately half this value for \(\lambda \simeq 800\) Å. According to the exposure data of No. 2 of the fourth flight, this estimate gives a value of the order of \(5\cdot 10^{11}\)—\(3\cdot 10^{12}\) quanta/\(\text{cm}^2\) sec.

In the work of Friedman, Lichtman, and Byram\({}^{19}\), photon counters were used as radiation receivers. The counters registered radiation in the spectral regions 0—10 Å, 1100—1350 Å, 1425—1650 Å, and 1725—2100 Å. The flight was carried out on September 29, 1949, at 10 h 00 min. The altitude of the Sun was \(43^\circ\). During the flight no significant changes in solar activity were observed. The counter readings were continuously transmitted to the Earth during the entire flight (336 sec.). The rocket reached an altitude of 150 km. Engine operation ceased 64 sec. after launch. During the first 60 sec. the position of the rocket was stable, after which the rocket began slowly to rotate, approximately, on average, with a 12-second period.

Two sets of six counters were enclosed in two cassettes, which were placed on opposite sides of the nose section of the rocket; the counter windows were parallel to the surface of the casing. In each cassette there was also placed a counter sensitive only to cosmic rays, and a photoelement for determining the position of the rocket relative to the Sun during its rotation.

Each counter consisted of a round shell (of chromed steel) 18 mm in diameter and 50 mm long, which served as the cathode, and an anode—a filament 0.1 mm in diameter. Glass leads for the anode were welded to the shell at both ends. Along the cathode a flattened portion was made, in which a hole 4.7 mm in diameter was drilled. The hole was closed by the corresponding filter, fastened to the shell by means of cement.

The counter for X-rays had as its filter beryllium foil 0.125 mm thick. The transparency of the filter was practically complete for \(\lambda = 2\) Å, 10% for \(\lambda = 6.5\) Å, 1% for \(\lambda = 8.5\) Å, and 0.1% for \(\lambda = 9.5\) Å; for \(\lambda > 10\) Å the filter was practically opaque. Approximately \(1/5\) of the photons passing through the filter produced a count.

The counters for ultraviolet rays had, as filters cutting off the short-wavelength part of the radiation, LiF (1100 Å), corundum (1425 Å), and crystalline quartz (1725 Å). To cut off the long-wavelength radiation the authors used an interesting effect they had discovered, namely that when small amounts of \(\mathrm{Cl}_2\), \(\mathrm{Br}_2\), or halogen compounds of hydrocarbons are added to the noble gas filling the counter, the long-wavelength threshold of the sensitivity of the counters shifts toward shorter wavelengths. A counter with an LiF window

was filled with a mixture of neon at a pressure of 300 mm Hg and 15 mm of Cl₂, which gave a threshold at 1350 Å. The counter with a corundum window was filled with a mixture of neon at a pressure of 300 mm Hg, with 0.5 mm of Br₂, and the counter with a quartz window—with a mixture of argon at a pressure of 10 mm Hg and ethylene at the same pressure. The curves of the spectral sensitivity of the counters are shown in Fig. 12. The efficiency of the counters at the sensitivity maximum in the presence of the filter was \(10^{-6}\)—\(10^{-8}\) counts per quantum. The absolute sensitivity of the counters, as well as the region of spectral sensitivity, changed with time and with use of the counters. The counters for the region 1150—1350 Å possessed

Fig. 12. Spectral sensitivity of the counters¹⁹. The scale along the ordinate is different for all three counters. The sensitivity at the maximum is expressed in the number of counts per one quantum.

stability over the course of several months. However, for most counters the long-wavelength threshold had a tendency to shift toward the long-wavelength side of the spectrum. About a month elapsed between the testing of the counters and the flight. Although control experiments did not show substantial changes in the counters, the sensitivity of the counters that participated in the flight can be estimated only to order of magnitude.

The counting rate of each counter was determined from the integral value of the charge passing through the counter, by means of an \(RC\)-circuit with a time constant of 2 sec. The mean voltage value, proportional to the counting rate, was fed by means of a cathode follower to the telemetric system without additional amplification. The data from all the counters were transmitted over four channels by means of the corresponding commutators. In Fig. 13, as an example, a sample of the telemetric recording is presented. At each revolution of the rocket, the corresponding

peak. The signal switching rate made it possible to obtain eight points for each peak.

The main factor affecting the accuracy of the measurements consisted in the onset of “saturation” of each peak at high altitudes.

To determine the true maximum of the “saturated” peaks, it was assumed that the shape of each peak is determined only by the relative position of the counter window and the Sun. Having determined the peak shape (the ratio between the maximum amplitude and the amplitude at a specified moment before and after the maximum) from peaks that had not reached saturation, the authors extrapolated the initial and final portions of the “saturating” peaks. This correction sometimes doubled the “saturated” count. A second correction was introduced to compensate for the distortions introduced by the large time constant of the integrating circuit. Finally, a correction was introduced for the precession of the rocket about its axis.

Fig. 13. Telemetric recording of counter signals for X-radiation4. The curves are drawn through the experimental points.

In Fig. 14, as an example, the counts of two identical counters located along the diameter of the rocket nose section during the ascent and descent of the rocket are given. The scatter of the points is due mainly to the error in extrapolating the “saturated” maxima and to the inaccurate knowledge of the rocket’s position. In general, the authors estimate the measurement error to be of the order of $\pm 50\%$. Despite the scatter of the points, the authors believe that, for the three shorter-wavelength regions of the spectrum, the altitudes corresponding to the maximum rate of change of absorption, and the region of maximum penetration

radiation. The counter, which recorded the region \(\simeq 2000\) Å, reached saturation already at an altitude of \(7\) km, which made further extrapolation of the “saturated” peaks impossible.

The following results were obtained: X-ray radiation was registered beginning at an altitude of \(87\) km; the resulting course of the change in atmospheric transparency for X-ray radiation with altitude is shown in Fig. 15. The flux of radiation beyond the limits of the Earth’s atmosphere, calculated from the experimental data, is \(10^{-4}\) erg/cm\(^2\) sec. This flux should probably be attributed to a small spectral interval near

Fig. 14. Counts obtained with two approximately identical counters for the region 1100–1350 Å during the ascent and descent of a rocket\({}^{19}\). A correction has been introduced for “saturation” and for the time constant of the circuit.

\(\lambda = 8\) Å. Fig. 15 also shows the change in atmospheric transparency for X-rays of various wavelengths, calculated from the data of Compton and Allison\({}^{20}\) on the magnitude of absorption of X-rays and from Havens’s data on the pressure distribution with altitude\({}^{21}\). In addition, the atmospheric transparency calculated under the special assumption of the distribution of continuous radiation in the solar corona at a temperature of \(10^{6}\) degrees\({}^{22}\) is shown.

Ultraviolet radiation in the region 1100–1350 Å was registered beginning at an altitude of \(70\) km \(\pm 5\) km; the resulting course of the change in atmospheric transparency for this spectral region is shown in Fig. 16. On the same figure is plotted the calculated

Text in the figure: vertical axis — “Altitude in km”; horizontal axis — “Transparency in %”; curve labels include “68 Å; 22.5 Å,” “37 Å,” “8.00 Å,” “4.6 Å,” and “V-2 No. 49.”

Fig. 15. Change of atmospheric transparency with altitude for X-ray radiation4: — experimental results; – – calculated data for different wavelengths; – · – calculated data for coronal radiation at \(T = 10^6\) degrees.

Text in the figure: vertical axis — “Altitude in km”; horizontal axis — “Transparency in %”; curve labels include “1300 Å,” “V-2 No. 49,” “1216 Å (sun at zenith),” and “1216 Å (sun altitude 45°).”

Fig. 16. Change of atmospheric transparency with altitude for ultraviolet radiation in the region \(\lambda \sim 1200\ \text{Å}\)4: — experimental results; – – calculated data.

the transparency of the atmosphere for \(\lambda = 1216\) (\(L_\alpha\)) and \(\lambda = 1300\) Å according to the data of Preston \(^{23}\) and Ladenburg \(^{24}\), under the assumption that oxygen is not dissociated. The experimental curve is in good agreement with the absorption curve for \(L_\alpha\).

Radiation in the region \(\lambda \approx 1500\) Å was recorded beginning at an altitude of approximately 50 km; however, the authors consider it possible that the counter readings at this altitude were caused by electrical interference. Therefore the authors indicate an altitude of 95 km. Figure 17 shows the change, measured for this spectral region, in atmospheric transparency with altitude, and gives the computed curves under the assumption that oxygen is not dissociated.

For the spectral region 1750–2100 Å the authors confine themselves merely to stating that the radiation was recorded at an altitude of 7 km.

The calculation, from the experimental data, of absolute radiation fluxes, in view of the comparatively large width of the spectral intervals used, requires knowledge of the law of energy distribution in the spectrum of the Sun. Assuming that this distribution corresponds to black-body radiation, the authors consider that the radiation intensity in the spectral region near 1200 Å corresponds to a solar temperature of 6000° K, and in the region 1500 Å to a temperature of 4500° K. The total radiation flux in the region 1150–1350 Å is estimated by the authors at \(1\text{–}10\ \mathrm{erg}/\mathrm{cm}^2\,\mathrm{sec}\).

Fig. 17. Change of atmospheric transparency with altitude for ultraviolet radiation in the region \(\lambda \sim 1500\) Å \(^{19}\): — experimental results; – – computed data.

In a short note by Burnight \(^{25}\), the results obtained in the photographic recording of X-ray radiation are set out very briefly. The photographic film was placed in a cassette with windows of aluminum and beryllium foil.

In the flight of 5.VIII 1948, film covered by a beryllium window \(0.076\) cm thick recorded an unexpectedly high intensity of radiation with \(\lambda < 4\) Å. In the flight of 18.XI 1948, films covered by aluminum windows \(0.00076\) cm and \(0.00153\) cm thick and by a beryllium window \(0.0254\) cm thick recorded no radiation.

In a similar experiment on 9.XII 1948, a noticeable blackening was found behind an aluminum window of thickness 0.00076 cm; for a beryllium window of 0.0254 cm, no blackening was found. The altitude of the flights is not indicated.

IV. DISCUSSION OF RESULTS

Let us first consider the results of the spectrographic measurements. The use of a slit in the spectrograph of Hopfield and Clearman provided a spectrum of better quality in comparison with the spectra obtained by Durand, Oberly, and Tousey. This makes the determination of the positions of the lines by Hopfield and Clearman more reliable. The measurements of the wavelengths of the lines, however, were apparently carried out in this work less carefully, which somewhat lowers the quality of the results obtained. In the work of Hopfield and Clearman, moreover, as was already indicated above, an incomplete list is given of the lines observed by the authors and, in particular, the wavelengths of the unidentified lines are not given, which is of greatest interest. A detailed comparison of the two works is therefore difficult.

With regard to establishing the presence in the spectrum of the Sun of the lines Fe I, Fe II, Si I, V I, V II, Mg I, Mg II, the data of the two works agree and are based on a sufficiently large number of lines. Thus the presence of the lines of these atoms and ions in the ultraviolet region of the solar spectrum may be considered reliably established (the presence of these atoms and ions in the atmosphere of the Sun, as was indicated above, is not in doubt). The situation is evidently analogous for the lines P I, although here only two lines are involved.

With regard to the lines of the remaining atoms and ions, the data of the two works diverge. Thus, the line which, according to the measurements of Durand and others, has wavelength 2835.5 Å and is identified by them as the strongest Cr II line \((\lambda = 2835.64\ \text{Å})\), according to the measurements of Hopfield and Clearman has wavelength 2836.0 Å and is identified by them as an Fe II line \((\lambda = 2835.72\ \text{Å})\). The line which, according to the measurements of Hopfield and Clearman, has wavelength 2985.8 Å may, according to the table of these authors, belong to Cr I \((2985.86—86.00—86.47\ \text{Å})\) or Fe II \((2985.55\ \text{Å})\); in the table of Durand and others the corresponding line is absent. Comparison of these data compels one to regard the establishment of the presence of Cr I and Cr II lines in the ultraviolet region of the spectrum as unreliable.

The line which, according to the measurements of Durand and others, has wavelength \(\lambda = 2669.4\ \text{Å}\) and is identified by them as an Al II line \((2669.116\ \text{Å})\), and the lines \(\lambda = 2661.5\ \text{Å}\) and \(\lambda = 2650.7\ \text{Å}\), identified as possibly belonging to Al I, are absent in the table of Hopfield and Clearman, evidently as weak lines. Therefore the question of the Al lines remains open.

The line \(\lambda = 2928.8\) Å, identified by Durand and others as the strongest Co I line \((\lambda = 2928.812\) Å), lies in a poorly resolved band. According to Hopfield’s measurements, this line has wavelength \(2428.7\) Å and is identified as an Mg II or Fe II line, which appears the most probable. The situation is analogous with the Ti lines. The line \(\lambda = 2825.8\) Å, identified by Durand and others, possibly, as a Cu line \((\lambda = 2824.7\) Å), lies in a poorly resolved region; in Hopfield’s table there is no line of close wavelength. The lines \(\lambda = 2593.7\) Å and \(\lambda = 2680.7\) Å are identified by Hopfield as Na lines \((\lambda = 2593.82/92\) and \(2680.33/43\) Å); in the table of Durand and others the first line has wavelength \(2593.6\) Å and is identified as an Mn II line \((2593.73\) Å), while the second line is absent. The presence of Na lines is therefore doubtful. Finally, the lines appearing in the table of Durand and others as a Cl line, and the lines possibly belonging to Ni I and As I, have wavelengths below \(2500\) Å and fall outside the limits covered by Hopfield’s table. Of special interest here is the question of As I, whose presence in the solar spectrum, as indicated above, has not yet been established. Against Russell’s suggestion that the line 2456.1 belongs to As is the circumstance that the excitation potential of this line is 6.37 volts, whereas the highest excitation potential so far observed in the spectrum of the reversing layer of the Sun is 5.75 v. The remaining As lines lying in the investigated spectral region also have excitation potentials of 6.6–6.7 v, but are strongly blended.

Finally, it must be noted that in the investigated spectral region there lie the resonance lines of Hg \((2536.5\) Å), Au \((2427.95\) and \(2675.95\) Å) and the intense lines of Bi \((2897.975\) Å), Te \((2385.76\) and \(2383.25\) Å). The question of searching for these lines, however, is not discussed in either work. In the table of Hopfield and Kliarman there is the line \(\lambda = 2536.2\) Å, identified as belonging to Fe I \((2535.60\) Å) or Fe II \((2536.67\) Å); in Durand’s table the nearest line is \(\lambda = 2535.5\) Å. Lines close to the Au \(2675.95\) Å line are absent in both tables. In Durand’s table there is likewise no line close to Au \(2427.95\) Å. The Te and Bi lines fall in a very strongly blended region of the spectrum. Therefore the available material does not permit any definite conclusions to be drawn about the presence or absence of Hg, Au, Bi, and Te lines in the spectrum.

Thus it may be stated that these investigations have added nothing to our knowledge of the qualitative chemical composition of the solar atmosphere. To obtain reliable results in this region of the spectrum, it is necessary to use spectral apparatus with considerably greater dispersion.

Very interesting are the concordant results of both works concerning the large width of the Mg II lines and the existence of emission lines at the centers of these lines. The latter, as has already been noted, had been known only for the Ca II line—lines very intense in absorption and, at the same time, strong in emission in the upper parts of the solar atmosphere. The possibility of such a phenomenon also for the Mg II lines was noted by Menzel.

Finally, mention must be made of the decline, in the ultraviolet region of the spectrum, of the total energy of the Sun’s radiation in comparison with the radiation of a black body at 6000°, which is caused by the considerably greater density of spectral lines than in the visible region, and which well confirms theoretical expectations.

Fig. 18. Computed heights to which X-ray radiation penetrates in the region 1–10 Å (transmission 1%).

Let us now turn to the consideration of results obtained in the short-wavelength ultraviolet region without spectral resolution.

Of greatest interest are measurements of X-ray radiation. A comparison of the data of the work of Tousey, Watanabe, and Purcell and the data of the work of Friedman, Lichtman, and Byram gives a rather consistent result concerning the penetration of X-ray radiation with wavelengths lying in the interval 0–8 Å to a depth of 80–90 km. The upper boundary of this interval (8 Å) is determined by the transparency of the beryllium foil used. It should be noted that, according to Shklovsky’s calculations3, the entire region of X-ray radiation with \(\lambda < 75\) Å penetrates to a height of approximately 90 km. This height corresponds precisely to the height of the atmospheric \(E\)-layer. Thus, the theoretical prediction that the principal agent causing the formation of the \(E\)-layer is X-ray radiation finds good experimental confirmation.

On the other hand, the experimentally found penetration depth of X-ray radiation, \(\sim 80\)—90 km, leads to the conclusion that the Sun’s X-ray radiation is apparently rather sharply bounded on the short-wavelength side. Indeed, the effective cross section for absorption of X-ray radiation rapidly decreases with decreasing wavelength. In Fig. 18 is shown the penetration depth of X-ray radiation lying in the region 10—1 Å, calculated from Shklov-

... for effective cross sections and Havens’ formula for pressures (the calculation was carried out for a transmission value of 1%).

As follows from this figure, the penetration depth of radiation in this wavelength interval increases rapidly as $\lambda$ decreases. The detection of X-ray radiation at an altitude greater than 80 km thus leads to the conclusion that the lower wavelength limit at which this radiation still has appreciable intensity lies near 5–6 Å. This result is to a considerable extent unexpected; however, attempts to interpret it are premature—careful verification of the experimental results is necessary.

As for quantitative data on the intensity of the X-ray radiation, here, first of all, it should be noted that, on the basis of the results of the works considered, it appears very probable that there are short-lived flashes of this radiation, possibly connected with the well-known flashes of radio emission from the corona.

Quantitative data on the radiation flux are available only in $^{19}$; the authors estimate the radiation flux outside the Earth’s atmosphere to be of the order of $10^{-4}$ erg/cm$^2$ sec. In accordance with what has been said, this flux should be assigned to the wavelength interval $\Delta \lambda \sim 2$ Å near $\lambda \sim 8$ Å.

According to Shklovsky’s calculation, at a coronal temperature $T \sim 10^6$ degrees the radiation flux with $\lambda < 75$ Å amounts to 15% of the continuous emission of the corona beyond the limit of the Lyman series, i.e.
$0.15 \cdot 5.6 \cdot 10^{-2}$ erg/cm$^2$ sec $= 8.5 \cdot 10^{-3}$ erg/cm$^2$ sec, and the intensity of this radiation in the wavelength region of interest to us depends little on wavelength. It follows that the radiation flux falling within the interval $\Delta \lambda \sim 2$ Å will be approximately

$$ 2 \cdot \frac{8.5 \cdot 10^{-3}}{75} \simeq 2 \cdot 10^{-4}\ \text{erg/cm}^2\text{sec}. $$

The agreement between the experimental and theoretical data proves to be very good, although it is possible that it is accidental.

Let us now turn to the results for ultraviolet radiation proper. Table IV brings together the experimental data of $^{17}$ and $^{19}$ for the penetration depth of radiation in various spectral intervals, as well as the data calculated from Fig. 2 under the assumption that oxygen at an altitude of 100 km and above is dissociated. The table also gives experimental data on the magnitude of the radiation flux and data calculated for black-body radiation at $T = 5700^\circ$ and $T = 4800^\circ$.

A comparison of the experimental data of the two works is possible only for the spectral region 1050–1350 Å. For this region there is fairly good agreement both with respect to the penetration depth and with respect to the magnitude of the radiation flux. With respect to the remaining portions of the spectrum, the expe-

Table IV

experimental data on the depth of penetration of radiation agree well with the theoretical data. The existence of a “window” near \(\lambda \sim 1200\) Å and of a second window near \(\lambda \sim 2100\) Å is confirmed. Thus, the assumption, very important for the theory of the ionosphere, that \(L_\alpha\) radiation (1215 Å) penetrates deeper than the \(E\)-layer receives experimental confirmation*). It should also be noted that the sharp increase, found experimentally in \(^{19}\), beginning at an altitude of \(\sim 95\) km, in the transparency of the atmosphere for \(\lambda \sim 1500\) Å (which corresponds to the absorption maximum of molecular oxygen) indicates that at this altitude complete dissociation of oxygen begins. If oxygen were not dissociated, this radiation would penetrate only to an altitude of about 130 km.

With regard to the magnitude of the radiation flux, both works give for the spectral interval 1100—1350 Å data which, in general (if one takes into account the possible measurement errors cited by the authors), agree satisfactorily in order of magnitude. These data also agree satisfactorily with the theoretical expectation that the principal radiation in this region is due to the photosphere of the Sun, with more satisfactory agreement between the experimental and theoretical data obtained if the temperature of the Sun is taken to be \(T = 4800^\circ\).

For the region 795—1050 Å the experimentally found value is several orders of magnitude higher than the corresponding photospheric radiation. Thus, for this region of the spectrum it is necessary, in accordance with existing ideas, to assume that the principal radiation is determined by the chromosphere and the corona. Shklovskii’s calculations, however, give a considerably smaller flux than that observed experimentally, even if one assumes that the monochromatic radiation Ne VIII with \(\lambda = 768/776\) Å falls within this interval.

*) According to the new data of Weissler \(^{10}\), the absorption coefficient of molecular oxygen near \(\lambda \sim 1300\) Å is less than \(10\ \mathrm{cm}^{-1}\). Lyman notes that in this region of the spectrum air at normal pressure transmits radiation even for a layer thickness of several centimeters \(^{26}\).

It should be noted, however, that the authors of the works considered did not take into account one quite possible cause of substantial errors in the experimental results. The fact is that both the phosphors and the counters used for recording the radiation have, on the long-wavelength side, a sensitivity that is not sharply limited, but merely has one or another decline in sensitivity. The flux of solar radiation, considered as that of a black body, increases extremely rapidly toward long wavelengths. Therefore it is not impossible that the phosphors and counters recorded not only the radiation in the expected short-wavelength interval of the spectrum, but also considerably longer-wavelength radiation.

For illustration of what has been said, Fig. 19 gives: \(S_\lambda\)—the spectral sensitivity of the phosphor according to the data of Fig. 9; the extrapolation of this curve from 1400 to 2400 Å is shown by the dashed line; \(I_\lambda\)—the distribution of energy in the radiation of a black body at \(T = 4800^\circ\) (the radiation at \(\lambda = 4600\) Å is taken as 1); \(S_\lambda I_\lambda\)—the product of the two curves.

Fig. 19. \(S_\lambda\)—the spectral sensitivity of the phosphor according to the data of Fig. 9; the extrapolated part of the curve is shown by the dashed line. \(I_\lambda\)—the distribution of energy in the spectrum of a black body at \(T = 4800^\circ\) (the radiation at \(\lambda = 4600\) Å is taken as 1). \(E_\lambda = S_\lambda I_\lambda\)—the energy recorded by the phosphor at various wavelengths.

As is seen from this figure, if the extrapolation of the phosphor sensitivity curve is correct, the energy recorded by the phosphor in the wavelength region \(\lambda > 1400\) Å is of the same order as the energy recorded in the region \(\lambda < 1400\) Å. To this must also be added that, since radiation in the region 1500–2500 Å penetrates much more deeply than the region \(\lambda < 1400\) Å, the effective exposure time of this longer-wavelength region of the spectrum is considerably greater than ...

short-wavelength boundary. Judging from Fig. 12, counters also have an analogous sensitivity “tail”; moreover, as indicated above, the long-wavelength boundary tends to shift toward longer waves. This state of affairs compels one to treat with caution the results set forth above, both with respect to the magnitude of the recorded energy and with respect to the depth of penetration of radiation of different wavelengths. It is possible that this very circumstance, in particular, explains the discrepancy between the results given in Tables II and III, as well as the registration of radiation by counters in the region \(\lambda \sim 1500\ \text{Å}\), beginning at altitudes below \(50\ \text{km}\), and in the region \(\lambda \sim 1900\ \text{Å}\), beginning at an altitude of \(7\ \text{km}\).

Undoubtedly, substantially more complete and reliable results can be obtained by using spectral decomposition of the Sun’s radiation. Here, however, two serious difficulties arise. The first of them consists in the need to eliminate scattered light in the instrument. In view of what has been set forth above, the scattered light from the long-wavelength part of the spectrum must be reduced to \(10^{-4}\)—\(10^{-5}\) of the incident light, which is an exceptionally difficult task. Secondly, the exposure time must be considerably increased.

Recently there appeared\(^{15}\) a detailed description of a tracking system that ensures the pointing of the optical axis of a spectrograph at the Sun in two dimensions during the rotation and precession of a rocket, which should have made the exposure sufficient for photographing spectra down to \(\lambda \sim 500\ \text{Å}\).

The article indicates that the first model of the instrument was destroyed in a rocket accident and that a simplified instrument is being manufactured. However, no further reports on this question have yet been published.

In conclusion—a few words about the possibility of ground-based observations. Undoubtedly, daily observations of the Sun’s ultraviolet radiation with a stationary installation would have exceptionally great value. As was already indicated in the introduction, the only region of the spectrum where such measurements are at present conceivable is the region near \(\lambda \sim 2100\ \text{Å}\). Although this region of the spectrum does not make it possible to observe the radiation of the corona and chromosphere, which is now of greatest interest, systematic observations of the ultraviolet radiation of the photosphere would undoubtedly also be useful. The question of the possibility of observations from small altitudes in the “window” near \(2100\ \text{Å}\) has been discussed more than once\(^{11}\). There have also been several experimental attempts at such observations. Thus, in the communication by Meyer, Shein, and Stoll\(^{27}\), observation of a positive effect was reported during work at an altitude of \(3500\ \text{m}\) (Jungfrau saddle) using a monochromator and photon counters. Positi-

... effect was also observed by Moller^28. Kiepenheuer at first did not obtain a positive result when using a double monochromator; then, on one occasion, he detected traces of the effect and, finally, when working with a single monochromator, obtained what in his opinion was an undoubted effect^29. Mayer, Stoll, and Moller^30 deny the existence of the effect in Kiepenheuer’s results, while at the same time they regard the presence of the effect in their own experiments as reliably established. In some of these investigations, control experiments were made to exclude the influence of longer-wavelength radiation; however, these experiments are not entirely convincing, and it is very probable that the observed effect was caused precisely by this radiation. Regener^31 (with careful filtration of the long-wavelength radiation) did not detect radiation near \(\lambda \simeq 2100\ \text{\AA}\) at an altitude of 25 km (balloon sondes). Regener, however, worked with spectral decomposition and photographic recording, i.e., with a sensitivity considerably lower than in the work of Friedman, Lichtman, and Byram. If the results of this work for \(\lambda = 2100\ \text{\AA}\) are not distorted by long-wavelength radiation, then the use of a condenser system would make it possible to measure this radiation at an altitude of 3–4 km.

The author expresses his gratitude to A. B. Severny and I. S. Shklovskii for valuable advice and comments in the preparation of the present review.

Cited Literature

1. M. Waldmeier, Results and Problems of Solar Research, IL, Moscow, 1950.
2. C. Moore, A. King, Pub. ASP 55, 36 (1943); H. Babcock, A. King, Pub. ASP 55, 111 (1943).
3. I. S. Shklovskii, Bulletin of the Crimean Astrophysical Observatory 4, 80 (1949).
4. Götz, Strahlentherapie 40, 690 (1931).
5. Fabry et Buisson, Gerlands Beitr. 24, 1 (1929).
6. C. Moore, A. Multiplet Table of Astrophysical Interest, Princeton, New Jersey (1945).
7. J. Hopfield, Astrophys. J. 104, 208 (1946).
8. E. R. Mustel and A. B. Severny, DAN 80, 867 (1951).
9. E. Warburg, Berliner Sitz. Ber. 230 (1915); W. Heilpern, Helv. Phys. Acta 19, 245 (1946).
10. G. Weisslerg and Po, Phys. Rev. 83, 888 (1951).
11. E. Meyer, Helv. Phys. Acta 14, 625 (1941); E. Vassy, Rev. d’Optique 15, 81 (1936).
12. W. Baum, F. Johnson, J. Oberly, C. Rockwood, C. Strain and R. Tousey, Phys. Rev. 70, 781 (1946).
13. E. Durand, J. Oberly and R. Tousey, Phys. Rev. 71, 827 (1947).
14. E. Durand, J. Oberly and R. Tousey, Astrophys. J. 109, 1 (1949).
15. H. Clark, Electronics, October 1950, p. 71.
16. J. Hopfield and H. Clearman, Phys. Rev. 73, 877 (1948).

1. R. Tousey, K. Watanabe and J. Purcell, Phys. Rev. 83, 792 (1951).
2. K. Watanabe, Phys. Rev. 83, 785 (1951).
3. H. Friedman, S. Lichtmann and E. Byram, Phys. Rev. 83, 1025 (1951).
4. A. Compton and S. Allison, X-Rays in Theory and Experiment, N. Y., 1947.
5. Best, Durand and Havens, Phys. Rev. 70, 985 (1946).
6. F. Hoyle, Some Recent Research in Solar Physics, Cambridge, Massachusetts, 1949.
7. W. Preston, Phys. Rev. 57, 887 (1940).
8. R. Ladenburg and C. Van Voorhis, Phys. Rev. 43, 315 (1933).
9. T. Burnight, Phys. Rev. 76, 165 (1949).
10. T. Laymann, Phys. Rev. 48, 149 (1935).
11. E. Meyer, M. Schein, B. Stoll, Nature 134, 535 (1934); Helv. Phys. Acta 7, 670 (1934).
12. O. Mohler, Astronom. J. 46, 33 (1937).
13. K. Kiepenheuer, Naturwiss. 26, 678 (1938).
14. E. Meyer, B. Stoll und Muller, Helv. Phys. Acta 12, 415 (1939).
15. V. Regener, Naturwiss. 26, 141 (1938).

4. Reference number as printed in the source. ↩↩↩